ACTORS AFFECTING STAFFING LEVELS ND a OF Misi G PERSONNEL i | ff | . i te = i; i 3 I FE i | Lk 0 i \ ! y fim it 1 , A y 1 Ig 4 ) : 3 hy He 3 ) " 4 | I i 1 iy iil ifn ] | A i | = ACTORS AFFECTING STAFFING LEVELS IND PATTERNS OF NURSING PERSONNEL Harry D./Leving, M.A. and P. Joseph Phillip, Ph.D. BUREAU OF RESEARCH SERVICES AMERICAN HOSPITAL ASSOCIATION 7 February 1975 ——————————————————— DHEW Publication No. (HRA) 75-6 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service ® Health Resources Administration Bureau of Health Resources Development @ Division of Nursing Bethesda, Maryland 20014 Pub / b1919275 The research reported in this publication was performed under Public Health Service Contract NTH 72-4169 from the Division of Nursing, Bureau of Health Resources Development. Division of Nursing project officer is Stanley E. Siegel, Assistant Chief, Man- power Analysis and Resources Branch. DISCRIMINATION PROHIBITED—Title VI of the Civil Rights Act of 1964 states: “No person in the United States shall, on the ground of race, color, or national origin, be excluded from partici- pation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assis- tance.” Therefore, the manpower analysis program, like every pro- gram or activity receiving financial assistance from the Department of Health, Education, and Welfare, must be operated in compliance with this law. Eugene Levine, Ph.D. Scientific Editor For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 — Price §1.45 cents Stock Number 1741-00080 ii A A R 972 oD L471 FOREWORD PUBL Nurse staffing is a major concern of the Division of Nursing and its activi- ties, and publications in this area cover more than a quarter of a century. The Division has conducted and supported research on optimal staffing and other studies to assist nursing service administrators to improve the utilization of nursing personnel. A recent Division effort was the sponsorship of an ex- tensive critique of the available literature on nurse staffing methodology followed by a conference on this subject. Concern for the quality of hospital care and the increasingly high cost of this care emphasized the need for a detailed analysis of the existing data, and for development of models which could assist nursing service directors, hospital administrators, and researchers to address these concerns. The Divi- sion therefore contracted with the American Hospital Association for the data and model analyses contained in this report. Data from four sources were used: the 1970 American Hospital Association-Division of Nursing joint survey of nursing personnel employed in hospitals; the Association’s 1970 annual survey of hospitals, Health Resources Statistics, 1971; and Census of Population: 1970. The employment of nursing personnel in community hospitals was analyzed from the perspectives of demand, need, and potential of hospital care for the community. These analyses provide new perspectives for looking at the maldistribution of nursing personnel. Regression models were developed for each of six personnel categories based on the considera- tion of 21 independent variables. To facilitate their reference and use, the hospital models were categorized into three strata: (1) nonprofit/non- teaching; (2) nonprofit/teaching; and (3) profit. This study adds significantly to our understanding of the factors affecting the level and patterns of nurse staffing. Because of the size of the universe and comprehensiveness of the analyses, the several models provide a speci- ficity not previously available to guide formulation of staffing policies. We hope that future studies and efforts to improve nurse staffing will make use of this information. Sn . csv Jessie M. Scott Assistant Surgeon General Director Division of Nursing ACKNOWLEDGMENTS This study represents the contributions of several persons. Bernard Ferber, Sc.D., Director, Bureau of Research Services, American Hospital Association, is responsible for initiating the idea, for obtaining partial funding under Contract No. NIH-72-4169 from the Division of Nursing, Bureau of Health Resources Development, and for exercising overall super- vision. That this study would not have come to fruition without his initiative is perhaps the best testimony of his role. Allen Stone, Assistant Director, Bureau of Research Services, handled all matters relating to the negotiation of the contract. We are indebted to Eugene Levine, Ph.D., Chief, Manpower Analysis and Resources Branch, Division of Nursing. We have profited, in no small mea- sure, from his experience, expertise, and insight into the nursing field. Stan- ley E. Siegel, Assistant Chief, Manpower Analysis and Resources Branch, was most helpful at all times. So was Sallie Manley, Project Director, Survey of Nursing Personnel. Ranjan Raj assisted with the computer programming. Finally, we wish to thank Jessie M. Scott, Director, Division of Nursing, for her helpful suggestions at the conferences held before, during, and after the consummation of the contract. Harry D. LEVINE, M.A. P. Josepn PHirLLip, Ph.D. iv CONTENTS Page Foreword coco mek od visas mem om ww om wen ae win wie wre 008 win wid 0 die view iii Acknowledgments ......c.cocuuunnunnninncrrirtsrrrecsedesnens iv Tables «ouvir aa vii Chapter 1. The Problem ccecimisusssnnsmesmemsmenornsafonrens 1 Chapter 2. The Need for Benchmarks .............cooiiiiaae, 3 Some Preliminary Definitions ...........coouun. 3 Regional Maldistribution .........coooiiiiiannn. 4 Horizontal Maldistribution .............ccouuin. 4 Vertical Maldistribution ...................u.e. 6 Regional Maldistribution—Vis-a-vis Individual Hos- PED coccrsiininsnminrnnimrmsnrinsnetapans 7 Inter-Hospital Disparities ........ovvvvuinna... 8 Chapter 3. Universe and Data Characteristics ........coviiunnnn. 11 The Universe «cocoveevrrrrnnenes SERRE | I 11 Sources of Data cc cinrvversnimnsvimssnin dad ans 11 Chapter 4. The Editing Process and the Description of Variables ... 13 Matching and Merging Records .................. 13 The Variables .....oovvviiiiiiiiiiiiiiian., 14 Preliminary Regression Model ................... 16 Treatment of Outliers and Missing Data ........... 16 Stratification. «,;..conesinrsssnrnsrinsvnsnsnshefsns 17 Chapter 5. Definition of Nursing Hours per Adjusted Patient Day... 21 Nursing Hours ...covvvvviiiiiiiiiiiinnn., 21 Adjusted Patient Days ............cooiiia... 21 Nursing Hours per Adjusted Patient Day .......... 23 Chapter 6. The Regression Models ............ccoiiiiiina., 25 Interrelationship Among Hospital Indices ......... 26 The Interaction Between Supply Variables and Nurse x11 eT, REPS ppp ps PO pS 26 Homogeneity of Criteria .....vovvvveiuiiiiian... 27 The Models and Their Application ............... 28 An Illustrative Example ..........cooiiiiiiit, 28 The Computation of Nursing Hours per Adjusted Patient Day «om eines vn mum mn wom ss vs vs wn wl olen 30 Chapter 7. Findings: Regression Models ......ovvvvvinnennnnn 31 Categories of Explanatory Variables .............. 31 Stratum I: Nonprofit, Nonteaching Hospitals .... 32 Stratum II: Nonprofit, Teaching Hospitals ....... 35 Stratum III: For-Profit Hospitals ............... 38 Chapter 8. Summary of Findings .....ovvviiiiiiiiiiiiinnnns, 41 Regional Analysis cconvssrnvsnsnsnsssnsnsn vss ne 41 Regression Models ......ccoviiiiiiiiiiiiiinnn, 41 Inter-Stratum Variations .......ceveeeneeeneneans 41 Impact of Independent Variables .....ooovvvvenn. 42 Organizational Characteristics «vveeeeeeeusnnnnans 43 Product Characteristics +..vvevvierneenneeennns 44 Socioeconomic Characteristics «veveeeevennneenens 45 Demographic Characteristics «v.uvvveieenennnnnns 45 Supply Characteristics «.oveveuuieeeeeennnnennns 45 Chapter 9 Descriptive StatiSHOS ««vvvevrrerreeennreenvsesnrses 47 Tables os covsvunvinunnsrrsnenninenss nnn sss 06s esses yess ens 51 REJorerites +515 15515 v6 5:45 4% 04 F 08 56.5 405 5 ON $16 00s Fe © 208 010 5 430 Wik 4 0 101 Technical Appendix ........eeeeueeeeeeessesessseeseerenonnns 103 Working Data Files + cuvrnsmrrnvvninssnrnverenenrnmeasenens 105 Some Notes on the Regression Computations ......covvevenennnn 105 Some Numerical Considerations ......eeeeeereeeniieeneenannns 106 Deletion of Independent Variables .......coovvvuiiiiiinnnnn.. 107 Stopping Rule for Elimination of Independent Variables .......... 107 Dropping Outliers ....vvviiiiiiiiiiiiieieinninnsnnneneannns 108 Combination of Strata +...vveevieiiiieiiierenseernneennneens 108 Computations Based on Less than a Full Year’s Data ............ 109 Application of Regression Models «....ovviviveiiiiiiiinnennn. 110 vi TABLES Table 1. Regional distribution of total nursing hours: continental United SIRES wu 000 00 300 50 wom bs wi Sos 50 wi wit 0 0s 000 # 20 i 100 4 RN 2. Ranking of States by total nursing hours ..........covvinnn. 3. Regional distribution of nursing hours by individual categories: continental United States .....oovvveiiiiiiieniiiiinenenns 4. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: Total nursing hours per adjusted patient OY LX) «win wn wine wou win win 0 450 050% 00 000 04 0 500 ww in wR won ik k 5. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: RN hours per adjusted patient day (Y2) .. 6. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: LPN hours per adjusted patient day (Y3) . 7. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: Aides, orderlies and/or attendant hours per adjusted patient day (Ty) csssvsnnravassssnssnnnnsnes . 8. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: General duty nurses hours per adjusted patient day (Y5) voveereenrninnneerenernnnssseesonnsnnns 9. Regression model for nonteaching, nonprofit hospitals, stratum I. Dependent variable: General duty and head nurses hours per adjusted patient day (Yg) veeevereneennennninnennnonennns 10. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: Total nursing hours per adjusted patient AY (V3) savniinosvseinssrsnoeiiaomninsensnsnnavosvsns 11. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: RN hours per adjusted patient day (Ys) .. 12. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: LPN hours per adjusted patient day (Y3) 13. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: Aides, orderlies and/or attendants hours per adjusted patient day (Ya) vovervirvienniniininnnnnnnns 14. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: General duty nurses hours per adjusted Patient Qay LY6Y viviusw iin vs vans vs vor va nan smn sim ven os 15. Regression model for teaching, nonprofit hospitals, stratum II. Dependent variable: General duty and head nurses hours per adjusted padent day (Yo) «cicconsnsnvorrssssannssseessns Page 52 53 54 56 57 58 59 60 61 62 63 64 65 67 Table 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. . SMSA (coded 1) versus non-SMSA (coded) : Variable Xi; .... 39. 40. 41. 42. 43. Regression model, for-profit hospitals, stratum III. Dependent variable: Total nursing hours per adjusted patient day (Y;) .. Regression model, for-profit hospitals, stratum III. Dependent variable: RN hours per adjusted patient day (Y2) ........... Regression model, for-profit hospitals, stratum III. Dependent variable: LPN hours per adjusted patient day (Ys) ......... Regression model, for-profit hospitals, stratum III. Dependent variable: Aides, orderlies and/or attendants hours per adjusted patient day (Ya) «ovine Regression model, for-profit hospitals, stratum III. Dependent variable: General duty nurses hours per adjusted patient day {V5) sonornrmmuiminromins i usmosntorhid seein sume weno Regression model, for-profit hospitals, stratum III. Dependent variable: General duty and head nurses hours per adjusted pa- HEL BAY (Ya) + 00 v 006 00 4 om wis 519% 5% mh 3 wk 010% 05 baw 500 6 48 510 @ 310 Regression coefficients for use in computing estimated values of nursing hours per adjusted patient day. Stratum I ........... Regression coefficients for use in computing estimated values of nursing hours per adjusted patient day. Stratum II ......... Regression coefficients for use in computing estimated values of nursing hours per adjusted patient day. Stratum III ......... Total number of facilities: Variable Xj «.ovvurrnnnnnn.n. Number of advanced technology facilities: Variable Xs ....... Number of advanced technology facilities as a fraction of total facilities: Variable Xg ovo iii Total admissions in 100,000’s: Variable Xg + ovvveeeennnnn... Length of stay in days: Variable X5 .........ccooviuue.... Occupancy rate in proportions: Variable Xg ...ovvuvvuvnn... Bassinets per statistical bed: Variable X7 .............o..u... Births per statistical bed: Variable Xg ......cc.vvvvnn..... Emergency visits per admission: Variable Xg ............... Total visits per admission: Variable X10 «vvvvvrerrnnrnnnnn. Surgical operations per admission: Variable Xj; ............ Statistical beds: Variable Xia ..oovvvuviiiiiiiinii..... Adjusted patient days in 100,000’s: Variable Xi3 +vveuuun.... Total nursing hours per adjusted patient day: Variable Y; ... RN hours per adjusted patient day: Variable Ya ............ LPN hours per adjusted patient day: Variable Yg ........... Aides, orderlies and/or attendants hours per adjusted patient day: Variable Yq ..vvvnniiii iii General duty nurses hours per adjusted patient day: Variable WE 00am nr msn ann am nh Br 4 en 0h hk 4g mw ve we Page 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 Table Page 44. General duty and head nurses hours per adjusted patient day: Variable Yq «ovr eee 96 45. Simple correlations among independent variables X; to Xjy: SIT I ue mom vans wome svsnom en esos sons sows ews 97 46. Simple correlations among independent variables X; to Xjq: Stratum II «oui i i ee ea 98 47. Simple correlations among independent variables X; to Xj4: Stratum TIL ..iniivimiisivrnonssnsmorisnirsivensavonsns 99 48. Ranked beta coefficients .....ovvuin iin iii iii, 100 ix Chapter 1 THE PROBLEM Studies focusing upon the internal organizational characteristics of hos- pitals are particularly apropos at a time when the mounting cost of hospital care has become a matter of public concern. Like all productive enterprises, hospitals employ labor and nonlabor inputs for the production of their outputs—hospital care. Current estimates place labor’s share somewhere between 50 and 60 percent of the total cost of hospital care. Nursing personnel, that is, professional and practical nurses and ancillary personnel such as aides, orderlies, attendants, and ward clerks, constitute about one- half of the total personnel employed by a typical hospital. This means that nursing staff account for about 25 to 30 percent of total hospital expendi- ture. There can be no dispute, therefore, that the “level” and “pattern” of nurse staffing are major determinants of hospital cost. By level we mean the personnel per unit of output (adjusted patient day); and by pattern we mean the personnel mix. If the level is high, costs are likely to be high, and if the pattern is characterized by a preponderance of higher level per- sonnel, costs are likely to be high. These are, of course, tentative statements, because they are subject to an important qualification: both the level and pattern of nurse staffing are related to the quality and adequacy of care. Nevertheless, they have essential validity. Increasing the employment of nursing personnel will, after a point, lead to diminishing returns. It is also well known that there is considerable scope for “upward” substitution of hospital personnel, that is, the substitution of lower level personnel for the performance of certain tasks traditionally done by higher level per- sonnel. In the employment of nursing personnel, hospitals should strive toward optimality. That is to say, the number and mix of nursing personnel should be such as to maximize productivity consistent with quality. However, what is optimum for one hospital may not be optimum for another. A large multispecialty hospital, for instance, may have a different optimum than a small hospital providing a narrow range of services. Similarly, the physical layout of the hospital, the degree of automation, institutional routines, pre- vailing wage rates, turnover rates and the amplitude of fluctuations in patient arrivals are some of the factors that might affect optimality. In short, the conditions surrounding the operation of individual hospitals are so diverse that a single optimum would be inappropriate. Ideally, optimum nurse staffing should be determined on a hospital-by- 1 hospital basis, through indepth studies conducted by a team of industrial engineers, time, motion and fatigue study specialists, personnel experts, and so forth. Alternatively, one may identify key organizational and product characteristics which affect optimality and cluster hospitals on the basis of these characteristics.! Optimum nurse staffing for each cluster could then be established through indepth studies conducted on sample hospitals drawn from each cluster. The preceding approaches are extremely costly and time consuming. In addition, the norms established may become obsolete when the conditions surrounding the operation of hospitals change. A more feasible alternative— the one that is attempted in this study—proceeds as follows: * Hospitals are stratified into three broad groups based on the following characteristics: teaching status and control. The purpose of this stratification—which has been established after detailed statistical in- vestigations, discussed later—has been to bring about a measure of comparability among hospitals within each stratum. * Key organizational, product, socioeconomic, demographic, and supply characteristics that influence staffing levels and patterns are identified from an original group of 21 characteristics by a “step-down” pro- cedure. * For each stratum, least-square regression equations, which explicitly “allow for” differences in the key characteristics, are constructed for the composite nursing personnel, as well as for each category of per- sonnel. The limitations of this approach are discussed later. Suffice it to say, at this point, that the “norms” derived from the regression equations are not optima in the rigorous sense in which they are described earlier. 1A major research project to cluster the Nation’s short-term, general hospitals is under way at the American Hospital Association. The clusters derived in this project might well serve as substantially improved strata for a study such as the present. 31t should be mentioned parenthetically that since teaching status and control are strongly correlated with bedsize, the stratification has led, to some extent, to a grouping based on size as well (see table 36, p. 88). 2 Chapter 2 THE NEED FOR BENCHMARKS There is general consensus that the overall supply of nursing personnel continues to fall short of overall demand. (1,21,8,7)! This shortage is accentuated by regional maldistribution of available nursing personnel. Some Preliminary Definitions The question that arises immediately is: Supply in relation to what? There are several perspectives from which supply of nursing personnel may be viewed: need, want, demand, and potential. The first three concepts are due to Jeffers, et al. (10) the last one is ours. Jeffers et al. define “need,” “want,” and “demand” as follows: Need —*“The quantity of medical services which expert medical opinion believes ought to be consumed over a relevant time period in order for its members to remain or become as ‘healthy’ as is permitted by existing medical knowledge.” Want—*“The quantity of medical services which its members feel they ought to consume over a relevant time period based on their own psychic perceptions of their health care needs.” Demand —“A multivariate functional relationship between the quantities of medical services that its members desire to consume over a relevant time period at given levels of prices of goods and services, financial resources, size and psychological wants of the population as reflected by consumers’ tastes and preferences for all goods and services.” Our definition of “potential” is as follows: Potential —The hospital management’s perception of the actual or an- ticipated demand by the community for medical services over a relevant time period. There are, of course, some interrelations and interdependence among these concepts. Demand, for example, is determined, to some extent at any rate, by the interaction between the consumers and the professionals. Similarly, there is a strong correlation between “want” and demand. In fact, demand is that part of want backed by the ability and willingness on the part of Numbers in parentheses refer to literature cited in reference list, p. 101. 3 the population to consume. Nevertheless, analyzing the supply or availability of health manpower in relation to need, want, demand, and potential can provide new perspectives on maldistribution of health manpower. Unfor- tunately there are no measures that faithfully reflect the need, want, de- mand, and potential parameters. We have, therefore, used three surrogates: adjusted patient days, statistical beds, and civilian resident population. Adjusted patient days represent the actual consumption of health care by the community. Consequently, this measure is indicative of the demand for health care by the community. Need, as previously defined, is independent of consumers perceptions. It is, in addition, independent of economic and convenience considerations such as financial resources, price, accessibility and so forth. Therefore, it will be correct to say that, assuming no major differences in the demographic characteristics of two population groups, the larger of the groups need more health care on the aggregate. The civilian resident population can thus be regarded as a surrogate for need. Hospital management provide beds on the basis of their perceptions about the actual or anticipated demand for health care by the community. There- fore, statistical beds may be regarded as a surrogate for potential.? Regional Maldistribution According to the theory of industrial agglomeration, industries tend to concentrate in areas where they can reap internal and external economies. (9,13,30) This implies that there are marked regional variations in the price and/or availability of men, machinery, material, transportation net- work, market potential, and so forth. In the context of the hospital industry which is labor-intensive, certain regions possess definite advantages over others with respect to the price and availability of labor. However, since hospitals provide a unique service, locational advantages are generally subordinated to social welfare considerations. In other words, hospitals are, more often than not, located in close proximity to the community they service regardless of whether or not the locations are advantageous from the standpoint of price and availability of labor. This aspect of the hospital industry explains, in part, the persistence of regional maldistribution of nursing personnel. This maldistribution appears along two dimensions which we may call horizontal maldistribution and vertical maldistribution. Horizontal Maldistribution Horizontal maldistribution is a condition where there are significant spatial disparities in the supply of nursing personnel. For example, one 2 We have used no surrogate for want. However, since want is strongly correlated with demand, patient days comes closest to being a surrogate for want. 4 region may have a more plentiful supply of nursing personnel than an- other. Table 1 (p. 52) presents, by regions and States, total nursing hours ex- pressed on a per-patient-day, per-bed, and per-population basis. Two ob- servations must be made at the very outset. First, the extent of regional differences is substantial regardless of the perspective from which one looks at it. Second, the distributional pattern that emerges when viewed from one perspective is somewhat different from the distributional pattern that emerges when viewed from another perspective. Thus, the perceptions of hospital management, the consumers and the professionals do not always run parallel. Total nursing hours per adjusted patient day are highest in Arizona (7.17), followed by Washington (7.00), Oklahoma (6.85), and New Hampshire (6.82) in that order. We may say that these States have more hospital-based nursing personnel in relation to the demand generated by the consumers. Utah has the lowest ratio (5.25), followed by New York (5.32), Montana (5.35), and Pennsylvania (5.40) in that order. Total nursing hours per statistical bed are highest in Arizona (2,264), followed by Rhode Island (2,186), Georgia (2,155), and New Hampshire (2,058). These States have more hospital-based nursing personnel in rela- tion to the potential for hospital care as perceived by the management. Montana has the lowest total nursing hours per statistical beds (1,389), followed by Wyoming (1,433), North Dakota (1,558), and Iowa (1,570) in that order. The District of Columbia stands all by itself with the highest ratio of total nursing hours per civilian resident population (11.50). Perhaps this geographic entity should not be compared with the States because, unlike the States, it does not have rural areas where the nurses-per-population ratios are generally low. Among States (excluding the District of Columbia), North Dakota has the highest ratio of total nursing hours per civilian resi- dent population (9.04), followed by Nebraska (8.22), Vermont (7.89), and Kansas (7.83) in that order. We may say that these States have more hospital-based nursing personnel in relation to need. Utah has the lowest ratio (4.29), followed by Louisiana (4.67), and New Mexico (5.40). It is interesting to note that these are also States with very low per capita in- come. Aside from the regional disparities brought out by table 1, the juxta- position of three ratios provides interesting insights into the extent to which the availability of nursing personnel viewed from the perspectives of de- mand, potential, and need run parallel, and the extent to which it diverges. To facilitate comparison, we have presented an additional table (table 2, p. 53) in which the ratios given in table 1 are transformed into rank values. That is, the State with the highest ratio is assigned a rank of 1, the State 5 with the next highest ratio is assigned a rank of 2, and so on.? In the last column of table 2 are provided the average absolute differences among the three ranks. This value is indicative of the degree to which the availability of nursing personnel in a State differs when viewed from the perspectives of demand, potential, and need. We have also given, at the bottom of table 2, three values, pap, pac; and pp. These are Spearman’s rank correlation coefficients. For example, p ab shows the degree of agreement, across States, between total nursing hours per adjusted patient day and total nursing hours per statistical bed. The larger these coefficients, the greater the agree- ment. (A negative coefficient indicates disagreement.) Turning to the last column of table 2, we find that there are several States with some degree of tripartite agreement. They include South Carolina, Colorado, Florida, Missouri, Massachusetts, North Carolina, and Illinois. However, disagreement is more common. North Dakota, Washington, Nebraska, New Mexico, Louisiana, and Iowa lead the rest in this respect. Turning to the rank correlation coefficients, we find that the only two ratios which show a measure of agreement across States are total nursing hours per adjusted patient day and total nursing hours per statistical beds. In other words, if we consider the Nation as a whole, the availability of nurs- ing personnel in relation to demand is, to some extent, in agreement with the availability of nursing personnel in relation to potential. However, there is no such agreement between demand and need or between potential and need. In fact, the negative values of p 4 and pp. suggest that disagree- ment is more typical than agreement. We may also note certain instances of major divergence in the availability of nursing personnel viewed from different perspectives. In certain States, notably Louisiana, Washington, Georgia, Arizona and New Mexico, the availability of nursing personnel in relation to demand is high, while the availability in relation to need is low. A contrary situ- ation is observed in States such as Pennsylvania, Minnesota, Iowa, and Vermont. In States such as Georgia, Delaware, Arizona, and California, the avail- ability of nursing personnel in relation to potential is high, whereas, the availability in relation to need is low. On the other hand, in some States, notably North Dakota, District of Columbia, Kansas, and Iowa, the avail- ability of nursing personnel in relation to need is high, while the avail- ability in relation to potential is low. Vertical Maldistribution Vertical maldistribution refers to spatial disparities in personnel mix. For example, one region may have a higher ratio of professional nurses to total nursing personnel than another region. Such spatial disparities in personnel 3 When tied ranks occur, each of them is assigned the average of the ranks which would have been assigned had no ties occurred. 