93d Congress } 2d Session COMMITTEE PRINT — NATIONAL NUTRITION POLICY BACKGROUND READING DOCUMENT PREPARED BY THE SUBPANEL ON NUTRITION AND DISEASE OF THE PaneL on Nutrition AND Hearts TO THE SELECT COMMITTEE ON NUTRITION AND HUMAN NEEDS | UNITED STATES SENATE ) A Vam bd Health, blbrer % Joe: ) Printed for the use of the Select Committee on Nutrition and Human Needs JUNE 1974 U.S. GOVERNMENT PRINTING OFFICE 34-621 O WASHINGTON : 1974 TR For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 - Price $1.20 Cos Ix Te) \) VN VO os Ra a i a | i i { i ih \ \ Hn il GEORGE MCGOVERN, South Dakota, Chairma HERMAN E. TALMADGE, Georgia CHARLES H. PERCY, PHILIP A. HART, Michigan WALTER F. MONDALE, Minnesota EDWARD M. KENNEDY, Massachusetts GAYLORD NELSON, Wisconsin ALAN CRANSTON, California HUBERT H. HUMPHREY, Minnesota KENNETH SCHLOSSBERG, Staff Director ROBERT DOLE, Kansa: HENRY BELLMON, Oklahoma RICHARD S. SCHWEIKER, Pennsylvania ROBERT TAFT, Jr., Ohio COMMITTEE PRINTS (93d Cong., 2d Sess.) To Save The Children. Publications List and Witness Reference. Index to Hearings During 93d Cong., 1st Sess. Food Price Changes, 1973-74, and Nutritional Status—Part I. Reference Material to Part I—Food Price Changes. Food Program Technical Amendments. National Nutrition Policy. Guidelines for a National Nutrition Policy. National Nutrition Policy : National Nutrition Policy Experiences. National Nutrition Policy : Nutrition and the International Situation. National Nutrition Policy : National Nutrition Policy : National Nutrition Policy : National Nutrition Policy : National Nutrition Policy : National Nutrition Policy : National Nutrition Policy : Programs. National Nutrition Policy Availability. National Nutrition Policy : Nutrition and Food Availability. Nutrition and the Consumer. Nutrition and Health. Nutrition and Special Groups. Nutrition and Government. Nutrition, Health, and Development. Selected Papers on Nutrition Information and : Selected Papers on Food Security and Selected Papers on Technology, Agriculture Advances and Production. National Nutrition Policy : Nutrition and the Consumer—II. National Nutrition Policy : Nutrition and the International Situation—1II. National Nutrition Policy : Background Reading Document. National Nutrition Policy Study: Report and Recommendations—I (Panel on Nutrition and Food Availability). Report and Recommendations—II (Panel on Nutrition and the Consumer). Report and Recommendations—III (Panel on Nutrition and Government). Report and Recommendations—IV (Subpanel of Health Care Systems of the Panel on Nutrition and Health). Report and Recommendation—V (Subpanel on Popular Nutrition Education of the Panel on Nutrition and the Consumer). : (II) Re 628 N28 pul. Alnifed Diafes Denale SELECT COMMITTEE ON NUTRITION AND HUMAN NEEDS SENATOR GEORGE MCGOVERN, CHAIRMAN SENATOR CHARLES H. PERCY, RANKING MINORITY MEMBER KENNETH SCHLOSSBERG, STAFF DIRECTOR GERALD S. J. CassiDY, SPECIAL COUNSEL NATIONAL NUTRITION POLICY STUDY JEAN MAYER, General Coordinator Professor of Nutrition, Harvard University COCHAIRMEN: PANEL ON NUTRITION AND THE INTERNATIONAL SITUATION PETER G. PETERSON D. GALE JOHNSON Former Secretary of Commerce Professor and Chairman Chairman, Lehman Brothers, Inc. Department of Economics New York, N.Y. University of Chicago COCHAIRMEN : PANEL ON NUTRITION AND FOOD AVAILABILITY TERRANCE HANOLD RAY A. GOLDBERG Chairman, Executive Committee Moffett Professor of Agriculture and Business The Pillsbury Co. Harvard University Minneapolis, Minn. COCHAIRMEN : PANEL ON NUTRITION AND GOVERNMENT D. MARK HEGSTED WiLLIAM CAREY Professor of Nutrition Vice President School of Public Health Arthur D. Little Co. Harvard University Washington, D COCHAIRMEN : PANEL ON NUTRITION AND HEALTH WiLLiAM E. CONNOR WILLIAM J. MCGANITY Professor of Internal Medicine Professor and Chairman Director, Clinical Research Center Department of Obstetrics and Gynecology University Hospitals, University of Iowa University of Texas COCHAIRMEN : PANEL ON NUTRITION AND THE CONSUMER ESTHER PETERSON HOWARD SCHNEIDER President, National Consumers League Director Consumer Adviser to President Institute of Nutrition Giant Food, Inc. University of North Carolina COCHAIRMEN : PANEL ON NUTRITION AND SPECIAL GROUPS DORIS CALLOWAY RONALD POLLACK Professor of Nutrition Director Department of Nutritional Sciences Food Research and Action Center University of California at Berkeley New York, N.Y. EXECUTIVE DEPARTMENT LIAISON Department of State. EDWIN MARTIN Department of Agriculture CLAYTON YEUTTER Department of Health, Education, and Welfare CHARLES EDWARDS COOPERATING NATIONAL ORGANIZATIONS American Academy of Pediatrics American School Food Service, National Congress of Parents and Teachers American Dental Association Association National Dairy Council American Dietetic Association CARE National Foundation March of Dimes American Heart Association Grocery Manufacturers of America The Ameri for Clinical Nutrition American Institute of Nutrition Junior Chamber of Commerce The Pees: of Sooty Technologists American Public Health Association National Association of Food Chains The National Consum The Society for Nutrition Education National Nutrition Policy Study Hearings, June 19-21, 1974—U.S. Senate Office Bldgs. (1m) 6043 CONTENTS Page Subsection on Obesity and Disease... rem — vi IOC ON cn oo en i ae aE ae 1 Enclosures of key publications: The effects of overweight on cardiovascular diseases. Geriatrics 28 :80— OR, dT a cr ee a pe aii te Io ee 27 Overweight and hypertension. Circulation. 39 :403—421, 1969. _____ 37 Obesity and diabetes. Medical Clinics of North America, 48:1387- he IROL 00 or on i S m ioe — SR ea C 57 Obes and diabetes : The odd couple. Am. Jr. Clin. Nut. 21 :1434-1437, he” Obesity and the body image : I. Characteristics of disturbances in the body image of some obese persons. Amer. J. Psychiat. 123 :1296-1300, OT i mm pe a ee il i Obesity and coronary heart disease. Nutrah. Nutrition/American Heart. Vol, X, Ne. 8, 1078. i a Ki Social factors in obesity. JAMA 192:97-102, 1965___________________ 81 Influence of social class on obesity and thinness in children. JAMA Sy T0888 NOT oi mm it et i i ba we aR ea 87 Effects of weight reduction on obesity. Clin. Invest. 53 :64-67, 1974__ 93 New therapies for the eating disorders. Arch. Gen. Psychiat. 26:391- BO OT en mm ee in mon An mh Be me 107 A therapeutic coalition for obesity: Behavior modification and pa- tient self-help. Am. J. Psychiatry 131:4, 423-427, 1974_____________ 115 v) NATIONAL CONFERENCE ON NUTRITION POLICY June 19-21, 1974 Background Reading Document Panel on Nutrition and Health Subpancz)l on Nutritiom smd Disease Subsection on Obesity and Disease, prepared by John W. Farquhar, M.D., Professor of Medicine and Director, Stanford Heart Disease Prevention Program Stanford University, Stanford, California Jules Hirsch, M.D., Professor The Rockefeller University, New York, New York Albert J. Stunkard, M.D., Professor and Chairman, Department of Psychiatry and Member of Advisory Committee, Stanford Heart Disease Prevention Program Stanford University, Stanford, California (VD I. Introduction Obesity is considered to be an important contributor to many different disorders such as coronary heart disease, hypertension, stroke, diabetes, gall bladder disease, arthritis, pulmonary dysfunction, sleep disorders, social disability, and decreased ability to withstand trauma or surgery. Many, if not all, of these hazards to health can be decreased by weight reduction or by prevention of weight gain during childhood and middle age. The principal preventable health hazard related to obesity that this section will cover is that of the cardiovascular disorders of coronary heart disease (including myocardial infarction, sudden death, and angina pectoris) and of fatal and non fatal cerebrovascular accidents or strokes. The reason for emphasis on cardiovascular disease is because the impact on the population at large will be greatest on this category were an effective reduction in obesity achieved. However, the social and medical implications of the other disorders is also great indeed. Many of them will be covered by other panelists. It will also be clear that considerable overlap exists between the material covered in this section and the sections on diabetes, coronary heart disease, hypertension, and gall bladder disease. The material will be presented in two forms, first, excerpts from key publications with brief comments when needed to clarify their meaning and second, an enclosure of some of the publications in full. II. References 1. "Arteriosclerosis'" - A Report by the National Heart and Lung Institute Task Force on Arteriosclerosis. Vol. II, June 1971. DHEW Publication No. (NIH) 72-219. 2.) Gordon,T. and Kannel,W. The effects of overweight on cardiovascular ‘diseases. Geriatrics 28:80-88, 1973. 3. Chiang, B.N., Perlman, L.V. and Epstein, F.H. Overweight and hyper- tension. Circulation 39:403-421, 1969. 4. Smith, M., Levine, R. Obesity and diabetes. Medical Clinics of North America 48:1387-1394, 1964. 5. Bierman, E.L., Bagdade, J.D., and Porte, D.,Jr. Obesity and diabetes: The odd couple. Am. Jr. Clin. Nut. 21:1434-1437, 1968. 6. Stunkard, A.J. and Mendelson, M. Obesity and.the body image: I. Characteristics of disturbances in the body image of some obese persons. Amer. J. Psychiat. 123:1296-1300, 1967. (1) 7. Coronary Heart Disease in Seven Countries. Edited by Ancel Keys, Ph.D. American Heart Association Monograph No. 29, 1970. 8. Kannel, W. Obesity and coronary heart disease. Nutrah. Nutrition/ American Heart. Vol. I, No. 3, 1973. x : 9. * Kannel, W. The Framingham Study, Section 29, Washington, D.C. Superintendent of Documents, 1973. 10. Goldblatt, P.B., Moore, M.D., and Stunkard, A.J. Social factors in obesity. JAMA 192:97-102, 1965. 11. Stunkard, A.J., d'Aquili,E., Fox, S., and Gilion, R.D.L. Influence of social class on obesity and thinness in children. JAMA 221:579-584, 1972, 12, Olefsky, J., Reaven, G.M., and Farquhar, J.W. Effects of weight reduction on obesity . Clin. Invest. 53:64-67, 1974. 13. Stunkard, A.J. New therapies for the eating disorders: behavior modification of obesity and anorexia nervosa. Arch. Gen. Psychiat. 26:391-398, 1972. 14, Levitz, L.S. and Stunkard, A.J. A therapeutic coalition for obesity: Behavior modification and patient self-help. Am. J. Psychiatry 131:4, 423-427, 1974. ’ a ; : . 15. A.H.A. Abstract, Meyer, A.J., Maccoby, N., et. al. A multifactor education campaign to reduce cardiovascular risk in three communities: results in high risk subjects. To appear, Circulation, Vol. 69, November 1974. I1I. Magnitude of the problem of the arteriosclerotic and hypertensive cardiovascular diseases A. Selections from Reference #1. The term "arteriosclerotic and degenerative heart disease " as used in this publication is synonymous with the term "coronary heart disease’ described in the Introduction. MORTALITY FROM ARTERIOSCLEROSIS National mortality statistics are the tabulation of the underlying cause of death entered on each death certificate filed in the United States, thie. Speed of death being coded to the International Statistical Classification of Diseases, Injuries, and Causes of Death. These statintics which are pre- sented below are the most impressive, the most reliable and the most readily available measure of the nation's problem of arteriosclerosis. It should be pointed out, however, that mortality statistics are not precise measures be- cause of the problem of accuracy of diagnostic information on death certifi- cates and the weakness inherent in the concept of a single underlying cause of death. Diseases of the Cardiovascular System In 1967, 54.1 percent of all deaths were attributed to diseases of the cardio- vascular system (Figure 1, Table 1). Of importance to our nation in terms of manpower and economic productivity, 30.5 percent of all deaths in the 35 to 44 yearn age group were due to these diseases (Figure 2). This percentage in- creases with advancing age s0 that these diseases account for 70.3 percent of the deaths among persons age 75 years and over. It is important Zo note that mone than one-third of atl deaths for persons under age 65 are due to cardio- vascular diseases. Arteriosclerosis and related diseases such as hypertension account for the : major portion of deaths from cardiovascular disease (Figure 3). Virtually all arteriosclerotic and degenerative heart disease is due to atherosclerosis, as is most aortic and peripheral arterial disease. A large proportion of strokes is due to hypertension and arteriosclerosis. Thus, over 80 percent of car- diovascular disorders can be ascribed to arteriosclerosis and its complications, FIGURE 1 Percent Distribution of Deaths From Specific Causes To Deaths From All Causes, United States, 19677 ~ Cardiovascular Disease 54.1% Z \ Other Causes 23.0% TABLE 1 Number of Deaths by Cause of Death and Broad Age Groups United States, 19672 Under Age 65 Cause of Death All ages age 65 and Over#* All Causes 1,851,323 715,161 1,136,162 Cardiovascular diseases (330-334, 1,002,111 250,977 751,134 400-468) : Arteriosclerotic and degenerative 625,850 167,495 458,355 heart diseases (420-422) Vascular lesions affecting central 202,184 36,761 165,423 nervous system (330-334) Diseases of arteries (450-456) 53,326 6,911 46,415 Hypertensive heart disease and 61,126 15,542 45,584 hypertension (440-447) Other diseases of the heart (430-434) 31,267 10,302 20,965 Rheumatic fever and rheumatic 14,176 8,930 5,246 heart disease (400-416) Diseases of the veins and other 14,182 5,036 9,146 diseases of the circulatory system (460-468) Other Causes of Death 849,212 464,184 385,028 * Includes age not stated Note: Numbers in parenthesis are codes of the International Classification of Diseases, Seventh Revision FIGURE 2 to Deaths cular Disease Percent Breakdown of from all Causes by Age Group, Deaths from Cardiovas 0 54 55-64 65- fe oN 1 < 45 United States, 19672 Under 5-14 15-24 2 os! (38) L322 \o2l Ad P ANE i FIGURE 3 Percent Breakdown of Deaths from Cardiovascular Disease by Major Sub-Group and The Proportion Due to Arteriosclerosis*: United States, 1967 RTERIOSCLEROTIC AND DEGENERATIVE HEART 2.5% OTHER CV § DISEASE HYPERTENSIVE \\ fq DISEASE \ / X4.2%/, 0 DISEASES OF STROKE Bal 5 : ARTERIES »" IV. Evidence that problem of the coronary heart disease component of cardiovascular disease is relatively a more serious problem in the United States than in many other countries on which such data has been collected. A. Note that this graph depicts the death rates in men aged 45-54. This highlights the fact that relatively young people are affected in the United States and that preventive efforts will surely be of greater effect in the under 65 age group. (See following page) FIGURE 6 - x Death Rates from Arteriosclerotic and Degenerative Heart Disease and fro All Causes; Selected Countries, 1967 -~ Men aged 45-54 years” "Rate Per 100,000 Population 0 200 400 600 800 1000 1200 I I I I I 1 1 Finland A UNITED STATES 727A | New Zeolond BZA Scotland Eh No. Ireland 4 Australia p74 Canada RA England & Wales Israel Y._. Norway a | Belgium aa. 0. Netherlands ve ] Denmark zy Germany (Fed rp) 2 Czechoslovakia ZA | Austria. rE... Hungary : ea... Sweden © Italy - eT Switzerland ea... Japan : : 27) Arteriosclerotic and Degenerative Heart Disease 3 Other Causes of Death *ICD 420, 422 10 V. Direct and indirect economic costs of the cardiovascular diseases in the U.S.A. A. Note that the total annual costs estimated for the year 1967 were 23.9 billion dollars, this figure in 1974 will be con- siderably greater due to the combined factors of inflation and to the immense cost of the recent growth of the use of the procedure of sur- gical bypass grafting of narrowed coronary arteries. TABLE 1 Estimated Economic Costs to the Nation due to Morbidity and Mortality from Arteriosclerotic and Hypertensive Diseases, United States, 1967% Amount (in millions) Total Morbidity Mortality Diagnosis Costs** Direct Indirect Costs Costs Total (Arteriosclerotic or $23,887 $4,291 $3,133 $18,463 hypertensive) Arteriosclerotic Heart Disease 15,545 2,072 370 13,103 Stroke **% 4,605 971 235 3,399 Diseases of Arteries 1,082 354 31 697 Hypertensive Diseases ***#% 2,655 894 497 1,264 * From Table 2, 3 and 4. ** Present value of lifetime earnings (discounted at 6%) of persons who died of arteriosclerotic and hypertensive diseases. *** Includes 20% not due to arteriosclerosis a *%%* Does not include all arteriosclerotic disease in which hypertension was involved. 