853 C55 Q36 1996 PUBL WONAL 1996 ANCER Quality of Life in Clinical Cancer Trials Number 20 INSTITUTE 4L OF THE NATIONAL CANCER INSTITUTE — Contents Introduction . vii Claudette G. Varricchio, Mary S. McCabe, Edward Trimble, Edward L. Kom Trial-Related Quality of Life: Using Quality-of-Life Assessment to Distinguish Among Cancer 1 Therapies Carolyn Cook Gotay Quality-of-Life End Points in Cancer Clinical Trials: The US. Food and Drug Administration 7 Perspective Julie Beitz, Clare Gnecco. Robert Justice Costs of Quality-of—Life Research in Southwest Oncology Group Trials 1 1 Carol M. Moinpour Modeling Health-Related Quality of Life: the Bridge Between Psychometric and Utility-Based 17 Measures Pennifer Erickson Taking Quality of Life Into Account in Health Economic Analyses 23 Jane Weeks Measuring Quality of Life in Culturally Diverse Populations 29 Richard B. Wamecke, Carol Estwing Ferrans, Timothy P. Johnson, Gloria Chapa-Resendez, Diane P. O’Rourke, Noel Chavez, Susan Dudas, Eva D. Smith. Lucy Martinez Schallmoser, Roger P. Hand, Thomas Lad Empirically Selected Instruments for Measuring Quality-of—Life Dimensions in Culturally 39 Diverse Populations , Frank Baker, David Jodrey, James Zabora, Charlene Douglas, Patricia Fernandez-Kelly Quality-of-Life Research in the Pediatric Oncology Group: 1991-1995 49 Andrew S. Bradlyn, Brad H. Pollock Model for Quality-of—Life Research From the Cancer and Leukemia Group B: the Telephone 55 Interview, Conceptual Approach to Measurement, and Theoretical Framework Alice B. Komblith, Jimmie C. Holland A Cooperative Group Report on Quality-of-Life Research: Lessons Learned 63 Mary S. McCabe Cancer and Leukemia Group B (CALGB) 67 Alice B. Komblith Eastern Cooperative Oncology Group (ECOG) 73 Diane L. Fairclough, David F. Cella Gynecologic Oncology Group (GOG) 77 Donald G. Gallup, David F. Cella North Central Cancer Treatment Group (NCCTG) 79 Charles L. Loprinzi Radiation Therapy Oncology Group (RTOG) 81 Todd Wasserman, Deborah Bruner. Charles Scott Southwest Oncology Group (SW OG) 83 Laura C. Loll, Carol M. Moinpcur, Polly Feigl Childrens Cancer Group (CCG) 87 William E.» MatcLean, Jr. ( C onlents continued on back cover) mm mm mm ftiEEE-E—L—EY L I B R A R Y UNIVERSITY or CALIFORNIA Become a National Cancer Institute Information Associate and receive one copy of this and each upcoming Monograph at no additional charge! As an Information Associate, you customize your own package of benefits to meet your individual needs. In addition to Journal Monographs, you can receive the distinguished Journal of the National Cancer Institute (available by subscription exclusively to members) . . . communicate with your peers in the oncology community through our dial-up Bulletin Board System . . . or search PDQ, the National Cancer Institute’s comprehensive cancer information database, via Internet. Best of all, the cost of becoming an Information Associate is surprisingly modest — just $100 for US. members and $150 for non-US. members per year. And if you’re not completely satisfied, we’ll refiind your money. To find out more about NCl’s Information Associates Program, simply complete the form below and return it to NCI today. Or, tojoin today, call 1-800-NCI—7890 within the United States or 301-496-7600 outside the United States. QINSC [3% YES! I want to learn how I can access National Cancer . , . . Institute 5 network of oncology Information resources. ASSOCIAIES FROG RAM Name []Dr.[ ]Mr.[ ]Ms.[ ]Mrs. Organization Address Address Phone Fax E-mail Return this form to NCI Information Associates Program, 9030 Old Georgetown Road, Bethesda, MD 20814-1519, USA, or fax to 301-231-6941. MONOG JOURNAL OF THE NATIONAL CANCER HS INSTITUTE Number 20 ISSN 1052-6773 Richard D. Klausner Director, National Cancer Institute Susan Molloy Hubbard Director, International Cancer Information Center Julianne Chappell Chief, Scientific Publications Branch EDITORIAL BOARD Barnett S. Kramer Editor-in-Chief J. Gordon McVie European Editor Eric J. Seifter Book Reviews Editor J. Paul Van Nevel News Editor Douglas L. Weed Reviews Editor Martin L. Brown Economics Editor 1996 U3. DEPOSITORY ASSOCIATE EDITORS Susan G. Arbuck Frank M. Balis William J. Blot Peter M. Blumberg John D. Boice, Jr. Louise A. Brinton Bruce A. Chabner Ross C. Donehower Susan S. Ellenberg Suzanne W. Fletcher Michael A. Friedman John K. Gohagan Frank J . Gonzalez Michael M. Gottesman Peter Greenwald Donald E. Henson Susan M. Hubbard Frederic J. Kaye Hynda K. Kleinrnan EDITORIAL ADVISORY BOARD James O. Armitage Bruce C. Baguley Laurence H. Baker William T. Beck Clara D. Bloomfield Benjamin Bonavida George J. Bosl Barry Brown C, Norman Coleman 0. Michael Colvin Thomas H. Corbett Pelayo Correa Stephen P. Creekmore Johanna T. Dwyer Merrill J. Egorin Soldano Ferrone Isaiah J. Fidler Richard 1. Fisher David FitzGerald Dystein Fodstad Antonio Fojo Lori J, Goldstein Harvey M. Golomb Elieser Gorelik Jean L. Grem Arnold H. Greenberg Maureen M. Henderson Gloria H. Heppner Ronald B. Herben'nan Waun Ki Hong William J. Hoskins Alan N. Houghton James N. Ingle David H. Johnson V. Craig Jordan John S. Kovach Theodore S. Lawrence W. Marston Linehan Marc E. Lippman Scott M. Lippman Dan L. Longo Douglas R. Lowy Susan G. Nayfield David L. Nelson Kenneth Olden Oliver W. Press Alan S. Rabson Robert H. Shoemaker Richard M. Simon Michael B. Spom Maryalice Stetler—Stevenson J. Paul Van Nevel Douglas L. Weed Noel S. Weiss Margaret L. Kripke Mark G, Kris Donald W. Kufe Bernard Levin Brian R. Leyland-Jones Allen S. Lichter Guy McClung Frank L, Meyskens Anthony B. Miller Malcolm S. Mitchell James J. Mule C. Kent Osborne John J. O’Shea David M. Ota David F. Paulson Henry C. Pitot Igor B. Roninson Edward A. Sausville APR 1 8.1995 STATISTICAL EDITORS Donald A. Berry Barry W. Brown Susan S. Ellenberg Scott S. Emerson Eric Feuer Laurence S. Freedman Edmund A. Gehan Sylvan B. Green Susan G. Groshen Richard A. Olshen Philip C. Prorok Philip S. Rosenberg Richard M. Simon Donald M. Stablein Robert E. Tarone Peter F. Thall Thomas J. Sayers David Schottenfeld Herman A, J. Schut Richard K. Severson William R. Shapiro Roy E. Shore Paul M. Sondel Patricia S. Steeg Herman D. Suit Sandra M. Swain Mario Sznol Raymond Taetle Ian Tannock Joel E. Tepper J. Tate Thigpen Peter R. Twentyman Larry M. Weisenthal EDITORIAL STAFF Scientific Editors: Claudette G. Varricchio, D.S.N., R.N. Mary S. McCabe, R.N., B.S. Edward Trimble, M.D. Edward L. Korn, Ph.D. Monograph Coordinator: Elaine Price Beck Manuscript Editors: Elaine Price Beck Joan O’Brien Rodriguez Editorial Assistants: Olga P. Sisson Chandra McNeil Smith MARKETING STAFF Marketing Director: Jean Griffin Baum Press Contact: Kate Nagy EDITORIAL POLICY. Manuscripts from key conferences dealing with cancer and close- ly related research fields, or a related group of papers on specific subjects of importance to cancer re— search, are considered for publication. with the understanding that they have not been published previously and are submitted exclusively to Journal oft/16 National C ant'c'r Institute Monographs. All material submitted for consideration will be subject to review, when appropriate, by at least one outside reviewer and one member of the Editorial Board of the Journal oft/w National Cancer In— stitute. Opinions expressed by the authors are not necessarily those of the publisher or the editors. Proposals for monographs should be submitted to the Editor-in-Chief. Journal of the National Cancer Institute. National Cancer Institute, 9030 Old Georgetown Rd.. Bethesda. MD 20814. Journal of the National Cancer Institute Monographs are available on request to members of the National Cancer Institute Information Associates Program (one copy of each monograph per mem- ber). Monographs are also available through the US. Government Printing Office. To request a copy of a monograph, or for more information about the lnformation Associates Program. call 1—800—624- 7890 (301—496-7600 outside the United States). Quality of Life in Clinical Cancer Trials Proceedings of a Workshop Held at the National Institutes of Health Bethesda, Maryland March 1—2, 1995 Sponsors National Cancer Institute. Division of Cancer Treatment, Diagnosis, and Centers and Division of Cancer Prevention and Control Workshop Faculty Frank Baker David Machin Julie Beitz Mary S. McCabe Andrew S. Bradlyn Carol M. Moinpour Pennifer Erickson David Osoba Clare Gnecco Leslie Robison Carolyn C. Gotay Edward Trimble Jimmie C. Holland Claudette G. Varricchio Penelope Hopwood Richard B. Warnecke Gwendoline M. Kiebert Jane Weeks Edward L. Kom Quality of Life in Clinical Cancer Trials 1996 Number 20 Contents Introduction vii Claudette G. Varricchio, Mary S. McCabe, Edward Trimble, Edward L. Korn Trial-Related Quality of Life: Using Quality-of-Life Assessment to Distinguish Among Cancer 1 Therapies Carolyn Cook Gotay Quality-of—Life End Points in Cancer Clinical Trials: The US. Food and Drug Administration 7 Perspective , Julie Beitz, Clare Gnecco, Robert Justice Costs of Quality-of-Life Research in Southwest Oncology Group Trials 1 1 Carol M. Moinpour Modeling Health-Related Quality of Life: the Bridge Between Psychometric and Utility-Based 17 Measures Pennifer Erickson Taking Quality of Life Into Account in Health Economic Analyses 23 Jane Weeks Measuring Quality of Life in Culturally Diverse Populations 29 Richard B. Warnecke, Carol Estwing Ferrans. Timothy P. Johnson, Gloria Chapa—Resendez. Diane P. O’Rourke, Noel Chavez, Susan Dudas, Eva D. Smith, Lucy Martinez Schallmoser, Roger P. Hand, Thomas Lad Empirically Selected Instruments for Measuring Quality-of-Life Dimensions in Culturally 39 Diverse Populations Frank Baker, David Jodrey, James Zabora, Charlene Douglas, Patricia Fernandez-Kelly Quality-of-Life Research in the Pediatric Oncology Group: 1991-1995 49 Andrew S. Bradlyn, Brad H. Pollock Model for Quality-of—Life Research From the Cancer and Leukemia Group B: the Telephone 55 Interview, Conceptual Approach to Measurement, and Theoretical Framework Alice B. Kornblith, Jimmie C. Holland A Cooperative Group Report on Quality-of—Life Research: Lessons Learned 63 Mary S. McCabe Cancer and Leukemia Group B (CALGB) f. 67 Alice B. Kornblith i : {’9' ’ —’ Eastern Cooperative Oncology Group (ECOG) ' ‘ fl 73 Diane L. Fairclough, David F. Celia ~ ‘ ' T Gynecologic Oncology Group (GOG) 1 f K x 77 Donald G. Gallup, David F. Cella ’ North Central Cancer Treatment Group (NCCTG) / , -’ .- 79 Charles L. Loprinzi ,/ f . ‘ Radiation Therapy Oncology Group (RTOG) / . 81 Todd Wasserman, Deborah Bruner, Charles Scott / V , ’ Southwest Oncology Group (SWOG) 83 Laura C. Loll, Carol M. Moinpour, Polly Feigl Childrens Cancer Group (CCG) 87 William E. MacLean, Jr. Pediatric Oncology Group (POG) Andrew S. Bradlyn, Brad H. Pollock Quality of Life in Clinical Cancer Trials: Experience and Perspective of the European Organization for Research and Treatment of Cancer Gwendoline M. Kiebert, Stein Kaasa Assessment of Quality of Life in Clinical Trials of the British Medical Research Council David Machin United Kingdom Cancer Research Campaign Approach to Quality-of—Life Research in Cancer Clinical Trials Penelope Hopwood Health-Related Quality-of—Life Studies of the National Cancer Institute of Canada Clinical Trials Group David Osoba, Janet Dancey, Benny Zee, James Myles, Joseph Pater 89 91 97 107 List of Workshop Participants 113 Introduction Claudette G. Varricchio, Mary S. Mc'Cabe, Edward Trimble, Edward L. K0rn* In July 1990 (I). the National Cancer Institute (NCI) held a quality-of-life (QOL) meeting (a) to define elements of QOL that are relevant to clinical decision making and serve as end points in cancer clinical trials, (b) to evaluate currently available instruments for QOL assessments and strategies for implementa- tion. (r) to identify site—specific questions of high priority. and (d) to examine issues regarding the integration of findings from therapeutic evaluations and QOL measurements. The meeting reported in this monograph, “Workshop on Quality of Life in Clinical Cancer Trials." represents the next step in advancing QOL research in NCI trials through an assess- ment of the progress that has been made in NCI-sponsored QOL research during the last 5 years. The general goal of this meeting was to focus on and to refine expectations for QOL research so that essential evaluations can be done in QOL as an addition to other clinical information. particularly in a time of limited re- sources. The specific focuses of the current workshop were (a) to re-evaluate clinical research areas in which QOL questions are a priority. (b) to address issues of implementation and col- lection of QOL data in NCI—sponsored clinical trials, and (c) to focus on new methods in QOL research, such as outcome studies and methods of assessing QOL in culturally diverse populations. The meeting participants included the QOL researchers from the NCI’s Cooperative Groups and the Community Clinical On— cology Program (CCOP) research bases. representatives from the British Cancer Research Campaign, the Medical Research Council (U.K.), NCI-Canada. and the European Organization for Research and Treatment of Cancer. As a formal part of the meeting, participants were asked to review current QOL re- search in their respective groups and to present the groups’ plans for future QOL investigations. This monograph presents the papers from the meeting with selected discussion. A summary of NCI—sponsored QOL pro— tocols from the groups is included along with the QOL instru- ments used to address the research questions. When available, citations of published findings are listed. Many issues were explored in the papers presented. Some of these issues were as follows: ls QOL needed in every trial? What are the advantages and disadvantages of including QOL? Given the resources needed for QOL studies. how can groups best set priorities on the use of group resources? Since inclusion of women and minorities is mandated in the National Institutes of Health (NIH) Revitalization Act of 1993 (Public Law l()3-43) (2). how do the groups plan to comply with the NIH guidelines? What is the best way to develop and refine measurement scales. especially disease—specific modules and cultural adaptations/ Journal of the National Cancer Institute Monographs No. 20. [996 translations that are valid and reliable? These issues require fur- ther thought and investigation. Other areas of interest or speculation came from the discus- sions. They included the manner in which QOL research is presented and the forums where results are presented. Inclusion of QOL results reported with clinical end points often leads clinicians to question the rationale for QOL research and the clinical application of the findings. The question “What are QOL data and how does one use them?” must be answered e10- quently and cogently. The question of multiple measures in trials versus standard measures used across trials was raised. Is comparability across trials warranted at this point in the devel- opment of QOL measures? Is there a need for depth (disease- specific question) in measurement as well as a need for breadth (all domains of QOL) in conceptual issues? A goal must be to understand not only who are doing better, but also why they are doing better. Criteria for excluding a subject from a trial must be looked at closely and must be justified if the trials are to be rep- resentative and generalizable. How does one design QOL studies to be inclusive? Special populations include those with linguistic and cultural diversity, persons with low literacy ability. children. the elderly, and hearing-impaired or visually impaired persons. What should the sample size be for QOL find- ings to be meaningful? Is it the same, larger, or smaller than that needed to answer a treatment question? Operational issues are of concern to the groups who are faced with the pragmatic reality of limited resources and manpower. These questions will no doubt need to be addressed through the conduct of trials. The results of QOL assessment in clinical trials should focus on interventions to lessen the negative impact of cancer and its treatment on QOL; these interventions must build on the descriptive QOL research findings. QOL will continue to ex- pand in clinical cancer research, such as prevention trials in which risk assessment (e.g., genetic or environmental exposure) and notification methods are developed and tested. There will be a need for assessment of the effect of knowledge of risk status on the person’s QOL. The translation of QOL findings into valid, effective clinical applications is the most important con— cern of researchers and clinicians at this time. It is important *Afli‘liutimzs (Ifrlll!ll()l‘.\'.‘ C. G. Varricchio (Division of Cancer Prevention and Control), M. McCabe. E. Trimble. E. L. Kom (Division of Cancer Treatment, Diagnosis. and Centers). National Cancer Institute, Bethesda, MD. (‘orresptindent-c In: Claudette G. Vanicchio, D.S.N., R.N.. National Institutes of Health. 6l30 Executive Blvd, MSC 7340, Bethesda, MD 20892-7340. Soc “Note" section following “References." vii that QOL research continues to develop and address cancer re- (2) Faderal Regis‘“ VOL 59. N0~ 59, Mmh 28, 1994- search questions of clinical importance to patients. Note References Workshop supported by the Division of Cancer Prevention and Control and Division of Cancer Treatment, Diagnosis and Centers, National Cancer Institute. (I) Nayfield SG, Hailey BJ, McCabe M. Quality of life assessment in cancer clinical trials: repon of the Workshop on Quality of Life Research in Cancer Clinical Trials. July 16—17, 1990. Bethesda (MD): US DHHS, 1991. viii Journal of the National Cancer Institute Monographs No. 20, 1996 Trial—Related Quality of Life: Using Quality—of—Life Assessment to Distinguish Among Cancer Therapies Carolyn Cook Goray* Issues in selecting quality-of-life (Q()L) measures that are best suited to assessing differences among treatments in can- cer clinical trials. as well as challenges to interpreting QOL outcome data. are discussed. When used in the context of randomized trials of cancer therapies. Q()L assessments must provide an answer to the question, “Did the treatments differentially affect patient well-being?" In order to detect differences in treatment efficacy against a background of great similarity. the broad concept of Q()L needs to be refined to reflect “trial-related QOL.” In many cases. this will entail emphasis on actual patient experience of symptoms and functional changes. as opposed to emphasis solely on evaluation and satisfaction. A model is proposed to identify cognitive. emotional. and sociocultural factors that influence a patient’s QOL evaluation and that need to be considered in understanding the meaning of QOL data. [Monogr Natl Cancer Inst 1996;20:1-6] The potential contributions of quality-of—Iife (QOL) data to cancer therapy evaluation are increasingly recognized. Only a handful of phase III clinical trials that include results of the QOL assessments have been published to date (/,2). Active portfolios of trials including QOL outcome measures. however. are maintained by all of the multisite clinical cooperative groups in the United States (3). as well as trial groups in Canada (4). Europe (including the UK.) (5). and Australia (6). The results of a number of these ongoing trials will begin to be available in the next few years. For example. the results of the first Southwest Oncology Group QOL studies will be reported at the plenary session of the October 1995 group meeting (Moinpour C: per- sonal communication. 1995). With increased interest in QOL assessment and concomitant resources devoted to this activity come increased expectations for the contributions of Q()L data to interpreting trial data. Some of these expectations may not be met. In the enthusiasm for using QOL measures. limitations to their interpretation have not been fully considered. This article discusses issues in select— ing QOL measures that are best suited to assessing differences between treatments in cancer clinical trials. as well as chal— lenges to interpreting QOL outcome data. Discussion will focus on the purpose of assessing QOL in a given trial. specific chal- lenges to QOL assessment posed by randomized trials. and dif- ficulties in interpreting QOL data. We will propose a model to explain QOL and to identify needs for additional research. Journal of the National Cancer Institute Monographs No. 20, 1996 For What Purpose Is QOL Being Assessed in 3 Given Trial? There are numerous reasons why QOL assessment might be included in a particular cancer treatment study (7-9). Cella and Tulsky (9) have provided a parsimonious taxonomy of purposes for measuring QOL: l) to identify the full range of side effects and impacts of the treatments in order to assess rehabilitation needs. 2) to compare treatments in a trial. and 3) to use QOL ratings as a predictor of response to future treatment. Data col- lected for the first purpose can identify patients at risk to pro— vide supportive interventions and to modify treatment regimens. Data collected for the second purpose can be used to determine which treatment should be the standard of care. With regard to the third purpose. base-line QOL scores can serve either as a prognostic indicator or as a basis for stratification in the random assignment of patients to treatments. These purposes cannot necessarily be achieved through the same approach to measurement. For example. documenting the impact of treatments may require a comprehensive assessment that includes questions about patient experience in the multiple dimensions that make up QOL. Virtually all researchers agree that QOL involves a number of relatively independent domains. including. at a minimum. physical. functional. psychological. and social well-being. Some researchers also emphasize other areas. such as symptoms. sexuality. spiritual concerns. and satis— faction with health care (/0). A broad and comprehensive ap— proach to assessment is likely to be particularly useful for treatments for which little is known about potential effects on patient well—being. New treatments may have an impact in areas that are not expected by the investigators. In one of the earliest studies in this area. Sugarbaker et al. (1/) demonstrated that radiation therapy used in limb—sparing procedures had an unan— ticipated negative impact on patient sexual functioning. For dis- tinguishing among treatments. the same broad approach to assessing QOL that is useful in understanding patient experience may not provide the specific information that is necessary to distinguish between treatments. This point will be discussed in more detail in the next section. *('nrrmpondwn'1’ Io: (‘arolyn Cook Gotay. Ph.D.. University of Hawaii Can- cer Research Center. 1236 Lauhala St.. Honolulu. HI 96813, See "Notes" section following “References." The use of QOL data for prognosis or stratification is suffi- ciently recent that the best approach to assessment has not yet been identified. Several studies (12.13) have shown that patient- rated overall QOL assessments predict survival better than physician—rated performance status. Coates et al. (14) made a head-to-head comparison of patient and physician QOL ratings. Breast cancer patients participating in a clinical trial of chemotherapy completed a QOL questionnaire (including ques— tions on physical well-being. mood. pain. nausea and vomiting. and appetite as well as an overall rating). and their physicians also completed the Quality of Life Index (QLI). a multidimen- sional QOL assessment (15). and a performance status measure. Results showed that patient ratings of physical well-being (but not overall QOL) and physician-rated QLI were both statistical- ly significant and independent predictors of length of survival. This study points out the need for additional research to identify the best approach to QOL assessment for prognostic purposes (16). It should not be inferred from the above discussion that QOL assessment can be used for one purpose and one purpose only in a given study. Often there are multiple reasons why QOL as- sessment should be conducted and several different ways this in- formation can be applied. QOL assessment. however. generally takes place in the context of limited resources. Not only are the data management and analysis capabilities of the cooperative groups limited. but also the patient‘s ability to provide informa- tion may be limited by fatigue and motivation. Priorities need to be set to ensure that the appropriate QOL data are available to address the study purpose. If resources are sufficient to permit additional data collection. such data are invariably likely to pro— vide useful information for patient care. At a minimum. how- ever. the researcher needs to be certain to collect data appropriate to study the hypotheses. What Are Constraints to Measuring QOL in Clinical Trials When the Goal Is to Distinguish Among Treatments? It is reasonable to consider that one of the primary reasons that QOL assessment has been accepted and adopted in therapeutic and drug development research derives from the cur— rent status of clinical research in cancer. For many cancers. there have not been major improvements in therapeutic cure rates since the success of chemotherapeutic agents in the l960s. Many patients are living longer. however. even if they ultimate— ly die of their disease. In this context. additional measures of treatment efficacy. such as QOL assessments. assume increased importance in determining standards of care in cancer treatment and the approval of new pharmaceutical agents. Phase III trials are the gold standard for evaluating new can— cer treatments. In phase III trials. patients are randomly assigned to one of two or more treatments. generally including the stan— dard “best treatment” and a new treatment that is believed to be at least as good and perhaps better. Eligibility criteria are used to ensure patient safety. as well as to restrict participation to a well-defined group of patients for whom treatment effects may most likely be detected. As a result of these eligibility restric- tions. most participants in phase III clinical trials constitute a selected and nonrepresentative group of patients: as a conse- quence of randomization. the patients in all treatment arms should be equivalent on any important variables related to out- comes except for the treatment they receive. The implications of this design are that detecting differences in QOL between treatment arms is apt to be very difficult. The patients reflect great similarity in their disease status on entry to the study. since site and stage of diagnosis will be identical across treatments. In addition. many aspects of the treatments will be identical. For example. monitoring schedules. tests per— formed. symptom control regimens. and numerous other aspects of treatment are specified and controlled in the study protocol. In addition. the majority of phase III studies currently ongoing in the cooperative groups involve comparisons of different chemotherapeutic regimens. This is also true for many phase II studies. especially those that test new drugs: QOL assessment may also be considered in these studies. as witnessed by the ac— tive interest and participation of the US. Food and Drug Ad- ministration and pharmaceutical industries in this field (17). This situation clearly poses challenges for QOL assessment. since a QOL measure would need to be very sensitive to detect differences in treatment efficacy against a background of great similarity in patient populations. The ability to detect differen- ces between identical patient groups and/or among treatment regimens that are alike on many dimensions requires focused QOL assessment strategies. Most investigators in the cancer field restrict their definition of QOL in clinical trials to health—related QOL (HRQOL). Specifically. most investigators agree that it makes sense to limit the QOL end point of interest in a study of a treatment in— tervention or in a population with compromised health status such as cancer patients to the dimension(s) that are likely to be affected by the intervention or health status of the patient. As a result. there may be aspects of QOL that are very important in an individual‘s “subjective evaluation of life as a whole“ [a fre- quently cited definition of QOL offered by DeHaes (18)] that are excluded from consideration when HRQOL is assessed. These aspects of QOL include dimensions such as the environ- mental quality. physical safety of the neighborhood. and quality of public schools. While these are critical aspects (and in fact are key components of comparative ratings of QOL in other contexts. such as comparing QOL in cities across the country). they are outside the realm of being affected by treatment inter- ventions or health status. We would like to propose that a similar “funneling” occurs in selecting measures of QOL used in clinical trials of cancer treat- ment in order to develop an assessment of trial—related QOL (TRQOL). Specifically. when one wishes to detect differences between two or more treatment arms. a number of the domains commonly included in QOL assessments are unlikely to be dif- ferentially affected by the treatments under study. For example. a comparison of two chemotherapies may be unlikely to have different impacts on spiritual concerns and family functioning. At the same time. the treatments might differ in. for example. their effect on sleep patterns or fatigue. Since these two areas are not assessed beyond a single question (if that) on most HRQOL questionnaires. standard tools would be unlikely to be sensitive enough to show differences among treatments if in- Journal of the National Cancer Institute Monographs No. 20. 1996 deed they existed. Given that there is a limit as to how many questions can be included in a given trial, researchers need to ensure that they cover the critical aspects of TRQOL. If they are fortunate enough to have the resources to asseSs HRQOL, or even QOL, they will gain additional valuable information. (In fact, virtually no research has investigated the relationship be- tween overall QOL and HRQOL. These data would help to put into perspective the relative weight patients attribute to their health concerns in the context of their life as a whole.) However, when QOL assessment is included in order to distinguish among treatments. the most important objective is to be able to answer the question, “Did the treatments differentially affect patient well-being?“ The specific assessments made need to be suffi— ciently sensitive to answer this question. What Are Difficulties in Interpreting QOL Findings? Most of the models of QOL that have been presented to date consist of identifying different dimensions that may influence QOL. However, little attention has been directed toward specifying the relationships between symptoms and functioning, performance and satisfaction, occurrence of a symptom and ex- perience of the symptom as a problem, and most of the other concepts that are loosely described as relating to QOL. Similar~ ly, virtually no attention has been given to identifying variables that predict QOL. apart from determining if QOL varies as a function of treatment. However. the question “What factors are associated with high levels of QOL?" remains unanswered. Part of the difficulty in exploring such a question derives from the way that many researchers define QOL. Most discus— sions of cancer-related QOL stress its subjective nature, em— phasize that QOL can be assessed only from the perspective of the patient, and stress that the patient’s evaluation is the “gold standard" (19). However valid this approach may be for under— standing an individual patient, it is difficult to accept as a criterion for success of cancer treatment. Consider the following case: Mr. G. is an 87—year—old prostate cancer patient. He lives with his wife of 58 years and his daughter and her family in a comfortable home. In addi— tion to his cancer diagnosis. he has several other comorbid diseases. which have left him with one leg amputated above the knee. a brain tumor, a weakened right side. paralysis of half his face. and an eye sewn shut. In addition. he is half-blind in his other eye. He is largely confined to a hospital bed. When he completed the QOL questionnaire and was asked, "Do you have any trouble taking a short walk outside the house?" he responded. “No. If someone helps me out of bed. and I use my two pros- theses, and a couple of canes. I can walk." When he was asked to rate the overall quality of his life from 1 to 10, he selected 10 and remarked. "I have a wife that‘s the tops. two daughters who are quite successful. and my mother and dad were really great. My quality of life is hard to beat. because I‘ve been getting everything I wanted. as far as I‘m concerned. and no problems." [From (20), cited with patient permission] It is clear that the patient’s evaluation of his QOL as “excel- lent” and “a 10" is a candid and accurate reflection of his perspective. Efforts to modify treatments to mitigate symptoms or enhance other outcomes, however, are still to be strived for if possible, despite the patient‘s satisfaction and experienced high levels of QOL. Relying completely on patient evaluations without equal attention to more objective aspects of well-being Journal of the National Cancer Institute Monographs No. 20. 1996 limits the usefulness of QOL data as an outcome measure in cancer treatment. It is clear that there are easier ways to make patients happy than by giving them cancer therapy. In addition, Mr. G’s excellent QOL was certainly influenced by his personal and social resources, as well as his own values and attitudes. Multiple perceptual. motivational, and external factors intervene between an experience related to cancer and/or its therapy (such as a symptom or a change in functioning) and its evaluation by the patient as a problem or an effect on QOL. All of these variables are largely outside the realm of cancer therapy. Models are needed to identify and link independent variables to patient—assessed QOL. Such models will facilitate the development of interventions that incorporate individual patient factors with QOL assessments. Fig. 1 presents a model that attempts to identify factors that may affect a patient‘s evaluation of HRQOL as related to can- cer. This schema elaborates on models presented by Selby (2) and Wilson and Cleary (2]). The model assumes that the patient who is asked to provide a rating of his or her HRQOL engages in a multistep process. Psychological (including cognitive and emotional factors) and sociocultural filters affect how the patient experiences and assesses the effects of cancer diagnosis and treatment. The model begins when cancer is diagnosed and treated. As a consequence of the disease and/or treatment. the patient may ex- perience symptoms and functional limitations. This is one junc- ture when data assessing patient experience provide direct information about whether or not various symptoms are ex- perienced as well as the functional limitations that may result. CANCER DIAGNOSIS CANCER THERAPY FUNCTIONAL LIMITATIONS Individual Sociocultural Resources factors factors DISTRESS SATISFACTION Weighting Synthesis HEALTH—RELATED QUALITY OF LIFE Fig. 1. Conceptual model of health-related quality of life. Measures of functional capacity—whether it is possible for a patient to perform a defined task—as opposed to assessment of performance during everyday activities are more likely to yield data that reflect the effects of treatment as opposed to motiva- tions and lifestyle. (Measures of actual functioning rather than capacity may provide more useful information to guide in- dividual patient rehabilitation.) A number of available QOL assessment questionnaires do not emphasize patient experience; instead. they ask directly about patient evaluations. The specific questions vary according to QOL questionnaire. For example. the QOL Index of Ferrans and Powers (22) asks cancer patients to indicate their satisfaction and rated importance of aspects of functioning. while the majority of questions in the CARES (23) address the degree to which cancer patients have difficulty in different areas. How- ever. whether or not symptoms or functional changes result in distress (or possibly have a positive impact) depends on other variables. These variables include individual factors. such as the patient‘s motivation. interpretation. expectations. and per- sonality. With respect to motivation. some people attempt to supersede potential limitations and find ways to surmount the challenges they face (such as Mr. G. described above); such people would probably not experience distress to the same degree as others. The role that motivation can play in whether a disability be- comes a handicap is well recognized in the rehabilitation litera— ture (24). An individual‘s interpretation of symptoms and other limita— tions also affects the degree of distress that is experienced. For example. patients may be willing to experience many objective- ly negative side effects of treatment (e.g.. chemotherapy-as— sociated emesis) on a short—[em] basis if they believe that they are going to get better as a result of the treatment. This kind of patient interpretation may help to explain the finding of Coates et al. (25) that QOL. as well as tumor response. was better for patients who were given continuous rather than intermittent chemotherapy. Treatment was administered to the patients in the continuous-therapy arm until their disease progressed; hence. receiving treatment may have signified to them that they were doing well. which in turn may have engendered more favorable QOL ratings. This explanation is conjectural. since Coates et al. did not collect data about patient interpretation. but it offers one way to explain somewhat puzzling findings. Expectations may have a great deal to do with whether or not an individual can accept limitations. For patients who expect to be “back to normal." a functional limitation may be much more distressing than for individuals who expect that they might need to live with limitations. The relationship between experience and expectations has been identified as a key determinant of QOL by Calman (26) and Cella and Cherin (27). Many other individual differences may affect the degree to which distress is experienced. For example. there is evidence to support the existence of a dispositional complaining style. Some people express dissatisfaction across situations and would be ex- pected to report lower levels of well—being regardless of the cir— cumstances (28). Premorbid health conditions also need to be considered. For example. the importance of considering pre- vious psychopathology in understanding dysfunctional coping with cancer has been amply demonstrated (29). Comorbid physical problems also affect health ratings. For example, in our experience assessing QOL. patients frequently make comments such as. “l have pain. but I don’t know if it’s because of the can- cer or my arthritis." Collecting data about concurrent and pre- vious health problems would aid in untangling the impact of cancer and cancer therapy from more general concerns. Sociocultural factors also have an influence on whether or not a particular symptom or functional disability gives rise to patient—rated distress. Perhaps the clearest illustration can be found with respect to pain. where cultural variation has been ex- tensively studied. While the ability to detect pain stimuli ap- pears to be equivalent across cultures (30). both the meaning and expression of pain are culturally bound (3/). as research during the past 40 years has demonstrated Ie.g.. (32—34)]. One of the earliest studies in this area (34) showed that while “old Americans" (U.S.-born Anglos of third or greater generation status) tended to be stoic and unemotional in their responses to pain. Italian-American and Jewish patients were expressive and verbal in communicating discomfort. Such ethnocultural dif- ferences could have a major effect on QOL ratings. The resources possessed by the patient constitute another category of individual factors that may influence whether a symptom leads to distress. Economic resources are the most straightforward. For example. if a patient can pay for someone to obtain groceries and clean the house. functional limitations may not be as distressing. Social support is another important resource. A third kind of resource is whether appropriate health care has been provided. For example. a patient may be suffering from depression in response to the cancer diagnosis and be receiving psychoactive medications or other support by his or her physician. The resultant QOL rating may reflect no mood dysfunction because the symptom has been adequately remediated. However. the need to inquire about such matters in the course of assessing HRQOL has not been discussed in the literature. In addition. existing QOL questionnaires do not build in questions about whether a patient is currently receiving sup— port to mitigate symptoms or functional limitations. There is yet a further evaluative step that patients need to take when they make an overall assessment of their HRQOL. Al- though it is an unconscious process for most patients. they need to make a number of judgments in order to render an overall QOL rating. Weights must be assigned according to the subjec- tive importance of different dimensions of well-being. ex- perienced distress must be multiplied by these weights. and these factors need to be synthesized in order to determine a final answer. Depending on individual values. the same distress ratings may give rise to different overall HRQOL scores. The strength of assessing global HRQOL is its emphasis on the importance of individual differences in QOL and the need to consider each patient’s perspective (35). At the same time. this poses a considerable difficulty when one is attempting to draw conclusions about HRQOL as a function of cancer treatment. If QOL is completely subjective and is affected by a multiplicity of factors well outside the jurisdiction of influence by cancer therapy. then it seems a very difficult task to detect differences due to treatment. Random assignment of patient to treatment condition should ensure that the conditions are balanced; i.e., in- Journal of the National Cancer Institute Monographs No. 20. 1996 dividual variations in patient motivations. values. and the like should be equally represented across treatments. Such variables, however. may give rise to so much error in outcome measure— ment that differences cannot be detected. The fact that studies have found differences between treat— ment groups in randomized studies indicates that. despite the considerable individual variation among patients, some differen- ces among treatments are apparently large enough that variation in HRQOL can be detected. Since individual patient values. per- sonality. and so forth are less likely to have been affected by the treatment. such differences in treatments. however. are likely to stem from variation in symptoms and/or functioning. [As we proposed earlier in discussing the study by Coates et al. (25). it is possible that individual factors such as expectations may be differentially affected by treatment ann.] By the same token. some treatments with considerable differences in symptoma- tology and function have not been demonstrated to have a con— sistent and demonstrable difference in HRQOL. Consider the impact of mastectomy versus conservative surgery for breast cancer. where the confluence ofevidence points to no consistent HRQOL advantage for either treatment (36). In this instance. it is likely that individual patient variables (such as the importance of physical appearance). as well as the impact of a diagnosis of cancer apart from treatment. mediate differential HRQOL rat— ings. Attention to factors like those represented in Fig. 1 will aid in interpreting the findings. either differences between treat— ments or lack thereof. and avoid coming to an erroneous con- clusion such as “the kind of surgery received for breast cancer does not affect QOL." What Are Implications for Future Research? Research assessing QOL as a consequence of cancer has made enormous strides in the past decade. HRQOL has been recognized and incorporated as an end point in trials of therapy. procedures have been developed to ensure quality control of the data in multisite studies. and questionnaires to assess HRQOL have been developed and continue to be validated. Several areas. however, deserve additional attention to ensure that HRQOL data fulfill their potential. 1) Basie research is needed to understand more fully the eon— tributions of HRQOL data to cancer therapy evaluation. The relationship between HRQOL data and other measures needs to be clarified to identify the distinct contributions of HRQOL data. For example, what are the relationships among toxicity ratings. symptoms. functioning. and HRQOL? Consider toxicity ratings and HRQOL. We would not expect perfect correlations between patient HRQOL ratings and clinician-rated toxic effects. based on demonstrated differences between patient and observer ratings (37). However. do toxicity ratings demonstrate differences among treatments in the same direction and of the same magnitude as HRQOL ratings? How much additional predictive validity do HRQOL data provide over and above toxicity ratings? Is it possible for toxicity and HRQOL to be noncorrelated or even negatively correlated? The cooperative groups maintain careful records of an extensive bat- tery of toxic effects. which could be compared with patient ratings in those studies that include QOL assessments. This kind Journal of the National Cancer Institute Monographs No. 20. 1996 of analysis could be relatively easily accomplished and would constitute an important contribution to the field by indicating areas where detailed patient reports are most critical. 2) HRQOL measurement needs to be tailored to the study purpose. Given resource constraints. an assessment strategy that addresses multiple purposes may not be possible. For studies in which HRQOL is used to distinguish among cancer treatments. the assessment tool must be sensitive and focused enough to detect small differences against a background of considerable similarity in patients and. frequently. treatment regimens. This requirement may necessitate the use of TRQOL (trial—related QOL) questionnaires in addition to (preferably) or instead of more standard assessment tools. 3) Attention needs to be directed at assessing more objective efl'eets ofeaneer diagnosis and treatment. in addition to patient evaluation. The majority of questionnaires that have been developed to measure HRQOL in cancer are heavily weighted to patient evaluations of distress or satisfaction. However. as we have discussed. patient ratings of distress are affected by many variables that fall well outside the scope of cancer therapy. It is recommended that investigators adopt HRQOL measures that include assessment of symptoms and functional deficits as well as patient—evaluated distress and problems. In fact. the most commonly used measure in health assessment outside of cancer is the Sickness Impact Profile (38). a scale that emphasizes functional assessment. HRQOL assessments in cancer patients should ensure the inclusion of questions that address functional status (39.40). In addition. consideration needs to be given to alternative ap- proaches to HRQOL assessment. Most assessment to date has utilized patient self-reports. Self-reports. however, are no less subject to methodological biases than other sources of data. Each approach to data collection has its own distinct strengths and limitations. and a triangulation approach. which relies on the convergence of data from different sources (4]). is the op— timal approach when possible. Alternative sources of data. such as observer reports. medical records. and behavioral ratings. should be considered in addition to self-reports (42); see (43) for a useful example of a behaviorally based scale assessing several aspects (speech and eating behavior) of HRQOL particularly important for head and neck cancer patients. 4) Models onOL need to be developed and tested. In order to understand why a patient experiences high or low levels of QOL. influences on this evaluation need to be identified and quantified. Little attention has been directed at identifying fac- tors that influence QOL ratings and which cancer patients ex— perience notably high or low HRQOL. As Till (44) pointed out. ultimately. “it may be much more important to try to understand and learn from the process used by the patient to construct his or her report than to obtain the outcome of the process.” The model outlined in this article was presented to stimulate thinking and empirical efforts toward this goal. References (I) Osoba D. Lessons leamed from measuring health»related quality of life in oncology [see comment citation in Medline]. J Clin Oncol 1994;12:608-16. (2) Selby P. Measurement of quality of life in cancer patients. J Pharm Phar- macol 1993:45 Suppl 12384-6. LN (4 (5 (6) (7) (8) (9 (/0) (II (/2 (/3) (/4) (/5) (I6) (/7) (/8) Nayfield SG. Ganz PA. Moinpour CM. Cella DF. Hailey BJ. Report from a National Cancer Institute (USA) workshop on quality of life assessment in cancer clinical trials. Qual Life Res 1992:1z203-10. Osoba D. The Quality of Life Committee of the Clinical Trials Group of the National Cancer Institute of Canada: organization and functions. Qual Life Res 1992:12211-8. Maguire P. Selby P. Assessing the quality of life in cancer patients. Br] Cancer 1989;60:437-40. Olweny CL. Quality of life in cancer care. Med] Aust 1993;158:429-32. Gotay CC. Kom EL. McCabe MS. Moore TD. Cheson BD. Quality-of—life assessment in cancer treatment protocols: research issues in protocol development. J Natl Cancer Inst 1992;84:5759. Moinpour CM. Feigl P. Metch B. Hayden KA. Meyskens FL Jr. Crowley J. Quality of life end points in cancer clinical trials: review and recommen- dations. J Natl Cancer Inst 1989;81:485—95. Cella DF. Tulsky DS. Quality of life in cancer: definition, purpose. and method of measurement. Cancer Invest 1993:112327—36. Donovan K. Sanson-Fisher RW. Redman S. Measuring quality of life in cancer patients. J Clin Oncol 1989;72959-68. Sugarbaker PH. Barofsky I, Rosenberg SA. Gianola FJ. Quality of life as- sessment of patients in extremity sarcoma clinical trials. Surgery [982:9]: 17-23. Ruckdeschel JC. Piantodosi S. the Lung Cancer Study Group. Quality of life assessment in lung surgery for bronchogenic carcinoma. J Thor Surg l99l;6:201-5. Ganz PA. Lee J]. Siau J. Quality of life assessment. An independent prog— nostic variable for survival in lung cancer. Cancer 1991;67:3131-5. Coates A. Gebski V. Signorini D. Murray P. McNeil D. Byme M. et al. Prognostic value of quality—of—life scores during chemotherapy for ad- vanced breast cancer. Australian New Zealand Breast Cancer Trials Group [sec comment citation in Medline]. J Clin Oncol 1992;10:1833-8. Spitzer WO. Dobson A]. Hall J. Chesterman E. Levi J. Shepherd R. et al. Measuring the quality of life of cancer patients: a concise QL—index for use by physicians. J Chronic Dis 1981;34:585—97. Weeks J. Quality-of—life assessment: perfonnance status upstaged? [edi— torial: comment] [see comment citations in Medline]. J Clin Oncol 1992;10:1827—9. Johnson JR. Temple R. Food and Drug Administration requirements for approval of new anticancer drugs. Cancer Treat Rep 1985:69:1 155—9. de Haes JC. Quality of life: conceptual and theoretical considerations. In: Watson M. Greer S. Thomas C. editors. Psychosocial oncology. Oxford: Oxford Univ Press. 1988:61—70. Gill TM. Feinstein AR. A critical appraisal of the quality of quality-of—life measurements [see comment citations in Medline]. JAMA 1994;272:619-26. Gotay CC. Quality of life in cancer patients in Hawaii. Ongoing study funded by the National Cancer Institute. 1995. Wilson IB. Cleary PD. Linking clinical variables with health-related quality of life. A conceptual model of patient outcomes. JAMA 1995: 273:59-65. Ferrans CE. Powers MJ. Quality of life index: development and psychometric properties. ANS Adv Nurs Sci 1985;8:15—24. Ganz PA. Schag CA. Lee JJ. Sim MS. The CARES: a generic measure of health—related quality of life for patients with cancer. Qual Life Res 1992;1219—29. Wright BA. Physical disability: a psychological approach. New York: Har» per & Row. 1960. Coates A. Gebski V. Bishop JF. Jeal PN. Woods RL. Snyder R. et al. Im— proving the quality of life during chemotherapy for advanced breast can- (26) (27 (28) (29) (j()) (31 (3'2) (33) (34) (35) (36) (37) (38 (39) (40) (41) (42) (43) (44) cer. A comparison of intermittent and continuous treatment strategies. N Engl J Med 1987;317:1490-5. Calman KC. Quality of life in cancer patientsvan hypothesis. .1 Med Ethics 1984;10:124-7. Cella DF. Cherin EA. Quality of life during and after cancer treatment. Compr Ther 1988;14:69—75. Watson D. Clark LA. Negative affectivity: the disposition to experience aversive emotional states. Psychol Bull 1984; 96:465-90. Andersen BL. Psychological interventions for cancer patients to enhance quality of life. J Consult Clin Psychol 1992;60:552—68. Trill MD. Holland JC. Cross-cultural differences in the care of patients with cancer. Gen Hosp Psychiatry 1993;15:21-30. Kleinman A. Culture. the quality of life and cancer pain: anthropological and cross< durationchemu + ”TWiST periop >< durationTWiST + ummp >< durationcomp + um. >< duration”... The data will also be presented graphically for each treatment arm as shown in the hypothetical plot in Fig. 2. Pilot data suggest that laparoscopic-assisted colectomy may be more expensive than open colectomy despite shorter hospital lengths of stay because of increased operative times and costs (29). At best. laparoscopic-assisted colectomy may be expected to produce equivalent survival and better quality of life. The cost—utility analysis is designed to permit a determination of whether the magnitude of any observed quality-of—life benefit is sufficient to justify the additional cost of the minimally invasive approach. Much additional methodologic work is needed to identify op- timal approaches to measuring utilities in the clinical trial set- ting. to refine techniques for calculating quality-adjusted survival from observed survival data. and to establish standards for what constitutes reasonable cost—utility ratios. It is critical that this work proceed as quickly as possible. The relevant data must be available to legislators and regulators if quality-of—life considerations are to receive the recognition they deserve as these individuals make tough choices about how to spend our shrinking health care dollar. References (I) Detsky AS. Naglie [0. A clinician's guide to cost-effectiveness analysis [.m' comment citation in Medlincl. Ann Intern Med l99();l 13:147-54. (2) Eisenberg JM. Clinical economics. JAMA 1989;262:2879—86. (3) Karnofsky DA. Abelman WH. Craver LF. Burchenal JH. The use of nitrogen mustards in palliative treatment of carcinoma. Cancer 1948: l :634- 56. DFS Percent LATE COIVPLDATIONS Months from FBndomization Fig. 2. Hypothetical plot of quality-adjusted survival determined using the Q— TWiST (quality-adjusted time without symptoms of disease and toxicity of treat- ment) methodology for one arm of the randomized trial of laparoscopic-assisted colectomy versus open colectomy. PERIOP = perioperative. CHEMO = chemo— therapy. REL : relapse. OS = overall survival. and DFS = disease—free survival. Journal of the National Cancer Institute Monographs No. 20. 