6 mix should be construed as maldistribution in the sense that a rearrange- ment of personnel mix will improve overall productivity consistent with quality of care. Table 3 (p. 54) presents a breakdown of total nursing hours by specific categories. Our focus of attention is registered nurses (RN's), licensed practical nurses (LPN’s), and ancillary (i.e., aides, orderlies, and/or attendants). These three categories constitute the total in the first column of the table. We have also included in the table, two additional categories based on appointment status—general duty nurses, and general duty and head nurses. As can be seen from table 3, personnel mix shows substantial regional differences. The proportion of RN’s relative to all nursing personnel is highest in Massachusetts (58 percent). All New England and Middle Atlantic States, Washington, and Colorado have higher proportions of hospital- based RN’s. The State with the lowest proportion of RN’s is Arkansas (23 percent). All West Central and East South Central States have a uniformly low proportion of RNs. Idaho ranks first in terms of LPN employment (38 percent). Arkansas and Texas, each with 35 percent LPN’s, tie for second place. The State with the lowest proportion of LPN’s is Wyoming (11 percent). Indiana and Kansas, each with 14 percent, are the next two States with the lowest proportion of LPN's. Regional disparity in the employment of ancillary personnel is as wide- spread as that observed for RN’s. Generally speaking, States with low pro- portions of professional and practical nurses are the States with high proportions of ancillary personnel. With 53 percent, Oklahoma ranks first in the employment of ancillary personnel. Kansas (51 percent) and Wy- oming (50 percent) rank second and third. The State with the lowest proportion of ancillary personnel is Rhode Island (18 percent), followed by Washington (22 percent), Vermont (22 percent), and Idaho (23 percent). Hospitals in Massachusetts have the highest ratio of general duty nurses (43 percent). Connecticut, Vermont, Pennsylvania, and Colorado, each with 40 percent tie for second rank. Arkansas has the lowest ratio of gen- eral duty nurses (11 percent). Generally speaking, West Central and East South Central States have low ratios of general duty nurses. As one might expect, the employment of general duty and head nurses follows, more or less, the same pattern as that observed for general duty nurses. Regional Maldistribution Vis-a-Vis Individual Hospitals From the standpoint of individual hospitals, regional maldistribution (both *The three categories may not add up to the total due to minor reporting incon- sistencies. horizontal and vertical) is an uncontrollable factor to which they should, willy-nilly, adjust themselves.® Any attempt to bring about a more optimal regional distribution of nursing personnel should, therefore, be initiated at local, State, or national levels. It is hoped that a meaningful outline of the distribution of nursing personnel across the country will emerge from the present study, so that steps designed to bring about a more judicious utilization of available nursing personnel can be initiated. Section 222 (b)(3) of the Health Manpower Act of 1968 is a legislative mandate in this direction. The act authorizes up to 100 percent cancellation of nursing student loans at the rate of 15 percent per year for services as professional nurses in a public hospital in an area with a substantial shortage of nurses. To enable the Division of Nursing to administer the act, a survey of nursing personnel in hospitals is conducted biennially by the American Hospital Association. Eligibility for loan cancellation is determined on the basis of some cut-off points such as the mean or the median staffing levels for various categories of hospitals. The present study will, hopefully, provide a more adequate basis to determine which hospitals meet the standards of eligibility set forth in the Health Manpower Act. Inter-Hospital Disparities The staffing levels and patterns of hospitals within a region or within comparable regions may exhibit variations. As pointed out earlier, a part of these variations may be “justifiable” or “warranted” in the sense that they reflect differences in organizational, and/or product characteristics of hospitals. However, variations may also arise from factors which by their nature could ideally be corrected. For example, a hospital which has an inefficient staffing policy may have a staffing level or pattern which is out of line with what may be appropriate for that hospital. Another hospital may have an undesirably low staffing level or a staffing pattern characterized by too high a proportion of lower level personnel if that hospital lacks the financial resources or the environment necessary to attract the needed personnel. Still another hospital may have an undesirably high staffing level, or a “top heavy” staffing pattern if it does have the financial resources and the environment to attract professional nurses. Any rational approach toward optimum staffing should initially draw a clear distinction between “justifiable” variations as opposed to variations that call for corrective action. If we examine an individual hospital, the chances are that the staffing level and/or pattern of this hospital exhibit variations from the stratum averages. These variations may be due to: a. “justifiable” factors, ®We are making an assumption that hospitals operating in regions of shortage are unable to attract more personnel due to financial and/or environmental reasons. 8 b. factors that demand corrective action, or c. some combination of the two. If we could decompose an observed variation in this manner, we will have a basis to identify hospitals which are “deviant” or “aberrent” from the standpoint of staffing level and/or pattern. This line of reasoning is the underlying rationale for the regression models developed in this study. TT Ee a rT Chapter 3 UNIVERSE AND DATA CHARACTERISTICS The Universe The universe on which this study is based is defined as “short-term, gen- eral hospitals in the continental United States registered with the American Hospital Association.” Thus, the universe does not include the following categories of hospitals: Nonregistered hospitals Federal hospitals Long-term hospitals Special hospitals Osteopathic hospitals Hospitals outside the continental United States Nonregistered hospitals are excluded for obvious reasons—their data are not available to the American Hospital Association. Federal hospitals’ definition of “short-term, general” is not consistent with the definition used by the Association. This will be clear from the fact that, in 1970, the average length of stay in Federal, short-term, general hospitals was 18.3 as opposed to 8.2 in non-Federal, short-term, general hospitals. The type and intensity of services provided by long-term and special hospitals (and hence, the level and pattern of nurse staffing) are quite different from those of short-term, general hospitals. Our preliminary analysis showed that the conditions surrounding the operation of hospitals outside the continental United States are so different from their continental counterparts that they constitute, so to speak, a different universe. There were, initially, 5,453 hospitals in the universe. Of these, 1,653 hospitals had to be dropped for one reason or another (discussed later). Thus, the models in the present study are based on 3,800 hospitals. Sources of Data 1. The annual survey of hospitals, 1970 The annual survey of hospitals is conducted to gather the necessary *These include the following categories: Maternity; Eye, Ear, Nose, and Throat; Children’s; Orthopedic; Chronic; All other. 11 data for the publication of the Guide Issue of Hospitals, Journal of the American Hospital Association. The hospital characteristics used as independent (explanatory) variables in this study were drawn from this source. 2. The survey of nursing personnel, 1970 This survey is conducted to gather data on nursing personnel employed in the Department of Nursing Services of hospitals. The data are ob- tained in some detail such as by category of personnel, employment status, that is, full time or part time, number of hours in a regularly scheduled work week and so forth. The dependent (criterion) variables used in this study were developed from this source. 3. The U.S. census, 1970 Selected demographic and socioeconomic characteristics of the area in which hospitals are located were obtained from the 1970 census. (16, 17) These data were gathered as potential independent variables. 4. Health Resources Statistics, U.S. Department of Health, Education, and Welfare, 1971. Data reflecting the supply of nursing personnel were extracted from the following tables of this publication: Table 95: Location of registered nurses according to activity status and ratio to population: 1966. Table 101: Location of licensed practical nurses according to activity status and ratio to population: 1967. Table 106: Location of aides, orderlies, and attendants employed in hospitals in relation to hospital beds: 1968. These data were gathered as potential independent variables. An explana- tion as to why data pertaining to 3 different years were used is in order. Data pertaining to the three categories of personnel are not available for the same year. However, a preliminary analysis using data reported in Health Resources Statistics, 1965, 1968, 1969, and 1971 showed that the relative standings of States with respect to nursing personnel per 100,000 population have remained, more or less, unchanged. 12 Chapter 4 THE EDITING PROCESS AND THE DESCRIPTION OF VARIABLES Since the data used in this study were drawn from a variety of sources, it was necessary to gather all data for a given hospital into a single record in our magnetic tape file. Furthermore, several measures had to be refor- matted, standardized, or synthesized; the problems introduced by missing information and the presence of data vectors and/or data elements which looked spurious had to be resolved; and a detailed statistical procedure to establish a manageable number of hospital strata had to be applied. We have gone to considerable lengths to retain as many hospitals and as much data as possible in a manner that minimizes the potential threat to validity and generalizeability occasioned by selective attrition, and to develop an optimum stratification scheme. Indeed, a substantial part of our total re- search effort has gone into this phase of the study. An outline of the major steps involved follows: Matching and Merging Records The data from the annual survey of hospitals, 1970, and the survey of nursing personnel, 1970, were available on two source files of magnetic tape. It was necessary to match records from these files in order to extract the data we required. At this point, a variety of computations were made to standardize and synthesize variables to yield an output data set. Our next task was to extract and assemble into suitable form the required information from the U.S. Census, 1970, and Health Resources Statistics, 1971, so that this information could be merged with the data set developed earlier. In the case of the census data, the smallest area that was prac- tically usable was the county. In other words, all hospitals in a county were assigned the values of variables for that county. There were some exceptions, however. Baltimore (Md.), St. Louis (Mo.) and several munici- palities in the State of Virginia are designated as “independent cities.” In these instances, the county nearest to the municipality was determined, and the data pertaining to the county and the municipality were pooled. This approach is consistent with the coding on the annual survey of hos- pitals and thus permitted appropriate demographic/socioeconomic data to be assigned to hospitals. Washington, D.C. was treated as an independent geographic entity. In the case of the data derived from Health Resources 13 Statistics, 1971, the smallest area that was practically usable was the State. In other words, all hospitals in a State were assigned the values of variables for that State. Having assembled the data from the U.S. Census, 1970, and Health Resources Statistics, 1971, they were merged with the data set already developed. At this stage our data set contained information on employment of nursing personnel, hospital characteristics, demographic and socio- economic characteristics, and supply characteristics. The Variables This section presents the variables used in the study including their defini- tions, wherever necessary. The order in which these variables are numbered is maintained consistently throughout the study. The symbols in parentheses below denote the sources from which the variables are derived: S = survey of nursing personnel, 1970, A = annual survey of hospitals, 1970, C = U.S. Census, 1970, H = Health Resources Statistics, 1971. The Dependent or Criterion Variables Y:: Total nursing hours per adjusted patient day (S+A) Y2: RN hours per adjusted patient day (S+A) Ys: LPN hours per adjusted patient day (S+A) Ys: Aides, orderlies and/or attendants hours per adjusted patient day (S+A) Ys: General duty nurse hours per adjusted patient day (S+A) Ys: General duty and head nurse hours per adjusted patient day (S+A) The Independent or Explanatory Variables Xi: Total number of facilities and services (A) Xo: Number of advanced technology facilities (A) These are: Intensive care unit Intensive cardiac care unit Open heart surgery facility X-ray therapy facility Cobalt therapy facility Radium therapy facility Radioisotope facility Organ bank Xs: Number of advanced technology facilities as a fraction of total facilities (A) X4: Total admissions (A) 14 Xs: Length of stay (A) Xe: Occupancy rate (A) Xz: Bassinets per statistical bed (A) Xs: Births per statistical bed (A) Xs: Emergency visits per (total) admission(s) (A) Xi0: All visits per (total) admission(s) (A) Xi1: Surgical operations per (total) admission(s) (A) Xj2: Statistical beds (A) Xis: Adjusted patient days (A) X14: SMSA (Coded as 1 if the hospital is located in an SMSA; coded 0 . otherwise) (C) Xi5: Percent of population under 18 years of age (C) X16: Percent of population 65 years and over (C) X17: Number of RN’s per 100,000 population (H) Xig: Number of LPN’s per 100,000 population (H) X19: Number of aides, orderlies and/or attendants per 100,000 population (H) Xa9: Per capita income (C) X21: Percent of families below poverty level (C) The manner in which nursing hours and adjusted patient days are com- puted is explained in chapter 5. “Statistical Beds” is a weighted average of the number of beds. The weighting is by days that those beds were available. If a hospital has experienced no bed changes during the reporting year, statistical beds equally total beds. If there were changes, a weighted average is obtained. For example, assume that a hospital had 100 beds at the beginning of the year. On July 1 it added 10 beds, but on October 1 it removed 40 beds. Then the statistical beds for the year will be: 100 + 6/12(10) — 3/12(40) = 95 beds? Except in those cases where a hospital has made, or intends to make significant changes in its bed complement during a reporting year, we sug- 1 An alternative formula is: n statistical beds = Yves. 1 where Bt: number of beds available during period t. We: period t expressed as a fraction of total period (usually 12 months) Thus, for the present example: Statistical beds = 100(6/12) + 110(3/12) + 70(3/12) = 50 + 27.5 + 175 = 95 beds 15 gest that total beds be used in place of statistical beds. This is a matter of convenience in making all the computations required both for standardizing variables and for use in place of Xjs3, statistical beds, in our models. Average daily census is computed as inpatient days divided by number of days in the reporting period, i.e., 365. Occupancy rate is computed by the formula: average daily census = statistical beds. It may also be computed by the formula: patient days + statistical beds X 365. Average length of stay may be computed in one of two ways. Where dis- charges and discharge days are available, average length of stay may be computed by the formula: discharge days =~ discharges. Where these data are unavailable, the following approximate formula may be used: inpatient days + admissions. Preliminary Regression Model From the data set developed, hospitals which had complete information on all independent variables and the composite dependent variable, namely, total nursing hours per adjusted patient day, were segregated. There were about 2,800 such hospitals. A preliminary regression model was developed from the data of these hospitals. The purpose of this step was to ascertain whether any of the independent variables could be eliminated at this stage. However, since all independent variables showed promise, they were re- tained. Treatment of Outliers and Missing Data Researchers who attempt to apply sophisticated techniques to survey data almost invariably come up against a vexatious problem. The nature of the rapport between the surveying institution and the responding units vary, and this variation is reflected in the extent of cooperation extended by the latter. Further, the responding institutions’ standards of record-keeping vary; so does the degree of care exercised by them in submitting data. All these lead to marked unevenness in the quality of the data assembled (assuming that the response rate meets acceptable standards). While this unevenness appears in a variety of forms, the following are typical: (1) the data vector of a respondent, when considered in its entirety, seems spurious, (2) the data vector of a respondent has certain elements which seem spurious, and (3) the data vector of a respondent has one or more missing elements. Each of these represents an almost intractable problem: inclusion of spurious values distorts the results of the study, and the missing values are not permissible in most sophisticated techniques. The best solution to the problem, of course, is to obtain fresh data from the respondents concerned. However, this is often not possible. Considera- tions of time and effort apart, impugning the veracity of a respondent is 16 certainly not politic. An alternative sometimes employed is to weed out the respondents of variable(s) in such a fashion that the researcher will have a sub-sample which is complete. However, this approach is unsatisfactory for several reasons: first, the detection of spurious data vectors and data elements is by no means easy, especially when the number of respondents and data elements are large. Second, the elimination of respondents, if carried too far, would either leave a residue too small for meaningful analysis, or would result in selective attrition so that the residue would be a biased sub-sample. Finally, the variable(s) that must be eliminated may be crucial to the purpose of the study. We shall, in what follows, outline the procedure applied in the present study which would, hopefully, be of some use to others who face similar situations. Independent variables 4 through 13 and dependent variables Y; through Ys being continuous, permit frequency distributions. These were made up for each of these variables. For each such distribution, hospitals falling into the upper and lower tails were identified. Based on the shape of the frequency distribution and on judgment, anywhere from one-half of 1 percent to 5 percent in both tails were selected for scrutiny. Questionnaires from these hospitals were located and examined visually. Using a combina- tion of judgment, a priori reasoning, and certain consistency tests such as the interrelationships between admissions, discharges, average daily census, and patient days, we were able to make judgments about the quality of data. In each case, we accepted a datum, rejected a datum, or rejected the entire hospital. When a datum (or data) was (were) rejected, and the remaining data for that hospital were judged acceptable, the hospital was retained with the rejected datum (or data) expunged. At this stage there were 5,453 hospitals in the universe. Stratification In an attempt to arrive at an optimal stratification scheme, hospitals were initially classified along three dimensions: control, teaching status, and bedsize. A three-digit code identified the stratum to which each hospital belongs: The first digit indicated the type of control. 1 signified a government hospital, municipal, county, or State. 2 signified a nongovernment, nonprofit hospital. 3 signified a for-profit hospital. The second digit indicated teaching status. 1 designated a teaching hospital. 0 designated a nonteaching hospital. The third digit designated bedsize. 0 indicated 0 to 20 beds. 1 indicated 21 to 40 beds. 17 2 indicated 41 to 60 beds. 3 indicated 61 to 80 beds. 4 indicated 81 to 100 beds. 5 indicated 101 to 140 beds. 6 indicated 141 to 200 beds. 7 indicated 201 to 300 beds. 8 indicated 301 to 400 beds. 9 indicated 401 or more beds. The resulting stratification matrix is displayed below. Non-Government Government Nonprofit For-profit Bedsize [Code Non- Teach. Non Teach. Non- Teach. Totals teach- . teach- . teach- : : ing . ing . ing ing ing ing O« 20.0004 0 78 —_ 52 — 59 —_— 189 21- 40...... 1 384 —_ 384 1 200 —_ 969 41- 60...... 2 329 —_ 348 1 163 1 842 61- 80...... 3 183 —_ 288 1 72 1 545 81-100...... 4 129 _ 258 4 78 —_ 469 101-140...... 5 146 3 355 9 69 —_ 582 141-200...... 6 105 12 364 43 46 — 570 201-300. ..... 7 53 22 281 177 27 1 561 301-400...... 8 15 25 96 193 3 —_ 332 400+ .....00e 9 7 88 35 263 a 1 394 Totals........ 1,429 150 2,461 692 n7 4 || 5453 Out of 60 possible strata, there were 27 with more than 20 hospitals having usable data. Out of the original 5,543 hospitals that met our initial criteria, there were 791 for which data on nursing hours were completely unavailable. These were dropped, perforce, leaving us with 4,662 hospitals having usable data. Then we singled out the “larger” or “better” strata as those in which there were at least 60 hospitals with one or more nursing variables present and for which all other variables were also present. There were 17 such strata. In these strata there were additional hospitals where only a few variables were missing. Variables Xi4 to Xo; were, of course, always present. If half or more of the first 13 variables were present in hospitals for only this group, these were deemed salvageable. Strata averages were put into the places where data were missing for such hospitals. It should be emphasized that this was done only where such averages are based upon 60 or more hospitals and only for the first 13 variables in our list. For strata with 20 to 60 hospitals, we took only those which had complete data for the 18 first 13 variables. In all cases, availability of one or more of the nursing hours variables was sufficient to permit us to consider such a hospital on the other criteria. The number of hospitals that were salvaged varied from stratum to stratum. Also, some variables were missing more often than others. In any case, the group of hospitals that were salvaged was a small portion of any stratum. This operation gave us a total of 3,800 hospitals in 27 strata, and left 1,653 hospitals which were not usable. It should be noted at this point that there will be some small variations among the numbers of hospitals and the nursing data pertaining to them in the various tables in the results section. This is due to the fact that a few hospitals had some of the nursing hours variables missing. These hospi- tals have been included wherever it was feasible and appropriate, but they do unbalance certain totals, In reviewing the distribution of hospitals in the strata described above, there are two areas where an obvious bias is seen. One concerns the data for the for-profit hospitals. The proportion of missing data is larger for these hospitals than for the other two control categories. Also, this problem seems to be somewhat more severe in the smaller bedsize classes. The second concerns the smallest bedsize class, 0 to 20 beds, where a disproportionate amount of data are found to be missing. This is true for all three control classes. Such a large number of strata (27) is operationally undesirable. In order to make our results fairly easy to use, computationally and otherwise, and reasonably understandable for an individual, say a hospital administrator, it seemed quite necessary to decrease the number of strata drastically. In order to combine strata, we adopted the criterion that if our models were alike in two strata, or two groups of strata, we would combine them; if not, we would not combine them. This criterion is formalized in the appendix. We have compared a sizable number of strata and attempted alternative ways of collapsing strata. In those instances where the number of hospitals within the strata were large, formal statistical tests served as a guide. How- ever, the final decisions were made on the basis of a combination of data characteristics and our knowledge as to what represents a traditional group- ing in the hospital system. The three strata finally established for the study are as follows: Stratum I: Nonprofit, Nonteaching Hospitals: This stratum is com- posed of 2,988 hospitals of all sizes, although there is a preponderance of small and medium size hospitals. The average bedsize is 109. Stratum Il: Nonprofit, Teaching Hospitals: There are 633 hospitals in this stratum. Of these, 52 are large (over 400 beds) governmental hos- pitals, and the remainder nongovernmental hospitals with 141 or more beds. The average bedsize is 421. 