1 VI. Contribution of obesity to the cause of cardiovascular disease in the United States. A. Figures 3 and 4,page 82 from Reference #2. The "Relative Odds" are the relative probabilities of future development of the disorders listed. . - Spas: opps talent .- ae Cr mea meee ry inn =a Figure 3. Odds according to relative weight: 3 Figure 4. Odds according to relative weight: Men, Framingsn Stowe, 16 year ar istionup : Women, Framingham Study, 1 year follow- ‘ & ° nN on nN LB oN og o al Relative odds of specified event HE o Relative odds of specified event oS wu 9% 100 110 120 130 140 150155 - 90 100 110 120 130 140 150155 _ Ratio: actual weight/ Metropolitan Life Insurance os I : Ratio: actual weight/Mstropolitan Life Insurance TE ’ Company ideal aight (percent) = “ : Company J ideal weight (percent) - = *Congestive heart failure {Atherothrombotic brain infarction 2, Birt = = Coronary heart disease “” Sintermittent claudication ~ 82 GERIATRICS, August 1973 Comments. It can be seen that all three cardiovascular disorders of coronary heart disease, acute brain infarction (i.e. "stroke'') and congestive heart failure are increased progressively as overweight increases. For intermittent claudication (I.C. on the graph) there was an inverse relationship for men; possibly because overweight men are less active than are lean mean and this symptom (cramps in the legs that appear on walking) requires physical activity to become manifest. 34-6210 - 74-2 VII. Contribution of obesity to the problem of hypertension. A. Excerpts from the review article of Chiang, Perlman and Epstein, Reference #3. 1. Summary, page 403 SUMMARY The inierrelationships between hypertension and obesity, two common and major health hazards, are reviewed. Comparisons of simultaneous intra-arterial and cuff blood pressure measurements indicate in general that the association between blood pressure and body weight is real and independent of arm circumference. Hypertension is more common among the obese than among the nonobese and, conversely, a significant proportion of hypertensive persons in the population are overweight. Obese hypertensive subjects experience a greater risk of coronary heart disease than the nonobese, and mortality rates for obese hyperiensive persons are higher than for those with obesity alone or hypertension alone. Weight reduction has been shown to lower blood pressure, 2nd it may bring about a more favorable prognosis in obese hypertensive persons. Pos- sible mechanisms that may be responsible for the frequent association between obesity and hypertension have been discussed. Irrespective of the underlying pathophysiologic mechanisms, the adverse melabolic and hemodynamic cffects of obesity upon hyper- tension impose an extra burden and strain on the circulatory system 2nd compromise its functional adequacy. Although it is not precisely Jmown 1o what extent weight reduction alone may be effective in controlling or preventing the lesser degrees of hypertension, the control of obesity should be an intrinsic part of any therapeutic or preventive antihypertensive regimen. Additional Indexing Words: 3 28 mm 29%, YUGOSLAVIA 28 %, ITALY 32%, NETHERLANDS NARROW, SOLID BARS SHOW CHD INCIDENCE RATE Figure S6 Percentage of men with values of 28 or more mm for the sum of the skinfolds over the triceps muscle and over the tip of the scapula. Comment. The "sum of skinfolds" is a measure of subcutaneous fat which correlates well with obesity defined by other means. 18 XI. Calculation of estimated impact on health of weight reduction (or of obesity prevention). A. Benefits of a drastic decrease of 17% in weight of adults aged 50 in U.S.A. Assuming an average 30 pound loss in males (from 180 to 150 1b), one can calculate an estimated decrease of 28% in cardiovascular risk. In lives saved this translates into 54,000 male and 8,500 female lives saved per year for a total of 65,000 lives per year. In non fatal cardio- vascular events (including non fatal strokes and myocardial infarction and angina pectoris) the saving would be approximately four times as great. The economic benefits can be calculated as a simple 27% fall in costs of the $23 billion annual cost to the nation, or $6.2 billion per year. B. Benefits of a modest decrease of a 9% fall in weight of adults aged 50 in the U.S.A. This assumes a 16 pound fall (from 180 to 164 1b) for males and a 14 pound fall (from 150 to 136 1b) for females. The estimated fall in cardiovascular risk is 14%. The impact on lives saved per year is 26,625 males and 4,313 females for a total of 31,000 per year. Again one can predict about four times the mortality figure (124,000) in non fatal and often dis- abling illnesses prevented. The economic benefits are estimated as 14% of $23 billion or $3.9 billion. C. Assumptions necessary for A and B above. One must assume that individuals who lose wéight immediately assume the risk of those who were never heavy. This is overly optimistic. One must assume that such in- dividuals would not change in other deleterious health habits (such as smoking, lack of physical activity, diets high in sugar, salt, saturated fat and cholesterol). This is overly pessimistic - therefore one can have an optimistic view toward the impact of obesity prevention or of weight loss. XII. Evidence for secular trends in obesity in the U.S.A. during the past two decades: (An increase among men and a decrease among women in the Framingham population). A. Excerpt from Reference #8, page 3 "Yet, despite the wide recognition of the health consequences of obesity, its high prevalence in affluent societies con- tinues unabated. In Framingham more than 15% of men and 20% 19 of women are at least 35% above "ideal weight." Men are heavier than their counterparts were two decades - ago. Most women though lighter than formerly, are still distinctly overweight, particularly in middle age and beyond.’ Comment. Further data is contained in Reference #9. B. Excerpt from Reference #2, Figure 2, page 81 Figure 2. Changes in average relative weight in ~~ : 18 years, Framingham Study: Persons aged 45 to 48 130— s—a—ee Men Women Ratio: actual weight/ Metropolitan Life Insurance Company ideal weight (percent) 8 13 Comment, This is evidence for the changes (increased for men and decreased for women) in relative weight that occur in the same cohort over time. Item A above referred to the fact that in 1970 men dged 45-59 were heavier than weré a similar age group in 1950. XIII. Evidence that the prevalence of obesity is highest in the lowest socioeconomic groups in the U.S.A. A. Summary from Reference #10 (See following page) The relationship between obesity and several social factors was investigated among 1,660 adulls representa- five of a residential area in midtown Manhatlan. An in- verse relationship previously described between obesity and parental socioeconomic status was also found be- tween obesity and one's own socioeconomic status. Obes- ity was six times more common among women of low status as compared fo these of high status. Furthermore, upwardly mobile females were less obese (129%) than the downwardly mobile (229%). Finally, the longer a woman's family had been in this country, the less likely she was to be obese. Similar but less marked trends obtained for the men. Suggestive relationships between ethnic and religious faclors and obesity were also found for both sexes. These findings suggest opportunities for more effective weight control measures through programs specially tailored for populations at high risk. B. Summary from Reference #11 Several social factors have been closely linked 10 obesity and thinness in adults. This study, based on 3,344 measurements of triceps skin-fold | thickness found similar relationships in white urban children. Obesity was far more prevalent in the lower-class girls than in those of the upper class—nine times as prevalent by age G. Similar though less striking differences were found between boys of upper and lower socioeconomic status. The patiern of thinness among girls was similar to that previously reported in women, with significantly more thinness in the upper-class group. Among boys, as among men, there were no such differences. The remarkably early onset of class-linked differences in prevalence of obesity underlines the importance of attempts to prevent the disorder in childhood. These attempts should be directed particularly toward those at high risk because of their lower socioeconomic status. XIV. Evidence that weight reduction is of benefit in reducing cardio—- vascular disease risk factors. A. Excerpt from Reference #2. Figures 5 and 6, page 84 (See following page) 21 Overweight and cardiovascular discases : ao ; : / ~ Figure 5. Changes in serum cholesterol level with Figure 6. Changes in systolic blood pressure level with changes in relative weight, Framingham Study © changes in relative weight, Framingham Study OF A. ! 30 4 ~_ eli $13 £ t ££ . E) @ p 0 “ 9 °3 al 8 by | se af 8 3 o \ 1 ® > oS \ 52 £ Men ! 52 £ 5 £2 4 Women 22 Men % ©2 10} / \ ©3 10 Women © e Pd i \ —— : Increases —' } =“ Increases —» ! deeb) 30 10 20 30 Change in serum | cholesterol / Change in systolic / blood pressure | § i i f g 7. 8 / 20 2 Zz ! 8 Zi 8 nd ; i —20F 8 7 + o i a2 el - . ar : -30L 4 i —3L¥_- : . . } . B. Excerpt from Reference #12, Figure 2, page 70 2 5. ! before offer before after 300 ; 1 3 1 P<0.000! —— | p<0.0001 T — i o . € 200 Hy € y 100 Ficure 2 Plasma TG (A) and cholesterol (B) concen- trations in 36 patients before and after weight reduction. Data are given as means=SEM. Comment. These changes in plasma lipids (TG = triglyceride) occurred following an average of 20 1b of weight loss per subject. 22 C. Excerpt from Reference #3. Table 3, pages 412-414 Table = Effect of Weight Reduction on Blood Pressure Senior nuthor Study group Results Benedict 25 nurmal men studied; on B.P. drop observed in 11 men. Average B.P. 1918° Jow calorie diet until 10-12% drop from 120/83 to 102/69 mm Hg. 3 (86) weight loss, then maintained borderline hypertension: B.P. before weight reduced weight Joss 142/90, 130/90, 120/90; after weight loss B.P. on 22nd day: 100/65, 100/70, 90/70 mm Hg ” Preble 1) 194 obese patients, wt. There was uniform B.P. drop with wi. Joss; 1923 loss at least 10 lbs. Mean B.P. drop 154/96 10 133/86 mm Hg; (87) 2) 1,000 cases, with 62 hyper- afier weight loss, inean B.P. drop from tension, B.P. over 200 mm 219/129 to 170/108 mm Hg Terry 63 obese patients, mean age ~~ One year after weight reduction, mesn B.P. 1923 45, mean wt. 199 lbs. 58%, drop from 196/103 to 170/95 mm Hg (88) with hypertension, mean B.P. = 196/103 mm Hg; on 1,200 calorie diet for 1 yr. Bauman 183 obese patients, on low- 48 out of 101 with significant systolic B.P. 1928 calorie and low-salt diet, reduction; 72 out of 161 with significant (89) observed .for 9.8 months diastolic B.P. reduction Master 91 females and 8 males from Wt. reduction associated with B.P. drop, 1929 obesity clinic, 10-58 years of (more marked with systolic pressure), also with (78) age, wt. 170-225 lbs; 67% slowing of heart rate. In 53 patients, wt. loss with systolic pressure over 25-30 lbs., B.P. drop from 20-30 systolic and 150 mm Hg 15-20 diastolic. Regain of wt. followed rise of B.P. Wt. Joss also associated with improved exercise {olerance Fellows 294 obese Metropolitan Life =~ Weight Joss associated with significant B.P. 1931 Insurance Co. employees, on drop; 30 hypertensives showed favorable (90) 1,200 calorie diet; in 18 course 5 years after weight reduction. cases thyroid was added 75% of 294 with initial weight Joss. 5 years later, 193 re-examined, only 219% maintained low weight, 79% regained weight in whole or part Wood - Animal experiment Marked rise of systolic blood pressure when 1939 8 dogs: 4 normal, 4 with caused 1o gain large amount of wt. by feeding a (74) experimental hypertension diet composed of beef fat in both normal and . hypertensive dogs. B.P. drop with diet restriction and Joss of wi. Less change in diastolic pressure Evans 100 consecutive obese 75% of the hypertensives showed a fall of B.P. 1952 patients, G1 are hypertensive to normal of less than 135 nm Hg with (82) (systolic B.P. 140-180 weight reduction. Weight reduction -has no mm Hg) effect on obese individusls with severe hypertension Fletcher 38 obese hyperiensive fcmales For 38 obese hypertensives with systolic over 1954 (20% overweight or more), 150 mm Hg, mean weight loss was 32 pounds (81) B.P.over 150 systolicand/or over 6.4 months, mesn systolic pressure drop 100 diastolic mm Hg. Initial diet GOO calories/day, later to 1,000 calories/day. Average was 32 mm Hg. For 30 obese females out of the 38 with dizstolic over 100 mm Jig as well, mean wt. Joss over 6.5 months was 33 Ibs, and 23 Table 3 (continued) > Effect of Weight Reduction on Blood Pressure Senior suthor Study group Results nge 54.9 years. 21 obese diastolic pressure drop was 16.5 min Hg. No hypertensive females who significant B.P. changes observed in the failed to Jose weight and 20 control group of 21 obese hypertensives and obese normotensive feinales 20 obese normotensive females. A group of were uscd for control obese normotensive females who lost weight also showed B.P. drop ; Dubl 12 obese hypertensive Group 1: Low-calorie, no salt restriction only 1958 patients 20-100% overweight one out of four with B.P. reduction (849) having hypertension for 2-14 Group 2: Low-salt, no calorie restriction all years. Low-salt diet with four patients showed a significant 85-105 mg Na/daily, end low B.P. drop more than usual calorie 600-800 calorie/daily. experience. Obese hypertensive The Jow-salt diet and low- seems more sensitive to salt calorie diet were alternated to restriction make 2 single change at a Group 3: Low-calorie diet at first and Jow- {ime, calorie or salt salt diet sdded later: Only one out of four with B.P. reduction with low calorie diet alone at first; when low salt diet was added, all four showed significant reduction of blood pressure Salzano 16 normotensive obese Average systolic and diastolic pressure 1958 persons, 12 females, 4 male reductions were 12/8 mm Hg respectively; (80) with 14-85% overweight. 