1996 (7 (8 (9 (/0) (ll) (/3) (/4) (l5) (/6) (/7) von Neumann J. Morgenstem 0. Theory of games and economic behavior. New York: Wiley. 1953. Ganz PA. Schag CA. Lee JJ. Sim MS. The CARES: a generic measure of health-related quality of life for patients with cancer. Qual Life Res 1992;1:19—29. Schipper H. Clinch J. McMurray A. Levitt M. Measuring the quality of life of cancer patients: The Functional Living Index-Cancer: development and validation. J Clin Oncol 1984:2z472-83. Spitzer WO. Dobson A]. Hall J. Chesterman E. Levi J. Shepherd R. et al. Measuring the quality of life of cancer patients: a concise QL-index for use by physicians. J Chronic Dis 1981;34:585-97. Aaronson NK. Bullinger M. Ahmedzai S. A modular approach to quality- of—life assessment in cancer clinical trials. Recent Results Cancer Res 1988111 1:231-49. Stewart AL. Hays RD. Ware JE Jr. The MOS short-form general health survey. Reliability and validity in a patient population. Med Care 1988;26:724-35. Priestman TJ. Baum M. Evaluation of quality of life in patients receiving treatment for advanced breast cancer. Lancet 1976:12899-900. Cella DF. Tulsky DS. Gray G. Sarafian B. Linn E. Bonomi A. et al. The Functional Assessment of Cancer Therapy Scale: development and valida- tion of the general measure. J Clin Oncol 1993;11:570-9. Andrykowski MA. Altmaier EM. Barnett RL. Otis ML. Gingrich R. Henslee-Downey PI. The quality of life in adult survivors of allogeneic bone marrow transplantation. Correlates and comparison with matched renal transplant recipients. Transplantation 1990; 50339406. Levy SM. Herberman RB. Lee JK. Lippman ME. d'Angelo T. Breast con- servation versus mastectomy: distress sequelae as function of choice [50? comment citation in Medline]. J Clin Oncol 1989:71367-75. Weinstein MC. Stason WB. Foundations of cost-effectiveness for health and medical practices. N Engl J Med 1977;296:716-21. Tsevat J. Weeks JC. Guadagnoli E. Tosteson AN. Mangione CM. Pliskin JS. et al. Using health-related quality~ofvlife infonnation: clinical encounters. clinical trials. and health policy. J Gen Intern Med 1994:9:576-82. Torrance GW. Measurement of health state utilities for economic ap« praisal. A review. J Health Econ 1985:521-30. Sumner W. Nease R. Littenberg B. U—titer: a utility assessment tool. In: Proceedings of Symposium on Computer Applications in Medical Care. Washington. DC. 1992:701-5. Journal of the National Cancer Institute Monographs No. 20. 1996 (/8) (/9) (20) (21 V (22 (23) (24 Morss SE. Lenen LA. Faustman WO. The side effects of antipsychotic drugs and patients' quality of life: patient education and preference assess— ment with computers and multimedia. In: Proceedings of Symposium on Computer Applications in Medical Care. Washington. DC, 1994: 17—21. Torrance GW. Social preferences for health states: an empirical evaluation ofthree measurement techniques. Socioecon Planning Sci 1976;10:129—36. O'Leary JF. Fairclough DL. Jankowski MK. Weeks JC. Comparison of time-tradeoff utilities and rating scale values in cancer patients and their relatives: evidence for a possible plateau relationship. J Med Decis Making 1995;15:132-7. Kaplan RM. Anderson JP. A general health policy model: update and ap- plications. Health Serv Res 1988;23:203-35. Torrance GW. Zhang Y. Feeny D. Furlong W. Barr R. Multi-attribute preference functions for a comprehensive health status classification sys— tem. Paper 92—18. Hamilton. ON. Canada: Centre for Health Economics and Policy Analysis, McMaster University. 1992. Euroqol Group. EuroQol—a new facility for the measurement of health-re- lated quality of life. Health Policy 1990: 16: 199-208. Weeks J. O‘Leary J. Fairclough D. Paltiel D. Weinstein M. The “Q—tility Index": a new tool for assessing health—related quality of life and utilities in clinical trials and clinical practice. Proc ASCO 1994; l 3:436. Nelson H. Weeks JC. Wieand HS. Proposed phase III trial comparing laparoscopic-assisted colectomy versus open colectomy for colon cancer. Monogr Natl Cancer Inst 1995; 19151—6. McCorkle R. Quint-Benoliel J. Symptom distress. current concems and mood dis- turbance after diagnosis of life—threatening disease. Soc Sci Med 1983;17:431—8. Gelber RD. Goldhirsch A. Cavalli F. Quality-of—life-adjusted evaluation of ad— juvant therapies for operable breast cancer. The International Breast Cancer Study Group [my comment citation in Medline]. Ann Intern Med 1991;] 14:621—8. Glasziou PP. Simes RJ. Gelber RD. Quality adjusted survival analysis. Stat Med 199019125976. Senagore AJ, Luchtefeld MA. MacKeigan JM. Mazier WP. Open colec— tomy versus laparoscopic colectomy: are there differences? Am Surg 1993; 59:549-53; discussion 553—4. Note Supported in part by the American Society ofCIinical Oncology. 27 #__— Iri’_ _ _ Measuring Quality of Life in Culturally Diverse Populations Richard B. Warnecke, C aral Esrwin g F errans, Timothy P. Johnson, Gloria C hapa-Resena’ez, Diane P. O’Rourke, Nae! C ha’vez, Susan Dua’as, Eva D. Smith, Lucy Martinez Schallmaser. Roger P. Hand, Thomas Lad* If the US. National Cancer Institute is to meet its goals for reducing cancer incidence and mortality. it has become increas- ingly evident that there will have to be a major focus on minority populations. Cancer statistics consistently indicate that. compared with the cancer incidence and survival in the general population. the incidence is higher and the survival lower in minority groups. In recognition ofthese statistics, it is now man- dated that all federally sponsored research explicitly include females and minorities or provide a specific explanation for why they have been excluded. One result of these concerns has been the need to develop valid assessment instruments. appropriate for ethnic minority populations, for use in clinical trials and other research. Despite earlier assertions to the contrary (1.2). researchers in- creasingly recognize the pitfalls of uncritically applying stand- ard measures in studies of minority populations (3-14). Culturally mediated differences in cognition and interpretation are now regarded as responsible for many of the systematic dif- ferences that have been observed in cross—cultural surveys (7.13). In a series of studies specifically relevant to quality of life. for example. it has been well documented that there are sig— nificant cross-cultural differences in how quality of life is as- sessed (15-19). in the perception and reporting of pain (20-23). and in illness behavior (24,25). Moreover, other ongoing re- search. funded by the National Center for Health Statistics. has recently shown the effects of culture and educational attainment on a whole range of standard health questions taken from the Health Interview Survey and other major federal health surveys (26). The present article describes the initial phase of a two—phase project designed to assess the applicability of the Ferrans and Powers Quality of Life Index (QLI) (27) among cancer patients with a high school education or less who were selected from two minority populations. In the research described here. cognitive methods were used to 1) identify the meaningfulness of specific items in four content domains of the QLI for both male and female adult African-American and Mexican—American cancer patients with low educational levels and 2) determine the capability of cancer patients to form judgments about their satis— faction with and the importance they attribute to life aspects as— sociated with quality of life. Based on these assessments. this article describes a process by which the QLI was modified to be more appropriate for these patients. In a second phase of this Journal of the National Cancer Institute Monographs No. 20. 1996 project. the resulting instrument is being tested in clinical trials with a larger sample of patients with the same ethnic and educa- tional attributes. Culture and Assessment of Quality of Life Virtually every cooperative group now incorporates quality- of—life end points into their clinical trial research protocols. This approach represents a significant departure from the traditional approach in which the end points have been tumor response and duration of survival, while the patient‘s well-being was not con- sidered (28.29). The increasing pressure to ensure that ethnic minorities are included in clinical trials and the growing interest in adding quality of life as an outcome require more explicit attention to how cultural, ethnic. religious, and other values influence judg- ments about quality of life (30-33). In addition. there is growing agreement that the patient‘s perceptions provide the most impor— tant indicator of quality of life (24-26). The Ferrans and Powers QLI (2734-36) was designed to take into account individual values in measuring quality of life. Quality of life in the QLI is defined as “a person‘s sense of well—being that stems from satis— faction or dissatisfaction with areas of life that are important to him/her" (27). Judgments about satisfaction were selected be- cause satisfaction implies a cognitive judgment of experiences based on comparisons of desired and actual conditions of life (28.30.37). Conceptual Basis of the Ferrans and Powers QLI Conceptually. for the Ferrans and Powers QLI. quality of life is multidimensional. composed of the following four domains: *Afliliuriuns ofuurhors: R. B. Wantecke. T. P. Johnson. G. Chapa-Resendez. D. P. O'Rourke (Survey Research Laboratory. College of Urban Planning and Public Affairs). C. E. Ferrans. S. Dudas. E. D. Smith (College of Nursing). N. Chavez (Community Health Sciences. School of Public Health). L. M. Schallmoser. T. Lad (Loyola University of Chicago). R. P. Hand (College of Medicine). University of Illinois at Chicago. (YWI'es/mmlcm‘v to: Richard B. Wamecke. Ph.D.. Survey Research Laboratory. College of Urban Planning and Public Affairs. University of Illinois at Chicago, PO. Box 6905 M/C 336. Chicago. IL 60680. See “Notes" section following “References." 1) health and functioning, 2) social and economic, 3) psychological/spiritual, and 4) family (36). Thirty-three specific life aspects comprise these four domains (Table l). The QLI was derived from an extensive literature review, measurement based on patient interviews, and then factor analysis. Extensive psychometric assessment of the QLI indicates strong validity and reliability across both patients and diseases. Category Fallacy and Quality-of-Life Measurement Despite the strength of psychometric properties, the patient populations used to develop the QLI were primarily well-edu— cated middle and upper middle class individuals. While these populations did include African-American and Hispanic patients, these patients tended to be fairly well educated and. in the case of Hispanics, acculturated to the point of being bilin— gual. Although a Spanish-language version of the QLI was evaluated with at least one group of patients, the kind of testing done here was not done in this or other earlier versions (38). Measures developed by and for middle and upper middle class respondents are increasingly believed to misrepresent the thoughts, feelings, and behaviors of individuals from segments of the population where poverty is common, where reading ability is limited, and, in the case of Latinos, where ability to speak English is limited or nonexistent (39—41). This mis- representation due to variation in cultural understanding of the question has been described as the “category fallacy.” It is a problem with much of the health-related survey data collected in the United States among these populations (26). The category fallacy results from the failure to distinguish be— tween etit' concepts that are truly universal and accepted across multiple cultural groups and emit' concepts that have meaning only within a specific cultural group or socioeconomic context. When an emit- construct is used as if it were wit, the resulting construct is described as pseudoetic", and the measure results in a category fallacy (42). Whether a concept is em or emic' is believed to be related to its level of abstraction (43,44). Triandis and Marin (45) proposed a strategy that emphasizes distinguishing between culturally specific measures (emit) and those that are universally relevant (etic‘). They called for using probes designed to assess and understand when unique aspects of the culture influence interpretation and response to questions and when the concept underlying the question is culturally transcendent (45). The “emit" + etic" methodology avoids the pitfalls of the pseudoetic approaches that adversely affect ques- tionnaire design. This is the approach used in this study. Methods The Cognitive Strategy Cognitive research on the validity of responses to survey questions has iden— tified four steps in the response process (46-48). Although not every step is fol— lowed by every respondent when answering every question. the four steps are I) question interpretation, 2) information retrieval, 3) judgment formation, and 4) response editing. Question interpretation. Culturally influenced language and interpretation differences are likely to influence how respondents understand questions dealing with quality of life (49.50). In some instances. the language into which a ques- tion is translated does not contain the concept. In other instances, cultural media— tion or cultural experience influences the meaning or validity of the question. In either case, when the respondent replies, the reply is not to the same question that has been asked: hence, it is not valid (51.52). Cognitive assessment seeks to understand what the question means to the respondent and, if the meaning is dif- ferent from that intended by the questioner. to guide the choice of more cultural- ly or educationally appropriate wording. Information retrieval. By this process, either an answer or information relevant for constructing an answer is retrieved from memory. One major area of inquiry has been how question wording can be modified to provide cues to facilitate recall (53). Individuals differ in how they access their memories for such information (54). Recall of infomiation about regular. recurring events is likely to be “semantic." involving schemas in which information about classes of events is retained in memory rather than information about specific events. In contrast. events that occur sporadically or very occasionally are subject to “episodic“ recall where recalled information is focused on specific episodes or events (54-56). It has been observed that semantic schema are likely to be cul- turally influenced by the individual‘s community or larger culture. Accuracy of recall may also be culturally related (57-59). Judgment formation. Based on information retrieved from memory. judg- ment formation is an important aspect of attitude formation. It is the process by which the importance of events and satisfaction with one's current life status are translated into an assessment of quality of life (27.28.30.34-37). Most often when a respondent is asked about the value attributed to a life aspect such as an event. experience. or action. the response is a synthesis of information retrieved from memory about relevant experience. The more frequently such information Table 1. Specific aspects of quality—of—life domains Social and economic domain Flealth and functioning domain Specific aspects by domain Family domain Psyichologictil/ ' spiritual domain Usefulness to others Physical independence Standard of living Financial independence Responsibilities Home Own health Job/unemployment Stress and worry Neighborhood Leisure activities Friends Retirement Travel Live a long life Sex life Health care Pain Energy (fatigue) Emotional support Education Life satisfaction Happiness Self Goal achievement Peace of mind Personal appearance Faith in God Control over life Family happiness Children Spouse Family health 30 Journal ofthe National Cancer Institute Monographs No. 20, 1996‘ is used. the more accessible it is in memory: hence. the more readily it is used for developing judgments (60.6] ). Perhaps of most relevance to quality-of—life research are the findings suggest- ing that. when individuals search their memory for specific information. they use contextual cues such as references to specific persons. events. or locations. These cues lead the respondent to access particular sets of memories that are likely to mediate the response ((72.63). These retrieval cues can only help access memories that have been previously encoded (64). Because of the overwhelming amount of information that is available. not all of it may be important enough to be encoded in memory (51). Questions about satisfaction and importance assume that the individual has stored in memory relevant experiences that will be available for forming judg— ments about the importance of and current satisfaction with the life aspects that are the point of each question. If there are no memories of specific and relevant events to cue the patient‘s responses, however. the actual responses to questions about how much the patient is satisfied with a particular life aspect or how im- portant it is may be based on motivation to be a "good respondent." Hence. the response may be subject to editing rather than reflecting the respondent's true assessment. (See the following section on response editing.) Cultural condition» ing may also directly influencejudgment formation. For example. some research on responses to health opinion surveys suggests that Hispanics in the United States are more fatalistic than Anglo respondents regarding cancer (65) because of the culturally specific concept offatalismn. or the belief that little can be done to alter one's fate. Other research suggests that cultural variation in the likelihood of probabilistic thinking (i.e.. the ability to express thoughts in terms of uncertainty) may also influence how judgments are formed (66.67). Finally. the validity of scales requires a common frame of reference for map— ping judgments onto a common metric. For example. African-American and Hispanic survey respondents are less likely than Anglo-Americans to qualify their answers on rating scales. whereas Asians are less likely to prefer extreme responses (ox-71). Preference for extreme versus cautious response styles has been interpreted as being a consequence of cultural variation in emphasis on sin- cerity versus modesty in social interaction (7I.72). Use of modifiers tends to in- crease among Hispanics with acculturation (70). Response editing. This editing is a commonly encountered phenomenon when survey respondents feel that certain answers are more socially desirable than others (73). For example. socially desirable behaviors such as exercise and nutrition are frequently overreported. whereas such undesirable behaviors as drinking or smoking are frequently underreported. Available information sug- gests that definitions of socially desirable behavior vary culturally (5/.74-77). Being Mexican and being a member of a minority group have been correlated with giving socially desirable responses (78,79). Socially desirable response pat- terns are compatible with the commonly observed pattem of social interaction itt Hispanic cultures referred to as .timpa'liu. or the expectation that interpersonal relations will be guided by harmony and the absence of confrontation (80). Such cultural expectations also seem to influence Asian survey respondents (XI). A related phenomenon is respondent acquiescence. or the tendency to agree with a statement regardless of its content (82). Acquiescence is observed as a strategy of self-presentation most commonly. although not universally (23.24). among low-status Hispanics and African-Americans (3,68.7().72.79.82). Altema- tively. it may be that acquiescence occurs because of too much emit- question content, leading respondents who are unsure about what is being asked to "play it safe" and acquiesce rather than to look foolish or admit they do not understand the question (26). Editing also occurs in situations where there is social or cultural distance be- tween the interviewer and respondent because of ethnicity. gender. educational level. or other status indicators (83-90). There is also evidence that bilingual respondents may answer differently, depending on the language used by the questioner and the cultural significance of the question (76,96). which may af— fect the tendency for acquiescence. social desirability. cultural understanding. or cross-cultural accommodation (76,97). It may also be that language variation produces differences in response cues that affect recall and/orjudgment (98). Cognitive Methods The Ferrans and Powers QLl was evaluated by use of cognitive probes designed to explore whether African-American and Mexican»American cancer patients varied in the way in which they understood the various components of the QLI. how they retrieved information and formed judgments regarding the importance of and their satisfaction with each life aspect. and whether they Journal ofthe National Cancer Institute Monographs No. 20. W96 edited their responses. Individual respondents are selected because they have educational or cultural characteristics that might affect how they may interpret the questionnaire content. Thus. they are recruited and interviewed. The cognitive interview has been developed over the last decade in research on questionnaire design by teams of survey methodologists and cognitive psychologists (99-102). During the cognitive interview. the respondent is asked the question to be evaluated. Once he or she answers the question. then standard probes or follow-up questions are used to explore understanding. retrieval. judg- ment t‘omtation. and editing effects in the answer. These probes help the respon- dent reconstruct or “think aloud" about the thought processes used to respond to the question. The interaction between the cognitive interviewing process and the question- naire is iterative. As more is learned about how these processes affect response through the cognitive process. the question content is revised. Further interviews are conducted until there is apparent consensus among respondents regarding the meaning of individual questions. which is the final objective of the process. When the final revisions are made. the questionnaire is finalized and pretested. Thus. we kept using a question in the interviews until the responses became redundant. When respondents had interpretive problems. questions were revised (sometimes several times) and then retested until no further linguistic or inter- pretive problems were identified. Before we began the cognitive interviews. the questionnaire was reviewed by a reading literacy laboratory in the College of Education at the University of 11- linois. The purpose of this process was to revise or eliminate any questions where overall reading level might interfere with the cognitive processes being evaluated through the “think-aloud" probes. Patient Selection The purpose of this study was to assess the validity of the Ferrans and Powers QLI among African-American and Mexican—American patients with low education and. in the case of the Mexican—American subjects. poorly acculturated. African- American patients selected for cognitive interviews were recruited from outpatient clinics affiliated with the University of Illinois and Mount Sinai hospitals in Chicago. They were eligible if they had a high school education or less. and the range in education among subjects varied from third grade to high school diploma or equivalent. Interviews were conducted between February and September 1994 with 23 African-American patients (nine females and 14 males). Fifteen Mexican-American patients (11 females and four males). selected ac— cording to the same criteria as used for the African—American subjects. were inter— viewed in October 1994 at The University of Texas M. D. Anderson Cancer Center in Houston. TX. These interviews were conducted in Spanish and, with patients from whom Spanish was the primary language. by bilingual interviewers. Patients were selected with the knowledge and consent of their attending physicians. While awaiting chemotherapy. patients were recruited by nursing staff in the clinics. All respondents were informed that their participation in the interview was voluntary. Respondents were given an honorarium for participating. Results The analysis focused on the content of each specific element of life quality in the four domains of the Ferrans and Powers QLl (Table l) and on the overall scaling used to obtain the respondents’ ratings of the importance of and satisfaction with each element. We will first consider the domain content and then the scales. Most of the problematic questions related to education and reading level. In the results reported below, the questions were initially evaluated and altered during interviews with the African-American patients who were interviewed first. This pat- tern was deliberate because of the costs associated with transla- tion into Spanish. Thus. we attempted to resolve the issues related to literacy and the interaction between cognitive respon- ses and education before we translated the questionnaire. When we evaluated the questions using cognitive probes with Mexican—American patients at The University of Texas M. D. 31 Anderson Cancer Center. we discovered additional problems due to linguistic issues. For the most part. however. the educa- tional level needed by the Mexican-American patients to under— stand and form judgments about the questions was the same as that required by the African-American patients. Question word- ings that were changed as a result of the interviews conducted in Spanish were retested in English on African—American patients. In the summary of results below. questions that were problematic for Hispanic patients are identified in the text. Domain 1: Health and Functioning Seven of the [3 items related to health and functioning re— quired some revision based on the readability assessment and cognitive interviewing process. Interpreting the question was the problem most commonly en- countered by the respondents. Three questions were revised based on the literacy evaluation. These questions were as fOIIOWs: l ) Original: How satisfied are you with your usefulness to others." Revised: How satisfied are you with how useful you are to others." 2) Original: How satisfied are you with your leisure time aetiyities." Revised: llow satisfied are you with the things you dofor fun." 3) Original: How satisfied are you with your potential for a happy old age/retirement." Revised: How satisfied are you with your eham'es‘f'or a lulppyfitture? Problems of understanding the underlying concept emerged from the probing process in a fourth question: 4) Original: How satisfied are you with your physical indepeiu/enee." Probe: What do the words “physical independence" mean to you." The responses to the probe indicated that the term “physical independence" was being interpreted as financial independence or as not being reliable for others who depended on the respon— dent. In one interview. the respondent simply said he or she did not know what the term meant. Revised: How satisfied are you with your ability to take eare of'yourself'without help." With that revision. subsequent respondents quickly achieved consensus. Respondents were clearly able to describe the “ability to do things without help." The revised form of the question was incorporated into the QLI. 5) Original: How satisfied are you with the amount of'stress or worries in your life." Probes: Can you tell me in your own words what this question is asking about." What does / the word] “stress" mean to you ." What does / the word] ”worries” mean to you." Did you answer this question in terms of'stress. worries. or both." WoulaI it have been easier to answer. harder to answer. or about the same if'we did not inelude both worry and stress in the same question." To the African-American respondents. there was considerable overlap in the meaning of the terms “worries” and "stress." but the results from the probes for this question indicated that using “worries" produced greater validity than when the term “stress" was used. Our decision regarding validity was based on the number of respondents who indicated that they understood what the question was asking during the probes and who based their responses on that understanding. Moreover. during the Spanish language interviews. it became clear that. linguistically. there is no term in Mexican Spanish for “stress" and using synonyms for stress changed the translation of the question. Revised: How satisfied are you with the amount of" worries in your life." Information retrieval problems occurred with one question from this domain. 6) Original: How satisfied are you with your ability to travel on vacations." What determines your ability to travel." Do you take vacations." lfyou wanted to take a vacation and had the money to do so, would your health be good enough to do so." What do you think we mean by vacation." Probes: The concepts of vacation and travel for pleasure had no equivalent meanings that would allow retranslation or refor— mulation of the question that asked about these things. Neither the concept of “vacation" nor the concept of “travel for pleasure” had meaning for the African—American and Mexican— American respondents because neither had relevance to their lifestyle or experience. For example. if they traveled. it was to visit family and only for a family emergency. The question was dropped when it became clear that there was no way the concept could be written that would cue relevant memories on which to base a response. The relevant elements of the concept were covered by the question discussed above: “How satisfied are you with the things you do for fun?" Judgment formation issues also arose in one question where the probing indicated variation in the anchoring point associated with the age ofthe respondent. 7) Original: How satisfied are you with your potential to live a long time." What do you think we mean by “your potential to live a long time" ." Probes: What do you consider “a long time" to be" The problem of establishing a common scale for forming judg— ments arose because the response to the original question depended on the age of the respondent. In point of fact. this issue is probably relevant to all who use this scale, regardless of education or accul- turation. In response to the probes. older respondents replied that they already had lived a long time: younger respondents responded Journal ofthe National Cancer Institute Monographs No. 20. 1996 in terms of the future. We revised the wording to cue respon— dents to respond in terms of whatever expectations about con— tinued longevity they might have. Revised: How satisfied are you with your ('hanee of'liying lo the age you would like." Domain 2: Social and Economic Four of the 10 questions in this domain were revised follow— ing probing. Question interpretation was a problem with two items in this domain. One item was revised following assessment for reading level as follows: I ) Original: How satisfied are you with _\‘o1njfinam'ial independence." How satisfied are you with how well you (on take rare of'yourfinam‘ial needs." Revised: For the second revised question. the problem was clearly in— terpretation. and the question created problems for both the African-American and Mexican—American patients. 2) Original: Ilow satisfied are you wit/1 your standard of living." What do you think we mean by "standard of living" ." What kinds of'things do you think about in answering this question about your standard of' living." Probes: In response to the probes. several African—American respon— dents interpreted the question as addressing a moral issue. “standards of living.“ Moreover. there was no straightforward translation into Spanish of the concept standard of living. The item was dropped because the elements that it was intended to address were adequately covered by other questions dealing with financial needs and satisfaction with home and neighbor- hood. Inforrnation retrieval was combined with question interpreta— tion in the two remaining problematic questions from this domain. 3) Original: How satisfied are you witlt the amount of" emotional support you getfrom others." What do you think we mean by ”emotional support" ." Probe: In response to the probes. especially the query about "others." it was clear that. as written. the question did not offer specific cues about whose support was relevant: some thought “others" referred to friends. and some thought the term referred to fami— ly. Two questions were ultimately used: one about family and one about friends. This procedure produced consensus to the probes. Thus. the final question wording used was as follows: Revised: How satisfied are you with the emotional support you getfrom your family ." How satisfied are you with the emotional support you getfi'om people other than your family." Journal ofthe National Cancer Institute Monographs No. 20. 1996 4) Original: Probes: How satisfied are you with your home." In your own words. what do you think this question is asking about your home." Why are you /satisfied/dissatisfied] wit/1 your home." What things about your home did you think about when answering this question." The last probe. requiring information retrieval. indicated problems with this item. The question was designed to address the physical aspects of the home environment. As the respon- dents "thought aloud" about the things about their homes that influenced their judgments regarding importance and satisfac— tion. they described the ambience. especially interactions with children in the home and the neighborhood. Finally. at least one respondent did not interpret the question as applicable to apart- ment dwellers. Two strategies improved consensus around this question. First. the question was relocated in the QLI to follow other questions that asked specifically about satisfaction with children and the neighborhood. By placing these items before the home question. the respondents were cued that we wanted them to think about other aspects of their home besides children and the neighborhood. Second. we rewrote the question to refer to several types of dwelling. Revised: How satisfied are you with your home. apartment. or plaee where you lire." Domain 3: Psychological/Spiritual There were no problems with this domain among the African— American patients. An item. regarding “personal belief in God.” proved quite wordy in the Spanish version. When it was retranslated without the word “personal." it worked well in both English and Spanish. 1 ) Original: How satisfied are you with your persona/faith in God." How satisfied are you with you/"faith in God." (Sufe en Dios") Revised: Domain 4: Family Question interpretation caused problems with one question in this domain. It was first evident in the Spanish translation. ()n review. however. it was also a problem with the English version. 1 ) Original: How satisfied are you with your relationship wit/1 your spouse/signifit'ant other." What parts of'the relationship did you think about when you answered this question." Probes: Has your relationship (hanged in any way sinee you got eaneer." In response to the probes. both the African—American and Mexican-American respondents indicated that they thought about the sexual aspects of the relationship. The Spanish transla- tion cued Mexican-American respondents to think about the sexual aspects of the relationship because the term “relation- ship" in Spanish is translated as "relaciones." which specifically means sexual intercourse. As the QLI was originally designed. 33 the questions about satisfaction and importance of sex life fol— lowed the questions about satisfaction with one‘s spouse and the importance of that relationship. The ordering of these two ques- tions was changed so that the sexual satisfaction question preceded the spousal satisfaction question. This order cued respondents that the question on spousal satisfaction did not refer to the sexual relationship. There was some confusion in both languages about who was a “significant other." so the wording was changed. Reworded: How satisfied are you with your spouse. lover. or partner." Measurement Scales Respondents were asked to form judgments about their satis— faction with each of the 33 life aspect items (see Table 1) and then to weight the importance of each item. Both of these scal- ing tasks required the respondents to form judgments and format their responses using bipolar scales ranging from “very satis- fied" to “very dissatisfied“ and from “very important" to “very unimportant." respectively. Early in the cognitive interview process. it became apparent that the African-American respondents experienced difficulty with the importance and satisfaction scales and the task of recording their judgments using the labeled. six—item. bipolar scales. The labels were presumed to form an equal interval scale that discriminated between levels of satisfaction and impor- tance. Upon further examination. the intervals were not per— ceived as discrete but rather as overlapping. The satisfaction scale asked: “How satisfied are you with . . . [each life aspect)?" It then asked the respondent to describe their satisfaction by selecting one of the following terms: very satis- fied. moderately satisfied. slightly satisfied. slightly dissatisfied. moderately dissatisfied. and very dissatisfied. The importance scale asked: “How important is . . . [each life aspectl?“ Respondents were again asked to select from a series of phrases: very important. moderately important. slightly impor— tant. slightly unimportant. moderately unimportant. and very unimportant. The respondents did not understand the bipolar scaling that they were being asked to perform. To assess the nature of the problem. a thermometer scaled from “0" to "100” was presented to 23 additional African-American and Spanish-speaking respondents who were selected on the basis of educational level but who were not cancer patients. The respondents were then asked to locate a variety of descriptors of “satisfaction" and “importance" on the thermometer (Fig. l). Inasmuch as the process and results were the same for both scales. we will report the results of the “importance" scale here. On the thermometer presented to respondents. “0" was labeled “as unimportant as something could ever be.“ “100" was labeled “as important as something could ever be." and “50" was labeled “neither important nor unimportant." Respondents were given a series of terms that they were asked to place on the thermometer between 0 and 100. The following temts reflected importance: “very important." “totally important." “important. somewhat im- portant." “moderately important.“ “a little important." “fairly im- u as 34 IMPORTANCE 'l‘ll El{ M()M IC'I‘ICR 100 AS IMPORTANT AS SOMETHING COULD l-ZVliR nr-z 90 80 70 6O 50 NEITHER IMPORTANT NOR UNIMPORTAN’I~ 40 30 20 10 0 AS UNIMPOR'I‘AN'I~ AS SOMETHING (‘OULD [EVER Bli Fig. l. Thermometer used by respondents to describe “satisfaction" or “impor- tance." u H portant." “slightly important. neither important nor unimpor— tant." “not very important. somewhat unimportant." “fairly unimportant." “moderately unimportant." “slightly unimpor— tant." totally unimportant. unimpor— n u u .. u .. a little unimportant. tant.” “not at all important." and “very unimportant.“ Table 2 presents the range of values assigned to the key terms actually used in the scale. Two things are evident from this table. First. based on the range of values. there was considerable overlap in the numeric rating of each descriptor. That is. the range for “a little important” (90-05) overlapped with the range Table 2. Range of responses to scale of importance or unimportance Scale Range of responses Scale of importance Very important IOU-9t) Import ant IOU-80 Somewhat important 100-45 A little important 90-05 Scale of unimportance Very unimportant 0-70 Unimportanl 0-50 Somewhat unimportant [0—95 A little unimportant 0-60 Journal ofthe National Cancer Institute Monographs No. 20. 1996 assigned to “important" (100—80). "A little important" (90-05) almost completely encompassed “somewhat important" (100- 45). The same pattern could be observed for the various charac- terizations of "unimportant." The range of values assigned to “very unimportant" (0-70) totally encompassed the ranges for “unimportant" (0—50) and for “a little unimportant" (0-60). which in turn had a wider range of values than “unimportant." The second point is that the respondents did not see “very im- portant“ and “very unimportant“ as polar opposites on a single scale from 0 to 100. On the contrary. the respondents treated the range of importance and the range of unimportance as two separate and overlapping scales. For example. the range of values assigned to "very important" was 100-90. However. the range assigned to "very unimportant" (0-70) overlapped “a little important" (90-05). “Somewhat unimportant" (10—95) over— lapped “very important" (100-90). “important“ (100—80). “some- what important" (100-45). and "a little important" (90-05). If the point of this scale is to discriminate. the bipolar scale clearly did not achieve that objective with these subjects. An additional matter that arose among the Spanish-speaking respondents was that there are no direct Spanish linguistic equivalents for either “unimportant" or “dissatisfied.“ Hence. the negative pole of each scale as previously translated was not cognitively understood in the way it was worded in English. Bilingual respondents solved this problem by devising a transla- tion from Spanish into English that was consistent with the scale. Thus. although sin im/mrtunt'ia. for example. is directly translated from Spanish as “not important" or “without impor— tance.“ the bilingual respondents understood that to be equivalent to unimportant. Because there was no direct transla- tion for "unimportant" which is the negative pole of the scale. however. the Spanish speakers did not understand the bipolar contrast. Thus. the only solution was to change the English to match the Spanish and to use a single dimensional scale anchored by “very important" and “not at all important” and to use a similar 6—point scale anchored by “very satisfied“ and “not at all satisfied." The graphic used for the self-administered ver— sion of the QLI and as an aid in the interview version is a bar graph indicating an "amount" of satisfaction or importance from 1 to 6 (Fig. 2). When the scale was retested using the graphics. subjects with low education were able to complete the entire in— terview without problems; moreover, they used the entire range of the scale. including the midpoint.‘ Discussion and Conclusions Several important points are evident from this analysis. First. there was clear evidence of problems during three of the four cognitive stages. (No editing problems were encountered.) The greatest problems were with question interpretation. In part. these problems resulted from questions that were written to re- quire a level of verbal comprehension that was too high for the likely respondents. These problems were easily corrected by rewriting the question to require lower levels of verbal comv prehension. Other problems of question interpretation came from language. In some instances. the English concept was not meaningful linguistically in Spanish. Of importance was the fact that. when these questions were asked of bilingual respondents. Journal ofthe National Cancer Institute Monographs No. 20. 1996 QUALITY OF LIFE INDEX IIIII 23456 VERY SATISFIED 3 4 5 6 VERY IMPORTANT 1 NOT AT ALL MORTANI' 1 NOT AT ALL SATISFIED Fig. 2. Bar graph indicating an "amount" of "satisfaction" or "importance" from I to 6. as they had been in earlier evaluations of the QLI (38). the respondents were able to translate them into English. arrive at an answer. and back-translate their responses into Spanish. If the respondent did not speak any English. these questions either were not understood and were left unanswered or. more often. were answered as a way of acquiescing t0 the interviewer. This was the only form of editing that we were able to observe in the way these scales were used. Some concepts. such as standard of living and traveling on vacation. had meaning but no recallable memory content. That is. the respondents could not retrieve information relevant to making an answer because they had never experienced the events in question. These questions had to be dropped. In some cases. the meaning was not well specified in the item as written. By paying close attention to the relationship between questions and to the wording. the meaning could be more clearly specified. This was the case with the question regarding relationship with one‘s spouse. This question was initially inter— preted as a question about satisfaction with the sexual aspects of the relationship until we changed the order of the question to follow the specific question regarding satisfaction with sex life. Once the order was changed. it was clear that the two questions referred to different aspects of the spousal relationship. Finally. there was the response scale. It is clear that bipolar and labeled scales were less successful with these respondents than unipolar scales where only the end points are labeled. Using a graphic also aided response. With it. the scaling task was understandable even to respondents with very low levels of verbal skills. and these respondents were able to use the entire scale including the midpoints appropriately. This result il- lustrates the importance of clarifying ambiguous or vague quan- tifiers such as “moderately satisfied“ for respondents with weak verbal skills. Another important finding was the consistent evidence that respondents will answer questions that are cmic' in order not to reveal their inability to understand a question. That is. they will “satisfiee.” It is important. therefore. at a minimum to translate and back-translate. both of which were done with the original scale. When scales are used with respondents who are not bilin- gual. however. it is also important to test the scales with respon- dents who are not bilingual. Only when the scale was tested with persons who spoke only Spanish did some of the linguistic problems emerge. Overall. the results demonstrate the impor— 35 tance of this kind of assessment in all scales before they are translated to a different language or administered to a population different from the one for whom the instrument was developed. Finally. our experience with both the individual items and the scales indicated the importance of both conceptual equivalence across cultural groups and the way the question is phrased. The failure to consider the conceptual equivalence and wording across language groups will limit the generalizability and validity of the results. It was most efficient to deal with the potential problems in order. starting with reading level. then conceptual understandability. and finally linguistic problems. It is also important to recheck changed questions with the groups where consensus as to meaning has been established. Thus. in the final phase. we interviewed our final group of African— American respondents after we did the interviews in Houston to ensure that the changed items were still valid in the African— American population. It is also important to emphasize that our data represent data on samples of patients who were selected because they repre- sented the ethnic groups on which the QLI was to be tested and because they had relatively low educational levels. Their responses cannot be generalized to any larger population. The selection of these two patient groups was driven by the funding source that specifically requested proposals to examine the generalizabilily ofquality-of—life scales to these populations. The QLI is now being evaluated on larger populations of patients at the University of Illinois Hospitals and Clinics and at The University of Texas M. D. Anderson Cancer Center. It is also being administered at several points during the therapy. so that we will have data on how the scale changes during treat- ment. Following this trial. we intend to conduct psychometric studies on the data to establish norms for these populations. These data will be compared with data obtained from a variety of other populations of patients on whom the scale has already been evaluated. In all probability. the QLI will be revised. References (I) Freetnan DM. A note on interviewing Mexican Atnericans. Soc Sci Q 1969;49:909—114. (2) Welch S. Comer J. Steinman M. Interviewing in a Mexican American cotnrnuttity: an investigation of sortie potential sources of response bias. Public Opin Q 197323701536. _ J Aday LA, Chiu (iY. Anderson R. Methodological issues in health care surveys of the Spanish heritage population. Am J Public Health 1980;70:367-74. Flaherty JA. Appropriate and inappropriate research methodologies for Hispanic mental health. In: Gaviria M. Arana JD. editors. Health and be» havior: research agenda for Hispanics. The Simon Bolivar Research Monogr Series No. 1. Chicago: Univ Illinois at Chicago. 1987:1774“). Marsella AJ. Thoughts on cross—cultural studies on the epidemiology of depression. Cult Med Psychiatry 19782234357. (0) Manfredi C. Lacey L. Wamecke R. Balch G. Allen K. Complementary in- formation from survey data and focus group insights. Paper presented at the Annual Conference of the American Association for Public Opinion Research. Lancaster. PA. 1990. Marin G. Marin BV. Research with Hispanic Populations. Newbury Park (CA): Sage Publications. 1989. Milburn NU. Gary LE. Booth JA. Brown DR. Conducting epidemiologi- cal research in a minority community: methodological considerations. J Commun Psychol 1991;19:342. Montero D. Research among racial and cultural minorities: an overview. J Social Issues 1977:3321—10. (IU) Myers V. Survey methods for minority populations. J Social Issues 1977;33:11—9. l (5 (7 (8 (9 36 (ll) Salber EJ. Bela AG. The health interview survey and minority health. Med Care 1980;28:319—26. Vernon SW. Roberts RE. Lee ES. Ethnic status and participation in lon— gitudinal health surveys. Am J Epidemiol 1984;119:99—1 13, (/3) Word CO. Cross-cultural methods for survey research in black areas. I Black Psychol 1977;}:72-87. (l4) Zmud JP. Arce CH. Language cue management: techniques for improving response rates and data quality in surveys of Hispanics. Paper presented at the Annual Conference of the American Association for Public Opinion Research. Phoenix. AZ. 1991. (I5) Rogler 1H. The meaning ofculturally sensitive resesarch in mental health. Am J Psychiatry 1989;146:2963-3113. (l0) Vaughan DA. Kashner JB. Stork WA. Richards M. A structural model of subjective well-being. Social Indicators Res 1985;16:315—32. (/7) Mukherjee M. Ray A. Rajyalaksme C. Physical Quality of Life Index. So- cial Indicators Res 1979;61:283—973. (/8) Mukherjee R. On the quality of life in India. Social Indicators Res 1981:9:455-76. (I9) Mastekaasa A. Moum T. The perceived quality of life in Norway. Social Indicators Res 1984;14:385-420. (20) Morse JM. Morse RM. Cultural variation in the inference of pain. J Cross— Cultural Psychol 1988;19:232—412. (2/) Zborowski M. Cultural components in responses to pain. J Social Issues 195218: 16—30. (22) Zborowski M. People in pain. San Francisco: Jossey-Bass. 1969. (21) Zola 1K. Culture and symptoms—an analysis of patients‘ presenting corn— plaints. Atn Sociol Rev 1966:31:615»30. (24) Good B. Kleirnnan A. Culture and anxiety. In: Tuma AH. Maser JD. editors. Anxiety and anxiety disorders. Hillsdale (NJ): Lawrence Erlbaum. 1985. (25) Harwood A. editor. Ethnicity and tnedical care. Cambridge (MA): Har- vard Univ Press. 1981. (26) Johnson T. O'Rourke D. Chavel. N. Sudrnan S. Warnecke R. Lacey L. et a1. Social cognition and responses to survey questions among culturally diverse populations. Presented at the Inlemational Conference on Survey Measurement arid Process Quality. Bristol. UK. 1995. (27) Ferrans CE. Development of a quality of life index for patients with catt— cer. Oncol Nurs Forum I99():17(3 Suppl): 15-9. (28) Caires N. List M. Lansky S. Article reviewed: Quality of life: what is it‘.’ How should it be measured? Oncology 1988276. (29) Tannock IF. Boyer M. When is a cancer treatment worthwhile? N Engl J Med 199(11323z989-9t). (30) Andrews F. Withey S. Social indicators of well—being. New York: Plenum Press. 1976. (.t’l) Calman K. Definitions and dimensions of quality of life. In: Aronson N. Beckmann O. editors. The quality of life of cancer patients. New York: Raven Press. 1987:1-9. (5’2) Ferrans C. Quality of life of breast cancer survivors. Oncol Nurs Forum l990;(216 Suppl). (33’) Oleson M. Subjectively perceived quality of life. Image J Nurs Sch 1990;22:187-90. (34) Ferrans CE. Powers MJ. Quality of life index: developtnent anti psychometric properties. ANS Adv Nurs Sci 1985;8:15»24. Ferrans CE. Powers MJ. Employment potential of hemodialysis patients. Nurs Res 1985;34:273-7. (36) Ferrans CE. Powers MJ. Psychometric assessment of the Quality of Life Index. Res Nurs Health 1992;15:29-38. Campbell A. Converse C. Rodgers W. The quality of American life. New York: Russell Sage. 1976. (38) Martinez-Schalltnoser L. Perinatal depressive symptoms. quality of life. social support. and risk factors in Mexican-American women [disserta— tion]. Chicago: Univ Illinois. 1992. (39) Alibo D. Understanding the proper and improper usage of the compara- tive research framework. J Black Psychol 1988;15:81-91. (40) Adams-Esquivel H. Conceptual adaptation vs. back-translation of multi- lingual instruments: how to increase the accuracy and actionability of multilingual surveys. Proceedings of the American Association of Public Opinion Research. May 1991. Ann Arbor (MI): Am Assoc Public Opin Res. 1991. Berry JW. On cross—cultural comparability. Int J Psychol 1969;4:2()7»29. Triandis HC. The analysis of subjective culture. New York: Wiley-Inter- science. 1972. Scheuch EK. The cross»cultural use of sample surveys: problems of corn- parability. Historical Soc Res 1993118: 104-38. (44) Triandis HC. Some universals of social behavior. Personality and Soc Psychol 1978:4:1-16. (45) Triandis HC. Marin G. Etic plus emic versus pseudoetic: a test of a basic assumption of contemporary cross-cultural psychology. J Cross-Cultural Psychol 1983;14:489-500. (/2 :w J! (37 (4/ (42 (43’ Journal of the National Cancer Institute Monographs No. 20. 1996 (46) (47 (48) (49) (50) (5/) (52 ) (53 ) (54 ) (55 ) (56) (57) (58) (59) (60) (6/) (62) (63 (64) (65) (66) (67) (68) (69) (70) (7/) (72) (73) (74) Strack F. Martin LL. Thinking. judging. and communicating: a process account of context effects in attitude surveys. In: Hippler HJ. Schwarz N. Sudman S. editors. Social information processing and survey methodol— ogy. New York: Springer-Verlag. 198712348. Tourangeau R. Attitude measurement: A cognitive perspective. In: Hip— pler HJ. Schwarz N. Sudntan S. editors. Social information processing and survey methodology. New York: Springer-Verlag. 