19 Stratum Ill: The For-Profit Hospitals: There are 179 hospitals in this stratum, all of which are small or moderate in size. The average bedsize is 64. 20 Chapter 5 DEFINITION OF NURSING HOURS PER ADJUSTED PATIENT DAY Throughout this study staffing levels of nursing personnel are expressed as “hours per adjusted patient day.” It is, therefore, necessary to explain, in some detail, how this measure is derived. Nursing Hours In the 1970 survey of nursing personnel, hospitals were asked to report the number of full-time and part-time nursing personnel they empoly in the department of nursing by specific categories such as RN’s, LPN’s, and ancillary staff (aides, orderlies and/or attendants).! The number of “full- time equivalent” personnel within each category was derived by adding one-half the number of part-time personnel to the number of full-time per- sonnel. Thus, if a hospital has reported 20 full-time and 10 part-time RN’s, the “full-time equivalent” RN’s for that hospital will be 20 + 14(10) = 25. The hospitals were also asked to report the number of hours in a regularly scheduled workweek. If a hospital has reported, say, a 3714-hour work- week, then the annual hours attributable to one full-time equivalent RN will be 3714 X 52 = 1,950 hours; and the annual hours attributable to, say 25 full-time equivalent RNs, will be 3715 X 52 X 25 = 48,750 hours. The full-time equivalent hours of other categories are estimated similarly. The sum of the hours estimated for RN’s, LPN’s, and aides, orderlies and/or attendants is the total nursing hours. Adjusted Patient Days The volume of output varies from hospital to hospital depending as it does on the amount of care provided on an inpatient and outpatient basis during a given time period, namely, a year. Therefore, in order to bring the staffing levels and patterns into comparable relationship, it is necessary to develop an overall measure of hospital output. The measure used in this Ward clerks and other personnel, if any, employed in the department of nursing are excluded. 21 study is adjusted patient days (ADJPD).2 The ADJPD is defined as in- patient days plus outpatient visits expressed in units equivalent to an inpatient day. A hospital can easily compute the ADJPD by any of the following formulas: ADJPD = IPD + OPV ve] IPREV PER DAY — IPD + OPV fuanziay TIPREV/IPD = IPD ef [1 + IPREV — pp [TOTREV IPREV IPD: Inpatient Days OPV: Outpatient Visits IPREV: Inpatient Revenue OPREV: Outpatient Revenue TOTREV: Total Revenue ® ® ©® Oo The following is an illustrative application of the formulas to the data of a hypothetical hospital: Inpatient Outpatient Total Inpatient Outpatient Inpatient Outpatient days visits revenue revenue revenue revenue revenue per per day visit 1,000 600 $86,000 $80,000 $6,000 $80 $10 _ 10 © ADIPD = 1,000 + 600 (go = 1,075 _ 6,000/600 _ ® = 1,000 + 00 ( B00 7100) = 1,075 _ 6,000 _ ® = 1000 ( 1 + Ser) = 1,075 _ 86,000 _ ® = 1,000 (ar = 1,075 It should be noted in this connection that with the growing emphasis on outpatient care, inpatient days has ceased to be a satisfactory measure of hospital output. It is for this reason that the American Hospital Association began reporting, since the late 1960s, two additional measures, adjusted average daily census and expenses per adjusted pa- tient day. (4) 22 Nursing Hours Per Adjusted Patient Day The nursing hours per adjusted patient day is obtained by dividing nursing hours by adjusted patient days. For example, if the RN hours are 48,750 and the adjusted patient days are 19,500, then RN hours per adjusted patient day = 48,750/19,500 = 2.5 23 Chapter 6 THE REGRESSION MODELS The regression models developed in this study are presented in tables 22 to 24. These are the final models which should be used to determine the expected staffing levels and patterns. An illustrative application of the model appears on pages 52, 53. We have also furnished, for each model, a sup- plementary table (tables 4 to 21) which provides the following additional details: a. The initial version of the model with all 21 explanatory variables. The standard error of the mean. c. The biased (uncorrected) multiple correlation coefficient. d. The unbiased (corrected for degrees of freedom) multiple correlation coefficient. e. The degrees of freedom. f. The beta (or the standardized regression) coefficients. The initial version of the model is presented as a matter of interest for those who wish to explore the relationship between various explanatory variables and nurse staffing levels. The standard error at the mean may be used as a criterion to determine whether the actual staffing level of an individual hospital differs significantly from the expected level, that is, the level predicted by the equation. The higher the multiple correlation coefficient, the better the model. In other words, a high coefficient implies that a large portion of the variations in staffing levels is explainable by the independent variables in the regression model.? The beta coefficients show the relative importance or explanatory power of the independent variables. In the discussion of individual models, we shall refer to tables 22 through 24 and the relevant supplementary tables 4 through 21. 1 This is a rough criterion because of the bi-convex shape of the confidence belt. 2 Technically, the extent of variation in staffing level explained by the model is the square of the correlation coefficient. For example, if the coefficient is 0.50, it means that (0.50)? = 0.25. That is, 25 percent of the variation is explained by the model. 25 Interrelationship Among Hospital Indices A pattern of influence which has appeared with unfailing consistency in almost all the models presented in this chapter is the following: The Nature of The Independent (Explanatory) Variable Influence on Staffing Level X4: Total admissions (A) ....... soi 58 5 4 wiv 00 wo oe POSHHVE Xs: Length of stay (S) vevvvvvnninennnnnns. .... Negative Xg: Occupancy rate (OR) .........ovvveenn. «vee... Negative Xis: Adjusted patient day (PD) ...... sacsnasrsaes oe Nogalive The preceding explanatory variables are indices of hospital operation/per- formance viewed from different perspectives. Consequently, they are closely interrelated. An understanding of the nature of this interrelationship will be helpful in interpreting the regression models. For a hospital with a given “patient days,” larger admissions necessarily mean shorter length of stay.? Shorter length of stay generally means higher intensity of services because ancillary services are high during the first few days of admission, but taper off as the patient stays longer. Therefore, a positive regression (or beta) coefficient for admissions (X4), or a negative coefficient for length of stay (Xj5) should be interpreted to mean that the level of nurse staffing increases with service intensity. All dependent variables (Y; through Yg are expressed on a “per adjusted patient day” basis. It is, therefore, natural to expect that the lower the (adjusted) patient days, the higher generally will be nursing hours per adjusted) patient day. Similarly, the lower the (adjusted) patient days, the lower will be occupancy rate. These relationships explain, in large part, why (adjusted) patient days (X;3) and occupancy rate (Xg) appear with negative regression (or beta) coefficients. We realize that the use of these measures as independent variables of the regression models may not be very enlightening in a casual sense. Nonetheless, a decision was made to include them because of their high predictive power. The Interaction Between Supply Variables and Nurse Staffing In our discussion pertaining to individual models we have frequently 2 Note that: PD = A-S. ¢ Note that: OR = PD/ (Beds) (365) 5In fact the theorists may argue that the inclusion of these variables is likely to dis- tort the casual significance of the remaining variables. We take the position of the empiricists and contend that this argument ignores the primary purpose at hand, namely, prediction of nurse staffing levels with as much accuracy as possible. We may add that this is an unresolved controversy dating back to the 1940's. See, for example, Koopmans’ critique on the pioneering work of A. F. Burns and W. C. Mitchell, Measuring Business Cycles. (22) 26 made two types of statements. The first type of statement is to the effect that more personnel of a specific category—say, RN’s—tend to be em- ployed by hospitals if the supply or availability of that category of per- sonnel is plentiful. The second type of statement is to the effect that when certain categories of personnel are more plentiful than others, a “substitute effect” is in operation. That is, the category that is plentiful tends to be substituted for the category that is in short supply. The causal sequence of events implied in these statements is speculative. It is conceivable that if hospitals in a region historically employ more nursing personnel of a given category, say, RN’s, then this will manifest itself in a high rate of RN’s per 100,000 population. If this is indeed the case, the causal sequence of events implied in these statements becomes untenable and the relationship between staffing levels and availability of personnel demonstrated by the models becomes almost tautological. However, the geographic and urban rural disparities in the distribution of nursing personnel described earlier (chapter 2) seems to indicate that nurses tend to gravitate toward areas where living conditions are more congenial. If this is, in fact, the case, the so-called “Say’s Law of the Market” becomes operative, that is, supply would create its own demand.” Also, increased supply of a category of nursing personnel would depress wage levels making it worthwhile for the hospitals to hire more personnel in that category.? Homogeneity of Criteria The multiple correlation involving registered nurses (Y2) was con- sistently higher than those for total nursing personnel (Y;), licensed prac- tical nurses (Y3) and aides, orderlies and/or attendants (Y4). This was true across all strata. This phenomenon may be explained in terms of the homogeneity or purity of the criterion variable. The RN’s have relatively clearly defined roles as against LPN’s and ancillary personnel. Consequently, the dependent variable, Ys, is a more homogeneous measure than Y3 and Ys. It is also a more homogeneous measure than Yj, which is a mixture of three disparate categories. One may think of this impurity as an additional “noise” or “random variation,” which is not represented in the model. This will de- crease the correlation in a manner analogous to that for “observational error.” ® ¢ See, for example, Smith, and U.S. Department of Agriculture. 7 See, for example, Altman, page 25, also, Shain and Roemer. 8 Under certain assumptions about the structure of the market. a firm typically hires labor up to the point where the marginal cost (the cost of hiring an additional em- ployee) equals the marginal revenue (the additional revenue derived from hiring that employee). ® See, for example, Kendall and Stuart. 27 The Models and Their Application For a given hospital stratum a set of six regression equations—one for each of the following categories of nursing personnel—has been con- structed: 1° Y:: Total nursing hours per adjusted patient day Y2: RN hours per adjusted patient day Y;: LPN hours per adjusted patient day Ys: Aides, orderlies and/or attendants hours per adjusted patient day Y;: General duty nurses hours per adjusted patient day Ys: General duty and head nurse hours per adjusted patient day Each regression equation has a set of independent (explanatory) variables. The initial versions of these equations have 21 such variables (see tables 4 to 21). Using a “step-down” procedure, variables whose contributions to predictions were negligible were eliminated. The final versions of the equations appear in tables 22 to 24. An Illustrative Example We shall now present an illustrative example of the application of the regression equations. From table 24 we reproduce below the regression equation for total nursing hours per adjusted patient day pertaining to Stratum III. For-Profit Hospitals: Y, = 5.61727 + .25544X; + .07595X;2 — 29.38419Xis R 617 Sd. error 907 These equations do not constitute a simultaneous-equations model with endogenous and exogenous variables, Such a model would probably have provided a better insight into the dynamics of the system, that is, the causalities and interactions within the sys- tem. However, the primary purpose of this study is to develop a usable tool to detect marked deviations from expected staffing levels and patterns. The equations constructed in this study will, in our judgment, serve this purpose. 28 The table below gives the staffing levels predicted by this equation as well as the actual staffing levels of three hospitals whose identities are withheld: Actual and Predicted Staffing Levels of Three Hospitals in Stratum III (Total Nursing Hours) Actual Predicted Difference Hospital (Y:-Y1)/Sd.Error Y: Y, Y:-Y: 10.431 6.487 3.944 2.094 7.122 7.128 0.006 —0.003 5.822 9.613 -3.791 -2.012 Notice that the actual staffing level of hospital A is markedly higher than the predicted level; that of hospital B is about the same as the predicted level; and that of hospital C is markedly lower than the predicted level. The last column of the table shows to what extent the discrepancy between the actual and predicted staffing levels exceeds the standard error. Thus, the regression equation has identified two hospitals, A and C, whose overall staffing levels deviate markedly from expected levels. As demonstrated, one can predict the overall staffing level of any hospital within the universe. The staffing level so predicted may be higher or lower than the stratum average depending on the particular configuration of characteristics associated with that hospital. We may think of the predicted value as an “adjusted” staffing level, that is, a level that “allows for” the peculiarity of the hospital in question. If the actual staffing level is markedly higher or lower than the predicted level, there is prima facie evidence to suggest the need for a reexamination of the staffing level of the hospital in question. The expected staffing level of specific categories of personnel—RN’s, LPN’s, aides, orderlies/attendants, general duty nurses, and general duty and head nurses—can be obtained by using the appropriate equations. We have earlier defined staffing pattern as the proportion in which the various levels of personnel are employed. The expected staffing pattern of a hospital can be determined in the following way: Using regression equations for Xz, X3, and X4, one can predict, for a given hospital, the expected hours per adjusted patient day of RN’s, LPN’s, aides, orderlies/attendants, and therefrom the expected ratio. Let this ratio be 3:2:4.1* Here again, this ratio may be different from the stratum ratio inasmuch as the regression equations have “allowed for” the peculiarities of the hospital. For example, a hospital which specializes in providing sophisticated services may have a different 1 That is, for every 3 hours spent by RN’s, 2 hours are spent by LPN’s and 4 hours are spent by aides, orderlies and/or attendants. 29 ratio, say 4:2:1. If the actual ratio differs markedly from the expected ratio, there is prima facie evidence to suggest the need for a reexamination of the staffing pattern of the hospital in question. A limitation of this approach should be emphasized at this point. The staffing levels and patterns predicted by the models should not be con- strued as optima because they reflect what is typical of all hospitals within the stratum. Therefore, interpreting the predicted levels and patterns as optima involves the untested assumption that the staffing levels and patterns of the generality of hospitals within the stratum meet the criterion of optimality. Nevertheless, the models can be used to detect “outliers,” that is, hospitals whose staffing levels and patterns are out of line with comparable hospitals. The Computation of Nursing Hours Per Adjusted Patient Day If it is desired to use the models for computation, and a full year’s data are available, one may use the formulas of chapters 4 and 5 and follow the sample on page 28. One should use the five-digit values of the coefficients given in tables 22, 23, and 24. If less than a full year’s data are available, one may make the computa- tion, but a few adjustments are necessary. These adjustments are discussed in the Technical Appendix. It should be borne in mind, however, that the longer the period for which data are available, the more reliable and stable the measures will be. 30 Chapter 7 FINDINGS: REGRESSION MODELS Categories of Explanatory Variables In this section references are frequently made of generic categories such as “organizational characteristics,” “product characteristics,” and so on. These references are based on the following taxonomy: Organizational Characteristics Xe: Occupancy rate Xj2: Statistical beds Xi3: Adjusted patient days Product Characteristics X;: Total number of facilities/services Xz: Number of advanced technology facilities/services X3: Number of advanced technology facilities/services expressed as a percentage of all facilities/services X4: Total admissions? Xs: Length of stay?! X7: Bassinets per statistical bed Xg: Births per statistical bed Xo: Emergency visits per admission Xo: Total visits per admission Xi1: Surgical operations per admission Socioeconomic Characteristics Xi4: SMSA vs. non-SMSA Xoo: Per capita income Xo;: Percent families under poverty level Demographic Characteristics Xi5: Percent population under 18 Xie: Percent population 65 and over 1See page 26 for an explanation as to why these variables are treated as product characteristics. 31 Supply Characteristics X17: RN's per 100,000 population Xis: LPN’s per 100,000 population X19: Aides, orderlies and/or attendants per 100,000 population Stratum I: Nonprofit, Nonteaching Hospitals Group Characteristics There were, initially, 3,890 hospitals within this stratum. However, the regression equations developed for this stratum are based on approximately 2,900 hospitals, because several hospitals had to be excluded for one reason or another as discussed in chapter 4. A detailed profile of this group of hospitals is provided in tables 25-44. Here we indicate the more salient characteristics of the group. The average bedsize of this group is 109; the average annual patient days (adjusted) are 32,600; the average annual admissions are 4,000; and the average length of stay is 7.8 days. Hospitals in this group have, on the average, 0.15 bassinets per bed; 3.33 outpatient visits per inpatient admission; and 0.35 surgical operations per inpatient admission. Hospitals in this group maintain, on the average, an occupancy rate of 72 percent. The average staffing levels and patterns of this hospital group are given below: Hours per adjusted Category patient day Yi: Total nursing personnel «.vueveuseeneeennruneeenneennnnns 6.553 Y:: Registered nurses «.....oevevenenen. wor 2.377 3: Licensed practical nurses ......... oe rp 1.351 Yi: Aides, orderlies, and/or attendants - PR 2.868 Ys: General duty nurses .........ooe... cee 1.449 Yo: General duty and head nurses «vuveveruneerennneeeennnnn. 1.879 The Regression Models 1. Total Nursing Hours per Adjusted Patient Day (Y;) Turning to table 4, we find that this model has 10 independent (explana- tory) variables. The unbiased multiple correlation coefficient is 0.50, which should be adjudged moderate to good. Notice that all independent variables used in this model, except one, are organizational or product characteristics. In other words, variations in total staffing levels are due primarily to the volume of activity (including capacity utilization) and product characteristics. The only socioeconomic variable that 32 significantly influences total staffing levels is the SMSA vs. non-SMSA dichotomy. Hospitals located in the SMSA tend to employ more nursing personnel. The beta coefficients given in table 4 (Final Model Section) show the relative importance of each explanatory variable. Adjusted patient days (Xi3) ranks first in order of importance. Other things being equal, the lower the patient days, the higher the overall staffing level. This is under- standable because total nursing hours, and indeed, all dependent variables, are expressed on a “per adjusted patient day” basis (see page 26). The variable which ranks second in importance is total admissions (Xj). The larger the admissions, the higher the overall staffing level. This relation- ship may seem, at first sight, intriguing. The explanation lies in the inter- relationship between certain measures. Admissions X length of stay = pa- tient days. This means that with a given patient days, more admissions necessarily imply shorter length of stay. And it is well known that shorter length of stay implies higher intensity of services (see page 26). That overall staffing level rises with broader repertoire of facilities and services is evident from the beta coefficients of total number of facilities and services (X;), number of advanced technology facilities (Xs), and surgical operations per admission (Xj1). 2. Registered Nurses Hours per Adjusted Patient Day (Y:) This model (shown in table 5) has 10 independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.72, which should be adjudged very good. Unlike the preceding model, this one has explanatory variables pertain- ing to organizational, product, socioeconomic and supply characteristics. Turning to the beta coefficients in table 6, we notice that total admissions (Xs) and adjusted patient days (X;3) are nearly equal in importance. Notice that the nature of the relationships is consistent with those observed in the preceding model. The impact of supply characteristics and the prev- alence of substitution effect are discernible from this model. The beta coefficient of RN’s per 100,000 population indicates that when the supply of registered nurses is plentiful, more registered nurses tend to be employed. At the same time, the beta coefficient for aides, orderlies and/or attendants per 100,000 population (Xjs) indicates that when the supply of ancillary personnel is low, more registered nurses tend to be employed. The model also supports the hypotheses that hospitals located in affluent areas attract more professional nurses (see variable X29), and that the more sophisticated the services provided, the higher the level of professional nurse staffing (see variable Xz). 3. Licensed Practical Nurses Hours per Adjusted Patient Day (Ys) This model (shown in table 6) has 10 independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.51, which should be adjudged moderate to good. 33 The explanatory variables of this model pertain to organizational product, socioeconomic, and supply characteristics. The beta coefficients of table 6 show that adjusted patient days (Xi3) ranks first, and total admissions (X4) ranks second in order of importance. The nature of their relationship to LPN staffing level is consistent with the relationship observed in the preceding models. The relationship between the availability of LPN’s and their employment is rather strong. Indeed, the t-value associated with the supply variable (X1g) is the highest. Furthermore, the model provides striking confirma- tion of the substitution hypothesis. Notice that the beta coefficients pertaining to RN’s per 100,000 population (X;7) and aides, orderlies and/or attendants per 100,000 population (X;g) are negative and highly significant. 4. Aides, Orderlies and/or Attendants Hours per Adjusted Patient Day (Y4) This model (shown in table 7) has nine independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.48, which should be adjudged moderate. This model demonstrates that the factors influencing staffing levels of ancillary personnel are somewhat different from those of professional and practical nurses. First, product characteristics do not seem to exert a major influence on the staffing level of ancillary personnel. Second, although ad- justed patient days (Xs) does affect staffing level, its effect is less pro- nounced. Reference to table 7 will show that supply characteristics (Xi7, Xis, and X19), demographic characteristics (X16) and socioeconomic characteristics (X21) are the more important determinants of ancillary staffing. In general, hospitals located in ghettos and poverty pockets where the supply of pro- fessional and practical nurses is scarce tend to employ more aides, orderlies and/or attendants. This supports the statement made earlier (chapter 2) about the prevalence of “vertical maldistribution” of nursing personnel. 5. General Duty Nurse Hours per Adjusted Patient Day (Y;) This model (shown in table 8) has 10 independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.68, which should be adjudged very good. The explanatory variables of this model represent organizational, product, socioeconomic, demographic, and supply characteristics. The beta coefficients for total admissions (X4), adjusted patient days (X13) and number of high technology facilities (X:) indicate that more general duty nurses are generally employed by hospitals providing high- intensity, sophisticated services. The large beta coefficient and associated t-value for the supply variable, RN’s per 100,000 population (X;7) are noteworthy. More general duty 34 nurses are employed by hospitals if the supply of professional nurses is plentiful. 