819% showed significant systolic pressure Weight reduction 2 Jbs/wk reduction, 65% showed significant diastolic pressure reduction Olson 83 normotensives and 38 No reduction of blood pressure in normoten- 1959 hypertensive subjects . sive group with a weight Joss of 14 Ibs. over 4-8 ©1) months period; 75% of hypertensive group 3 showed significant reduction of blood pressure after weight reduction Adlersberg 54 obese hypertensive © Average weight loss was 23% Ibs.; 72% with 1946 patients on 1,200 cslorie diet, significant B.P. drop; 28% with no change of (83) ~ ~ "no drug or dehydration, B.P. Those who maintained low weight bad seen 1940-1941, 15 patients favorable prognosis. Wt. reduction had no re-examined 1944 effect on retinopathy Keys Minnesota Experiment Mean B.P. drop 106/69 to 97/64 mm Hg. 1947 34 healthy men on 1,600 Measn pulse rate 56 to 37 per minute 7 calorie diet for 6 months, Mean VO, reduction 228 to 145 ml/min (control wt. 69.4 kg) Heart size also reduced Green 1,260 obese patients. Weight reduction brought 3 of the obese 1948 149 (11.8% hypertensive) hypertensives to normal B.P. Influence of (48) age 30-60 weight reduction on B.P. is inconsistent Brozek Observations on European 1. Incidence of bypertension decreased 1948 countries: Russia (Leningrad), 2. Hypericnsive complications decreased; (76) Holland snd Germany during clinical severity reduced World War IT with drastic 3. Autopsy findings of hypertensive disease food shortage 4 decreased - 4. Hypotension common among prisoners of war Kempner High carbohydrate diet, Jow Reduction of blood pressure with this diet, 1949 salt and high potassium probably due to Jow salt content as well as (85) weight reduction Wilhelmj Animal experiment. 4 normal 1. B.P. change during fasting period: 1951 dogs with 13 fasting periods. dog 1: 127/62 to 86/35 mm Hg (75) Mean wt. Joss 4 kg dog 2: 112/66 Lo 86/51 nm Hg (16-12 kg) dog 3: 102/60 to 74/45 mm Hg dog 4: 111/43 to 82/37 mm Hg 2. Associated with slowing of heart rate 3. The fall of B.P. to stable state is influenced by preceding nutritional stste, longer to achieve stable B.P. if previous diet is luxurious, shorter if previous diet is poor Martin 37 middle-aged obese patients 13 of 18 normotensives showed no change of 1952 wt. 212-218 Ibs; 18 hyper- B.P., 4 with systolic and or diastolic pressure (79) tensive, 19 normotensive; drop, one with B.P. rise; 12 of 19 hypertensives Average wi Joss 42 bs, for showed no change of B.P., 2 with both normotensive, and 36 Ibs. for systolic and diastolic pressure drop, 5 with hypertensives systolic fall alone * Date of publication XV. Evidence that new forms of medical care and public health education are of value in achieving weight reduction in normal populations or in high risk overweight groups. A. Excerpt from Reference #13. Summary, page 391 New approaches to psychotherapy, which appear more effective than traditional ones in modifying several kinds of disturbed behavior, have recently been applied to the eating disorders. Patients with both anorexia nervosa and obesity have responded to behavior modi- fication, and experience with obesity has already been sufficient to permit the development and description of relatively specific behav- ioral programs. These programs have been used to compare behav- ior modification in a systematic manner with a variety of alternate treatment methods. Every one of eight such studies has reported re- sults favoring behavior modification, an unusual example of unanim- ity in this heterogeneous and complex disorder. Furthermore, some new experi | designs developed in these studies are making a significant contribution to the study of psychotherapy and the eluci- dation of its effective components. B. C. 25 Summary of Reference #14 The effectiveness of a self-help organization Jor the obese was significantly increased by behavior modification techniques. Sixteen chapters of TOPS (Take Off Pounds Sensibly), with a 101al of 234 members, received one of Jour treatments: behavior modification conducted b y a professional therapist, behavior modification conducted by the TOPS leader, nutrition education conducted by the TOPS leader, and continuation of the usual TOPS program. During the three-month treatment period, be- havior modification produced significantly lower attrition rates and significantly greater weight losses than did the alternate treatment methods. At nine-month follow-up, the differences among treatments were even greater. Reprint of Reference #15 A MULTIFACTOR EDUCATION CAMPAIGN TO REDUCE | { CARDIOVASCULAR RISK IN THREE COMMUNITIES: -RESULTS IN HIGH RISK SUBJECTS ‘A.J. Meyer, N. Maccoby, J.W. Farquhar, S.H. Russell and M.P. Stern, Stanford Heart Disease :Prevention Program, Stanford, Calif. : : To assess the effects of behavior modification and media campaign techniques on risk reduction, ‘three groups of subjects at high risk for card- :iovascular disease were selected accordihg to ‘the Cornfield formula from randomly selected ‘subject pools in Watsonville (W), Gilroy (G) and .Tracy (T). W subjects (n=87) were treated with .a behavior modification protocol for the reduc-; tion of risk associated with overweight, smok- ing, lack of physical activity and a diet high ‘in cholesterol, saturated fat, sugar and salt. The protocol was administered in group setting or in subjects' homes over a three-month period ‘during an on-going risk reduction media cam- : ‘paign. A comparable group in G was exposed to initial screening and media campaign only; in T, to initial screening only. Significant differ- .ences in indicators of risk reduction among W, G and T subjects support the efficacy of behav- ‘ior modification and in some instances the media ‘campaign for risk reduction. For example, mean W cholesterol reduction =-11.4mg; G=-1.2mg*; T= 3.9mg*. Mean W systolic blood pressure reduc- ° tion =-9.5; G=-6.3; T=1.5%., Mean W reduction in relative weight =-5%; G=-1%Z%*; T=-2%*., The mean W reduction in eggs consumed per week = -3.3; ‘G=-2.6; T=-.8%, (“between-group comparison with Weep 30r- : » > ah ix 32? : 7 52 E- SE 1 4 B $ = Decreases 30 —20 -10 / - ih 9 aN Change in serum cholesterol greater cardiovascular risks of smoking. The impact of relative weight on blood pressure is not dramatic from an individual point of view but is significant for groups. Yeople 50 percent above ideal weight (par- ticularly women) are not rare; on the aver- age, they have a systolic pressure 14 mm Hg higher than those at ideal weight. This association is not adventitious. Weight gain is associated with a rise in both blood pressuse and serum cholesterol, and weight oss with a fall in these atherogenic attri- butes (figures 5 and 6). Again, the impact of such changes is substantial when one analyzes group statistics. 2 Weight fluctuation in the Framingham group has not been significant enough to produce substantial secular changes in car- diovascular disease incidence. While the average difference between the highest and lowest weights for individuals over 18 years of observation is large (over 21 Ib), this chiefly reflects short-term fluctuation. Per- sistent changes occur very slowly, and weight at one age is closely related to weight later in life for most people. Weights 18 years after entry had a correla- tion of 0.75 or 0.76 with weights at entry (table 3). Sy 7 Py A ar Fe 6. Changes in systolic blood pres3ure level with changes in relative weight, Framingham Study RE } i I 1 ow le 3 cm 2 3 02 2 8 oo Qo = 2 5:2 Men os 10} 7 : ® ot 3 ‘¢ Decreases =“ Increases ’ A fe a -30 0 2 Change in systolic blood pressure 1 S T T « Decreases An interesting subgroup of weight changes were the large weight losses (10 percent or greater) not immediately pre- ceded or followed by large weight gains. People who had such large weight losses fell into three major classes of approximately equal size: (1) weight loss preceded by a medical event that accounts for the loss, (2) loss due to deliberate dieting, 50 per- cent at a doctor’s suggestion and 50 percent self-determined, and (3) weight loss not explained by medical history. The first group had a large excess in subsequent mortality, at least four times that for the other groups, generally ascribed (as might be anticipated) to the medical condition preceding weight loss. These conditions covered a wide range of ills, including car- diovascular diseases. The other two cate- ~gories had no discernible excess mortality. It is difficult to judge whether weight affects the risk of cardiovascular disease . only by its influence on blood pressure and serum cholesterol. If we take into account other variables related to cardiovascular disease (specifically, blood pressure and the associated findings of eh ventricular hypertrophy seen in electrocardiographic studies, serum cholesterol, glucose intoler- 32 ance and cigarette smoking), a residual as- sociation of relative weight with cardiovas- Table 2. Averses correlation of relative weight 1 3 : and ce Nn risk factors cular disease independent of these still re- (Examination 2) mains that is statistically significant for CHD in men and for ABI and CHF in women (table 4). Thus, we have some in- dication that weight operates through other modalities as well, although these are as yet undetermined. Finally, what is the overall impact, direct or indirect, of overweight on cardiovascu- lar disease? If everyone above ideal weight at Framingham were at ideal weight in- stead, the incidence of CHD should be about 25 percent less, and the incidence of ABI and CHF, approximately 35 percent lower. If we were certain that an over- weight person who reduces would achieve the same risk status as if he were originall oth at the lower weight, weight es Table 3. Joriarion In ody an fs Jears would certainly have a powerful impact on cardiovascular disease. But since such ran reductions are difficult to attain and are Highest minus Men Women sometimes achieved by hazardous methods, Jowes! weight avoiding overweight in the first place would certainly be preferable. Discussion Conflicting findings have been reported concerning the role of obesity in cardiovascular disease in general and in the development of CHD in particular. Many investigators have been unable to support the belief that obesity is associated with myocardial infarction.*’* Others have found a relationship, but only in those in- cidents occurring under age 45.° Autopsy studies have in general failed to show a quantitative relationship between the de- gree of adiposity and the extent of coronary atherosclerosis found at necropsy, particu- larly in nonhypertensives and women.'*"" Also, coronary angiographic findings in pa- tients with hyperlipidemia and established CHD have not correlated particularly well with the degree of obesity of patients un- dergoing this examination.’ It thus would appear that physique in terms of relative weight or its equivalent might be accorded a relatively minor role in the etiology of coronary disease despite the striking. rela- tionship noted in insurance data.'® 33 A number of factors contribute to these conflicting results. For one thing, in the Framingham data obesity seems more strik- ingly related to sudden death and angina pectoris than to myocardial infarction, al- though the differences among the various clinical manifestations are not statistically significant.” This would explain the re- sults of clinical studies largely confined to persons who have sustained myocardial in- farctions. Also, patients with recent myo- cardial infarctions tend to lose weight, so that retrospective studies (including autop- sies) underestimate the usual weight status. Conflicting results also have been at- tributed to differences in the technics em- ployed for assessing adiposity since it may be confounded with other features of body build. Some have found no relationship of CHD to relative weight but could show an association with body fatness assessed by “skinfold” measurements.’ In Framing- ham, skinfold thickness correlates well with relative weight (table 5); it is approximate- ly as good as the correlation of skinfold measurements in one area of the body with another; for example, of triceps with sub- scapular skinfolds. Both subscapular skin- fold thickness and relative weight also show a correlation with risk of CHD (table 6). It is not related to triceps skinfolds, however. Furthermore, weight gain after age 25 is also related to the rate of coronary events. It thus appears that adiposity, how- ever assessed, is distinctly related to coro- nary morbidity and mortality. The mechanism by which adiposity con- tributes to cardiovascular incidence is still not fully established. That it contributes by providing increased cholesterol, higher blood pressure, and possibly higher blood sugar values seems clear, but whether it makes some unique extra contribution needs further investigation. Even if there is Table 4. T values of additional contribution of relative weight to discrimination of heart conditions by sex and age, Framingham Study, 16 year follow-up ABI wid 34 \ Table §. Correlations of various measures of obesity with relative weight* 3 (Examination 5) Skinfold Arm Ponderal Triceps Subscapular girth index Table 6. Standardized univariate regression coefficients for CHD incidence on various measures of adiposity, Framingham Study, 10 year follow-up Relative Skinfold Arm Ponderal weight Triceps Subscapular girth index no additional effect in either sex, however, it would be illogical to conclude that over- weight is unimportant in the evolution of cardiovascular disease. It is more reason- able to argue that the effect of weight is important but mediated through its in- fluence on blood lipids, blood pressure and carbohydrate tolerance. While the cardiovascular consequences of obesity are well established, the patho- genic mechanisms are not; obese people have been found to have an increased intra- vascular volume and increased cardiac work load for reasons that are not clear.” This may explain the unique contribution of overweight to cardiovascular disease. The mechanism by which accumulation of adipose tissue raises blood pressure, in- ‘creases the cardiac work load,* impairs glu- cose tolerance”? and alters blood lipid metabolism®* is poorly understood. The chief determinants of adiposity in affluent populations require further elucidation, al- though it seems clear that the inactivity and overeating prevalent in affluent societies must play a prominent role. Overfeeding in childhood seems to condition the adult to a predilection for obesity. *” Medical science would be well served by greater research attention to overweight and its effects on metabolism and the car- diovascular apparatus. Overweight is a ma- jor public health problem, contributing not only to cardiovascular but to other diseases such as diabetes, gallbladder dis- ease, arthritis and gout; that this is well known hardly justifies its neglect by re- searchers in cardiovascular disease. Obesity is probably the most common metabolic disorder affecting mankind. It is a serious condition adversely affecting sev- eral organ systems, causing decades of dis- ability and contributing to premature death. Effective means for preventing and correcting obesity have yet to be