1987:149-62. Tourangeau R. Rasinski KA. Cognitive processes underlying context ef- fects in attitude measurement. Psycho] Bull 1988;103:209-314. Angel R. Gronfein W. The use of subjective information on statistical models. Am Sociol Rev 1988:53z4o4-73. Dressler WW. Viteri FE. Chavez A. Grell AC. DosSanIOs JE. Compara— tive research in social epidemiology: measurement issues. Ethn Dis 1991;12379—93. Angel R. Thoits P. The impact of culture on the cognitive structure of ill- ness. Cult Med Psychiatry 1987:] 1:465—94. Pepitone A. Triandis HC. On the universality of social psychological theory. J Cross-Cultural Psycltol 1988;18:471—97. Sudman S. Schwarz N. Autobiographical memory and the validity of retrospective reports. New York: Springer-Verlag. 1994. Tulving E. Elements of episodic memory. New York: Oxford Univ Press. 1983. Blair E. Button S. Cognitive processes used by survey respondents to answer behavioral frequency questions. J Consumer Res 1987;14:280—8. Menon G. Judgments of behavioral frequencies: memory search and retrieval strategies. In: Schwarz N. Sudtnan S. editors. Autobiographical memory and the validity of retrospective reports. New York: Springer- Verlag. 1994:161-72. Engle PL. Lutnpkin JB. How accurate are time-use repons‘.’ Effects of cognitive enhancement and cultural differences on recall accuracy. Ap— plied Cognitive Psycho] 1992:62141-59. Fendrich M. Vaughn CM. Diminished lifetime substance use over time. Pub Opin Q 1994;58:961-123. Warnecke RB. Havlicek PL. Manfredi C. Awareness and use of screening by older-aged persons. In: Yancik R. Carbone PP. Patterson WB. Skel K. Terry WD. editors. Perspectives on prevention and treatment of cancer in the elderly. New York: Raven Press. 1983:275—87. Dovidio JF. Fa/io RH. New technologies for the direct and indirect as- sessment of attitudes. In: Tanur J. editor. Questions about questions: in- quiries into the cognitive bases of surveys. New York: Russell Sage Foundation, 1992:204—37. Sudman S. Bradbum N. Schwarz N. Thinking about answers: the applica— tions of cognitive processes to survey methodology. San Francisco: Jos- sey-Bass. 1995. Brewer WF. Memory for randomly sampled autobiographical events. In: Neisser U. Winograd E. editors. Remembering reconsidered: ecological and traditional approaches to the study of memory. Cambridge: Cambridge Univ Press. 1988:21—90. Wagenaar WA. My memory: a study of autobiographical memory over six years. Cognitive Psycho] 1985;18:225-52. Strube G. Answering survey questions: the role of memory. In: Hippler HJ. Schwarz N. Sudman S. editors. Social information processing and survey methodology. New York: Springer-Verlag. 1987:8640]. Perez-Stable. EJ. Sabogal F. Otero—Sabogal R. Hiatt RA. McPhee SJ. Misconceptions about cancer among Latinos and Anglos. JAMA 1992;268:3219-23. Vaughn E. Nordenstam B. The perception of environmental risks among ethnically diverse groups. J Cross-Cultural Psycho] 1991;22:29-60. Wright GN. Phillips LD. Whalley PC. Choo GT. Ng KO. Tan 1. et a]. Cultural differences in probabilistic thinking. J Cross-Cultural Psycho] 1978:92285—99. Bachman JG. O‘Malley PM. Black-white differences in self esteem: are they affected by response styles? Am J Sociol 1984;90:624—39. Hui CH. Triandis HC. Effects of culture and response format on extreme response style. J Cross-Cultural Psycho] 1989;20:269—309. Marin G. Gamba RJ. Marin BV. Extreme response style and acquiescence among Hispanics: the role of acculturation and education. J Cross-Cul- tural Psycho] 1992;23:498—509. Zax M. Takahashi S. Cultural influences on response style: comparisons ofJapanese and American college students. J Soc Psycho] 1967:7113—10. Bachman JG. O‘Malley PM. Yea-saying. nay-saying. and going to ex- tremes: black»white differences in response styles. Pub Opin Q 1984;48:491-509. Bradbum NM. Sudman S. Improving interview method and questionnaire design: response effects to threatening questions in survey research. San Francisco: Jossey-Bass. I979. Dohrenwend B. Social status and psychological disorder: an issue of sub— stance and an issue of method. Am Sociol Rev 1966;31:14-34. Journal of the National Cancer Institute Monographs No. 20. 1996 (75) (76) (77) (78 (79) 32 N (82) (83 ) (84 ) (85) (86) (87) (88) (X9) (90) (9/) (96) (97) (98) (99) (I00) (IUI) (I02) Groves RM. A total survey error approach to AIDS—related survey re— search. In: Fowler FJ. editor. Conference Proceedings on Health Survey Research Methods. Washington. DC: DHHS Pub] No. (PHS) 89-3447. 1989:265-7(). Marin G. Triandis HC. Betancourt H. Kasima Y. Ethnic affirmation ver» sus social desirability: explaining discrepancies in bilingual responses to a questionnaire. J Cross—Cultural Psycho] 1983;14:173-86. Presser S. Pretesting: A neglected aspect of survey research. In: Fowler FJ. editor. Conference Proceedings on Health Survey Research Methods. Washington. DC: DHHS Pub] No. (PHS) 893447. 1989:35-37. Ross CE. Mirowsky J. The worst place and the best face. Social Forces 1983;62:529-36. Ross CE. Mirowsky J. Socially/desirable response and acquiescence in a cross-cultural survey of mental health. J Health Soc Behav 1984;25:189- 97. Triandis HC. Marin G. Lisansky J. Betancourt H. Simptiliu as a cultural script for Hispanics. J Pers Soc Psychol 1984: 47: 1 363—75. Deutchcr I. Asking questions: linguistic comparability. In: Warwick D. Osherson S. editors. Comparative research methods. Englewood Cliffs (NJ): Prentice Hall. 1973:163-86. Schuman H. Presser S. Questions and answers in attitude surveys: experi~ ments on question fornt. wording and context. San Diego: Academic Press. 198]. Carr LG. The scale items and acquiescence. Am Sociol Rev 197136387- 93. Alers JO. Interviewer effects on survey response in an Andean estate. Int J Comparative Sociol 1970:] 1:208-19. Anderson BA. Silver BD. Abramson PR. The effects of the race of the in— terviewer on race-related attitudes of black respondents in the SRC/CPS National Election Studies. Pub Opin Q 19885222899324. Andersen RM. Mullner RM. Comelius LJ. Black-white differences in health status: methods or substance? Milbank Q 1987;65(Supp| l):72-99. Campbell BA. Race-of—interviewer effects among southem adolescents. Pub Opin Q 1981;45:231-44. Cotter PR. Cohen J. Coulter PB. Race—of—intervicwer effects in telephone interviews. Pub Opin Q 1982;46:278-86. Finkel SE. Guterbock TM. Borg MJ. Race»of—interviewer effects in a preelection poll: Virginia 1989. Pub ()pin Q 1991;55:313-30. Hatchett S. Schuman H. White respondents and raceeof—interviewer ef- fects. Pub Opin Q 1975;39:523-28. Reese SD. Dattielson WA. Shoemaker PJ. Chang TK. Hsu HL. Ethnicity- of—interviewer effects among Mexican»Americans and Anglos. Pub ()pin Q 1986;50:503-72. Schaeffer NC. Evaluation race-of—interviewer effects in a national survey. Sociol Methods Res 1980;18:400-19. Schutnan H. Converse J. The effects of black and white interviewers on black responses in 1968. Pub Opin Q 197 l :35:44»68. Sudman S. Bradbum NM. Response effects in surveys: a review and syn— thesis. Chicago: Aldine. I974. Summers G. Hammonds A. Effect of racial characteristics of investigator on self-enumerated responses to a Negro prejudice scale. Soc Forces 1966;48:525-8. Weeks MF. Moore RP. Ethnicity of interviewer effects on ethnic respon— dents. Pub Opin Q 1981;45:245-9. Bond MH. Yang KS. Ethnic affirmation versus cross-cultural accom- tnodation: the variable impact ofquestionnaire language on Chinese bilin- gual in Hong Kong. J Cross-Cultural Psycho] 1982;13:169—85. Feldtnan RH. The effect of administrator and language on traditional— tnodern attitudes among Gusii students in Kenya. J Soc Psychol 1975;96:141-2. Jabine TB. Slaf ML. Tanur JM. Tourangeau R. Cognitive aspects of sur- vey methodology: building a bridge between disciplines. Washington. DC: Natl Acad Press. 1984. Jobe JB. Mingay DB. Cognition and survey measurement: history and overview. Applied Cognitive Psychol 1991 :5: 175—92. Lessler J. Tourangeau R. Salter W. Questionnaire design in the Cognitive Research Laboratory. National Center for Health Statistics. Vita] and Health Statistics. DHHS Pub] No. (PHS)89-15()1. 1989. Royston PN. Using intensive interviews to evaluate questions. In: Fowler FJ. editor. Conference Proceedings on Health Survey Research Methods. Washington. DC: DHHS Pub] No. (PHS) 89-3447. 1989:3-7. Notes lWe thank David Cella. Ph.D.. Rush University. Chicago. IL. for suggesting this approach. Supported by Public Health Service grant ROICA61698-02 from the National Cancer Institute. National Institutes of Health. Department of Health and Human Services (C. E. Ferrans. Principal lnvestigatorl We thank the following individuals for their help: Arthur Boddie. Henry Briele. Tapas Das Gupta. Thomas Lad. Michael Regan. Roobollah Sharit‘i. Michael Warso. Barry Wenig. and the staff at the participating clinics (Univer- 38 sity of Illinois Hospital and Clinics. Cook County Hospital and Clinics. West Side Veterans Administration Hospital and Clinics College of Medicine) and Robert Levin and his staff(Mount Sinai Hospital and Clinics). We also thank the following research assistants who helped with both respondent recruitment and the cognitive interviews: Shaunda Bonds, Charles Bright. Ana N. Chapa. Fran— cisco Perez. and Jonathan VanGeest. Journal of the National Cancer Institute Monographs No. 20, I996 Empirically Selected Instruments for Measuring Quality—of—Life Dimensions in Culturally Diverse Populations Frank Baker, David J 0dre_v, James Zabora, Charlene Douglas, Patricia F ernandez-K elly* We describe a process for developing and testing the cul- tural equivalence of quality-of—life (Q()L) instruments that may be used across culturally diverse populations. Q()L in- struments dealing with satisfaction with various life domains, psychological distress, and physical health and functioning were reviewed by African-American and Hispanic community advisory boards, translated into Spanish and back-translated to ensure translation adequacy. administered to samples of 100 patients from each 0fthe eth- nic minority populations by indigenous nurse interviewers. and examined for psychometric adequacy. Ten Q()L measures showed adequate reliability and validity for fur- ther use in the assessment of Q()L with African-American and Hispanic patients. Three other measures failed to meet the defined standards. A dimension shown to be particularly difficult to address across culturally diverse groups is family functioning. Procedures for achieving cultural equivalence of QOL measures have been shown to be practical and productive. Measures are identified that may be used with some confidence to assess varied dimensions of Q01. with culturally diverse groups. [Monogr Natl Cancer Inst 1996; 20:39-47] As the number of individuals surviving cancer and other life- threatening diseases has increased during the last decade. there has been increasing recognition of the importance of conducting research on the psychosocial adaptation and quality of life (QOL) of long—term survivors (1,2). There has also been in— creased acceptance that QOL should be considered a major out- come criterion for the assessment of the medical effectiveness of demanding treatments. The extent to which interest in QOL in cancer patients has in- creased dramatically is demonstrated by the large number of reviews of this area [e.g.. (3—11)]. These reviews generally agree that QOL is an important criterion in studies of the consequences of treatment of cancer patients. They also agree that QOL is a multidimensional concept that includes psychological. function- al. and social dimensions and that its assessment should include self—report measures from patients. However. the reviews also point out that QOL measurement still poses a variety of problems that trouble those responsible for assessing the effec— tiveness of cancer treatments. Among these problems is the Journal of the National Cancer Institute Monographs No. 20. 1996 question of appropriateness of these measures for use with patients from diverse cultural and socioeconomic backgrounds. Significant evidence exists that minority populations ex— perience higher mortality rates and suffer higher incidences of diseases than other populations in the United States (12.13). Given differential rates in incidence and mortality, one can con- clude that minority populations may lack adequate access to the health care system. Among the various factors that interfere with receiving adequate health care are the generally recognized factors related to adequacy of employment. income. and health insurance coverage. Cultural factors. including attitudes, beliefs. customs. and practices, also affect whether these population groups seek care and how they participate in and respond to care. Barriers also exist related to difficulties with the majority language and educational requirements. These problems not only contribute to underservice of pa- tients from minority and low socioeconomic groups. but also af- fect their inclusion in clinical trials in cancer treatment and supportive care. QOL measurement has grown in acceptance as an important component in medical decision making and in the evaluation of medical treatment effectiveness. The absence of nomtative data from special populations for QOL measures has limited the use of these measures in clinical trials. Problems in Multicultural Research The problem of developing tests and measures that can be used across culturally diverse groups is not limited to medical research: much relevant experience exists in psychological and anthropological research over a considerable period of time. In the United States. the concern for developing adequate tests and measures to use cross-culturally was greatly stimulated by social and political developments in the second half of this century. However. the general problem of developing cross-cultural *Afii‘l/‘uliom ar'uullmrs: F. Baker. D. Jodrey. Department of Environmental Health Sciences, The Johns Hopkins University School of Hygiene and Public Health. Baltimore, MD: J. Zabora. The Johns Hopkins Oncology Center and The Johns Hopkins School of Medicine. Baltimore: C. Douglas. Center for Health Policy. George Mason University School of Nursing and Health Science. Fair- fax. VA: P. Fernandez-Kelly. The Johns Hopkins University Institute for Policy Studies and the Department of Sociology, The Johns Hopkins University. Ct)rres'lmmlt'm'e m preren/ address: Frank Baker. Ph.D.. American Cancer Society. 1599 Clifton Rd.. NE. Atlanta. GA 30329-425 I. See “Notes" section following “References." psychological tests was recognized as early as l91() during the attempt to develop tests for use in research on the comparative abilities of the large groups of immigrants coming to this country at the turn of the century (/4). Those who have criticized multicultural research have iden— tified a number of problems with regard to the instruments used. They include the following: 1) Instruments have been developed on the white middle-class population (/5). and 2) the conceptual base used in creating these instruments has been Eurocentric. and it may be inappropriate to use these instruments with non— white. non-middle—class respondents ( [6—19). Achieving Cultural Equivalence It is possible to reduce cultural distortion in order to make comparisons. such as evaluating the effects of treatment on can— cer patients in culturally diverse groups. This requires a process of adapting instruments to achieve what Flaherty et al. (20) have called “cultural equivalence." Flaherty et al. (20) have identified the following five major dimensions as relevant to establishing the cultural equivalence of measures to be used cross-culturally: l) content equiv— alence whether the content of each item is relevant to the phenomena of each culture being studied; 2) semantic equiv— alence—whether the meaning of each item is the same in each culture after translation into the language and idiom (written or oral) of each culture: 3) technical equivalence whether the method of assessment (e.g.. pencil and paper. interview) is com- parable in each culture with respect to the data it yields; 4) criteria equivalence whether the interpretation of the measure— ment of the variable remains the same when compared with the norm for each culture studied: and 5) conceptual equivalence“ whether the instrument is measuring the same theoretical con— struct in each culture. These five types of equivalence that are necessary for achieving measures that work across cultural boundaries offer a basis for examining QOL instrument development for use with culturally diverse populations. Culturally Equivalent Q()L Instruments Funded by the National Cancer Institute. we have been developing instruments for assessing Q()L that will provide comparable data in cancer patients from culturally diverse populations. including African-Americans and Hispanic Americans who vary in levels of literacy and socioeconomic status. During the first phase of this 3—year project. existing QOL measures have been examined and modified as necessary to achieve cul- tural equivalence. The research design has been structured to deal with each of the five issues of cultural equivalence identified by Flaherty et al. (20). To achieve content equivalence. advisory groups from the two racial/ethnic minority groups reviewed the measures and rated their content. To establish semantic equivalence. the English language ver- sions of measures were translated into Spanish by one bilingual person. and the Spanish version was back-translated into English by others. To deal with differences in colloquial Ian— 40 guage used in different national groups of Spanish speakers. an attempt was made to use “broadcast Spanish." the type of Spanish used on the radio and television. These translations and back-translations were reviewed by an Hispanic American ad- visory board. including members with Central American. South American. and Puerto Rican backgrounds. With regard to technical equivalence. these two advisory groups of individuals familiar with the subcultures of relevance provided input on the procedures of administration of the measures and critiqued the response formats. To reduce the effect of interviewer differences on responses to the QOL questions. the suggestions of Choi and Comstock (2/) were used and included selecting interviewers with similar characteristics and backgrounds. training them adequately and conducting periodic field assessment of their performance. simplifying the questions and reducing the number of possible responses per question. and allocating various types of subjects to the interviewers as uniformly as possible. The measures were administered as part of a face—to—face interview with large com- munity samples of patients from Baltimore. MD. Washington. DC. and Northern Virginia by nurses from the same racial/eth- nic minority groups. who also completed rating fomis on how well the measures were understood and whether there were any problems in acceptability. content. format. or wording. Criteria equivalence and conceptual equivalence are being ex- amined through psychometric and other statistical analyses of the data obtained from samples from the two culturally different groups in relation to data obtained from earlier administrations of the instruments with other groups. Three-Phase Study Our overall research design involves a three—phase process. The first phase. as outlined above. involves review of available instruments. initial testing with community samples of African— American and Hispanic patients with chronic disease. and psychometric evaluation of the performance of these measures. The second phase involves administration of the measures that performed adequately in the first phase to cancer patients from the special populations. The third phase will consist of testing the measures resulting from the first two phases in clinical trials with special population cancer patients. Data from the first phase are currently available as a basis for comparing instru— ments for measuring QOL in culturally diverse populations. In the first—phase study. the initial step was to assess the ade- quacy of these measures by having them reviewed by advisory boards made up of people knowledgeable about the culture and language of the special population groups. Instruments in this review process included measures of satisfaction with various life areas. global QOL. psychological well-being. depression. mood states. level of physical functioning. health status. pain. family functioning. and current concerns. The selection of in— struments was based on a review of the literature and the re- searchers' experience in conducting QOL research with cancer patients and other groups of patients with chronic disease in varied community settings during the past [5 years. QOL measures were reviewed by each of the two advisory boards as described above and then administered to a sample of Journal ofthe National Cancer Institute Monographs No. 20. I996 each population. A sample of 100 African-American and a sample of 100 Hispanic patients with various chronic diseases who resided in Baltimore. Washington, or Northern Virginia were recruited to complete the 10 QOL measures that had sur- vived the advisory board review process. The QOL scales were administered by specially trained inter- viewers recruited from the African—American and Hispanic patient population groups. The interviewers were predominantly nurses who were familiar with the respective communities. The Hispanic interviewers were all fluent in Spanish. Interviewers were given a training manual that described each of the instru— ments and instructed the interviewers with regard to the ad— ministration of these scales and the accompanying interview questions. Separate group-training sessions were held for the African—American and Hispanic interviewers. The booklet that presented the interview questions and the structured scales for the Hispanic patients provided both English (on the left) and Spanish (on the right) versions of the questions on facing pages. Hispanic patients were allowed to respond in English if they preferred to. but the great majority chose to use Spanish in the interview and in answering the scales. Patients were recruited through hospitals, public health clinics. churches. and a variety of other community organiza— tions. The subjects gave informed consent before the interviews. After completing the interview/questionnaire, they were paid a nominal sum for their time and transportation costs. Interviews were conducted at the patients’ homes or at several offices that were made available at Clinics. churches. and other community organizations in Baltimore. Washington. and Northern Virginia. Study Samples Table 1 presents the demographic characteristics of the African—American and Hispanic samples studied. The initial sample of 100 African-American patients included approximate- ly equal proportions of males and females. with an overall mean age of 54.27 years (range. 24—84 years). Almost two fifths (38%) of the patients included in this sample were married or living with someone. With regard to educational level. 3% had only an elementary education. 6% had only a middle school education. 32% had some high school. 27% had graduated from high school, 21% had some college or university education. 7% were college graduates. and 4% did not provide this inforrna— tion. With regard to the initial sample of 100 Hispanic patients. 49% were male. The mean age of Hispanic patients was some— what lower (47.39 years). Almost two thirds of this group were married or living with someone. With regard to citizenship status. 17% were US. citizens. 54% were permanent residents (i.e.. had green cards), 25% were without documents or in some other “informal or temporary” status. and 4% did not provide such information. With regard to education. the Hispanic sample reported less schooling than the African-American sample. QOL Instruments Ten QOL measures have been successfully put through the advisory board review and translation process and administered Journal of the National Cancer Institute Monographs No. 20. 1996 Table 1. Demographic characteristics of the samples Sample African— American. % . Hispanic. % Characteristic (n = 100) (n = 100) Sex Male 48 49 Female 52 5| Age. y Mean 54.27 47.39 Range 24-84 18-82 Marital status Married/living with someone 38 62 Widowed 23 6 Separated/divorced 20 12 Single 19 20 Highest educational level attained Elementary 3 21 Middle school 6 7 Some high school 32 28 High school graduate 27 18 Some college 21 15 College graduate 7 9 Missing" 4 2 *Did not provide information. to the initial samples of 100 African-American and 100 Hispanic patients. Table 2 presents basic data on these measures. including the following characteristics: 1) number of items. 2) type of response format. 3) availability of alternative versions. 4) rating of ease of response. 5) acceptability rating. and 6) reliability (Cronbach alpha). The measures that have been tested for use with culturally diverse populations are described below. Scales l and 2 are general measures of QOL; scales 3-6 assess dimensions of psychological distress; scales 7- 10 deal with physical health and functioning dimensions of QOL. l) The Satisfaction With Life Domains Scale for Cancer (SLDS-C) is a broad measure of QOL that asks about multiple aspects of life. It is based on an earlier Satisfaction With Life Domains Scale developed by Baker et al. (22,23) to assess the QOL of chronic psychiatric patients. The SLDS-C asks those completing the scale to indicate their satisfaction with a number of different life domains relevant to the QOL of cancer patients using a picture response format. Respondents are asked to ex- press their feelings about 17 life areas by choosing one of seven faces. ranging from a “delighted” face with a large smile (scored 7) to a “very unhappy" face with a deep. down-tumed frown (scored 1). a response format shown by Andrews and Withey (24) in national QOL surveys to be easily used and well ac— cepted by most respondents. The “smiley face" response format. which may be presented to the respondent on a card without printed words. has been shown to work well with interviewees who have limited language or conceptual capabilities. The development of the earlier Satisfaction With Life Domains Scale for Mental Illness (SLDS-MI) was undertaken to obtain a QOL measure that could be used in the evaluation of com— munity-based support services for deinstitutionalized mental patients. who were low in levels of socioeconomic status. 41 Table 2. QOL measures for culturally diverse populations Reliabilityf No. of Response Alternative Ease of AfricawAmertcan/ Instrument items fomiat versions response Acceptability Hispanic General QOL Satisfaction With Life 17 Smiley faces General bone +++ +++ 93/90 marrow transplant Domains Scale for Cancer Breast Mentally ill Cantril Ladder of Life 1 Ladder Past +++ +++ NA Present Future Psychologic distress Center for Epidemiologic Studies— 20 No. of +++ +++ .89/91 Depression Scale days in past week felt this way Shacham Profile of Mood States 37 How much Total negative mood ++ ++ 95/96 felt like Subscales adjective ()4 Tension .81/83 Depression 91/92 Anger .87/97 Fatigue 85/92 Vigor .75/.86 Confusion .74/.X1 Bradbum Positive Affect Scale 5 No Affect balance scale + + .65/73 Sometimes Often Bradbum Negative Affect Scale 5 No Affect balance scale + + .8()/.75 Sometimes Often Physical health and functioning Self-rated Kamofsky Performance 7 Check one ++ + Scale statement MOS+ SF-20 Physical Functioning 6 + + Scale MOSWL SF—20 General Health 5 ++ ++ Perceptions Scale MOS’r SF-20 Bodily Pain Scale 1 +++ +++ *The reliability data presented here are alphas based on administrations of the scales to community samples of 100 African-American and 100 Hispanic patients (Spanish language version used). NA = not applicable. rMedical Outcomes Study. literacy, and ability to handle abstractions. The SLDS-C shows only partial overlap in the life domains rated in completing the scale. Sample items include “your relations with friends,“ “your body,” “how comfortable you feel,” and “your ability to attain sexual satisfaction.” Evidence of the reliability of the SLDS—C was first obtained from analysis of the responses of a sample of 109 cancer pa- tients. The coefficient alpha for this group was .93, and evidence of the SLDS-C’s concurrent validity was also demonstrated on the basis of its correlation with several other QOL measures (25). Evidence of its sensitivity to change was obtained by com- paring the responses of 64 cancer patients who completed the measure as part of a resurvey 2 years after completing an initial mailed questionnaire. A statistically significant gain in average SLDS-C was observed in the subset of cancer patients who indi- cated that their health had improved (25). The construct validity of the bone marrow transplant version of the scale (which has an additional item that asks about bone marrow transplantation) has been supported by a study showing that the ability of 135 cancer survivors who had had bone marrow transplants to maintain their valued social roles was significantly related to a higher 42 QOL as measured by the 18—item SLDS—BMT version of the scale (26). 2) The Cumril Ladder afL/fe is another graphic method for studying QOL. It asks about overall life satisfaction (27). Respondents are shown a “ladder of life“ with l() rungs; 0 repre- sents the worst possible life (as the person conceives it), and 10 represents the best possible life. Respondents indicate where they are on the ladder at the present time. Other versions ask patients to indicate where they were in the past (e.g.. before they had cancer) and where they expect to be in the future. This method has been described as “self—anchoring," since the respondents establish where they are at a particular time on the ladder of life and can use that judgment as a basis for com— patibility rating their lives at other time points. This simple measure has been widely used in the study of life satisfaction. 3) The Centerfnr Epidemiolngic Studies—l)epression Scale (CES-D) is a self-report measure of the frequency of depression rated for the past week (28). The CES—D consists of 20 items for which patients are asked to circle a number on a scale of 1—4; 1 is defined as “rarely or none of the time (less than 1 day)," and 4 is defined as “most or all of the time (5—7 days)" The CES—D Journal of the National Cancer Institute Monographs No. 20, 1996 was developed initially for use in epidemiologic surveys with the general population. and its use for screening people for symptomatology related to depression has been well established (29.30). The CES—D is unusual among the QOL measures dis- cussed here. in that it has already been translated into Spanish by its developers and has undergone reliability and validity test- ing with general samples of several US. ethnic minority groups (31). Roberts (32) in a comparison of samples of white subjects of non-Hispanic origin. African-Americans. and Mexican— Americans found no differences among the three groups with regard to missing data or alpha coefficient reliability. The CES- D has frequently been used to evaluate depression symptoms in cancer patient populations (33-35) and has the advantage over other measures of depression of being less biased by the in- clusion of items asking about physical concerns that might be expected to reflect symptoms of cancer or its treatment rather than depression (36). 4) The Profile ofMoocl States (POMS) assesses transient. dis- tinct mood states by self-report on an adjective checklist (37). yielding an overall score of total negative mood and six factor scores including the following: 1) tension—anxiety. 2) depres- sion—dejection. 3) anger—hostility. 4) fatigue—inertia. 5) vigor— activity. and 6) confusion—bewilderment. The original 65—item POMS has shown internal consistencies ranging from .84 to .95 and test—retest reliabilities ranging from .65 to .74 for 20 days and .43 to .53 for 9 weeks (37,38). The version we are using is the shortened 37-item POMS developed with cancer patients by Shacham (39). and it has shown correlations with the original full-length scale of .95. indicating stability of the shorter ver- sion. The POMS has frequently been used to assess the psychological status of patients with cancer [e.g.. (40—44)] and has also been employed to examine the psychosocial impact of cancer on the family (45-48). Graydon (49) found that. for can— cer patients (while adjusting for diagnosis and age). the tension— anxiety component was the best predictor of functioning after therapy. Cassileth et a1. (44) have provided comparative POMS scores for cancer patients and next of kin. In prior research by members of this research group with cancer patients surviving bone marrow transplantation. the obtained alpha reliability coef- ficient for total negative mood on the short Shacham form of the POMS was .94 (26). 5 and 6) The Brat/burn Positive and Negative Ajj‘ect Scales comprise a set of 10 questions (five negative and five positive) that ask respondents about their recent affective experiences. In- tended by Bradburn (50) to be a single measure of psychological well—being. the two five-item clusters have been found to be in— dependent in a number of studies (23.24.51) and are often used as separate measures of affect. The two measures. the Positive Affect Scale and the Negative Affect Scale. have been shown to be useful outcome measures for chronically ill patients (22). In previous research by members of this research group with cancer patients surviving bone marrow transplantation. the ob— tained alpha reliability coefficients were .83 for the Positive Af- fect Scale and .60 for the Negative Affect Scale (26). Other researchers have also found this measure to be a useful one in studying the affect dimension of QOL among cancer patients [e.g.. (52)]. Journal of the National Cancer Institute Monographs No. 20. 1996 7) The Self-Rated Karnofsky Pezfarmance Scale (SR—KPS) is a measure of physical functioning for cancer patients. It was developed to provide a self-report version of the classic physician-rated Karnofsky scale (53). Patients rate themselves by a 10-point increment from 40 (low-level functioning requir- ing help) to 100 (high—level functioning requiring no help). The categories of 10. 20. and 30 have been deleted because patients at these lower levels of functioning would be unable to par- ticipate in such a study. In a survey of 70 cancer patients after bone marrow transplantation. the SR-KPS was validated against a physician‘s ratings using the traditional Karnofsky scale. and statistically significant kappas were obtained (54). 8) The MOS SF-20 Physical Functioning Scale is from the Medical Outcomes Study (MOS) and is a 20—item short form (54). which was designed at the RAND Corporation (Santa Monica. CA) as a quick (<5 minutes) self-administered ques- tionnaire for use in large-scale patient surveys. The alpha ob— tained for the Physical Functioning Scale in the original study was .86 (54). It has been shown to be a useful measure of physi- cal functioning in the study of health and functional status of adult cancer survivors after bone marrow transplantation by Wingard et al. (53). Responses range from 1 (“all of the time") to 6 (“none of the time") regarding how much of the time during the last month the respondents” health has limited them in each of six types of activities they can do. ranging from vigorous ac- tivities such as “lifting heavy objects. running or participating in strenuous sports." to activities of daily living such as “eating. dressing. bathing or using the toilet." 9) The MOS SF-2() General Health Perceptions Scale (54) is also from the MOS SF-20 and consists of five overall ratings of current health in general. Four of the items ask the respondents to circle a number from I (“definitely true”) to 5 (“definitely false”) indicating whether each statement is true or false for them. Items include such statements as: “I am somewhat ill" and “My health is excellent." A fifth item asks the respondent to circle a number from 1 (“excellent") to 5 (“poor") in rating how their overall health is at the present time. The alpha obtained for this measure in the original study was .88 (54). This scale was successfully employed in the study by Wingard et al. (53) in as- sessing the perceived health status of bone marrow transplant survivors. Although a longer form of the MOS measures (i.e.. MOS SF-36) is now available. these scales from the earlier short form have the advantage of simplicity and of having already been used with cancer patients in several studies (53). 10) The MOS SF-ZO Bodily Pain Scale is a single item from the MOS SF-20 (54) that asks the following question: How much bodily pain have you had during the past week? Respon- dents are asked to circle a number from 1 (“none”) to 5 (“severe”). We used a modified wording of this item that asked how much pain the respondent was “feeling now.” This version has been shown to be a simple but useful measure of self- reported level of pain in the study of bone marrow transplant survivors noted above (53). Analysis Plan Our analysis plan was first to examine the psychometric per- formance of the various scales. A standard approach to 43 reliability of multi—item scales is to calculate each scale‘s coeffi— cient alpha (55). which has been described as a measure of inter— nal consistency. Reliability as estimated by the alpha coefficient has been considered to be acceptable for a scale at as low a level as .50 for making group comparisons (56). although Nunnally (57) has recommended using a standard of reliability of .70 or higher. Preliminary tests of validity were also conducted with this pilot dataset. Convergent validity was examined in terms of the correlation among measures hypothesized as measuring the same dimension of QOL. Discriminant validity was gauged by the extent to which absolute values of correlations were lower for measures presumed to be assessing different dimensions of QOL than for those concerned with the same dimension. Scales were also assessed in terms of the amount of missing data and inconsistency in factor structure for those scales that were developed as having a particular factor structure. Results Rejected Instruments Some measures that we put through the advisory board review and initial testing with special population samples did not survive the process. Three of these measures are of par— ticular interest. Our experience with assessing QOL had reminded us that cancer is not just a disease that affects the patient, but. in a sense. it is a “family disease" that has an impact on those who are close to the patient and are affected by his or her ordeal (58). After reviewing ll scales related to family functioning, we selected the 20-item FACES 111 (Family. Adaptation. and Cohesion Evaluation Scales) developed by Olson et al. (59), based on Olson's circumplex model of marital and family sys- tems. Unfortunately, although this instrument was based on sur- veys of thousands of adults (60), it was not well accepted by either the African—American or Hispanic samples of patients whom we asked to complete it. Both the African—American and the Hispanic patient samples particularly objected to items that seemed to assume that children share power with their parents in family decision making. such as the following items: “Parents and children discussed punishment together": “In solving prob— lems. the children's suggestions were followed": and “Children had a say in their discipline." Even though the respondents had the option of indicating that the particular behavior almost never happened. many of the subjects rejected these items as inap— propriate. As a result, they either refused to answer the scale al- together or left a number of the items unanswered. The Inventory of Current Concerns (lCC), a classic self— report measure built to assess multiple QOL dimensions (Weis- man AE. Worden JW, Sobel HJ: unpublished report). was examined. The [CC is a 72—item, self—report inventory that focuses on the patient's “current concerns." These concerns are categorized into seven areas: 1) health, 2) family. 3) work— finance. 4) friends, 5) religion, 6) existential. and 7) self—ap- praisal. The lCC got through the review by the two advisory boards. although there were concerns about its length and the wording of some of the items: however. it encountered major difficulties at the testing stage of the study. Missing data were 44 higher for items in the scale than any other scales except the FACES 111. Finally. a third scale that was reviewed for possible use was rejected early in the process of measurement review. In order to ascertain the social desirability response bias, the tendency to give answers that make the respondent look good. a shortened version of the Marlowe—Crowne Social Desirability Scale (M-C SDS) (61) was considered for inclusion in our study. However. this lO—item scale was eliminated during the advisory board process because members of the African-American Advisory Board objected to this measure in the strongest terms. suggest— ing that it was extremely insensitive and implied that minority patients were not trusted to tell the truth. The Hispanic Advisory Board did not object in such strong terms, but they too felt that the measure was somewhat offensive. Acceptable Measures Table 2 summarizes the results from the first phase of our re— search in developing culturally equivalent QOL measures. For each of the 10 instruments, the table presents l) the number of items in each scale, 2) the nature of each scale’s response for- mat, 3) whether there have been alternative versions of the measure developed and/or whether it has subscales, 4) how easy using a particular scale’s format was for the special populations we studied, 5) how acceptable each instrument appeared to be, and 6) the coefficient alpha reliability obtained from analysis of the responses to each scale of the initial community samples of 100 African-American and 100 Hispanic patients. Most of the scales are multi-item scales, except for the Cantril Ladder of Life and the MOS SF-20 Bodily Pain Scale. Four of the scales have alternate forms, and one has specific subscales as well as a total scale score. The SLDS has several alternative forms. but only the general cancer version, the SLDS-C, was ex— amined in this study. The Cantril Ladder of Life has been used to rate one’s life as it was in the past and as one expects it to be in the future, but only the versions asking the respondents to rate their lives right now were used. The POMS originally was 65 items long, but we are using the shorter Shacham 37-item form developed with cancer patients; however, both the whole scale score for total negative mood and the six subscale scores were examined here. The Bradburn Positive and Negative Affect Scales were originally added together algebraically to create an Affect Balance Scale score (50), but we have treated them as two different scales in our analysis, consistent with the way they have generally been used in studies with cancer patients (26). While all 1() measures were found to be reasonably accept- able to the groups to whom we administered them, the measures scoring the highest average rating of ease of use were the SLDS, the Cantril Ladder of Life, the CES-D, and the MOS SF-20 Bodily Pain Scale. reflecting their one—item simplicity or their use of picture response formats. The interviewers observed that these measures seemed most acceptable to respondents as well. The coefficient alphas ranged from .65 to .93, indicating accept- able reliability for all the measures. The scales with highest coefficient alphas were the SLDS, the CES—D. and the Total POMS. The Bradburn Positive Affect Scale had the lowest alpha levels with both ethnic minority groups. Journal of the National Cancer Institute Monographs No. 20, 1996 Table 3. Correlations across QOL measures for African-American patients Medical Medical Outcomes Medical Center for Shacham Outcomes Study SF-ZO Outcomes Cantril Epidemiologic Profile Bradbum Bradbum Self-rated Study SF-ZO General Study SF~20 Ladder Studiesi of Mood Positive Negative Kantofsky Physical l leallh Bodily of Life Depression States— Affect Affect Perfonnance Functioning Perceptions Pain (now) Scale total Scale Scale Scale Scale Scale Scale Satisfaction With Life 59* —.74* 776* 231' v.68* .6 l *‘ 55* .61* #.48* Domains Scale for Cancer Cantril Ladder of Life (now) —.56* —.55* .43* —.54* 49* 44* 518* 739* Center for Epidemiologic 34* —.3()* 69* —.4X* ~.4()* —.53* 36* Studies—Depression Scale Shacham Profile of Mood 732* .76* 754* —.47* 154* 36* States—total Bradbum Positive Affect Scale —.2l1L 28* .14 .25'1' —.2()‘l’ Bradbum Negative Affect Scale —.54* ~.46* 754* 39* Self-rated Kamofsky Performance .74* .(15'* 755* Scale Medical Outcomes Study SF—ZO .71 "‘ —.53* Physical Functioning Scale Medical Outcomes Study SF—ZO —.49* General Health Perceptions Scale *Significance, PSO l. ’rSignificance, P305. Validity of QOL Measures Tables 3 and 4 present intercorrelation matrices for the African—American and Hispanic samples, respectively. The SLDS shows high correlations (about .5 or above) with every measure considered here. except that, for the African-American sample, the correlation with positive affect is smaller (r = —.23). Presumably, this reflects the fact that the SLDS is a comprehen— sive measure that covers a broad range of QOL domains. Originally, the ICC was to be used to test conversant validity with SLDS-C, but this could not be done because of the ICC’s poor performance and low acceptability. as discussed below. Future studies are planned to examine the convergent validity of this multidomain measure with another QOL measure that measures multiple QOL dimensions. The one-item Cantril Ladder of Life for the present has its highest correlation with the SLDS for the African-American sample and scored highest for the Hispanic sample. Like the other general QOL measure. the SLDS, it shows high negative correlations with the measures of psychological distress. Establishing convergent and discriminant validity for the measures of psychological distress used in these samples re— Table 4. Correlations across QOL measures for Hispanic patients Medical Medical Outcomes Medical Center for Shacham Outcomes Study SF-ZO Outcomes Cantril Epidemiologic Profile Bradbum Bradbum Selllratcd Study SF-2() General Study SF-20 Ladder Studies7 of Mood Positive Negative Karnofslx'y Physical Health Bodily of Life Depression State s7 Affect Affect Performance Functioning Perceptions Pain (now) Scale total Scale Scale Scale Scale Scale Scale Satisfaction With Life .66* — 72* —.75* .45* —.61 * 117* 59* .65* —.49* Domains Scale for Cancer Cantril Ladder of Life (now) —.63* —.64* 50* v.51* 56* .57* 74* 752* Center for Epidemiologic .85* 745* 73* -.51* —.50* 7.64‘“ 48* Studies—Depression Scale Shacham Profile of Mood A.42* 80* #.47* —.44* 758* .51 * States—total Bradbum Positive Affect Scale —.35+ 38* 23+ .40'l' —.l7 Bradbum Negative Affect Scale —.35* —.30* 412* 42* Self-rated Kamofsky Performance .74* .67* 750* Scale Medical Outcomes Study SF-20 53* —.59* Physical Functioning Scale Medical Outcomes Study SF—20 —.58* General Health Perceptions Scale *Significance. P = .01. 1'Significance, P = .05. Journal of the National Cancer Institute Monographs No. 20. 1996 45 quires examining the pattern of correlations for the CES-D Scale. the POMS total score. and the Bradbum Negative Affect Scale. As shown in Tables 2 and 3. the CES—D and the POMS scales have their highest correlations with each other. In addi- tion. the Bradbum Negative Affect Scale has its highest correla- tions with these two other measures of negative mood. All these scales have considerably smaller correlations with the Bradbum Positive Affect Scale. which is appropriate. as the two Bradbum scales are intended to be (approximately) independent of each other. As these tables show. that is not quite the case here, but the correlation of positive affect and negative affect is smaller than most of the correlations that each has with other QOL measures. The measures that deal with physical functioning also show a pattern of higher intercorrelations than with other QOL measures. The SR—KPS measure of physical functioning and the MOS SF-20 Physical Functioning Scale show their highest cor- relations with each other. The lowest absolute values for cor- relations with the SR-KPS measure are the Bradbum Positive Affect and Negative Affect Scales. This is also true of the MOS SF-20 Physical Functioning Scale. Current pain shows moderate correlations with nearly every measure. with the exception of weak or nonsignificant correla— tion with the Bradburn Positive Affect Scale. For the Hispanic sample. its strongest correlations are with the MOS SF-20 Physical Functioning Scale and the MOS SF-20 General Health Perceptions Scale. In the African-American sample. the strongest correlations are with the SR-KPS scale and the MOS SF-ZO Physical Functioning Scale. The MOS SF-20 General Health Perceptions Scale has its strongest correlations in the Hispanic sample with the Cantril Ladder of Life evaluating one‘s current life (r = .74). Close be- hind are the SR-KPS (r = .67) and the MOS SF-ZO Physical Functioning Scale (r = .63). as well as the SLDS Scale (1‘ = .65) and the CES-D Scale (r = —.64). In the African—American sample. the MOS SF-20 General Health Perceptions Scale has its strongest correlations with the MOS SF—20 Physical Functioning Scale (r = .71) and the SR-KPS (r = .65). The cor- relation of the MOS SF-20 General Health Perceptions Scale with the Bradbum Positive Affect Scale is notably stronger among Hispanics (r = .40. P305) than among African- Americans (1' = .14, not significant). The validity of measures for particular populations requires more than a single study. However. the initial findings are at least encouraging that there is good support for both convergent and discriminant validity for those broad QOL dimensions where several concordant measures are available for com- parison. Advantages of Graphic Response Formats In facilitating response by respondents who have limited literacy levels. it is helpful to use pictures. cartoons. or other graphic response formats that minimize dependence on ability to read text. Our experience with the SLDS—C with its use of the “smiley-face" response format and with the Cantril Ladders of Life with their use of a ladder fonnat has shown the advantages of using response formats that do not depend much on language facility. 46 The Dartmouth COOP Charts provide another example of an approach that uses pictures as a basis for eliciting information from patients. The COOP Chart system includes nine charts to screen and monitor patients' functioning and was specifically designed for use in physicians‘ offices during the doctor—patient interview (62.63). Stick—figure drawings are used in these charts to indicate different levels of functioning. C. C. Gotay (personal communication. 1995) and colleagues are developing the COOP Charts for assessment of QOL with several Asian-American and Hawaiian populations. They have developed new charts to answer additional demands and have made minor modifications to the COOP Charts on the basis of their pretesting. Conclusions The results of this analysis of our initial first-phase study offer encouragement regarding the potential for some available measures of different QOL dimensions to be used as culturally equivalent measures across African-American and Hispanic patient populations. Ten measures have been identified that appear to be worthy of further reliability and validity studies with culturally diverse cancer patient samples and eventual test— ing in multiple clinical trials. One particular QOL dimension has been identified that re- quires special attention because of differences in values and nor— mative assumptions across culturally diverse groups. That is the QOL dimension of family functioning; one of the most popular measures of family functioning. the FACES III. apparently failed to be acceptable because its middle-class. Eurocentric as- sumptions about the role of children in family decision making seemed too foreign to be taken seriously. Attaining culturally equivalent measures in the domains of family functioning offers a particularly interesting challenge for further research. Finally. the procedures described here for assessing the cul- tural equivalence of QOL measures have been shown to be prac- tical and productive. Techniques to establish cultural equivalence. utilization of special advisory boards. and comple- tion of interviews in the community offer methods to evaluate available measures of QOL in diverse populations. These tech- niques are equally applicable to measures other than QOL and for use in other special populations. References (1) Holland JC. Silverfarb P. Tross S. Cella D. The Cancer and Leukemia Group B (CALGB) experience. In: Ventrafridda V. Van Dam FS. Yancik R. Tamburini M. editors. Assessment of quality of life and cancer treat- ment. Amsterdam: Excerpta Medica. 1986. (2) Aaronson NK. Beckmann .l. The quality of life of cancer patients. New York: Raven Press. I987. (3) Fayers PM. Jones DR. Measuring and analysing quality of life in cancer clinical trials: a review. Stat Med 1983;2z429-46. (4) Holland JC. Methodology in behavioral and psychosocial cancer research. Need for improved psychosocial research methodology: goals and poten- tials. Cancer l‘)84;53( ll) Suppl):22]8~2(). (5) de Haes JC. van Knippenberg FC. The quality of life of cancer patients: a review ofthe literature. Soc Sci Med 1985220180947. ((1) Ventafridda V. Van Dam FS. Yancik R. Tamburini M. editors. Assessment of quality of life and cancer treatment. Amsterdam: Excerpta Medica. I986. (7) Clark A. Fallowl'ield LJ. Quality of life measurements in patients with malignant disease: a review. J R Soc Med l986:79:|65-9. Journal ofthe National Cancer Institute Monographs No. 20. I‘)% (8) (9) (/0) (ll) (/4) ([5 (lo) (/7) (/8) (26) (27) (28 (29) (30) (3/) (32) (3’3 (34) (35) (36) (37) Selby P. Robertson B. Measurement of quality of life in patients with can— cer. Cancer Surv 19872652143. Donovan K. Sanson-Fisher RW. Redman S. Measuring quality of life in cancer patients. J Clin Oncol 1989;72959-68. Aaronson NK. Meyerowitz BE. Bard M. Bloom JR. Fawzy FI. Feldstein M. et a1. Quality of life research in oncology. Past achievements and future priorities. Cancer 1991:67(3 Suppl):839-43. Moinpour CM. Savage M. Hayden KA. Sawyers J. Upshurch C. Quality of life assessment in cancer clinical trials. In: Dimsdale JE. Baum A. editors. Quality of life in behavioral medicine research. Hillsdale (NJ): Lawrence Erlbaum Associates. 1995:79-95. Payne KW. Ugarte CA. The Office of Minority Health Resource Center: impacting on health related disparities among minority populations. Health Educ 1989:2026—8. Baquet CR. Hunter CP. Patterns in minorities and special populations. In: Greenwald P. Kramer BS. Weed DL. editors. Cancer prevention and con- trol. New York: Marcel Dekker. 1995:23-36. Anastasi A. Psychological testing. 5th ed. New York: Macmillan. 1982. Ponterotto JG. Racial/ethnic minority research in the Journal qf‘Cozinsel- ing Psychology: a content analysis and methodological critique. J Counsel— ing Psychol 1988;35:410-8. Casas JM. Policy. training and research in counseling psychology: the ra- cial/ethnic minority perspective. In: Brown S. Lent R. editors. Handbook ofcounseling psychology. New York: Wiley. 