6. General Duty and Head Nurse Hours per Adjusted Patient Day (Ye) This model (shown in table 9) has 10 independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.71, which should be adjudged very good. This model is very similar to the one for general duty nurses. However, the positive relationship between staffing level and the provision of high- intensity, sophisticated services noted in the preceding model operates here a fortiori. Similarly, the availability of professional nurses is more strongly related to the employment of general duty and head nurses. Stratum II: Nonprofit, Teaching Hospitals Group Characteristics There were, initially, 842 hospitals in this stratum. However, certain hospitals had to be excluded for reasons stated in tables 25-44. Here the data base is approximately 630 hospitals. A detailed profile of this group is presented in chapter 9. The more salient characteristics of this group is indicated here. The average bedsize of this group is 421; the average annual (adjusted) patient days are 142,000; the average annual admissions are 14,100; and the average length of stay is 9.10 days. Hospitals in this group have, on the average, 0.104 bassinets per bed; 5.89 outpatient visits per inpatient admission; and 0.53 surgical operations per inpatient admission. These hospitals maintain, on the average, an occupancy rate of 84 percent. In general, these are relatively large hospitals. The higher proportion of surgery and the longer length of stay suggest that these hospitals generally handle more complex cases. The average staffing levels and patterns of this group are presented below: Hours per adjusted Category patient day Y;: Total nursing personnel .......coviiiiiiiiiiiiiiiiiinns 6.232 Yao: Registered nurses «vuvvve een iiiiineeernnneeenennnns 2.952 Ys: Licensed practical nurses ........oeeiiiiuurennneneneenns 1.184 Yi: Aides, orderlies, and/or attendants ......ovviieinnnnnnnnnn 2.093 Ys: General duty BUrses coviuvessssessnnunmunviss is sunvmmnns 2.230 Ye: General duty and head nurses .....eevvvvvniveennnnnnnnnn 2.652 35 The Regression Models 1. Total Nursing Hours per Adjusted Patient Day (Y,) This model has seven independent (explanatory) variables (see table 10). The unbiased multiple correlation coefficient is 0.48, which should be adjudged moderate. The independent (explanatory) variables of this model represent organi- zational, product, and demographic characteristics. The beta coefficients given in table 10 show the relative importance of each explanatory variable. Adjusted patient days (Xs) ranks first in im- portance. In fact, the rank and direction of effect of this variable are con- sistent with those observed for the corresponding model of Stratum I. The size of the hospital (measured in beds) ranks second in importance. In general, the larger the hospital, the higher the level of total nurse staffing. However, this relationship should be viewed with the utmost caution. The reason is that the stratification of hospitals has led to truncation of range with respect to bedsize. What would have been the relationship if hospitals were not stratified, it is difficult to say. 2. Registered Nurses Hours per Adjusted Patient Day (Y:) This model (shown in table 11) has seven independent (explanatory) variables. The unbiased multiple correlation coefficient is 0.54, which should be adjudged good. If one examines the beta coefficient of RN’s per 100,000 population (X;7) in conjunction with the beta coefficient of per capita income (Xz), it will be apparent that the level of RN employment tends to be higher in affluent areas with a high RN-per-population ratio. The shorter the length of stay, the higher the level of RN employment; and the more advanced technology facilities/services a hospital has, the higher the level of RN employment. Thus, the provision of high-intensity, sophisticated services is associated with more employment of professional nurses (see page 26). 3. Licensed Practical Nurse Hours per Adjusted Patient Day (Ys) This model (shown in table 12) has five independent (explanatory) variables. The unbiased multiple correlation coefficient is 0.39, which should be adjudged fair. The variable that ranks first in order of importance is RN’s per 100,000 population (Xz). The sign of the beta coefficient of this variable (negative) suggests the operation of “substitution effect.” When one considers this in conjunction with the negative effect exerted by per capita income (X20), it will be apparent that more LPN’s tend to be employed in low income areas characterized by scarcity of professional nurses. 36 4. Aides, Orderlies and/or Attendants Hours per Adjusted Patient Day (Y,) This model (shown in table 13) has six independent (explanatory vari- ables. The unbiased multiple correlation coefficient is 0.43, which should be adjudged moderate. This model bears some similarity with its counterpart in Stratum I in two respects: First, product characteristics have little influence on the staffing levels of ancillary personnel. Second, supply characteristics (X17, Xi, and Xie) and the socioeconomic characteristic (X2;) emerged as important determinants of staffing level. In general, hospitals located in ghettos and poverty pockets where the supply of professional and practical nurses is scarce tend to employ more aides, orderlies and/or attendants. Thus, the prevalence of “vertical maldistribution” of nursing personnel is again substantiated. The size of the hospital is also related to ancillary staffing. The larger the hospital the higher generally is the staffing level. S. General Duty Nurse Hours per Adjusted Patient Day (Y;) There are 10 independent (explanatory) variables in this model (which is shown in table 14). The unbiased multiple correlation coefficient is 0.55, which should be adjudged good. The explanatory variables of this model represent organizational, product, and supply characteristics. In this respect, this model is similar to the corresponding model for Stratum I. The beta coefficients for total admissions (X4) and adjusted patient days (X13) suggest that more general duty nurses are generally employed by hospitals providing high-intensity services. The large beta coefficient and associated t-value for the supply variable, RN’s per 100,000 population (X;7) are noteworthy. More general duty nurses are employed by the hospitals if the supply of professional nurses is plentiful. 6. General Duty and Head Nurse Hours per Adjusted Patient Day (Ye) This model (shown in table 15) has seven independent (explanatory) variables. The unbiased multiple correlation coefficient is 0.54, which should be adjudged good. The variable that ranks first in importance is RN’s per 100,000 popula- tion (Xi7). More general duty and head nurses tend to be employed in areas where the supply of professional nurses is plentiful. The variable that ranks second in order of importance is length of stay (Xs). The negative sign of the beta coefficient pertaining to this variable seems to suggest that employment of this category of personnel increases with the intensity of services provided (see page 26). 37 Stratum III: For-profit Hospitals Group Characteristics There were, initially, 721 hospitals in this stratum. The quality of report- ing by hospitals in this stratum was generally poorer than that of the other strata. Consequently, many hospitals had to be dropped. The models pre- sented in this section are based on the data of approximately 175 hospitals. A detailed profile of this group appears in tables 25-44. The more salient characteristics of the group is indicated here. The average bedsize of this group is 64; the average annual patient days (adjusted) are 17,800; the average annual admissions are 2,500; and the average length of stay is 7.2 days. Hospitals in this group have, on the average, 0.09 bassinets per bed; 4.65 outpatient visits per inpatient admission; and 0.36 surgical operations per inpatient admission. Hospitals in this group maintain, on the average, an occupancy rate of 70 percent. In general, these are relatively small hospitals with a somewhat shorter length of stay. The average staffing levels and patterns of this group are given below: Hours per adjusted Category patient day Y:: Total nursing personnel covvons ess snsnrnsrsosrres cpprves 6.811 Y:: Registered nurses «.....ceevvueeannn ve 2.327 Ys: Licensed practical nurses ........ sR 5 1.494 Y.: Aides, orderlies, and/or attendants 3.040 Ys: General duty nurses ............. cee oe 1.125 Ys: General duty and head nurses ...... 1.678 The Regression Models 1. Total Nursing Hours per Adjusted Patient Day (Y;) Turning to table 16, we find that this model has three independent (ex- planatory) variables. The unbiased multiple correlation coefficient is 0.61, which should be adjudged very good. The explanatory variable, adjusted patient days (Xi3), ranks first in im- portance, followed by statistical beds (X;2) and total number of facilities/ services (X;) in that order. The inverse relationship between adjusted patient days and total nurse staffing is understandable in view of the manner in which the dependent variable (Y;) is expressed. For profit-seeking hos- pitals, the other key determinants of total staffing level are size and the repertoire of facilities/services. 38 2. Registered Nurses Hours per Adjusted Patient Day (Y;) This model (shown in table 17) has four independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.68, which should be adjudged very good. The explanatory variable that ranks first in order of importance is surgical operations per admission (Xj1), followed by total number of facilities/services (Xi), occupancy rate (Xe) and adjusted patient days (X13) in that order. It is apparent that insofar as for-profit hospitals are concerned, the level of RN employment is determined, to a large extent, by the scope and complexity of services provided. The broader the scope and the more complex the services, the more RN’s tend to be employed. 3. Licensed Practical Nurses Hours per Adjusted Patient Day (Y;) This model (shown in table 18) has four independent (explanatory) variables. The unbiased multiple correlation coefficient is 0.44, which should be adjudged moderate. The explanatory variable that ranks first in order of importance is per capita income (X29), followed by LPN’s per 100,000 population (Xs), occupancy rate (Xg) and number of advanced technology facilities (Xz) in that order. Particularly noteworthy is the strong influence of per capita income. Hospitals located in low income areas tend to employ more LPN's, probably as substitutes for RN’s. 4. Aides, Orderlies, and/or Attendants Hours per Adjusted Patient Day (Y.) There are four independent (explanatory) variables in this model (which is shown in table 19). The unbiased multiple correlation coefficient is 0.47, which should be adjudged moderate. The explanatory variable which has the strongest influence is adjusted patient days (Xi3). As explained earlier, this may be due to the fact that the dependent variable (Y4) is expressed on a per-adjusted patient-day basis. Why statistical beds (X12) which reflects the size factor, and total admissions (Xs), which reflects the intensity factor, have emerged as im- portant explanatory variables, it is difficult to say. 5. General Duty Nurses Hours per Adjusted Patient Day (Y;5) This model (shown in table 20) has four independent (explanatory) variables. The unbiased multiple correlation coefficient is 0.65, which should be adjudged very good. In order of importance, the explanatory variables which rank first and second are total number of facilities/services (X;) and surgical operations per admission (Xj). These variables reflect the scope and complexity of the services provided, and therefore, we may say that the broader the scope 39 and the more complex the services provided, the more general duty nurses tend to be employed. 6. General Duty and Head Nurse Hours per Adjusted Patient Day (Ye) This model (shown in table 21) has five independent (explanatory) vari- ables. The unbiased multiple correlation coefficient is 0.70, which should be adjudged very good. As noted earlier, the negative relationship between adjusted patient days (X13) and the dependent variable (Yg) is explainable by the fact that the dependent variable is expressed on a per-adjusted-patient-day basis (see page 26). Other explanatory variables which are related, in an important way, to the employment of general duty and head nurses are total number of facilities (a measure of scope of services), total admissions (a measure of intensity of services) and surgical operations per admission (a measure of the complexity of services). We may say, therefore, that the level of general duty and head nurse employment will be high in those hospitals that pro- vide a broad repertoire of complex, high-intensity services. Chapter 8 SUMMARY OF FINDINGS Regional Analysis Regional analysis of hospital-based nursing personnel was done in terms of adjusted patient days, statistical beds, and civilian resident population. Total nursing hours per adjusted patient day vary from a high of 7.17 in Arizona to a low of 5.25 in Utah. Total nursing hours per statistical bed vary from a high of 2,264 in Arizona to a low of 1,389 in Montana. Total nursing hours per civilian resident population vary from a high of 11.50 in the District of Columbia to a low of 4.29 in Utah. While significant regional disparities in nursing hours exist in relation to patient days, beds and population, they are most glaring in terms of popu- lation. Since population is the best available surrogate of “need,” which we have earlier defined as an objective assessment by professionals of the health care needs of the community, the observed regional disparities in total nursing hours per civilian resident population merit corrective action at the local, State and national levels. It is hoped that a sustained imple- mentation of Section 222(b) (3) of the Health Manpower Act will go a long way toward ameliorating the situation. Personnel mix, too, is characterized by substantial regional differences. All New England and Middle Atlantic States have high proportions of pro- fessional nurses, while the proportions are uniformly low in West South Central and East South Central States. On the other end of the spectrum, States such as Oklahoma, Kansas, and Wyoming have high ratios of ancillary personnel. Since extreme variations in personnel mix detract from overall productivity consistent with quality, this too is a matter that calls for cor- rective action at the local, State and national levels. Regression Models In this section we will summarize the more important findings that have emerged from the regression analyses. Inter-stratum variations are discussed first, followed by a discussion of the impact of each independent (explan- atory) variable used in the regression models. Inter-Stratum Variations To facilitate inter-stratum comparisons, average staffing levels and pat- terns for the three strata are presented below: 41 Average Staffing Levels and Patterns by Stratum Hours per adjusted patient day Category Stratum Stratum Stratum I II 111 Y,: Total nursing personnel ..........c..... 6.553 6.232 6.811 Y:: Registered nurses .....ooeeeeeeveeneennns 2.377 2.952 2.327 Ys: Licensed practical nurses ............... 1.351 1.184 1.494 Y.: Aides, orderlies and/or attendants ....... 2.868 2.093 3.040 Ys: General duty nurses ......covvnnnnnnnnns 1.449 2.230 1.125 Ye: General duty and head nurses ........... 1.879 2.652 1.678 When all categories of nursing personnel are lumped together, staffing level is highest in Stratum III (for-profit hospitals) and lowest in Stratum II (teaching hospitals), with Stratum I (nonprofit, nonteaching hospitals) occupying the middle ground. These differences seem to be largely due to the level of employment of ancillary personnel. It is also worth mentioning that the levels and patterns seen in the preceding table are in general agreement with previous studies. (6, 12) There are notable inter-stratum differences in staffing pattern or staff mix. For teaching hospitals, the ratio among RN’s, LPN’s and ancilliary personnel is roughly 3.0:1.2:2:1. This group has the highest ratio of RN’s and the lowest ratio of ancillary personnel. We have indicated that teaching hos- pitals have more advanced technology facilities and that they generally handle more complex cases. These features probably explain the need for more professionals among the nursing staff. For nonprofit, nonteaching hospitals, the ratio is roughly 2.4:1.4:2.9. This group of hospitals has a lower ratio of RN’s and a higher ratio of ancillary personnel than teaching hospitals. For-profit hospitals’ ratio is roughly 2.3:1.5:3.0. Notice that this hospital group has the lowest ratio of RN’s and the highest ratio of ancillary person- nel. “Upward substitution,” that is, the substitution of lower level personnel for the performance of tasks traditionally done by higher level personnel is, thus, very much in evidence. If the substitution is optimal, this is all to the good in that it contributes toward lower hospital costs. On the other hand, if “upward substitution” is being carried beyond the optimal level, some reduction in quality of care may result via the “Human Capital Argu- ment.” (5) The level of employment of general duty nurses as well as general duty and head nurses is highest in teaching hospitals and lowest in for-profit hospitals with nonprofit, nonteaching hospitals occupying the middle ground. Impact of Independent Variables In an attempt to provide an overall view of the importance of each explanatory variable across strata and across models within stratum, table 42 48 has been constructed. It was pointed out earlier (chapter 6) that the beta coefficients shown in tables 4 through 21 indicate the relative importance of the explanatory variables included in each regression model. Table 48 displays the ranked values (ignoring signs) of these beta coefficients. For example, the column headed “Y;” in Stratum I shows the ranked values of beta coefficients pertaining to the model for total nursing hours per adjusted patient day. Notice that the beta coefficient for the explanatory variable, adjusted patient days, has a ranked value of 1 with a negative sign in parenthesis. This means that “adjusted patient days” exerts the strongest influence and that the nature of the influence is negative. The explanatory variable that ranks second in order of importance—total ad- missions—has a ranked value of 2 and so on. By reading across rows, one can gain an overall view of the impact of each explanatory variable across strata and across models within stratum. For instance, reading across the row pertaining to “adjusted patient days,” the following generalizations are possible: a. This variable appears in most models (14 in all). b. The impact of this variable is very strong. c. The direction of the relationship (negative) is consistent across strata and models. Organizational Characteristics Statistical beds (X;2): This variable represents the scale of operation, and does not have a major impact on nurse staffing levels. This statement is made with an important caveat. The stratification of hospitals has substan- tially reduced the variability of bed size. If the models were constructed for the entire universe, statistical beds might have emerged as an important determinant of staffing levels. Five out of the 18 models do show a relation- ship—albeit a weak one—between staffing level and bedsize. In all these cases, the relationship is positive, indicating that staffing levels tend to rise with the scale of operation.! Adjusted Patient Days (X;3): In all but four of the models, this vari- able appears with large negative regression (or beta) coefficients. The smaller the adjusted patient days, the higher the staffing level. However, as explained elsewhere, this relationship is due, in large part, to the manner in which the dependent variables are defined. Occupancy Rate (Xe): This variable appears in most models (11 in all). The regression (or beta) coefficients are, in all cases, negative. The lower the occupancy rate, the higher generally is the staffing level. Here 11t is difficult to speculate what would have been the extent and direction of the effect if the hospitals were not stratified. 43 again, this relationship is due, in large part, to the manner in which the dependent variables are defined (See footnote 2, page 25). Product Characteristics Total Number of Facilities/Services (X;): This variable represents the scope and range of services provided by hospitals. There are seven models in which this variable appears with significant regression (or beta) coefficients. These models pertain to the staffing levels of higher level person. nel: general duty and head nurses, RN’s and LPN’s. In every case, the re- lationship is positive suggesting that the staffing level of higher level nursing personnel generally rises with the scope of services provided by hospitals. Number of Advanced Technology Facilities/Services (X:): This variable represents the extent of sophistication of services provided by hos- pitals. It appears with significant regression (or beta) coefficients in six models, all of which pertain to staffing levels of higher level personnel. Number of Advanced Technology Facilities/Services Expressed as Fraction of Total Number of Facilities/Services (X3): This vari- able is similar to X, except that it shows the extent of sophisticated facilities/ services relative to the total facilities/services provided by hospitals. This variable appears with significant regression (or beta) coefficient in only one model. Perhaps variable X; is sufficient to bring out the impact of advanced technology facilities and services on nurse staffing levels. Total Admissions (X4): We have indicated elsewhere that more ad- missions during a fixed time span generally imply higher intensity of care. Total admissions can, therefore, be construed as a product characteristic. This variable appears with significant regression (or beta) coefficients in nine models. Particularly noteworthy is the fact that it appears in all models belonging to Stratum I (nonprofit, nonteaching hospitals). It appears once in Stratum II (nonprofit, teaching hospitals) and twice in Stratum III (for-profit hospitals). Wherever this variable appears, the relationship is positive and strong, suggesting that staffing levels tend to increase with service intensity. Length of Stay (X;): As explained elsewhere, shorter length of stay generally implies higher intensity of care. This variable appears in six mod- els. In all cases, the relationship is negative, indicating that staffing levels generally rise with intensity of services. Bassinets per Statistical Bed (X;) and Births per Statistical Bed (X;): These two variables reflect the extent of obstetrical cases handled by hospitals. One of these variables appears in seven models. In all cases, the relationship is positive. Generally speaking, more nurses are employed by hospitals handling a larger proportion of obstetrical cases. Emergency Visits per Admission (Xo) and Total Visits per Ad- mission (Xjo): These variables reflect the extent of outpatient activity in 44 hospitals. One of these variables appears in several models (eight in all) pertaining to Stratums I and II. The relationship is positive, suggesting that more nursing personnel are generally employed by hospitals which provide more outpatient services. It is significant that these variables do not appear in any of the models pertaining to Stratum III (for-profit hospitals). Surgical Operations per Admission (Xi): This variable appears with positive regression (or beta) coefficients in seven models. Hospitals which handle a large proportion of surgery cases tend to employ more pro- fessional nurses. Socioeconomic Characteristics SMSA vs. Non-SMSA (Xi4): This dichotomous variable captures, in a broad way, the urban-rural distinction among the areas in which hospitals are located. There is only one model in which this variable appears with sig- nificant regression (or beta) coefficients. It would seem that the other socio- economic variables included in this analysis have brought out more clearly the socioeconomic differences. Per Capita Income (X2) : This variable has emerged as an important determinant of the staffing levels of RN’s and/or LPN’s in all strata. In the regression models for RN’s, this variable appears with large positive coeffi- cients, while in the regression models for LPNs, this variable appears with negative weights. There is, thus, a strong association between high-income areas and RN employment, and low-income areas and LPN employment. Percent Families Under Poverty Level (X31): This variable has appeared with positive regression (or beta) coefficients in the models per- taining to aides, orderlies and/or attendants. The implication is noteworthy. More lower level nursing personnel tend to be employed in hospitals located in ghettos, poverty pockets and economically depressed areas. What bearing this has on quality of care is a subject that merits further study. Demographic Characteristics Percent of Population Under 18 (X,;) and Percent of Popula- tion 65 and Over (X;;): These two variables have appeared with nega- tive regression (or beta) coefficients in eight models. One may hypothesize that patients belonging to the excluded age category—18-64—generally come to the hospital with more serious ailments demanding closer attention by the nursing staff. Supply Characteristics Registered Nurses per 100,000 Population (Xi): Except for Stratum III (for-profit hospitals), the impact of this variable is strong and 45 consistent with a priori reasonings. Employment of professional nurses in- creases, and the employment of LPN’s and aides, orderlies and/or attendants decreases with the increased availability of RN’s. However, this pattern is not observed in for-profit hospitals. Licensed Practical Nurses per 100,000 Population (Xs): Like variable X,7, the impact of this variable is strong and consistent with a priori reasonings. The employment of LPN’s increases, and the employment of aides, orderlies, etc., decreases with the increased availability of LPN's. This pattern is discernible in all strata. Aides, Orderlies and/or Attendants per 100,000 Population (X19) : Like the two supply variables just discussed, the impact of this vari- able is strong and consistent with a priori reasonings. The employment of aides, orderlies and/or attendants increases, and the employment of profes- sional and practical nurses decreases with the increased availability of an- cillary personnel. This pattern is discernible in all strata. 