developed. No other field of medicine has been so filled with faddism or quackery. Obesity is a complex disorder and more information concerning its epidemiology, physiology, biochemistry and genetics is much needed. REFERENCES i. Build and blood pressure study. Society of Actuaries, Chicago, 1659, vol 1 2. Marks HH: Influence of obesity on morbidity and mortality. Bull NY Acad Med 36:296- 312, 1960 3. Lew EA: Importance of overweight in life insurance. Presented before the eleventh international conference of COINTRA, 1969 4. Keys A, Aravanis C, Blackburn H, et al: Coronary heart disease: Overweight and obesity as risk factors. Ann Intern Med 77:15-27, 1972 5. Gordon T, Kannel WB: The Framingham Study 20 years later. In Kessler II, Levin ML (Editors): The com- munity as an epidemiological laboratory: A casebook of community studies. Baltimore, Johns Hopkins Press, 1970, pp 123-144 8. Walker SH, Duncan DB: Estimation of the probability of an event as a function of several in- dependent variables. Biometrika 54:167-179, 1967 7. Shurtleff D: Some characteristics related to the incidence of cardiovascular disease and death. In Kannel WB, Gor- don T (Editors): The Framingham Study, Section 26. Washington, DC, Superintendent of Documents, 1970 8. Trulson MF, Stare FJ: Epidemiologic approach to coro- nary art disease.” Postgrad Med 30:67-73, 1961 8. Acheson RM: The etiology of coronary heart disease: A review from the epidemiologic standpoint. Yale ] Biol Med 35:143-170, 1962 10. Epstein FH: The epidemiology of coronary heart disease. A review. J Chronic Dis 18:735- 774, 1965 11. Sanders K: Coronary artery disease and obe- sity. Lancet 2:432-485, 1959 12. Stamler J, Berkson DM, Lindberg HA: Coronary risk factors. Med Clin North Am 50:229-254, 1966 13. Paul O, Lepper MH, Phelan WH, et al: Longitudinal study of coronary heart disease. Cir- culation 28:20-31, 1963 14. Gertley MM, White PD, et al: Coronary heart disease in young adults. Cambridge, MA, Harvard University Press, 1954 15. Pell S, D'Alonzo CA: Acute myocardial infarction in a large industrial popula- tion: Report of a 6 year study of 1,356 cases. JAMA 185: 831-838, 1963 16. Spain DM, Nathan DJ, Gellis M: Weight, body type and the prevalance of coronary athero- sclerotic heart disease in males. Am J Med Sci 245:63, 1968 17. Wilkins RH, Roberts JC Jr, Moses C: Autopsy studies in atherosclerosis. 111. Distribution and severity of atherosclerosis in the presence of obesity, hypertension, nephrosclerosis, rheumatic heart disease. Circulation 20: 527,1959 18. Cramer K, Pawlin §, Werko L: Coronary angiographic findings in correlation with age, body weight, blood pressure, serum lipids, and smoking habits. Circula- tion 33:888, 1966 19. Kannel WB, LeBauer EJ, Dawber TR, et al: Relation of body weight to development of coronary heart disease.- Circulation 35:734, 1967 20 Alex- ander JK: Obesity and cardiac performance. Am J Cardiol 14:860, 1964 21. Kannel WB, Brand N, Skinner I et al: The relation of adiposity to blood pressure and develop- ment of hypertension. The Framingham Study. Ann Intern Med 67:48, 1967 22. Heald F, Mueller P, Daugela M: Glu- cose and free fatty acid metabolism in obese adolescents. Am J Clin Nutr 16:256, 1965 23. Kalkhoff R, Ferrou C: Metabolic differences between obese overweight and mus- cular overweight men. N Engl J Med 284:1236, 1971 24. Salans LB, Hirsch J, Knittle J: Obesity. carbohydrate me- tabolism and diabetes mellitus. In Ellinberg M, Rifkin H (Editors): Diabetes Mellitus, Theory and Practice. New York, McGraw-Hill Book Company, Division of McGraw- Hill, Inc, 1970, pp 424-485 25, Whyte HM, Nestie PJ, Goodman DS: Cholestercl distribution and turnover in obesity in man. Israel J Med Sci 5:644, 1969 26, Albrink M], Meigs JW: Interrelationship between skinfold thick- ness, serum lipids and blood sugar in normal men. Am J Clin Nutr 15:255, 1964 2], Salans LB, Knittle JL, Hirsch J: The role of adipose cell size and adipose tissue insulin sensitivity in the carbohydrate intolerance of human obe- sity. J Clin Invest 47:153, 1968 37 Overweight and Hypertension A Review By Benjamin N. Cuianc, M.D., Lawrence V. Perim:X, M.D, AnD Freperick H. EpstemnN, M.D, SUMMARY The interrelationships between hypertension and obesity, two common and major health hazards, are reviewed. Comparisons of simulianeous intra-arterial and cuff blood pressure measurements indicate in general that the association between blood pressure and body weight is real and independent of arm circumference. Hypertension is more common among the obese than among the nonobese and, conversely, a significant proportion of hypertensive persons in the population are overweight. Obese hypertensive subjects experience a greater risk of coronary heart disease than the nonobese, and mortality rates for obese hypertensive persons are higher than for those with obesity alone or hypertension alone. Weight reduction has been shown to lower blood pressure, 2nd it may bring about a more favorable prognosis in obese hypertensive persons. Pos- sible mechanisms that may be responsible for the frequent association between obesity and hypertension have been discussed. Irrespective of the underlying pathophysiologic mechanisms, the adverse metabolic and hemodynamic effects of obesity upon hyper- tension impose an extra burden and strain on the circulatory system 2nd compromise its functional adequacy. Although it is not precisely known to what extent weight reduction alone may be effective in controlling or preventing the lesser degrees of hypertension, the control of obesity should be an intrinsic part of any therapeutic or preventive antihypertensive regimen. Additional Indexing Words: Obesity Eevee blood pressure and obesity —£_4 predispose -to coronary heart" disease. * The control of these two factors is, therefore, an essential part of any coronary heart disease prevention program. It is likely that the detri- mental effect of obesity is mediated, at least. in part, by the association between blood pressure and weight levels. In fact, the fre- quent coexistence of these two common con- From the Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan. Dr. Chiang is a recipient of an Intemational Postdoctoral Research Fellowship (5 FOJ-TW 1025) and Dr. Epstein is the recipient of a Research Carcer Award (HE-KG-6748) from the U. S. Public Health Service, Bethesda, Maryland. Blood pressure measurement Epidemiology - Adrenocortical steroids Hypertension ~~ Hemodynamic effects Weight reduction Prognosis ditions suggests that there is a causal relation ~-— between them so that weight gain constitutes one kind of environmental stress that brings a genetic predisposition toward hypertension into the open. A parallel situation exists in the case of diabetes, which may sometimes be controlled by weight reduction, implying that, conversely, overeating can precipitate it. To prevent coronary heart disease, there __ must be a constant search for environmental factors that may be modified to delay the on- set and progression of atherosclerosis and its consequences. The present review attempts to summarize some of the key evidence that links hypertension and obesity and to provide a scientific basis for the belief that weight control is likely to be an important ingredient of any program for the control of hyperten- sion in the community and in individual pa- tients. * t Prevalence of Overweight and Hypertension In the United States about 30% of men and 40% of women over the age of 30 are 20% or more above “desirable weight” as defined in the tables published by the Metropolitan Life Insurance Company.? The 1959 Build and Yiood Pressure Study of the Society of Actu- aries showed that 6% of men (ages 15 to 69) and 11% of women (ages 15 to 69) were 20% or more in excess of “average weight” (for height and age).* The difference in the percentage of persons 20% overweight by one of these standards depends obviously on the definition of overweight and the standard used. Weight is gained more frequently at certain ages or physiological periods. In women, this is most likely after the completion of growth— about age 20, during pregnancy, and after the menopause.” Men tend to gain weight between the ages of 25 and 40 years.® It has been shown that persons overweight as children, tzenagers, or young adults are likely to remain or become more overweight throughout life.! * Overweight, as defined above, ‘is the ratio of actual weight to average or desirable weight (specific for age, sex, height, and body build). An individual may be overweight on account of musculature or bony structure rather than excess fat, so that overweight and obesity are not necessarily synonymous. Aver- age weight tables for men and women aged 15 to 65 years are provided in the 1959 Build and Blood Pressure Study of the Society of Actuaries. Desirable weight tables have been constructed by the Metropolitan Life Insur- ance Company,® derived from the 1959 Build “and Blood Pressure Study. The National Cen- ter for Health Statistics, Public Health Service, has published data on height, weight, and selected body dimensions based on a nation- wide adult population sample instead of an insured population? Since the component of overweight that can be reduced is fat, determination of rela- tive fatness is of practical significance. Obesity, 38 defined as excessive body fatness, can be measured in many ways. Some methods are quite complex, such as densitometer * or hydrometry studies and determination of whole body potassium content; others are relatively simple, including measurement of subcutane- ous skin folds, and soft-tissue x-rays.*1¢ The Commitiee on Nutritional Anthropometry recommended triceps and subscapular skin- folds as methods for characterizing over-all fatness’ since subcutancous fat makes up half of the total body fat content.”” For prac- tical purposes, triceps skinfold measurements alone are considered adequate as being sim- ple, reasonably precise, and reproducible.’ Beyond 9 years of age, fat accounts for a high- er percentage of total body weight in females than in males. Body fatness for both sexes increases with age and, in adult life, increases at a faster rate for men than women; yet, women at all ages are on the average fatter than men. The extent to which overweight measures obesity depends on the correlation between these two measurements, which varies according to body build. The level at which obesity assumes clinical and prognostic significance is not easy to define. Much of this information comes from insured popula- tions and cannot be easily extrapolated to the population at large. It would probably be conservative to estimate the prevalence of significant obesity in the United States, de- pending on age and sex, as being at Jeast somewhere between 20 and 30%. The prevalence of hypertension, like the prevalence of obesity or overweight, depends on the level one chooses as the line of division between “normal” and “abnormal.” If a systolic blood pressure of 160 mm Hg or greater and/ or diastolic blood pressure of 95 mm Hg or greater are chosen,’® data from the Tecumseh Study indicate that 27% of men and 37% of women aged 40 to 5S are “hypertensive.”" Relationship between Blood Pressure and Body Weight In this section, three questions will be dis- cussed: (1) what is the over-all correlation between blood pressure and body weight, (2) what proportion of hypertensive persons: are overweight, and (3) is the correlation between blood pressure and body weight a methodologic artifact? Correlations between Blood Pressure and Budy Weight Clinical and epidemiological investigations of the relationship between body weight and blood pressure from different parts of the world are summarized in table 1. Most popu- lation studies tend to show a rise of blood pressure with increase of body weight or adiposity. The correlation tends to be higher in countries where obesity is more common. Epstein reviewed geographic differences in various parts of the world, classified the age- blood pressure trends into groups according to the slope of pressure rise with age and suggested that populations who show the Jeast rise of blood pressure with age are those with lower average weight.?® Several studies that do not show a constant relationship between body weight and blood pressure have been reported from primitive societies or in social groups where either the range of body weight is narrow or hypertension is rare.?-*! Seasonal variation of blood pressure must be taken into account, at least in subarctic climates, where seasonal changes in skinfolds were correlated with variations in blood pressure? . The magnitude of the association between blood pressure and body weight in countries 39 relation coefficient between systolic blood pressure and relative weight at ages 30 to 59 was 0.3 for both sexes?” In an extensive Scandinavian study,?® the blood pressure difference between those 20% underweight and those 20% overweight averaged 16.9 mm Hg for systolic and mm Hg for diastolic pressure (11,063 men and 3,721 women, age 20 to 60). In another large study by Boe and his colleagues, in Norway, among 67,976 adult men and women, there was an increase in systolic and diastolic pressure of 3 and 2 mm . Hg, respectively, for a 10 kg increase in body where obesity is very common as in the United States, or common as in Scandinavia, may be illustrated by data from four studies. In the Tecumseh Study? in Michigan, the cor- relation coefficients between systolic pressure and relative weight were 0.29 for those aged 20 to 29; 0.33 for the group aged 30 to 39; 0.23 for those 40 to 49; and 0.21 for those 50 to 59; in the older age groups 60 to 69 and 70 to 79 correlation coefficients were only 0.13 and 0.05 respectively. For diastolic blood pressure?® correlation coefficients were highest for those aged 30 to 39 (0.32) and lowest for those aged 60 to 69 (0.05); for the other age groups, the values were 0.28 (20 to 29), 0.25 (40 to 49) and 0.26 (50 to 59). In the Framingham study in Massachusetts, the cor- weight. The relationship between blood pressure and body weight has been shown to be greater in women,®® 32.34 in those with a family his- tory of obesity and hypertension, 3¢ in the extremely obese, 3% %8 and in those less than ° 60 years old.2:2%27 Systolic blood pressure shows a greater association with body weight than diastolic pressure.33® Higher blood pressure was found in nonsmokers than smok- ers who are less heavy® In other studies, hypertensive patients were heavier than nor- motensive controls.*:42 In addition, relative weight was correlated significantly with the prevalence of hypertensive retinopathy,! as well as cardiovascular complication in hyper- tensive patients. #! Blood pressure has been shown to correlate with the ponderal index*® and skin fold thickness. Both Whyte?! and Kannel and his associates?” have noted blood pressure to be more strongly and primarily correlated with body weight rather than body fat and concluded that the correlation with body fat is a secondary phenomenon due to the association between weight and degree of fatness. It is of interest that in at least one popula- tion in which hypertension and obesity are both very common, no correlation between the two variables was found; these findings from a study among American Negroes in South Carolina,* remain to be explained and confirmed by other investigations. In the Evans County (Georgia) study, blood pressure was found to be higher in Negro women, but not in men with higher body weight! Miall and 40 Table 1 Significant findings Relation of Body Weight and Blood Pressure Senior author Study population Larimore 417 factory workers; 1923¢ B.P. and Lody build (44)t Marks Metropolitan Life Insur.; 1924 men age 28-42 years, U.S.A. (45) Huber 1,332 healthy military 1927 personnel, U.S.A. (46) Hartman 2,042 persons age over 15 1929 years, males, 959; females, (39) 1,083; Mayo Clinic Robinson . 