1984:785-831. Lonner RP. Ibrahim FA. Assessment in cross—cultural counseling. In: Pedersen PB. Draguns JG. Lonner WJ. Trimble JE. editors. Handbook of psychotherapy and behavior change: an empirical analysis. 2nd ed. New York: Wiley. 1989:903—38. Sue DW. Effective multicultural counseling: proposed research directions. In: Fouad NA. Chair. Cross—cultural research and training: focus on critical issues. Symposium presented at the annual meeting of the American Psychological Association. New Orleans. August 1989. Sue S. Psychotherapeutic services for ethnic minorities. Two decades of research findings. Am Psychol 1988;43:301-8. Flaherty JA. Gaviria FM. Pathak D. Mitchell T. Wintrob R. Richman JA. et a1. Developing instruments for cross—cultural psychiatric research. J Nerv Ment Dis 1988;176:257-63. Choi 1C. Comstock GW. Interviewer effect on responses to a questionnaire relating to mood. AmJ Epidemiol 1975;101:84-92. Baker F. Intagliata J. Quality of life in the evaluation of community support systems: a review of the literature. Eval Program Planning 1982:5:69—79. Baker F. Jodrey D. Intagliata J. Social support and quality of life of com— munity support clients. Community Ment Health] 1992;28:397—411. Andrews FR. Withey SB. Social indicators of well—being: Americans’ per— ceptions of life quality. New York: Plenum Press. 1976. Baker F. Curbow B. Wingard J. Development ofthe Satisfaction With Life Domains Scale for Cancer (SLDS-C). J Psychosocial Oncol 1992;10:75-90. Baker F. Curbow B. Wingard JR. Role retention and quality of life of bone marrow transplant survivors. Soc Sci Med 1991;32:697-704. Cantril H. The pattern of human concerns. New Brunswick (NJ): Rutgers Univ Press. 1965. Radloff LS. The CES-D scale: a self—report depression scale for research in the general population. Appl Psychol Measurement 1977;1:385—401. Myers JK. Weissman MM. Use of a self—report symptom scale to detect depression in a community sample. Am J Psychiatry 1980;137:1081—4. Roberts RE. Vernon SW. The Center for Epidemiologic Studies Depres— sion Scale: its use in a community sample. Am J Psychiatry 1983;140:41- 6. Naughton MJ. Wiklund I. A critical review of dimension—specific measures of health-related quality of life in cross»cultural research. Qual Life Res 199322397432. Roberts RE. Reliability of the CES—D Scale in different ethnic contexts. Psychiatry Res 198():2:125-34. Zondcrman AB. Costa PT Jr. McCrae RR. Depression as a risk for cancer morbidity and mortality in a nationally representative sample [sec com- ment citation in Medline]. JAMA 1989126221 191-5. Grit] ER. Wellisch DR. Siau J. Wang H]. Long-term effects of testicular cancer on marital relationships. Psychosomatics 1990131 :301-12. Devins GM. Orme CM. Costello CG. Binik YM. Measuring depressive symptoms in illness populations. Psychometric properties of the Center for Epidemiologic Studies Depression (CES—D) Scale. Psychol Health 1988;22139-156. Fobair P. Hoppe RT. Bloom J. Cox R. Varghese A. Spiegel D. Psychoso- cial problems among survivors of Hodgkin's disease. J Clin Oncol 1986:4:805-l4. McNair DM. Lorr M. Droppleman LF. EITS manual for the Profile of Mood States. San Diego (CA): Educational and Industrial Testing Service. 1971. Journal of the National Cancer Institute Monographs No. 20. 1996 (38) (39) (40) (41) (42 ) (43) (44) (45) (4O) (47) (48) (49) (54) (55) (56) (57) (58) (59) (60 (61) (62) (63 ) Eichman WJ. Review of the Profile of Mood States. In: Buros OK, editor. The eighth mental measurements yearbook. Highland Park (NJ): Gryphon Press. 1978. Shacham S. A shortened version of the Profile of Mood States. J Pers As- sess 1983:47:305»6. Achterberg J. Lawlis GF. A canonical analysis of blood chemistry vari- ables related to psychological measures of cancer patients. Multivariate Exp Clin Res 1979:421-10. McCorkle R. Quint-Benoliel J. Symptom distress. current concerns and mood disturbance after diagnosis of life threatening disease. Sci Med 1984;17:431-8. Taylor SE. Lichtman RR. Wood JV. Attributions. beliefs about control. and adjustment to breast cancer. J Pers Soc Psychol 1984;46:489—502. Worden JW. Weisman AD. Preventative psychosocial intervention with newly diagnosed cancer patients. Gen Hosp Psychiatry 1984;62243-9. Cassileth BR. Lusk E]. Walsh WP. Doyle B. Maier M. The satisfaction and psychosocial status of patients during treatment for cancer. J Psychosoc Oncol 1989:7:47-57. Cassileth BR. Lusk E]. Strouse TB. Miller DS, Brown LL. Cross PA. A psychological analysis of cancer patients and their next—of—kin. Cancer 1985;55:72-6. Giacquinta B. Helping families face the crisis of cancer. Am J Nurs 1977;77:1585-8. Huth CM. Illness and the family. Ann Intern Med 1978;89:132-3. Plumb MM. Holland J. Comparative studies of psychological function in patients with advanced cancer—I. Self—reported depressive symptoms. Psychosom Med 1977;39:264—76. Graydon JE. Factors that predict patients’ functioning following treatment for cancer. Int J Nurs Stud 1988;25:117-24. Bradbum NM. The structure of psychological well-being. Chicago: Aldine, 1969. Beiser M. Components and correlates of mental well-being. J Health Soc Behav 1974;15:320-7. Coward DD. Self—transcendence and emotional well-being in women with advanced breast cancer. Oncol Nurs Forum 1991;18:857-63. Wingard JR. Curbow B. Baker F. Piantadosi S. Health. functional status. and employment of adult survivors of bone marrow transplantation. Ann Intern Med 1991;114:113—8. Stewart AL. Hayes RD. Ware JE. The MOS short—form general health sur- vey: reliability and validity in a patient population. Unpublished working draft. Santa Monica (CA): The RAND Corporation. 1987. Cronbach LJ. Coefficient alpha and the internal structure of tests. Psychometrika1951;16:297-334. Helmstadtler GC. Principles of psychological measurement. New York: Appleton—Century—Crofts. 1964. NunnaIIy JC. Psychometric theory. 2nd ed. New York: McGraw-Hill. 1978. Zabora J. Smith ED. Baker F. Wingard J. Curbow C. The family: the other side of bone marrow transplantation. J Psychosoc Oncol 1992;10:35-46. Olson DH. Portner J. Lavee Y. FACES 111. St. Paul (MN): Family Social Science. Univ Minnesota. 1985. Olson DH. Circumplex Model VII: validation studies and FACES III. Fam Process 1986;25:337-51. Crowne D. Marlowe D. A new scale of social desirability independent of psychopathology. J Consulting Psychol 1960;24:349-54. Nelson E. Wasson J. Kirk J. Keller A. Clark D. Detrich A. et a1. Assess— ment of function in routine clinical practice: description of the COOP Chart method and preliminary findings. J Chronic Dis 1987;40 Supp112555—69S. Nelson EC. Landgraf JM. Hays RD. Wasson JH. Kirk JW. The functional status of patients. How can it be measured in physicians” offices? Med Care 1990;28:1111-26. Notes Supported by Public Health Service grant IROICA61673 from the National Cancer Institute. National Institutes of Health. Department of Health and Human Services. We acknowledge and give our special thanks to the members of the two com- munity advisory boards who have participated in the instrument review process described in this article. The members of the African American Advisory Board included Ms. Nina Harper. Dr. Miles Ham'son. Dr. Melva Owens. Pastor Mar- shall Prentice. Ms. Demetria Rogers. and Rev. Melvin Tuggle II. The members of the Hispanic Advisory Board included Sister Mary Neil Corcoran. Ms. Libby Garcia-Herd. Dr. Elmer Huerta. Mr. Pedro Ponceano. Ms. Haydee Rodriguez. and Dr. Ruth Zambrana. 47 Quality—of—Life Research in the Pediatric Oncology Group: 1991—1995 Andrew S. Bradlyn. Brad H. Pol/0016* Quality-of-life end points for cancer clinical trials have received much attention in the adult literature. However, within pediatric cancer clinical trials, the inclusion of these alternate end points has only recently been considered. We review the Pediatric Oncology Group’s approach to re- search in this area and describe our guidelines and protocols that incorporate quality-of-life end points and several of the methodologic barriers that must be addressed. [Monogr Natl Cancer Inst 1996;20:49-53] Over the past 10 years. there has been increasing emphasis on the role of alternate end points. such as quality of life (QOL). in cancer clinical trials. For example. QOL data have been used in a variety of phase III clinical trials that have compared different treatment modalities (l). regimens hypothesized to reduce toxicity (2). and the long-term psychosocial adjustment of sur- vivors (3). Additionally. QOL data have been examined as prog— nostic factors in attempts to identify patient characteristics associated with differential responsiveness to therapeutic regimens (4). However. QOL end points have not typically been incorporated into pediatric cancer clinical trials. A recent retrospective review of phase III pediatric trials reported that less than 5% included QOL outcomes by almost any definition (5). Since 1991. the Pediatric Oncology Group (POG) has made a concerted effort to examine the potential contribution of end points. such as QOL. in the evaluation of alternate therapies. These end points are currently being incorporated in a variety of clinical trials. Background In 1980. the pediatric sections of the Cancer and Leukemia Group B and the Southwest Oncology Group were merged to form the POG. The two National Cancer Institute—sponsored pediatric cooperative groups (POG and the Children‘s Cancer Group) provide protocol-driven treatment for the vast majority of children and adolescents in the United States who are diag- nosed with a malignancy (6). As a result of continued advances in diagnosis and treatment. the prognosis for many of these patients has changed dramatically over the past 30 years. For ex- ample. while the 5-year survival rate for patients diagnosed with acute lymphoblastic leukemia in the early 1960s was less than 5%. more recent data project a greater than 70% 5—year survival rate for these patients (7). However. while our effectiveness in improving the quantity of children‘s survival has been substan- Journal of the National Cancer Institute Monographs No. 20. 1996 tial. our understanding of the quality of their survival has been limited. Approximately 4 years ago. the POG established a formal mechanism for examining the role that QOL end points might play in its clinical trials. At the direction of the Supportive Care and Cancer Control Committee. a subcommittee on QOL was established and convened its first meeting at the fall 1991 group meeting. Accomplishments of the Committee The committee comprises representatives from a variety of disciplines. including pediatric oncology. psychology. nursing, data management. epidemiology. and biostatistics. Initially. it was agreed that we needed to establish a consensus definition of QOL in pediatric oncology and to provide written guidelines to steer the POG‘s efforts in this area. Therefore. we developed a set of guidelines (reviewed below) that define QOL. discuss the selection and timing of measures. and address the quality con— trol of this data collection. The objectives of these guidelines were to arrive at an organized. uniform, and coherent framework for QOL assessment within the FCC and to employ methodologically sound data collection procedures and instru- ments to address these research questions. Additionally. we wished to answer QOL questions in an efficient manner. maxi- mizing the benefits in relation to their cost. in terms of both financial and human capital. Definition We adopted the following definition of QOL. based in large part on the World Health Organization‘s definition of health (8): Quality of life is a multidimensional construct. incorporat— ing both objective and subjective data. including (but not limited to) the social. physical. and emotional functioning of the child and. when indicated. his/her family. QOL measurement must be sensitive to changes that occur throughout development [POG; unpublished manuscript. 1993]. *Afiiliulimn (fun/Inn's: A. S. Bradlyn. Department of Behavioral Medicine and Psychiatry. Robert C. Byrd Health Sciences Center. West Virginia Univer~ sity. Morgantown: B. H. Pollock. Department of Health Policy and Epidemiol- ogy. University of Florida College of Medicine. Pediatric Oncology Group Statistical Office. Gainesville. Carla's/nint/rm'e In: Andrew S. Bradlyn. Ph.D.. Department of Behavioral Medicine and Psychiatry. Robert C. Byrd Health Sciences Center. West Virginia University. 930 Chestnut Ridge Rd. Morgantown. WV 26505. 49 This definition is consistent with those of other investigators (9,10) and was intended to provide a focus for our research ef- forts. so that they would be planned. coordinated. and respon- sive to the rigors of the scientific method. It highlights the unique importance of the role of child development in our patients and their response to treatment. the impact of treatment on development, and the impact on family members. Prioritization of Clinical Trials for QOL Assessment Because of the drain on limited resources imposed by the col- lection of these types of data. a number of factors were specified to aid in the identification of clinical trials in which QOL end points would be most relevant. This serves the purpose of prioritizing which trials would include an assessment of QOL. Consistent with the results of other investigators (1/). ran— domized phase III trials are identified as being the most relevant setting for inclusion. While phase I and II trials are noted to have incorporated potentially relevant QOL questions. single- ann trials and those without randomization are considered to be problematic from a QOL research standpoint. Trials that are ex— pected to accrue a sufficient number of patients for statistical power considerations are also identified as potentially promis- ing. The guidelines note that QOL end points should be con- sidered in trials of therapeutic equivalence. in comparative investigations of new treatment modalities (or comparisons of alternate treatment modalities). in those protocols with a suppor- tive care or rehabilitation focus and in those that are likely to have a major impact on clinical practice. Instrumentation One of the critical responsibilities of the committee has been the identification of instruments that are consistent with our con- ceptual definition of QOL and protocol-specific research goals. This has been a dynamic process where the performance of in- struments is reviewed on a regular basis: newly developed and published instruments are also critically reviewed for possible inclusion. A guiding principle of the committee is that QOL measurement must reflect current scientific standards for ques- tionnaire development, psychometric properties, and practical use. We elected to adopt a core set of QOL measures. including generic and cancer-specific measures. Investigators were free to include additional. specific modules (e.g.. treatment or disease specific) to these core instruments for new protocols. We felt that recommending a consistent. core set of instruments would provide for the greatest degree of comparability of data across protocols. would increase quality control in a multi-institutional setting by reducing the number of different data forms. and would also enable us to continually update our knowledge and understanding of both the instruments and children‘s QOL. Un— fortunately. the state of the art in measuring child and adolescent QOL is significantly behind that of adult QOL. where there are numerous instruments for both generic (e.g.. SF—36; [2) and cancer specific (e.g.. Functional Assessment of Cancer Therapy; 13) assessment. Based on our review of the literature. we recommended the following core instruments: 1. Generic health SIafllS.’ The instruments derived from the Rand Health Insurance Experiment (14.15) were identified as 50 being the most appropriate. currently available instruments for our patient population. A major logistical barrier to the inclusion of QOL end points has been the age range that our patients rep— resent (from birth to over 21 years of age for most malignancies. and up to 30 years of age for bone tumor protocols) and the lack of a single instrument that could cover that great an age span. It was recognized early in our discussions that we would need to be able to collect data across more than one developmental stage (e.g.. 0-4 and 5-13 years of age) and that some degree of con— ceptual consistency was important in our choice of instruments. The Rand scales (0-4. 5-13. and 14 years of age and above) were therefore recommended because of their multidimensional focus. sound development and standardization. and desirable psychometric properties. Other scales. such as the Quality of Well-Being Scale (16). were considered but not recommended because of the nature of their administration (most typically ad- ministered by a trainer interviewer) and the associated cost of those procedures in a multi-institutional setting. 2. Cumm'—.rpec1_'fic' instrzmmzts: One frequently noted limita- tion of generic health status instruments is the potential lack of sensitivity to issues that might be of great importance in a par— ticular population (17). Cancer-specific instruments are recom- mended for inclusion to maximize the probability that the most relevant information will be obtained. Unfortunately. there is a paucity of such instruments currently available. but we have focused on two: the Pediatric Oncology Quality of Life Scale (9) and the University of Miami Quality of Life Scale (Armstrong FD; unpublished manuscript. I995). both of which were developed and standardized with pediatric oncology patients and their families. Both of these instruments are com— pleted by parents. and investigators are directed to choose one for inclusion in the study. 3. Other recommended core u.\'.ws.\'n1wzt.s‘.' A variety of other. brief instruments are recommended in the core battery. These include the Play Perfomiance Scale for children (18). which is a 0-100 parents-completed rating of their child's activity over the past 2 weeks (similar in form to the Karnofsky Perfonnance Status scale for adults). and two single—item ratings. The first of these ratings is one of overall QOL (from the parents’ or patient‘s perspective), and the second asks for a rating of overall health. These ratings are included to provide an additional. categorical evaluation of the patients that may be used in sub- sequent analyses. For adult survivors. the Karnofsky Perfor- mance Status Scale (19) is substituted for the Play Perfonnance scale. A number of measurement issues cloud the assessment of QOL for pediatric patients. As mentioned previously. it is not unusual for therapeutic trials to cover a wide age range (birth to 2| years) and to include a substantial follow—up period while on study. which may range up to 5 years. To highlight the many decision points. consider a child who initially registers in a protocol at age 4 years and is then followed for 5 years. In terms of QOL assessment. the parents would complete the ()—4-year- old version of the Rand scales and the single-item ratings. How- ever. when the child is reassessed at age 6 years and beyond. should the parents complete the 0—4 Rand or the more age—ap— propriate version (5-13 years)? While there is item content over— lap between the 0-4 and 5— l 3 versions. the earlier version has an Journal of the National Cancer Institute Monographs No. 20. 1996 emphasis on acquisition of developmental milestones. while the scale for older children includes mental health items and school and social functioning as well. Also consider the situation in which the child is placed in the study at age 13 years: his or her parents complete the 5—13-year-old Rand and the other recom- mended core instruments. However. at age 14 years. the patient could potentially serve as his or her own informant and com- plete the 14 and above version of the Rand. A change in instru- ments at this point would introduce two potential confounds: instrument (5-13 versus 14 Rand) and informant (parent versus patient). Generally speaking. our guidelines recommend that situations such as these be dealt with by having the same instrument com- pleted at each data collection point. For all intents and purposes then. this means that whatever instrument was completed at registration should be completed at each subsequent assessment. recognizing that there will be a group of patients who are “out of bounds" for the instrument. However. given the emphasis on measuring change over time. maintaining the consistency of the instrument is a higher priority. An additional and troubling issue is the use of proxy respon- dents (i.e.. parents) in pediatric trials. There is widespread ac— ceptance of the patient as the best provider of QOL data (20). but pediatric patients provide for many exceptions to this maxim. As noted above. these trials may include patients not judged competent to provide this information as a function of their age (e.g.. 3—year—olds). The lower age at which children can competently provide these types of data is not well estab- lished. but there is a recent report of pediatric cancer patients as young as 5 years of age being able to do so in a reliable and valid manner (Kaplan S. Barlow S. Spetter D. Sullivan L. Khan A. Parsons S. et al: unpublished manuscript. I995). Proxy respondents are also necessary in those situations for which child or adolescent self-report measures are not available. For example. at this time there are no published self-report cancer— specific instruments for school-aged children and adolescents or for the generic health status assessment of younger. school—aged patients. The agreement between patient and proxy measures in this population is not well understood and may. in fact. be quite poor (Kaplan et al.; unpublished manuscript. 1995). The POG QOL guidelines recommend that to the extent that it is feasible. the patient‘s perspective should be included. Toxicity measures. in and of themselves. are not appropriate as proxy measures. since they include little if any assessment of the impact of those toxic effects on the patient's functioning. Other Recommendations Included in the Guidelines There are several other issues that merit brief dicussion. At the time the guidelines were initially developed. it was typical for QOL questions to be addressed in a companion protocol to a therapeutic protocol. At that same time. however. there was dis- cussion about the difficulties arising from such a strategy. the primary one being the situation where only a small percentage of patients were registered on the QOL study in comparison to those actually receiving treatment. This introduced a number of difficulties into the interpretation of the data. not the least of which was potential selection bias. Our guidelines now recom— mend that QOL questions be included in the therapeutic Journal of the National Cancer Institute Monographs No. 20. I996 protocol and that they be labeled as specific cancer control ob- jectives. This has the benefit of allowing for joint review by both the treatment (Cancer Therapy Evaluation Program) and cancer control (Division of Cancer Prevention and Control) divisions of the National Cancer Institute. with assignment of cancer control credits (that provide financial support for institu- tional data management). Protocols with a QOL component should also have an in- dividual identified as the coordinator for those particular re- search questions. This person is responsible for the design and implementation of the data collection strategy. as well as the on- going monitoring of the data quality. Each participating institu- tion has also been asked to identify an individual on site who will be responsible for ensuring that patients complete the re- quired QOL assessments at the specified times and in the specified manner. A manual was recently developed and dis- tributed to these individuals to maximize the quality and quan- tity of data collection. Protocols With QOL End Points A number of phase III trials that are currently accruing patients or are expected to begin accrual in the near future in- clude QOL end points. Several earlier trials included a modified or shortened version of our core battery. most typically bundled with a series of psychologic or neuropsychologic evaluations. Protocols that include the core battery. as recommended by POG include: 1. P00 9485/9585: “Evaluation of the Role of Minimal Ac- cess Surgery in the Treatment of Childhood C ant-ers." For this pair of intergroup studies (with the Children‘s Cancer Group). the objectives are to evaluate the role of minimal access surgery versus standard open approach surgery in the diagnosis. staging. and treatment of a wide range of tumors. The primary end points are treatment—related morbidity and mortality. QOL. and economic costs. QOL assessments are scheduled at base line (registration), postsurgical day 3. and postsurgical day 30. al- lowing for examination of the differential effects of these surgi- cal procedures on short-term or acute QOL. 2. P00 9421: “Evaluation of Standard vs. High-Dose Ara-C Induction Followed by the Randomized Use of C yclosporine as an MDR Reversal Agent, Compared to Allogeneic' EMT in Childhood AML (Phase [1])." The objective of this study relat- ing to QOL is to compare the event-free survival between patients randomized between allogeneic bone marrow transplan— tation and chemotherapy. Serial QOL assessments and neuro- psychologic evaluations are scheduled to identify the effects of the two treatments on the patient’s QOL. This study is projected to accrue 150 patients per year. for each of4 years. 3. P06 9315: “A Phase III Study of Large—C ell Lymphomas in Children and Adolescents: Comparison ofC_\'turabine. Predv nisone. and Vincristine Versus C _\'tarabine. Prednisone. and Vineristine Plus Methotrexute, H_\‘dro.\'.\'ureu. and Cytarabine and Continuous Versus Bolus Infusion of Dororubicin." The primary objective of this investigation is to study whether inter- mediate—dose methotrexate and high—dose cytarabine ad— ministered during the maintenance phase can improve the event—free survival of patients with advanced-stage large cell 51 lymphoma. The objective relating directly to QOL focuses on the impact of doxorubicin infusion time (continuous versus bolus) on subsequent efficacy, cardiotoxicity. and QOL. This protocol is projected to accrue approximately 240 patients over a 5-year period. 4. P00 9480: “After/(Md Reduction for Late Anthracyc/inc C unlintarit'ity." The primary objective of this placebo-controlled. double-blind. randomized intervention trial is to investigate the role of enalapril in ameliorating the late cardiotoxic effects of doxorubicin for long-term survivors of childhood cancer. One of the specific objectives is to determine the impact of enalapril therapy on QOL. Additionally. we will compare. using the Q- TWiST method (quality-adjusted time without symptoms and toxicity). the tradeoff between treatment impact on QOL and cardiac functioning. This study is projected to accrue a total of 150 subjects who will be followed for a period of 4 years. Future Goals and Directions As part of our overall cancer control objectives and activities. we have come a long way toward including alternate end points in therapeutic protocols. However. as in any other developing area of research. we anticipate significant advances in our knowledge over the next several years. As interest in QOL as- sessment with children has burgeoned. new instruments are like— ly to become available and we will have gained further information regarding the performance of instruments in our core battery. Thus. the core set of instruments that is currently recommended is likely to change as more experience is gained in the group. Our QOL efforts include providing ongoing education and consultation to the cooperative group membership in general. We have planned a variety of educational activities. including roundtable discussions at upcoming group meetings and con— tinuing presentations and discussions in disease committees as new protocols are developed. It is a primary goal that protocols with relevant QOL questions be identified in the conceptualiza— tion and design phases so that these end points can be well in- tegrated in protocols from the beginning and not included as a post hoc consideration. Early incorporation allows for the op- timal integration of the QOL or cancer control objectives in the protocol and provides pivotal opportunities for full discussion regarding the hypothesized effects on QOL and how and when they may best be measured. It is important to note that in many of our protocols, QOL end points do not stand alone and may be incorporated into a larger package of alternate end points. including factors such as economic analyses and conventional end points such as survival. Protocols with multiple dependent variables. such as survival. QOL. and cost. will generate datasets that provide the oppor- tunity for exploration of the interrelationships among these end points. their relationship to independent variables. and the roles that they play in ultimate patient outcome. Additionally. as described in the enalapril preventive investigation. we are ex- ploring the applicability of utility assessments. such as the Q- TWiST methodology. as a means to integrate quantity and quality of survival. 'JI to A final. and ultimately critical. goal is to understand how these data should be interpreted and how they will be integrated into clinical decision making. The decision-making process is straightforward in situations where survival rates between two groups are equal but there are differences in QOL or where the survival and QOL of patients in one group are both superior to that in the other group. The most problematic situation. how- ever. is when one therapeutic arm has poorer survival but better QOL or when QOL improves but quality deteriorates. Investigation of QOL issues in pediatric oncology (as com- pared to adult oncology) presents a number of unique situations in terms of integrating these data into clinical decision making. For example, we have previously noted (21) that inclusion of QOL end points in clinical trials has been hampered by the at— titude that therapy for pediatric patients is curative in intent. while palliation may more often be the focus in adult clinical tri- als. Thus. there may be a willingness to trade quality of life for quantity of life in certain pediatric settings. How “treatment in- tent" should or can be integrated into decision making is un- known at this time and is an area for future investigation. Additionally. given the relatively positive prognosis for many pediatric malignancies. investigators must also examine the long-term or late effects of treatment and disease on QOL. While there are numerous reports of late effects on organ functioning. few if any have examined the actual impact of treatment on functioning. QOL data have the potential to go beyond the description of toxicity or late effects to actually describe how those effects are related to the patient‘s day-to-day abilities in their social, emotional, and physical functioning. These are important issues that are only beginning to be ex- plored. Because of the relative rarity of pediatric cancers. QOL re- search with this population typically occurs in the context of a cooperative group. multi-institutional setting. Approximately 1.5 million adults in the United States are diagnosed with cancer each year. which is orders of magnitude larger than the ap- proximately 8000-10 000 children and adolescents diagnosed with cancer each year. However. the number of years of life saved is substantial in the pediatric population. and as the num- ber of survivors increases. the study of the quality of their sur- vival will come more to the forefront and become predominant in the development of new therapies. Multi-institutional re- search investigations are costly. but the potential payoff for society. particularly in regard to the contributions that these sur- vivors may make. is likely to be substantial. References (I) Sugarbaker PH. Barofsky l. Rosenberg SA. Gianola FJ. Quality of life as- sessment of patients in extremity sarcoma clinical trials. Surgery 1982; 9:17-23. Selby P, Robertson B. Measurement of quality of life in patients with can- cer. Cancer Surv 1987:62521—43. Kornblith AB. Anderson J. Cella DF. Tross S. Zuckerman E. Cherin E. et al. Hodgkin‘s disease survivors at increased risk for problems in psychoso— cial adaptation. The Cancer and Leukemia Group B. Cancer 199227022214— 24. (4) Kaasa S. Mastekaasa A. Lund E. Prognostic factors for patients with in- operable non—small cell lung cancer. limited disease. The importance of patients‘ subjective experience of disease and psychosocial well-being. Radiother Oncol 19893523542. 13 V (3 V Journal of the National Cancer Institute Monographs No. 20. 1996 (5 (6 V (7 (8 (9) (/0) (/1) (/2) (l3) Bradlyn AS. Ham's CV. Spieth LE. Quality of life assessment in pediatric oncol— ogy: a retrospective review of phase III reports. Social Science Med. In press. Pediatric Oncology Group. Progress against childhood cancer: the Pediatric Oncology Group experience. Pediatrics 1992;89:597-600. Boring CC. Squires TS. Tong T. Cancer statistics, 1991. CA Cancer J Clin 1991;41:19-36. World Health Organization The first ten years of the World Health Or- ganization. Geneva. 1958. Goodwin DA. Boggs SR. Graham—Pole J. Development and validation of the Pediatric Oncology Quality of Life Scale. Psycholog Assess l994:6:321-8. Mulhem RK. Ochs J. Armstrong DF. Horowitz ME. Friedman AG. Copeland D. et al. Assessment of quality of life among pediatric patients with cancer. Psycholog Assess 1989;12130-8. Ware JE. Sherboume CD. The MOS 36-item Short—Form Health Survey (SF— 36). 1. Conceptual framework and item selection. Med Care 1992;30:473—83. Celia DF. Tulsky DS. Gray G. Sarafian B. Linn E. Bonomi A. et al. The Functional Assessment of Cancer Therapy scale: development and valida- tion of the general measure. J Clin Oncol 1993;11:570—9. Nayfield SG. Hailey BJ. McCabe M. Report of the Workshop on Quality of Life Research in Cancer Clinical Trials. Bethesda. MD: National Cancer Institute. 1990. Journal of the National Cancer Institute Monographs No. 20. 1996 (/4) (/5) (/6) (/7) (/8) (/9) Eisen M. Donald C. Ware JE Jr. Brook R. Conceptualization and measure— ment of health for children in the health insurance study. Santa Monica. CA: Rand Publication R-2313 HEW. 1980. Brook RH. Ware JE Jr. Davies»Avery A. et al. Conceptualization and measurement of health for adults in the Health Insurance Study. Volume 8: Overview. Santa Monica. CA: Rand Publication R—l987/8 HEW. 1979. Kaplan RB. Anderson JP. A General Health Policy Model: Update and ap— plications. Health Serv Res 1988;23:203—35. Patrick DL. Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. Med Care 1989:27. 5217—32. Lansky SB. List MA. Lansky LL. Ritter-Sterr C. Miller DR. The measure- ment of performance status in childhood cancer patients. Cancer 1987: 60: 165 1-6. Kamofsky D. Burchenal J. Clinical evaluation of chemotherapeutic agents in cancer. In: Macleod CM. editor. Evaluation of chemotherapeutic agents. New York: Columbia Press. 1949:191»205. Osoba D. Lessons learned from measuring health—related quality of life in oncology [see comment citation in Medlinel. J Clin Oncol 1994;12:608—16. Bradlyn AS. Ritchey AK. Harris CV. Moore AK. O’Brien RT. Parsons SK. et 211. Quality of life research in pediatric oncology: research methods and barriers. Cancer. In press. Model for Quality-of—Life Research From the Cancer and Leukemia Group B: the Telephone Interview, Conceptual Approach to Measurement, and Theoretical Framework Alice B. Kornblith, Jimmie C. H()lland* Cancer and Leukemia Group B Quality-of-Life Research Experience: 1976-1995 Historical Background: 1976-1985 In 1976. the Cancer and Leukemia Group B (CALGB). the oldest cancer cooperative clinical trials group. established the Psycho-Oncology (originally Psychiatry) Committee as a part of its multimodality structure. This created the first opportunity to ask psychological. social. and quality-of-life (QOL) questions in large patient groups in which medical variables were controlled and biases of geographic location. investigator. and treatment were reduced. Thus. the CALGB has provided a setting for more definitive testing of psychosocial questions relevant to cancer patients than could be achieved in studies conducted from single institutions. The Psycho-Oncology Committee began by assessing relevant psychosocial variables in specific clinical trials and later began to add QOL as one of the outcome variables as— sessed in clinical trials. A core of instruments was used to measure the impact of treatment on physical. psychological. and social functioning and to assess the role of psychosocial and demographic factors on survival (1-3). In the first 8 years. more than 1000 patients treated in eight CALGB protocols were studied with the core battery of measures. permitting com— parison of patients‘ psychosocial function and QOL across dis— ease site and treatment. This database provided the early studies of the impact of sociodemographic (education and income) (4) and psychological characteristics on survival. controlling for disease. treatment. and prognostic variables (5.6). Furthermore. it enabled the comparison of psychological distress of patients with advanced pancreatic cancer to those with advanced gastric cancer when receiving similar treatment protocols (7). the development of norms for the Profile of Mood States. a widely used measure of psychological distress. as well as the creation of a brief. valid version of the measure (8.9). and the examina- tion of the relationship of disease stage and performance status to psychological state in patients with lung cancer ([0). By the early l980s. the CALGB had accrued one of the largest cohorts of leukemia and Hodgkin‘s disease survivors who had been treated in CALGB phase III protocols. In 1985. the Psycho—Oncology Committee used this unusual resource to examine the long-term psychosocial adaptation of these sur— vivors. using a core of instruments. The first studies explored Journal of the National Cancer Institute Monographs No. 20. 1996 the negative impact of cranial radiation in childhood leukemia (ll). followed by examination of the QOL of adult-onset Hodgkin‘s disease survivors (12-15). leukemia survivors (16). and 15-year survivors of breast cancer (in progress). Over the past 5 years. there has been an explosion of interest by the cooperative clinical trials groups in assessing QOL. As a consequence of many clinical trials detecting only marginal dif- ferences in survival among treatment arms. QOL outcomes have assumed greater importance. In I985. partly in response to this issue. the Food and Drug Administration required that for new anticancer agents to gain approval for use. either a primary sur- vival gain or a secondary QOL benefit must be demonstrated (17). Concurrently. there has been a rapid increase in the num- ber. reliability. and validity of QOL measures with the “Hand— book of Quality of Life Measures" (l8). a compendium of QOL measures commonly employed in cancer research today. bearing testimony to this expansion. The enhanced interest in QOL research. coupled with strong. stable. collaborative ties between CALGB psycho—oncology and oncology investigators. led the group to address the need to im— prove QOL data collection procedures. A major limitation of QOL research in cooperative clinical trials groups has been that the responsibility for data collection was placed on busy data managers and research nurses. without designated time or addi- tional funding. Data often were not collected at the correct time. due to hectic clinic schedules. Data managers often did not have time to explain or instruct patients in the use of questionnaires. which frequently resulted in missing data and invalid scores. Patients. anxious prior to being seen by their oncologist or receiving treatment. provided ratings of their psychological state that may not have reflected their usual condition. resulting in skewed results; others were often reluctant to fill out question- naires at all. As a consequence of these problems. there were frequent missing data points. typically 50% at base line. with substantial further attrition over the course of the study. raising serious concerns as to the representativeness of the sample. *Afiiliariun of authors: Memorial Sloan-Kettering Cancer Center. New York. NY. Corresprmdcm"c to: Alice B. Komblith. Ph.D.. Psychiatry Service. Box 266. Memorial Sloan-Kettering Cancer Center. 1275 York Ave. New York. NY IUOZI. See “Note" section following “References." 'Jl ’Jl validity. and generalizability of the findings. In 1985. in an ef— fort to deal with these plaguing problems. the Psycho—Oncolog Committee changed the data collection method to interviewing patients at home by telephone. Data Collection by Telephone Interview The decision to collect data by telephone interview was based in part on research over the past 25 years that had demonstrated the efficacy of telephone interviewing for the collection of psychosocial data. Because telephones are present in over 95% of American households (19). a largely representative sample could be captured using this method of data collection. For spe— cial segments of the population who could not be interviewed by telephone either because of lack of access to a phone or hearing problems. as is more likely with the socioeconomically disad- vantaged and the elderly. respectively (21)). a mixed-mode method of data collection (2122) could be applied. using mailed questionnaires in place of telephone interviewing. Studies have shown that there are few substantive differences between results obtained from telephone and in-person interviewing. that the interrelationship among variables is maintained with both methods. and that the level of missing data is comparable (23— .3()). Further. less distortion in reporting socially undesirable acts occurs with the anonymity provided by the telephone interview (2529). Response biases that have been reported to occur more frequently in telephone than in in—person interviewing are greater use of extreme ends of response categories (e.g.. "never" and “extremely") and briefer responses to open—ended questions. For questions requesting ratings of agreement with given state— ments. there is a greater willingness to agree (i.e.. acquiesce) with telephone interviewing. regardless of the statements‘ con- tent (26.28). When costs incurred by the two methods have been analyzed. telephone interviewing has consistently been shown to be less expensive than in—person interviewing (1926.27.31.32). Hodgkin’s disease survivor study. In 1986. telephone inter— viewing was initiated in a study of survivors of advanced Hodgkin‘s disease (CALGB 8561 and 8562) (12—14). which was supported by the CALGB budget from the National Cancer In— stitute for a trained telephone interviewer. Procedurally. patients were sent a letter from the Principal Investigator of the institu— tion where they were treated that explained the study and in- formed them that a research interviewer would call within 1-2 weeks to discuss the study with them. Upon calling the patient. the interviewer answered questions concerning the study. ob- tained the patient‘s consent to participate. and made an appoint— ment for the telephone interview within 7-10 days. A packet was mailed to the patient that contained two copies of the consent form (one of which was signed and returned). along with a copy of the psychosocial measures. The patient was instructed to read through the material and answer as many of the questions as possible prior to the interview. At the time of the telephone in- terview. the interviewer clarified questions the patient had and entered his/her answers on an identical form. The telephone in- terview usually lasted 45—60 minutes. Of the 369 eligible patients. 273 (74%) survivors of Hodg— kin‘s disease were interviewed. This 74% participation rate was high and well within the range of 49%—82% reported for tele- phone interviewing (24.26—28,3(),32). The refusal rate was only 56 9%. near the lowest level in reported refusal rates of 4%-29% (26.28.31). An additional 15% were not interviewed because they were lost to follow-up and could not be located. Missing data were considerably diminished (12). The results of the CALGB study were compared with the study by Fobair et a1. (33) of early and advanced Hodgkin‘s disease survivors whose data—collection method had been by in—person interviews and self—report questionnaires. On equivalent questions concerning problems patients‘ attributed to cancer. findings were remark- ably similar between the CALGB (12) and Fobair et a1. (33) studies: decrease in sexual activity (21% versus 20%. respec— tively). divorce or separation (56% versus 49%. respectively). and loss ofjob (5% versus 6%. respectively). This provided fur- ther evidence of the comparability of data obtained by telephone and in-person interview (12). Importantly. it became evident that many patients enjoyed the interview. Some openly stated that the interview provided an opportunity to discuss a broad range of illness-related issues with an interested. caring individual. They found it gratifying that the CALGB had a continued interest in their well-being. While similar comments have been made by patients in active treatment in our other studies involving telephone interviews. they have been particularly frequent from cancer survivors who no longer maintain frequent contact with their oncologists. Expansion of the telephone interview method to other CALGB studies. The use of the telephone interview for QOL studies for patients during treatment was first tested in a QOL study of stage IV breast cancer patients in a phase III dose— response trial of megestrol acetate (CALGB 8864) (34). Patients were interviewed by telephone three times over a 3‘month period using a battery of measures significantly shortened from that of the Hodgkin’s disease survivor study to accommodate their limitations due to advanced—stage disease. Only 4% refused to participate after consent had been given. While there was at- trition over the 3-month period because of disease progression resulting in termination from the clinical trial (21%). sickness (3%). and death (2%). most patients who were able to par— ticipate were assessed successfully by telephone interview. The combined experience of the studies of Hodgkin's disease survi- vors and breast cancer patients on megestrol acetate has resulted in the telephone interview becoming the standard data collection method for CALGB QOL studies (35). Attrition due to illness demonstrated in the breast cancer megestrol acetate study underscores the limitations placed on QOL research in patients with advanced stages of disease using any self—report methodology. Our QOL study of cachectic patients in the terminal stage of illness. treated with megestrol acetate to increase their appetite and weight (CALGB 8971). met with this overriding limitation also. Experience suggests that obtaining self-report data from patients either in the ter— minal phase of their disease or with a poor performance status are highly limited and inappropriately intrusive. In these patients. behavioral observations of patient functioning made by a family member. caregiver. or health professional must suffice. Approach to Measurement Our measurement approach has been similar to that of Aaronson et al. (36). with major QOL dimensions assessed by a core of in- Journal of the National Cancer Institute Monographs No. 20. 1996 struments supplemented by measures specific to the disease site and treatment protocol. This approach has been applied to the development of two core sets of measures for two distinct patient populations: those in active treatment and cancer survivors. Most recently, individuals at high genetic risk for cancer have emerged as a new study population, requiring a third core set of measures to adequately assess relevant QOL issues (Table 1). Assessment of QOL of patients in active treatment. Be- cause treatment protocols address different sites of cancer at dif- ferent disease stages, assessing QOL of patients in active treatment requires a broad spectrum of measures. In the late l980s, the Functional Living Index-Cancer (FLIC) (37) was used in several breast and prostate cancer trials (CALGB 8864, 9181, and 9182) as the core QOL measure. However, the FLIC provided only an overall QOL total score, without subscale scores for different QOL dimensions. We therefore switched our core QOL measure for several lung cancer phase III trials [CALGB 8931 (38); CALGB 9033] to the EORTC Quality of Life Questionnaire (EORTC QLQ-C30) (39) and the Lung Can- cer Module (EORTC QLQ-LC13) (40) when they became avail- able. The EORTC measures were considered well suited for these studies because they had been originally developed in patients with lung cancer, had demonstrated reliability and validity on large samples, had subscale scores for the essential domains of QOL, and were brief, posing less respondent burden for very sick patients. To strengthen the social functioning com- ponent of the EORTC QLQ-C3O measure, the Duke—University of North Carolina Social Support Questionnaire (41) was ap- pended. Because of its demonstrated strengths, the EORTC QLQ-C30 has also served as the core measure for other CALGB studies involving patients with breast cancer (CALGB 9364), myelodysplastic syndrome (CALGB 9221), and pleural ef— fusions treated by talc thoracoscopy versus talc slurry (CALGB 9334). As new measures are developed, they are routinely reviewed for their potential use in our trials. For example. the Functional Assessment of Cancer Therapy Scale (FACT) (42), with its core QOL component supplemented by modules for nine disease sites and specific treatment regimens (e.g., bone marrow transplant), is being considered for use in studies cur- rently in development. For studies involving patients at an earlier stage of disease or with better performance status, expanded measurement has been possible, enabling assessment of additional variables or more in- depth measurement of important constructs. Because of the centrality of psychological state to understanding patients’ QOL. the Mental Health Inventory (MHI) (43.44) has been fre- quently used to obtain an assessment of both positive and nega- tive affect (CALGB 8864. 9221, and 9364). Particular QOL dimensions or related constructs have been assessed through the addition of the following measures: the McCorkle Symptom Distress Scale (45,46), to assess physical symptoms in several breast and prostate cancer clinical trials (CALGB 9066, 9181, and 9182); the Memorial Symptom Assessment Scale (47), for a colorectal cancer trial in development (CALGB 9481); the Wis- consin Brief Pain Inventory (48,49), for an extended assessment of pain in several prostate cancer clinical trials (CALGB 9181, Table 1. Measures used in selected CALGB studies Measure* MOS McCorkle EMP/ Rand Social Symptom Sexual INS Cond Socio Func Support Distress Additional Study No. 881 POMS IES problems problems N&V PAIS dcm FLIC MHI Limit EORTC Survey Scale measures Active treatment Breast 8864 X X X X X 9066 X X X X Lung 9033 X X X 9334 X X X Prostate 9l8| X X X X X 9182 X X X X X Myelodysplastic 9221 X X X X syndrome Other, cachexia 8971 X X X X X Survivors Hodgkin‘s 8561 and X X X X X X X X disease 8562 Leukemia 8963 X X X X X X X X X Breast X X X X X X X X X (proposed) Hodgkin‘s X X X X X X X disease telephone counseling (proposed) High»risk X X X screening (proposed) Special issues Bereavement 9364 X X X X X Hydrazine sulfate 8931 X X X *BSI = Brief Symptom Inventory: POMS = Profile of Mood States; IES = Impact of Event Scale; EMF/INS = Employment and Insurance problems; Cond N&V = conditioned nausea and vomiting; PAIS = Psychosocial Adaptation to Illness Scale; Socio dem = sociodemographic characteristics; FLIC = Functional Living Index— Cancer; MHI = Mental Health Inventory; and Rand Func Limit = Rand Functional Limitations Scale (modified). Journal of the National Cancer Institute Monographs No. 20, 1996 57 9182, and 9480); and the Body Image Subscale (50). to assess the effects of increasing weight on patients’ body image in the breast cancer megestrol acetate trial (CALGB 8864). Assessment of QOL of cancer survivors. A core set of measures can be applied as well in survivors of the commonality of issues across survivors of any neoplasm. In both the Hodgkin’s disease (CALGB 8561 and 8562) and adult leukemia (CALGB 8963) studies and the proposed breast cancer survivor study (to begin in the fall of 1995), the following measures con- stituted the core. The Psychosocial Adaptation to Illness Scale (PAIS) (51) was used to assess the impact of having had cancer on survivors’ current psychosocial and sexual functioning. Psychological state, both in general (Brief Symptom Inventory) (52) and specific to cancer (Impact of Event Scale) (53), was as— sessed in depth because of its central importance to adaptation. Measures were created to assess the continuing impact of cancer on survivors‘ lives, in terms of their employment, income, ob— taining health and life insurance, sexual problems, and condi- tioned nausea and vomiting in response to treatment-related stimuli (12,16). Because sterility was an important issue for many patients with Hodgkin’s disease who had been treated with alkylating agents, a detailed section of the questionnaire was devoted to assessing the prevalance of pregnancy outcomes, child deaths, and illnesses (15). By applying a core set of measures across survivor studies, as well as a set of disease- specific measures, there is an opportunity to compare the psychosocial adaptation of different groups of survivors and to speak meaningfully to issues that are important to each. Assessment of individuals at high genetic risk. With the in— creasing development of DNA testing, tumor markers. and other presymptomatic cancer-screening methods, QOL issues relevant to this set of individuals at high genetic risk have assumed greater importance. Our initial study, in development, will evaluate the psychological consequences of an intensified screening program for relatives of patients with colon cancer. Theoretical models guiding measurement will include the Health Belief Model (54,55), related models (56.57), and the transtheoretical model of change (58). Outcome variables and measures of importance are psychological state (e.g., Mental Health Inventory), particularly general anxiety and anxiety specific to high-risk individuals (e.g., Kash‘s Breast Cancer Anxiety Scale, modified for colon cancer) and adherence to screening recommendations. To test the influence of mediating factors on an individual’s adherence to screening recommenda- tions, as suggested by the theoretical models, the following vari— ables will be assessed: family history of cancer; history of compliance to colon cancer diagnostic testing; social support (e.g., MOS Social Support Survey) (59); and preventive health practices and health beliefs about cancer and screening, such as susceptibility to having cancer and potential costs and benefits of screening (e.g., General Health Motivation and Practices Scale) (60). The proposed study is the first in a series. as the CALGB develops a research program in the molecular genetics of solid tumors and hematopoietic malignancies. With the development of a CALGB registry of patients with breast cancer who consent to genetic testing, the psychological. ethical. and health behavior issues of genetic testing for both the patients and their relatives will become the focus of intensive study. 