46 Chapter 9 DESCRIPTIVE STATISTICS In all, 27 variables were used in this study. These variables are defined and details of computation set forth in chapter 4. Independent variables Xi; through Xo; are demographic, socioeconomic, and supply characteristics. Since all hospitals in any geographic area (county or State) are assigned the values pertaining to the area, distribu- tions, means, standard deviations, etc., would not be very meaningful insofar as these variables are concerned. Consequently, a decision was made to develop descriptive data for only the remaining 20 variables (X; through X14 and Y; through Yg). Descriptive statistics pertaining to the 20 variables are shown by the original 27 strata and also by the final three strata in tables 25 through 44. As noted in chapter 4, data were missing in a variety of ways and places. When feasible and appropriate, average values were substituted in place of the missing ones in order to salvage as much information as possible. The descriptive statistics do not include data which were not available before the salvaging process. For this reason, the numbers of hospitals differ some- what among tables 25 through 44. The tables show the following statistics: number of hospitals, arithmetic means, standard deviations, minimum values, maximum values, and range. Table 25 presents data on total number of facilities and services. As might be expected, total number of facilities and services increases with size and teaching status. Consequently, large teaching hospitals (Stratum Code 119) have the largest number of facilities and services. Table 26 displays number of advanced technology facilities. This vari- able shows a pattern identical with the preceding variable, with large teaching hospitals having the largest number of advanced technology facilities. Advanced technology facilities expressed as a fraction of total facilities are presented in table 27. This fraction increases with size. Data on total admissions are shown in table 28. As might be expected, admissions (indicated by the mean values), and variability (indicated by the standard deviations) increase with bedsize. Since teaching hospitals are generally large, and for-profit hospitals are generally small, Stratum II has the highest and Stratum III the smallest average admissions. Length of stay is shown in table 29. As can be seen from the mean values, length of stay tends to increase with hospital size and teaching status. 47 Consequently, the longest stays are recorded by large, teaching hospitals (Stratum Code 119). For length of stay, the ranges are large.compared with the standard deviations. If this data were “normally distributed,” one should expect the range to be from four to six times the standard deviation. Here the ranges run up as high as 10 to 20 times the standard deviations, and this indicates a badly skewed distribution. In other words, there is a dis- proportionately large number of hospitals with lengthier stays as compared with the arithmetic mean. Data on occupancy rate are presented in table 30. Aside from very small hospitals which generally have lower occupancy rate, no distinguishable pattern is discernible. There are a few hospitals with an occupancy rate of over 1,000. This is due to several reasons, such as hospitals changing their beds after submitting the survey questionnaire, setting up temporary beds to meet emergency situations and so forth. Table 31 shows bassinets per statistical bed. Smaller hospitals have more bassinets in relation to their bed complement. Also, for-profit hospitals have fewer bassinets per statistical bed. Births per statistical bed are presented in table 32. There seems to be no correlation between births per statistical bed and bassinets per statistical bed. Table 33 presents data on emergency visits per admission. It is note- worthy that teaching hospitals handle more emergency visits in relation to admissions. For-profit hospitals, on the other hand, handle the least number of emergency visits. Total outpatient visits per admission are shown in table 34. Teaching hospitals generally handle more outpatient visits. Aside from this, no other pattern is discernible. Table 35 presents surgical operations per admission. Two patterns are dis- cernible: First, surgical operations per admission tend to increase with bedsize. This may be explained by the fact that there is a high correlation between bedsize (X;2) on the one hand, and total number of facilities and services (X;) and number of advanced technology facilities (Xz) on the other. See tables 45 through 47. Second, governmental hospitals perform fewer surgical operations per admission. Table 36 presents data on statistical beds. As can be seen from the arithmetic means, Stratum II (teaching hospitals) is mostly composed of large hospitals, Stratum III (for-profit hospitals) is mostly composed of moderate and small sized hospitals. Adjusted patient days are shown in table 37. For convenience this measure is expressed in units of 100,000. As one might expect, adjusted patient days increase with bedsize. The locations of hospitals, that is, whether hospitals are located in SMSA’s or otherwise, are presented in table 38. Since SMSA is coded “1” and non- SMSA “0,” the arithmetic means shown in the table should be interpreted as the proportion of hospitals located in SMSA, and 1.000 minus the pro- 48 portion shows the number of hospitals located in non-SMSA. The table shows that the larger the hospital, the more the likelihood that it is located in an SMSA. (One should ignore the last four columns of this table.) Tables 39 through 44 present data on the dependent variable Y; through Ys. One is pleased to note that the standard deviation and ranges are small. Tables 45, 46, and 47 display, by stratum, the intercorrelations among all independent variables. These intercorrelations are meaningful in their own right. They show the extent to which measures traditionally used in hospital research are interrelated. Furthermore, they will help one to inter- pret certain relationships that have appeared in the regression models. 49 Fs aS TT Pas == TT = TABLES 51 Table 1.—Regional distribution of total nursing hours: continental United States ! Total nursing hours? Region and State Per adjusted Per statisti- Per civilian resi- patient day® cal bed* dent population® New England ....ccovvivvitiininnsensncnancnnnes 6.18 2,012 7.830 Connecticut .vevvencananes veensae 6.14 2,049 6.77 Maing. vorevrrvirernenione peessee 6.07 1,797 6.01 Massachusetts ........co000e cereeen 6.18 2,019 7.82 New Hampshire ......... sas 6.82 2,058 6.71 Rhode Island ............ sieves 6.16 2,186 7.34 Vermont .evesesaserevsnss ereesrsareanes reesane 5.77 1,779 7.89 Middle Atlantie .covcvevrnciernsnisirncesnrrncens 5.37 1,818 6.37 New Jersey ..vcovevncnsns . 5.45 1,839 5.71 New York ....evveaveeese 5.32 1,822 5.95 Pennsylvania ....ecieeeniiicinieniiinnans naive 5.40 1,804 7.43 South Atlantie ......evvvvevuirirrerenrannsnenes 5.93 1,903 6.32 Delaware ....... “3 . 5.81 2,036 6.15 District of Columbia. ‘ 5.94 1,875 11.50 Florida 6.05 1,871 6.90 Georgia . 6.75 2,155 5.87 Maryland .. 5.92 1,616 5.44 North Caroli . creresvanee 5.44 1,748 5.81 South Carolina ........ svevsvies 5.70 1,704 5.91 iit in ETT TE 5.99 1,978 6.28 West Virginia ....... SEER SRA ee 5.42 1,774 7.33 East North Central ..........coiiiiiiiiniinnnes 5.89 1,883 6.79 NBNois covvrenveravenanes 5.98 1,886 7.24 Indiana .....evnenn cere 5.88 1,906 6.76 Michigan ....civeniinnnn 5.83 1,901 6.03 Ohio +evervsvevrnsenses 5.91 1,958 6.98 Wisconsin ..uesvsivsrsnsnstsionttsnsnssniennnes 5.74 1,661 6.77 East South Central ........ccoviiviiniiienianenns 5.71 1,817 6.23 Alabama ...........e00 6.02 ,899 5.81 Kentucky covvivvennnns 5.78 1,909 6.61 Mississippi +.eivenenn 5.73 1,757 5.38 TENNESSEE +r vvernrnennenennessensssssesnsnnssans 5.44 1,721 6.77 West North Central ..........cooiiiiiiinnnnnnns 6.42 1,883 7.34 JOWR seisvessinsvescrvunrverivensssssssssvarne 5.58 1,570 7.54 Kansas ...... cereeen 5.80 1,652 7.83 Minnesota . . 5.55 1,626 7.72 Missouri . 5.82 1,851 6.25 Nebraska .. 5.86 1,614 8.22 North Dakot 6.11 1,558 9.04 South Dakota .......... . 6.13 1,599 6.72 West South Central .. 6.42 1,883 5.69 Arkansas ........vvuns 6.42 1,941 6.45 Louisiana ....cvvvunnn . 6.41 1,893 4.67 Oklahoma ..... teesersetseistrsessirenee ees 6.85 1,949 7.06 Texas ..vvesvrvens False ves A AYE 6.30 1,854 5.57 Mountain ........cevvtririririeritiittiticinnes 6.27 1,849 6.31 Arizona ...cvevvenvenss Crerereseen 717 2,264 6.31 Colorado +....vuovuvnee . Creare 6.49 1,960 7.42 1dabho sinscivnenvens “ee sesrssssnsesnnne ens 6.05 1,706 5.96 Montana ....coivinenn [I 5.35 1,389 6.86 Nevada «vvevvevvnennes vesverese 6.00 1,777 6.72 New Mexico ......... reeertenes 6.63 1,859 5.40 Utah vvervvvverrnens . revessereeene 5.25 1,766 4.29 WYOMING cvsorcsrssessvsmrsnsnssnsevsarsururses 5.76 1,433 7.38 PRelfio «evevueniieriernirrrerrrirnrersrarcisness 6.68 1,992 5.73 California ... 6.72 2,042 6.01 Oregon «.....cvvevess 5.92 1,686 6.16 WasBINZION sus conuisvessnssnsvosnsssvsvvnsssvnes 7.00 1,931 5.56 "Based on 4,662 hospitals. See page 41. 2? Excludes ward clerks and personnel employed outside Department of Nursing. ® For definition, see pp. 45, 46. * For definition, see pp. 33, 34. ® Population Estimates and Projections. U.S. Department of Commerce, Series P. 25, No. 468, October 5, 1971. 52 Table 2.—Rankings of States by total nursing hours Total nursing hours Per civilian Region and State Per adjusted Per statis- resident Averagel patient day tical bed population difference Rank Rank Rank — (A) (B) (€) IDI New England Connecticut «eevveseressvsvsnsnsnen 14 5 20 10.0 Maine ........... Crvnreriree 17 29 34.5 11.7 Massachusetts ... Levee 12 8 6 4.0 New Hampshire .. TATE we 4 4 25 14.0 Rhode Island ......ccoviviveninnnnns 13 2 12 7.3 Vermont sevesesresesravasessaresense 35 30 4 20.7 Middle Atlantie New Jersey .eeeececscsscassnnes 42 26 42 10.7 New York ....... 48 27 37 14.0 Pennsylvania ......... 46 28 9 24.7 South Atlantie Delaware ......co.us Cerseseressir eae 32 7 32 16.7 District of Columbia 24 21 1 15.3 Florida . 18.5 22 17 3.3 Georgia .. 5 3 39 24.0 Maryland .. 25.5 43 45 13.0 North Carolina “ 43.5 35 40.5 5.7 South Carolina ........covvee suEwy 39 38 38 0.7 Virginia coves invanuvvssivnvin sine FN 22 9 29 13.3 West Virginia ...ovevvvincnncnnnnnes 45 32 13 21.3 East North Central Illinois 23 20 14 6.0 Indiana ... 28 16 22 8.0 Michigan . 30 17 33 10.7 HO vueee . 27 11 16 10.7 Wisconsin .eevecivnninsensearerenens 37 40 20 13.3 East South Central Alabama .......... 20 18 40.5 15.0 Kentucky .. 34 15 26 12.7 Mississippi +evivviininnnans treesenas 38 34 47 8.7 Tennessee ......oevveeeveiennnnennns 43.5 36 20 15.7 West North Central Jowa .......... teesereetesntnanianes 40 46 8 25.3 Kansas cvococrsnenivnessvreserie 33 41 5 24.0 Minnesota . 41 42 1 23.3 Missouri ... 31 25 30 4.0 Nebraska 29 44 3 27.3 North Dakot 16 47 2 30.0 South Dakota .. 15 45 23.5 19.7 West South Central Arkansas .. 9 13 27 12.0 Louisiana .... 10 19 48 25.7 Oklahoma «.ovvevevrsenenersss 3 12 15 8.0 TeXB8 sovvesenretnnrenneatnnnsensene 11 24 43 21.83 Mountain ATiZONA ..vieveieiinrinctinsnrsinsenne 1 1 28 18.0 Colorado 8 10 10 1.3 Idaho ..... 18.5 37 36 12.3 Montana 47 49 18 20.7 Nevada 21 31 23.5 6.7 New Mexico . 7 23 46 26.0 Utah ...... 49 33 49 10.7 Wyoming ceeveveernconnnss Cretrasnes 36 48 1 24.7 Pacifie California ..... 6 6 34.5 19.0 Oregon v 25.5 39 31 9.0 ‘Washington ....... 2 14 44 28.0 * Average difference |D|={|A—B|+|A—C|+(B—-C)}|3 NOTE: p w=0.596 Pac= —0.105 53 p ve=—0.141 vs Table 3.—Regional distribution of nursing hours by individual categories: continental United States ! Region and State Nursing hours? Total number3 RN's No. LPN's No. Ancillary No. General duty No. Pet. General duty and bead nurse No. New England ..... Connecticut . Maine ...... . Massachusetts New Hampshire . Rhode Island .. Vermont ........ Middle Atlantic ........coonvenenn treestrenas New Jersey .... New York ... Pennsylvania South Atlantic . Delaware z District of Columbia ..... Florida Georgia ... Maryland North Carolina South Carolina Virginia wean West Virginia .... East North Central . Illinois . 2 Indiana East South Central ....... sans Alabama Kentucky Mississippi Tennessee 86,033,024 20,499,632 5,922,754 44,396,115 4,956,305 6,728,400 3,529,818 236,977,317 40,765,314 108,498,429 87,713,573 190,690,964 3,347,257 8,469,584 46,535,588 26,538,781 21,075,186 28,927,331 14,917,342 28,073,883 12,806,012 273,761,477 80,337,849 35,191,529 53,650,382 74,561,267 30,020,450 79,201,072 19,867,776 21,053,415 11,813,848 26,466,033 48,374,123 10,954,219 2,909,270 25,970,621 2,766,742 3,048,974 1,924,296 119,446,239 21,026,482 52,514,106 45,905,651 76,070,150 1,509,494 3,737,620 19,483,584 8,374,130 9,671,649 11,793,169 5,148,733 11,722,956 4,628,816 115,346,450 36,343,340 13,647,681 20,670,051 32,831,189 11,854,188 23,209,113 5,431,190 6,976,270 3,231,514 7,570,140 16,281,639 3,803,280 1,177,144 7,811,372 993,591 1,680,976 815,276 44,775,818 7,341,526 19,582,120 17,852,172 39,583,772 681,428 1,735,027 9,652,851 5,828,438 3,162,678 5,978,055 3,311,289 6,503,059 2,730,948 51,353,698 12,547,613 5,008,531 12,818,540 16,434,681 4,544,334 21,491,352 6,005,346 4,484,686 2,994,116 8,007,205 22,604,487 5,742,132 2,012,019 11,580,444 1,281,196 1,198,450 90,245 75,599,898 12,688,343 38,338,416 24,573,139 77,599,549 1,156,335 3,441,646 17,714,673 12,619,202 8,687,322 11,558,395 6,477,063 10,210,666 5,734,247 110,372,516 32,292,872 16,661,106 20,593,503 27,100,185 13,724,850 34,621,702 8,431,240 9,638,982 5,588,218 10,963,262 33,729,132 39 8,097,717 40 1,746,607 29 19,207,107 43 1,978,590 40 1,286,760 19 1,412,351 40 87,701,457 37 15,884,360 39 36,913,026 34 34,904,072 40 51,897,568 27 1,155,449 35 2,888,099 34 13,640,599 29 5,459,622 21 6,604,330 31 7,383,802 26 3,472,684 23 8,117,100 29 3,175,885 25 83,205,234 30 27,102,817 34 9,419,161 27 14,304,809 27 24,241,204 33 8,137,243 27 14,266,542 18 3,386,360 17 4,540,315 22 1,836,688 16 4,503,179 17 40,531,181 9,791,178 2,439,408 22,722,749 2,396,554 1,509,973 1,671,320 105,163,209 18,681,664 45,791,619 40,689,925 64,669,239 1,373,850 3,329,061 16,892,413 10,267,330 3,929,248 100,007,144 31,730,233 11,717,335 17,673,625 29,217,666 9,668,285 18,732,594 4,333,691 5,816,814 2,523,106 6,058,982 SS Table 3.—Regional distribution of nursing hours by individual categories: continental United States !—Continued Nursing hours? Region and State RN's LPN's Ancillary General duty General duty Total and head nurse number? No. Pct No. Pct No. Pct No. Pct. No. Pct. West North Central ..........coooiiiiiinnns 119,352,032 49,343,332 41 20,981,810 18 51,104,083 43 35,543,437 30 42,619,368 36 IOWA vuvyenunvvnvns 21,340,072 9,199,258 43 3,233,865 15 9,355,009 44 6,776,540 32 7,975,497 37 Kansas 17,323,340 6,813,762 39 2,369,527 14 8,873,032 51 4,689,369 27 5,814,631 34 Minnesota 7 29,489,333 13,942,728 47 5,894,284 20 10,032,810 34 10,514,176 36 12,402,559 42 Missouri .. % 29,120,807 10.282,069 35 5,907,145 20 12,931,594 44 7,127,654 24 8.755,018 30 Nebraska .... . 12,149,931 5,254,989 43 1,884,546 16 5,320,212 44 3,732,413 31 4,443,282 37 North Dakota . a 5,483,715 2,071,719 38 913,583 17 2,598,197 47 1.419.923 26 1,731,938 32 South Dakota «.eeeeeennuusueeresnnnnnneennn ; 4,444,834 1,778,807 40 778.860 18 1,993,218 45 1,283,363 29 1,496,442 34 West South Central ................ cernnne ..| 108,918,107 30,727,369 28 33,245,552 31 46,082,128 42 17,171,558 16 24,234,861 22 Arkansas 12,375,801 2,860,277 23 4,284,615 35 5,264,171 43 1,321,932 11 2,027,006 16 Louisiana 16,834,861 4,837,163 29 3,785,214 22 8.229,103 49 3,101,669 18 3,963,595 24 Oklahoma 17,912,613 4,856,196 27 3,620,993 20 9,505,086 53 2,714,451 15 3,843,931 21 TOXRS wovurrrssnsastvahinsnsh savas sssmnnnmmn 61,794,832 18,173,733 29 21,554,730 35 23,083,767 37 10,033,506 16 14,400,328 23 J CY 51,957,784 23,803,778 46 10,355,700 20 19,267,097 37 17,751,355 34 20,930,741 40 Arizona 11,135,890 5,345,686 48 1.832,707 16 4,113,475 37 4,041,520 36 4,838,176 43 Colorado 16,157,554 8,026,089 50 2,727,350 17 5,485,201 34 6,387,992 40 7,292,813 45 Idaho 4,250,393 1,812,921 43 1,602,896 38 970,731 23 1,157,575 27 1,471,669 35 Montana 4,743,328 2,063,467 44 854,244 18 2,166,588 46 1,472,704 31 1,788,887 38 Nevada ... 3,246,671 1,289,333 40 911,137 28 1,516,149 47 911,056 28 1,103,461 34 New Mexico 5,407,963 1,952,374 36 1,147,211 21 2,324,851 43 1,311,475 24 1,599,565 30 Utah ....... . 4,572,074 2,178,682 48 999,704 22 1.462,313 32 1,667,153 36 1,909,479 42 WYOMING + entanenenanenseaaaaneeseseeaeeeen 2,443,911 1,135,225 46 280,451 11 1,226.789 50 801,879 33 926,690 38 Preifle suiiiniadessnes ssi iiiinnsmsmintmsnsiioh 149,518,638 73,214,798 49 26,913,990 18 51,047,452 34 52,967,915 35 64,472,007 43 California 117,997,808 57,202,405 48 19,816,742 17 41,884,946 35 41,290,467 35 50,329,311 43 Oregon 12,936,854 6,360,254 49 2,156,584 17 4,986.261 39 4,353,483 34 5,584,206 43 Washington 18,583,976 9,652,139 52 4,940,664 27 4,176,246 22 7,323,965 39 8,558,491 46 * Based on 4,662 hospitals. See page 00. ? Exclude personnel employed outside Department of Nursing, and ward clerks. *RN, LPN and ancillary hours may not add up to total because of minor reporting inconsistencies by hospitals (see p. 42). 9g Table 4.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: total nursing hours per adjusted patient day (Y;) Initial model Final model 21 independent variables 10 independent variables Biased multiple correlation ... S51 .50 Unbiased multiple correlation 51 50 Standard error @ mean . 1.38 1.38 Degrees of freedom .. . 2839 2850 Intercept. wuiscinsicesess I .orevies rrreeesatnees tT cernes 7.66 7.99 Independent variables b beta t b beta t 1. .03 .10 2.52 04 BE] 4.31 2, 06 07 1.09 —_— —_— —_ 3. 80 06 1.65 1.29 .10 4.91 4. Total admissions .......eevvenevuneennennnas 29.74 .63 9.36 31.37 .66 10.35 5. Length of stay ... - .07 -.12 —5.62 - 07 -.12 — 5.86 6. Occupancy -2.17 -.19 —7.64 -2.53 -.22 —11.66 7. Bassinets per statistical bed 1.19 .06 2.58 _ — — 8. Births per statistical bed ...... ny .06 10 4.11 07 a1 5.39 9. Emergency visits per admission .............. sererirrrinne ovinmn I - 07 .05 2.42 .08 .05 3.30 10. Total visits per admission .............. "owweve coveevebeniiany SESE 01 .03 1.60 _ _ —_ 11. Surgical operations per admission 68 .09 4.87 .62 09 4.56 12. Statistical beds ......ccvvvineniinnnan... 00 21 1.98 — —_— —_— 13. Adjusted patient days ............. esa EEY ESE VRS NETRA RR en ee -5.79 -1.01 —8.88 -4.75 - .82 -12.74 MM. SMSA ...coovercrsisrssnsiiene .35 .10 4.59 35 10 5.52 15. Percent of population under 18 ....... pene - 01 - .03 -1.44 = _ _ 16. Percent of population 65 and over . .02 .05 2.02 —_ —_ —_ 17. RN’s per 100,000 population ..... . - .00 - 02 =1.08 — ad —_— 18. LPN's per 100,000 population .........eeseueeeunnneesnnnneeennnnneennnnnnnns .00 .03 1.70 — — —_ 19. Aides, etc. per 100,000 population .............. Ceetrasenanees ene -.00 - 04 -1.85 -— -_— —_ 20. Per capita income ............... Couns en ¥ 10.66 .04 1.45 —_ — —_ 21. Percent of families under poverty level .... .01 .03 1.14 -_ — — LS Table 5.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: RN hours per adjusted patient day (Y2) Initial model 21 independent variables 10 independent variables Final model Biased multiple correlation ........ccoevees Take 73 Jq2 Unbiased multiple correlation . J3 72 Standard error @ mean ..... 5 a7 Degrees of freedom .... . 2942 2953 INIETCEPE cevvevrinnrnnvensrvnsannsssess EARS Te Cervera 2. .33 Independent variables beta t b beta t 1. Total number of facilities ............ce0n. Ceseesrerraresiasssensstensrrees .01 06 1.85 = - —_ 2. Number of high tech. facilities .. . 07 a1 2.34 .10 a7 8.21 3. Fraction high tech. to total ....ccovviinevinnnnns CRs sR vase ven 22 .03 .88 —_— — —_ 4. Total admissions ... 9.98 21.43 66 15.54 5. Length of stay ... —2.06 — —_ —_— 6. Occupancy ..eeeeeerevssenesvsenss 9.05 —=1.08 - 14 —9.60 7. Bassinets per statistical bed ........cc0innnnn Th vevves 1.08 .08 4.38 1.54 J1 7.58 8. Births per statistical bed SREY .03 .06 3.39 —_ —— _ 9. Emergency visits per admission .........ccieiuiinnninanns EY 01 .01 94 — —_ od 10. Total visits per admission .......ceceeveennenns reeceseanenaranaes rrr es vay .01 07 4.94 .02 10 7.37 11. Surgical operations per admission .55 1 7.28 65 13 8.49 12. Statistical beds ................ - .00 - 13 —1.58 — _ = 13. Adjusted patient days ....... -1.79 —- 45 -5.11 —2.60 —- .66 —15.29 14, SMSA iii . 22 09 5.35 —_— — — 15. Percent of population under 18 .......ceeiniiiiiiiiiiiiiiiieiiiniiiiianennes —- .03 - .09 -5.27 — _ —_ 16. Percent of population 65 and OVEr ........eeeeeeneeensennssscsnscnneennns .. - .03 -.11 -5.80 — — — 17. RN’s per 100,000 population ... .01 42 25.75 .00 39 25.95 18. LPN's per 100,000 population - .00 - 05 —3.46 —_ — _ 19. Aides, etc. per 100,000 population ........ee..n EY RET eee - - 11 -=7.05 - .00 -.13 - 9.39 20. Per capita income .........ieeieininnnnnn 24.07 14 6.07 48.74 28 17.17 21. Percent of families under poverty level ... -= .01 - .06 ~-3.17 _ -_— — 8S Table 6.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: LPN hours per adjusted patient day (Y;) Initial model Final model 21 independent variables 10 independent variables Biased multiple correlation ... 52 «51 Unbiased multiple correlation . wwe oe 51 51 Standard error @ mean ....... .. 77 97 Degrees of freedom ..... we 2869 2880 INtercept .ouueniiiiniiiinnineiiarnnannnnns SSO woe EE 1.97 1.80 Independent variables b beta t b beta I. Total number of facilities ..........c..cuuue — .02 11 2.88 .02 12 2. Number of high tech. facilities % - 03 - .06 -1.00 — — 3. ‘Fraction high tech: 10 1012] tives invamnsmnees meni vtis vemie Elvis Re .50 07 1.87 — —_ 4. Tor] admisslons courses vevvvons Sorina sss enantio Faso sd ns nse semssns os 7.37 .28 4.16 9.26 35 5. Length of stay . .. - .02 - .07 —3.33 - .02 - .07 0: OCCUpBNCY vue venavsnss danatisn sat insane arenes wo - .02 — .00 - .16 — — 7. Bassinets per statistical bed - 38 - .03 —1.48 — a 8. Births per statistical bed ....... oe .02 .06 2.62 — — 9. Emergency visits per BAMIBBION +.sesimiroremenesvei vive vrvisiver cs ty iveniy .04 .05 2.57 .04 05 10. Total visits iper admission ..vuivvvvarsnsvinnvvsvs GEN rR Se - .00 - .00 - 12 — — 11. Surgical operations per admission . .e a1 .03 1.39 _ — 12. Statistical beds ....... 0 ccuiuuiiriiniirieriiiritir trite irene .00 33 3.08 .00 .29 13. Adjusted patient days ..........cvnniennenn AAA —-2.18 - 67 —6.00 -2.25 - .69 HM. SMSBA. ...covvnivn we rvnaminie vevmmns - - 04 - .02 - 92 — — 15. Percent of population under 18 ....... Cres er ert LIers trata sara sree - .01 - .02 —1.08 — —_ 16. Percent of population 65 and over .. —- .01 - .02 — .86 — —_ 17. RN’s per 100,000 population ........ a - .00 -.20 —9.58 - .00 -.20 18. LPN’s per 100,000 population .........ceeevuruenenensnenenennnnenenennnnnns .01 .33 17.99 01 32 19. Aides, etc. per 100,000 population - .00 -.11 =5.71 - .00 = 13 20, Per: capita INCOME +usipevsrrsnsnmsmanpos we —9.90 - 07 —2.40 —8.82 - .06 21. Percent of families under poverty level .00 .01 31 _ — 6S Table 7.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: aides, orderlies, and/or attendants hours per adjusted patient day (Y.) Initial model 21 independent variables Final model 9 independent variables Biased multiple correlation ...ucsuresecussnssin sens vss sassssevansnasassesons .49 .48 Unbiased multiple correlation .48 .48 Standard error @ mean ...... 1.12 112 Degrees of freedom ..... - 2933 2945 IEICE PE ett ieee ieee eee ta eee aaa erararaaanens 3.67 4.23 Independent variables beta t beta t 1. Total number of facilities ..... - .01 - .04 — .88 — = = 2. Number of high tech. faciliti .05 .07 113 _ ne — 3. Fraction high tech. to total - .28 - .03 - 73 _ — 4. Total admisSIons . cvversimss vou vinsasins Sos anEaes sven (00a rans ttn snne ns 5.99 .16 2.33 8.18 .21 3.61 5. Length of stay .. —- .03 —- .06 =3.05 07 =3. 6. Occupancy ;...ssosisresss -1.22 - 14 —5.44 -1.22 - 14 -7.39 7. Bassinets per statistical bed . 52 .03 1.42 — = — 8. Births per statistical bed ...... soe 01 02 3 — o— -_ 9.. Ertergency visits’ per admission ..csviscssssinicssrsssnsmorvesssnissainsane .03 .02 1.23 _ — hd 10; Total Visiis per Admission wus vines dens tvs dna STR TTR Sor Rese we dn aia aid - .01 —- .04 —2.42 — — —_ 11. Surgical operations per admission A. .02 .00 21 a= — - 12. Statistical beds ....uniiiiiii iii iii eres .00 .05 43 — — == 13. Adjusted DRUERLE QEYS «vruvnaninsninvrms sis sns ies as vi sss savas ss esas -1.38 —- .30 —2.62 —-1.36 -.29 —4.91 14 J5 .05 2.49 —~ — — 15 .01 .04 1.75 — — ~~ 16. Percent ‘of population 65 and OVEr «i uivvsvvivividnanivsvivsissvnssnnissini .05 43 6.21 .04 a1 6.29 17. RN's per 100,000 population - .00 —- .24 —11.64 - .00 - .24 —12.46 18. LPN's per 100,000 population . — .00 -.15 — 8.08 - .00 -= .15 — 8.08 19. Aides, etc. per 100,000 POPULATION: vo vuvwnicuvn vm eas sn avis sn sa asmnms a saegaens .00 14 6.83 .00 14 7.41 20. Per capita INCOME covuvivas bevavwnms ins —4.32 - 02 - 73 — —_— —_— 21. Percent of families under poverty level .01 .06 2.64 01 .08 4.28 Table 8.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: general duty nurses hours per adjusted patient day (Y;) 09 Initial model Final model 21 independent variables 10 independent variables Biased multiple correlation .............. ST EN a. .70 .68 Unbiased multiple correlation .. “ 69 .68 Standard error @ mean ........ . 76 7 Degrees of freedom ... 2914 2925 Intercept ......... 3 1.04 -.57 Independent variables beta b beta 1. Total number of facilities .. .01 06 1.84 — —_ — 2. Number of high tech. faciliti 04 07 1.47 09 16 1.52 3. Fraction high tech. to total ........ Pr Th SE EE pn .30 04 1.17 —_ — == 4, Total AQIRIGRIONG convrvncvwmimimins ea AE ES SORA SENSE SE EE Tea EE 15.76 .50 9.04 18.65 .60 13.36 5. Length of stay . we -.01 -.03 -1.82 — — ee 0: Occupancy vusivsvssuiovrivininsasisvares Cerseeesrteetretrrerrrnrenanrane -.B9 -.12 —=5.86 -.59 -.08 - 5.20 7. Bassinets per statistical bed ........c.ccvueunenn ee etereseeetetatattatnennn 84 .06 3.36 1.05 .08 5.11 8. Births per statistical bed .02 .04 2.13 — — —_ 9. Emergency visits per admission 02 02 1.04 —_ — — 10. Total visits per admission ......... .01 07 4.51 02 09 6.23 11. Surgical operations per admission .. ooo 28 .06 3.61 —_ _ — 32: Statistical beds cinmvivin nmi RRR SC RENT aA Sere A San ae -.00 -.14 -1.58 _ _ — 13. Adjusted patient days ............ curren enn Sess ey crvrens -.97 -.26 -2.72 -1.73 —.46 -=10.10 19 .08 4.53 — —_ — -.02 —.06 -3.18 —_ _ —_— 16. Percent of population 65 and over ........ccovuvennnn -.03 -.09 —4.57 - .03 -.10 —- 6.35 17. RN’s per 100,000 population 00 41 23.67 .00 24.85 18. LPN's per 100,000 population -.00 -.08 -5.32 _ _ — 19, Aides, etc. ‘per 100,000 POPUIRLIONR: 4uvisssnrsmmimssivisr sass swans uvss sassy -—.00 -.07 —-4.12 - .00 -.05 - 3.19 20. Per capita. Income: cocenrvevsvrrmssvsoes 18.86 a1 4.67 38.18 22 13.31 21. Percent of families under poverty level -.01 -.05 -2.47 _ —_— re Table 9.