10,883 healthy persons; 1939 7,478 males, 3,405 females, (33) U.S.A. Short * 3,516 healthy persons 1939 applying for periodic (47) health examination; U.S.A. Boynton 3 75,258 University students 1948 (U. of Minnesota), (35) 42,800 males, 31,258 females Green 1,260 obese patients © 1948 age 11-70, U.S.A. (48) Master 1,000 private patients 1953 age 20-64; 169 with (49) hypertension ; Thomas 266 Johns Hopkins Medical 1955 students; study of family (36) relation of overweight and hypertension Bjerkedal 14,784 adult population, 1957 11,063 males, (28) 3,721 females, Norway Boe 67,976 adult population, 1957 27,718 males, 40,258 females; (29) Bergen, Norway The asthenic had the lowest blood pressure; the sthenic had the highest blood pressure B.P. 7/6 mm Hg higher in persons 25% over- weight; overweight had higher blood pressure 2/3 of persons over age 40 with 10 lbs. over- weight or more had elevated blood pressure; 499 of those 10% underweight or more with low systolic pressure of 110 mm Hg or less; 18%, of those overweight with high systolic blood pressure over 140 mm Hg Systolic B.P. increases with increase of weight in both sexes; diastolic B.P. increases with increase of weight in females only Obese females and males have 6 and 1} {limes more hypertension respectively; B.P. increases with overweight, and low in under- weight in all age groups of both sexes More hypertension among overweight persons; B.P. difference of 11/4 mm Hg between overweight and underweight in the 50-59 age group Mean systolic and diastolic B.P. rises as weight increases in each age group; more marked in systolic B.P.; systolic B.P. higher in persons with positive family history Hypertension more {requently noted in mild to moderzie overweight; age distribution of obese hypertension did not differ from essential hypertension in general In males, more hypertension in overweight than the general population, ratio: 32 : 14; in females, no difference in hyperiension Family study showed high correlation between hypertension and obesity among parents, grandparents, uncles, and sunts > ‘Small increase of mean systolic and diastolic B.P. with increase of relative weight in all age groups; no excessive accumulation of hypertension in overweight persons Overweight has negligible effect on hyper- tension; 3/2 mm Hg increase in systolic- diastolic pressure for each 10 kg increase of body weight 41 Table 1 (continued) Relation of Body Weight and Blood Pressure Senior author Study populstion Significant findings Padmavati 1,132 low socioeconomic; * In low S-E class, B.P. rise with body weight, 1959 224 high socioeconomic; but not with age; in high 8-E class, B.P. (50) Delhi, India rises with both weight and age; systolic pressure rises 0.2-0.3 mm Hg per pound weight increase, diastolic 0.2 mm Hg per pound weight increase in both groups Build & B.P. Metropolitan Life Insurance Excessive mortality of overweight persons esp. Study 4,900,000 policy holders, in younger age group, male sex and 1959 U.S.A. hypertensive; mortality mainly due to (4) cardiovascular causes Nutrition 1,333 military personnel Both systolic and diastolic pressures Survey of incresse with each increment of relative Armed Forces 1960 (52) Stamler 1961 (51) Epstein 1965 (26) Doyle 1961 (53) Hunter 1962 (35) Truedsson 1962 (39) Palmai 1962 (25) Whyte 1959, 1965 —(21, 22) Reid 1966 (40) Shaper 1967 (23) age 20-49, Philippines Peoples Gas Co. employees, 1,329 men age 4049; Chicago, Ill, U.S.A. 8,641 adult population, age 20-79; Tecumseh, Michigan, U.S.A. Waterside worker, Australia Polynesians, Cook Island Healthy individuals, Sweden Subarctic area study of seasonal variation of B.P. 100 healthy men 20-24 years, ‘Australia ET 676 van drivers of General Postoffice age 40-59, England Three east African tribes, Northern Kenya weight Twofold increase in hypertension in over- weight persons than in underweight persons Significant correlation between B.P. and relative weight and skin folds; hypertensive heart diseese significantly correlated with relative weight in females only B.P. rises with skinfolds in younger age, up to 40 years; not in older people over-60 years Significant correlation between obesity and hypertension and hypercholesteremia but coronary heart disease is rare Skin folds correlate with blood pressure in females, but not in males Scasonal changes of B.P. correlation with changes of subcutaneous skinfolds B.P. correlates significantly with bulk in young Australians, no correlation in older “Australians. In New Guinea, B.P. does not rise with age, not related to relative weight; people are lean. B.P. of nonsmokers higher than smokers; the body weight is also higher in the non- smokers : B.P. no rise with increase of age or weight in two tribes, the third tribe showed B.P. increase with weight, also increased prevalence of hypertension; the last tribe showed increased ponderal index and consumed more carbo- hydrates 42 Tabie 1 (continued) Relation of Body Weight and Blood Pressure Eenior suthor Study population Significant findings Boyle 2,184 persons, B.P. rises with increase of weight in white 1967 Charleston Heart Study, people, but no significant difference in - (30) USA. weight/B.P. relation in Negro population who have more hypertension McDonough 3,102 subjects age 15-74, Kfore hypertension in Negro population than 1967 Evans County, Ga., USA. white. In the Evans County (Georgie) study (31) blood pressure waz found to be higher in Negro 3 fernales, but nol males with higher body weight Chiang Taiwan Cardiovascular Hypertensive subjects average 5 kg heavier 1967 Study, 1,822 men age 40-59, than age-matched normotensive controls; the (42) Taiwan hypertensives have significantly thicker skin- {olde 3 Tibblin Hypertension, 855 men, B.P. correlated with obesity; hypertensive 1967 age 50, Goteborg, Sweden retinopathy and LVH more in the obese {41) group (3-fold increase) Keys Cooperative study B.P. significantly correlates with skin folds 1967 in 7 countries except Serbian village of Velike Krsna, (24) Men, age 40-59 where there was no correlation; the men 3 are underweight and thin Kannel The Framingham study, More hypertensives in the overweight men 1967 5,127 men and women and women; B.P. correlates with relative 27) _age 30-62, U.S.A, weight througboui, al! age groups (30-62) Aleksandrow 10% of adult population, B.P. rises with weight increase; more 1967 Poland hypertension in obese group (54) * Date of publication. § Reference numbers are in parentheses. his associates®? have made a careful analysis of the relation between blood pressure and body weight, based on a large population sample in South Wales; a correlation was confined to younger persons, in accordance’ with a similar tendency already noted above in the Tecumseh population?® Their state- ment®? that the cause and effect relationship between the two parameters needs further investigation is well justified. In the meantime, summarizing the data presented in this sec- tion, the evidence is overwhelmingly in favor of the view that blood pressure and body weight are positively correlated. Frequency of Overweight among Hypertensive Persons According to most reports, overweight oc- curs more frequently in hypertensive than normotensive subjects.?)-26, 29-36. 39, 42, 44-55 J the Framingham study, the over-all prevalence of “obesity,” defined as a “Framingham Rela- tive Weight” of 120 or greater (that is, 20% above the height and sex-specific median), ‘was about 9% among men aged 30 to 59. From the data presented, it may be calculated that about 6% of the men were hypertensive, ac- cording to the definition used. Among these men, the prevalence of obesity was between 13 and 19%, depending on age, whereas the prevalence of obesity among normotensive controls (disregarding borderline hypertensive subjects in the population) varied between 2 and 4% in this age range. A similar trend has been demonstrated in the Tecumseh Com- munity Health Study.?® In the latter study, 33% of the participants (both sexes, age 40 to 59) in the upper quintile for blood pres- sure were in the upper quintile for relative weight.?% 4 Overweight women generally have more hypertension than overweight men. Robinson and associates,*® in a study of 10,833 persons, showed that obese women have hy- pertension 1% times more often than obese _ men. The frequency of overweight among hy- pertensive persons will obviously vary ac- cording to the definitions used for these two variables. Using definitions that, from the public health point of view, are neither too liberal nor too conservative, one may estimate that at least between one fifth and one third of all hypertensive subjects in the adult popu- lation are overweight. Since such persons have an increased mortality from cardiovascular and cerebrovascular diseases, this fact has preventative implications as far as blood pres- sure lowering through weight reduction is con- cerned, as will be discussed later. Comparison of Direct Intra-Arterial and Indirect Ausculiatory Methods for Blood Pressure Measurement Most observers agree that blood pressure increases with body weight or skin folds. The problem is whether the increase is true or an artifact resulting from the influence of large arm girth, thick subcutaneous fat, or both. Ragan and Bordley®® compared 51 per- sons by simultancous measurement of intra- arterial blood pressure in one arm and the indirect auscultatory method in the other; 13 of the 51 cases had aortic insufficiency; it was found that indirect blood pressure tended to be lower than the direct readings for subjects with thin arms and higher for subjects with thick arms. Pickering and his colleagues’? used the same data, excluding the 13 cases with aortic insufficiency, to draw up a regres- sion equation and a table for correction of the indirect sphygmomanometer readings. Miall and Oldham?®® used this equation for correct- ing cuff readings by arm circumference in an epidemiological study but also showed a high correlation between arm circumference and weight (r=0.99). In 1952, Bjerkedal?® studied the blood pressure of 14,784 adults in Norway. He found a small increase of blood pressure with body weight and an excess of hyperten- sion in obese persons, concluding that this effect was due to the influence of amm girth as postulated by Ragan and Bordley,’® and Pickering and associates.’ However, careful review of his paper indicates that there are more individuals with substantially increased 34-621 O - 74 - 4 43 blood pressure than could be expected from the small artifact attributed to increased arm girth. Recent studies by Raftery and Ward® and Holland and Humerfel,®® comparing the di- rect intra-arterial and indirect auscultatory blood pressures failed to confirm the observa- tions of Ragan and Bordley; the difference between intra-arterial and cuff blood pressure, on the one hand, and arm circumference or skinfold thickness, on the other, were not correlated but a significant correlation be- tween arm circumference and the direct arteri- al pressure was found. Berliner and co-work- ers® studied simultaneous intra-arterial and indirect cuff blood pressure of 100 subjects, 61 were obese and 39 nonobese. While the indirect method generally, but not consistent- ly, underestimated systolic and overestimated diastolic pressure, among lean and moderately obese persons the discrepancies were usually of a minor degree; even in the extremely obese persons, large arm girth did not regular- ly cause falsely high readings so that a correc- tion for arm circumferences could not be ap- plied routinely. Similar studies by Karlefors and his associates? and Nagle®® showed good agreement between indirect and direct blood pressures at rest and during exercise. Kannel and associates,” in the Framingham study, - compared upper arm and forearm pressures in the same persons and found similar readings whether or not their arm girths were large. Alexander’ 38 studied a group of grossly obese patients with average weight over 300 pounds with simultaneous direct intra-arterial and indirect cuff pressure measurement; the blood pressure agreed to within 20 mm Hg systolic and 10 mm Hg diastolic on half of the occasions, and the cuff method gave false low readings as frequently as high readings; most of the patients were frankly hypertensive on the basis of intra-arterial readings. Reid and co-workers,* in a survey of British postal workers, found no significant partial correla- tion between blood pressure and arm circum- ferences. Khosla and Lowe® observed from a regression analysis of systolic and diastolic _ pressure, age, weight, and arm circumference in e population sample of employed men aged 45 to 69 years that the relationship . between arm circumference and sphygmo- manometer reading is indirect and due to the high correlation of arm circumference with body weight, which, in turn, is related to blood pressure. Lowe® analyzed indirect blood pressure readings in a population of over 5,000 employed men, aged 15 to 69 years; he found that, for a given age, sphygmomanome- ter readings increase with body weight and that weight and arm circumference are highly correlated. When the blood pressure is cor- recled for weight, the effect of arm circum- ference upon blood pressure is negligible. Karvonen® and later King®” have suggested that discrepancies between direct and indirect readings could be minimized by using a wide cuff in obese individuals. Tibblin,* in a cohort study of hypertension in men, found body weight and skinfolds not only related to the level of blood pressure, but also to the severity of hypertensive retinopathy. The prevalence of subjects with hypertensive eye-ground changes increased about three times, with an increase in subscapular skin- folds and body weight; the fundus changes were independent of arm circumference. Stamler! also found body weight to be cor- related not only with blood pressure, but also with hypertensive heart disease, as demon- strated by electrocardiographic or x-ray evi-. dence of cardiomegaly. In-summary, sources of variation in measur- ing blood pressure by the indirect sphyg- momanometer method, as pointed out by Rose and others,’ are multiple, arm circumference being only one source of possible variation in the recording of blood pressure. Recent studies on the effect of arm circumference on pressure readings appear to have shown that the dis- crepancy is minor in the majority of cases, and the difference between direct intra-arterial and indirect cuff readings varies in either direction in obese as well as nonobese sub- jects (table 2). The relationship between arm girth and sphygmomanometer pressure read- ings is likely to be an indirect one due to high correlation between body weight and arm 44 circumference, so that the cuff blood pressure reading should not be corrected for arm circumference. Correcting blood pressure by arm circumference will obscure the important influence of body weight on blood pressure in epidemiological studies.l 41: 01. 