58 Special QOL research questions addressed by the CALGB. Some major psychological questions are ideally ad— dressed in the cooperative group setting because of the large numbers of patients treated in clinical trials in which treatment is by protocol, with medical, treatment, and treatment-related outcome data stored in a common database. One area in which cooperative clinical trials groups are particularly valuable is the testing of the efficacy and QOL impact of an alternative therapy in a rigorously controlled trial. The CALGB conducted a phase III clinical trial of hydrazine sulfate, in which all patients with advanced non—small-cell lung cancer were entered in a ran- domized study to receive either a standard chemotherapy regimen and hydrazine sulfate or the chemotherapy regimen and placebo in a double-blind fashion (CALGB 8931) (38). The EORTC QLQ-30 (39) and Social Support LCl 3 (40) were used as the QOL measures for the study, supplemented by the Duke— University of North Carolina Questionnaire (4/). No differences in survival or disease-free survival were found between the two groups. with the hydrazine arm ofthe study actually demonstrat- ing worse physical functioning, greater fatigue. and worse lung cancer—specific symptoms than the control group at the 2-month assessment (38). A second study currently under way. supported by the John D. and Catherine T. MacArthur Foundation. is quite different from others we have conducted in that it will test whether a major stressor, defined as loss of a spouse or child. is associated with a significantly increased risk of recurrence or death from breast cancer (CALGB 9364). The question of the relationship of stress to disease onset and recurrence is one that preoccupies many patient‘s concerns and is the focus of much research. How- ever. studies of smaller cohorts have resulted in contradictory findings (61). By strictly defining the stressor in terms of what is universally accepted as a major stress, the loss of a spouse or child, a more definitive answer to this question may be provided. As a secondary objective to this study, the role of so- cial support and spiritual beliefs in modulating the trauma of cancer will be examined. In this case—control study, in which all women were treated for stage II breast cancer over 10 years ago in CALGB 8541. case subjects were defined as women who had disease recurrence subsequent to their treatment in CALGB 8541 or who died; control subjects were those who were alive without disease progression. The odds of bereavement will be statistically compared between the two groups. with the odds ratio reflecting the increased risk of disease recurrence or death from breast cancer due to bereavement. The objectives led to the use of a psychosocial battery containing the MOS Social Sup- port Survey to measure social activities and emotional and instrumental support from family and friends (59); the Life Ex- periences Survey (62) to assess stressful life events: the Systems of Belief lnventory to assess spirituality (Holland and Kash: per- sonal communication): and the Texas Revised Inventory of Grief (63) to assess severity of bereavement for those who had experienced the death of a spouse or child. Theoretical framework. In the past 5 years, the stress—ill- ness vulnerability theory has emerged as a theoretical model for understanding patient adaptation (64). In this model, adapted for application to cancer patients (Fig. l) (65). cancer and its treat- ment are the stressors, and QOL or patient adaptation is the out- Journal of the National Cancer Institute Monographs No. 20, 1996 INDEPENDENT VARIAQLES lEDIATING VARIABLES OUTCOME SOCIAL SUPPORT Family Friends Comunlty MEDICAL CARE RELATIONSH PS ReIationship with HD/RN ADAPTATION PsychoIogical ECONOMIC RESOURCES Vocational Social INDIVIDUAL CHAR. Sexual Sociodemographic A Perceptions: Cancer. Health Personality Prior Adjustment CANCER 1 TREATN NT | | OTHER srnrssons | other Illnesses other Life Events I l | | Psvcuosocut INTERVENTION Fig. I. Vulnerability model of patients‘ adaptation to cancer. come. Mediating factors that may influence patient adaptation are included in the model, such as social support. relationship to the health care team, economic resources. personality charac— teristics, concurrent stressful life events. and comorbid condi— tions. In addition to assessing social support as a potential buffer serving to protect patients from the impact of stress. the suppor— tive role of the medical team to a patient’s adjustment. often overlooked by researchers. is highlighted in this model. Psychosocial interventions must be developed to affect these mediating variables or QOL itself. While no single study can incorporate the measurement of all mediating variables, the vulnerability model has served to guide the selection of variables and, consequently. instruments of im— portant mediating factors of adaptation in a number of our studies: family environment and health beliefs in long-term psychosocial adaptation of leukemia survivors (CALGB 8963) (16): preference for control over health care in a study of patient-controlled analgesia for patients with severe pain (CALGB 8872. in collaboration with the Cancer Control Com— mittee) (66); the role of stressful life events, social support. and spirituality in the psychosocial adaptation of breast cancer patients. as described above (CALGB 9364); and the patient‘s relationship with his or her medical team in a study of the treat- ment of fever and neutropenia at home by antibiotics (CALGB 9170, in collaboration with the Cancer Control Committee). By identifying critical factors that exacerbate patients‘ vulnerability to the stresses of cancer, as well as those that are protective, and examining the balance of these forces within the context of this model. wide differences in patient adaptation to the same dis— ease and treatment can be better understood. Journal of the National Cancer Institute Monographs No. 20, 1996 Guidelines for Future QOL Research Prioritizing Clinical Trials for QOL Study Cost containment today mandates prioritizing which studies receive a QOL component. In the CALGB, outside funding must be obtained for any study that exceeds the capacity of the single, National Cancer Institute-supported research interviewer, which is approximately two studies at a time, with several as- sessments during and off treatment. Given the volume of inter- est in QOL research, one interviewer has not been adequate to meet the demand. When there have been additional studies of interest. support of approximately $25 000 per year has been sought to cover the cost of telephone—derived QOL data collec- tion. Apportioning QOL studies by disease site or modality (i.e., one per committee) is one way to place a limit on QOL studies, but this does not take into account the possibility of several studies with important QOL issues becoming simultaneously ac- tive within a single committee. A mechanism must be estab- lished to prioritize QOL studies, involving both oncologists and psycho—oncologists within the cooperative clinical trial group, similar to the designation of high-priority clinical trials. How- ever. prioritizing QOL studies will not eliminate the need for financial support for this research. As of today, the CALGB remains the only cooperative group with a budget dedicated to QOL research. Although minimal. that budget has been pivotal in our research effort over the years. With the prioritization of QOL studies. in conjunction with the appropriate support. resources can then be allocated so that studies are conducted with the proper methodologic rigor and depth of measurement. Data Collection Method We consider the telephone interview as the QOL data collec- tion method of choice for cooperative clinical trials groups. QOL information is validly obtained via telephone interview. yields minimal missing data. results in excellent retention of patients for follow-up assessments. and has high patient satisfac- tion. The high rate of successfully completed interviews in the breast cancer megestrol acetate study (CALGB 8864) (34) was felt, in part. to be due to the rapport that developed between in- terviewer and patient over the course of the three assessments. Retaining patient compliance to repeated assessments in QOL studies could thus be enhanced as a consequence of the rapport between interviewer and patient. Computer-assisted telephone interviewing (CATI) (26.67) could be used to reduce some of the additional costs of telephone interviewing by increasing ef— ficiency in coding and data processing. without the visible presence of computer technology interfering with the interview process. The use of mailed questionnaires may be appealing at the outset because of ease of administration and lowest cost of all the data collection methods (2750—32). Self—administration of questionnaires in the clinic is similarly viewed favorably by those in the cooperative clinical trials groups for those reasons, although there are hidden costs that render this approach not so inexpensive. as Moinpour’s paper in this monograph suggests. Excellent completion rates using self-administered question- naires in the clinic have been reported by the Southwest Oncol- ogy Group (68) and the National Cancer Institute of Canada Clinical Trials Group (69). However. that certainly has not been 59 our experience using this method of data collection. which proved quite scientifically costly to the CALGB in its early studies. nor has it been the experience of other cooperative clini- cal trials groups. such as the European Organization for Re- search and Treatment of Cancer (70.71). While mixed—mode methods would certainly improve compliance by an average of 504-15% (2]). it is not clear if that would sufficiently offset the magnitude of the problem of missing data. particularly as patients‘ functioning deteriorates and clinic attendance becomes sporadic. Therefore. when the budget for a QOL study is being developed. a broader view of cost needs to be adopted that in— cludes our confidence in obtaining results on which the scien- tific community can build. Measurement There is a current wave of conservatism and demand for simplification in QOL measurement. with the suggested use of a single measure across all clinical trials. with perhaps a few addi- tional items specific to a protocol. This trend is due to multiple factors: I) frustration stemming from our current inability to compare results across trials due to variability in measurement (72). 2) a lack of understanding as to how to evaluate different measures for their appropriateness in measuring specific QOL issues in the different trials. and 3) an increasing limitation in financial resources devoted to QOL research. Indeed. the use of a core measure will allow for comparisons across clinical trials and provide much needed information conceming QOL issues for specific patient populations through the ongoing develop- ment of a database. However. the paramount consideration when selecting QOL measures for a clinical trial is that it serve as a valid test of the QOL research question. specific to that trial. Despite the clear benefits of using a common measure across trials. this issue can never supersede in importance the primary scientific mandate: answering the QOL question for that trial. As the oncology community‘s confidence in the availa— bility of valid QOL measures has increased. attention has begun to shift to understanding the clinical significance of statistically significant results. For many QOL measures. the clinical significance of findings is not readily apparent, nor are normative data available from either large community samples or relevant cancer patient populations to provide a frame of ref- erence for interpreting patients’ scores. The clinical significance of QOL scores will be determined as normative information is obtained for these measures and correlated with clinically well— understood. disease-relevant measures. such as the Kamofsky perfonnance status scale. psychiatric diagnoses. and behavioral indicators of psychosocial functioning (65). The cooperative clinical trials groups are the ideal context within which to con- duct this research. given the broad representation of patient populations and documentation of disease and treatment vari- ables. Cost Analysis The cost—conscious atmosphere in health care in the past 5 years has resulted in an increased interest in including cost analyses in relation to survival. toxicity. and/or QOL in the evaluation of cancer treatments (73-76). However. as of yet. it is 60 rare to see all four end points included in the research design. The CALGB’s newly created Clinical Economics Working Group. in collaboration with the Psycho-Oncology Committee. will select studies in which there are clear cost as well as QOL implications for different treatments. Our first effort in this area will be a study of the hepatic arterial infusion pump. in which colorectal cancer patients with hepatic metastases will be ran- domly assigned to receive chemotherapy either by a surgically implanted pump or systemic therapy (CALGB 9481). The sig- nificance of this model approach to treatment evaluation is that four major parameters are included: survival. toxicity. QOL. and cost. creating an enriched dataset from which to understand the impact of a cancer treatment on patients’ lives. Conceptual Issues Concerning QOL Research QOL research has not been theory driven. but rather. it has been guided by hypotheses as to which treatment arm will result in worse QOL. based on expected side effects. treatment ef- ficacy. or other treatment-related effects. This has been ap- propriate. given the nature of the research questions and the measurement limitations imposed by patients’ level of illness. However. by having a paucity of theoretical issues tested within the trials. little light has been thrown on identifying specific psychosocial mechanisms by which cancer patients adjust to their disease and treatment. Because planning rational interven- tions will require such information. theoretical models need to be considered in the development of QOL research in clinical trials. By concentrating resources in identified high-priority studies. expanded measurement is made more possible. enabling the testing of theoretically based questions. Conclusion The most significant clinical application of QOL research in phase Ill clinical trials will be to assist both patients and their oncologists in making treatment decisions by providing them with relevant information concerning a specific treatment’s im- pact on QOL. QOL issues are routinely taken into account in decision-making: oncologists. on the basis of their clinical ex— perience. and patients, on the basis of their judgment about potential efficacy versus expected side effects and disruption in function. When QOL research is conducted with the proper thought and methodologic rigor, the combined effect of obtain- ing QOL with survival. toxicity. and cost data in the evaluation of cancer treatments in clinical trials will be to guide treatment decisions from a more rational perspective. References (I) Holland JC. Silberfarb P. Tross S, Cella D. Psychosocial research in can- cer: the Cancer and Leukemia Group B [CALGB] experience. In: Ven- tafridda V. van Dam FS. Yancik R. Tamburini M. editors. Assessment of quality of life and cancer treatment. Amsterdam: Excerpta Medica 1986:8940]. Ahles TA. Silberfarb PM. Rundle AC. Holland JC. Komblith AB. Canel— los GP. et al. Quality of life in patients with limited small—cell carcinoma of the lung receiving chemotherapy with or without radiation therapy. for the Cancer and Leukemia Group B. Psychother Psychosom 1994;62:193-9. Silberfarb PM. Holland JC. Anbar D. Bahna G. Maurer LH. Chahinian AP. et al. Psychological response of patients receiving two drug regimens for lung carcinoma. Am J Psychiatry l983:l4():l 10-]. (2 (3 Journal of the National Cancer Institute Monographs No. 20. 1996 (4) (5 (6 (7 V (8 (9 V ([0) (ll) (/2) (/3) (l4 (/5) (/6) (I7) (/8) (19) (20) (2]) (22) (23) (24) (25 (26) (27) (28) (29) (30) (31) Cella DF. Orav EJ. Komblith AB. Holland JC. Silberfarb PM. Lee KW. et al. Socioeconomic status and cancer survival. J Clin Oncol 1991;9z1500—9. Holland JC. Korzun AH. Tross S. Cella DF. Norton L. Wood W. Psychosocial factors and disease free survival (DFS) in stage 11 breast car. cinoma. Proc ASCO 1986252237. Silberfarb PM. Anderson KM. Rundle AC. Holland JC. Cooper MR. Mc— lntyre OR. Mood and clinical status in patients with multiple myeloma. J Clin Oncol 1991:9:2219—24. Holland JC. Korzun AH. Tross S. Silberfarb P. Perry M. Comis R. et a1. Comparative psychological disturbance in patients with pancreatic and gastric cancer. Am J Psychiatry 1986;143:982—6. Cella DF. Jacobsen PB. Orav EJ. Holland JC. Silberfarb PM. Rafla S. A Brief POMS measure of distress for cancer patients. J Chronic Dis 1987;40:939—42. Cella DF. Tross S. Orav EJ. Holland JC. Silberfarb PM. Rafla S. Mood states of patients after the diagnosis of cancer. J Psychosocial Oncol 1989; 7(1/2):45-54. Cella DF. Orofiamma B. Holland JC. Silberfarb PM. Tross S. Feldstein M. et al. The relationship of psychological distress. extent of disease. and per~ formance status in patients with lung cancer. Cancer 1987;6(121661-7. Rowland JH. Glidewell OJ. Sibley RF. Holland JC. Tull R. Bemtan A. et al. Effects of different forms of central nervous system prophylaxis on neuropsychological function in childhood leukemia. J Clin Oncol 19842: 1327-35. Komblith AB. Anderson J. Cella DF. Tross S. Zuckerman E. Cherin E. et a1. Quality of life assessment of Hodgkin‘s disease survivors: a model for cooperative clinical trials. Oncology 1990:4:93—1()1. Komblith AB. Anderson J. Cella DF. Tross S. Zuckerman E. Cherin E. et al. Hodgkin’s disease survivors at increased risk for problems in psychoso- cial adaptation. The Cancer and Leukemia Group B. Cancer 19927012214 24. Komblith AB, Anderson J. Cella DF. Tross S. Zuckerman E. Cherin E. et al. Comparison of psychosocial adaptation and sexual function of survivors of advanced Hodgkin‘s disease treated by MOPP. ABVD. or MOPP alter- nating with ABVD. Cancer 1992;70:2508—16. Janov A]. Anderson J. Cella DF. Zuckerman E. Kornblith AB. Holland JC. et al. Pregnancy outcome in survivors of advanced Hodgkin‘s disease. Cancer 1992;70:688-92. Greenberg DB. Hemdon JE. Komblith AB. Zuckerman E. Schiffer CA. Weiss RB. et al. Long»term psychosocial adaptation of survivors of acute leukemia. Proc ASCO 1995;14:508. Johnson JR. Temple R. Food and Drug Administration requirements for approval of new anticancer drugs. Cancer Treat Rep 1985:69zl 155—7. Komblith AB. Holland JC. Handbook of measures for psychological. so— cial and physical function in cancer. Volume I: Quality of life. New York (NY): Memorial Sloan—Kettering Cancer Center. 1994. Lavrakas PJ. Telephone survey methods: sampling. selection and super- vision. Newbury Park (CA): Sage. 1987. Herzog AR. Rodgers WL. Kulka RA. Interviewing older adults: a com- parison of telephone and face—to-face modalities. Public Opin Q 1983; 47:405-18. Herzog AR. Rodgers WL. The use of survey methods in research on older Americans. In: Wallace RB. Woolson RF. editors. The Epidemiologic Study of the Elderly. New York: Oxford Univ Press. 1992:60-90. Dillman DA. Mail and telephone surveys: the total design method. New York: John Wiley & Sons. 1978. Aneshensel CS. Frerichs RR. Clark VA. Yokopenic PA. Telephone versus in-person surveys of community health status. Am J Public Health 1982;72:1017-21. Aneshensel CS. Yokopenic PA. Tests for the comparability of a causal model of depression under two conditions of interviewing. J Pers Soc Psychology 1985;49:1337-48. Colombotos J. Personal versus telephone interviews: effect on responses. Public Health Rep 1969;84:773—232. Groves R, Kahn M. Surveys by telephone: a national comparison with per— sonal interviews. New York: Academic Press. 1979. Hochstim JR. A critical comparison of three strategies of collecting data from households. J Am Stat Assoc 1967;62:976~89. Jordan LA. Marcus AC. Reeder LG. Response styles in telephone and household interviewing: a field experiment. Public Opin Q 1980;44:210-22. Locander W. Sudman S. Bradbum N. An investigation of interview method. threat and response distortion. J Am Stat Assoc 197617 1 :269-74. Siemiatycki J. A comparison of mail. telephone. and home interview strategies for household health surveys. Am J Public Health 1979;69:238- 45. Warner JL. Berrnan JJ. Weyant JM. Ciarlo JA. Assessing mental health program effectiveness: a comparison of three client follow-up methods. Eval Rev 1983;7:635-58. Joumal of the National Cancer Institute Monographs No. 20. 1996 (32) (33) (34) (3 7) (38) (39) (40) (41) (43) (44) (45) (46) (47) (48) (49) (50) (5/) (52) Walker AH. Restuccia JD. Obtaining information on patient satisfaction with hospital care: mail versus telephone. Health Serv Res 1984;19:83-98. Fobair P. Hoppe RT. Bloom J. Cox R. Varghese A. Spiegel D. Psychoso- cial problems among survivors of Hodgkin‘s disease. J Clin Oncol 198614: 805-14. Komblith AB. Hollis DR. Zuckerman E. Lyss AP. Canellos GP. Cooper MR. et al. Effect of megestrol acetate on quality of life in a dose-response trial in women with advanced breast cancer. J Clin Oncol 1993; 1 12081-9. Holland JC. Komblith AB. Zuckerman E. A centralized model for quality of life (QOL) data collection by telephone interview in multicentcr clinical trials. Proc ASCO 1992: l 1:406. Aaronson NK. Bullinger M. Ahmedzai S. A modular approach to quality- of—life assessment in cancer clinical trials. Recent Results Cancer Res 1988;111:231-49. Schipper H. Clinch J. McMurray A. Levitt M. Measuring the quality of life of cancer patients: the Functional Living Index—Cancer: development and validation. J Clin Oncol 1984:22472-83. Kosty MP. Fleishman SB. Hemdon JE 2d. Coughlin K. Komblith AB. Scalzo A. et al. Cisplatin. vinblastine. and hydrazine sulfate itt advanced. non-small~cell lung cancer: a randomized placebo—controlled. double-blind phase 111 study of the Cancer and Leukemia Group B. J Clin Oncol 1994;12:1113-20. Aaronson NK. Ahmedzai S. Bergman B. Bullinger M. Cull A. Duez NJ. et al. The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of—life instrument for use in international clinical tri- als in oncology. J Natl Cancer Inst 1993;85:365-761. Bergman B. Aaronson NK. Ahmedzai S. Kaasa S. Sullivan M. The EORTC QLQ—LC13: a modular supplement to the EORTC Core Quality of Life Questionnaire (QLQ—C30) for use in lung cancer clinical trials. EurJ Cancer 1994;3()A:635-42. Broadhead WE. Gehlbach SH. de Gruy FV. Kaplan BH. The Duke»UNC Functional Social Support Questionnaire. Measurement of social support in family medicine patients. Med Care 1988;26:709-23. Cella DF. Tulsky DS. Gray G. Sarafian B. Linn E. Bonomi A. et al. The Functional Assessment of Cancer Therapy Scale: development and valida- tion of the general measure. J Clin Oncol 1993:112570—9. Brook RH. Ware JE Jr. Davies»Avery A. Stewart AL. Donald CA. Rogers WH. et a1. Conceptualization and measurement of health for adults in the Health Insurance Study: overview (Vol. VIII). Santa Monica (CA): Rand Corporation. October 1979. Veit CT. Ware JE Jr. The structure of psychological distress and well- being in general populations. J Consult Clin Psychol 1983;51:730—112. McCorkle R. Young K. Development of a symptom distress scale. Cancer Nurs1978;1:373—8. McCorkle R. Quint—Benoliel J. Symptom distress. current concems and mood disturbance after diagnosis of life—threatening disease. Soc Sci Med 1983;17:431-8. Portenoy RK. Thaler HT. Komblith AB. Lepore JM. Friedlandet'«Klar H. Kiyasu E. et al. The Memorial Symptom Assessment Scale: an instrument for the evaluation of symptom prevalence. Characteristics and distress. Eur JCancer1994;3(1A:1326-36. Cleeland CS. Syrjala KL. How to assess cancer pain. In: Turk D. Melzack R. editors. Pain assessment. New York: Guilford Press. l992:36()—87. Daut RL. Cleeland CS. Flanery RC. Development of the Wisconsin Brief Pain Questionnaire to assess pain in cancer and other diseases. Pain 1983;17:197-210. Derogatis LR. Melisaratos N. The DFSI: a multidimensional measure of sexual functioning] Sex Marital Ther 1979:5:244-81. Derogatis LR. The Psychosocial Adjustment to Illness Scale (PAIS). J Psychosom Res 1986;30:77-91. Derogatis LR. Melisaratos N. The Brief Symptom Inventory: an introduc- tory report. Psychol Med 1983;13:595—605. Horowitz M. Wilner N. Alvarez W. Impact of Event Scale: a measure of subjective stress. Psychosom Med 1979141 :209—18. Becker MH. Maiman LA. Sociobehavioral determinants of compliance with health and medical care recommendations. Med Care 1975:18(1):l()—24. Kirscht JP. The Health Belief Model and predictions of health actions. In: Gochman DS. editor. Health behavior: emerging research perspectives. New York: Plenum Press. 1988:27-41. Leventhal H. Diefenbach M. Leventhal EA. Illness cognition: used com- mon sense to understand treatment adherence and affect cognition interac- tions. Cognitive Ther Res 1992: 16: 14363. Weinstein ND. Testing four competing theories of health-protective be- havior. Health Psychol 1993;12:324-33. Prochaska JO. DiClemente CC. The transtheoretical approach: crossing traditional boundaries ofchange. Homewood (IL): Dorsey Press. 1984. Sherboume CD. Stewart AL. The MOS Social Support Survey. Soc Sci Med 1991;32:70544. 61 (60) (61) (62) (63) (64) (65) (66) (67) (68) (69) Halper MS. Winawer SJ. Brody RS. Andrews M. Roth D. Burton G. Issues of patient compliance. In: Winawer SJ. Schottenfeld DM. Sherlock P. editors. Colorectal cancer: prevention. epidemiology. and screening. New York: Raven Press. 1980:299-310. Fox BH. The role of psychological factors in cancer incidence and prog- nosis. Oncology l995:9:245-53. Sarason IG. Johnson JH. Siegel JM. Assessing the impact of life changes: development of the Life Experiences Survey. J Consult Clin Psychol 1978;46:932—46. Fashingbauer TR. Zisook S. DeVaul R. The Texas Revised Inventory of Grief. In: Zisook S. editor. Biopsychosocial Aspects of Bereavement. Washington. DC: American Psychiatric Press. l987:l l l-24. Dohrenwend BS. Dohrenwend BP. Life stress and illness: formulation of the issues. In: Dohrenwend BS. Dohrenwend BP. editors. Stressful life events and their contexts. New York: Prodist. I98 I : I -27. Komblith AB. Thaler HT. Wong G. Vlamis V. McCarthy»Lepore J. Loseth DB. et al. Quality of life of women with ovarian cancer. Gyn Oncol. In press. Citron M. Conaway M. Zhukovsky D. Komblith AB. Berkowitz I. Pascal] V, et al. Efficacy of patient—controlled analgesia (PCA) vs. continuous in- travenous morphine (CIVM) for the treatment of severe cancer pain: CALGB 8872. Proc ASCO 1993;12:433. Patrick DL. Erickson P. Health status and health policy: allocating resour— ces to health care. New York: Oxford Univ Press. I993. Hayden KA. Moinpour CM. Metch B. Feigl P. O‘Bryan RM. Green S. et al. Pitfalls in quality—of-life assessment: lessons from a Southwest Oncol- ogy Group breast cancer clinical trial. Oncol Nurs Forum 1993120: 1415-9. Sadura A. Pater J. Osoba D. Levine M. Palmer M. Bennett K. Quality—of- lil‘e assessment: patient compliance with questionnaire completion. J Natl Cancer Inst 1992;84:1023-6. (70) (7]) (72) (73) (74) (75) (70) da Silva FC. Quality of life in prostatic cancer patients. Cancer 1993; 72:3803-6. Aaronson NK. Quality of life research in cancer clinical trials: a need for common rules and language. Oncology l990:4(5):59-6o. Gill TM. Feinstein AR. A critical appraisal of the quality-of-life measure- ments. JAMA 1994;272:619-26. Feeny D. Labelle R. Torrance GW. Integrating economic evaluations and quality of life assessments. In: Spilker B. editor. Quality of life assess- ments in clinical trials. New York: Raven Press. 1990:71-83. Goodwin PJ. Feld R. Evans WK. Pater J. Cost~effectiveness of cancer chemotherapy: an economic evaluation of a randomized trial in small—cell lung cancer. J Clin Oncol 1988;10:1537—47. Bennett CL. Armitage JL. Armitage GO. Vose JM. Bierman PJ. Armitage JO. et al. Costs of care and outcomes for high-dose therapy and autologous transplantation for lymphoid malignancies: results from the University of Nebraska l987 through l99l.JClin Oncol 1995;13:969—73. Weeks JC. Tierney MR. Weinstein MC. Cost-effectiveness of prophylactic intravenous immune globulin in chronic lymphocytic leukemia. N Engl J Med 1991;325:81-6. Note Supported by Public Health Service grant CA31946 from the National Cancer Institute. National Institutes of Health. Department of Health and Human Ser» vices; the T. J. Martell Foundation: the John D. and Catherine T. MacArthur Foundation: the Bristol—Myers Phannaceutical Company; and the Chemotherapy Foundation. Journal of the National Cancer Institute Monographs No. 20. I996 A Cooperative Group Report on Quality-of—Life Research: Lessons Learned Mary S. Mc'Cab€* Introduction This paper is a summary of the presentations given by repre- sentatives of the National Cancer Institute Cooperative Groups at the March 1-2. 1995. meeting. “Workshop on Quality of Life in Clinical Cancer Trials." The individual sections convey the diverse interests. unique patient populations. modality focus. and overall thoughtful approaches that each of the cooperative groups brings to oncology quality-of—life (QOL) research (Table I). This summary provides a unique opportunity to review the prob- lems. successes, and plans for the conduct of QOL research from a national perspective. The study—specific discussions are intended to provide clinically valuable information and to assist in the planning of future QOL evaluations that will advance the field and ultimate- ly answer questions that will benefit patients with cancer. *Cnrrespmidem'e to: Mary S. McCabe. R.N., National Institutes of Health. EPN. Rm. 715, Bethesda. MD 20892. Table 1. Active cooperative group treatment trials with QOL end points*.+ Protocol/ coordination center No. Title Phase QOL instrument Disease group—AIDS E1493 Sequential chemotherapy and radiotherapy for AIDS—related primary central 11 Functional Assessment of HIV nervous system lymphoma Infection. HIV QOL Survey, HIV Questionnaire Disease group—breast CALGB—9342 Study of Taxol (paclitaxel) at three dose levels in the treatment of patients 111 Functional Living Index—Cancer, with metastatic breast cancer Symptom Distress Scale El 193 Trial of doxorubicin versus paclitaxel versus paclitaxel plus adriamycin plus “I Functional Assessment of Cancer— granulocyte—colony stimulating factor in metastatic breast cancer Breast Cancer INT-()121/EST-2190 Study of conventional adjuvant chemotherapy versus high-dose chemotherapy Ill Breast Chemotherapy and autologous bone marrow transplant as questionnaire/adjuvant intensification therapy following conventional adjuvant chemotherapy in patients with stage II and Ill breast cancer at high risk of recurrence INT-()l42/E-3l93 Comparison oftamoxifen versus tamoxifen with ovarian ablation in III Functional Assessment of Cancer premenopausal women with axillary node-negative receptor-positive breast Therapy—Breast, ECOG cancer <2 cm Menopausal Symptom Form T90-0180/PBT-l Rrandomized comparison of maintenance chemotherapy with CTX, [11 Medical Outcomes Study Short Form- MTX, and S-FU versus high—dose chemotherapy with CTX. thiotepa. and 20, Symptom Distress Scale. Profile CBDCA and ABMT support for women with metastatic breast cancer of Mood States, Mental Adjustment responding to conventional induction chemotherapy to Cancer Scale Disease group—central nervous system INT-()l40/POG-933l Treatment of children with early-stage medulloblastoma: standard-dose III POG QOL Questionnaire craniospinal irradiation versus reduced-dose craniospinal irradiation plus adjuvant chemotherapy with cisplatin. cyclophosphamide. and vincristine lNT~0l49fRTOG~9402 lntergroup randomized comparison of radiation alone versus pre-radiation III Kamofsky Performance Scale. Mini— chemotherapy for pure and mixed anaplastic oligodendrogliomas Mental State Examination. EORTC—B QOL Questionnaire NCCTG-‘B-"llSZ Trial ofBCNU and cisplatin versus BCNU alone and standard 11] Mini-Mental State Examination, radiation therapy versus accelerated radiation therapy in patients with Neurologic Function Status high~grade glioma RTOG-93—(l5 Trial comparing the use of radiosurgery followed by conventional III Mini-Mental State Examination, radiotherapy with BCN U to conventional radiotherapy with BCNU for Spitzer QOL Index supratentorial glioblastoma inultiforme RTOG—94-l | Ttumor volume-influenced dose escalation of accelerated hyper— Il Mini—Mental State Examination fractionated radiotherapy to 64.0 and 70.4 Gy with BCNU for newly diagnosed radiosurgery—incligiblc glioblastoma multifonne patients Journal ofthc National Cancer Institute Monographs No. 20. I996 Table 1 (continued). Active cooperative group treatment trials with QOL end points*,T Protocol/ coordination center No Title Phase QOL instrument RTOG-94- l 7 Single—arm. open label. study of intravenously administered tirapazamine Il Mini-Mental State Examination plus radiation therapy for high-grade glioblastoma multiforrne Disease group—gastrointestinal INT-()l4o/NCCTG—93-46—53 Prospective randomized trial comparing laparoscopioassisted colectomy III Symptom Distress Scale, QOL-Index. versus open colectomy for colon cancer Q-TWiST INT—()l47/RTOG-9401 Intergroup randomized trial of preoperative versus postoperative combined [11 Anorectal Function Assessment Tool, modality therapy for resectable rectal cancer Functional Assessment of Therapy Cancer NCCTG—9 I —46—51 Salvage protocol for patients with advanced colorectal cancer who have I] QOL Uniscale relapsed following surgical adjuvant chemotherapy Disease group—genitourinary CALGB.9182 Randomized comparison of low-dose steroids and mitoxantrone versus low- Ill Functional Living Index—Cancer, dose steroids in patients with “hormone refractory" stage D2 carcinoma of Symptom Distress Scale. Sexual and the prostate Urologic Functioning Questionnaire, Problems in Daily Activities E7892 Randomized. double-blinded trial of adjuvant hormonal therapy for Ill Functional Assessment of Cancer surgically treated pathologic stage C carcinoma of the prostate Therapy—Prostate INT—()162/SWOG-9346 Intermittent androgen deprivation in patients with stage D2 prostate cancer 111 Medical Outcomes Study Short Form- 36. Medical Outcomes Study Short Form-20. Symptom Distress Scale. Linear Analogue Self Assessment RTOG-94—(l8 Phase Ill trial ofthe study of endocrine therapy used as a cytoreductive and III Sexual Adjustment Questionnaire cytostatic agent prior to radiation therapy in good prognosis, locally confined adenocarcinoma of the prostate T94—0l IO/NCIC—CTG lntergroup (NCIC CTG, and ECOG) phase III randomized trial comparing III EORTC Core Questionnaire—33. total androgen blockade versus total androgen blockade plus pelvic Functional Assessment of Cancer irradiation in clinical stage T3—4, N0. M0 adenocarcinoma of the prostate Therapy—Prostate Disease group—gynecologic E2E93 Clinical trial of an outpatient paclitaxel and carboplatin regimen in the II Functional Assessment of Cancer treatment of suboptimally debulked epithelial carcinoma of the ovary Therapy—Ovarian GOG-145 Randomized study of surgery versus surgery plus vulvar radiation in the III Functional Assessment of Cancer management of poor prognosis primary vulvar cancer and of radiation versus Therapy—General. GOG Symptom radiation and chemotherapy for positive inguinal nodes Inventory, Groningen Arousability Scale, Groningen Body Image Scale GOG-152 Randomized study ofcisplatin (NSC ll9875) and paclitaxel (NSC 125973) III Functional Assessment of Cancer with interval secondary cytoreduction versus cisplatin and paclitaxel in Therapy—Ovarian patients with suboptimal stage III and IV epithelial ovarian carcinoma GOG-9l02 Effect of alopecia on cancer patients‘ body image and the role of audiovisual Other Secourd and Jourard Body Cathexis infomtation on body image lndex GOG-LAPl Orientation and evaluation study of surgeon proficiency in performing a GOG Other Functional Assessment of Cancer standardized procedure for laparoscopic FIGO staging in adenocarcinoma Therapy—General, Medical of the endometrium Outcomes Survey—Physical Functioning Subscale. Wisconsin Brief Pain Inventory, Fear of relapse/ Recurrence Scale, Sexual Functioning Scale, Body Image SWOG-9324 Trial of vinorelbine tartrate (navelbine) for patients with relapsed ovarian II Medical Outcomes Study Short Form- cancer 36 Disease group—head and neck RTOG-90-03 Randomized study to compare twice daily hyperfractionation, accelerated III Functional Assessment of Cancer hyperfractionation with a split and accelerated fractionation with concomitant Therapy-«Head and Neck, List boost to standard fractionation radiotherapy for squamous cell carcinomas Performance Status Scale. Dische of head and neck Morbidity Scoring Tool RTOG—9l-l 1 Trial to preserve the larynx: induction chemotherapy and radiation therapy III Functional Assessment of Cancer— versus concomitant chemotherapy and radiation therapy versus radiation Head and Neck, Symptom Scale therapy 64 Journal of the National Cancer Institute Monographs No. 20, 1996 Table 1 (continued). Active cooperative group treatment trials with QOL end points".‘r Protocol/ coordination center No. Title Phase QOL instrument Disease group—leukemia CCG- 1941 Bone marrow transplantation versus prolonged intensive chemotherapy for [I] Ontario Health Survey children with acute lymphoblastic leukemia after an initial bone marrow relapse FOG-942] Evaluation of standard versus high—dose ARA—C induction followed by the III Bayley Scales of Infant Development, randomized use of cyclosporine A as an MDR reversal agent, compared Vineland Adaptive Behavior Scales. with allogeneic BMT. in childhood AML POG QOL Battery, Family Environment Scale. Wechsler Intelligence Scale for Children-III. Beery Test of Visual-Motor Integration, Achenbach CBC. Wide Range Achievement Test Disease group—lung E3592 Cisplatin plus etoposide versus daily oral etoposide in elderly patients with 11] Functional Assessment ofCancer extensive-stage small cell lung cancer Therapy—Lung, ECOG Neurotoxicity—Related QOL Questionnaire E4593 Study of hyperfractionated accelerated radiation therapy for advanced. 11 Functional Assessment of Cancer unresectable non-small-cell lung cancer with or without G—CSF Therapy—Lung E7593 Cisplatin plus etoposide versus cisplatin plus etoposide followed by topotecan Ill Funcational Assessment of Cancer in extensive-stage small cell lung cancer Therapy—Lung INT-013] Randomized study of CODE plus thoracic irradiation versus alternating lll EORTC Quality of Life Questionnaire CAV and EP for extensive stage small—cell lung cancer NCCTG-X9—2()~5] Study in extensive—disease small-cell lung cancer to evaluate the addition [[1 Functional Living IndexACancer of megestrol acetate to the etoposide/cisplatin regimen Disease group—lymphoma SWOG-9l33 Randomized trial of subtotal nodal irradiation versus doxorubicin, vinblastine lll CARES—SF, Symptom Distress Scale. and subtotal nodal irradiation for stage I-IIA Hodgkin‘s disease Medical Outcomes Study Short Fonn~36 swoc—9zos Health status and QOL in patients with early stage Hodgkin‘s disease Other Symptom Distress Scale, Cancer INT»() l 43/RTOG-93 l() CALGB—922] Disease group—multiple sites Intergroup phase 11 combined modality treatment of primary central nervous system lymphoma Disease group—myelodysplastic syndrome Randomized phase III controlled trial of subcutaneous 5—azacytidine (NSC #102816) versus observation in myelodysplastic syndromes IIl Rehabilitation Evaluation System— Short Form. Medical Outcomes Study Short Form—36 Mini—Mental State Examination EORTC QOL Questionnaire, Revised Rand General Well-Being Scale *Source: Cancer Therapy Evaluation Program database ofcooperative group trials. tCTX = cyclophosphamide; MTX = methotrexate: 5-FU = fiuorouracil; CBDCA = carboplatin; BCNU = carmustine; NCIC CTG = National Institute of Canada- Clinical Trials Group; ECOG = Eastern Cooperative Oncology Group; GOG = Gynecologic Oncology Group: ARA—C = cytarabine: MDR = multidrug resistance; BMT = bone marrow transplant: AML = acute myeloid leukemia; CAV : cyclophosphamide. doxorubicin. and vincristine; and EP = etoposide and cisplatin. Journal of the National Cancer Institute Monographs No. 20. 1996 65 Cancer and Leukemia Group B (CALGB) Alice 8. Kornblz'th Overview The past 5 years have been very active for the Psycho—Oncol- ogy Committee. Studies were conducted by or are currently ac— tive with five Disease Committees: 1) breast (CALGB-8082, CALGB-8864, CALGB-9066, CALGB-9342, and CALGB- 9364), 2) lung (CALGB-8931 and CALGB—9033). 3) lymphoma (CALGB-8561, CALGB—8562, and CALGB-9497), 4) leukemia (CALGB-8963 and CALGB-9221), and 5) prostate (CALGB- 9181 and CALGB—9l82). Three studies are currently in development with the Gastrointestinal Committee. Three major areas of research have been pursued since 1990: 1) quality of life (QOL) of patients on active treatment (CALGB-8083, CALGB-8534, CALGB—8864, CALGB-8872, CALGB-8931. CALGB-8971, CALGB-9033, CALGB—9066, CALGB-918l, CALGB—9l 82, CALGB-9221, and CALGB-9342), 2) psychoso- cial adaptation of leukemia and Hodgkin’s disease survivors (CALGB-8963, CALGB—8561, CALGB-8562, and CALGB- 9497), and 3) psychosocial and sociodemographic factors as predictors of survival (CALGB-7761, CALGB—8082, and CALGB—9364). Eight journal articles (1-8) and six abstracts (9- 13,16) have been published. New initiatives for the Psycho—On— cology Committee address the development of interventions: l) telephone counseling to improve patients” adjustment during ac- tive treatment or upon completing treatment, 2) improving doc— tor—patient communication, and 3) use of patient advocates to improve minority participation in clinical trials. Furthermore, we will be collaborating with the newly created Clinical Economics Working Group in selected studies in which there are clear cost as well as QOL implications of different treat— ments (e.g., hepatic arterial infusion protocol CALGB-9481). Last, with the increasing development of methods for genetic testing and other cancer-screening methods, QOL issues con- cerning individuals at high risk for cancer have assumed critical importance. The study of the psychological consequences of in- tensified screening of relatives at high risk for colon cancer will serve as a paradigm for this research area (9X6Q). Additional QOL research related to patients” participation in genetic re- search is currently being explored across the Psycho-Oncology, Cancer Control, and Oncology Nursing Committees. QOL During Active Treatment Active Protocols CALGB-9182—Randomized comparison of low-dose steroids and mitoxantrone versus low-dose steroids in patients with hormone refractory stage D2 carcinoma of the prostate. This study uses telephone interviewing as the method for data collection. The QOL measures in CALGB—9181 and Joumal of the National Cancer Institute Monographs No. 20, I996 CALGB-9182 are identical. QOL is being evaluated in both protocols by use of the Functional Living Index: Cancer Mc- Corkle Symptom Distress Scale, sexual and urological function- ing subscales of the EORTC (i.e., European Organization for Research and Treatment of Cancer) Prostate Questionnaire, Rand Functional Limitations Scale (modified), and the inter- ference of pain with daily functioning subscale of the Wisconsin Brief Pain Inventory. CALGB-9221—A randomized phase III controlled trial: subcutaneous 5-azacytidine versus observation in myelo- dysplastic syndromes. The QOL hypothesis in this study is that those randomly assigned to the 5-azacytidine arm will ex- perience a better QOL because of better symptom control (e.g., fewer hospitalizations, fewer infections, and decreased fatigue) than those in the control group. CALGB-9334—Sclerosis of pleural effusions by talc thoracoscopy versus talc slurry: a phase III study. This study will compare the QOL of patients with pleural effusions ran- domly assigned to receive talc slurry, administered at the bed- side, or talc thoracoscopy, conducted in the operating room. Patients’ QOL will be assessed using the EORTC QLQ-C3O Questionnaire, and items will be developed to assess patients’ satisfaction with these two procedures. In addition, pain will be assessed daily with the use of a visual analogue scale, until the chest tube is removed. CALGB-9342—Phase III study of paclitaxel (Taxol) at three dose levels in the treatment of patients with metastatic breast cancer. The objectives of the QOL component of this study are to examine the prognostic value of QOL scores at base line and to compare patients’ QOL on 175, 210, or 250 mg/m2 paclitaxel. QOL is assessed by the Functional Living Index— Cancer (FLIC) and the McCorkle Symptom Distress Scale. CALGB-9497—Health status and QOL in patients with early stage Hodgkin’s disease: a companion study to CALGB-9391/SWOG (i.e., Southwest Oncology Group) 9133. This intergroup trial under SWOG evaluates the QOL of patients with early stage Hodgkin’s disease over a 7-year period. Measures include the CARES—SF (Cancer Rehabilitation Evaluation System-Short Form), McCorkle’s Symptom Distress Scale, and the MOS SF-36 (Medical Outcome Study-36 Item Short Form Health Survey) Vitality and Health Perception Sub— scales. In Collaboration With Cancer Control Committee CALGB-9170—A multicenter trial of hospital versus early discharge therapy of low-risk patients with fever and neutropenia: a phase III study. This study will compare the QOL of patients with fever and neutropenia randomly assigned to continued hospitalization or early hospital discharge with 67 continued care at home with antibiotics. The research nurse as— sesses patients’ psychological state, satisfaction with medical care, and overall QOL at base line and at the end of treatment (generally within 5-7 days). CALGB-9490—Does an oral analgesic protocol improve pain control for patients with cancer? In this intergroup trial under Eastern Cooperative Oncology Group (E4Z93/CALGB- 9490), the efficacy of prescribing analgesic management of pain by protocol will be tested as a mechanism for improving pain control and the QOL of patients with metastatic or recurrent non—small-cell lung cancer, breast cancer, or multiple myeloma. Institutional sites will be randomly assigned to either the pain protocol or a standard care condition. Patients” pain, other physical symptoms, and emotional state will be assessed at base line and days 15 and 29, by use of the Brief Pain Inventory, Profile of Mood States (POMS), and the McCorkle Symptom Distress Scale. Closed Protocols CALGB-8083—Localized small-cell carcinoma of the lung: simultaneous chemotherapy and radiotherapy, chemotherapy versus sequential therapy (chemotherapy, radiotherapy, chemotherapy) versus chemotherapy alone. QOL and neuropsychological function of 57 patients with small- cell lung cancer were evaluated using the Trail-Making B Test (global indicator of cognitive impairment), POMS, and the Handicap Rating Scale, a physician-rated measure of five dimensions of psychosocial functioning. Patients receiving chemotherapy plus radiation therapy to both lung and brain had a significantly worse emotional state (POMS total score) and Handicap Rating Scale score at the beginning of cycle 4 of chemotherapy than those receiving chemotherapy plus pro— phylactic radiation therapy to the brain alone (P<.05). No sig- nificant differences in neuropsychological functioning were found between treatment arms (I). CALGB-8534—Combination chemotherapy with inten- sive ACE/PCE (doxorubicin, cyclophosphamide, and etopo- side/cisplatin, cyclophosphamide, and etoposide) and radi- ation therapy to the primary tumor and prophylactic whole- brain radiation therapy with or without warfarin in limited small-cell carcinoma of the lung: phase III. This study ac— crued 369 patients, and follow-up data collection has been com- pleted. No significant differences in psychological status and neuropsychological functioning (as measured by the POMS and Trail-Making B Test) were found by the end of cycle 5 (after radiotherapy) between the two treatment arms, indicating that warfarin had no significant effect on patients" QOL. CALGB-8864—Assessing QOL during a dose—response trial of megestrol acetate in patients with advanced breast cancer (companion to CALGB-8741). The QOL of patients with advanced breast cancer randomly assigned to receive three different doses of megestrol acetate was examined over a 3— month period. At 3 months, women treated with the lowest dose (160 mg/day) reported significantly less severe side effects (P<.0005) (including appetite increase, weight gain, fatigue, and feeling bloated), better physical functioning (P<.0005), and less psychological distress (P = .008) from study entry than those 68 treated on the highest dose (1600 mg/day). No differences in body image were found among the three dose groups (6). CALGB-8931—Cisplatin, vinblastine, and hydrazine sul- fate (NSC-150014) in treatment of advanced non—small-cell lung cancer: a randomized, placebo-controlled, double- blinded phase III study. Patients’ QOL was assessed while they were receiving either hydrazine sulfate or placebo, at 2— month intervals as long as they remained on study. Measures in- cluded the EORTC Quality of Life Questionnaire and the Duke-University of North Carolina Functional Social Support Questionnaire. At 2 months, patients receiving hydrazine sulfate had significantly worse physical symptoms and physical functioning than those receiving placebo; there were no other quality differences between the two arms (7). CALGB-9033—Oral versus intravenous etoposide in com- bination with intravenous cisplatin in extensive small-cell lung cancer. The question of interest in this study was whether oral administration of chemotherapy improved the QOL of patients with extensive small-cell lung cancer compared with in— travenous administration as a result of the ease of administration and potentially fewer side effects. Patients’ QOL was assessed by telephone interview. No significant differences in QOL were found between the two treatment arms over the 3-month time period, as measured by the EORTC Quality of Life Question- naire, the MOS Social Support Scale, and the CES-D (Center for Epidemiologic Studies—Depression Scale) Depression Scale. CALGB-9066—QOL and psychosocial adjustment of patients with stage II or III breast cancer randomly assigned to receive high-dose CPA/cDDP (cyclophosphamide/cis- platin)/carmustine with autologous bone marrow support versus standard-dose CPA/cDDP/carmustine as consolida- tion to adjuvant CAF (cyclophosphamide, doxorubicin, and fluorouracil) (companion to CALGB-9082). The study’s primary objective was to assess the QOL and psychosocial adaptation of stage II or III breast cancer patients randomly as- signed to receive either autologous bone marrow transplant or conventional chemotherapy. Patients were interviewed by telephone using a battery of measures: the PAIS (Psychosocial Adjustment to Illness Scale), FLIC, and McCorkle Symptom Distress Scale. Follow-up data collection will continue for the next 3 years. CALGB-9181—Randomized phase II study comparing standard-dose with moderately high-dose megestrol acetate in patients with advanced prostate cancer. The methodology used in CALGB-9181 involving telephone interviewing as the method for data collection and all QOL measures were identical to those used for CALGB-9182 (see CALGB-9182 above). Fol- low-up data collection is continuing. In Collaboration With the Cancer Control Committee CALGB-8872—Randomized study of patient-controlled analgesia versus continuous intravenous morphine for severe pain. The study’s objective was to compare the efficacy and impact on QOL of two forms of pain control: continuous in- travenous infusion of morphine (IV) versus patient-controlled analgesia (PCA). Pain and sedation as well as patients” psychological distress and preference for having personal con- trol over the administration of morphine were assessed. While Journal of the National Cancer Institute Monographs No. 20, 1996 the PCA group was found to have used significantly less mor- phine (P<.05) and reported significantly greater pain intensity (P<.05) than the IV group, there was an equivalent rating of pain relief in both arms. Furthermore, those on PCA reported significantly less psychological distress at day 5 than those in the 1V arm (P<.05). Those on PCA reported the least sedation and had the least distress of all subgroups, controlling for other sociodemographic and pain characteristics (9). CALGB-8971—A dose—response trial of megestrol acetate for the treatment of cachexia in patients with advanced Jung or colorectal cancer. The objective of this study was to evaluate the effect of three different levels of megestrol acetate on weight gain and QOL of cachectic patients. The QOL com— ponent of this study was closed prior to completion as a conse- quence of a 75% drop—off in patient assessment by the 1st month assessment due to illness, death, and interviewer error. Although severely compromised by attrition, data analysis revealed no significant differences in QOL due to dose level of megestrol acetate, for the entire sample or by disease site. Long-Term Psychosocial Adaptation of Cancer Survivors Closed Protocols CALGB-8561—Comparative assessment of psychosocial sequelae in long-term Hodgkin’s disease survivors. The ob- jective of this study was to examine the long-term psychosocial adaptation of 273 survivors of advanced Hodgkin’s disease who had been treated in any of nine CALGB clinical trials. Psychological distress was found to be elevated by one standard deviation above that of healthy respondents, as assessed by the Brief Symptom Inventory, with 22% reporting distress at a level requiring further psychiatric evaluation. Furthermore, a range of psychosocial “re—entry” problems was reported as a conse- quence of having had Hodgkin’s disease: denial of life insurance (31%) and health insurance (22%), sexual problems (37%), con- ditioned nausea in response to reminders of chemotherapy (39%), and a negative socioeconomic impact on their lives (36%) (3,4). This study established the value of the telephone interview as the method for QOL data collection in the coopera- tive clinical trials group and served as the foundation for the proposed telephone counseling study to improve adaptation in survivors upon completion of their oncology treatment (CALGB-9360). CALGB-8562—Comparative assessment of psychosocial and psychosexual sequelae in three treatment regimens for advanced Hodgkin’s disease (companion study to CALGB- 8251): a randomized phase III trial comparing MOPP (i.e., mechlorethamine + vincristine + procarbazine + pred- nisone), ABVD (i.e., doxorubicin + bleomycin + vincristine + dacarbazine), and MOPP alternating with ABVD in treat- ment of advanced Hodgkin’s disease. CALGB-8562 was a subset of CALGB-8561, involving 92 patients who had been treated for