—Regression model for nonteaching, nonprofit hospitals, stratum I, dependent variable: general duty and head nurse hours per adjusted patient day (Ys) 19 Initial model Final model 21 independent variables 10 independent variables Biased multiple correlation ........ceevevineennenennnn srr cretienren 74 71 Unbiased multiple correlation 74 J Standard error @ mean .... 73 J Degrees of freedom ... pa ey ¥ 2014 2925 Intercept ciovivianiamssessravaEe Ey Ee 1.67 .66 Independent variables beta t beta 1. Total number of facilities ....... eesenrreen teerttesstrtsnarranns eerrrerrane 01 .05 1.65 — —_ — 2. Number of high tech. facilities . 07 Jd2 2.54 Jl 18 8.54 3. Fraction high tech. to total .....coeiniiniiniinennnneenennnnns eeeeaeeeaaes 17 .02 .68 —_ — —_ 4. Total admissions 17.52 55 10.52 22.03 69 15.85 5. Length of stay . - .01 -.03 -1.90 — — —_ 6. Occupancy ...... - .89 -]2 -6.13 -.72 -=.10 — 6.38 7. Bassinets per statistical bed ..... srervenee teerereiiniieaan terresaanes FN .86 .06 3.59 a2 05 3.53 8. Births per statistical bed ..... “ .02 .06 3.07 — — — 9. Emergency visits per admission .........c.viiiiiiniiiiieriiiinan. verse 02 .01 1.04 —_ _ —_— 10. Tora] visits. per adMISBION uv vvrvnvssnrsrsnsesssnvvnsiesssssrsssinssvavness 01 07 5.08 .02 a1 8.52 11. Surgical operations per admission @ 42 .08 5.71 67 14 8.86 12. Statistical ‘beds vusvevvivivirintiims sev a Ei RSE RR ASR RRR - .00 -.09 —1.03 —_ — —_ 13 =1.65 -.42 —4.81 —2.24 -.58 -13.12 14. 21 .09 5.39 — -_ — 15. - .02 —.08 —4.48 — — — 16. Percent of population 65 and OVEr ........civveeeisennensenncnsesscncnnenns - .03 -.10 -=5.40 - .04 -.13 -— 9.03 17. RN's per 100,000 population .... . .01 43 26.89 01 .50 35.40 18. LPN's per 100,000 population ......... CeEre Tee eEsRR Cee cersene ras aeas - .00 -.09 —6.08 _ —_— — 19. Aides, etc. per 100,000 population ........ec.uus PER ASS - .00 -.10 -6.30 - .00 -.08 — 5.43 20. Per capita income 22.10 13 5.73 —_ — — 21. Percent of families under poverty level ......ccovuiiiiiiieiiiririnncinnnnnnns - .01 —.06 -2.95 —_ _ — 4°) Table 10.—Regression model for teaching, nonprofit hospitals, stratum II, dependent variable: total nursing hours per adjusted day (Y;) Initial model Final model 21 independent variables 7 independent variables Biased multiple correlation .......c.civiiiiniiiiiriiiinirenrsieensinaens Peeves 52 49 Unbiased multiple correlation .. ’ . .50 48 Standard error @ mean ........ . 1.02 1.03 Degrees of freedom ... . 600 614 1BIEICEPE wuvvrrwnnnevnvnnvina rasan EER EE TA 10.54 9.44 Independent variables b beta t b beta t 1. Total number of facilities .......... eresertsssenisrrreans trsserrsreertranees - .06 -.23 -1.57 — — — 2. Number of high tech. facilities .48 64 3.17 14 19 4.45 3. Fraction high tech. to total .. -7.32 - .36 —-2.65 — —_ — 4. Total admissions .00 .00 .00 _— = — 5. Length of stay .. - .10 -.17 -3.02 -.10 - .16 -—3.86 6. Occupancy 65 04 72 —_— eves ad 7. Bassinets per statistical bed i.iiissinsssissessevrivcavssnnnee snsprserrevsen 2.02 .09 1.47 —_— — -_ 8. Births per statistical bed ...... » .06 J3 2.06 08 19 4.50 9. Emergency visits per admission .04 .03 a7 ~~ — - 10. Total visits per admission .......co0veuun rl - .00 - .01 -.27 —_ —_ cd 11. Surgical operations per admission .. Fo .55 .06 1.63 — — a] 12. Statistical beds ..........oiiniinen ees FN .01 1.69 6.28 .01 1.50 8.60 13. Adjusted patient days ........eeeines ne -—3.35 -1.92 —-7.61 -2.94 —-1.69 —-9.74 14. SMSA ....coviummurmsiinen inns we 19 04 1.08 = —_ — 15. Percent of population under 18 .......cviiiiiriiiinitinriniennennennenarnnnns - .08 - .23 —3.88 - .06 -.19 -3.79 16. Percent of population 65 and over .............. RR AREA AAR -.]12 -22 -3.93 - .10 - .18 -3.58 17. RN's per 100,000 population ik .00 .05 1.09 _ _ ~~ 18. LPN’s per 100,000 population - .00 - .06 -1.52 —_— _ -— 19. Aides, etc. per 100,000 population .........c...uue ARIE EEN .00 .08 2.05 _ —_— — 20. Per capita INCOME cuccivvivissvsvvvsvvens —16.11 - .08 —1.63 i — — 21. Percent of families under poverty level .... .00 .01 16 _ —_ — Table 11.—Regression model for teaching, nonprofit hospitals, stratum II, dependent variable: RN hours per adjusted patient day (Y.) €9 Initial model Final model 21 independent variables 7 independent variables Biased multiple correlation ............. .60 .55 Unbiased multiple correlation . .58 .54 Standard error @ mean ....... 71 73 Degrees of freedom ..... 607 621 Intercept ......... SFA 5.97 2.95 Independent variables beta t b beta 1. Total number of facilities ....... Crtestetenstenrrrranrans A - .02 -=.13 - .95 Saw — — 2. Number of high tech. facilities 22 .39 2.07 09 16 4.00 3. Fraction high tech. to total ............... —3.01 -.20 -=1.57 — _ — 4. Total admissions 1.92 13 1.30 _ —_ — 5. Length of stay . v — .08 =.16 -3.20 -— 12 -.27 -7.38 Gr DCouDRNLY wrsivversinisssssouvisrvvin a -1.24 -.10 —2.00 —-1.26 -.11 -3.09 7. Bassinets per statistical bed - .28 -.02 -.29 — -_— —_ 8. Births per statistical bed ...... oe .04 a1 1.83 — — — 9. Emergency visits per admission ....... SE, ASR CERES - .08 -.10 —2.56 - .09 -.11 -3.06 10. Total visits per admission ....... Vein SINAC I TSW SERRE SO .01 .05 1.23 _ — _ 11. Surgical operations per admission . “esse i .60 .09 2.56 bd _ —_ 12. Statistical beds ...ccercsvivninins Khare satus TR TERR Sea et .00 .26 1.05 _ _ — 13. Adjusted patient days ............ cereenes 8 Tarvin Te Taree Sees ee don re ae TA JR -.74 -.57 —2.41 - .20 -.15 —3.69 14. SMSA ...vvvvivnerrinrinsnnnes 14 04 1.19 — — — 15. Percent of population under 18 ... —- 05 -.21 -3.77 — —~ s— 16. Percent of population 65 and over - .08 -.19 —-3.67 —_ — — 17. RN's per 100,000 population .00 .40 9.35 .00 .33 9.23 18. LPN's per 100,000 population - .00 -.03 - .83 — _ — 19. Aides, etc. per 100,000 population ........ceveeeeeeeneenns A a - .00 07 =2.01 _ —_ —_ 20. Per capita income ..........ecviviininnn eee . 9.98 07 1.46 29.76 .20 5.80 21. Percent of families under poverty level ...... RE - .01 -=.07 -1.94 _ —_— -_ Table 12.—Regression model for teaching, nonprofit hospitals, stratum II, dependent variable: LPN hours per adjusted patient day (Y;) Initial model 21 independent variables Final model 5 independent variables Biased multiple correlation ..........ceevveunen. ss TERE TREE ER aE 45 .40 Unbiased multiple correlation .. 42 39 Standard error @ mean ........ 57 57 Degrees of freedom ... 602 618 Intercept -.09 1. Independent variables b beta t b beta 1. Total number of facilities .........e..... trirevenee sr errrena ny —— wwe .01 a1 69 02 15 3.90 2. Number of high tech. facilities . .05 4 | 54 —_ — — 3. Fraction high tech. to total .. - 98 - .09 - 64 — oe] — 4: Total 28MISSiOns wwenevrersssins onions es Ea 1.21 12 1.03 —_ _ — 5. Length of stay .. - .00 - .01 - 13 _ _ _— 6. Occupancy 1.02 J2 2.06 —_ —_— — 7. Bassinets per statistical bed ........ J CaP Ces SSR ves ET EYRE 2.01 16 2.64 2.05 16 4.35 8. Births per statistical bed ..... - 01 -.02 - 37 —_ == — 9. Emergency visits per admission 04 07 1.64 — ~~ _ 10. Total visits ‘per admission sesseesivsversvriavsisnvs cries .00 .01 .35 — —_— rs 11. Surgical operations per admission - 17 - 04 - 91 _ — — 12. Statistical beds .00 1.01 3.58 — —~— — -1.10 -1.19 —4.49 _ _ —_— .06 02 .62 —_— _ — 02 10 1.65 ~ ~~ — 16. Percent of population 65 and over 01 .05 .85 —_— -— —_ 17. RN’s per 100,000 population ...... . - .00 -.24 -5.02 - .00 -.25 -5.93 18. LPN's per 100,000 population ............c.... ENN, .00 16 3.67 00 16 3.77 19. Aides, etc. per 100,000 population ¥ -= .00 - .04 - 99 — —_— — 20. Per capita income ................. . —13.91 -.13 —-2.54 —-20.27 -.19 —4.83 21. Percent of families under poverty level ............ .00 .03 2 —_ — — S9 Table 13.—Regression model for teaching, nonteaching, nonprofit hospitals, stratum II, dependent variable: aides, orderlies, and/or attendants hours per adjusted patient day (Y,) Initial model 21 independent variables 6 independeat variables Biased multiple correlation ........oeeeeeeseeeeneeneesosscnsencenenns EN 47 Unbiased multiple correlation . 44 Standard error @ mean ..... .69 Degrees of freedom ... 610 Intercept civsvererininne 4.06 Independent variables beta 1. Total number of facilities - .30 =2.00 Ee —_ — 2. Number of high tech. fa . we 45 2.16 — _ _ 3. Fraction high tech. 10 total ...uuiuiiruiniineiueneereneseeennenensennennenns - 26 —1.83 _ — —_ 4: Total AAMISSIONS «.ovsmmurvrvnmrsmessraens cor orisss sess ess ynre wate resndy -.24 -=2.16 —_ — = 5. Length of stay . wie - .05 —- .86 — —_ —_ 6: OCCUPANCY wis smneviimn acon ss sas Se AR CRORES ER SA Ae SE al .10 1.68 — — — 7. Bassinets per statistical bed 04 64 _ — _ 8. Births per statistical bed ....... .06 97 —_ _— — 9. Emergency visits per admission .................. 09 2.00 _ — — 10. Total visits per admission ........ - .07 -1.62 em — _ 11. Surgical operations per admission . " .01 .33 — — — 32. Statistical Beds .oovvevervime soni van asee SE e S S 1.47 5.32 .00 .88 5.13 13. Adjusted patient days ........ieeuriinirieiiirtentatattttctratateetatrtanenes -1.34 —-5.14 .09 .95 —5.58 14. .01 21 — — — 15. - .16 -2.58 _ _ _ 16. Percent of population 65 and over - 13 -2.38 — — 17. RN’s per 100,000 population —- .18 -3.82 - .00 - —4.65 18. LPN's per 100,000 population ... - 17 —3.98 .00 -. -3.% 19. Aides, etc. per 100,000 population ... ....... hbase waa. .20 5.43 .00 5.91 20. Per capita income ..............o.... . - 07 ~1.35 _ — 21. Percent of families under poverty level ........... .08 1.94 .02 2.90 99 Table 14.—Regression model for teaching, nonprofit hospitals, stratum II, dependent variable: general duty nurses hours per adjusted patient day (Y;) Initial model 21 independent variables Final model 10 independent variables Biased multiple correlation ........cciiiiiiiaenn rarer eee Crrerrerenees .58 .56 Unbiased multiple correlation . .56 55 Standard error @ mean ....... 67 67 Degrees of freedom .. . we 601 612 IDLEICEPE sucivsunsnnrversronvnnises VEER a. Tes ErTEs yy 4.95 4.87 Independent variables b beta t beta t 1. Total number of facilities ............. wR CAREERS seven seners -.02 -.11 - .82 — — — 2. Number of high tech. facilities . 42 34 1.76 —_ —_ —_ 3. Fraction high tech. to total ...... -2.80 -.20 —1.56 —_— _ —_ 4. Total admissions ........cvvviivviinnnnnnnnes 2.67 .20 1.92 5.14 38 4.90 5. Length of stay - .05 -.13 -2.45 — — re 6. Occupancy - .87 -.08 —=1.48 -1.31 -.12 -3.29 7. Bassinets per sf ical bed ........ -.76 -.05 —- 84 — — — 8. Births per statistical bed ....... 05 JAS 2.41 04 13 3.31 9. Emergency visits per admission .... -.11 -.15 -3.59 a1 -.14 -3.81 10. Total visits per admission ..........cecvvevnns Carrere eh SR .00 .03 .84 —_— er 11. Surgical operations per admission . oe .62 10 2.81 69 a1 3.13 12. Statistical beds ................. J vere a artes rr .00 13 48 == — 13. Adjusted patient days . CER - 57 -—.48 -1.98 - 53 -.45 -6.02 14. SMSA ........... sees 07 .02 .58 — — — 15. Percent of populat nder 18 . PEAR REN AY - .05 -.22 -3.17 - .06 -.27 -5.79 16. Percent of population 65 and over ........ Crrterrrrersennee - .06 -.16 -2.98 - 07 -.18 -3.67 17. RN’s per 100,000 population . . .00 .36 8.28 .00 .36 9.56 18. LPN's per 100,000 population - .00 -.01 = .16 _ —_ _ 19. Aides, etc. per 100,000 population ...... Crverseevetee ens CHER EREREEREe - .00 —.06 -~1.72 — -_— — 20. Per capita income . . 4.49 03 70 — _ — 21. Percent of families under poverty level ....... Tr ee - 02 -.11 —-2.90 - .02 -.13 -3.58 L9 Table 15.—Regression model for teaching, nonprofit hospitals, stratum II, dependent variable: general duty and head nurse hours per adjusted patient day (Ys) Initial model 21 independent variables Final model 7 independent variables Biased MUNIPle: COTLEIRLION: ivvvvrevme snvuumnwinivsn sismiviase se sin saws vases swans 59 54 Unbiased multiple correlation .. .58 54 Standard error @ mean ........ .69 71 Degrees of freedom ... 601 615 Intercept ...... Nets eetsaeeesatettaatttatetattt esate tattttatteatrtttnterateraan 5.70 6.98 Independent variables beta beta 1. Total number of facilities ........ccvviiiiniinineiiiiiiiiiniiiriiintinsenens - .02 -.10 - — —_ — 2. Number of high tech. facilities . : .20 37 1.94 — —_ — 3. Fraction high tech. 10 1018] vcore nmmnsmmnoormsormssins sosmseers vom -2.91 -.20 -=1.57 —— _ — 4. Total admissions 2.09 15 1.46 —_ — —_ 5. Length of stay .. * - 07 -.16 -3.10 - 13 -.20 -—8.23 6. ‘OCOUPARCY cv viens v3 vassve vs vs soreuvE om wa wares ss vos Varese Foss esses 49 -1.16 -.10 -1.92 -1.27 -.11 -3.25 7. Bassinets per statistical Ded .....cicoviirrniirviisviinrnirsenivsrrrsrerins - .52 -.03 - 56 —_ —_ — 8. Births per statistical bed ..... 04 BE] 2.25 —_ rn — 9. Emergency visits per admission - .09 -.11 —2.85 = .11 -.14 —3.95 10. ‘Total ‘visits per admission iv ccvrsncrsveversvrensn essen BESET .00 .03 .85 —_— a= -_— 11. Surgical operations per admission ’ .60 .09 2.64 —_ ~~ -_ 132. Statistical beds isrenermcurani URE I ME NR SR ERR .00 .21 .81 -_ -— — 13. Adjusted patient days .. - 65 -.52 -2,17 — -_ — 18, SMSA eunuurninnnnsinnnsennnnes 209 03 76 j— f — 15. Percent of population under 18 - 05 -2 -3.80 - .06 -.24 -5.18 16. Percent of population 63 and Over ....civsiveviirarcrsrrvssrsrreviccirionve -.07 -.18 —3.42 - .09 -,22 —4.46 17. RN’s per 100,000 population .00 .38 8.87 .00 36 9.63 18. LPN’s per 100,000 population - .00 -.04 - 9 —_— ~~ — 19, Aides, ete. per 100,000 population: visas inrssnias ass isvais ve sh vase ve - .00 —.06 -1.82 _ — —-— 20. Per capita INCOME ...c.cevrvrnrrrncnrenees 6.08 .04 91 _ —_ — 21. Percent of families under poverty level .... - .02 -.11 -2.72 - .03 -.16 —4.39 89 Table 16.—Regression model, for-profit hospitals, stratum III, dependent variable: total nursing hours per adjusted patient day (Y;) Initial model 21 independent variables Final model 3 independent variables Biased multiple correlation .......... teterterttrttettstesterstrrtranaanarnrann .68 62 Unbiased multiple correlation . 62 61 Standard error @ mean ....... 1.88 1.91 Degrees of freedom .... 146 164 Intercept .......... eeererereeeeaaen 6.54 5.62 Independent variables b beta b beta 1. Total number of facilities ............. 14 .20 1.55 26 .36 4.63 2. Number of high tech. facilities 15 .08 .48 —_— _ — 3. Fraction high tech. to total .............. .76 05 43 — — _ 4. Total admissions ....... SE 48.69 .31 1.30 —_ ee —_ 5. Length of stay . - 13 w- 12 -=1.15 —_— me — 6. OCCUPANCY 4 tetenenineenanusnensnsesesesensnssassenensesnsesnsnessnenenens - 24 - .02 - 13 _ _ —_ 7. Bassinets per statistical bed ............... EE TT AREER -1.37 - .05 - 42 _— en —_ 8. Births per statistical bed ...... .03 .04 33 _ em _ 9. Emergency visits per admission -.13 - .05 - .80 —_ — -_— 10. Total visits per admission .............. Sires tess. reverse. .02 J2 1.78 _ —_ _ 11. Surgical operations per admission . . .20 .02 17 — — — 12. Statistical beds ..... — ws SE 07 .98 3.98 .08 1.04 7.31 13. Adjusted patient days -33.83 -1.44 —4.81 —29.38 -1.25 —9.05 14. SMSA - 25 - 05 - .50 —_ —_ — 13: Percent of population under 18 ouriiriivisrinnvissesiasnesmmensisinndve vine .06 08 .78 — — — 16. Percent of population 65 and over ........ SAAR ERT . .05 09 70 — —_ —_ 17. RN's per 100,000 population ....... - .00 00 .05 —_ —_ —_ 18. LPN’s per 100,000 population ............ - 00 - 04 - .54 —_— —_— — 19. Aides, etc. per 100,000 population .............. CesT Ye TRY veaeee .00 .02 .26 — —_ —_ 20. Per capita IBCOmE isvescasvinsveiverans -27.81 - .09 - 51 — — —_ 21. Percent of families under poverty level - 07 -.24 -1.61 —_ _ -— 69 Table 17.—Regression model, for-profit hospitals, stratum III, dependent variable: RN hours per adjusted patient day (Y:) Initial model Final model 21 independent variables 4 independent variables Biased multiple correlation ... a5 69 Unbiased multiple correlation . a1 .68 Standard error @ mean ....... eererecenenas . « .98 1.01 Degrees of freedom .... -e 156 173 Intercept ............ ereeaen EN er e——_— PO RERATA S ors 2.91 2.56 Independent variables b beta t b beta 1. Total number of facilities «sivivverrunisnavenssnrinivennine sestnerrarnerans a1 27 2.45 16 42 2. Number of high tech. facilities .... 08 07 .50 — = 3. Fraction high tech. to total ........... -.31 -.03 - 35 — o— 4. Total admissions ...evssersessenirrnrnsssrarstssssrersrnmnerrnsensrrensssss 22.49 25 1.20 —_ — 5. Length of stay ... vee - .05 -.08 - .90 — — 6. OCCUPANCY +uvvnruenrnnnannenenensnsnenennns oe -1.35 -.16 -1.56 -2.57 -.30 7. Bassinets per statistical bed . 05 00 .03 —_ — 8. Births per statistical bed ..... - .01 -.03 - .30 —_ —. 9. Emergency visits per admission - .08 -—.06 —1.01 — a 10. Total visits per admission ........ mw he hop — . .00 .01 23 _ soi 11. Surgical operations per admission “eens . 1.51 21 2.54 3.17 45 Y2.. Statistical Beds: .ovvenrvvsvvrvnsnorsnsss ersnovs dass iss nas ani RE .01 25 1.16 —_ — 13. Adjusted patient days ............ . —9.02 - 67 —2.49 —-3.35 =.25 14; SMSA icine vo vs .02 .01 06 — - 15. Percent of population under 1 revs Rese - .00 - .01 - 07 —_ — 16. Percent of population 65 and over .............. EEE RR - 02 - 06 - .56 —_ es 17. RN’s per 100,000 population ..... “ .00 21 2.55 —_ —_ 18. LPN's per 100,000 population ........ envrarens eR Gee vs Eres Svea - .00 - .10 -1.49 — — 19. Aides, etc. per 100,000 population ........ serve EA SOR, . - .00 - .06 - 9 — — 20. Per capita income ................ severens 11.08 06 40 —_ _ 21. Percent of families under poverty level .........ccevvuuunn... -= .01 - 04 - 32 — — 0L LPN hours per adjusted patient day (Y3) Table 18.—Regression model, for-profit hospitals, stratum III, dependent variable: Initial model 21 independent variables Final model 4 independent variables Biased multiple correlation .. .59 46 Unbiased multiple correlation S51 44 Standard error @ mean .. .78 .81 Degrees of freedom ...... 148 165 JIERCEPY. covivvivsvsensniv ivr vienie CEE CARER Cn sm ae eee ommnmninn 6.83 2.59 Independent variables b beta b beta t 1. Total number of facilities ............... trererearirannenen I — - 09 -.35 -2.49 _ — —_— 2. Number of high tech. facilities .. ove 22 .30 1.71 12 17 2.23 3. Fraction high tech. to total .......cc0uu..... CE CR - 12 -.02 - .16 — — _— 4. Total admissions ........evvveenennnnnnnnnn TT TTT rR 24.20 42 1.62 _ _— so, 5. Length of stay ... - .03 -.09 - .80 — _— —_ 6: OCCUPANCY ivvvmuvmimmnnmsna sper se a i A Fae SAA ir Core - 69 -.12 - .96 -1.32 -.23 -3.26 7. Bassinets per statistical bed ..........coeeveuniiiieinniiiirnnnineirenaans. .45 .05 34 _ -_ = 8. Births per statistical bed ...... - .07 -.24 -1.85 _ -_— a 9. Emergency visits per admission .............eeeiunerunnrennnnnnnnnnnnnnnn. .02 .02 .33 —_ — — 10. Total visits per admission ............eeevviuevnrunnnnn. TLL Tra .00 .03 48 _— —_— — 11. Surgical operations per admission -.75 -.17 -1.55 — — — 12: Sntistical Beds ...orvvriciiissnsirvararonersmnsen ten sirs tn sane in stoner 01 .35 1.32 _ _ ~ 13. -3.67 —-.42 -1.26 _ —_ -— 14. 24 13 1.16 _ —_ =~% 15. - .05 -.18 -=1.58 — ~~ oe 16. Percent of population 65 and over . - .01 -.02 - .18 _ _ ad 17. RN’s per 100,000 population ... lion - 00 -.00 - .03 _ _ — 18. LPN's per 100,000 popnlatlon «s.crsssiasisvssnns sion suis issetn ism tns ams .00 .20 2.44 01 27 3.84 19. Aides, etc. per 100,000 POPUlBION ....uernrneernneuneneenesnennsnnsnnnnns -— -.09 -1.16 _— _ ei 20. Per capita income .........c0vvviunnnnn —81.34 —.66 —3.63 —34.48 -.28 —3.66 21. Percent of families under poverty level .............oeeeeueennrnennnnnnnns. - -.36 -2.23 -_ -_ es 1 Table 19.—Regression model, for-profit hospitals, stratum III, dependent variable: aides, orderlies, and/or attendants hours per adjusted patient day (Y,) Initial model 21 independent variables Final model 4 independent variables Biased multiple correlation ......ueueeeenenenenenenenenenenenenenenenenenenens .59 49 Unbiased multiple correlation S1 47 Standard error @ mean ...... 1.34 1.38 Degrees of freedom ... 154 171 Intercept corsinssvsnnunsinsb ETE Cae SEA —4.41 1.25 Independent variables b beta t b beta t 1. Total number of facilities .......uiuniniinienininiienenenernrennnnnnnnnenns .10 23 1.70 — — m= 2. Number of high tech. facilities we - .08 - .07 - .38 — — = 3. Fraction high tech. to total ....ovuiiiiuniuiiiiiiiieieiieieeneenannennnns 85 .08 69 — —_ = 4, Total admISSIONE ...vevvercvorsnarens sosseni stem smmres ss mise eR Caen eR se 29.68 29 1.11 53.15 .53 3.37 5. Length of stay .. .00 00 .03 —_ _ _ 6. OCCUPANCY wvnn smi nnn Vap eR HELV 1.41 14 1.13 — —_— — 7. Bassinets per statistical bed ......c.iiuiiiiiiiiiiniiiiiii iii iia -2.93 37 -1.30 = — — 8. Births per statistical bed ...... 10 .20 1.60 — —_ — 9. Emergency visits per admission - .03 - .02 -.26 — — — 10. “Total ‘visits Per AdmIBEION ..ucu sanminmrsapsrsnssws vp vv anes vara ws .02 14 2.02 — — —— 11. Surgical operations per admission - 39 - .05 - 49 —_ — a 12. Sratistical beds .vvererrinisserrenanaisins 05 95 3.63 04 80 5.21 13. Adjusted patient days .......uiiiiiiieiiieeraiteanattaaaaat eaten —23.14 -1.52 —4.66 —21.48 —=1.41 —7.08 14. SMSA oie - .56 - .18 —1.63 — — — 15. Percent of population under 18 .10 22 2.00 — em — 16. Percent ‘of population 63 ABA OVEr waaauveans sevrssn vine step ssievs dpse sess .08 21 1.56 — — —_ 17. RN's per 100,000 population . wr -— .00 - .18 -1.72 _ — — 18. LPN's per 100,000 population .svssssnssvsms sussmssnssssnnonnivssn is snniias - .00 - .05 -~ 57 — — — 19. Aides, etc. per 100,000 population ......eeeeereeseneenenneneeneenensennnnns .01 21 2.72 01 19 2.64 20. Per capita income ..........oivvennan. 56.36 26 1.48 _ — _ 21. Percent of families under poverty level - .02 — .08 - 52 _ — cL Table 20.—Regression model, for-profit hospitals, stratum III, dependent variable: general duty nurses hours per adjusted patient day (Ys) Initial model 21 independent variables Final model 4 independent variables Biased multiple correlation ............ A .66 Unbiased multiple correlation . 67 .65 Standard error @ mean ..... .83 .85 Degrees of freedom 173 Intercept ......... . 1.19 1. Independent variables beta t b beta 1. Total number of facilities .08 25 2.17 11 .35 5.28 2. Number of high tech. facilities . - .03 - .03 -.19 —_ _— — 3. Fraction high tech. to total .....civvevinnnnnennnns veerrennn SPER -.19 - .03 -.25 —_ on — 4. Total admissions 38.89 55 2.46 —_ _ —_ 5. Length of stay .. .01 02 16 — —_ — 6. Occupancy .....eeeeens -.72 - .10 - 98 -1.38 -.20 —3.43 7. Bassinets per statistical bed ......cieeeiinrnnciniiiiiiiiiiiiiiiiiiiiiiinne. -.19 - .02 - 14 — —_ 8. Births per statistical bed ....... “ - .03 - .08 -.76 —_ — _ 9. Emergency visits per admission ........ciieiteiiirtiiiiiiiiiiiiiiittnaaenens .06 .05 89 —_ — — 10. Total visits per admission .........c0uns .00 05 .83 — —_ _ 11. Surgical operations per admission . 97 A7 1.93 1.91 .34 4.88 12. Statistical beds .....eouvvuvvenvinninreninrnennns trerresrrrtnens tessaresaans 01 21 94 —_ —_ — 13. Adjusted patient days ......... TT senvee -6.51 - 60 -2.13 —~ -~ — 14. SMSA .........0 eee . .00 .00 02 —_ _ —_ 15. Percent of population under 18 ..... CER ARTA SAARC Ey CrrEeee - .01 — 03 - .36 —_ I. = 16. Percent of population 65 and over ..........cevveeens - .01 — 02 -.21 _ —_ _ 17. RN’s per 100,000 population .. .00 23 2.59 —_ — — 18. LPN’s per 100,000 population . - .00 -.15 -2.23 - .00 -.18 -3.13 19. Aides, etc. per 100,000 population .......... SCRE ees SSR TERR - .00 -.12 —1.82 —_ _ — 20. Per capita income ............0nunn ven 9.21 .06 .39 —_ —_ —_ 21. Percent of families under poverty level 01 04 29 _ —_ — gL Table 21.—Regression model, for-profit hospitals, stratum III, dependent variable: general duty and head nurse hours per adjusted patient day (Ys) Initial model 21 independent variables Final model 5 independent variables Biased multiple correlation .........cou0ue.n CARRERE sree Crna ene 74 71 Unbiased multiple correlation . .70 70 Standard error @ mean ..... .86 .87 Degrees of freedom ..... tr 156 172 1 Ves sein R Tee SRR ST Cer Ree eis 1.88 —.46 Independent variables beta b beta 1. Total number of facilities ......ccovvueununn. 09 26 2.29 12 34 4.51 2. Number of high tech. facilities . .e ol .04 04 .28 — — — 3. Fraction high tech. 10 total ...iuuiiuiueineeiniieenneneenresensenecannnens —- 40 -.05 -.51 _ — _ 4. Total admissions .cvevervrenssraven 23.14 .30 1.41 41.06 53 3.96 5. Length of stay . - .03 —.05 - .56 _ — — LT -.25 -.03 -.32 —_ —_ —_ 7. Bassinets per statistical bed ........ci0iuinnunn NE eee - .28 -.02 -.20 —_ —_ — 8. Births per statistical bed ....... . 01 02 16 _ —_ — 9. Emergency visits per admission ..vsvesivseessnsevsrnsinssrsnssansessin rene — .01 -.01 - 21 —_ — -_— 10. Total visits per admission ........ccvveiuvurununnenes Es PRL PR C R .00 .00 .05 — — _ 11. Surgical operations per admission . 1.30 21 2.50 1.94 32 4.42 12. Statistical beds wisvsivrvrvsenivsimeaniias river esns ss trian da sasnssanesensns .01 .26 1.22 _ — _ 13. Adjusted patient days . —6.88 —-.59 -2.17 —17.89 —.67 -5.19 1: SMSA: corvircvimrvevinvurenvnevy . .08 .03 37 pe — ~~ 15. Percent of population under 18 ........ccvviuuennnnn CPSs EE YER - .00 —-.00 - .00 SE —_— — 16. Percent of population 65 and over ..........c... ERE Cs ee ee - .00 -.01 - .07 — —_ — 17. RN’s per 100,000 population ...... . .00 25 2.99 .00 29 4.26 18. LPN’s per 100,000 population ........ceeeveuunes SER A pr os - .00 -.18 -2.78 —_— _ _ 19. Aides, etc. per 100,000 population ............ EE EERE Se - .00 -.12 -191 —_— _— _ 20. Per capita Income ....ccvvvvenvennens -2.26 -=.01 - .09 _ — _— 21. Percent of families under poverty level - .01 -.04 - 32 -_— — _— PL Table 22.—Regression coefficients for use in computing estimated patient day, stratum I values of nursing hours per adjusted Yi Ys Ye Total Ys Ya General General duty Predictor or nursing Yz LPN Aides, etc. duty and head nurse independent variable hours RN hours hours nursing hours nursing hours hours Intercept 7.98881 .32580 1.80328 4.23146 - .57319 66162 1. Total number of facilities ....... SENT, eh .04118 — 102310 — ~~ en 2. Number of high tech. facilities . —_ 10222 — — .09368 .10699 8. Fraction high tech. 10 1018] ivvivvs vinrmvi issn ss ss EAE RSET E EN SRT ERED 1.28870 -— —_ = hd — 4. Total admissions ... 31.37044 21.42647 9.25729 8.17763 18.64793 22.02927 5. Length of stay .. —- 07134 —_ — .02466 — .03445 — —_ 6. Occupancy — 2.52897 — 1.08030 —_ —1.22063 —- .58731 - .72145 7. Bassinets per statistical bed .....coiviiiiiiiiiiiinen PETE PRN — 1.53998 — — 1.04876 71568 8. Births per statistical bed ......... .06533 —_ _ —_ — 9. Emergency visits per admission .08385 -_ .04296 —_— i == 10. Total visits per admission ......iveveerueeesessssesssssssceanssnns wees A — .02009 — — 01713 02316 11. Surgical operations per admission .61886 64625 — — — 66667 12. Statistical beds ............ cd — .00316 — — = 13. —4.74797 — 2.60254 —2.24665 —1.36189 — 1.73240 —- 2.24320 14. SMSA ....cescivinvesstinsiosios i .34858 — sn fi — — 15. Percent of population under 18 ...iiiviiiiiiiiiiiiiiiiiii ities — — —_ — — — 16. Percent of population 65 and OVer .......ceevevsnncnnnns Sennen errreneer _— —_ —_ .04110 — .02826 — .03941 17. RN’s per 100,000 population oe — .00482 — .00198 — .00341 .00466 .00606 18. LPN's per 100,000 population ........eevsvetvasensssarsssnnssnsnssnnssannes —_ —_ .00660 — .00430 —_ — 19. Aides, etc. per 100,000 population ...... —_— — .00232 — .00192 .00309 — .00082 - .00139 20. Per capita income — 48.74370 —8.81834 _ 38.18416 —_ 21. Percent of families under poverty evel ‘er — _ _ .01254 -_ —_ SL Table 23.—Regression coefficients for use in computing estimated patient day, stratum II values of nursing hours per adjusted Y1 Ys Ye Total Ys Ys General General duty Predictor or nursing Yaz LPN Aides, etc. duty and head nurse independent variable hours RN hours hours nursing hours nursing hours hours Intercept 9.43547 2.95467 1.56376 2.16786 4.87008 6.98436 1. Total number of facilities .... aii —_ .01971 -— —— —_ 2. Number of high tech. facilitie: 13979 .08830 _ — -_— - 3. Fraction high tech. to total ... — _ — — _ —_ 4. Total admissions .......c.ccvvuuunnn — —_ — — 5.14461 _ S. Length of stay — .09822 — .12268 — — — —- 12751 6. Occupancy ..... monn EE REA CREO _ —1.26385 — — —1.30600 —1.27158 7. Bassinets per statistical bed ............... TIT Cre a ne — 2.05029 — — _ 8. Births per statistical bed . 08457 —_ — —_ 104090 —_ 9. Emergency visits per admission ............iieiiiiiiiiiiiiiiiiiiiiieiiaaens —_ — .08890 _ — - .10780 = .11078 10. Total visits per admission ....... —_ — ms — —_ —_ 11. Surgical operations per admission . wy — — — _ .68688 -_ 12. Statistical beds .......... SHR Se ee SEER ER SR eR Se ee Ee .00909 _ —_ .00347 —_— —_ —2.93541 — .19950 _ —1.08579 — .53378 —_— — .06498 _ on —_ — .06470 — .06054 16. Percent of population 65 and OVEr ......c.ceeeeverierernnrensecnccssnnnnnse — .09651 _ — — - .06715 — .08591 17. RN’s per 100,000 population ....... —_ .00326 - .00180 - .00171 .00329 .00346 18. LPN’s per 100,000 population ............ SR CELA ER ARR v en — .00225 — .00284 —_— —_ 19. Aides, etc. per 100,000 population ........cceeevereinrnieirnnieneieienenns — — — .00301 = _ 20. Per capita income ...........ceeveiennnnan — 29.75701 -—20.27120 — _ — 21. Percent of families under poverty level — —_ — 01726 - .02086 — .02694 9L Table 24.—Regression coefficients for use in computing estimated patient day, stratum III values of nursing hours per adjusted Predictor or independent variable Y1 Total nursing hours Yz RN hours Ys LPN hours Ya Aides, etc. nursing hours Ys General duty nursing hours Ye General duty and head nurse hours Intercept 5.61727 2.56179 2.58701 1.24510 1.33448 — .45670 . Total number of facilities ....cevuievnennnnnnns Pee eR ee crrrsaraen . Number of high tech. facilities . Fraction high tech. to total ..... 25544 16492 12368 11282 11725 Total admissions Length of stay ... ow OCCUPANCY .ovvernevnne —2.57420 — 1.31859 53.14815 —1.38461 41.06314 . Bassinets per statistical bed ........ . . Births per statistical bed ... . Emergency visits per admission ........ Il [rr o-oo . Total visits per admission . Surgical operations per admission .. a . Statistical beds ........ sestereerrtrertrtrerenrrennirn brane ve smmen 07595 3.17284 .03820 1.90803 1,94128 Yt aw . Adjusted patient days . SMSA . Percent of popula ion under 18 —29.38419 —3.35121 —21.47713 — 7.88737 bt bd bk ON . Percent of population 65 and over ........ceevevenennnns Sree . RN’s per 100,000 population > —_— . LPN's ‘per 100,000 population sc cesisssnsvassnssnsvessnsnessssvssssvsns . .00513 — .00430 NR = 00 L Aides, etc. per 100,000 popRISHOR vv cvervnsvasvnnsnsvansnvetsvanonsvnssnse L'Per capita Income ....ccuciiviinvenenas . Percent of families under poverty level —34.48297 8 - © = I LL Table 25.—Total number of facilities: variable X; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 2.657 1.392 1.000 6.000 5.000 101 21- 40 258 3.872 1.889 0.000 13.000 13.000 102 41- 60 235 5.770 2.324 1.000 13.000 12.000 103 61- 80 144 6.785 2.814 1.000 16.000 15.000 104 81-100 108 7.769 2.702 1.000 14.000 13.000 105 101-140 127 9.425 2.785 2.000 17.000 15.000 106 141-200 85 12.047 3.236 1.000 19.000 18.000 107 201-300 34 14.971 3.252 9.000 24.000 15.000 Non-Government 201 21- 40 273 3.971 1.847 0.000 9.000 9.000 202 41- 60 262 6.076 2.551 1.000 16.000 15.000 203 61- 80 232 7.405 2.498 1.000 18.000 17.000 204 81-100 222 9.108 3.235 0.000 25.000 25.000 205 101-140 304 10.750 2.622 4.000 18.000 14.000 206 141-200 311 13.209 3.169 1.000 26.000 25.000 207 201-300 250 15.604 3.214 6.000 24.000 18.000 208 301-400 83 17.470 3.405 3.000 26.000 23.000 209 401+ 25 19.480 3.949 14.000 31.000 17.000 Total vcives tesrseesattetestnssantrrrenees 2,988 8.910 4.880 0.000 31.000 31.000 II. Teaching hospitals Government 119 401+ 52 25.192 4.116 15.000 31.000 16.000 Non-Government 216 141-200 24 15.042 3.316 10.000 21.000 11.000 217 201-300 154 17.312 3.437 9.000 27.000 18.000 218 301-400 173 19.936 3.601 1.000 29.000 28.000 219 401+ 230 23.300 3.641 15.000 31.000 16.000 Total..ovviuiniiinenennnnns cesssersrsrsans 633 20.766 4.611 1.000 31.000 30.000 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 3.327 2.102 0.000 10.000 10.000 302 41- 60 51 5.647 2.763 1.000 11.000 10.000 303 61- 80 23 6.739 3.018 2.000 12.000 10.000 304 81-100 27 9.519 2.792 3.000 15.000 12.000 305 101-140 26 9.038 2.919 4.000 16.000 12.000 Total... cvvisvinvinnes Ernie ures Creesnserenes 179 6.190 3.505 0.000 16.000 16.000 8L Table 26.—Number of advanced technology facilities: variable X, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government , 100 0-20 35 143 .430 0.000 2.000 2.000 101 21- 40 258 318 592 0.000 3.000 3.000 102 41- 60 235 .583 .782 0.000 4.000 4.000 103 61- 80 144 722 904 0.000 4.000 4.000 104 81-100 108 1.046 1.062 0.000 5.000 5.000 105 101-140 127 1.472 1.240 0.000 5.000 5.000 106 141-200 85 2.788 1.597 0.000 7.000 7.000 107 201-300 34 4.147 1.395 0.000 6.000 6.000 Non-Government 201 21- 40 273 .267 .541 0.000 3.000 3.000 202 41- 60 262 .626 847 0.000 5.000 5.000 203 61- 80 232 927 1.023 0.000 5.000 5.000 204 81-100 222 1.590 1.375 0.000 6.000 6.000 205 101-140 304 2171 1.363 0.000 6.000 6.000 206 141-200 311 3.251 1.499 0.000 7.000 7.000 207 201-300 250 4.152 1.392 0.000 7.000 7.000 208 301-400 83 4.723 1.300 1.000 8.000 7.000 209 401+ 25 5.120 1.166 3.000 7.000 4.000 Totalicvvsseviinrivsvssrnsminvonrsrvsvovssss 2,988 1.687 1.786 0.000 8.000 8.000 II. Teaching hospitals Government 119 401+ 52 6.442 1.335 3.000 8.000 5.000 Non-Government 216 141-200 24 3.542 1.532 1.000 6.000 5.000 217 201-300 154 4.227 1.360 1.000 7.000 6.000 218 301-400 173 5.220 1.355 0.000 8.000 8.000 219 401+ 230 6.217 1.143 2.000 8.000 6.000 Total.vevvnes tessersesttrestasattrtesnarernen 633 5.378 1.568 0.000 8.000 8.000 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 173 A474 0.000 2.000 2.000 302 41- 60 51 .922 1.074 0.000 4.000 4.000 303 61- 80 23 1.000 1.044 0.000 4.000 4.000 304 81-100 27 2.111 1.050 0.000 4.000 4.000 305 101-140 26 2.077 1.412 0.000 5.000 5.000 Totals evevssrens terre trrr errs srs erry 179 1.061 1.236 0.000 5.000 5.000 6L Table 27.—Number of advanced technology facilities as a fraction of total facilities: variable X; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0-20 35 .033 .095 0.000 .333 .333 101 21- 40 258 064 117 0.000 .500 .500 102 41- 60 235 .088 A117 0.000 .500 .500 103 61- 80 144 .087 .099 0.000 .364 .364 104 81-100 108 116 104 0.000 .364 .364 105 101-140 127 147 Jd11 0.000 .455 .455 106 141-200 85 217 .106 0.000 429 429 107 201-300 34 .281 .098 0.000 .556 .556 Non-Government 201 21- 40 273 .053 .108 0.000 .500 .500 202 41- 60 262 .087 115 0.000 667 667 203 61- 80 232 110 J12 0.000 .500 .500 204 81-100 222 156 J22 0.000 .556 556 205 101-140 304 192 104 0.000 .462 .462 206 141-200 311 241 .094 0.000 .455 455 207 201-300 250 .264 .073 0.000 .438 .438 208 301-400 83 W272 .062 .059 .385 .326 209 401+ 25 267 .062 176 375 .199 Total........... LS EET - 2,988 .146 129 0.000 667 667 II. Teaching hospitals Government 119 401+ 52 .258 L045 143 .353 .210 Non-Government 216 141--200 24 .228 .076 .091 .353 .262 217 201-300 154 245 067 .071 .500 .429 218 301-400 173 .261 .059 0.000 438 .438 219 401+ 230 269 .049 J25 412 .287 i 633 .258 .060 0.000 .500 .500 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 .047 131 0.000 .500 .500 302 41- 60 51 138 179 0.000 1.000 1.000 303 61- 80 23 J22 159 0.000 .333 .333 304 81-100 27 217 102 0.000 .400 .400 305 101-140 26 W222 127 0.000 571 571 Totaly vvevnmivenss SEES OY sesvenase 179 134 154 0.000 1.000 1.000 08 Table 28.—Total admissions in 100,000: variable X, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 .006 .003 .002 011 009 101 21- 40 258 011 .004 .003 027 .024 102 41- 60 235 .018 .006 .005 035 .030 103 61- 80 144 025 .009 .003 L061 .057 104 81-100 108 .031 .011 .011 073 .061 105 101-140 127 .041 014 014 .085 071 106 141-200 85 062 017 .016 099 .082 107 201-300 34 094 .030 .016 144 .128 Non-Government 201 21- 40 273 .010 .004 001 .027 .026 202 41- 60 262 .018 .006 .003 .037 034 203 61— 80 232 .024 .008 005 054 .049 204 81-100 222 .032 .010 .006 067 .060 205 101-140 303 044 L013 .012 .085 073 206 141-200 311 .063 .017 015 118 .103 207 201-300 250 090 L022 023 157 134 208 301-400 83 129 025 .076 .218 142 209 401+ 25 171 L031 110 .227 116 Total....... teetieetttteetttensttrrssnerannns 2,987 .040 .034 .001 sa .226 II. Teaching hospitals Government 119 401+ 52 .165 055 .070 .306 .236 Non-Government 216 141-200 24 072 L021 .031 125 094 217 201-300 154 .090 .021 .035 153 118 218 301-400 172 124 025 062 192 130 219 401+ 230 189 059 046 532 .486 Totalevsssnsorsnrrsrnssnrressssorsvsvnenss ness 632 141 060 .031 532 .500 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 011 .005 .001 027 .026 302 41- 60 51 .019 .007 .006 .039 .033 303 61- 80 23 .028 011 .011 .048 .037 304 81-100 27 .039 011 .015 054 .038 305 101-140 26 046 014 .015 .082 .066 Total covnivesvrvines PEERS SCAR CR 179 025 024 .001 .082 .080 18 Table 29.—Length of stay in days: variable Xj Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0-20 35 6.377 2.134 1.002 13.106 12.105 101 21- 40 206 7.131 2.777 1.010 29.117 28.107 102 41- 60 198 7.263 2.800 3.371 27.245 23.874 103 61- 80 113 7.545 2.623 3.896 19.108 14.212 104 81-100 84 7.953 2.733 4.372 19.082 14.710 105 101-140 115 8.178 3.281 2.807 22.122 19.315 106 141-200 72 8.035 3.472 4.790 31.519 26.729 107 201-300 34 9.365 6.141 5.577 36.069 30.492 Non-Government 201 21- 40 225 7.249 2.411 1.584 21.303 19.719 202 41- 60 219 7.557 2.847 3.203 25.728 22.525 203 61- 80 210 8.311 4.220 4.374 56.706 52.332 204 81-100 195 8.139 2.590 3.842 19.999 16.157 205 101-140 274 7.943 2.393 1.127 26.456 25.329 206 141-200 274 8.030 2.446 4.078 25.210 21.132 207 201-300 226 8.148 1.869 4.771 19.560 14.789 208 301-400 69 8.021 1.370 4.828 11.419 6.591 209 401+ 25 8.177 1.401 6.199 10.832 4.633 Total. eeueeieeuienenseiinesnsnnsensnsananens 2,574 7.802 2.843 1.002 56.706 55.704 IL. Teaching hospitals Government 119 401+ 52 10.833 4.123 7.098 32.756 25.659 Non-Government 216 141-200 24 8.459 2.374 4.877 14.986 10.109 217 201-300 129 8.664 1.386 5.718 11.800 6.082 218 301-400 140 8.846 1.588 5.600 17.419 11.820 219 401+ 196 9.192 1.588 1.507 14.668 13.161 Totalcvevrisvonivinny SEER Ee 541 9.102 2.058 1.507 32.756 31.249 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 6.939 2.185 4.234 15.304 11.070 302 41- 60 51 7.497 2.749 3.866 19.418 15.553 303 61- 80 23 7.008 2.074 4.239 11.819 7.580 304 81-100 27 6.684 1.866 4.001 12.313 8.312 305 101-140 26 7.510 1.835 4.284 10.119 5.835 Total.everrenennnnnns CTC cvrenves 179 7.152 2.262 3.866 19.418 15.553 [4] Table 30.—Occupancy in proportions: variable Xg Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0-20 35 571 206 .286 1.000 714 101 21- 40 258 .617 171 125 1.486 1.361 102 41- 60 235 .689 152 275 1.075 .801 103 61- 80 144 696 135 .297 1.150 853 104 81-100 108 715 116 .362 .976 .614 105 101-140 127 741 133 .187 1.038 .851 106 141-200 85 .768 .102 433 979 .546 107 201-300 34 779 11 .486 .969 .483 Non-Government 201 21- 40 273 637 .168 227 1.391 1.164 202 41- 60 262 687 147 214 1.153 .938 203 61- 80 232 W713 126 .373 1.028 .655 204 81-100 222 .739 110 400 1.150 750 205 101-140 304 754 118 .336 1.138 .802 206 141-200 310 .781 .105 .380 1.162 782 207 201-300 250 .805 .091 270 1.053 .783 208 301-400 83 .810 .081 .610 .992 .382 209 401+ 25 832 057 674 915 .241 Total......... awn ser erBE TREE EE , 2,987 719 145 125 1.486 1.361 II. Teaching hospitals Government 119 401+ 52 72 .081 .506 925 419 Non-Government 216 141-200 24 .849 .100 .588 .985 .397 217 201-300 154 .830 .080 .541 1.000 .459 218 301-400 173 .846 064 .660 1.000 .340 219 401+ 230 .853 .058 .658 1.051 .393 CREE 633 .839 .074 506 1.051 .545 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 646 186 .071 .950 .879 302 41- 60 51 727 .148 .419 1.055 .636 303 61—- 80 23 .701 129 .410 947 .536 304 81-100 27 W740 .149 232 1.022 790 305 101-140 26 W737 141 272 .930 .658 Total. vv vv vervrivvimcvsin s 0s SR RAH IHR SHA RES 179 .703 162 071 1.055 .983 €8 Table 31.—Bassinets per statistical bed : variable Xj; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0-20 35 .276 .073 .100 .400 .300 101 21- 40 257 .218 .076 0.000 .500 .500 102 41- 60 233 182 .059 .069 444 375 103 61- 80 143 159 .057 0.000 .362 .362 104 81-100 108 J42 .057 0.000 .364 364 105 101-140 127 128 .051 0.000 .264 .264 106 141-200 85 126 .058 0.000 422 422 107 201-300 34 118 .050 0.000 227 .227 Non-Government 201 21- 40 272 194 104 0.000 .783 .783 202 41- 60 261 156 .080 0.000 .360 .360 203 61—- 80 229 134 .074 0.000 .533 .533 204 81-100 220 126 .068 0.000 .358 .358 205 101-140 302 J22 .062 0.000 317 317 206 141-200 310 Jd12 .059 0.000 .310 310 207 201-300 250 115 .053 0.000 .523 .523 208 301-400 83 101 .032 0.000 .195 195 209 401+ 25 .098 .033 0.000 157 157 TOMBE ev vacvmacn vwmwmninn was sense ves FEEEEEES EOS 2,974 .148 .078 0.000 .783 .783 II. Teaching hospitals Government 119 401+ 52 .087 .043 0.000 .189 .189 Non-Government 216 141-200 24 135 .070 0.000 292 .292 217 201-300 153 115 .055 0.000 .263 .263 218 301-400 173 105 .050 0.000 .325 .325 219 401+ 229 .097 .041 0.000 228 .228 i EL er. 631 104 .051 0.000 325 325 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 129 .109 0.000 .457 .457 302 41- 60 51 .090 .078 0.000 W222 W222 303 61- 80 23 .083 .082 0.000 .263 .263 304 81-100 27 .081 .063 0.000 .162 162 305 101-140 26 .058 .086 0.000 .343 .343 Total: on crvraiii cn sh Eadie ns sommmmmnmmmnircnms 179 .094 .092 0.000 457 .457 12] Table 32.—Births per statistical bed: variable Xs Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 2,865 1.900 .050 7.667 7.617 101 21- 40 257 3.424 2.410 0.000 17.700 17.700 102 41- 60 233 3.823 2.306 .067 13.596 13.529 103 61- 80 144 3.819 1.905 0.000 9.725 9.725 104 81-100 108 4.282 2.573 0.000 18.160 18.160 105 101-140 127 4.127 2.261 0.000 11.274 11.274 106 141-200 85 4.554 2.365 0.000 13.391 13.391 107 201-300 34 4.923 2,624 0.000 14.038 14.038 Non-Government 201 21- 40 270 3.074 2.602 0.000 13.964 13.964 202 41- 60 261 3.688 2.942 0.000 21.347 21.347 203 61- 80 228 3.731 3.052 0.000 17.230 17.230 204 81-100 221 3.585 2.702 0.000 19.326 19.326 205 101-140 302 4.028 2.830 0.000 18.730 18.730 206 141-200 308 4.201 2.765 0.000 18.121 18.121 207 201-300 249 4.846 2.816 0.000 15.354 15.354 208 301-400 83 4.549 2.527 0.000 20.822 20.822 209 401+ 25 4.496 1.869 0.000 8.850 8.850 TotaYsceviscunnne CR n AVR wen 2,970 3.907 2.675 0.000 21.347 21.347 II. Teaching hospitals Government 119 401+ 52 3.745 2.137 0.000 9.952 9.959 Non-Government 216 141-200 24 4.963 3.725 0.000 14.798 14.798 217 201-300 152 4.514 2.609 0.000 13.991 13.991 218 301-400 173 4.636 2.832 0.000 21.463 21.463 219 401+ 229 4.325 2.313 0.000 18.154 18.154 Total russ vviverrsorsvivsnverrernssenyrrennve 630 4.432 2.592 0.000 21.463 21.463 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 2.617 3.136 0.000 14.286 14.286 302 41- 60 51 2.449 3.259 0.000 15.745 15.745 303 61- 80 23 2.535 3.118 0.000 10.974 10.974 304 81-100 27 3.293 3.298 0.000 10.393 10.393 305 101-140 26 2.334 3.634 0.000 15.306 15.306 Total ivi nisms CVE A re Re SR 179 2.619 3.248 0.000 15.745 15.745 S8 Table 33.—Emergency visits per admission: variable X, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 1.686 1.760 .146 7.188 7.042 101 21- 40 226 1.388 1.060 0.000 6.408 6.408 102 41- 60 201 1.402 912 0.000 6.302 6.302 103 61- 80 128 1.451 .994 .064 6.111 6.047 104 81-100 98 1.431 937 0.000 5.678 5.678 105 101-140 118 1.456 775 232 4.172 3.940 106 141-200 80 1.516 807 0.000 4.638 4.638 107 201-300 34 1.674 .855 0.000 3.629 3.629 Non-Government 201 21- 40 246 1.291 1.093 0.000 5.813 5.813 202 41- 60 238 1.423 1.501 0.000 16.992 16.992 203 61— 80 205 1.371 1.407 0.000 15.507 15.507 204 81-100 183 1.333 1.180 0.000 12.064 12.064 205 101-140 276 1.419 1.006 0.000 6.515 6.515 206 141-200 279 1.522 1.163 0.000 9.518 9.518 207 201-300 224 1.576 962 .008 6.028 6.019 208 301-400 Ki! 1.242 .588 0.000 3.109 3.109 209 401+ 25 1.310 617 410 3.490 3.080 Total. .vee. CRESS SESE wens 2,667 1.426 1.109 0.000 16.992 16.992 II. Teaching hospitals Government 119 401+ 52 2.920 1.740 0.000 7.161 7.161 Non-Government 216 141-200 24 2.097 1.364 255 5.768 5.513 218 301-400 147 2.071 1.105 0.000 6.659 6.659 7 201-300 160 1.767 865 .007 5.338 5.331 219 401+ 222 1.506 .708 103 4.141 4,038 Total...... tererrrarerrertererteees emrreye 605 1.857 1.078 0.000 7.161 7.161 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 1.049 1.154 0.000 6.134 6.134 302 41- 60 51 1.054 1.110 0.000 4.364 4.364 303 61- 80 23 1.027 W722 037 2.835 2,797 304 81-100 27 1.079 1.077 0.000 4.024 4.024 305 101-140 26 .891 696 0.000 2.565 2.565 Totaly irrviverpvn “eens SRR ER 179 1.029 1.019 0.000 6.134 6.134 98 Table 34.—Total visits per admission: variable Xj Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 3,303 2.332 327 9.757 9.430 101 21- 40 251 2,692 5.376 0.000 83.554 83.554 102 41- 60 229 2.366 1.916 0.000 13.577 13.577 103 61- 80 144 2.420 1.500 151 10.811 10.660 104 81-100 106 2.910 2.398 .585 14.961 14.375 105 101-140 127 3.584 6.270 579 68.906 68.327 106 141-200 85 3.419 3.876 784 28.051 27.267 107 201-300 34 3.318 2.298 .360 9.251 8.891 Non-Government 201 21- 40 270 3.765 11.579 0.000 137.438 137.438 202 41- 60 257 3.857 9.560 0.000 144.623 144.623 203 61— 80 228 3.323 4.114 0.000 50.953 50.953 204 81-100 216 3.316 4.482 0.000 60.594 60.594 205 101-140 302 3.214 2.349 0.000 14.679 14.679 206 141-200 307 4.010 5.502 0.000 52.126 52.126 207 201-300 249 3.938 3.951 J24 42.334 42.210 208 301-400 83 2.976 2.700 .607 20.185 19.578 209 401+ 25 2.554 1.680 .621 5.281 5.902 FOUR] wvvemnvnriinn wmimmimsimace win srbinimiereinm wine e sins 2,948 3.332 5.811 0.000 144.623 144.623 II. Teaching hospitals Government 119 401+ 52 9.746 4.998 2.058 20.482 18.424 Non-Government 216 141-200 24 5.893 4.119 720 15.807 15.087 217 201-300 152 6.418 8.474 0.000 76.450 76.450 218 301-400 171 4.930 3.260 .392 19.358 18.966 219 401+ 229 5.383 4.428 .502 45.207 44.705 Totals vis vsvvsin Gers ervenonsnevess wa ve E Ee 628 5.891 5.623 0.000 76.450 76.450 III. For-profit, nonteaching hospitals For-profit 301 21- 40 S52 8.677 22.444 0.000 150.418 150.418 302 41- 60 51 3.699 6.539 0.000 36.619 36.619 303 61- 80 23 3.672 5.261 .278 21.490 21.212 304 81-100 27 2.073 2.011 0.000 7.976 7.976 305 101-140 26 2.031 2.393 .029 11.839 11.810 Total vu sansr ses ras tvsnsisssovinsesssnse sonny 179 4.654 12.974 0.000 150.418 150.418 L8 Table 35.—Surgical operations per admission: variable X;; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 .165 126 0.000 .629 .629 101 21- 40 251 .188 118 0.000 1.110 1.110 102 41- 60 228 224 J21 0.000 .660 660 103 61- 80 142 .282 .241 0.000 2.704 2.704 104 81-100 105 .304 .140 0.000 1.045 1.045 105 101-140 127 .331 120 102 719 617 106 141-200 82 .400 130 .070 .780 .710 107 201-300 34 444 J27 .001 .639 .638 Non-Government 201 21- 40 265 .238 .313 0.000 4.880 4.880 202 41- 60 260 .316 172 0.000 1.216 1.216 203 61- 80 226 .329 159 0.000 917 917 204 81-100 215 .411 .383 .055 5.506 5.451 205 101-140 296 .426 .146 .036 1.007 971 206 141-200 301 .471 144 .105 1.209 1.103 207 201-300 245 .506 121 .251 1.009 .758 208 301-400 81 .503 115 .296 .871 575 209 401+ 25 .501 .096 .301 7 .416 Total......... sess eevee eserreans FR 2,918 349 .220 0.000 5.506 5.506 Il. Teaching hospitals Government 119 401+ 52 .429 124 145 .695 550 Non-Government 216 141-200 24 .507 128 .309 .857 .549 217 201-300 152 .526 J21 0.000 .988 .988 218 301-400 170 .545 130 .345 1.181 .835 219 401+ 229 .559 123 .282 1.399 1.117 Total. euueneenenienneerirensnseeseisanennnnns 627 534 131 0.000 1.399 1.399 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 229 171 0.000 .591 591 302 41- 60 51 .355 .186 0.000 726 726 303 61- 80 23 .384 154 114 631 517 304 81-100 27 463 146 .104 .707 .603 305 101-140 26 522 .166 197 .908 711 Total....ocmvrnninn essen Ee 179 .363 .198 0.000 .908 .908 88 Table 36.—Statistical beds: variable X12 Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 17.286 3.277 8.000 20.000 12.000 101 21- 40 258 31.027 5.420 21.000 40.000 19.000 102 41- 60 235 49.974 5.487 41.000 60.000 19.000 103 61- 80 144 70.201 6.045 61.000 80.000 19.000 104 81-100 108 90.343 6.213 81.000 100.000 19.000 105 101-140 127 117.654 11.297 101.000 140.000 39.000 106 141-200 85 165.000 17.747 141.000 200.000 59.000 107 201-300 34 250.029 28.544 201.000 298.000 97.000 Non-Government 201 21- 40 273 30.996 5.373 21.000 40.000 19.000 202 41- 60 262 50.359 5.931 41.000 60.000 19.000 203 61- 80 232 70.927 5.806 61.000 80.000 19.000 204 81-100 222 91.090 6.306 81.000 100.000 19.000 205 101-140 304 120.556 11.578 101.000 140.000 39.000 206 141-200 311 169.019 17.836 141.000 200.000 59.000 207 201-300 250 241.856 29.542 201.000 300.000 99.000 208 301-400 83 344.711 29.820 304.000 400.000 96.000 209 401+ 25 452.920 44.933 401.000 535.000 134.000 Totalesusresnsneseererenverercerennersncenes 2,988 108.980 83.056 8.000 535.000 527.000 II. Teaching hospitals Government 119 401+ 52 616.077 196.653 408.000 1326.000 918.000 Non-Government 216 141-200 24 181.792 12.901 150.000 200.000 50.000 217 201-300 154 252.844 28.359 201.000 300.000 99.000 218 301-400 173 348.341 28.665 301.000 400.000 99.000 219 401+ 230 568.043 185.954 402.000 1564.000 1162.000 Totalusvsnincovinsrmovsvsves ve 633 420.616 193,862 150.000 1564.000 1414.000 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 30.038 5.409 21.000 40.000 19.000 302 41- 60 51 48.804 5.370 41.000 60.000 19.000 303 61— 80 23 72.304 5.269 61.000 80.000 19.000 304 81-100 27 93.630 5.732 81.000 100.000 19.000 305 101-140 26 122.538 11.389 101.000 139.000 38.000 Total..uvvvuruvnnnsnss enrens tesrrerrerrenees 179 63.844 33.011 21.000 139.000 118.000 63 Table 37.—Adjusted patient days in 100,000: variable X;; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 .041 .016 017 .076 .059 101 21- 40 256 077 L041 012 .583 W571 102 41- 60 234 134 .034 054 .245 190 103 61- 80 143 191 .045 .104 .367 263 104 81-100 106 .256 .045 139 369 229 105 101-140 127 343 .070 119 516 397 106 141-200 85 497 .088 .251 .740 .489 107 201-300 34 770 144 407 .938 W531 Non-Government 201 21- 40 2n .079 .026 .020 207 187 202 41- 60 262 .138 .036 W055 254 .199 203 61- 80 231 .201 044 .095 338 .243 204 81-100 220 .266 049 .148 .503 .355 205 101-140 302 .358 .070 Jd12 .580 .468 206 141-200 308 .521 .095 W252 795 .542 207 201-300 247 775 .138 .243 1.366 1.123 208 301-400 82 1.094 152 .780 1.462 682 209 401+ 25 1.460 172 1.223 1.867 643 Total. csvsinnesinn testtettstetenarranne cerens 2,968 .326 277 012 1.867 1.855 II. Teaching hospitals Government 119 401+ 52 2.010 716 903 4.638 3.736 Non-Government 216 141-200 24 643 119 421 1.017 .596 217 201-300 153 .845 142 457 1.182 725 218 301-400 171 1.179 146 .836 1.783 947 219 401+ 230 1.929 659 1.150 5.477 4.327 Total. visvenmevnesnnnovsns seetisurenan renee 630 1.420 674 421 5477 5.056 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 .080 .030 .011 173 162 302 41- 60 51 138 .038 .059 .220 161 303 61- 80 23 196 W052 078 .267 189 304 81-100 27 267 .063 .069 374 .305 305 101-140 26 .340 074 126 .469 34 Total cvs conenninninsnine trsresvenes rrrrreans | 179 178 120 011 469 458 06 Table 38.—SMSA (coded 1) versus non-SMSA (coded 0) : variable X,, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 114 .323 0.000 1.000 1.000 101 21- 40 258 .058 .234 0.000 1.000 1.000 102 41- 60 102 077 .267 0.000 1.000 1.000 103 61- 80 144 .097 297 0.000 1.000 1.000 104 81-100 108 176 .383 0.000 1.000 1.000 105 101-140 127 .228 421 0.000 1.000 1.000 106 141-200 85 .353 .481 0.000 1.000 1.000 107 201-300 34 441 .504 0.000 1.000 1.000 Non-Government 201 21- 40 273 114 .318 0.000 1.000 1.000 202 41- 60 262 .240 .428 0.000 1.000 1.000 203 61- 80 232 .293 456 0.000 1.000 1.000 204 81-100 222 .297 .458 0.000 1.000 1.000 205 101-140 304 .431 .496 0.000 1.000 1.000 206 141-200 311 .508 .501 0.000 1.000 1.000 207 201-300 250 724 .448 0.000 1.000 1.000 208 301-400 83 735 444 0.000 1.000 1.000 209 401+ 25 .800 .408 0.000 1.000 1.000 Total. .uususesescsrsenvncnrsvens ens 2,988 .309 462 0.000 1.000 1.000 II. Teaching hospitals Government 119 401+ 52 942 .235 0.000 1.000 1.000 Non-Government 216 141-200 24 958 .204 0.000 1.000 1.000 217 201-300 154 877 .330 0.000 1.000 1.000 218 301-400 173 919 274 0.000 1.000 1.000 219 401+ 230 974 160 0.000 1.000 1.000 Totals vivir sr RAAT A 633 .932 .253 0.000 1.000 1.000 III. For-profit, nonteaching hospitals For-profit 301 21- 40 52 .500 .505 0.000 1.000 1.000 302 41- 60 51 .510 .505 0.000 1.000 1.000 303 61- 80 23 696 .470 0.000 1.000 1.000 304 81-100 27 .852 .362 0.000 1.000 1.000 305 101-140 26 .731 .452 0.000 1.000 1.000 Totalecrvvsswsnmmsmvnes Cosrersrnirnen tevennes 179 615 .490 0.000 1.000 1.000 16 Table 39.—Total nursing hours per adjusted patient day: variable Y, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0-20 26 7.962 2.405 4.620 14.905 10.286 101 21- 40 234 6.943 1.966 .530 14.756 14.225 102 41- 60 229 6.731 1.629 1.240 14.267 13.027 103 61- 80 140 6.417 1.574 2.679 11.255 8.576 104 81-100 107 6.435 1.557 1.640 9.720 8.081 105 101-140 122 6.320 1.629 1.591 10.076 8.484 106 141-200 85 6.608 1.350 3.125 10.554 7.429 107 201-300 32 6.769 1.297 3.436 11.030 7.594 Non-Government 201 21- 40 238 6.271 1.778 1.581 12.775 11.195 202 41- 60 245 6.575 1.664 1.662 11.769 10.107 203 61- 80 227 6.267 1.633 2.032 12.894 10.862 204 81-100 218 6.428 1.632 1.874 12.245 10.371 205 101-140 303 6.611 1.463 1.595 11.762 10.167 206 141-200 307 6.720 1.380 1.953 11.118 9.164 207 201-300 244 6.549 1.236 2.388 10.964 8.576 208 301-400 79 6.254 1.178 1.556 8.825 7.270 209 401+ 25 6.180 W730 4.959 7.765 2.806 Totals svenwnvpemnanenvseisvmnvanarve Fn 2,861 6.553 1.595 .530 14.905 14.375 II. Teaching hospitals Government 119 401+ S52 6.264 1.830 3.549 15.032 11.483 Non-Government 216 141-200 23 6.303 1.387 3.586 8.366 4.780 217 201-300 150 6.438 1.121 3.089 10.325 7.236 218 301-400 168 6.111 1.012 3.438 9.466 6.028 219 401+ 229 6.173 1.105 3.698 10.770 7.072 Total.eevvnvinninnnnnnn nba 622 6.232 1.178 3.089 15.032 11.943 III. For-profit, nonteaching hospitals For-profit 301 21- 40 45 6.348 2.220 1.444 15.248 13.805 302 41- 60 48 6.553 2.195 2.386 15.178 12.792 303 61- 80 23 7.178 3.721 4.152 19.329 15.178 304 81-100 27 7.594 2.267 4.611 16.723 12.112 305 101-140 25 6.958 1.427 4.565 10.494 5.929 Total ouicrnnnirmnnine sn even ssr snes ve 168 6.811 2.403 1.444 19.329 17.886 a6 Table 40.