62. 6¢ Ryriher- more, it is concluded that the relationship between body weight and blood pressure is not a methodological artifact, due to falsely high blood pressure readings in heavier per- sons with fatter arms. Overweight as a Risk of Developing Hypertension and Hypertensive Heart Disease: Longitudinal Studies Few longitudinal studies on the relation of overweight and the risk of subsequent hyper- tension have been published?” 9-72 Levy and his colleagues®-" did a retrospective study of the medical records of 22,741 officers in the United States Army. An officer was considered to be overweight if he was heavier by 20 pounds or more than the standard given in Amy regulations. Two and a half as many men developed sustained hypertension in the overweight as compared to normal controls after the age of 45 in a total experience of 33,921 persons-years. When transient tachy- cardia and transient hypertension were added to overweight, the ratio of subsequent develop- ment of sustained hypertension was 12.3 fold. The ratios of retirement and death from cardiovascular-renal diseases for the same group were 4.1 and 2.0 times higher, respective- ly, than for the normal controls. ®: 70. 71 Stamler™ analyzed data on 746 men in the labor force of a Chicago utility company over a period of 20 or more years. A strong relationship between the incidence of hyper- tensive disease in middle age and the rate of increase of body weight in a 20-year period was shown. Men at desirable weight as young adults and exhibiting little or no weight gain over the ensuing decades had the lowest in- cidence rate of hypertensive disease, whereas men above desirable weight as young adults and with significant weight gains over the next 20 years had a 4 to 5 fold higher inci- dence rate of hypertensive disease. In another group of 584 employees of the same company Comparisons Beiween Direct and Indirect Blood Pressure M. easurements Names of Patients Table 2 Methods Differences between direct and Conclusions * Alexander and Dennis (37) (38) Raftery and Ward (59) healthy men 35 extremely obese persons 50 young healthy females Medicine DRS recorder Cournand needle Inductance manometer and ultraviolet, recorder compare with direct; cuffs 26 cm or 42 cm long, 13 cm wide Not mentioned Cuff size 14 X 17.5 cm London Schocl of Hygiene Sphygmomanometer difference with shorter cuff and less significant difference with longer cuff Direct systolic generally higher, diastolic generally lower, relationship variable The indirect-direct differences were not related to the level of blood presstire, arm skin thick- ness or arm circumference. Good agreement between phase I indirect systolic pressure and direct systolic pressure. Phase V indirect pressure rather than phase IV agreed better with direct diastolic pressure authors studied Direct Indirect indirect blood pressure Holland 47 patients and Hansen capacitance London School of Hygiene; Direct blood pressure higher Direct blood pressure Humerfelt stafl (not obese) manometer left arm sphygmomanometer three than indirect correlated with arm (60) 12 cm X 24 cm cuff circumference Ragan 51 patients (13 with Optical manometer 13 em cuff and mercury Direct higher in individuals with Blood pressure over- Boerdley aortic ineufficiency) : manometer thin arms, lower in individuals estimated in obese and (53) with fat arms. When grossly under estimated in non- obese and aortic insufficiency obese A patients excluded 83% agreement Berliner 100 patients, Sanborn Elget B ter cuff Direct systolic is generally | Cuff pressure under- Fujiy 61 obese meter measured 13 X 20 em higher; in obese, however, cuffl estimates systolic and Lee tapered sleeve pressure is higher; indirect overestimates diastolic Yildiz dinstolic is always higher but Garnter differences are of small magnitude (61) except in extremely obese - King Number not Statham P23 Db strain- Specially made and Direct generally higher systolic, A cuff atleast 42 cm long (67) mentioned; young © gauge on Electronics for ~~ molded rubber bladders lower diastolic; considerable is recommended for measurement of B.P. in an adult A high reading by cuff usually indicates hypertension Indirect method under- records more at high level of systolic pressure than at normal or low level ev 46 Table 3 Effect of Weight Reduction on Blood Pressure Senior author Study group Resulis Benedict 25 normal men studied; on B.P. drop observed in 11 men. Average B.P. 1918°* low calorie diet until 10-12% drop from 120/83 to 102/69 mm Hg. 3 (86) weight loss, then maintained borderline hypertension: B.P. before weight reduced weight loss 142/90, 130/90, 120/90; after weight Joss B.P. on 22nd day: 100/65, 100/70, 90/70 mm Hg Preble 1) 194 obese patients, wt. There was uniform B.P. drop wah wt. loss; 1923 loss at least 10 Ibs. Mean B.P. drop 154/96 to 133/86 mm Hg; (87) 2) 1,000 cases, with 62 hyper- after weight loss, mean B.P. drop from tension, B.P. over 200 mm 219/129 to 170/108 mm Hg Terry 63 obese patients, mean age ~~ One year after weight reduction, mean B.P. 1923 3 45, mean wt. 199 lbs. 589, drop from 196/103 to 170/95 mm Hg (88) with hypertension, mean B.P. > 196/103 mm Hg; on 1,200 calorie diet for 1 yr. Bauman 183 obese patients, on low- 48 out of 101 with significant systolic B.P. 1928 calorie and low-salt diet, reduction; 72 out of 161 with significant (89) observed for 9.8 months diastolic B.P. reduction Master 91 females and 8 males from Wt. reduction associated with B.P. drop, 1929 obesity clinic, 10-58 years of (more marked with systolic pressure), also with (78) age, wt. 170-225 Ibs; 67% slowing of heart rate. In 53 patients, wt. loss with systolic pressure over 25-30 1bs., B.P. drop from 20-30 systolic and 150 mm Hg 15-20 diastolic. Regain of wt. followed rise of B.P. Wt. loss also associated with improved exercise tolerance Fellows 294 obese Metropolitan Life =~ Weight loss associated with significant B.P. 1931 Insurance Co. employees, on drop; 30 hypertensives showed favorable (90) 1,200 calorie diet; in 18 course 5 years after weight reduction. cases thyroid was added 75% of 294 with initial weight loss. 5 years ? later, 193 re-examined, only 21%, maintained low weight, 799, regained weight in whole or part Wood. ~~ Animal experiment Marked rise of systolic blood pressure when 1939 8 dogs: 4 normal, 4 with caused to gain large amount of wt. by feeding a (74) experimental hypertension diet composed of beef fat in both normal and . hypertensive dogs. B.P. drop with diet restriction and loss of wt. Less change in diastolic pressure Evans 100 consecutive obese 75%, of the hypertensives showed a fall of B.P. 1952 patients, 61 are hypertensive to normal of less than 135 mm Hg with (82) (systolic B.P. 140-180 weight reduction. Weight reduction has no mm Hg) effect on obese individuals with severe hypertension Fletcher 38 obese hyperiensive females For 38 obese hypertensives with systolic over 1954 (20% overweight or more), 150 mm Hg, mean weight loss was 32 pounds (81) B.P. over 150 systolicand for over 6.4 months, mean systolic pressure drop 100 diastolic mm Hg. Initial was 32 mm Hg. For 30 obese females out of diet 600 calories/day, later to the 38 with diastolic over 100 mm Hg as well, 1,000 calories/day. Average mean wt. loss over 6.5 months was 33 Ibs., and 47 : Table 3 (continued) Effect of Weight Reduction on Blood Pressure Senior author Study group Results age 54.9 years. 21 obese diastolic pressure drop was 16.5 mm Hg. No hypertensive females who significant B.P. changes observed in the failed to lose weight and 20 control group of 21 obese hypertensives and obese normotensive feinales 20 obese normotensive females. A group of were uscd for control obese normotensive females who Jost weight also showed B.P. drop : Dahl 12 obese hypertensive Group 1: Low-calorie, no salt restriction only 1958 patients 20-1009, overweight one out of four with B.P. reduction (84) having hypertension for 2-14 Group 2: Low-salt, no calorie restriction all years. Low-salt diet with * four patients showed a significant 85-105 mg Na/daily, and low B.P. drop more than usual calorie 600-800 calorie/daily. experience. Obese hypertensive The low-salt diet and low- seems more sensitive to salt calorie diet were alternated to restriction make a single change at a Group 3: Low-calorie diet at first and Jow- {ime, calorie or salt salt diet added later: Only one out of four with B.P. reduction with Jow calorie diet alone at first; when low salt diet was added, all four showed significant reduction of blood pressure Salzano 16 normoiensive obese Average systolic and diastolic pressure 1958 persons, 12 females, 4 male reductions were 12/8 mm Hg respectively; (80) with 14-85% overweight. 819%, showed significant systolic pressure Weight reduction 2 1bs/wk reduction, 659, showed significant diastolic pressure reduction Olson 83 normotensives and 38 No reduction of blood pressure in normoten- 1959 hypertensive subjects sive group with a weight loss of 14 Ibs. over 4-8 (91) months period; 75% of hypertensive group showed significant reduction of blood pressure after weight reduction Adlersberg 54 obese hypertensive © Average weight loss was 231 1bs.; 729%, with 1946 . patients on 1,200 calorie diet, significant B.P. drop; 28%, with no change of (83) "no drug or dehydration, B.P. Those who maintained low weight had seen 1940-1941, 15 patients favorable prognosis. Wt. reduction had no re-examined 1944 effect on retinopathy Keys Minnesota Experiment Mean B.P. drop 106/69 to 97/64 mm Hg. 1947 34 healthy men on 1,600 Mezan pulse rate 56 to 37 per minute 7) calorie diet for 6 months, Mean VO, reduction 228 to 145 ml/min (control wt. 69.4 kg) Heart size also reduced Green 1,260 obese patients. Weight reduction brought 3 of the obese 1948 149 (11.8% hypertensive) hypertensives to normal B.P. Influence of (48) age 30-60 weight reduction on B.P. is inconsistent Brozek Observations on European 1. Incidence of hypertension decreased 1948 countries: Russia (Leningrad), 2. Hypertensive complications decreased; (76) Holland and Germany during clinical severity reduced World War II with drastic 3. Autopsy findings of hypertensive disease food shortage . decreased 4. Hypotension common among prisoners of war 48 Kempner High carbohydrate diet, low 1949 salt and high potassium (85) Wilhelmj Aniinal experiment. 4 normal 1951 dogs with 13 fasting periods. (75) Mean wt. Joss 4 kg (16-12 kg) Martin 37 middle-aged obese patients 1952 wt. 212-218 lbs; 18 hyper- (79) - tensive, 19 normotensive; Average wt. loss 42 lbs., for normotensive, and 36 Ibs. for hypertensives Reduction of blood pressure with this diet, probably due to low salt content as well as weight reduction 1. B.P. change during fasting period: dog 1: 127/62 to 86/35 mm Hg dog 2: 112/66 to 86/51 mm Hg dog 3: 102/60 to 74/45 mm Hg dog 4: 111/43 to 82/37 mm Hg 2. Associated with slowing of heart rate 3. The fall of B.P. to stable state is influenced by preceding nutritional state, longer to achieve stable B.P. if previous diet is luxurious, shorter if previous diet is poor 13 of 18 normotensives showed no change of B.P., 4 with systolic and/or diastolic pressure drop, one with B.P. rise; 12 of 19 hypertensives showed no change of B.P., 2 with both systolic and diastolic pressure drop, 5 with systolic fall alone * Date of publication followed up from 1954 to 1957, the incidence rate of hypertension and hypertensive cardio- vascular disease was twice as high in the over- weight group.” Kannel and his associates? reported on the development of hypertension in the Framingham study. Of 5,127 men and women aged 30 to 62 years at the initial examination and followed for more than 12 years, subsequent development of hyperten- sion was related not only to gross obesity but to relative weight, as determined on the initial examination. Thus, the risk of developing hypertension in the group 20% overweight was eight times greater than in the group 10% underweight, and the risk of developing hypertensive cardiovascular disease was about 10 times greater in the group 20% overweight as compared to all others.?” Men at desirable weight as young adults who reported little or no weight gain over a 20-year period had one . fifth the incidence of hypertension (= 150/90) as compared to those who had a significant gain of weight in the same period. In the Framingham study, after a 12 year follow-up period, the incidence of hypertension (= 150/ 95) was also shown to be significantly higher in persons who gained weight after their twenty fifth birthday. While 8% of those who had gained from 4 to 15 pounds were hyper- tensive, almost 20% of the group gaining more than 28 pounds were hypertensive. The 12-year incidence of hypertension could be related to both obesity and to initial relative weight. It was also noted that the initially obese but otherwise normal subjects developed diabetes, cerebrovascular disease, and coronary heart disease more often. 2.73 In summary, with few exceptions, increase of “relative weight over time is associated with a rise of blood pressure in population studies. Obese hypertensive patients experience a greater risk of coronary heart disease, cere- brovascular disease and mortality than persons with either hypertension or obesity alone. The risk of subsequent development of sustained hypertension in young adults with obesity or those becoming obese in the ensuing decades is greater than among young adults with nor- mal or less than normal weight who remain lean. Effect of Weight Reduction on Blood Pressure Experimental and clinical observations o.. the effect of weight reduction in 19 studies are summarized in chronological sequence in table 3. Animal experiments in normal and hypertensive dogs have shown that weight reduction by fasting is followed by a fall in both systolic and diastolic pressure and slow- ing of heart rate.” When the animals regained weight .by feeding a rich diet, the blood pressure and heart rate returned to previous levels. The preceding nutritional state of the animal is related to the degree of blood pres- sure fluctuation during weight reduciion and duration of achieving stable low pressure level.” 7% The animals’ systolic blood pres- sure is affected more markedly than diastolic pressure. During a period of drastic food shortage in World War II, jit was observed that the frequency of hypertension and hypertensive cardiovascular complications was markedly re- duced in Russia, Holland, and Germany during’ the war and reversed thereafter.’® In one human experiment, 34 healthy men were on a low-calorie diet for 6 months and lost about a quarter of their control body weight; there was marked reduction of blood pressure, heart rate, oxygen consumption, and decrease of cardiac size with reversal when nonnal weight -was regained.’ z Clinical studies on obese hypertensive pa- tients have demonstrated a general but vari- able and inconsistent fall of blood pressure with weight reduction.”