advanced Hodgkin’s disease in one of the nine clini- cal trials, CALGB-8251. This study was undertaken to deter— mine if there were significant differences in survivors’ long-term psychosocial and psychosexual function as a conse- quence of differential gonadal damage from the three regimens, Journal of the National Cancer Institute Monographs No. 20, 1996 MOPP versus ABVD versus MOPP/ABVD. No significant long—term advantage was found for survivors of Hodgkin’s dis— ease treated by the less gonadally toxic ABVD regimen (5). CALGB-8963—Psychosocial adaptation of survivors of acute leukemia. This study was developed to examine the psychosocial adjustment of survivors of acute leukemia and to identify factors predictive of current distress. Initial analyses in- dicate that 14% of leukemia survivors reported psychological distress that was at a level requiring further psychiatric evalua- tion, as measured by the Brief Symptom Inventory. Survivors most likely to have heightened distress were younger (P<.05) and were less educated (P<.002), had a history of conditioned anticipatory distress prior to their chemotherapy treatment (P<.05), and had a worse family environment in conjunction with more medical problems subsequent to completion of treat- ment(P<.05)(11). Psychosocial and Socioeconomic Factors as Predictors of Survival Active Protocols CALGB-9364—Effect of bereavement on disease recur- rence and death in women with stage 11 breast cancer (com- panion study to CALGB-8541). This companion study is designed to determine whether bereavement (defined as the loss of a spouse or child) in stage II breast cancer patients sub- sequent to adjuvant treatment on CALGB-8541 is associated with an increased risk of recurrence or death due to breast can- cer. A secondary objective is to examine the relationship of stressful life events to current psychological status, as mediated by sociodemographic, medical, and social support factors. With a case—control research design, case subjects include patients who have had disease recurrence or have died subsequent to treatment completion of CALGB-8541; control subjects are women who are alive without disease recurrence. Case and con— trol respondents will be matched for age, menopausal status, time of entry to CALGB-8541, lymph node status, and family status (living children versus no living children). Closed Protocols CALGB-7761—A study to determine the effectiveness of single versus multiple alkylating agents with or without doxorubicin in the primary treatment of multiple myeloma. Psychosocial status at protocol entry was examined as a predic- tor of survival. Multiple myeloma patients were administered two measures of psychological state at base line: the POMS and the Multiple Affective Adjective Checklist (MAACL). Neither POMS nor the MAACL was a significant predictor of survival (8). CALGB-8082—Surgical adjuvant chemotherapy for breast carcinoma: two CMFVP regimens (i.e., cyclophos- phamide + methotrexate + fluorouracil + vincristine + pred- nisone) with or without a subsequent doxorubicin combination. This study also examines psychosocial status at protocol entry as a predictor of survival. Women with stage 11 breast cancer were administered the SCL-9O (Symptom Check- list, a measure of psychological state) at base line. After control- 69 ling for known medical prognostic factors. the SCL-90 total score was not found to significantly predict survival at 7 years. Future Plans Interventions to Improve Patient Adaptation CALGB-9360—Psycho-educationaI/interpersonal counsel- ing intervention to improve “re-entry” adjustment of Hodgkin’s disease patients upon completing active treat- ment: a pilot study (companion study to CALGB-8952). The major objective of this proposed study is to evaluate the feasibility of conducting a psychosocial intervention by telephone. The intervention, consisting of education, counseling, and emotional support, improves Hodgkin‘s disease patients’ adjustment upon treatment completion. This intervention is based on Interpersonal Counseling developed by Klerman and colleagues (14,15), with an expanded psycho—educational com— ponent, and has been adapted for a cancer patient population. This study will be a companion to CALGB—8952, in which patients with advanced Hodgkin’s disease are randomly as— signed to receive either MOPP/ABV (i.e., doxorubicin + bleomycin + vinblastine) or ABVD. Patients who completed treatment on CALGB—8952 within the past 4 months will be eligible to participate. The intervention will consist of six telephone counseling sessions, conducted biweekly by an oncol- ogy nurse. Typical problematic areas identified by Hodgkin’s disease patients upon completing treatment will be discussed. Relevant educational materials will also be distributed. CALGB-9363—“ProtoCaIl”: a randomized trial of a telephone-based supportive/educational intervention to im- prove QOL, satisfaction with care, and compliance. This study will test the hypothesis that cancer patients, randomly as- signed to receive a supportive counseling intervention provided over the telephone by a research nurse during active treatment, experience an improvement in their psychological and social functioning and compliance to treatment, compared with a con— trol group not receiving the intervention. Interpersonal Counsel— ing, the therapeutic model upon which this intervention is based, was developed by Klerman and colleagues (14.15) and has been adapted for a cancer patient population. Patients will be assessed by use of standardized measures of QOL. QOL of Patients During Active Treatment CALGB-9480—A phase III study of three different doses of suramin administered with a fixed dose schedule in patients with advanced prostate cancer. A QOL component to a dose—response trial of suramin has been drafted. CALGB-9481—A phase III study of hepatic artery floxuridine, leucovorin, and dexamethasone versus systemic fluorouracil and leucovorin as treatment for hepatic metas- tases from colorectal cancer. The QOL component of this phase III trial has been developed in conjunction with the newly created Clinical Economics Working Group, which will conduct a cost analysis for this study. Study of sexual function in postmenopausal women treated with tamoxifen. A pilot study was conducted of 67 postmenopausal women with early stage breast cancer treated 70 by tamoxifen to examine the magnitude of tamoxifen’s effect on sexual functioning (16). Patients were assessed for drug side ef- fects, sexual functioning, and depressed mood by use of ques- tionnaires and vaginal and Pap smears. Data analysis began in March 1995. The next step concerning this line of research will be discussed after analysis of the pilot data. Long-Term Adaptation of Cancer Survivors In development—Long-term psychosocial adaptation in breast cancer survivors treated with adjuvant therapy (com- panion study to CALGB-7581). The long-term psychosocial adaptation of 200 breast cancer survivors treated 15-20 years ago with the adjuvant therapy on CALGB—7581 will be studied. Survivors will be interviewed concerning their current psycho- logical, social, sexual. and vocational functioning; breast cancer detection behaviors; and problems they attributed to having been treated for cancer. An identical battery of measures that we have used in our Hodgkin’s disease survivor studies (CALGB- 8561/8562) and acute leukemia survivor study (CALGB-8963). supplemented by appropriate measures for this patient popula tion, will be used. All patients will be interviewed by telephone. Any patient in significant distress will be further evaluated by a psychiatrist via telephone interview and referred for treatment in her community. Quality of Life of Those at High Risk for Cancer In collaboration with the Cancer Control Committee: in development—9X6Q psychological consequences of colorec- tal cancer screening in first-degree relatives of colorectal cancer patients (companion study to CALGB-9173). The proposed study will serve as a companion to the colorectal screening trial of first—degree relatives of colorectal cancer patients (CALGB-9l73). Colon cancer patients will be random— ly assigned to receive either a direct letter from the physician sent to the patients’ relatives concerning screening recommen- dations, or a flyer concerning screening recommendations given to the patient to be sent to their relatives. High-risk relatives will be interviewed by telephone concerning psychological distress subsequent to an evaluation of compliance to screening recom- mendations. References (I) Ahles TA, Silberfarb PM, Rundle AC, Holland JC, Kornblith AB, Cartel- los GP, et al. Quality of life in patients with limited small—cell carcinoma of the lung receiving chemotherapy with or without radiation therapy, for the Cancer and Leukemia Group B. Psychother Psychosom 1994;62:193—9. (2) Janov AJ. Anderson J. Cella DF, Zuckerman E, Komblith AB, Holland JC. et a]. Pregnancy outcome in survivors of advanced Hodgkin’s disease. Cancer 1992;70:688—92. (3) Kornblith AB, Anderson J. Cella DF. Tross S, Henderson ES, Weiss RB, et al. Quality of life assessment of Hodgkin's disease survivors: a model for cooperative clinical trials. Oncology 1990:4z93-IOI. Kornblith AB, Anderson J, Cella DF, Tross S, Zuckerman E. Cherin E. et al. Hodgkin‘s disease survivors at increased risk for problems in psychoso— cial adaptation. Cancer 1992;70:2214-24. Kornblith AB. Anderson J. Cella DF, Tross S, Zuckerman E, Cherin E, et al. Comparison ot‘psychosocial adaptation and sexual function of survivors of advanced Hodgkin‘s disease treated by MOPP. ABVD. or MOPP alter— nating with ABVD. Cancer 1992;70:2508—16. Komblith AB, Hollis D. Zuckerman E. Lyss AP, Canellos GP, Cooper MR, et al. Effect of tnegestrol acetate upon quality of life in a dose— (4 (5 ((5 Journal of the National Cancer Institute Monographs No. 20, I996 (7 (8) (9 (I0) response trial in women with advanced breast cancer. J Clin Oncol 1993: l 1:2081-9. Kosty MP. Fleishman SB. Herndon IE. Coughlin K. Komblith AB. Scalzo A. et al. Cisplatin. vinblastine. and hydrazine sulfate in advanced. non» small-lung cancer: a randomized placebo»controlled. double-blind phase III study of the Cancer and Leukemia Group B. J Clin Oncol 1994: [2:] 1 [3-20. Silberfarb PM. Anderson KM. Rundle AC. Holland JC. Cooper MR. McIntyre OR. Mood and clinical status in patients with multiple myeloma. JClin Oncol 1991:9z2219—24. Citron M. Conaway M. Zhukovsky D. Komblith AB. Berkowitz I. Pascall V. et al. Efficacy of patient—controlled analgesia (PCA) vs. continuous in— travenous morphine (CIVM) for the treatment of severe cancer pain: CALGB 8872. Proc ASCO 1993;122abstr 1494. Fleishman SB. Kosty M. Herndon J. Komblith AB. Duggan D. Morris J, et al. Quality of life (QOL) predicts survival in advanced non-small cell lung cancer. A Cancer and Leukemia Group B (CALGB) study. Proc ASCO 1994:13:ahstr I479. Journal of the National Cancer Institute Monographs No. 20. 1996 (ll) (12) ([3) (I4) (I5) (lo) Greenberg DB. Herndon JE. Komblith AB. Zuckerman E. Schiffer CA. Weiss RB. et al. Long-term psychosocial adaptation of survivors of acute leukemia. Proc ASCO 1995;142abstr 1668. Holland JC. Komblith AB. Zuckerman E. A centralized model for quality of life (QOL) data collection by telephone interview in multicenter clinical trials. Proc ASCO l992;11:abstr 1421. Holland IC. Herndon J. Komblith AB. Cella DF. Cooper MR. Green M. et al. A sociodemographic data collection model for cooperative clinical trials. Proc ASCO 1992:] lzabstr 445. Klerrnan GL, Budman S. Berwick D. Weissman MM, Damico—White J. Demby A. et al. Efficacy of a brief psychosocial intervention for symptoms of stress and distress among patients in primary care. Med Care 1987;25:1078-88. Klerman GL. Weissman MM. Interpersonal psychotherapy. In: Paykel ES. editor. Handbook of affective disorders. 2nd ed. London: Churchill Livingston. 1992:501—10. Mortimer JE. Knapp D. Fracasso PM. Rowland JH. Komblith AB. Assess— ment of sexual function in women on tamoxifen. Proc ASCO 1994; l3:abstr1554. 71 Eastern Cooperative Oncology Group (ECOG) Diane L. Fairclough, David F. Celia History The Outcomes Subcommittee (formerly Quality of Life Sub- committee) is one of five subcommittees of the Health Practices Committee of the ECOG. It was established in 1990 to oversee the scientific integrity of quality-of—life (QOL) research activity within the group. Its core membership is comprised of social scientists, physicians. nurses, and statisticians. Its initial role was to stimulate and promote high—quality QOL investigations in selected clinical trials. That role quickly shifted to include quality assurance of QOL data collection efforts and has more recently emphasized scientific prioritization of study proposals, because demand for QOL research within the group has out— stripped the resource availability. The future of the expanded Outcomes Subcommittee promises to be very exciting as the outcomes field matures scientifically. The following is a brief description of events and activities that explain the evolution of QOL activity and priorities within the ECOG. The first ECOG QOL study predates the formation of the QOL subcommittee. It was initiated as a pilot feasibility study for patients with metastatic non—small—cell lung cancer in 1983 (I). A separate QOL pilot protocol (E4983—Assessment of Quality of Life in ECOG Patients) was written to accompany the primary therapeutic study (E1583-phase II-III Chemo- therapy of Metastatic Non—Small-Cell Bronchogenic Car- cinoma) using the FLIC (Functional Living Index—Cancer). Compliance to the QOL assessments dropped rapidly to 33% of survivors by 6 months. Anecdotal reports suggested that medical staff were reluctant to administer the questionnaire to seriously ill patients and that in future studies efforts to address com— pliance should include both patients and staff. In 1991, as a result of the early experience in the first pilot, the QOL subcommittee began actively addressing the com- pliance issues by sponsoring QOL data management training at each semiannual ECOG meeting, by producing a training video addressing collection, and by initiating a centralized quality as— surance program for QOL assessments within ECOG. As a result of these activities, the overall compliance in all studies ac- tivated since 1991 is approximately 85%. Building on the previous experience, a second QOL study (C0190-Qua1ity of Life on Breast Cancer Adjuvant Trials) was developed in a group of patients with a more favorable prog- nosis (E3189—Phase III Comparison of Cyclophosphamide, Doxorubicin, and Fluorouracil [CAF] and a 16—week Multi- Drug Regimen as Adjuvant Therapy for Patients with Hormone Receptor Negative: Node-Positive Breast Cancer) and limited the number of assessments to three (before, during, and after therapy). In addition, reasons for missing and incomplete assess- ments were prospectively monitored. Compliance (defined as a completed questionnaire) was considerably improved, dropping Journal of the National Cancer Institute Monographs No. 20, 1996 only from 98% to 93% over the three assessments (2). Notably, only 1% of all assessments were missing because of patient refusal and 4% were missing for other reasons. Half of these missing as- sessments occurred in patients who discontinued therapy early, demonstrating the need for explicit instruction for assessment of patients who have stopped therapy early or experienced disease progression. The majority of missing items were the result of the failure to copy both sides of the form or random skipping of the back side of two-sided forms. As a result of this experience, all QOL instruments are now distributed as one-sided copies. The conclusions of the second QOL study described above were that both the CAF and multidrug regimens have a sig- nificant impact on QOL during therapy where the magnitude of the change in Breast Cancer Questionnaire (BCQ) scores is roughly equivalent to the pretreatment difference between patients with ECOG performance status scores of 0 and 1, and by 4 months post-treatment BCQ scores on both arms recover to pretreatment levels (3). The impact on QOL of the shorter but more intensive 16—week multidrug regimen is greater than the 24-week CAF regimen; however, this impact seems justified by the improved disease-free (70% versus 64%) and overall (80% versus 73%) survival at 3 years for the multidrug arm (4). Final- ly, data from the BCQ complements Common Toxicity Criteria (CTC) data. The only significant treatment difference in CTC toxicity was stomatitis [20% versus 9% Grade III and IV for the 16-week multidrug versus CAF regimen (4)]; however, there was no difference in the related BCQ item. In contrast, the BCQ identified an additional, clinically relevant treatment difference related to fatigue. Rapid Growth Because of efforts of the QOL subcommittee and the nation- wide increased interest in QOL research within the cancer treat- ment community, the number of studies with QOL components increased dramatically over time from two active protocols in 1991 to nine in 1994. There has also been a corresponding in- crease in the number of patients and scheduled assessments that have more than doubled every year. There are currently (July 1995) eight active and three proposed ECOG-coordinated studies with QOL as a primary or secondary end point (Table 2). ECOG also currently participates in five other intergroup clini- cal studies that include a QOL component. Future Directions New Study Development and Prioritization With the increasing number of active and proposed QOL studies, there is a need to focus the resources of ECOG on 73 Table 2. ECOG coordinated studies Year Active studies Scheduled assessments 1991 2 81 1992 2 328 1993 5 962 1994 9 2000 studies where the QOL component will have a substantial im- pact on clinical practice. With this in mind, a QOL scientific review form has been developed with specific questions about the potential impact of the QOL results on treatment in the com— munity or on future trials. There also has been an attempt to in— corporate practical considerations into study design. For example, the length of follow-up is limited to 5 years and ac- crual to the QOL component of the trial is limited to the first half of the enrolled patients in a large prostate cancer trial (E7892—A Phase III Randomized, Double—Blind Trial of Ad- juvant Hormonal Therapy for Surgically Treated Pathologic Stage C Carcinoma of the Prostate.) Similarly, the QOL com- ponent of a large breast cancer trial (E3193-Phase III Com- parison of Tamoxifen versus Tamoxifen with Ovarian Ablation in Premenopausal Women with Axillary Node-Negative-Recep- tor Positive Breast Cancer) is limited to the first 367 of a total of 1684 patients. Certain diagnoses have been targeted for QOL evaluations, such as lung cancer, breast cancer, and Kaposi’s sarcoma within the AIDS-related malignancies. These three dis- ease priorities were chosen in 1992 because, at that time, they represented cancer diagnoses in which QOL was recognized as a significant issue to balance with treatment response and toxicity and because of the high degree of interest and support within those disease-oriented committees for QOL research. Since 1992, there has been considerable interest in QOL re— search from many other disease—oriented committees. Most notable is the Genitourinary Committee that currently leads or substantially participates in three QOL protocols. Compliance A target of 90% compliance for QOL assessments and 100% documentation of the reasons for mistimed or missing assess- ments has been set. The current rate of compliance is estimated at 85%, up from a base-line rate of approximately 70%. The im- provement is the result of the accumulation of experience by local data managers and an extensive QOL training and data— monitoring initiative. Prospective documentation of reasons for mistimed or missing QOL assessments are now included in all studies. In addition to past efforts, monitoring and provision of feedback on compliance to individual institutions and affiliates have begun. One component of this monitoring is an annual award to the institutional data manager with the best record of compliance. The award includes funding to attend an ECOG meeting where the data manager will make a short presentation to the QOL data management training session. 74 Areas of Investigation Completed Studies QOL assessments are complete for the adjuvant breast study (C0190). and the results were presented at the 1995 American Society of Clinical Oncology meetings (3). E5592—Phase III Trial Comparing Etoposide/Cisplatin versus Taxol/Cisplatin/G- CSF versus Taxol/Cisplatin in Advanced Non—Small Cell Lung Cancer has recently completed accrual, and all QOL assess- ments are scheduled to be completed in the summer of 1995; the final analysis of the primary outcome data is scheduled for 1996, at which time analysis of the QOL component will be completed. Ancillary Investigations In addition to the treatment comparisons within each of the clinical trials, there are numerous methodologic questions of in- terest to ECOG including: 1) Relationship of QOL and toxicity. Does QOL provide in- formation that toxicity data alone cannot? What toxic effects have the greatest impact on QOL (5)? , 2) Missing item in multi-item scales: What is the best method for handling assessments with missing items? 3) Analysis methods in the presence of missing assessments: What methods are practical for analysis of the QOL studies with missing assessments due to disease and treatment—related mor- bidity and mortality (6)? 4) Cross-cultural and multilingual validation of QOL instru- ments in clinical trials (7.8). 5) Testing the equivalence of commonly used QOL instru— ments to allow for the possibility of better comparison of data across trials and improved communication about QOL among health care professionals (9). Economic (Cost) Outcomes As a long—term objective, ECOG investigators are interested in the integration of QOL information into decision making at the levels of both individual clinical practice and health policy. Toward that end. the ECOG has expanded the scope of scientific inquiry to include economic outcomes that bear on the overall determination of the value of a given treatment in a given cohort of patients. Cost of treatment, patient preferences for various treatments, and patient values for health status outcomes of various treatments (“utilities“) are all of interest. To reflect the expanded scope. the name of the Quality of Life Subcommittee was changed in late 1994 to the Outcomes Subcommittee. References (I) Finkelstein DM, Cassileth BR, Bonomi PD, Ruckdeschel JC, Exdinli EZ, Wolfer JM. A pilot study of the Functional Living IndexiCancer (FLIC) Scale: the assessment ofquality of life for metastatic lung cancer patients. An Eastern Cooperative Oncology Group study. Am J Clin Oncol 1988; l 1:630—3. (2) Fetting .l. Fairclough D, Gonin R, et al. Compliance with a quality of life evaluation in a cooperative group trial. Proc ASCO 1994;1132abstr 1572. (3) Fairclough DL, Fetting J, Cella D, Wonson W, Gonin R. Grove—Conrad M, et al. Quality of life on a breast cancer adjuvant trial comparing CAF with a 16-week regimen. Proc ASCO 1995;114zabstr 890. Journal of the National Cancer Institute Monographs No. 20, 1996 (4) (5) (6) Fetting J, Gray R. Abeloff M, et al. CAF versus a week multidrug regimen as adjuvant therapy for receptor-negative. node positive breast cancer: an intergroup study. Proc ASCO 1995;] l4:abstr 83. Cella DF. Quality of life: concepts and definitions. J Pain Symptom Management l994:9:186-92. Fairclough DL. Gelber R. Quality of life: statistical issues and analysis. In: Spilker B, editor. Quality of life and pharmacoeconomics in clinical trials. 2d ed. New York: Raven Press. 1996. Journal of the National Cancer Institute Monographs No. 20, 1996 (8) (9) Cella DF. Lloyd SR. Data collection strategies for patient—reported infor- mation. Quality Management in Health Care 1994;2z28-35. Cella DF. Wiklund l. Shumaker A, Aaronson N. Integrating health-related quality of life into cross—national trials. Quality of Life Res 1994;2z433-40. Gonin R. Lloyd S. Cella D. Establishing equivalence between scaled measures ofquality of life. Quality of Life Research. In press. 75 Gynecologic Oncology Group (GOG) Donald G. Gallup, David F. Celia History The GOG is the only national multicenter clinical trials group devoted specifically to the treatment of cancer in women. Its primary cancer treatment committees include those for cancer of the ovary, uterine corpus, and cervix and vulva. Quality-of—life (QOL) considerations are important to the treatment of gyne— cologic malignancies for a number of reasons. First, in early stage disease, choices often must be made between very differ— ent treatment modalities (e.g., surgery versus radiation therapy), where traditional clinical outcomes may be equivalent or close to equivalent yet there may be a dramatic difference in acute and long—terrn effects. Second, in advanced disease, a treatment may have limited or no benefit to survival time and yet may improve the quality of that time by virtue of tumor—burden relief. The GOG currently has studies looking at QOL in both early stage disease and advanced disease phase III protocols. The Quality of Life Committee within the GOG is, however, young and funded only through the National Cancer Institute (NCI) Cancer Therapy Evaluation Program. Therefore, the number of active protocols is kept low to reserve resource use for only the highest priority studies within the group. The GOG has a governing structure through which scientific prioritization of protocols is accomplished by the Protocol Committee, after it receives recommendations from multidisciplinary committees, such as the Quality of Life Committee. Committee Membership The Quality of Life Committee is a multidisciplinary commit- tee comprising gynecologic oncologists, nurses, statisticians, psychologists, radiation oncologists, and medical oncologists. There are currently 18 voting members on the committee. Scientific Priorities The Quality of Life Committee of the GOG has selected two general areas for scientific priority. The first is the area of phase III clinical trials; the second is the area of delayed effects of the treatment of curable cancers. The committee has excluded phase I and phase II trials from consideration of QOL evaluation. Until now, the committee has placed less emphasis on symptom con- trol studies. However, now that the GOG has been funded as a Cancer Control Research Base by the Division of Cancer Prevention and Control, symptom control is likely to take on more importance. Symptom control studies will be handled by the newly formed Cancer Prevention and Control Committee. Journal of the National Cancer Institute Monographs No. 20, 1996 QOL Protocols Active There are currently four active QOL studies in the GOG. Since the committee has been in existence for only 2 years, there are no closed or completed protocols. The four active protocols are: 1) Protocol 147: Whole abdominal radiation therapy versus combination doxorubicin-cisplatin chemotherapy in advanced endometrial carcinoma (Treatment Study 122). This first QOL protocol in the GOG was activated as a com- panion protocol; however, it was changed to an integrated protocol to enhance accrual to the QOL component, which has been lagging behind accrual to the parent treatment study. 2) Protocol 152: Phase III randomized study of cisplatin and Taxol (paclitaxel) with interval secondary cytoreduction versus cisplatin and paclitaxel in patients with suboptimal stage III and stage IV epithelial ovarian carcinoma. The purpose of this study is to evaluate the value of secon- dary debulking surgery in patients with suboptimal ovarian can- cer. It is unclear whether this surgery improves survival time, but it remains possible that it improves the quality of survival by decreasing tumor burden and associated symptoms. The purpose of the QOL study is to contrast the relief of symptoms and im— provement of QOL associated with tumor debulking with the short-term disability caused by the surgery itself. 3) Protocol 9102: Effect of alopecia on cancer patient body image and the role of audiovisual information on body image. This study is open to only a limited number of institutions. 4) Protocol 145: Randomized study of surgery versus surgery plus vulvar radiation in the management of poor-prognosis pri- mary vulvar cancer and of radiation versus radiation and chemo- therapy for positive inguinal nodes. Proposed 1) Protocol 137: Randomized trial of estrogen replacement therapy versus no estrogen replacement in women with stage I or stage II endometrial adenocarcinoma. This study is somewhat controversial because of the concern about the potential carcinogenicity of hormone-replacement therapy in women with endometrial cancer. Discussion regard- ing sample size and appropriate end points is ongoing among the GOG, the NCI, and the US. Food and Drug Administration. 2) Late effects of therapy for germ cell tumor survivors. This protocol represents the GOG’s first initiative into the area of studying late medical and psychological effects of cura- tive cancer therapies. Because ovarian germ cell tumors are rather rare, a multicenter group such as the GOG is probably the only forum in which questions of late effects for this disease can 77 be addressed with sufficient sample size. Although there have been a considerable number of post-treatment cancer survivor studies in diseases such as leukemia. Hodgkin‘s disease, and tes- ticular cancer, there exist no comparable data for women who have been previously treated for germ cell tumors. The research protocol and questionnaire packet have been approved, and study activation is pending due to the need for external funding. 3) Protocol 99RB: Phase III randomized clinical trials with laparoscopy pelvic and periaortic node sampling, vaginal hyster- ectomy, and bilateral salpingo-oophorectomy (BSO) versus open laparotomy with pelvic and periaortic node sampling and abdominal hysterectomy and BSO in endometrial carcinoma clinical stage 1, IA, grades I, II, and 111. After surgeons become proficient in laparoscopy staging. the purpose of this phase III trial will be to demonstrate the clinical equivalence of laparoscopy compared with open laparotomy. 78 The QOL study is then pivotal in demonstrating that laparos— copy-assisted staging is superior by virtue of more rapid return to normal function and fewer problems with psychological well— being and body image during the short-term recovery period after surgery. This is an approved study that awaits completion of the surgical proficiency stage of the project; Future Plans The Quality of Life Committee will continue to place em- phasis on phase III trials and late treatment effects. Two priority areas for further investigation include cervical cancer (early stage disease) and bone marrow transplantation in ovarian can- cer. Because this is a relatively new committee in the GOG, there are no mature data from which to generate publications. Journal of the National Cancer Institute Monographs No. 20. 1996 North Central Cancer Treatment Group (NCCTG) C harles L. Loprinzi History While it can be argued that many of the cancer treatment trials in the adjuvant setting and in the advanced disease setting are indirectly related to improving the quality of life (QOL) of our patients (increasing disease-free survival time without recur- rent cancer and shrinkage of metastatic cancer may improve the QOL of patients). it is generally agreed that these studies are not QOL studies per se. Nonetheless. the NCCTG does have a large program dealing with symptom control trials, which we feel are more directly related to the QOL of our patients. These trials are aimed at controlling symptoms that come about from cancer and/or from cancer therapy. They are not designed to look at the quantity of life or shrinkage of cancers but. rather. are designed to look at means to decrease bothersome symptoms. and. through this mechanism. improve the QOL of the patients being studied. Past. present. and future NCCTG research related to symptom control trials include studies aimed at l) preventing or alleviating muc0sitis. 2) treatment of cancer anorexia and cachexia. 3) therapy of menopausal symptoms in patients where estrogen treatment is contraindicated. and 4) improving our ability to care for patients suffering from pain. In 1986. a protocol was developed to study whether an al- lopurinol mouthwash could prevent fluorouracil (5—FU)-induced stomatitis (NCCTG-864651). based on promising pilot infor- mation obtained elsewhere. This study clearly demonstrated that the allopurinol mouthwash was not useful in this situation (1). Subsequently. another trial (NCCTG-88-92-53) was able to clearly demonstrate that oral cryotherapy could markedly reduce 5-FU-induced mucositis (2). A followvup trial (NCCTG—89-92- 58) was developed to evaluate different durations of oral cryo- therapy in patients receiving bolus S—FU-based chemotherapy. The results from this protocol did not suggest any advantage for continuing oral cryotherapy longer than 30 minutes (3). Another protocol was developed to determine whether a chamomile preparation will be able to further ameliorate 5-FU-induced stomatitis (NCCTG—90-92-56). The data from this study did not suggest any benefit from chamomile (4). An additional protocol (NCCTG-909253) was developed to evaluate chlorhexidine and an oral nonabsorbable antibiotic lozenge to determine whether either will be helpful in alleviating stomatitis resulting from ir- radiation of the oral mucosa (5). Based on an interim analysis. the chlorhexidine arm was closed (due to lack of benefit) while the antibiotic lozenge arm versus a placebo arm is being analyzed. Two protocols were developed to evaluate whether sucralfate can 1) inhibit 5-FU-induced mucositis (NCCTG-92- 92-51). or 2) inhibit treatment—induced esophagitis (NCCTG—92- 94—51). Both of these trials rapidly accrued patients and both are Journal of the National Cancer Institute Monographs N0. 20. 1996 closed and being analyzed. Also related to treatment of therapy- related gastrointestinal mucosal injury. a protocol was opened to study whether osalazine can inhibit radiation-induced diarrhea (NCCTG-91-92-53). This trial was closed early because of ex- cessive drug toxicity (6). Currently, concepts approved by the National Cancer Institute (NCI) include 1) studying antibiotic lozenges for treatment of 5-FU-induced mucositis, 2) studying sucalfate for prevention of diarrhea in patients receiving pelvic radiation therapy. 3) studying glutamine for preventing 5-FU-in- duced mucositis, and 4) studying glutamine for preventing radiation-induced mucosal injury. Anorexia and Cachexia Another active area for the NCCTG Cancer Control Program has involved studies aimed at the treatment of cancer anorexia and cachexia. After an initial trial (NCCTG-87-92-51), Kardinal et al. (7) suggested that cyproheptadine was not very useful in this situation. A follow—up protocol clearly demonstrated that megestrol acetate could stimulate the appetite of, and cause weight gain in. patients with severe cancer anorexia and cachexia (NCCTG-88-92-51) (8). The results of this trial at- tracted substantial national interest. Accrual was subsequently completed. with 343 eligible patients being entered in another protocol that evaluated various doses of megestrol acetate and determined that there was a positive dose—response relationship for this drug for patients with cancer anorexia and cachexia (NCCTG-89-92—55) (9). Another trial (NCCTG-9l—92-54) determined that the drug, pentoxifylline, was not helpful to al- leviating cancer anorexia and cachexia (10). Currently, a proto- col is open to compare megestrol acetate to dexamethasone and to fluoxymesterone (NCCTG-91-92-52) in patients with cancer anorexia and cachexia. In addition, a recently closed clinical trial (NCCTG—89—20-51) was designed to determine whether megestrol acetate will improve the survival of previously un- treated small-cell lung cancer patients (11). Two other related trials evaluated a drug that has been purported to have nutritional— enhancing properties. hydrazine sulfate, in patients with 5-FU- resistant advanced colorectal cancer (NCCTG-89-49-51) (12) and in patients with lung cancer receiving concomitant chemo- therapy (NCCT-89-24-51) (13). Menopausal Symptoms Hot flashes can be a major problem in postmenopausal women and in male patients who have had a bilateral orchiec- tomy. especially since estrogen therapy is relatively contraindi- cated in both situations. We completed accrual on a protocol 79 (NCCTG-89—92-54) designed to evaluate the use of the anti- hypertensive medication, clonidine. in this disorder (14.15). Subsequently. another protocol was opened to evaluate low doses of megestrol acetate for the therapy for this problematic symptom in these patient populations (NCCTG-90-92-55) (16). A concept has been approved by the NCI to study vitamin E in breast cancer patients with hot flashes. Another concept has been submitted to the NC] for studying low-dose androgen therapy for symptomatic hot flashes. Another quite bothersome situation for some estrogen-deprived women is vaginal dryness and/or pruritis. Estrogen creams usually will relieve this symptom. but these are relatively contraindicated in patients with breast cancer. A study is now ongoing (NCCTG-91-39-51) to evaluate a new nonhormonal agent (Replens). which has ap— peared to be beneficial in some women with this problem. Analgesic Studies Completed and published analgesic studies include I) a placebo-controlled trial assessing the role of the psycho- stimulant drug. methylphenidate, in improving pain relief and general alertness in patients requiring a strong opioid drug (NCCTG-89-92—51) (I7). and 2) a placebo-controlled trial of a topical local anesthetic cream (EMLA cream) in the manage— ment of painful percutaneous access procedures in children (NCCTG-89—92—52) (18). Other Studies We have completed the pilot phase of a protocol designed to study the efficacy of the methods of measuring QOL in patients with advanced colorectal cancer. This project was initially a part of NCCTG-89—49-5 l . where we were studying hydrazine sulfate in patients with 5-FU—resistant advanced colorectal cancer. Patients entered in this trial were randomly assigned to receive their QOL measured by one of four different QOL measurement instruments (Uniscale. FLIC [Functional Living Index of Can- cer]. Categorical Quality of Life Index. and lnvestigational Pic- tureface scale). After approximately 130 patients were entered in this hydrazine protocol. the protocol entry was stopped be— cause of a preliminary analysis, which demonstrated no benefit for hydrazine sulfate. To complete this QOL project, a separate protocol was developed (NCCTG-93—92-5 l: a randomized com— parison of QOL measurement tools in patients with advanced in- curable colorectal cancer). However. this study was not completed because of inadequate funding to support this work. Thus. in summary. the NCCTG has been. is. and will con- tinue to be actively participating in research that is specifically designed to improve the QOL of patients with cancer. 80 References (I) Loprinzi CL. Cianflone SG. Dose AM. Etzell PS. Bumham NL. Thereau TM. et al. A controlled evaluation of an allopurinol mouthwash as prophylaxis against S-fluorouracil—induced stomatitis. Cancer I990; 65187940. (2) Mahood DJ. Dose AM. Loprinzi CL. Veeder MH. Athmann LM. Themeau TM. et al. Inhibition of fluorouracil-induced stomatitis by oral cryotherapy. JClin Oncol l991;9:449-52. (3) Rocke LK. Loprinzi CL. Lee JK. Kunselman SJ. Iverson RK. Finck G. et al. A randomized clinical trial of two different durations of oral cryotherapy for prevention of 5-fluorouracil-related stomatitis. Cancer 1993;72:2234—8. (4) Fidler P. Loprinzi CL. O'Fallon JR. Michalak J, Novotny P. Hayes D. et al. A controlled evaluation of chamomile for preventing stomatitis in patients receiving 5—fluorouracil based chemotherapy: a North Central Cancer Treatment Group trial. Proc ASCO 1995;14:534. Foote RL. Loprinzi CL. Frank AR. O‘Fallon JR. Gulavita S. Twefik HH. et al. Randomized trial of a chlorhexidine mouthwash for alleviation of radiation»induced mucositis. J Clin Oncol 1994;12:2630—3. (f2) Manenson J. Hylan G. Moertel C. Mailiard J. O‘Fallon J. Collins R. et al. Oslazine is contraindicated during pelvic radiation therapy: results of a randomized trial. Proc ASCO 1995;14:161. Kardinal CG. Loprinzi CL. Schaid DJ. Hass AC. Dose AM. Athmann LM. et al. A controlled trial of cyproheptadine in cancer patients with anorexia and/or cachexia. Cancer 1990;65:2657—62. Loprinzi CL. Ellison NM. Schaid DJ. Krook JE. Athmann LM. Dose AM. et al. Controlled trial of tnegestrol acetate for the treatment of cancer anorexia and cachexia. J Natl Cancer Inst 1990;82:1127—32. Loprinzi CL. Michalak JC. Schaid DJ. Mailliard JA. Athmann LM. Goldberg RM. et al. Phase III evaluation of four doses of megestrol acetate as therapy for patients with cancer anorexia and/or cachexia. J Clin Oncol 1993;1l2762-7. (l0) Goldberg RM. Loprinzi CL. Mailliard JA. O‘Fallon JR. Krook JE. Ghosh C. et al. Pentoxifylline for treatment of cancer anorexia and cachexia‘? Randomized. double-blinded. placebo controlled trial. J Clin Oncol 1995;13:2856-9. Rowland KM. Loprinzi C. Shaw EG. Maksymiuk AW. Kuross SA. Jung SH. et al. Randomized double blind placebo controlled trial of cisplatin and etoposide plus megestrol acetate/placebo in extensive stage small cell lung cancer: a North Central Cancer Treatment Group study. J Clin Oncol. In press. Loprinzi CL. Kuross SA. O'Fallon JR. Gesme DH Jr. Gerslner JB. Rospond RM. et al. Randomized. placebo-controlled evaluation of hydrazine sulfate in patients with advanced colorectal cancer. J Clin Oncol 1994;12:1121-5. ([3) Loprinzi CL. Goldberg RM. Su JQ. Mailliard JA. Kuross SA. Maksymiuk AW. et al. Placebo-controlled trial of hydrazine sulfate in patients with newly diagnosed non-small cell lung cancer. J Clin Oncol 1994:12:l 126—9. (I3) Loprinzi CL. Cianflone SG. Dose AM. Etzell PS. Bumham NL. Thereau TM. et al. A controlled evaluation of an allopurinol mouthwash as prophylaxis against 5—fluorouracil-induced stomatitis. Cancer 1990;65: l879-82. (I4) Goldberg RM. Loprinzi CL. O‘Fallon JR. Veeder MH. Miser AW. et al. Transdermal clonidine for ameliorating tamoxifen»induced hot flashes. J Clin Oncol 1994;12:155-8. (I5) Loprinzi CL. Goldberg RM. O‘Fallon JR. Quella SK. Miser AW. Mynderse LA. et al. Transdemtal clonidine for ameliorating postorchiec- tomy hot flashes. J Urol 1994;151:634-6. (l6) Loprinzi CL. Michalak JC. Quella SK. O'Fallon JR. Hatfield AK. Nelimark RA. et al. Megestrol acetate for the prevention of hot flashes. N Engl J Med 1994;331 1347-52. (I7) Wilweding MB. Loprinzi CL. Mailliard JA. O‘Fallon JR. Miser AW. van Haelst C. et al. A randomized. crossover evaluation of methylphenidate in cancer patients receiving strong narcotics. Support Care Cancer. In press. ([8) Miser AW. Goh TS. Dose AM. O'Fallon JR. Niedringhaus RD. Betcher DL. et al. Trial of a topically administered local anesthetic (EMLA Cream) for pain relief during central venous port accesses in children with cancer. J Pain Symptom Manage l994;9:259-64. (5 (7 V (8 E? (II x. \) Journal of the National Cancer Institute Monographs No. 20. 19% Radiation Therapy Oncology Group (RTOG) Todd Wasserman, Deborah Bruner, C harles Scott History In 1991. the RTOG Quality of Life (QOL) Subcommittee was established to oversee and facilitate QOL research. It is the com— mitment of the RTOG to have QOL in select phase III studies in each disease site (1.2). Studies are chosen where the therapeutic options most warrant a QOL investigation and where there are companion issues related to health economics. The RTOG QOL Subcommittee has a steering group that consists of the committee chairman. vice-chairman. disease—site coordinators. statistician. RTOG protocol manager. and RTOG research associate manager. The role of this group is to provide a review of all RTOG protocols, to sign off on those studies with QOL end points. and to establish policy decisions for the Quality of Life Subcommittee. It is not an objective of the group to develop new QOL instruments. but if member institutions are interested in developing new radiation—appropriate instruments. RTOG will provide them with a research arena to test these new instruments. The RTOG QOL initiative is separate and more global than late toxicity analysis. but there is significant interaction with the Late Effects Subcommittee (3). The QOL Subcommittee is in- volved in the testing of the Late Effects Normal Tissue Scales developed by the Late Effects Subcommittee. The RTOG is working to identify late radiation therapy effects and to evaluate interventions that diminish late effects (toxicity modifications). As part of the educational mission within the RTOG. a QOL Procedure Manual and a patient-oriented QOL video show the value of QOL research to patients and to investigators. QOL train- ing has been incorporated into the training session for RTOG re- search associates. One statistician coordinates all RTOG QOL studies to ensure consistency of design and analysis across the trials. The principal QOL researchers of RTOG institutions are nurses. In an effort to promote greater acquisition of QOL data, the RTOG has adopted the policy of putting patients in charge of their QOL data so that they are responsible for its completeness. RTOG QOL Research Objectives The research objectives are to: set priorities for QOL research within the group: use existing instruments for measuring QOL in a consistent manner: develop guidelines for QOL protocol development and training of RTOG investigators. interact with the statistical unit to develop realistic end points. develop proce- dures for data collection based on study timepoints. and initiate and develop interventional studies in response to QOL data (4). Journal ofthe National Cancer Institute Monographs No. 20. 1996 Research Issues QOL is a multidimensional construct that must incorporate the patients‘ perspective within its measurement (5). It parallels. but is distinct from. acute and late toxicity assessment. The RTOG has also established a Late Effects Subcommittee to study toxicity measurements and scales; the QOL and Late Ef- fects Subcommittees interact. As policy. the RTOG uses exist- ing QOL instruments in studies rather than develop new instruments and. when QOL is determined to be a study end point. all patients will be assessed with the use of a global QOL instrument. This consistent approach to instrument selection al- lows investigators. data managers. and statisticians to become knowledgeable about and familiar with using the instrument(s) relevant to radiation therapy questions (6—9). Use of existing in- struments also allows comparison of RTOG results with those currently in the literature. However. disease—specific questions are developed and added to the general questionnaire. when ap— propriate (10). All QOL research is approved by the Quality of Life Subcom- mittee. One committee member is assigned to act as liaison to each disease—site committee. The priorities established by the Quality of Life Subcommittee guide the use of resources. QOL studies require extensive resources in coordination. data collec- tion. and data analysis. The Research Associates Committee acts as a link to the QOL Subcommittee. since its members are the resource for actually conducting the QOL research. The nurse research associates have the interest. the coordination. and the patient—interview skills needed to conduct QOL research and to participate in the research in the following ways: by serving as coordinators for the conduct of QOL studies. by developing procedures for the collection of QOL data. by training investigators and research associates in interview techniques to obtain patient consent and compliance. and by developing guidelines to reduce patient at- Ir1t10n. References (1) Scott CB. Stew. J. Design. analysis and data management issues in quality of life trials (QOL) within the Radiation Therapy Oncology Group (RTOG). Drug Inf] 1993;27:854—5. (2) Wasserman T. McDonald A. Quality of life: the patient's endpoint. Int J Radiat Oncol Biol Phys 1995;33:965—6. (3) Bruner DW. Wasserman T. The impact on quality of life by radiation late effects [editorial]. lntJ Radial Oncol Biol Phys 1995;31:1353-5. (4) Scott CB. Stetz J. Bruner DW. Wasserman TH. Radiation Therapy Oncol- ogy Group quality of life assessment: design. analysis. and data manage— ment issues. Qua] Life Res 1994;}: “99-206. (5) Bruner DW. In search of the quality in quality—of—life research [editorial]. Int J Radiat Oncol Biol Phys 1995;31:191-2. (6) Choucair A. Scott C. Urtasun R. Nelson D. Coia L. Curran W. Quality of life (QOL) and neuropsychological evaluation (NSE) for patients with 81 (8) 82 malignant astrocytomas (MA). RTOG 91-14. Int J Radiat Oncol Biol Phys 1995;32:178. Murray KJ, Nelson DF. Isaacson S. Scott C. Fischbach AJ. Porter A. et a1. Quality-adjusted survival analysis of malignant glioma. Patients treated with twice-daily radiation (RT) and carmustine: a repon of Radiation Therapy Oncology Group (RTOG) 83-02. Int J Radiat Oncol Biol Phys 1993;27:207. Murray KJ, Nelson DF. Scott C. Fischbach AJ. Porter A. Faman N. et a1. Quality-adjusted survival analysis of malignant glioma. Patients treated with twice—daily radiation (RT) and carmustine: a report of Radiation (9) ([0) Therapy Oncology Group (RTOG) 83—02. Int J Radiat Oncol Biol Phys 1995;31:453-9. Scott C. Choucair A. Urtasun R, Nelson D. Coia L. Curran W. Mini-men- tal status exam versus the Radiation Therapy Oncology Group‘s (RTOG) Neurologic Function Status Scale. Cross-validation using patients from 91- 14. Int Soc Qual of Life Research. Accepted 1995‘ Watkins—Bruner D, Scott C. Lawton C. DelRowe J. Rotman M. Buswell L. et a1. RTOG 90-20: a phase II trial of external beam radiation with etanidazole for locally advanced prostate cancer. Int J Radiat Oncol Biol Phys. In press. Journal of the National Cancer Institute Monographs No. 20. 1996 Southwest Oncology Group (SWOG) Laura C. L011, Carol M. Moinpour, Polly Feigl History Increasing interest in cancer control research in general and effects of cancer treatment on patient quality of life (QOL) in particular led to the SWOG‘s first attempt at QOL assessment in SWOG—8313. an intergroup adjuvant breast cancer clinical trial. In this trial. a standard l-year chemotherapy regimen was com- pared with a shorter. more intensive regimen. In 1984. the QOL study was added to the ongoing therapeutic trial but was ter— minated in January 1989 because of inadequate questionnaire submission rates. As compliance problems became evident. there was concern about whether QOL research could be con- ducted in cooperative group trials. To evaluate these concerns. the SWOG initiated a review of QOL assessment issues and methods in November 1987. A draft position paper was circu— lated within the SWOG outlining how. and to what extent. QOL end points should be included in SWOG clinical trials given the special needs and constraints of cooperative group research. Input from reviewers outside of the SWOG was also incor- porated. In 1988. the members of the Quality of Life Subcom— mittee and its parent committee. the Cancer Control Research Committee. approved the QOL assessment recommendations. In April 1989. the QOL policy guidelines were approved by the SWOG’s Board of Governors: the results of this review were published (2). The approval of the relevant committees in the SWOG and its Board of Governors was important in recogniz— ing the legitimacy of this research in the cooperative group mechanism. The original QOL assessment guidelines addressed a number of areas: 1) QOL assessment should occur primarily in phase III trials. although including QOL assessment in phase II trials can inform the design of future phase III trials. It is not feasible to do QOL assessment in all phase III trials. so certain types of tri- als have been emphasized (e.g.. protocols in which the disease site is associated with poor prognosis and palliative care objec- tives are paramount). 2) Comprehensive assessment of QOL re— quires measurement of physical. emotional, and social functioning; symptom status (both disease- and treatment—re— lated); and global perception of QOL. Symptoms associated with comorbidity should also be assessed. 3) QOL assessments should emphasize a patient report as a supplement to physician- rated toxic effects. 4) The QOL assessments should be brief questionnaires. not interviews. to reduce patient and staff bur- den. Example questionnaires were suggested. 5) Patient-com— pleted QOL questionnaires should have adequate psychometric properties. 6) Categorical versus visual analogue scales are more practical for multicenter clinical trial research. 7) QOL should be assessed at least three times: before. during. and after treat- ment. 8) Special quality control procedures are required to monitor questionnaire submission and to enhance data quality. Journal of the National Cancer Institute Monographs No. 20. 1996 9) QOL studies are conducted as companion trials to therapeutic trials, and all proposals are reviewed by the Quality of Life Sub- committee. In 1994. the Quality of Life Subcommittee and Behavioral Sciences Subcommittee were combined in a single committee with a broad health outcomes focus. The new Behavioral and Health Outcomes Subcommittee will emphasize QOL. recruit- ment and adherence interventions. supportive care. and health economics: the subcommittee sees itself as a resource to the Cancer Control Research Committee and the disease commit- tees in the SWOG. In 1995. the QOL policy guidelines were updated to reflect the incorporation of QOL studies in therapeutic protocols versus separate companion protocols; a renewed emphasis on assess- ment in phase III trials; the elimination of the list of appropriate questionnaires. because the questionnaire pool is evolving; addi- tional quality control procedures; and the structural change in the subcommittee. Over the years. quality-control procedures evolved from responsibility at the study coordinator level to an increasing Statistical Center role. Incorporation of the QOL questionnaires in SWOG’s Expectation Report. a monthly listing of overdue data by institution. and increased monitoring by the Statistical Center Data Coordinators have improved both submission rates and data quality. Group Protocol Development Plan Concepts can be drafted by Behavioral and Health Outcome Subcommittee members or by members of other SWOG com- mittees. These concepts are reviewed by subcommittee mem- bers. members of the Cancer Control Research Committee, and relevant staff at the Statistical Center. If deemed to be of scien- tific value, which is feasible given the SWOG’s structural and resource constraints and consistent with the guidelines for can— cer control research, the concept is developed into a protocol by the investigator with assistance from the SWOG Statistical Cen- ter and Operations Office staff. Many levels of review occur. and the time frame from concept to protocol activation is typi- cally more than 1 year. QOL Protocols Active Studies SWOG-8994—Evaluation of QOL in patients with stage C adenocarcinoma of the prostate enrolled in SWOG-8794 (INT-0086). All patients registered in SWOG-8794 are regis— tered in SWOG—8994 until a total of 400 patients (200 per arm) are registered to SWOG-8994. The objectives of the study are as 83 follows: I) to compare three primary aspects of QOL (treat— ment-specific symptoms and physical and emotional function- ing) according to treatment assignment in SWOG—8794. and 2) to compare three secondary aspects of QOL (general symptoms. global perception of QOL. and social functioning) according to treatment assignment in SWOG-8794. The SWOG Quality of Life Questionnaire is a battery of scales including SF—20 and SF-36 scales. the Symptom Distress Scale. and disease- and treatment-specific items. As of January 1. I995. compliance to the submission of questionnaires has been good. The base-line QOL assessment has been submitted for 95% of the patients. The current submission rates for the 6-week. 6—month. l—year. 2- year. and 3—year follow-up questionnaires are 89%. 