—RN hours per adjusted patient day: variable Y, Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 2.699 1.280 718 6.211 5.493 101 21- 40 256 2.028 1.013 J77 6.076 5.899 102 41- 60 235 1.808 1926 424 7.334 6.910 103 61- 80 143 1.881 1.286 .193 12.973 12.780 104 81-100 108 1.964 969 129 4.928 4.799 105 101-140 126 2.135 1.208 354 7.197 6.843 106 141-200 85 2.353 1.016 .728 5.803 5.075 107 201-300 33 2.428 749 .832 4.227 3.395 Non-Government 201 21- 40 267 2.223 963 .073 5.760 5.687 202 41- 60 257 2.331 1.106 .108 6.766 6.658 203 61- 80 230 2.351 1.150 441 6.868 6.427 204 81-100 220 2.450 1.153 .187 6.425 6.238 205 101-140 304 2.659 1.055 .499 5.982 5.483 206 141-200 309 2.886 1.002 387 5.851 5.464 207 201-300 249 2.917 .891 279 5.316 5.028 208 301-400 82 2.608 743 847 5.007 4.160 209 401+ 25 2.550 .901 W951 4.946 3.996 Totahisvsersssrisssvsrennrisvursnnsossnnoners 2,964 2.377 1.097 073 12.973 12.900 II. Teaching hospitals Government 119 401+ 52 2.439 .885 .820 4.572 3.752 Non-Government 216 141-200 23 2.831 1.060 1.032 5.221 4.189 217 201-300 152 3.153 939 1.094 5.776 4.682 218 301-400 172 2.967 .826 443 5.716 5.273 219 401+ 230 2.935 794 1.431 6.126 4.695 Totalessssrnssrsrrrevrnsurersmressnennnmsss 629 2.952 875 443 6.126 5.683 III. For-profit, nonteaching hospitals For-profit 301 21- 40 51 1.927 1.269 456 7.625 7.170 302 41- 60 51 2.200 1.345 .220 7.112 6.891 303 61- 80 23 2.205 1.579 .591 6.180 5.589 304 81-100 27 3.121 1.439 626 7.985 7.359 305 101-140 26 2.641 1.169 .809 4.959 4.150 Totalevesesssssensesrsenssrsvasnvernnonnses . 178 2.327 1.393 .220 7.985 7.765 £6 Table 41.—LPN hours per adjusted patient day: variable Y; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 26 1.863 1.295 0.000 6.775 6.775 101 21- 40 239 1.377 .988 0.000 4.871 4.871 102 41- 60 230 1.452 964 0.000 5.188 5.188 103 61- 80 141 1.464 976 0.000 5.015 5.015 104 81-100 107 1.500 1.010 0.000 5.052 5.052 105 101-140 122 1.486 1.050 .064 5.198 5.133 106 141-200 85 1.507 923 131 4.894 4.763 107 201-300 33 1.492 812 .260 3.241 2.981 Non-Government ’ 201 21- 40 244 1.061 .821 0.000 4.543 4.543 202 41- 60 250 1.281 929 0.000 5.591 5.591 203 61- 80 229 1.205 .803 0.000 4.499 4.499 204 81-100 220 1.396 936 0.000 4.942 4.942 205 101-140 304 1.437 864 0.000 7.016 7.016 206 141-200 310 1.358 826 0.000 4.632 4.632 207 201-300 246 1.289 W720 .027 3.754 3.727 208 301-400 80 1.256 674 .108 3.977 3.869 209 401+ 25 1.237 .761 012 3.305 3.293 Total ven verse Vr ES RARE ARR TERS 2,891 1.351 899 0.000 7.016 7.016 II. Teaching hospitals Government 119 401+ 52 1.498 W735 0.000 4.008 4.008 Non-Government 216 141-200 23 1.350 643 279 2.818 2.540 217 201-300 150 1.154 600 .056 3.085 3.028 218 301-400 170 1.060 603 .210 3.977 3.767 219 401+ 229 1.209 .596 .207 3.646 3.440 Totalesueensnernrennenrnrnnnne J 624 1.184 624 0.000 4.008 4.008 III. For-profit, nonteaching hospitals For-profit 301 21- 40 47 1.566 1.123 0.000 4.482 4.482 302 41- 60 48 1.392 645 0.000 2.959 2.959 303 61- 80 23 1.479 677 .468 2.762 2.294 304 81-100 27 1.297 745 0.000 2.833 2.833 305 101-140 25 1.782 1.156 .164 5.516 5.352 Totaliuvwe vnsunnensen Ces vERL SETS SEE . 170 1.494 906 0.000 5.516 5.516 v6 Table 42.—Aides, orderlies, and/or attendants per adjusted patient day Y, Stratum Control Bedsize Number of Arithmetic Standard Minimum Maximum code class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals overnment 100 0-20 33 3.438 1.741 0.000 6.621 6.621 101 21- 40 253 3.669 1.604 0.000 9.140 9.140 102 41- 60 233 3.485 1.365 0.000 7.282 7.282 103 61- 80 143 3.193 1.272 0.000 6.791 6.791 104 81-100 108 2.974 1.147 .597 5.217 4.620 105 101-140 126 2.721 1.166 179 5.138 4.959 106 141-200 85 2.748 1.000 607 6.059 5.451 107 201-300 33 2.858 J22 1.673 4.902 3.229 Non-Government 201 21- 40 266 3.133 1.425 0.000 11.417 11.417 202 41- 60 259 3.004 1.274 0.000 7.810 7.810 203 61- 80 232 2.719 1.155 0.000 7.053 7.053 204 81-100 220 2.596 1.130 0.000 7.846 7.846 205 101-140 303 2.508 1.051 275 7.084 6.810 206 141-200 309 2.478 1.009 0.000 5.872 5.872 207 201-300 246 2.363 979 0.000 5.541 5.541 208 301-400 81 2.373 918 .428 5.075 4.647 209 401+ 25 2.394 869 1926 4.073 3.147 Totalvevsivsvvvven SSEAEEY TTL VA 2,955 2.868 1.279 0.000 11.417 11.417 II. Teaching hospitals Government 119 401+ 52 2.327 1.009 817 7.120 6.302 Non-Government 216 141-200 24 2.144 799 .636 3.580 2.943 217 201-300 154 2.129 .796 .148 4.129 3.981 218 301-400 172 2.057 707 .516 4.168 3.652 219 401+ 230 2.037 719 .591 4.331 3.740 Totalivicriniivrrcrrnrrrmsnsnssnensrmmonvney 632 2.093 .768 .148 7.120 6.972 III. For-profit, nonteaching hospitals For-profit 301 21- 40 50 2.987 1.487 0.000 6.787 6.787 302 41- 60 51 3.023 1.522 0.000 6.381 6.381 303 61- 80 23 3.493 2.258 1.050 10.783 9.732 304 81-100 27 3.176 1.503 1.333 6.864 5.531 305 101-140 25 2.621 .970 .576 4.351 3.775 Totalev suv vrvvnvervve Cessrersraresss ar 176 3.040 1.564 0.000 10.783 10.783 S6 Table 43.—General duty nurses hours per adjusted patient day: variable Y; Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 1.787 1.491 0.000 6.216 6.216 101 21- 40 252 974 1.044 0.000 6.082 6.082 102 41- 60 234 .865 .865 0.000 5.735 5.735 103 61- 80 142 .874 .762 0.000 3.159 3.159 104 81-100 106 1.044 965 0.000 3.848 3.848 105 101-110 125 1.292 1.141 0.000 6.120 6.120 106 141-200 85 1.521 .962 0.000 4.549 4.549 107 201-300 33 1.747 71 .068 3.329 3.260 Non-Government 201 21-40 264 1.204 1.020 0.000 4.833 4.833 202 41- 60 256 1.236 965 0.000 5.238 5.238 203 61- 80 228 1.324 1.048 0.000 5.724 5.724 204 81-100 218 1.490 1.033 0.000 5.185 5.185 205 101-140 301 1.750 L967 0.000 4.659 4.659 206 141-200 305 2.051 914 0.000 4.958 4.958 207 201-300 246 2.188 .840 0.000 4.464 4.464 208 301-400 81 1.988 747 .180 4.395 4.215 209 401+ 25 1.961 918 .193 4.242 4.049 2,936 1.449 1.056 0.000 6.216 6.216 II. Teaching hospitals Government 119 401+ 50 1.595 .748 .281 3.907 3.626 Non-Government 216 141-200 23 1.971 910 .453 4.145 3.692 217 201-300 151 2.356 916 0.000 4.817 4.817 218 301-400 169 2.305 696 .448 4.368 3.920 219 401+ 230 2.257 737 .562 5.323 4.760 2 623 2.230 .805 0.000 5.323 5.323 III. For-profit, nonteaching hospitals For-profit 301 21- 40 51 .598 710 0.000 3.117 3.117 302 41- 60 51 965 .986 0.000 4.249 4.249 303 61— 80 23 1.085 1.235 0.000 3.532 3.532 304 81-100 27 1.976 1.198 0.000 4.218 4.218 305 101-140 26 1.624 1.171 0.000 4.225 4.225 Total. vunvsunsnnonsiwsvaransa cas cravnss snes 178 1.125 1.118 0.000 4.249 4.249 96 Table 44.—General duty and head nurses hours per adjusted patient day: variable Ys Stratum Bedsize Number of Arithmetic Standard Minimum Maximum code Control class hospitals mean deviation value value Range I. Nonprofit, nonteaching hospitals Government 100 0- 20 35 2.052 1.420 0.000 6.216 6.216 101 21- 40 252 1.419 1.027 0.000 6.082 6.082 102 41- 60 234 1.202 .925 0.000 6.802 6.802 103 61- 80 142 1.251 .864 0.000 3.664 3.664 104 81-100 106 1.485 .958 0.000 4.357 4.357 105 101-140 125 1.693 1.190 0.000 6.878 6.878 106 141-200 85 1.964 975 .338 5.347 5.008 107 201-300 33 2.144 743 665 3.910 3.245 Non-Government 201 21- 40 264 1.652 975 0.000 5.156 5.156 202 : 41- 60 256 1.738 1.024 0.000 5.871 5.871 203 61— 80 228 1.773 1.067 0.000 6.141 6.141 204 81-100 218 1.956 1.089 0.000 5.988 5.988 205 101-140 301 2.196 1.026 0.000 5.207 5.207 206 141-200 305 2.506 .949 0.000 5.325 5.325 207 201-300 246 2.594 851 .268 5.064 4.795 208 301-400 81 2.328 756 .380 4.740 4.359 209 401+ 25 2.325 857 774 4.565 3.791 Total veins a sesres sesrerserenereeres 2,936 1.879 1.081 0.000 6.878 6.878 II. Teaching hospitals Government 119 401+ 50 2.115 .831 .585 4.393 3.808 Non-Government 216 141-200 23 2.476 1.020 779 4.863 4.085 217 201-300 151 2.783 .932 .500 5.376 4.876 218 301-400 169 2.710 755 .896 5.348 4.452 219 401+ 230 2.659 .780 1.204 5.850 4.646 Total.veeensrnseesersesssrrnnssrsenssnrsnnans 623 2.652 843 .500 5.850 5.350 III. For-profit, nonteaching hospitals For-profit 301 21- 40 51 1.194 .836 0.000 3.506 3.506 302 41- 60 51 1.524 1.180 0.000 5.414 5.414 303 61— 80 23 1.527 1.299 0.000 4.340 4.340 304 81-100 27 2.532 1.307 0.000 6.179 6.179 305 101-140 26 2.176 1.149 .502 4.496 3.994 Total..uveuversrnsnnsnnn tersseseraneeens serene 178 1.678 1.212 0.0000 6.179 6.179 L6 Table 45.—Simple correlations among independent variables, X; to X;4, stratum I Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1.00 .87 1.00 75 .54 1.00 -.02 -.06 -.09 1.00 32 24 45 J2 1.00 -.26 -.20 -.24 -.30 -.15 1.00 19 17 35 —.36 27 43 1.00 .03 .03 05 -.01 .05 07 13 1.00 .05 .01 .03 04 -.01 -.06 05 44 1.00 43 .36 42 =.01 25 -.22 15 02 02 1.00 75 .53 93 a1 .36 -.36 15 .03 04 40 1.00 74 52 94 a1 .48 -.33 .20 .07 .06 41 98 1.00 .35 25 43 04 22 -.25 J2 .08 10 .34 41 43 1.00 Table 46.—Simple correlations among independent variables, X; to Xj4, stratum II Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1.00 .68 1.00 49 17 1.00 8 .08 -.10 -o1l 1.00 -.02 -.02 .16 -.01 1.00 -.21 -.14 .00 -.36 16 1.00 -=.11 —.08 .18 —.43 23 .78 1.00 -.14 -.22 -.20 21 —.04 06 .00 1.00 04 =.11 -.03 a3 -.09 .05 .10 .37 1.00 .04 .04 13 -.14 21 -.05 -.03 —-.18 -.09 1.00 S52 Al .83 .30 -.01 -.23 -.13 —.06 .06 .03 1.00 51 .09 .84 .30 14 -.18 -.08 —=.01 .09 05 .98 1.00 07 .05 JA12 .03 .10 -=.01 .08 .01 .04 .05 11 13 1.00 66 Table 47.—Simple correlations among independent variables, X; to X;4, stratum III Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1.00 .78 1.00 50 .33 1.00 .00 .07 -.29 1.00 14 .03 .39 10 1.00 -=.19 -.12 -=.00 -.42 —.06 1.00 .03 04 34 -.47 J2 75 1.00 07 .00 .06 —.18 14 .10 .16 1.00 -.07 -.09 =.15 —.03 -.07 .04 -.03 .09 1.00 44 .30 .48 -.12 15 -.30 .00 .06 -17 1.00 S57 .41 .82 .04 14 -.30 -.02 =.05 -.19 .52 1.00 .56 .37 .88 .06 .47 -.20 .08 .05 -.12 .46 .89 1.00 24 a2 .23 -.09 -.06 -.28 —.06 12 -.09 .56 27 22 1.00 001 Table 48.—Ranked beta coefficients ! Independent variable Stratum I Nonprofit, nonteaching Stratum II Teaching, not-for-profit Stratum III For-profit Ys Ya Ys Y1 Ys Ys Yeo Yz Ys Ys Ys 1 Total number of facilities....... 2 Number of high tech. facilities. . 3 Fraction high tech. to total...... [al 5 I || ew 4 Total admissions ............... 5 Length of stay .. 6 Occupancy - | — = | I ~~ [ew 7 Bassinets per statistical bed..... 8 Births per statistical bed........ 9 Emergency visits per admission. . [le 3 [1 | =I I 10 Total visits per admission....... 11 Surgical operationss per adm. .. 12 Statistical beds ................. 0 [1 wl | wl | wl | ~ 13 Adjusted patient days ........... 14 8. M.S. A: suruinisorimnseaiats 15 Percent of population under 18.. I — (=n (=)4 - — (-N (=)4 |= 16. Percent of pop. 65 and over.... 17 RN’s per 100,000 population..... 18 LPN's per 100,000 population.... ~~ SEER] —— ae )5 [11 [ 19 Aides, etc. per 100,000 pop. .... 20 Per capita income .......... ie 21 Percent of families under poverty level I ® | eo ole (-)8 [1a * To assign ranks, decimal places beyond those shown in tables 4 through 21, was taken into account. NOTE: Y,—Total nursing hours per adjusted patient day Y>—RN hours per adjusted patient day Y:—LPN hours per adjusted patient day Y.—Aides, orderlies, etc., hours per adjusted patient day Y:s—General duty nurses hours per adjusted patient day Y+—General duty and head nurses hours per adjusted patient day 11. 12. 13. 14. 15. 16. REFERENCES Altman, S. H., Present and Future Supply of Registered Nurses (DHEW Publication No. (NIH) 72-134). U.S. Dept. of Health, Education, and Welfare, Government Printing Office, Washington, D.C., November 1971. American Hospital Association, Hospitals, Guide Issue, J.A.H.A. Chicago, Ill., August 1, 1971. American Hospital Association, The 1970 Survey of Nursing Personnel in Hospitals (mag. tape), November 1970. American Hospital Association, Hospital Statistics, 1971, Chicago, Ill, 1972. Becker, G. S., Human Capital, Columbia University Press, New York, 1964. Ferber, B., “An Analysis of Chain-Operated For-Profit Hospitals,” Health Services Research, Vol. 6, No. 1, Spring, 1971. Flagle, C. D., “Factors Related to Nurse Staffing and Problems of Their Measurement,” in Research on Nurse Staffing in Hospitals, Report of the Conference, May 1972, (DHEW Publication No. (NIH) 73-434), Government Printing Office, Washington, D.C. Hale, T., Jr., “The Five Sides of the Nursing Problem,” The Modern Hospital, Vol. 89, No. 1, July 1957. Isard, W., Location and Space-Economy, John Wiley, New York, 1956. Jeffers, J. R., M. F. Bognanno, and J. C. Bartlett, “On Demand Versus Need for Medical Services and Concepts of ‘Shortage’,” American Journal of Public Health, Vol. 61, January 1971. Kendall, M. G. and Stuart, A., The Advanced Theory of Statistics, Vol. 2, Hafner Publishing Co., New York, Second Edition, p. 414, 1967. Levine, E., S. Siegel, and J. de la Puente, “Diversity of Nurse Staffing Among General Hospitals,” Hospitals, J.A.H.A., Vol. 35, May 1, 1961. Losch, A., The Economics of Location (Translated by W. H. Woglom and W. F. Stopler), Yale University Press, New Haven, 1954. Shain, M. and M. I. Roemer, “Hospital Costs Relate to Supply of Beds,” The Modern Hospital, Vol. 92, No. 4, April 1959. Smith, P. M., Influence of Wage Rates on Nurse Mobility, Graduate Program in Hospital Administration, University of Chicago, Chicago, 1962. U. S. Bureau of the Census, Census of Population: 1970, General Popu- lation Characteristics, Final Report PC(1)-B, Table 16. 101 17. 18. 19. 20. 2]. 22. U.S. Bureau of the Census, Census of Population: 1970, General Social and Economic Characteristics, Final Report PC(1)-C, Table 124. U. S. Dept. of Agriculture, Economic Research Service “Urban Versus Rural Contrasts in Distribution of Health Workers,” Hospitals, J.A.H.A., Vol. 41, June 1, 1967. U. S. Dept. of Health, Education, and Welfare, Health Resources Sta- tistics, Government Printing Office, Washington, D.C., 1972. Weber, A., Alfred Weber's Theory of the Location of Industries (Trans- lated by C. J. Friedrich), Chicago University Press, 1929. Yett, D. E., “Causes and Consequences of Salary Differentials in Nurs- ing,” Inquiry, Vol. VIII, No. 1, March 1970. Koopmans, T. C., “Measurement Without Theory,” Review of Economic Statistics, Vol. 29, August 1947. 102 TECHNICAL APPENDIX 103 I. Working Data Files It has been convenient to assemble two data files which have served as our basic data. Each record on both files contains three fields of identifying data: AHA hospital identification number Stratum code (as detailed in chapter 4) An AHA State and county code These identifying fields are followed by 27 fields of data. The order is identical to that in chapter 4. One of our files contains records for all 5,453 hospitals. The other con- tains records for the 3,800 hospitals used to construct our models. Where data are missing we have used a value of 999999.999999 as an indicator of nonavailability of data. The 3,800 hospitals were selected from the larger group as detailed in chapter 4. The smaller file contains complete data for the 21 independent variables, although, in some cases, the data for the dependent variables may be lacking. Both files have been used exclusively with Fortran programs. In this context the unformatted or binary form was clearly more efficient than a formatted file, in terms of computer time. Also, records were blocked to give maximum efficiency. Without these two features the computer time would have been increased by a factor of three or more. II. Some Notes on the Regression Computations It seems appropriate to indicate in some detail the least squares, or multiple regression computations used to obtain our models. The procedures fall into five principal sections which were incorporated into Fortran sub- routines for convenience. The first subroutine accumulated the sums, the sums of squares, and sums of cross products for all variables in a data set. From these data means, variances and a complete matrix of simple correlations were com- puted. The first subroutine calls a custom subroutine for data input. This was of great convenience to us in that it permitted working with one file instead of 18 (3 strata X 6 dependent variables). The data input subroutine edited data as required and selected data according to any criteria we choose to program into it. The second subroutine selected from the means, variances, and correla- 105 tions, those required to build any given model. This provided input to the remaining subroutines. The next step was the inversion of a matrix of correlations. This was done using a Cholesky or square root decomposition algorithm. The inner products were computed in double precision arithmetic. The fourth subroutine did most of the remaining regression computations. These included the ordinary and standardized regression coefficients, t-values, analysis of variance, multiple correlations, etc. If desired a fifth subroutine computed “predicted” values. More precisely, these might be called estimates of means or expected values. Also this sub- routine computed residuals and standard errors for the mean values. When desired, a crude distribution of residuals was computed. ITI. Some Numerical Considerations Least squares computations are notorious for their instability, and hence we have taken great care to optimize our computations. Firstly, we examined the ranges and maximum values found in our data. The ranges and the maximum values did not exceed 2 X 10°. We had a very distinct advantage in that the computing system we used, the CDC 6400, has a Fortran compiler that uses approximately 14 digits of accuracy for single precision floating arithmetic. It would seem quite clear that there was no problem in the representation of numbers and in the accumulations of sums of squares and cross products. It is likely that use of an IBM series 360 or 370 machine would have necessitated double precision arithmetic throughout with very substantial increase in machine time. Next, the input to the matrix inversion subroutine was a correlation matrix. In general, the condition of such a matrix is very close to optimum, since all elements are scaled to the range —1.0 to +1.0. The matrix in- version using the Cholesky decomposition algorithm and double precision for inner products is an optimum procedure. In evaluating the condition of the matrixes to be inverted, we found it convenient to use the ratio of the largest eigenvalue to the smallest eigen- value. For the initial models we found condition numbers up to 560.0. For the final models they all were below 200.0 and some were very small indeed. Given that we had approximately 14 digits of accuracy, this seems quite reasonable. There was some “poor conditioning,” and to some degree, it was asso- ciated with three variables: X4: Total admissions Xj: Statistical beds X13: Adjusted patient days As one might expect, all these variables are strongly correlated with size of a hospital (see tables 45-47). In many cases, when one or two of these 106 variables were dropped, the condition of the final models was distinctly improved over that for the initial models. IV. Deletion of Independent Variables In all cases the initial models were too large. That is, they contained several variables which did not significantly contribute to the prediction. As is well known, inclusion of such variables may indeed make the estimates of expected values or mean values less reliable. That is, inclusion of such variables may increase the variances of estimated values. Hence, we were at great pains to eliminate the superfluous independent variables. This is the process that led from the initial models to the final models of chapter 3. The variables to be omitted were chosen for each model, one at a time, in a so-called step-down manner. This procedure does not guarantee an absolute optimum final choice. However, it is very close to optimum and this is seen in the fact that the multiple correlations and the analyses of variance are not significantly altered. This is formalized in the next section of this appendix. As is well known, the variable (at any given stage) with the smallest t-value is the one to drop. This is true because the variable with the smallest t-value will affect the analysis of variance and multiple correlations the least. The variables were dropped, one at a time, for each model individually. V. Stopping Rule for Elimination of Independent Variables At any stage in the deletion of independent variables, the current model can be compared with the initial model. If the fit of the model to the data is not significantly poorer in the current model than in the initial model, one can conclude that the variables which have been dropped are in- deed superfluous. Or put another way, if the analysis of variance and multiple correlations have not been significantly altered, then one may con- clude that the variables which have been omitted do not add, in a significant way, to the prediction of the dependent variable. In such cases we do want to omit such variables. In the contrary case, where the fit of the model is significantly worse, we do not want to drop some variables. This is formalized in the usual F-test. Considering the analysis of variance for any given model with, say, p independent variables and n cases, we denote the sum of squares due to error by Q. This is the sum of squares of the residuals. Omission of variables from the model cannot decrease the value of Q; in that situation Q can only remain unchanged or increase. Let us denote by Q; and p; the sum of squares and number of variables in the initial model, and by Q» and ps the corresponding quantities in a reduced model. Then (Q2 — Q1)/(P2 — Py) io Qz: — Q X n—p — 1 Qi/(n — pr — 1) P1 — Pp2 Q: 107 r= Given the standard assumption on the residuals [i.e., ND (0,02), in- dependent, and that the model is in some sense valid], we know the distribu- tion of F, and may compare it to the standard F-table. Since for this section of the work we computed a sizeable number of F-statistics, we chose to adopt a level of significance of .01. That is, if our F-value was smaller than the .01 tabular value, we would continue the process of deleting variables. When our computed value approximately equaled or exceeded the .01 tabular value, we would retreat one step. That is, we would put back the last variable dropped, and that yielded the final model. Since this is almost always a monotonic process in F and the degrees of freedom, this worked out very well. In a few cases this step-down procedure yielded a model which was still rather large and cumbersome. In such cases the step-down procedure was carried somewhat further to yield a final model. In all cases we dropped variables until the model contained at most 10 variables. VI. Dropping of Outliers There were 17 of the original strata which had a fairly large number of hospitals in each. For each such stratum we constructed preliminary models using the first 15 variables on our list and the total nursing hours per patient day (Y;). We found some 27 hospitals where these preliminary models fit poorly. That is, the residuals exceeded 4 times their standard errors. Such hospitals were judged to be atypical, and we believed they would do more harm than good. They were not used in any further com- putations. Nor are they used in the descriptive statistics. VII. Combination of Strata At one point there were 27 strata. This was judged to be entirely too large a number, and it was necessary to cut this down to an operationally reasonable number. This was done by combining strata. There were two criteria for combination of strata. If the model, as described in the previous section on outliers, fit the combined strata as well as two distinct models fit two separate strata, it makes sense to combine them. This can be for- malized with an F-statistic. Such statistics were used as a very rough guide in decisions to combine or not to combine strata. Also, and very importantly, we used our knowledge of the hospitals to attempt to arrive at reasonable groupings for the final strata. These were: I. Nonprofit, nonteaching hospitals II. Nonprofit, teaching hospitals III. For-profit hospitals 108 VIII. Computations Based on Less than a Full Year’s Data Where it is necessary or desirable to employ data for a time period less than the full reporting period of 1 year, several adjustments are necessary. Denote by t the number of days for which data have been reported and are available. We can then simply inflate several of the measures to the scale of a full year as follows: Let f = 365/t, be the inflation factor. If we have data for a single day, t = 1, and f = 365. If we have data for a week, t = 7 and f = 52 (rounded). If we have data for 3 months t = 90 and f = 4 (rounded). The variables to be inflated (multiplied by {) are: Total admissions Births Emergency visits All visits Surgical operations Adjusted patient days ‘Total nursing hours RN hours LPN hours Aides, orderlies and/or attendants hours General duty nurse hours General duty and head nurse hours After inflation the variables should be appropriately standardized by total admissions or adjusted patient days. Then one may follow the example on page —. 109 IX. Application of Regression Models Since the independent variables are expressed in different units, the fol- lowing table will serve as a ready guide for the application of regression models. Independent Variable Unit for Use in Regression Models Total number of facilities Number of “high” technical facilities Fraction of “high” technical facilities Total admissions Length of stay Occupancy Bassinets per statistical bed Births per statistical bed Emergency visits per admission Total visits per admission Surgical operations per admission Statistical beds Adjusted patient days SMSA Percent population under 18 Percent population 65 and over RN’s per 100,000 population LPN’s per 100,000 population Aides, etc. per 100,000 population Per capita income Percent families under poverty level Same Same Same In units of 100,000 Same In proportions Same Same Same Same Same Same In units of 100,000 SMSA = 1, Non-SMSA = 0 In percentages In percentages Same Same Same In units of 100,000 In percentages w U.S. GOVERNMENT PRINTING OFFICE: 1975 O—5635-048 110 DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE PUBLIC HEALTH SERVICE POSTAGE AND FEES PAID U.S. DEPARTMENT OF H.E.W. HEALTH RESOURCES ADMINISTRATION HEW-394 BETHESDA, MARYLAND 20014 OFFICIAL BUSINESS DHEW Publication No. (HRA) 75-6 U.S. DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE Public Health Service ® Health Resources Administration co292215eH