®®! It is more con- sistent for systolic blood pressure,’®# in obese woman,® with mild to moderate hyperten- sion,’ substantial and prolonged weight loss,®2 and in combination with salt restric- tion.® Prolonged weight reduction in some obese hypertensive patients was said to have a favorable effect on the course of the dis- ease.® In one study, 38 hypertensive women who lost 32.4 pounds on the average showed an average blood pressure fall of 32.8 mm Hg systolic and 16.5 mm Hg diastolic; a control group without any change of weight had no “decrease in blood pressure.! The effect of caloric restriction on obese hypertensive pa- tients is enhanced by low-salt intake.®* 8 Dahl and co-workers® found obese hyper- tensive patiems more sensitive to salt restric- tion than non-obese hypertensives; they con- cluded that most low-calorie diets also contain 49 less sodium chloride, and the pressure-lowering effect is mainly due to salt restriction. In summary, in spite of the fact that the effect of weight reduction varies and difficulty is encountered in keeping patients on diet, _ weight reduction lowers blood pressure in a considerable proportion of obese hypertensive patients. It is desirable to prescribe weight reduction in combination with salt restriction as part of a therapeutic regimen for obese hypertensive patients. Overweight and Ilypertcnsion: Common Mechanisins and Pathophysiological Changes The mechanism of the frequent coexistence of overweight and hypertension is unknown. Possible mechanisms and influencing factors are summarized in table 4. Alexander and associates,” 38 studying a series of 40 ex- tremely obese persons with average body weight of 300 pounds, showed that the hyper- tensive individuals in this series had increased cardiac output and normal peripheral resis- tance. They postulated that an increased stroke volume is required to pump blood through an expanded vascular bed in the obese individual. Blood volume, oxygen consump- tion, and the transverse diameter of the heart Table 4 Overweight and Hypertension: Summary of Possible - Mechanisms and Influencing Factors 1. Hemodynamic: 1) Increased Vo, and A-V Og difference, more so during exercise3%:92.93 2) Increased cardiac output with normal peripheral resistance 1:22.38 3) Increased blood volume and plasma volume 38 II. Hormonal: 1) Increased adrenocortical function 97:98 2) Imbalance of renin-angiotensin-aldosterone sys- tem 0 IIL. Environmental: 1) Physical inactivity leading to reduced energy expenditure and overweight 43-94 2) Overeating and high-salt intake #¢ IV. Genetic: 1) Endomorph body type 28:29:33.43.44.65 2) Family history of obesity and hypertension 38-3¢ also were noted to increase with body weight. $7.38 Proger- and Dennig® found in- creased oxygen consumption in four moderate- ly obese patients, but there was no significant difference in cardiac output and A-V oxygen difference as compared to the controls. Taylor and his colleagues,” in a study of healthy adults, found basal cardiac output related to body weight (r= 0.54); and A-V oxygen dif- ference positively correlated to body fat (r=0.63). Whyte ** has suggested that 50 high blood pressure in the obese is a result of increased cardiac output, which is forced into a reservoir (comprising the aorta and large arteries), which does not increase with body weight. In individuals with diseased inelastic aortae, the eflect of increased cardiac output on blood pressure would be disproportionate- ly greater. Fletcher® suggested.that the higher blood pressure in obese persons may be a physiological response to the greater metabolic burden and-mechanical handicap imposed by obesity. These suggestions indicate that the mechanism of the hypertension related to obesity may be different from essential hyper- tension in which cardiac output is normal and peripheral resistance increased. Morris and Crawford,® in a national necropsy survey of 5,000 cases, noted that hypertensive heart disease was present in light workers five times more frequently than in heavy workers; obese individuals tend to be more inactive physically. Tibblin?? has suggested that hypertension and obesity may be the result of a third set of antecedent events, such as physical activity. Hartroft,% in a comparative autopsy series, found renal weight in obese hypertensive persons to be normal, but significantly reduced in the whole series of hypertensive individuals. Obese per- sons consume not only food above energy requirement but may also ingest more salt. According to Dahl and associates, blood pressure levels of obese persons are especially sensitive to dietary salt restriction. Conn®® suggested that the common triad of obesity, hypertension, and impairment of carbohy- drate tolerance in middle aged men may be related to a form of primary aldosteronism that activates the renin-angiotensin-aldoste- rone system. Mild degrees of hypoxia and increased A-V oxygen difference associated with obesity lead to increased red-cell mass and increased blood volume. Another common finding in obesity is increased corticosteroid secre- tion.” *® In pharmacological doses, cortico- steroids increase cardiac output® It is not clear whether the modest increase in 17- hydroxysteroid and 17-ketosteroid secretion in obesity could explain vascular hypertension or whether the increase in steroid production is a response lo increased circulating blood volume.’ | 3 Regardless of the underlying pathophysio- logical mechanism, the hemodynamic and metabolic changes in obesity may exert an adverse effect on hypertension. The increased cardiac output and blood volume in obesity, when imposed wpon persons with hyperten- sion, may further strain the circulatory system and compromise its functional adequacy. These adverse effects are diagramatically illustrated in figure 1, which illustrates the various patho- physiological mechanisms that obesity may add to hypertension. Weight reduction in the obese hypertensive may remove the extra metabolic and hemodynamic burdens that Obesity - - N/a @/N ay Voy = oxygen conmumplion CO.* cordior outiput BP.« blood preswure VR = peripheral wosculor resistonce LV ~ left veniricuilor worklood (Vevolume ilood, P= pressure food) Figure 1 A suggested mechamism for the adverse hemodynamic effect of obesity in hypertensive subjects. obesity imposes upon hypertension. It seems plausible that weight reduction would dimin- ish circulatory strain, thus preventing acceler- ated decompensation in hypertensive heart discase among the obese, especially when coupled with appropriate antihypertensive therapy. Prognosis of Overweight and Hypertension The Build and Blood Pressure Study in 19594 showed excessive mortality in overweight persons among all age and blood pressure groups. When mortality is assessed by height and weight levels for three blood pressure groups (normal, moderate, and more severely elevated) and three age groups (15 to 39, 40 to 49, 50 to 69 at policy issue), the excess mortality is greatest in the younger age group, due primarily {o cardiovascular disease. Even slight and moderate overweight is associated with excess of mortality in the more severely hypertensive group. According to Lew,’ at levels of systalic blood pressure of 140 to 160 mm Hg and diastolic pressure of 90 to 100 mm Hg, overweight curtails the expectation of life among individuals who are both over- weight and hypertensive as compared to hypertensive subjects who are not overweight. However, longevity is much less affected among women than men at corresponding degrees of overweight and hypertension. Overweight and hypertension are also as- sociated with an excess of coronary heart disease and cerebrovascular disease, especially among men.:: 27 Levy and his colleagues®*-72 showed that retirement from active service and death from cardiovascular renal disease were more common among Army officers in the obese-hypertensive group than in the obese or hypertensive groups alone. Severity of hypertensive retinopathy and cardiovascu- lar complications were related to obesity in Tibblin's study of hypertensive disease in men aged 50.7 However, in other clinical series of patients with more severe hypertensive disease, Bechgaard,’? Frant,*® and Mathisen’® found mortality in obese hypertensive subjects no higher than in nonobese hypertensive persons. In fact, obese hypertensive women may have a relatively better chance of long-term survival. In a 10-year follow-up study of a small number of hypertensive patients without retinal change or only arteriolar narrowing at the Mayo Clinic, Breslin and colleagues? showed a slightly greater mortality in in- dividuals 30% overweight but in individuals with sclerosis and exudates, the prognosis was slightly better in the obese group. Clinical studies are highly selective. The difference between these studies and the life insurance company experience may be due to the fact that the actuarial study dealt with blood pressures in the range from 140 to 160/ 90 to 100 mm Hg, whereas blood pressures in the clinical investigations were higher. Blood pressures over 160/100 mm Hg apparently, overshadow the effect of overweight on mor- tality and the course of severe hypertensive disease may be allered but little by weight reduction.’ *2 However, in individuals with a moderate degree of blood pressure elevation associated with overweight, weight reduction may significantly change the course of the disease. Obese persons who are able to reduce their weight comprise a relatively small and *selected proportion of the total group, Never- theless, data from the Build and Blood Pres- sure Study? indicate a lowering of the mor- tality ratio among such men; there were too few women in the series for analysis. Among those called “moderately obese” (averaging 25% overweight), the mortality ratio was 109% in those who lost weight, as compared with a ratio of 128% in the total group; the cor- responding figures in the “markedly obese” {averaging 35 to 40% overweight) were 96% and 151%, respectively. It is difficult to extrapolate these findings to what might be the effect of weight reduc- tion in the population at large. For this reason, there is a great need for documenting the potential benefits of weight loss among the obese and obese hypertensive subjects by means of further study. However, the con- verging evidence from a large variety of sources summarized in this review strongly suggests . that the impact of weight reduction in the general population would be very considerable and lessen appreciably the burden of mortality and also morbidity from cardiovascular dis- cases. 10. 11. 12. 13. 14. . Bunp ano Broop Pressure Stuy, References . Stamvrem, J.: Lectures on Preventive Cardi- ology. New York, Grune & Stratton, Inc, 1967, p. 256. . Kanner, W. B., LeBauer, E. J, Dawses, T. R., AnD MacNamara, P. M.: Relation of body weight to development of coronary heart disease: Framingham study. Cir- culation 35: 734, 1967. . MemroroLrTAN LiFe Insurance Co., New York. Frequency of Overweight and Under- weight. Stat Bull 41: 4, 1960. 1959. Society of Actuaries, Chicago, Vols. 1 and 2, 1959. . Wate, P. L: The nutritionist. JAMA 186: 34, 1963. . NamioNaL CentER FOR HEALTH STATISTICS. Weight, Height, and Selected Body Dimen- - sions of Adults. Public Health Service, U. S. Dept. of Health, Education and Welfare, Washington, D. C., 1965, Series 11, No. 8. . NamonNar Center ror Hearts Stamstics: Weight by Height and Age of Adults, United States, 1960-1962. Public Health Service. U. S. Dept. of Health, Education, and Welfare. Washington, D. C., 1965, Series 11, No. 14. . Heaw, F. P., Hunt, E. D.,, Jr, Scuwarrz, R., Cook, C. D. Erwor, O., anp Vajpa, B.: Measures of body fat and hydration in adolescent boys. Pediatrics 31: 226, 1963. . Novak, L. P.: Age and sex differences in body " density and creatinine excretion of high school children. Ann" N Y Acad Sci 110: 545, 1963. Brozex, J, axp Kevs, A.: Evaluation of Jeanness, fatness in man: Norms and in- terrelationships. Brit J Nutr 5: 194, 1951. Pascare, L. R., Grossman, M. L, SLOANE, H. S., avo Franken, T.: Correlations be- tween thickness of skinfolds and body deansi- ty in 88 soldiers. Hum Biol 28: 165, 1956. Eperman, I. S.: Body Water and Electrolytes in Techniques for Measuring Body Composi- tion. Edited by J. Brozek and A. Henschel, Washington, D. C, National Academy of Sciences, 1961, p. 155. Forees, G. B., GaLLup, J, ano Humsg, F. B.: Estimation of total body fat from potassium- 40 content. Science 133: 101, 1961. CoynarrTEE ON NUTRITIONAL ANTHROPOMETRY, Foor ann Nutrition Boarp, Namonar Reseancy CounciL. A. Keys, Chairman. Rec- 15. 16. 17. 18. 19. 20. 21. 26. 27. 29. ommendations Concerning Body Measure- ments for the Characterization of Nutritional Status in Body Measurements for Human Nutrition. Edited by J. Brozek, Detroit, Wayne University Press, 1856, p. 10. ; Stuart, H. C, Huy, P., ano Suaw, C.: The Crowth of Bone, Muscle .and Overlying Tis- sues as Revealed by Studies of Roentgeno- grams of the Leg Area (monograph). Evans- ton, Ill, Socicty for Research in Childhood Development, Inc. 1949, Vol. 5, No. 3, Serial No. 26. Sevrzer, C. C.,, ano Maver, J.: Simple Crite- rion of obesity. Postgrad Med 38: A101, 1965. MAYER, J.: Some aspects of the problem of regulation of food intake and obesity. New Fng J Med 274: 610, 1966. Hypertension anp Coronary Heart Dis- Ease: Classification and Criteria for Epi- demiological Studies, WHO Technical Report Series No. 168, 1959. Tecumsern Covmuntry Heavra Stupy: Une published data. Epstein, F. H.,, ano Eckuory¥, R. D.: Epi- demiology of high blood pressure-geographic distributions and etiological factors, In The Epidemiology of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune & Stratton, pp. 155-166. Wayre, H. M.: Behind the adipose curtain. Amer J Cardiol 15: 66, 1965. . Wuvre, H. M.: Blood pressure and obesity. Circulation 19: 511, 1959. . Suarer, A. G.: Blood pressure studies in East Africa. In The Epidemiology of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune & Stratton, 1967, pp. 139-145. . Keys, A., er aL.: Epidemiological studies related . to coronary heart disease: Characteristics of men age 40-59 in seven countries. Acta Med Scand (suppl. 460), 1967. . Parmay, G.: Skin fold thickness in relation to body weight and arterial bloed pressure. Med J Aust 49: 13, 1962. EpsTEIN, F. H., £7 AL.: Prevalence of chronic dis- ease and distribution of selected physiclogical variables in a total community, Tecumseh, Michigan. Amer J Epidem 81: 307, 1965. Kanner, W. B, Brawe, N, Sinner, J. J, Dawser, T. RB, anp McNamara, P. M.: Re- lation of adiposity to blood pressure and de- velopment of hypertension: The Framingham study. Ann Intern Med 67: 48, 1967. . BjErkEpaL, T.: Overweight and hyperiens on. Acta Med Scand 159: 13, 1957. Bok, J., HumerserT, S., axp Webenvane, G.: Blood pressure in a population: Blood pres- 30. 31. 32. 34. Tiss 36. 38. 39. 40. 41. 