91%. 88%. 75%. and 71%. respectively. for those patients alive and in the study long enough for these assessments to be made. SWOG-9208—Health status and QOL in patients with early stage Hodgkin’s disease: a companion study to SWOG- 9133. It is anticipated that 288 patients will be accrued to this study before the treatment study. SWOG-9133. meets its accrual goal. The objectives of the study with respect to QOL are as fol— lows: I) to evaluate prospectively the health status and QOL in patients with early stage Hodgkin's disease receiving either sub- total nodal irradiation or short—course chemotherapy followed by subtotal nodal irradiation; 2) to describe the short-term. acute ef- fects of two treatments for patients with early stage Hodgkin's disease with the use of a patient report of symptoms and QOL: and 3) to evaluate the intermediate and long-term effects of these treatments with the use of patient QOL reports over 7 years. The Symptom and Personal Information Questionnaire. the Cancer Rehabilitation Evaluation System Short Form. and the cover sheet are completed prior to registration to SWOG- 9133. SWOG-9346—A phase III trial of intermittent androgen deprivation in patients with stage D2 prostate cancer. A primary objective of this trial is to compare three treatment- specific symptoms and physical and emotional functioning by treatment arm. A secondary objective is to compare general symptoms. role functioning. global perception of QOL. and so- cial functioning between treatment arms. This will be the first SWOG protocol where QOL is integrated into a phase III therapeutic trial. QOL is a primary objective of this trial. Closed Studies SWOG-9045—Evaluation of QOL in patients with ad- vanced colorectal cancer enrolled in SWOG-8905. A total of 287 patients were registered to this QOL companion study when the parent study. SWOG-8905. closed. The objectives of this study were as follows: I) to compare three primary aspects of QOL (treatment—specific symptoms and physical and emotional functioning) according to treatment assignment on SWOG— 8905; and 2) to compare four secondary aspects of QOL (general symptoms. role functioning. global perception of QOL. and social functioning) according to treatment assignment in SWOG—8905. The SWOG Quality of Life Questionnaire was used. At trial closure, the base-line questionnaire had been sub- mitted for 98% of the patients. The submission rates for the 6—. 11-, and 21-week follow—up questionnaires were 85%. 79%. and 79%. respectively, for those patients who were alive and in the 84 study long enough for these assessments to have been made. The results of this study have been presented at the 1995 SWOG Fall Meeting Plenary Session. SWOG-9039—Evaluation of QOL in patients with clinical stage D2 cancer of the prostate enrolled in SWOG-8894. A total of 739 patients were registered to SWOG—9039 when the parent study. SWOG—8894. closed. The objectives of the study were as follows: I) to compare three primary aspects of QOL (treatment—specific symptoms and physical and emotional func- tioning) according to treatment assignment in SWOG-8894; and 2) to compare four secondary aspects of QOL (general symp- toms. role functioning. global perception of QOL. and social functioning) according to treatment assignment in SWOG—8894. The SWOG Quality of Life Questionnaire was used. As of June 1995. 97% of the base—line QOL questionnaires had been sub— mitted. The submission rates for the 1-. 3-. and 6-m0nth QOL assessments were 87%, 86%, and 79%. respectively. for patients alive and in the study long enough for these assessments to have been made. Analyses are currently under way. SWOG-9248—Phase II trial of paclitaxel (Taxol) in patients with metastatic refractory carcinoma of the breast. At study closure, 135 patients had been registered to the therapeutic protocol; of these. 18 were ineligible. One hundred twenty-five patients had completed base-line QOL question— naires. Because of the phase [1 status of this trial. the QOL ob- jective was restricted to monitoring patient reports of symptoms during treatment with paclitaxel. The Patient Symptom Monitor— ing Questionnaire (Symptom Distress Scale and treatment- specific items) was collected at base line and prior to each course of therapy as long as the patient remained in the protocol treatment. Analyses are currently under way. SWOG-9235—Phase II trial of Casodex in patients with advanced prostate cancer who failed conventional hormonal manipulation. Fifty—three patients were accrued in 6.5 months prior to closure. Four patients were ineligible because of insuffi— cient infonnation. Because of the phase II status of this trial. the QOL objective was restricted to assessing the tolerance and toxicity of Casodex through a combination of physician and patient reporting. The Patient Symptom Monitoring Question- naire and the McGill Pain Questionnaire were collected at base line (prestudy) and every month for 6 months, then discon- tinued. These data have yet to be analyzed. SWOG-9021—Phase III study of postoperative radio- therapy for single-brain metastases. This study was closed prematurely because of poor accrual to the therapeutic portion of the trial. At the time of closure, 54 patients had been registered in the trial, 16 of whom were ineligible. The QOL 0b- jectives were to compare the two arms with respect to QOL and to evaluate the use of a QOL questionnaire specific for central nervous system malignancies. The Spitzer Quality of Life Index was filled out by the patient and a family member at each as- sessment. Portions of the SWOG QOL and symptom question— naire were completed by the patient at each assessment. Concordance of patient and proxy QOL report will be ex— amined. SWOG-8861—Evaluation of QOL in patients with clinical stage A2 or B adenocarcinoma of the prostate enrolled in SWOG-8890. This study was closed because of poor accrual to Journal of the National Cancer Institute Monographs No. 20. 1996 the therapeutic trial. The objectives of the study were as follows: I) to compare three primary aspects of QOL (treatment-specific symptoms and physical and emotional functioning) according to treatment assignment: 2) to compare four secondary aspects of QOL (general symptoms. role and social functioning. and global perception of QOL) according to treatment assignment; and 3) to assess the feasibility of collecting QOL data from patient report via self—administered questionnaires over a 5—year period in a cooperative group setting. Studies in Development SWOG-9327—Randomized phase II pilot study of pen- toxifylline (Trental) and placebo in patients with metastatic malignancy and the anorexia/cachexia syndrome. The objec- tives of this study with respect to QOL are as follows: 1) to evaluate the effect of pentoxifylline on the QOL of patients with the anorexia/cachexia syndrome related to malignancy; and 2) to evaluate the effect of pentoxifylline on the nutritional status of patients with cancer cachexia and on various laboratory meas- urements of nutritional status. The primary end points in this double-blinded. placebo-controlled trial are appetite and fatigue. The SWOG Quality of Life Questionnaire was modified to in— clude a physical functioning scale more sensitive to dysfunction of end-stage cancer patients (Self-Report Barthel Index) and the Energy/Fatigue scale from the SF-36 Health Survey. The Quality of Life Questionnaire, a nutritional status form, and a pill count form completed by the SWOG institution staff will be collected. This study will be activated in early 1996. Phase III trial of placebo versus megestrol acetate at a dose of 20 mg per day versus megestrol acetate at a dose of 40 mg per day as treatment for symptoms of ovarian failure in women treated for breast cancer (no SWOG No.). This study does not contain a comprehensive assessment of QOL but Journal of the National Cancer Institute Monographs No. 20. 1996 emphasizes menopausal symptoms. Patients experiencing hot flashes will be followed for 9 months. Assessment schedules and forms are under development. SWOG-9324—Phase II trial of vinorelbine tartrate for patients with relapsed ovarian cancer. The SF-36 question— naire and the Symptom Distress Scale will be used to describe the change in patient report of QOL (primarily symptom status) associated with salvage therapy. Abstracts and Publications Hayden KA. Moinpour CM. Metch B. Feigl P. O’Bryan RM. Green S. et al. Pitfalls in quality—of—life assessment: lessons from a Southwest Oncology Group breast cancer clinical trial. Oncol Nurs Forum 1993;20:1415—9. Moinpour CM. Feigl P, Metch B. Hayden KA, Meyskens FL Jr. Crowley J. Response. J Natl Cancer Inst 1989:8121 106»7. Moinpour CM, Hayden K. Thompson I. Feigl P. Metch B. Quality of life measurement in Southwest Oncology Group trials: policies and implementation. In: Tchekmedyian NS, Cella DF. editors. Quality of life in oncology practice and research. Williston Park (NY): Dominus Publishing Co., 1991:43—9. Moinpour CM. Quality of life assessment in Southwest Oncology Group clinical trials: translating and validating a Spanish questionnaire. In: Quality of life assessment in health care settings. WHO/IPSEN Foundation Series. Berlin: Springer-Verlag. 1994:8337. Moinpour CM. Savage M. Hayden KA. Sawyers J. Upchurch C. Quality of life issues in cancer. In: Dimsdale JE. Baum A. editors. Perspectives on be- havioral medicine. Hillsdale (NJ): Lawrence Erlbaum Assoc. l995:79~95. Thompson I. Crawford ED, Miller G. Paradelo J. Blumenstein B. Wolf M, et al. Adjuvant radiotherapy following radical prostatectomy for pathologic stage C adenocarcinoma of the prostate: initial evaluation of toxicity. Proc ASCO 1992;11:212, Reference (1) Moinpour CM. Feigl P. Metch B, Hayden KA. Meyskens FL Jr, Crowley J. Quality of life end points in cancer clinical trials. J Natl Cancer Inst 1989;81:485-95. I1 '1 Childrens Cancer Group (CCG) William E. MacLean, Jr. History During the past 30 years, significant advances have been made in pediatric cancer treatment as indexed by traditional study end points. i.e., disease—free survival, tumor response, and overall survival. The CCG, through its multicenter clinical trials, has been a major contributor to this success. Concurrently, the CCG has focused attention on the effects of various cancers and their treatments on children‘s physical health and psychosocial well-being in phase II and III therapeutic trials as well as retrospective studies of long-term survivors. These studies have included measures of physical growth, gonadal function, and cardiac and lung functions. as well as measures of neuro- psychologic and behavioral functioning, employability, in- surability. and educational attainment. This research has had a “toxicity" orientation for the purpose of establishing the “costs“ of various treatments. These results are then used as a guide to prepare subsequent frontline protocols and to inform patients and parents of potential late effects. These protocols include studies of acute lymphoblastic leuke— mia (ALL) in infants where high—dose systemic chemotherapy and intensive intrathecal therapy are used instead of cranial radiotherapy (CRT) to prevent relapse (CCG-lO7—intensive chemotherapy for infants with ALL; CCG—1883—treatment of newly diagnosed infants with ALL under 12 months of age); a study of children with intennediate-risk ALL who received variations of the BFM (Berlin—Frankfurt-Muenster) regimen and either CRT + intrathecal methotrexate (ITMTX) or lTMTX alone as central nervous system prophylaxis (CCG-lOS—studies of modifications in BFM therapy for intermediate-risk ALL; successor to CCG-162A); studies of childhood brain tumors that examine the effects of reduced radiotherapy (CCG-923—low stage medulloblastoma: a study of reduced neuraxis irradiation in newly diagnosed children; CCG—9891—low-grade astro— cytoma and CCG-9892—treatment of medulloblastoma and primitive neuroectodermal tumor in children older than 36 months to 10 years of age with reduced neuraxis radiotherapy and adjuvant chemotherapy); a study of brain tumors in infants that compares two chemotherapeutic regimens in conjunction with granulocyte colony—stimulating factor (CCG-992l—multi- agent chemotherapy and deferred radiotherapy in infants with malignant brain tumors): a study of bone marrow transplant (BMT) in first remission of ALL (CCG-l921—allogeneic EMT in first remission for children with high-risk features of ALL); and a retrospective study of fertility and psychosocial status in long—tenn survivors ofchildhood ALL (L-891). Although much of the research contained in the therapeutic protocols is ongoing, several preliminary reports have been pub- lished (1-6). The long‘term survivor study (L—891—retr0spec- tive cohort study of late effects in long-term survivors of Journal of the National Cancer Institute Monographs No. 20. I996 childhood ALL) has yielded several interesting findings. For ex- ample, survivors (ages 18-33 years) scored significantly higher (more anxiety and more depression) on the Profile of Mood States (POMS) than sibling controls (6). Female survivors had higher scores on the POMS than did male survivors or female and male siblings. Survivors who reported unemployment be- cause of the effects of their disease scored significantly higher on the POMS than did survivors who reported no disease-related employment problems and were fully employed. Similar effects were evident in relation to schooling. Interference with educa- tion was associated with higher POMS scores. In relation to both employment and education, the difference in scores was significantly greater for those survivors who were older at diag- nosis compared with those who were younger. These survivors were also questioned about their scholastic performance (1). After diagnosis, survivors were more likely than their sibling control subjects to enter a special education or learning disabilities program but just as likely to enter a pro- gram for gifted and talented children. The risk associated with special education and learning disabilities placement increased with increasing dose of cranial radiotherapy. Despite these problems, survivors generally had the same probability as their siblings of finishing high school, entering college, and earning a bachelor's degree. There was some indication that survivors treated with 24 Gy and those diagnosed before 6 years of age were less likely to enter college. QOL Protocols CCG currently has three protocols in varying stages of development that will include QOL end points. CCG-l94l—BMT versus prolonged intensive chemo- therapy for children with ALL after an initial bone marrow relapse. This phase III trial for children with ALL and an initial bone marrow relapse within 1 year of completion of therapy will compare prolonged intensive chemotherapy, conventional bone marrow transplantation using human leukocyte antigen/mixed leukocyte culture (HLA/MLC)-compatible sibling donors, and alternative bone marrow transplant strategies employing altema- tive stem cell sources, e.g., matched unrelated marrow donors, haploidentical family marrow donors, or purged autologous marrow. The study plan includes health status assessments with the use of the Ontario Health Survey at several time points. Additional measures of social, emotional, and physical function- ing are being considered for inclusion. CCG-l951—Extramedullary relapse and occult marrow involvement in childhood ALL. This is a phase III group-wide study of children with ALL whose first adverse event while on or off therapy is a central nervous system (CNS) or testicular relapse. Therapy will be determined by the time and site of oc- 87 currence of the extramedullary relapse. Patients developing an early relapse in CNS. less than 18 months from first complete remission. who have an available HLA/MLC-compatible sibling bone marrow donor will be eligible for allogeneic BMT. For patients developing an early CNS relapse without an available HLA/MLC—compatible sibling bone marrow donor. for late CNS relapse. and for all testicular relapse patients. induction therapy will be followed by four 6—week intensification cycles of chemotherapy and by four 12-week maintenance cycles. Patients with CNS relapse will be given craniospinal irradiation during the initial month of maintenance at dosages being deter- mined by current treatment regimen (BMT versus chemo- therapy) and previous CNS radiotherapy history. The health status assessment and social, emotional. and physical function— ing measures will be the same as those used in CCG-1941. S-942—Study of minimally invasive survey of the chest in children with cancer. This is a study comparing minimally in- vasive surgery (MlS) with conventional open—chest surgery in the management of cancer in children. A secondary aim of the study is to evaluate the impact of M18 and open surgery on short-term QOL. at 3. 7. and 30 days after surgery. Several domains of QOL will be examined. including surgery-related pain: physical. social and emotional functioning: and global ratings of health and overall QOL. Group Development Plan It has been argued that QOL is not synonymous with measures of intelligence. psychopathology. academic achieve— ment. peer social status. neuropsychologic functioning. health status, fertility. sensation. mobility. self-care. pain. or growth. Rather. QOL is defined in the literature as a multidimensional construct composed of social. emotional. and physical function- ing as perceived by the patient. Unfortunately. there are few measures of QOL consistent with this definition that are ap— propriate for the special conditions associated with pediatric on— cology. These conditions include a rapidly developing person in which functioning changes radically through the developmental age span. the need for informants or proxies for young children. the need to consider family and cultural context in assessing a particular child’s QOL. measurement of generic aspects of QOL and disease— or treatment-specific effects. the need to fit with large-scale multi-institutional protocols. and so on. Simply stated. what single measure could possibly encompass all of the dimensions of QOL across a developmental age span of 18 years or more. be sensitive to changes in functioning that result from cancer and its therapy. and be suitable for use in the cooperative group research context? Several CCG committees (e.g.. Nursing. Psychology. Cancer Control. and Supportive Care) have been discussing these issues while developing a group—wide plan on QOL. The CCG Execu- tive Committee is in the process of establishing a single strategy group that will determine research priorities for late effects. can- cer control. supportive care. and QOL. This strategy group will focus its future efforts on measurement issues and several high— priority research studies. 88 Measurement is a primary concern for QOL research in pedi- atric oncology. We are using the few existing measures in cur— rent studies to gain some experience with them and to assess issues related to compliance and respondent burden. Concur- rently. there is considerable interest in the development of new QOL measures appropriate for use in future protocols. Several candidate measures are being developed that warrant considera— tion after determining their psychometric characteristics and sensitivity to change in QOL over successive observations. In this regard. we plan a study of ALL patients that will yield im- portant infomiation regarding the Minneapolis—Manchester QOL measure in comparison with the currently available measures. This instrument. recently developed by M. Jenney from the U.K. and her colleagues at the University of Minnesota. is a refine- ment of several existing measures of health outcomes. The proposed study will provide important validity data and a demonstration of the feasibility of telephone interviewing for QOL data collection. There are plans to conduct three retrospective studies involv— ing three well—known patient cohorts: children with acute myelogenous leukemia who received BMT versus chemo— therapy. children with brain tumors who received either standard versus reduced radiotherapy. and children with non-Hodgkin’s lymphoma who received eight-drug combination chemotherapy versus those who received four—drug combination chemotherapy followed by low-dose regional radiotherapy. These studies will provide much needed data on long-term QOL in these patients. CCG will also be examining the appropriateness of existing QOL measures for all cancers. Some have argued that the avail— able measures are most appropriate for children with leukemia and that they are not particularly sensitive to the effects of brain tumors and their treatment. It could be that we will direct some effort to developing a brain-tumor-specific pediatric QOL measure for use in CCG studies. References (I) Haupt R. Fears TR. Robison LL. Mills JL. Nicholson HS. Zeltler LK. et al. Educational attainment in long-term survivors of childhood acute lym» phoblastic leukemia. JAMA 1994;272:1427-32. (3) Kaleita TA. MacLean WE. Reaman G. Whitt JK. Hammond GD. Neurodevelopmental studies of children less than 12 months old diagnosed with acute lymphoblastic leukemia. Proc Inter Soc Pediatric Oncol 1987; 19:159. (3) MacLean WE Jr. Noll RB. Stehbens JA. Kaleita TA. Schwartz E. Whitt JK. et al. Neuropsychological effects of cranial irradiation in young children with acute lymphoblastic leukemia 9 months after diagnosis. Arch Neurol 1995;52:156-60. (4) Stehbens JA. MacLean WE. Kaleita TA. Noll RB. Schwartz E. Cantor N. et al. Effects of CNS prophylaxis on the neuropsychological performance of children with acute lymphoblastic leukemia: nine months post diag— nosis. Child Health Care 1994;23:231-50. (5) Stehbens JA. Kaleita TA. Noll RB. MacLean WE Jr. O‘Brien RT. Wasker- will. MJ. el al. CNS prophylaxis of childhood leukemia: what are the long- term neurological. neuropsychological. and behavioral effects? Neuropsychol Rev 199 l :2: 147—77. (6) Zeltler L. Zhang F. Stuber M. Meadows A. Mills J. Byrne J. et al. Psychological sequelae in adult survivors ofchildhood acute lymphoblastic leukemia. Med Pediat Oncol 1994;23:169. Journal of the National Cancer Institute Monographs No. 20. 1996 Pediatric Oncology Group (POG) Andrew S. Bradlyn, Brad H. Pollock History The POG established a Quality of Life (QOL) Committee ap- proximately 4 years ago, and that group currently is organized as a subcommittee of the Cancer Control Committee. To date, QOL outcomes have been included in a small number of trials. Initially, there was debate regarding a number of conceptual and methodologic issues that evolved into the development of a set of guidelines to direct research efforts. In investigating the QOL of children being treated for, or ultimately surviving, a malig- nant cancer, POG investigators have faced many of the typical problems that confront all investigators in this field: recruiting subjects, determining the most appropriate time of assessment for a particular protocol, and dealing with missing data. How- ever, there has been a need to address a number of issues that are somewhat unique to children and families, such as estab- lishing a definition of QOL that is applicable to children, adoles- cents, and families; dealing with the rapid and variable developmental changes that occur throughout the 0-l8+ year life span of our patients; identifying instruments that reflect that QOL definition; and finally, dealing with the ever-present (and potentially paralyzing) problem of proxy respondents. Histori- cally, we know that QOL outcomes by almost any definition have only rarely been included in phase III trials by either of the two pediatric cooperative groups (1 ); POG has made a con- certed effort over the past several years to examine the potential contribution of alternate end points, such as QOL and economic factors (2). Definition The following definition of QOL was adopted by the POG on the basis of the World Health Organization’s definition of health (1958): “Quality of life is a multidimensional construct, incorporat- ing both objective and subjective data, including (but not limited to) the social, physical, and emotional functioning of the child and, when indicated, his/her family. QOL measurement must be sensitive to changes that occur throughout development.” (Pediatric Oncology Group: un- published definition.) This definition provides a focus for what is meant by the term “QOL,” so that the POG research efforts could be planned, coordinated, and responsive to the rigors of the scientific method. With limited resources (financial and human), the goal is to implement a standardized but flexible approach to the assessment of QOL with the use of a core set of measures along with additional QOL measures that are specific to the objectives of the protocol. Journal of the National Cancer Institute Monographs No. 20, 1996 High-Priority Trials for QOL Assessment Recognizing the limited resources that are available, certain types of protocols were identified as being of the highest priority for QOL end points, and these are consistent with those factors typically identified in the literature. For example, phase III trials were identified as being the most relevant to QOL assessment, especially trials comparing different treatment mo- dalities or trials expected to result in therapeutic equivalence. Additionally, it is specified that trials should be expected to ac- crue a sufficiently large number of subjects to ensure adequate statistical power for the QOL questions. The two areas in which the QOL Committee has focused its efforts are as follows: 1) standard measurement strategy for the measurement of QOL, including the specific instruments that are appropriate, and 2) how to deal with the issue of proxy respondents. QOL Measurement In terms of measurement strategy, the POG has adopted an approach that is based on the notion of a standard core group of measures that may be supplemented by other relevant modules. Group QOL Guidelines recommend the inclusion of several dif- ferent types of instruments across protocols but also allows for the inclusion of protocol-specific questions. The basic strategy is to include (at a minimum) a measure of generic health status, a cancer-specific measure, and a measure of performance status. Additionally, the inclusion of several single-item global ratings of QOL and health is recommended. Unfortunately, unlike QOL investigations with adult patients where there may be multiple, standardized, psychometrically sound instruments from which to choose, QOL research in pediatrics has been severely hampered by the relative paucity of appropriate instruments. For example, at this point in time, there is only one published measure of QOL that was developed with pediatric cancer patients and their families, although there are a number of others currently being developed (3). The issue of the proxy respondent is particularly problematic in investigations of pediatric populations. Because there are limitations associated with proxies, and studies have shown that patients are the best informants about their own QOL, pediatric trials present a unique challenge when devising a QOL com- ponent. It is not unusual for POG trials to identify eligible sub— jects as all patients under the age of 21 with a particular malignancy. Thus, we have to develop a measurement strategy for subjects who may range in age from less than 1 year to 21 years of age. This is further complicated by the fact that patients may cross previously set age ranges for particular instruments during the course of their participation. 89 Ongoing activities include an effort to provide information to each of the Disease Committees within the POG about how QOL questions might be identified and included in protocols under development. A list has been established of individuals at each institution who are responsible for QOL data-collection aspects of clinical trials. In addition, a manual is in preparation that addresses issues regarding the day-to-day management of the investigation, i.e., Institutional Review Board submissions and consent forms, standard administration instructions, and typical “problem situations” and solutions. Protocols With QOL Assessment To date, QOL measures have been included in the following POG protocols: POG-9202—ALinCl6: acute leukemia in children No. 16. This protocol, which is currently accruing subjects, includes a modified QOL component that is embedded within psychologic studies. The QOL data relate to the objective “to determine the feasibility of gathering neuropsychological data with magnetic resonance imaging and specified neuropsychological tests.” POG-9331—Intergroup low-risk medulloblastoma. The protocol also includes a modified QOL component that is em- bedded within psychological studies. The QOL data relate directly to the objectives. This protocol is currently accruing subjects. POG-9485/9585—Intergroup minimal-access surgery. This protocol includes the full QOL battery as recommended in the POG Guidelines. Additionally, questions relating to respon- dent burden are included to further understanding of this issue. The QOL data are end points in the primary objectives of this protocol, which are “to investigate the role of minimal access surgery in terms of short—term quality of life, economic factors, and perioperative morbidity and mortality." There are also a number of protocols in various stages of development or review that include QOL components, including 90 studies of the effect of Enalapril in reducing cardiotoxicity from anthracycline therapy, the effects of bone marrow transplanta- tion on QOL, and the relationship between doxorubicin infusion time and QOL. It is important to note that all of these efforts have been multidisciplinary and have originated from a variety of disease and/or discipline committees. Group Perspective The QOL Subcommittee has been fortunate to have the sup- port of the leadership of the group, which has resulted in earlier identification of relevant protocols and, importantly, the ad— ministrative and statistical support that is crucial to successfully bringing research questions of this type to fruition. In fact, the mission of the POG, as described in its constitution, has recently been amended to include not only the cure of childhood cancer but also the promotion of the quality of our patient's lives. While strides have been made toward this goal during the past 4 years, QOL research is clearly in an early phase within the FCC. There is a desperate need for the funding of basic research that addresses questions such as instrument development and ongo- ing validation. Given the relatively low-base rate of childhood cancers, many of these questions must be asked on a multi-in- stitutional or group-wide basis, and this cannot be accomplished without financial support. POG investigators are pleased with the progress thus far and are looking forward to contributing to the growing database regarding the QOL of children and adoles- cents with cancer as well as their families. References (I) Bradlyn AS, Harris CV. Speith L. Quality of life assessment in pediatric cancer clinical trials: a retrospective review of phase III cooperative group investigations. Soc Sci Med 1995;41:1463—5. (2) Land V, Pollock Bl-I. Economic outcomes in clinical trials. Symposium at the Pediatric Oncology Group meeting. Chicago, IL: 1994. (3) Goodwin D, Boggs S, Graham—Pole J. Development and validation of the Pediatric Oncology Quality of Life Scale. Psycholog Assess l994;6:321—8. Journal of the National Cancer Institute Monographs No. 20, 1996 Quality of Life in Clinical Cancer Trials: Experience and Perspective of the European Organization for Research and Treatment of Cancer Gwendoline M. Kiebert, Stein Kaasa* Background The European Organization for Research and Treatment of Cancer (EORTC) is an international nonprofit organization that was founded in 1962 by European cancer specialists to conduct, develop, coordinate, and stimulate research in Europe on the ex— perimental and clinical bases of cancer treatment and related problems. The ultimate goal of the EORTC is to provide the best state-of—the-art treatment to as many cancer patients as pos— sible in Europe. The fundamental structure of the EORTC Treat- ment Division is based on the input from 22 cooperative groups, who develop their clinical research through the direct input of the participating scientists. The development of this research is supervised by different committees. Research is accomplished mainly through the execution of large, prospective, randomized, multicenter cancer clinical trials. More than 2000 clinicians lo- cated in 350 medical institutions in 31 countries enter each year approximately 6000 patients in about 100 ongoing studies. The EORTC Data Center in Brussels is the nucleus of all the clinical research. It provides an optimal and unique European infrastructure to conduct multicenter and multidisciplinary clini- cal trials with expertise in data management, biostatistics, medi— cal monitoring, and quality-of—life and health economics evaluations. The EORTC Data Center is therefore concerned with all aspects of late phase II and phase III cancer clinical tri- als, the design and preparation of such trials, collection of data, statistical analysis, and publication of the final results, as well as quality—control procedures and legal and administrative respon- sibilities. Currently, more than 50 people with various scientific backgrounds are working at the Data Center. They include data managers, statisticians, medical doctors, medical fellows, nur- ses, economists, pharmacologists, psychologists, and computer specialists. To address these important issues, six specialty units have been created within the Data Center. Until recently, clinicians have mainly focused their attention on the more classical aspects of evaluating cancer treatment out— comes, such as response to treatment, relapse, and (disease—free) survival. It is now increasingly recognized that quality of life is an important outcome measure in the evaluation of cost/benefit ratios of new interventions, especially when the impact of medi— cal treatment on the length of life is expected to be small. The number of trials that include quality of life as an outcome parameter has increased rapidly during the last few years and is still increasing. Table 1 provides an overview of the EORTC tri- Journal of the National Cancer Institute Monographs No. 20, 1996 als that have included quality of life as an end point during the past 10 years. The rapid growth of the number of studies assessing quality of life emphasized the need for a coherent policy and a standard approach to conduct this research. For this reason, a Quality of Life Unit was established at the EORTC Data Center in 1993 with financial support from the European Community. Its main objective is to stimulate. enhance, and coordinate quality of life as a treatment outcome in cancer clinical trials. In this context, the principal tasks of this unit are to establish an adequate in- frastructure for the data management of quality-of—life studies; to undertake the design, collection, and analysis of quality-0f- life data in EORTC clinical trials; and to generate specific quality—of—life research questions. At present, the unit consists of a psychologist (PhD. and head), a statistician, a part-time quality-of—life administrator, and a part-time data manager. In the near future, we hope to welcome a research fellow. The Quality of Life Unit has a close collaboration with and builds further on the achievements of the EORTC Study Group on Quality of Life. This group was created in 1980 and from its inception has included a broad range of professionals with ex- tensive experience in quality-of—life research. In these past years, this group has performed much research to develop sound tools to measure quality of life in cancer patients. It has developed a modular approach to the assessment of quality of life by which a core questionnaire measuring a range of physi- cal, emotional, and social health issues is supplemented by diag- nosis-specific and/or treatment-specific modules (1,2). The core questionnaire, known as the EORTC QLQ-CSO, is currently available in 18 languages and is being used in more than 200 studies worldwide. It is a copyrighted instrument, and ad- ministration of the core questionnaire is handled by the Quality of Life Unit. Various modules (such as the lung, breast, head and neck, and colorectal cancer modules) have been developed or are currently being field tested [e.g., (3,4)]. For each module, *Afiiliutimzs (g/‘amlmrs: G. M. Kiebert, Quality of Life Unit, European Or- ganization for Research and Treatment of Cancer (EORTC) Data Center, Brus- sels, Belgium; S. Kaasa, EORTC Study Group on Quality of Life, Palliative Medicine Unit, University Hospital. Trondheim. Norway. Cm‘respondem'e m: Gwendoline M. Kieben, Ph.D., Quality of Life Unit, EORTC Data Center. Avenue Mounier 83/11, 1200 Brussels, Belgium. 91 Table l. EORTC trials (1985-1995) with quality—of—life evaluation as an end point Year Phase Protocol* No. of patients Status 1985 III Operable breast cancer in the elderly 413 Closed [11 Randomized trial on dose response in radiotherapy of low—grade gliomas 379 Closed III Orchidectomy versus LHRH analogue in metastatic prostate cancer 327 Closed 1986 [11 Long—term QoL of adult leukemia after bone marrow transplantation versus intensive consolidation 1057 Closed (acute myelogenous leukemia) III Radiotherapy versus no radiotherapy for cerebral gliomas of the adult 237 Open 111 Estracyt versus mitomycin C in hormone escaped advanced prostate cancer 171 Closed 1987 111 Development of EORTC core QoL questionnaire for cancer patients 985 Closed 1988 III Adjuvant trial in malignant melanoma comparing recombinant interferon alfa—2 with recombinant 755 Open interferon gamma with control 1990 111 Early versus late orchidectomy or early versus late treatment in asymptomatic nonmetastatic prostate cancer 537 Open III Endocrine treatment with flutamide versus cyproterone in good-prognosis patients with prostate cancer 286 Open III Orchidectomy versus orchidectomy + mitomycin C in poor—prognosis patients with metastatic prostate cancer 189 Open 1991 [11 Short, intensive preoperative combination chemotherapy versus similar therapy given postoperatively in 482 Open breast cancer patients III LD—ARA—C versus LD-ARA—C + GM—CSF versus LD»ARA—C + recombinant interleukin 3 for patients with 201 Open myelodysplastic syndromes and high risk of developing acute leukemia III Flutamide versus prednisolone in hormone—resistant metastatic prostate cancer 1 14 Open 1992 II Second-line chemotherapy with docetaxel in patients with breast cancer 83 Closed III Strontium chloride versus palliative local—field radiotherapy in patients with horrnone—resistant 70 Open metastatic prostate cancer 1993 III Dose—intensive chemotherapy as primary treatment in locally advanced inflammatory breast cancer 249 Open 111 Influence of dose intensity on survival in G-CSF~supported treatment of HIV—associated non—Hodgkin’s 209 Open lymphoma (high malignancy) Il-lll Comparison of Cisplatin-based chemotherapies in NSCLC 181 Open 11 Randomized paclitaxel versus doxorubicin as first-line chemotherapy for advanced breast cancer 230 Open 111 Chemotherapy with or without G-CSF in operable osteosarcoma 47 Open 111 Induction and intensive consolidation followed by bone marrow transplantation in acute myelogenous leukemia 615 Open 1994 III Role of booster dose of postoperative radiotherapy in patients with early stage carcinomas of head and neck 12 Open 111 Oral pamidronate versus placebo in breast cancer patients with newly diagnosed bone metastases 77 Open 111 Prospective radiotherapy versus chemotherapy in patients with locally advanced head and neck carcinoma 53 Open 111 Paclitaxel + platinum versus cyclophosphamide + platinum in advanced epithelial ovarian cancer 171 Open III 5»FU and L—leucovorin after liver or lung metastasis resection from colorectal cancer 4 Open 111 Cisplatin + cyclophosphamide versus abdomino~pelvic irradiation in high—risk epithelial ovarian cancer 5 Open II—III Cisplatin + 5-FU versus cisplatin + 5—FU with interferon alfa in metastatic pancreatic cancer 14 Open 111 Surgery versus radiotherapy in NSCLC after response to induction chemotherapy 13 Open 1995 ll First—line iv vinorelbin and Cisplatin in patients with metastatic epidermoid carcinoma ofesophagus 3 Open 111 3BEP versus 3BEP-1EP in good-prognosis gemi cell cancer 13 Open 111 Reliability and validity of QLQ-C30 (version 3.0) and head and neck cancer module 0 Open 111 Reliability and validity of QLQ—C30 (version 3.0) and breast cancer module 0 Open *Ll-IRH = luteinizing honnone—releasing hormone: QoL = quality of life; LD-ARA-C = low—dose cytosine arabinoside (cytarabine); GM-CSF = granulocyte—mac- rophage colony—stimulating factor; G-CSF = granulocyte colony-stimulating factor; NSCLC = non-small-cell lung cancer; 5—FU = fluorouracil; iv = intravenous: HIV = human immunodeficiency virus; 3BEP = three cycles of bleomycin + etoposide + Cisplatin; 3BEP»1EP = three cycles of bleomycin + etoposide + cispatin—fol— lowed by one cycle of bleomycin + etoposide. one member of the study group is the principal investigator responsible for its development. How Does the EORTC Integrate Quality-of-Life Questions in Clinical Trials? ' There are three principal channels to integrate quality of life as an outcome measure in EORTC clinical trials. The first channel is through training and education. Although many clinicians subscribe to the importance of quality—of—life evaluation in clinical trials, only a few have extensive knowl- edge and/or personal experience with quality—of-life assess- ments. Often there is a lack of familiarity with quality—of—life instruments as well as a lack of experience in solving practical problems in implementing quality-of—life assessments. The Quality of Life Unit tries to increase awareness and knowledge 92 of quality-of—life issues by having sessions on quality—of—life considerations at least once during one of the cooperative groups” meetings. These meetings are held every 6 months. Many sessions have already been organized, and it is our ex- perience that a l—hour presentation of the basic principles about the “when,” “why, who,” and “how” of quality—of—life meas— urements is usually sufficient to result in a lively, constructive discussion and prepares the groundwork for future collaboration with the Quality of Life Unit at the Data Center. The second channel for integrating quality-of—life issues is through assigning liaison members from the Quality of Life Study Group to the various disease-oriented cooperative groups. For the past few years some members from the study group, with a particular interest in involvement in a specific disease site. have been appointed as liaisons to offer their expertise to the cooperative groups on how to conduct quality-of—life assess- ” H Journal of the National Cancer Institute Monographs No. 20, 1996 ments in clinical trials. In principle, the liaisons attend all meet- ings of their respective cooperative group. Between meetings, they can be consulted concerning quality-of—life issues. Some cooperative groups have formed a Quality of Life Subcommittee consisting of clinicians with a special interest in quality-of-life issues. During the cooperative groups' meetings. subgroup meetings are held to discuss ongoing matters. A senior member from the Study Group on Quality of Life is also a member of the EORTC Protocol Review Committee. The third channel is the involvement of the Data Center staff at an early phase of protocol development. The existing protocol submission procedures require the early involvement of the Quality of Life Unit at two moments. The first moment is during the initial development of a new protocol. This consists of a two-page outline, which must be submitted to the Protocol Review Committee for approval of the basic idea of the study. Each two-page outline is seen and reviewed by the Quality of Life Unit before it is sent to outside referees. If this two-page outline of the protocol is approved by the Protocol Review Committee. a full protocol can be developed in which quality of life is an integral part of the study objectives. An accompanying letter to the principal investigator includes a recommendation to contact the Quality of Life Unit as soon as possible. It is ex- plained to the investigator that this part of the full protocol must be approved by the Quality of Life Unit before the protocol can be submitted to the Protocol Review Committee for final ap- proval. This is the second moment. The investigator is not obliged to contact the unit, but one runs the risk that a study cannot be opened for the patients’ entry because this part of the protocol has not been approved by the Quality of Life Unit. EORTC Criteria for Inclusion of Quality of Life in Clinical Trials Phase III Studies The general policy of the EORTC regarding the inclusion of quality—of—life issues in phase III cancer clinical trials is as fol— lows: Theoretically, it can be a relevant end point if ° no improvement in overall, recurrence-free, or sys- temic disease-free survival is expected, but when significant changes or differences in (at least) one aspect of quality of life are expected; ° one treatment results in a better survival but has more toxic effects; - the patients have an extremely poor prognosis with or without treatment; - treatment is known to be very burdensome to patients; ' a new (invasive) treatment is to be evaluated. If either one or a combination of these criteria applies to a proposed study. then it is up to the cooperative group in general and the principal investigator in particular to decide whether or not quality of life will be evaluated. Both the EORTC Protocol Review Committee and the Quality of Life Unit do not follow the policy of imposing quality of life as an end point. However. they can strongly advise to include it as an end point if they con- sider it to be a relevant issue. The EORTC adopted this policy Journal of the National Cancer Institute Monographs No. 20, 1996 for the following reason: Since quality of life is a relatively new field of research (and for many clinicians and institutions even an experimental field of research), it requires extra motivation on the part of the clinicians and other persons responsible for data collection before its assessment can become a fully in- tegrated part of clinical practice. Imposing an extra workload on already overloaded personnel, who may also doubt the useful- ness of evaluating quality of life. will only have a negative ef— fect on the quality of those data. Since there are about 100 trials ongoing every year, the EORTC prefers to have a limited num- ber of studies with good-quality data instead of a large number of studies with low-quality data. The best way to convince and motivate people with regard to the importance of quality—of-life research is by means of examples of successful and high—quality studies. In those studies in which quality of life has been accepted as an end point, this aspect of the study is mandatory in all institu- tions. The only exception to this rule is for countries for which no validated translation of questionnaires in that particular lan- guage exists. In general, the ability to fill in quality-of—life as- sessments is one of the inclusion criteria, but refusal or missed quality-of—life evaluations are not exclusion criteria for entering a study. If quality of life is an end point. then the protocol should pro- vide information on the following aspects: (a) rationale for the inclusion of measuring quality of life as a primary or secondary outcome measure; (h) formulation of both the study objectives and hypotheses; (c) justification for the quality-of—life aspects or dimensions that will be evaluated: ((1) description of patient eligibility criteria; (6) design and methods used; and (f) statisti- cal considerations such as sample size calculation and methods used to analyze data. Phase 11 Studies In principle. quality of life is not considered a relevant end point in EORTC phase II trials, since the primary aim of such a study is to determine anticancer activity as well as toxicity. Pre- vious studies (5.6), however. have shown that patients and their physicians can differ in their rating of toxic effects and burden of treatment. Moreover, the clinical evaluation of toxicity focuses mostly on its occurrence and severity and not on the duration of toxic symptoms or on the relative burden for patients. For these reasons, it may be important to include the patient‘s valuations of these factors in phase II studies. This may provide not only important information (and thus increase our understanding of the frequency, severity. and burden of the side effects), but also valuable information for deciding on the design of a subsequent phase III trial. It can provide a better in— dication of which aspects of quality of life should be examined. and it may be used to determine which interventions should be made early in the phase III trial in order to minimize symptoms and dysfunction. The subjective evaluation of the perceived bur- den of treatment—related side effects, however, is not to be con- fused or regarded as equivalent to the “classical" approach to the evaluation of quality of life. The latter entails more than just the assessment of treatment-related symptoms. There is one exception to the general rule not to measure quality of life in a phase II study: randomized phase II studies 93 that will continue as phase III studies, in which quality of life is regarded as an important outcome measure. Since the data on patients entered in a randomized phase II study will be included in the phase III comparison, quality-of—life assessment should have started already in phase II. The points mentioned above reflect the present policy of the EORTC. and they serve as theoretical guidelines for the integra- tion of quality-of—life issues in its late phase II and phase III clinical trials. In practice. however, the awareness. knowledge. and previous experience with quality-of—life evaluation appear to be stronger determinants for the integration of this end point than guidelines. These factors can be active at the following three levels: personal, group, and national. The personal level refers to the principal investigator. If this person has had positive experience with quality-of—Iife evalua— tions, then this optimizes the chances that quality of life will be evaluated in a new study if it is considered a relevant outcome. The same principle applies to the second level, which refers to the attitude toward and experiences of the cooperative group with quality—of-life issues. It is remarkable to note that there are cooperative groups who have a long history of quality-of-life evaluations in their trials. whereas other groups appear quite reluctant and resistant to consider quality of life as an outcome measure. This discrepancy seems difficult to explain. Although there are disease sites that have a long history of extensive quality-of—life research (e.g., breast cancer and genitourinary cancers). grounds do not seem to exist for the assumption that the relevancy of quality-of—life issues is different in the various cancer sites. The third level concerns the national policies of the various countries. EORTC studies are conducted on an international level. Cross—nationally. substantial differences do exist, not only with regard to familiarity with quality-of—life evaluations. but also with regard to the infrastructure for managing cancer clini- cal trials in general and for quality-of—life assessment in par— ticular. Some countries (e.g., The Netherlands) provide data management support to their large institutions. in the form of either data managers or research nurses. The presence or ab— sence of such an infrastructure substantially influences the motivation and capacity of clinicians and institutions to par- ticipate in high—quality and sophisticated cancer clinical trials. Lack of data management support can be a reason to limit the number of ongoing studies that include quality of life as an end point per disease site. How Does the EORTC Build on Successive Trials? Ideally. each new clinical trial builds on the results of the preceding study. The efficacy of a new treatment applied in an experimental group is compared with a control group who usually receives the treatment that is considered to be the cur— rent standard. The new treatment modality is tested for equivalence or for a difference. If the new treatment is proven to be better. it will become the new standard. In the next study. this treatment will then serve as the control arm. The same principle applies to quality—of—life issues. Ideally. new studies incorporate the results of the preceding ones. Obviously. this has been the case in the process of developing the EORTC core questionnaire 94 and the disease-specific modules. The first generation of the core questionnaire. consisting of 36 questions. was developed in 1987. Detailed results on the international field testing of this in— strument were published in 1991 (7). While the overall psychometric results were promising. they also pointed to some directions in which the questionnaire could be improved. In the next generation of the instrument, these areas were further developed. A major advantage of the EORTC approach is that the same core instrument is used in each study. This approach allows a sufficient degree of generalizability for cross-study comparison. The Quality of Life Unit is involved in a wide range of studies across cooperative groups and throughout all its phases from the design to the analysis and publication of the results. This unique characteristic allows us to get a good overview of the actual field of research; it enables the coordination of research projects at a European level. and it is extremely helpful in generating new research questions. This way. we hope to contribute sub- stantially by building on the results of successive trials within the EORTC. Priorities for the Near Future Although much progress has been made during the last decade. there is still a long way to go before quality-of—life evaluation can be regarded as an integrated part of standard can— cer clinical practice. The rapid growth in the number of EORTC studies that include quality of life as an end point may reflect the increasing awareness and importance of the subject on the part of the investigators, but it has also pointed out more clearly the flaws and shortcomings in this new field of research. The EORTC has set the following priorities for its activities related to quality-of—life issues: Good-Quality Studies Since EORTC trials are conducted in an international, multi- center setting. it is extremely important to have a good in- frastructure and a standard approach to the collection and analysis of quality-of—life data. To ensure adequate rates of patient accrual, compliance. and data quality, there is an urgent need for a number of standard data management strategies. These strategies include implementation procedures. detailed in- structions for data collection. explicit instructions on the ad— ministration of quality-of—life instruments, regulations on coding of data. and interpretation of missing data and incomplete forms. A standard training course for people who are responsible for data collection will be developed and conducted at regular inter- vals in all countries that participate in EORTC studies to ensure optimal benefit. Analysis and Interpretation of Data Despite the research efforts of the last two decades. a number of questions with regard to specific issues remain open. An im— portant fact is that there is no optimal method for analyzing quality—of—Iife data. Several methods can be used and perhaps should be used to provide insight into the data. However, each method has its advantages and disadvantages. and different models have different assumptions that are not always met. Journal of the National Cancer Institute Monographs No. 20. 1996 The interpretation of results is impeded by the lack of stan- dards concerning what can be considered as a clinically impor- tant change in any quality-of—life score and the absence of standard methods to define effect sizes and to calculate sample size requirements. An important step forward would be the availability of large datasets that could be utilized in future trials for the computation of expected differences and sample sizes. Since the EORTC QLQ-C30 is currently being used in many studies, reliable datasets should become available in the near fu- ture. A final methodologic issue relates to the integration of dif- ferent outcome measures. As stated previously, cancer clinical trials have a history of parameters. all related to length-of—life outcomes. Further development of methods to combine length- of-life with quality-of—life data is both warranted and a major challenge. Since resources for health expenditure are becoming more restricted, health economic issues have become increas- ingly important also in cancer clinical trials. Combining eco- nomic data with quality-of~life and length-of—life data, therefore, will become increasingly important. These issues will be ad- dressed in close collaboration with the EORTC Health Economics Unit. Theoretical Issues Although it has become virtually impossible nowadays to keep up with the stream of publications of empirical studies on quality-of—life issues. the theoretical foundation and framework on quality of life are still rather weak. Quality of life is a dynamic concept. like illness. However, the way in which and degree to which these two concepts interact with each other and what other additional factors may have an influence are still largely unknown. One such additional factor is the unknown role culture plays in quality-of—life issues. A unique charac- teristic of the EORTC is that its clinical trials are by definition cross-national studies. The total number of countries that are in- Journal of the National Cancer Institute Monographs No. 20, 1996 volved in EORTC studies is at present 31. This feature provides a treasure of information to investigate cross-cultural differ— ences. So far, this investigation has not been done, but cross- cultural differences will become one of the major new research questions in the near future. . In conclusion, this article has outlined the experience and perspective of quality-of-life research within the EORTC. Al- though much progress has been made, there is still a lot of work to do before quality of life achieves its rightful place in cancer therapy evaluation. With the present enthusiasm and motivation on the part of all parties involved, we are optimistic that this process will lead to a better understanding of the impact of an- ticancer therapy on patients’ quality of life. References (I) Aaronson NK, Ahmedzai S, Bergman B. Bullinger M, Cull A. Duez NJ, et al. The European Organization for Research and Treatment of Cancer QLQ- C30: a quality—of-life instrument for use in international clinical trials in on- cology. J Natl Cancer Inst 1993;85:365-76. (2) Aaronson NK. Cull A. Kaasa S. Sprangers MA. The EORTC modular ap— proach to quality-of—life assessment in oncology. Int J Mental Health 1994;23:75-96. (3) Bergman B. Aaronson NK, Ahmedzai S. Kaasa S. Sullivan M. The EORTC QLQ—LCI3: a modular supplement to the EORTC Core Quality of Life Questionnaire (QLQ-C30) for use in lung cancer clinical trials. EORTC Study Group on Quality of Life. EurJ Cancer 1994;30:635—42. (4) Bjordal K. Ahlner—Elmqvist M. Tollesson E, Jensen AB, Razavi D, Maher EJ. et al. Development of a European Organization for Research and Treat- ment of Cancer (EORTC) questionnaire module to be used in quality of life assessments in head and neck patients. EORTC Quality of Life Study Group. Acta Oncol 1994;33:879—255. (5) Presant CA, Quality of life in cancer patients. Who measures what? Am J Clin Oncol 1984;7z571-3. (6) Olver JN, Matthews JP, Bishop JF. Smith RA. The roles of patient and ob— server assessment in anti—emetic trials. EurJ Cancer 1994;30A21223-7. (7) Aaronson NK. Ahmedzai S. Bullinger M. et al. The EORTC core quality-of- life questionnaire: interim results of an international field study. In: Osoba D. editor. Effect of cancer on quality of life. Boca Raton (FL): CRC Press, 1991:185—203. Assessment of Quality of Life in Clinical Trials of the British Medical Research Council David Machin* This article describes aspects of the way the Cancer Therapy Committee of the British Medical Research Council incor- porates quality-of-life (QOL) assessments in randomized clinical trials in patients with cancer. The steps taken in in- corporating QOL assessments in individual trial protocols are described. The aspects described concern problems as- sociated with choice of instruments, time of assessment, sample size, and analysis. A protocol for patients with small- cell lung cancer that compares oral etoposide with in- travenous multidrug chemotherapy is used for illustration. [Monogr Natl Cancer Inst 1996;20:97-102] Cancer clinical trials sponsored by the British Medical Re- search Council (MRC) are usually organized under the auspices of either the Leukemia Working Party or the Cancer Therapy Committee. The latter has site-specific working parties who ad— dress therapeutic questions regarding solid tumors, and it is the work of the Cancer Therapy Committee that is described here. The Cancer Therapy Committee essentially is made up of the chairs of the site-specific working parties, an independent chair, several other independent assessors, including one with special interest in quality of life (QOL), and the chief medical statis- tician of the MRC Cancer Trials Office. Proposals for clinical trials are generated within the site—specific working parties, and a brief summary of these proposals is presented to the Cancer Therapy Committee for its approval. Approval is or is not given at a full meeting of the committee. The particular trial coor- dinator of the proposal under discussion attends this meeting to explain the rationale for the trial. The coordinator is absent when a decision on the particular protocol is made. Approval of the protocol at this stage guarantees the statistical support of the Cancer Trials Office and signals the development of a full protocol. This protocol, together with the appropriate data forms, is then subsequently put to an independent Protocol Review Committee for approval. The Protocol Review Commit— tee discusses the protocol line by line with the trial coordinator, statistician, and data manager assigned to that particular protocol. At this review, it is not usually expected that major changes will be made to the therapeutic questions being ad— dressed, as these have been examined in detail at the earlier stages. Rather, the review is to see that the protocol is indeed practicable. Once the Protocol Review Committee gives its ap— proval, the final protocol documentation is prepared and the trial is launched at a convenient date. Ethical approval of each protocol is given at a local level, usually by a committee of the institute where the participating Clinicians work. If a trial Journal of the National Cancer Institute Monographs No. 20, 1996 proposed is a pragmatic one, which may require many thousands of patients, then this trial is usually coordinated through the U.K. Coordinating Committee for Cancer Research, and this type of trial is not considered further here. [See for ex- ample, details of the AXIS trial, 1994(1)] Until recently. each working party had the responsibility to decide whether or not QOL was appropriate for the particular study in question; again. until relatively recently. the major use of QOL measures was confined to the Lung Cancer Working Party. This particular group has a long history of using QOL measures and was responsible for developing the MRC patient diary card (2-4). As will be illustrated below, this working party has made extensive use of the Rotterdam Symptom Checklist (5) and the Hospital Anxiety and Depression Scale (6) and has recently started to use the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 (7). In 1993, the MRC reached a concordat with the U.K. Depart- ment of Health. One of the consequences of that concordat was that QOL (and health economic assessment) ought to be an in— tegral pan of clinical trials. Thus, the site-specific working par- ties of the Cancer Therapy Committee now have to state why QOL should not be included in a particular trial. An example of a case in which QOL is not included is a trial in operable os- teosarcoma. Patients with operable osteosarcoma are mainly children. and no validated QOL instrument for children was available prior to the launch of the trial in 1993. In that case. the reason for not conducting QOL was at a very practical level. Work on an appropriate instrument is in progress. Of the 25 open, randomized phase III trials of the Cancer Therapy Committee, 12 involved QOL assessments of one form or another. QOL assessments were included in trials of cancers of the bladder. brain, colorectum, lung, prostate, kidney, and stomach. The specific reasons for inclusion of QOL assessment in a renal trial were documented (8). ln explanatory trials in which it is anticipated that the therapy being tested may bring more than modest therapeutic gain, as expressed in terms of patient survival, survival is used as the main outcome measure. and patient numbers are calculated on the basis of the anticipated survival benefit. In contrast, how- ever, in some of the palliative trials of the Lung Cancer Work- ing Party, in which attempts have been made to reduce therapy as compared with standard therapy, these trials aim for survival *Corroxpnmlcm"c tr): David Machin, M.D., Medical Research Council., Can- ccr Trials Office. 5 Shafteshury Rd., Cambridge C82 28W, U.K. 97 equivalence. As a consequence, the QOL issues become more prominent and indeed may be the major outcome variable. If this is the case. then QOL becomes the focus for the design and. in particular, determines the end points for calculations of patient numbers. Within the MRC and elsewhere. there is considerable ex— perience in estimating appropriate sample sizes on the basis of survival end points (9.10). This is not only because such cal- culations have been used frequently by the statisticians but also because the clinicians are able to balance the anticipated sur- vival gain against the weight of therapy in the patient groups and thereby determine a clinically worthwhile difference to be established by the trial. On the other hand. although QOL is clearly an important end point for the patient. experience of as- sessment and perhaps more importantly the “feel" for what con- stitutes an improvement in QOL are more problematic. Strategies that attempt to summarize subjective clinical opinion at the design stage do not appear to have been utilized (/1). Thus. objective definitions of what constitutes a clinically im— portant benefit in terms of QOL have not been identified. al- though work is in progress in this area (/2). To date. this problem has been circumvented somewhat for the purposes of sample size calculation by focusing on a single item or com— ponent of the QOL questionnaire and using this as a surrogate. Thus. for example. the 10 symptoms that most trouble patients with lung cancer have been identified. and the palliation of a prespecified number of these symptoms has been regarded as an indicator of a clinically important QOL improvement ([3). Clearly. other issues related to QOL have to be addressed. These issues include patient compliance. patient attrition. and missing data. All of these issues need to be considered at the design stage and may influence the number of patients to be recruited. An Example To illustrate the various aspects of development of a protocol involving QOL as an integral part. we use a randomized. con- trolled clinical trial of oral etoposide versus intravenous multi- drug chemotherapy for the palliative treatment of patients with small-cell lung cancer and a poor prognosis. This trial. referred to as LU 16. was begun in August 1992. and it is anticipated that it will close toward the end of 1996 after 500 patients are recruited. Design The trial design of the LU16 trial is shown in Fig. l. The fig- ure summarizes the eligible patients and the randomization to oral etoposide against the intravenous multidrug chemotherapy EV (i.e.. etoposide + vincristine) or CAV (i.e.. cyclophos- phamide + doxorubicin + vincristine) and details the follow-up for QOL assessments by means of the Rotterdam Symptom Checklist and the Hospital Anxiety and Depression Scale and the period during which the patient diary card should be com- pleted. The patient diary card was specifically included here in order to assess the influence of active therapy on those aspects of QOL that may be transitory during the treatment phase and caused either by an immediate benefit of therapy or as a conse— 98 quence of the side effects of the therapy. For example. the use of the patient diary card in a previous trial had indicated transient dysphagia between 10 and 21 days from the start of radiotherapy in patients with non-small-cell lung cancer receiv- ing a two-fraction course of radiotherapy. while such an excess was not noted for those patients randomly assigned to receive a single—fraction regimen (/4). The QOL assessments by means of the Rotterdam Symptom Checklist. the Hospital Anxiety and Depression Scale. and the patient diary card are completed following the schedule sum- marized in Fig. 1. Thus. immediately before the therapy is started and before the randomized treatment is allocated. the patient completes each of these three instruments. The daily diary card is then completed for 12 weeks. which covers the period until completion of the fourth cycle of chemotherapy. The Rotterdam Symptom Checklist and the Hospital Anxiety and Depression Scale are completed every 3 weeks immediately before the chemotherapy is administered and thereafter until 3 months. then monthly to 6 months. then every 2 months to 1 year. and every 3 months thereafter. Since the treatment was scheduled to be completed by the 12th week (3 months). this was believed to be not only an ap- propriate date for QOL assessment but also the key QOL assess- ment for evaluation and hence design purposes (Fig. 2). In many situations. it is not always clear when. for example. the Rotterdam Symptom Checklist or the Hospital Anxiety and Depression Scale questionnaire should be completed in order to make sensible comparisons between treatments. especially if the alternative therapies under test are of different types (e.g.. chemotherapy as opposed to radiotherapy) and/or of different duration. It is usually not desirable to ask for additional clinic visits to complete QOL instruments alone merely in order to maintain synchrony between treatment assessments. Usually some compromise has to be reached between the optimal time points that are best for the scientific question posed and the demands of everyday patient care. Number of Patients The stated objectives of LU16 trial are listed in Fig. 2. which identifies palliation as the major outcome variable. As already referred to. there is an intrinsic difficulty in defining benefit in these circumstances. To assess patient numbers. it was thought appropriate to select a series of symptoms from the Rotterdam Symptom Checklist to form the basis of a definition of palliation (/3). The symptoms identified were cough. pain. anorexia. and shortness of breath. Each of these symptoms is scored on an or- dered categorical scale from 0 (not at all) to 3 (very much). Thus. the score at presentation could range from 0 to 12. With appropriately selected patients. however. the lower limit is un- likely to be less than 2. This definition. albeit somewhat ar- bitrary. was then applied to patient data from previous trials and was found to achieve approximately 50% palliation (improve- ment in QOL) in the equivalent of the CAV arm. These calculations led to a sample size of 400 patients. but because of patient attrition it was believed to be appropriate to increase this sample size (/5). A judgment was then made sug- gesting that 500 patients would be more appropriate. The cor- responding statement made in the protocol is shown in Fig. 3. It Journal ofthe National Cancer Institute Monographs No. 20. 1996 Limited or extensive SCLC Performance status: WHO 2, 3, 4 [-134NDOMISEW Clinician's choice E {Ev—j Ea] E lflj LQAYl then monthly to 6 months then every 2 months to 1 year then every 3 months thereafter Follow—up Pretreatment, RSCL, HAD - / ' week Reports: ”r orFfl 0, E Q}! 3 ' _ E [ EV J lCAV J Follow up, RSCL, HAD Follow—up, RSCL, HAD 9 , Follow-up, RSCL, HAD Follow-up at 3 months - —————————————— JL Follow-up, RSCL, HAD Follow-up, RSCL, HAD Patient Diary Card Fig. l. LU16 trial design: British Medical Research Council randomized. controlled clinical trial of oral etoposide versus intravenous multidrug chemotherapy in the palliative treatment of patients with small—cell lung cancer (SCLC) and poor prognosis (dated August 1992). The panels that are presented in this section of the article are extracted from the LU16 protocol itself and have not been edited. E = oral etoposide: EV = etoposide + vincristine: CAV = cyclophosphamide + doxorubicin + vincn'stine'. RSCL = Rotterdam Symptom Checklist; HADS = Hospital Anxiety and Depression Scale: WHO = World Health Organization. was recognized that comparisons between treatments with respect to other aspects of QOL would be made on a more infor- mal (exploratory) basis. Comparison of Q()L Instruments Since the launch of the above trial, the copyright format of the EORTC core questionnaire (EORTC QLQ-C30) has become available. There is therefore sotne debate as to whether or not this format should replace the Rotterdam Symptom Checklist. As a consequence. rather than all patients on the LU16 trial receiving the Rotterdam Symptom Checklist as is indicated by Fig. l. half of the patients now receive this checklist and half receive the EORTC QLQ-C30 on a random basis. This ran— domization gives the trial a 2 X 2 factorial design format. al- though analysis of the two instruments cannot be made on this basis. In a certain sense. the best comparison of the two instru- ments should be a within—patient comparison. with both instru- Journal ol'thc National Cancer Institute Monographs No. 20. I996 merits completed almost simultaneously, albeit this design has obvious flaws. In any event. it is recognized that this is not pos- sible, at least within the context of a randomized, controlled trial involving many centers, as it clearly places an extra burden on both patients and staff. Since the primary objective of a trial is to compare treatments. it will be of interest to see which instru~ ment best reflects the (standardized) true treatment difference. There is some circularity here, since we do not know the true treatment difference (16). Such a comparison is also likely to in- volve other factors in the final choice of instrument for future use. These factors include. in particular. considerations of any major differential in compliance rates. Practical Considerations One of the major obstacles to recruitment to Clinical trials is often the complexity of the trials themselves in terms of the extra information on a patient that it is necessary to record over 99 PRINCIPAL ENDPOINT: SECONDARY ENDPOINTS: 2. Adverse effects of treatment 3. Quality of life 4. Survival 5. Response 1. Palliation of major symptoms at 3 months Fig. 2. End points and definition of prin~ cipal end point for the LUI6 protocol (dated August 1992). O Palliation of major symptoms defined as failures of palliation O Palliation is defined as having a reduction in the sum of the cough, pain, anorexia and shortness of breath scores at three months from randomisation 0 Patients who die before 3 months (whether or not they have palliation) are It is anticipated that major symptoms (cough, pain, anorexia and shortness of breath) will be palliated in 50% in the control group within the first 3 months of treatment. The oral etoposide treatment will be regarded as equivalent to the intravenous chemotherapy Pig. 3. Statistical considerations section of the LU16 treatment if palliation is achieved in not less than 37.5% of the patients. With this 12.5% level of equivalence, a one-sided test at 5% and 80% power would require a total of between 400 and 500 patients. protocol (dated August 1992). and above that recorded in routine clinical practice. Of course. there are other more difficult areas. including seeking informed consent from the patients (17). As a consequence of the recog- nized burden on the clinical team, a great emphasis. at least in Europe, has been on conducting minimum—forrns trials. The clinicians themselves have recognized and welcomed the need for such an approach. It is therefore somewhat counter to this trend that we now add (many) QOL assessments. Of course, these assessments are intended to be completed by the patients themselves. At the very least, however. the participating centers need to make appropriate arrangements for distribution. comple— tion, and return to the trials office. This is no small task. Sugges- tions to individual centers as to how they may facilitate completion of the QOL instruments are usually included in the study protocol. and some advice for the LU16 trial is sum— marized in Fig. 4. The burden on the trials offices themselves is also not easy to dismiss. The data received have to be processed, their quality needs to be assessed and queried. missing forms have to be pur- 100 sued, and finally analysis needs to be conducted. Thus, QOL as- sessments, albeit a desirable feature for the majority of ran- domized trials of treatments for cancer patients, should not be conducted without taking account of the resources required. Analysis Although it is not the purpose of this article to go into details of aspects of analysis of QOL data once collected, this is clearly an important issue, and steps that have been taken by the MRC in this respect have been outlined elsewhere (18). These steps follow lines similar to those suggested for the analysis of menstrual bleeding diaries (19). The patient-diary—card assess— ments have been reported both in terms of relative compliance between treatments and by means of a daily summary measure of individual symptoms (20). For example, Fig. 5 shows the patient-diary-card profile as recorded for activity in a randomized trial comparing ECMV chemotherapy (i.e., etoposide + cyclophosphamide + metho- Joumal of the National Cancer Institute Monographs No. 20. 1996 Application of the Quality of Life Questionnaires It is important to explain to the patient that the Rotterdam Symptom Checklist (RSCL) and the Hospital Anxiety and Depression (HAD) scale refer to how they have been feeling during the past week, and that all questions should be answered even if the patient feels them to Fig. 4. Application of the quality- of—life questionnaires in the LUlfi trial (dated August 1992). be irrelevant. Emphasise that the completion of these forms helps doctors find out more about the effects of the treatment. Also remind the patient to complete the back of the RSCL. The patient should complete the questionnaires, without conferring, whilst waiting to be seen in the clinic. Collect the questionnaires before the patient leaves and check that all questions have been answered, if necessary going back to the patient immediately and asking them to complete any missing items. trexate + vincristine) with selective palliative treatment in 162 patients with small-cell lung cancer. This profile was not in— cluded in the published report (2]). Activity was recorded on a 5—point scale. ranging from 1 (at work or active retirement) to 4 (confined to home or hospital) to 5 (confined to bed). Thus. Fig. the two treatment modalities. No formal testing of such profiles is attempted. The compliance for each treatment on a monthly basis is indicated beneath the horizontal axis in Fig. 5 and clear- ly indicates how poor it was. although this trial was conducted between 198] and 1985 and organizational details for encourag- ing completion of patient diary cards have since been improved (Fig. 4). There are certain hidden problems with this method of sum- mary. however. These problems include patient attrition and the 5 indicates that the ECMV treatment, which requires hospitalization. does indeed induce more inactivity than the selective therapy in the first few days following randomization. Thereafter. activity levels improve and are comparable between 1OO Activity Grade 4or5 VVVVVVVVVVVVVVVVVVVVVVVVVVVV (%) 75 """"""""""""""""" ECMV 50 ................................................................ I ' il 25 1' ir' " * * “Fill. l l ‘ l . Selective | i. W O O 1 2 3 4 5 6 Month ECMV 4 27 22 14 15 13 8 Selective 13 24 17 11 9 7 5 Fig. 5. Patient diary profiles of patients randomly assigned to receive either ECMV (i.e.. etoposide + cyclophosphamide + methotrexate + vincristine) or selective treatment with respect to activity as assessed by a patient diary card. Journal of the National Cancer Institute Monographs No. 20. 1996 101 “blur“ that occurs if. for example. patients receive their treat— ment on days other than those scheduled. Thus. if the patient diary card was assessing nausea or vomiting during an intensive chemotherapy regimen. this symptom would be greatest on treatment days but more or less absent on other days. Such a profile would be represented by spikes at the appropriate cycle day interspersed by a very low plateau if all therapy was on schedule but blurred otherwise. For other QOL instruments that are not recorded on a daily basis, it is therefore important that these analyses in a sense compare like with like. Thus. one strategy adopted is to make treatment comparisons between patients completing the same number of QOL questionnaires (and at the same time points) and then to combine these differences by means of a stratified analysis as one might do in any standard survival—type corn- parison. The summary statistics used in such comparisons have usually been the slope and intercept of a linear regression equa- tion fitted to the individual patient profiles. These summary statistics are then summed over patients, and treatment corn— parisons are made with these. This method can be extended to include orthogonal polynomial fits if changes over time are not even approximately linear. In this respect. we prefer the approach to repeated measures data advocated by Matthews et al. (22). who suggested that key features of each profile be identified, such as the area under the curve (AUC). rather than a formal repeated analysis of variance that can be utilized through standard statistical packages. The main reason for our preference is that it is important that any analysis focuses on aspects of QOL summary that have a rela— tively easy interpretation. It is recognized. however. that surn— marizing such complex data by means of relatively few parameters may hide more subtle treatment differences that nevertheless may have an important impact on a patient‘s well— being. Approaches to analysis suggested by Korn (23) also con- cemed the AUC, and work is in progress to confirm the utility of this particular approach. Discussion The introduction of QOL assessments into the conduct of ran— domized clinical trials in cancer raises issues that range from the choice (and perhaps development) of an appropriate instrument, choice of completion times. additional burden on the patient and the trial itself, appropriate sample size. analysis. and interpreta- tion. Of particular importance here is any “trade off" between QOL and survival. Increased survival should not necessarily dominate, particularly if it is at some considerable cost in terms of QOL, but neither should the opposite be seen to be the case. An appropriate estimate of survival and an equally reliable quantification of QOL (or at least aspects thereof) are likely to be jointly valuable guides to patient management. The role of QOL in other areas of MRC activities has been described in part by Johnson (24). 102 References (1) The AXIS color'ectal cancer trial: randomisation of over 2000pa1ients. The AXIS Steering Group. BrJ Surg 1994;81:1672. Fayers PM. Jones DR. (iirling DJ. Measurement of quality of life in cancer clinical trials. Cancer Treat Sympos [9853:2580 Jones DR. Fayers l’M. Simmons J. Measuring and analyzing quality of life in cancer clinical trials. In: Aaronson NK. Beckrnan J. editors. The quality oflife ofcancer patients. New York: Raven Press. 1987:41-61. Fayers PM. Bleehen NM. Crirling DJ. Stephens RJ. Assessment ofquality of life in small-cell lung cancer using a Daily Diary Card developed by the Medical Research Council Lung Cancer Working Party. Br J Cancer 1991;64:299-306. De Haes JC. van Knippenberg FC. Neijt JP. Measuring psychological and physical distress in cancer patients: structure and application of the Rotter- dam Symptom Checklist. BrJ Cancer 1990;62:1034~X. Zigmond AS. Snaith RP. The Hospital Anxiety and Depression Scale, Acta PsychiatrScarrd 1983;67:3olv70. Aarorison NK. Ahmed/ai S. Bergman B. Bullinger M. Cull A. Duez NJ. et al. The European Orgarri/ation for Research and Treatment of Cancer QLQ—C30: a quality-of—life instrument for use in international clinical tri— als in oncology. J Natl Cancer Inst 1993;85:365-761. (8) Fayers PM. Cook PA. Machin D. Donaldson N. Whitehead J. Ritchie A. et al. On the development of the Medical Research Council trial of alpha-in- terferon in metastatic renal carcinoma. Urological Working Pany Renal Carcinoma Subgroup. Stat Med 1994;13:2249-60. Freedman LS. Tables of the number of patients required in clinical trials using the logrank test. Stat Med [9811:121-9. (I0) Fayers PM. Machin D. How many patients are necessary? Br J Cancer 1995:7211-9. Spiegelhalter DJ. Freedman IS. A predictbe approach to selecting the size of a clinical trial based on subjective clinical opinion. Stat Med 1986:51- l3. (/2) George S. Julious SA. Campbell MJ. Sample sizes for studies using SF-36. J Epiderniol Cotnrnun Health. In press. Hopwood P. Stephens RJ. Symptoms at presentatiort for treatment in patients with lung cancer: implications for the evaluation of palliative treatment. The Medical Research Council (MRC) Lung Cancer Working Party. BrJ Cancer 1995;71:633-6. (/4) A Medical Research Council (MRC) randomised trial of palliative radiotherapy with two fractions or a single fraction irt patients with in- operable non»smallaccll lung cancer (NSCI.C) and poor performance status. Medical Research Council Lung Cancer Working Party. BrJ Can- cer 199216593441. (/5) Machin D. Campbell MJ. Statistical tables for the design of clinical trials. Oxford: Blackwell Scientific Publications, 1987:94-131. Machin D. Lewith (31.. Wylson S. Pain measurement in randomised clini- cal trials: a comparison oftvm pain scales. ClinJ Pain 1988;4z 161-8. (/7) Altman DU. Whitehead J. Parrnar MK. Stenning SP. Fayers PM. Machin D. Randomised consent designs in cancer clinical trials. EurJ Cancer. In press. 1/8) Hopvrood P. Stephens RJ. Machin D, Approaches to the analysis ofquality of life data: experiences gained from a Medical Research Council Lung Cancer Working Party palliative chemotherapy trial. Qual Life Res 1994:32339—52. (19) Machin D. Farley TM. Busca B. Campbell MJ. d‘Arcangues C. Assessing changes in vaginal bleeding patterns in contracepting women. Contracep- tion 1988;38:165—79. (20) Bleehen NM. Girling DJ, Machin l). Stephens R]. A randomised trial of three or six courses of etoposide. cyclopltOsphamide. methotrexate. and vincristine or six courses of etoposide and ifosfamide in small cell lung cancer (SCLC). 11: Quality ot~ life. Medical Research Council Lung Cancer Working Party. BrJ Cancer 1993;638:1157-6361. (2/) Survival. adverse reactions and quality of life during combination chemotherapy compared with selective palliative treatment for small-cell lung cancer. Report to the Medical Research Council by Its Lung Cancer Working Party. Respir Med 1989;83:51-8. (22) MattheWs JN. Altman DG. Campbell MJ. Royston P. Analysis of serial measurements in medical research [see comment citation in Medline]. BMJ 1990;300:230-5. (23) Korn EL. On estimating the distribution function for quality of life in canA cer clinical trials. Biometrika 1993;80:535-42. (24) Johnson AL. Some statistical issues in quality of life measurements. In: Trimble MR, Dodson WE. editors. Epilepsy and quality of life. New York: Raven Press 1994:65-84. (2 (3 (4 (5 (6 3' E (II (I3 (/6) Journal of the National Cancer lnstittrte Monographs No. 20. 1996 United Kingdom Cancer Research Campaign Approach to Quality—of—Life Research in Cancer Clinical Trials Penelope H0pw00d* Clinical trials of new anticancer therapies form an impor- tant part of the research activity of the Cancer Research Campaign (United Kingdom). and quality-of-life (QOL) end points are being increasingly used in the evaluation of new treatment approaches. The Campaign has a unique policy of supporting a broad range of scientific and clinical research. including psychosocial studies. and thus QOL research is generated in a variety of clinical settings. The focus of inter- est for the Cancer Research Campaign lies in QOL design and assessment rather than the routine application of QOL protocols. Clinical investigators are free to adopt an in- dividual approach, but the Campaign operates a strict peer- review system in protocol assessment. Some standardization of approach is being achieved through consensus of opinion and wide collaboration, both nationally and internationally. [Monogr Natl Cancer Inst 1996;20:103-5] The Cancer Research Campaign (CRC) (a United Kingdom national charity) supports a wide—ranging portfolio of research encompassing the nature and causes of cancer. new approaches to treatment and prevention. clinical trials of new therapies. psychosocial studies, and an educational program. Clinical and nonclinical training programs are also funded and a number of personal fellowships are awarded. The CRC is unique in the funding of cancer research in the United Kingdom because of its policy of supporting a broad range of scientific and clinical activities. This. in turn. means that quality- of-life (QOL) research can originate from many different clinical and/or academic sites. which are shown in Table 1. The CRC is rigorous in the application of peer review in the assessment of research. using the expertise of its own commit- tees and external. often international. referees. Therefore. when directly funded by the CRC. QOL protocols are also subject to this close scrutiny. which ensures that a high standard is achieved. The funding of the educational and psychosocial research program. which includes a small number of QOL projects. ac— counts for approximately 5% of the overall budget and is as- sessed and administered through a separate committee. While QOL protocols are more likely to arise within the clinical trials setting. some specific projects. such as the QOL study in the UK. Tamoxifen Chemoprevention Trial. have been funded directly from the educational and psychosocial research budget. Journal of the National Cancer Institute Monographs No. 20. [996 Table 1. Cancer Research Campaign Scientific and clinical research Educational and psychosocial research (EPR) Scientific committee EPR committee Funding ot'clinical and scientific research. via Clinical trials centers Research groups Program grants Project grants Fellowships Studentships Collaborative research initiatives Clinical trials \ ‘ QOL protocols Funding ofpsychosocial and QOL research. via Research groups Program grants Project grants Fellowships Studentships Collaborative research initiatives Clinical trials originate from individuals or groups within» universities. hospitals. and medical schools and also through project grants. The major trial centers now incorporate QOL end points in most trials. principally phase III randomized clinical trials. but also in some phase II studies. The CRC hosts an expert committee of leading scientists and clinicians involved in the development of new therapy for can- cer. i.e.. the Phase [/11 Clinical Trials Committee. This highly qualified committee advises on the development and testing of novel anticancer agents and carries out early (phase I and II) clinical trials. The committee‘s policy. like that of the European Organization for Research and Treatment of Cancer (EORTC). is not to conduct QOL research in these early stages of clinical testing of new drugs. A wide range of cancers is covered by CRC phase III trials. including all major solid tumor sites. lym- phomas. and hematologic cancers. One particular focus of activity is the CRC Cancer Trials Of— fice. located at Kings College Hospital. which supports the CRC Breast Cancer Trials Group. The group structure comprises four working parties (responsible for biological protocols. new *Ctu'rerpmzdem‘e to: Penelope Hopwood. M.D., Cancer Research Campaign. Psychological Medicine Group. Christie Hospital NHS Trust. Stanley House. Wilmslow Rd.. Withington. Manchester M20 4BX. United Kingdom. See “Note" section following “References.“ 103 studies. current adjuvant trials. and closed trials) that interact with the central trials group parent committee. An executive committee, which includes an expert on QOL (L. Fallowfield). coordinates the activity of the different subgroups. In this way. a consistent approach to QOL research is ensured. and the struc— ture facilitates the design. development. and analysis of QOL end points in a cohesive way. Several major trial centers (e.g.. Birmingham and Glasgow) have recently appointed a person to be responsible for QOL re— search so that this can be developed and coordinated efficiently. In addition. CRC clinicians are involved in collaborative re— search with the U.S. National Cancer Institute. the EORTC. the British Medical Research Council. and the Imperial Cancer Re— search Fund. The United Kingdom Coordinating Committee for Cancer Research (UKCCCR) acts as a coordinating committee for cancer research in many of the U.K. collaborative programs. which are also starting to include QOL protocols. An example is the UKCCCR Adjuvant Breast Cancer Trial. QOL Application While QOL research is primarily associated with the cancer trials that are described above. its application is much wider. QOL measures have been incorporated in CRC-funded research evaluating psychosocial interventions in controlled randomized trials (I) and are currently being used in psychosocial studies of women with a genetically high risk of cancer. In the field of cancer prevention. a battery of self-report ques— tionnaires is being administered to women in a randomized trial of tamoxifen versus placebo. Psychosocial researchers funded by the Campaign incorporate QOL measures in a wide range of projects. This adds to the overall expertise in generating and analyzing QOL data. to the development of subscales. and to the refinement of measures for use in the clinical trial setting. Incorporating QOL Into Phase III Clinical Trial Protocols To date. the CRC has not published a mission statement ad— vocating routine incorporation of QOL into phase III clinical tri- als. in contrast to the policy advocated. for example. by the National Cancer Institute of Canada. although the British Medi- cal Research Council now expects to see QOL assessments in trial protocols. Nevertheless. U.K. investigators in CRC clinical trials are being asked increasingly. by protocol review commit- tees and peer group referees. to consider adding QOL end points where appropriate alongside the more traditional outcome measures. There is also growing interest from purchasers and providers in generating these data. Consequently. there is evidence that the integration of QOL research in clinical trials has expanded considerably over recent years. and through infor— mal collaboration and the open exchange of ideas. a consider— able degree of overlap in approach has developed. In designing QOL protocols. there is agreement among QOL researchers that such studies should answer a specific research question and (where evidence exists from earlier research) should test a hypothesis. Table 2 shows elements of the decision process that may be considered when assessing the potential in— 104 Table 2. Deciding when to assess QOL in clinical trials: a decision tree Is there likely to be a difference in the treatments compared that —>What is the principal QOL research question? will have an impact on QOL? , ,, ./»/ . <—/T W 77 Has the impact on QOL been HDch it warrant replication? measured before? , 7. ,,// «I, ~ Is the expected effect of treatment easily measurable? —>Should a pilot study be conducted? Is there a suitable instrument? «1—» , Is the sample big enough to detect —>Should collaboration be considered? a difference? -,// , k’,//T ls the potential workload/cost to the patient/staff/institution acceptable? ->Will the results of the QOL study influence future patient care? clusion of QOL end points. This is more likely to lead to a proper consideration of the sample size. selection of appropriate measures. timing of assessment. duration of research. and other issues of methodology. It is also more likely to ensure that the results have clinical relevance and practical use. It is important that QOL end points are not included without this kind of plan- ning. since the research makes considerable demands on resour- ces and must be justified. Also. poorly planned research is more likely to generate incomplete data. precluding any useful inter- pretation of the results. Thus. wherever possible. QOL protocols should be designed in parallel with the clinical trial and in col- laboration with someone with specialist knowledge of QOL. The area of QOL design and assessment is of particular inter- est to the CRC and one where it is most willing to provide re- search funding rather than the routine application of QOL protocols. QOL Measures Investigators are free to select the most appropriate mea- sure(s) for any particular study: there is no overall policy to limit this. However. a number of groups may have been influ— enced by published recommendations made by a working party of the Medical Research Council Cancer Therapy Committee (2) that suggested that the Rotterdam Symptom Checklist (3) combined with the Hospital Anxiety and Depression Scale (4) provided the optimal approach at the time. The recommenda- tions were made in an effort to encourage some degree of com- monality of measures in trials. to ensure compatibility of results. Since that time. a number of other carefully developed and well- validated measures have been published: for example. the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) (5) and the Functional Assessment of Cancer Therapy—General Scale (FACT-G) (6). These. and a limited number of other instru» ments. are all currently being used in cancer trials. More specific measures of body image. sexual adjustment. and at— titudes to illness are being developed by CRC research fellows. In summary. the CRC is actively involved in QOL research in cancer clinical trials and. as a result of its broad support of psychosocial research. is also able to support the necessary Journal of the National Cancer Institute Monographs No. 20. I996 developmental and advisory functions through researchers in the psychosocial field. Active collaboration with other organiza— tions and institutions ensures that some degree of standardiza- tion and cross—fertilization of ideas is achieved and facilitates collaboration in large multicenter and, in some cases, multina— tional trials. References (I) (2) Greer S, Moorey S. Baruch JD, Watson M, Robertson BM, Mason A, et al. Adjuvant psychological therapy for patients with cancer: a prospective ran- domised trial [rec comment citations in Medlinel. BMJ 1992;304:675-80. Maguire P, Selby P. Assessing quality of life in cancer patients. BrJ Can- cer1989;60:437—40. Journal of the National Cancer Institute Monographs No. 20, 1996 (3) (4) (5) (6) de Haes JC. van Knippenberg FC, Neijt JP. Measuring psychological and physical distress in cancer patients: structure and application of the Rotter— dam Symptom Checklist. BrJ Cancer 1990;62:1034-8. Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 1983;67:361-70. Aaronson NK, Ahmedzai S. Bergman B. Bullinger M, Cull A, Duel NJ, et al. The European Organization for Research and Treatment of Cancer QLQ-C30: a quality—of—life instrument for use in international clinical tri- als in oncology. J Natl Cancer Inst I993;85:365»76. Cella DF, Tulsky DS, Gray G, Sarafian B, Linn E, Bonomi A, et al. The Functional Assessment of Cancer Therapy scale: development and valida— tion of the general measure. J Clin Oncol 1993;11:5724). Note Supported by the Cancer Research Campaign, United Kingdom. 105 Health—Related Quality-of—Life Studies of the National Cancer Institute of Canada Clinical Trials Group David Osoba, Janet Dancey, Benny Zee, James Myles, Joseph Pater* Since 1989, the National Cancer Institute of Canada Clinical Trials Group (NCIC CTG) has been successful in im- plementing and completing health-related quality-of-life (HQL) assessments as part of phase III clinical trials. Com- pliance rates for completing HQL instruments remain high, with a minimal amount of missing data. It is believed that this success is attributable not only to the high degree of commitment to measuring HQL by clinical trials inves- tigators, nurses, data managers, and central office ad- ministrative staff, but also to the educational process that was instituted after the development of a CTG policy for measuring HQL. From inception to May 1995, a total of 27 clinical trials with HQL assessment have been initiated or completed. In the majority of trials, the core HQL instru- ment is the European Organization for Research and Treat- ment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30). In addition to answering specific questions about HQL in these clinical trials, the trials provide the oppor- tunity to do research into the measurement of HQL. Thus, current clinical trials include research questions about the appropriate timing of assessments, the reliability and validity of the QLQ-C30 and other instruments, the role of HQL data in assessing toxicity, and the significance of the results of HQL assessments. It is anticipated that this ac- tivity not only will be a rich source of information about the effects of cancer and its treatment on HQL but also will lead to improvements in measuring HQL in oncology. [Monogr Natl Cancer Inst 1996;20:107-111 The National Cancer Institute of Canada Clinical Trials Group (NCIC CTG) formed a Quality-of—Life Committee in 1987 and adopted a policy for measuring health-related quality of life (HQL) in clinical trials in 1989 (1). The policy is that “there should be a statement about the anticipated impact on quality of life in every phase III clinical trial and whether or not quality—of—life measures will be incorporated in the protocol.” The concept underlying the policy is that HQL measurement should be an integral part of a phase III trial rather than an ac— tivity added to the trial, i.e., a companion study. The implica— tions for the implementation of this philosophy are that the central office administrative functions required for the HQL component of a trial are assigned to the same personnel who are Journal of the National Cancer Institute Monographs No. 20, 1996 responsible for the entire trial, and HQL activities are integrated into their usual activities rather than through a separate ad- ministrative structure. Thus, protocol development, form production, data management, and analysis are treated as in- tegral functions within the assigned job descriptions of the exist— ing personnel. The Quality-of—Life Committee, consisting of volunteers from cancer centers across the country, assumed the responsibility for assisting investigators with the integration of HQL assessments into the proposed trials by developing writing guidelines for protocols (1); providing educational seminars for investigators, clinical trials nurses, and data managers; and providing written instructions for collecting HQL data to be used by clinical trials personnel in the participating cancer centers. Particular attention was paid to these educational processes, since it was recognized that HQL data must be collected as completely as possible at the appropriate time points. Otherwise, the result would be missing data that would interfere with making valid conclusions. The or- ganization and functions of the Quality-of—Life Committee, the protocol—writing guidelines, and the instructions to clinical data managers have been presented in detail previously (1). There- fore, the remainder of this paper will concentrate on the HQL activities of the NCIC CTG since 1991. Clinical Trials With a Quality-of-Life (QOL) Component Current Studies Since the adoption of the policy for measuring HQL, a total of 27 NCIC-sponsored, phase III trials containing HQL assess— ments have been implemented (Table 1). They range across several anatomic sites. Five trials have been studies in symptom control and seven have involved the combined use of radiation therapy and chemotherapy or radiation therapy alone, whereas the remainder are chemotherapy trials. Brief titles of the trial *A/j‘iliarions of authors: D. Osoba, British Columbia Cancer Agency and University of British Columbia. Vancouver. Canada: J. Dancey, B. Zee, J. Myles, J. Pater, National Cancer Institute of Canada Clinical Trials Group, Kingston, Ontario, Canada. Carrespondent'e to: David Osoba, M.D., Communities Oncology Program. British Columbia Cancer Agency, 600 W. 10 Ave., Vancouver. British Colum« bia VSZ 4E6, Canada. 107 and the HQL instrument used are presented in Table 2. Five tri- Table 1- Summary “studies with HQL assessment als have been completed (two are still being analyzed), whereas three were closed because of lack of accrual. Currently, 17 trials Disease site No. Status are open and accruing patients. Breast 4 1 closed, 3 open The results from the completed trials are of interest. In ME.7, COIOFCCWm 3 A“ Open . - . , - ‘ - Genitourinary tract 2 Both open a trial comparing interferon gamma to levamisole in the ad— \ .g . \ ~ ' K . _ ' ' . Gynecologic site 4 2 closed, 2 open Juvant treatment of high—risk, surgically resected primary malig- Head and neck 1 planned nant melanoma, pretreatment global QOL predicted for Hemato'ogic site 3 1 Open. 2 Planned 4 - - , . . Lung 3 1 closed, 2 open subsequent on—treatment global QDL (2). Preliminary results Melanoma 1 Closed from two studies of the efficacy of H3T—antagonist antiemetics Sarcoma 1 Open with or without dexamethasone (SC.8 and SC.9) indicated that Symplom 00""01 5 4 closed. 1 0Pen ~ - - - T 2 27 ‘ .' 5 , _ ' patients who experienced postchemotherapy vomiting had lower 0‘" 9 dosed” 1 Open 1 planned physical, role, and social function, lower QOL, and more fatigue than did patients who did not have vomiting (Osoba D, Lee B, Table 2. Studies with HQL components (April 1995)* Disease site Symbol Brief title'l' lnstrumentsi Breast MA.5§ CMF versus CEF in patients with positive nodes BCQ MA.8|l Vr plus doxorubicin versus doxorubicin in metastatic and recurrent disease QLQ—C30 MA. I ()ll Dose—intensive chemotherapy for locally advanced/inflammatory cancer QLQ—CSO MA.l l|| Escalating FEC with G—CSF BCQ Gastrointestinal CO.7|| Adjuvant S-FU and leucovorin versus delayed therapy after resection of liver or lung metastasis QLQ-C30, SF-36 in colorectal cancer CO.9|I Adjuvant high-dose versus standard-dose levamisole + 5-FU and leucovorin in colorectal cancer QLQ—C30 CO.lOl| Immediate versus delayed 5-FU + leucovorin in asymptomatic advanced colorectal cancer QLQ—C30 Genitourinary PR.3|| Total androgen blockade i pelvic irradiation in localized carcinoma of the prostate QLQ»C30, FACT-P PR.51] Short radiation fractionation schedule for localized prostate cancer QLQ—C30 Gynecology CX.2|| Radiation i cisplatin for locally advanced squamous cell cancer of the cervix QLQ—C3t) CX.3# Cisplatin i etoposide and ifosfamide for carcinoma of the cervix QLQ—C30 OV.9** Paclitaxel in platinum-pretreated ovarian cancer QLI OV.1()|| Platinum and paclitaxel versus platinum and cyclophosphamide for advanced ovarian cancer QLQ-C30 Head and neck HN. l1] Elective neck dissection in early oral cancer QLQ—C30, SE36 Hematology HD.6|I Radiotherapy or ABVD + radiotherapy versus ABVD alone for early»stage Hodgkin’s disease QLQ—C30 LYSfl CHOP versus CHOP + G—CSF for intermediate and high-grade non—Hodgkin‘s lymphoma in QLQ—C30 the elderly MY.71I Melphalan + dexamethasone or prednisone for multiple myeloma QLQ-C30 Lung BR.8|I CODE versus alternating CAV and EP in extensive-stage small—cell lung cancer QLQ-C30 BR,9# Chemotherapy + surgery versus radiation therapy for stage III A non—small—cell lung cancer QLQ—C30 BRJOII Adjuvant Vr and cisplatin in resected non—small—cell lung cancer QLQ—C30 Melanoma ME.7'H' Human interferon gamma versus levainisole as adjuvant therapy for poor prognosis QLQ-C30 malignant melanoma Sarcoma SR.2|| Preoperative versus postoperative radiation therapy for soft tissue sarcoma SF—36, TESS Symptom control SC.8§ Ondansetron and dexamethasone in highly emerogenic chemotherapy QLQ-C30 SC.9§ Granisetron i dexamethasone in moderately emetogenic chemotherapy QLQ—C30 SC,1()# Clodrinate versus placebo for bone pain in metastatic cancer QLI SCI l§ Dolasetron mesylate versus ondansetron t dexamethasone for moderately emetogenic chemotherapy QLQ