42. . ALExanper, J. K,, sure readings and height and weight deter- minations in the adult population of the City of Bergen. Acta Med Scand (suppl 321), 1957. Bovig, E., Jr, Currey, W. P., Nicuaman, M. Z:, axp Taunent, C. R., Ja.: Epidemiological study of hypertension among racial groups in Charleston County, South Carolina. The Charleston Heart Study, Phase II, In The Epidemiology of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune & Stratton, 1967, pp. 193- 203. McDoxroucr, J. R., Garnison, G. E. ann Hames, C. G.: Blood pressure and hyper- tensive disease among Negroes and whites in Evans County, Georgia. In The Epidemiol- ogy of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune & Stratton, 1967, pp. 167-187. Murr, W. E,, Ber, R. A, anp Lover, H. G.: Relation between change in blood pressure and weight. Brit J Prev Soc Med 22: 73, 1968. . Rorpinson, S. C., BRucer, M., anp Mass, I.: Hypertension and obesity. J Lab Clin Med 25: 807, 1939. 2 TruepssoN, E.: Variation of arterial blood pres- sure and serum cholesterol levels in coconut- eating Polynesians. Fed Proc {suppl 11) 21: 36, 1962. Boynton, R. E., anp Topp, R. L.: Relation of body weight and family history of hyperten- sive disease to blood pressure levels in uni- versity students. Amer J Med Sci 216: 397, . 1948. Tuomas, C. B.,, ano Conen, B. H.: Familial occurrence of hypertension and coronary artery disease; with observations woncerning 53 obesity and diabetes, Ann Intern Med 42: 90, .. 1955. - : 3 ano Dennis, E. W.: Cir- culatory dynamics in extreme obesity. Circu- lation 20: 662, 1959. ALEXANDER, J. K.: Obesity and the circulation. Mod Conc Cardiovasc Dis 32: 799, 1963. Harrrman, H. R.,, anp Guwst, D. G.: Blood pressure and weight. Arch Intern Med 44: 877, 1929. Rem, D. O., Horrano, W. W. HumerreLT, S., anp Rose, GC. A.: Cardiovascular survey of British postal workers. Lancet 1: 614, 1966. Tispuin, G.: High blood pressure in men aged 50. Acta Med Scand (suppl 470), 1967. Chane, B. N, Ting, N, Ly, Y. B, Pao, Y. L,, Avexanper, E. R, Bruce, R. A, axp Cray- ston, J. T.: Maximal exercise responses in middle aged hypertensive and normotensive 43. 44. 45. 46. 47. 48. 49. 51. 52. 55. Chinese males; Preliminary Med J 14: 239, 1967. Jonnson, B. C., ano Rerincron, R. D.: Sam- pling study of blood pressure levels in white and Negro residents of Nassau, Bahamas, J Chron Dis 13: 39, 1961. LariMORE, J. W.: Study of blood pressure in relation to type of body habitus. Arch Intern Med 31: 567, 1923. : Manxs, H. H.: Influence of obesity on mor- bidity and mortality. Bull N Y Acad Med 36: 296, 1960. Hupen, E. G.: Systolic blood pressure of healthy adults in relation to body weight. JAMA 88: 1554, 1927. SuorT, J. J, Anp Jounson, H. J.: Evaluation of the influence of overweight on blood pres- sure of healthy men. Amer J Med Sci 198: 220, 1939. CnreeN, M. B., ann BeckMaN, M.: Obesity and hypertension. New York J] Med 48: 1250, 1948. Master, A. M., Jarre, H. L., AnD Cuesxy, K.: Relationship of obesity to coronary dis- ease and bypertension. JAMA 153: 1499, 1953. report. Chin . Papmavaty, S., AnD Gupta, S.: Blood pressure studies in rural and urban groups in Delhi. Circulation 19: 395, 1959. StamLER, ]., Berkson, D. M., Linoserg, H. A, MiLLer, W., axp Harv, Y.: Racial patterns of coronary heart disease: Blood pressure, body weight and serum cholestern] in whites and Negroes. Geriatrics 16: 382, 1961. NutrmoN Survey oF THE ARMED FORCES; A Report by Interdepzrtmental Committee on Nutrition for National Defense. Washington 25, D. C, U. 8S. Gov't Printing Office. Thai- Jand (1960); Republic of China (1960); Ethiopia (19858); Philippines (1957); Uru- guay (1963); Alaska (1959); Chile (1960). . Dovie, A. E, axp LoveLy, R. R. H.: Blood pressure and body build in men in tropical and temperate Australia. Clin Sci 20: 243, 1961. . ALexsanprow, D.: Studies on the epidemiol- ogy of hypertension in Poland. In The Epi- demiology of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune & Stratton, 1967, pp. 82- 97. Hunter, J. D.: Diet, body build, blood pressure and serum cholesterol levels in coconut-eating Polynesians. Fed Proc (suppl 11) 21: 36, 1962. . Racan, C, anp Bororey, J., Ja.: Accuracy of clinical measurements of arterial blood pres- sure. Bull Johns Hopkins Hosp 69: 504, 1941. S58. 59. 60. 61. 62. 70. 71. 72. . Rose, GC. A, HoLrLanp, W. W,, . Pickening, G. W., RoBeRrTs, J. A. F., AND Sow- ry, G. S. C.: Etiology of essential hyperten- sion: Effect of correcting for arm circumfer- ence on the growth rate of arterial pressure with age. Clin Sci 13: 267, 1954. Mia, W. E., axp Orpnam, P. D.: Study of arterial blood pressure and its inheritance in a sample of the general population. Clin Sci 14: 459, 1955. Rarreny, E. B., axp Warp, A. P.: Indirect method of recording blood pressure. Cardio- vasc Res 2: 210, 1968. Horrano, W. W., axp HuMERFELT, S.: Mea- surement of blood pressure: Comparison of intra-arterial and cuff values. Brit Med J 2: 1241, 1964. Berunen, K., Fujry, H.,, Leg, D. H, Ynorz, M., AnD Garnier, G.: Blood pressure mea- surements in obese persons: Comparison of intra-arterial and auscultalory measurements. Amer J Cardiol 8: 10, 1961. Karierors, T., Nmsex, R.,, axp WestLing, H.: On the accuracy of indirect auscultatory blood pressure measurements during exercise. Acta Med Scand (suppl. 449): 81, 1966. . Nacrg, F. J.: Comparisons of direct and in- direct blood pressure with pressure flow dy- namics during exercise. J Appl Physiol 21: 317, 1966. . Knosra, T., ano Lowe, C. R.: Arterial pressure and am circumference. Brit J Prev Soc Med 19: 159, 1965. . Lowe, C. R.: Arterial pressure, physique and occupation. Brit J Prev Soc Med 18: 115, 1964. . KARvoNEN, M. ].: Effect of sphygmomanometer cuff size on blood pressure measurement, Bull WHO 27: 805, 1962. King, G. E.: Errors in clinical measurement of blood pressure. Clin Sci 32: 223, 1967. AND CrowLrEey, E. A.: Sphygmomanometer for epidemiolo- gists. Lancet 1: 296, 1964. . Levy, R. L., Warrg, P. D., Stroup, W. D., AnD Hiriman, C. C.: Overweight: Prognostic significance in relation to hypertension and cardiovascular-renal diseases. JAMA 131: 951, 1946. Levy, R. L., Warrg, P. D,, Stnoun, W. D,, anp Hrinan, C. C.: Transient hyperten- sion: Relative prognostic importance of vari- ous systolic and diastolic Jevels. JAMA 128: 1059, 1945. Levy, R. L., Wate, P. D., Stroup, W. D., AND Human, C. C.: Transient tachycardia: Prognostic significance alone and in associa- tion with transient hypertension. JAMA 129: 585, 1945. STAMLER, J.: On the natural history and epi- 73. 74. 75. 76. 1. 78. 79. 81. 82. 85. 86. demiology of hyperlensive disease. In The Pathogenesis of Essential Hypertension. Edited by J. H. Cort, V. Fencl, Z. Hejl, and E. Jirka. Proc of the Prague Symposium, Prague, Czechoslovakia: State Medical Pub- lishing House, 1960, pp. 67, 107. Kanner, W. B,, Dawper, T. R., ano McNa- Mama, P. M.: Vascular disease of the brain —Epidemiologic aspects: Framingham study. Amer J] Pub Health 55: 1355, 1965. Woop, J. E., anp Cass, J. R.: Obesity and hypertension: Clinical and experimental ob- servations. Ann Intern Med 13: 81, 1939. WineLng, C. M.,, Wiabpmann, E. B., ano McGume, T. F.: Basal blood pressure of nor- mal dogs determined by an auscultatory method and a study of the effect of fasting. Amer J Physiol 166: 296, 1951. - Brozex, ]., Cuapman, C. B, ano Keys, A: Drastic food restriction: Effect on cardiovas- cular dynamics in normotensive and hyper- tensive conditions. JAMA 137: 1569, 1948. Keys, A., Henscuer, A., anp Tavior, H.: Size and function of human heart ai rest in semi- starvation and in subsequent rehabilitation. Amer J Physiol 150: 153, 1947. Master, A. M., AND OPPENHEIMER, A.: Study of obesity: Circulatory, roenigen-ray and elec- trocardiographic investigations. JAMA 92: 1652, 1929. Martin, L.: Effect of weight reduction on nor- mal and raised blood pressure in obesity. Lancet 2: 1051, 1952. . Sarzawo, ]. V.,, Gunning, R. V., MastorauLo, T. N., axp Turrre, W. W.: Efect of weight Joss on blood pressure. J Amer Diet Ass 34: 1309, 1958. Frercuer, A. P.: Effect of weight reduction upon blood pressure of obese hypertensive women. Quart J] Med 23: 331, 1954. Evans, F. A, AnD Stranc, J. M.: Treatment of obesity with low calorie diet JAMA 97: 1063, 1931. . Aprersserc, D., Corcuer, H. R., anp Lavar, J.: Effect of weight reduction on course of arterial hyperlension. J Mount Sinai Hosp NY 12: 984, 1946. . Dany, L. K., SiLver, L., axp Corre. R. W.: Role of salt in the fall of blood pressure ac- companying reduction of obesity. New Eng J Med 258: 1186, 1958. Kexspner, W.: Treatment of heart and kidney disease and of hypertensive and arterio- sclerotic vascular disease with rice diet. Ana Intern Med 31: 821, 1949. Bexepicr, F. G., Axo Romy, P.: Effect of a pro- longed reduction in diet on twenty-five men. Proc Nat Acad Sci 4: 149, 1918. 87. 88. 89. 90. 91. 92. 93. 95. 96. 97. Puence, W. E.: Obesity: Observations on one thousand’ cases. Boston Med Surg J 188: 617, 1923, Terry, A. H, In: JAMA 81: 1283, 1923. Bauman, L.: Obesity: Recent reports in the literature and results of treatment. JAMA 906: 22, 1928. Ferrows, H. A.: Studies of relatively normal obese individuals during and after dietary restrictions. Amer J Med Sci 181: 301, 1931. Owson, R. E.: Obesity as a nutritional disorder. Fed Proc 18: (Part I) 58, 1959. Procer, S. H., ano Dexnic, H.: Study of the circulation in obesity. J Clin Invest 11: 789, 1932. Tavror, H. L., Brozeg, J., anp Keys, A.: Basal cardiac function and body composition with special reference to obesity. J Clin Invest 31: 976, 1952. . Moris, J. N., axp Crawrorp, M. D.: Coronary heart discase and physical activity of work: Evidence of a national necropsy survey. Brit Med J 2: 1485, 1958. Hartrorr, W. S.: Pathology of obesity. Bull NY Acad Med 36: 313, 1960. Conn, J. W.: Hypertension, the potassium ion and impaired carbohydrate tolerance. New Eng J Med 273: 1135, 1965. ScureENcArT, D. E.: Characteristics of in- creased adrenocortical function observed in ‘many obese persons. Ann N Y Acad Sci 131: 388, 1965. Obesity and hypertension. 55 98. 99. 100. 101. 102. 104. 105. Scuteincant, D. E., CREGERMAN, R. I, AnD Conn, J. W.: Comparison of the characleris- tics of increased adrenocortical function in obesity and Cushing's syndrome. Metabolism 12: 484, 1963. Samprr, M. P., Wen, M. H.,, ano Ubnoj, « V. M.: Acute pharmacodynamic effects of glucocorticoids: Cardiac output and related hemodynamic changes in normal subjects and patients in shock. Circulation 31: 523, 1965. Ennucy, Epwarp: Aldosterone, the adrenal cortex and hypertension. Ann Rev Med 19: 373, 1968. Lew, E. A.: Blood pressure and mortality— Life insurance experience. In The Epidemiol- ogy of Hypertension. Edited by J. Stamler, R. Stamler, and T. N. Pullman. New York, Grune and Stratton, 1967, pp. 392-397. BecHcaarp, P.: Natural history of benign hy- pertension. In Essential Hypertension: An In- ternational Symposium. Edited by F. C. Reubi, K. D. Bock, and P. T. Cattier. Berlin, Springer-Verlag, 1960, p. 198. . FranT, F., AnD GROEN, J.: Prognosis of vascular hypertension: Ninc-year follow-up study of four hundred and eighteen cases. Arch Intern Med 85: 727, 1950. Mammsen, H. S., Jexsen, D., Loken, E., ano Loxen, H.: Prognosis of essential hyperten- sion. Amer Heart J 57: 371, 1959. BresLin, D. J., CGirroro, R. W., Jr., AND Fam- BARNES, J. F., II: Essential hypertension: A twenty year follow-up study. Circulation 33: 87, 1966. : 57 Obesity and Diabetes MORTON SMITH, M.D., FACP.* RACHMIEL LEVINE, M.D., F.A.C.P.** The aggregation of subjects with diabetes mellitus in families is an observation supported by a world-wide experience spanning many decades. However, the rarity of overt diabetes in infancy and its relative infrequency in childhood and adolescence suggest environmental factors to be operative. Equally as well appreciated as the hereditary nature of diabetes is the fact that obesity and diabetes are closely linked. Diabetes is diagnosed chiefly after the age of 40, accompanied in approximately 80 per cent of the cases by a present or past history of being overweight. Probably half of this number are substantially fat, i.e., in excess of 20 per cent above ideal body weight, and some of these may be massive.!®. 3 Diabetic children are seldom obese, the incidence being on the order of 5 per cent.” THE NATURE OF THE OBESITY-DIABETES RELATIONSHIP The nature of the kinship between obesity and diabetes is poorly understood. Obviously, not all of the obese develop diabetes. Older studies disclosed an abnormality in glucose tolerance with considerable regularity in the corpulent. An incidence of 50 to 75 per cent was found, depending on the age of the subject.®: 3. 39.58 Some current studies do not present quite as gloomy a statistical outlook. The frequency was closer to 10 per + cent, although the subjects in these investigations belonged to rather restricted population groups and may very well not be characteristic of other obese people.” 2? Nevertheless, the latter figure would seem more likely considering the fact that the overall incidence of diabetes in this country is approximately 2 per cent. Obesity is a considerably more common From the New York Medical College, Flower and Fifth Avenue Hospitals, Metropolitan * Hospital Center, New York, N. Y. * Associate Professor of Medicine ss Professor of Medicine and Chairman, Department of Medicine 58 finding in the United States, up to 30 per cent of the inhabitants being 10 per cent or more overweight.*® ITowever, a very recent investigation dealing with 100 obese men and women, each at least 25 per cent over standard weight, found the fasting blood sugar to be elevated in 22 per cent. The oral glucose tolerance test was impaired in 58 per cent. Of interest, the intravenous tolbutamide test was normal in 30 per cent of the latter group.® The Influence of Sex Five hundred unselected obese patients composed of approximately an equal number of males and females having oral glucose tolerance tests were investigated by Iimbleton.?® He employed a somewhat different technique and analyzed the curves in such a manner that comparison with the results of others is difficult. Nevertheless, there emerges from this study an interesting sidelight on the influence of sex on obesity. Up to the age of 35 “high” glucose tolerance tests occur with almost equal fre- quency in the two sexes, 25 per cent. Then, there is a very steep rise in the incidence of this type of curve in the male, and by the age of 45 about 80 per cent show this characteristic. Among the females, the rise in fre- quency occurs a decade or more later and also at a gentler rate to a peak of 70 per cent at age 65. In both sexes the percentage of “‘high’’ tests bears no relationship to the degree of obesity. Only 12 per cent of the males and 2 per cent of the feinales had a definitely characteristic diabetic curve. Morse et al.*® studied females who were at least 30 per cent above ideal weight and with obesity of at least 10 years’ duration. There was a negative personal and family history of diabetes. The oral glucose tolerance test was impaired and suggestive of diabetes. The intravenous tolerance test, though, revealed a normal net glucose removal rate and a normal rise in blood pyruvate. Moreover, the galactose removal rate after an intra- venous infusion was increased in this obese group. However, Dische and co-workers" had noted the oral galactose tolerance test to be relatively flat and considered this to be secondary to poor absorption of galactose in the obese. In view of the above finding of Morse et al.,*® accelerated hepatic uptake of galactose may also explain the flat oral curve. Further- more, this group expressed the view that the oral glucose tolerance test might not be reliable in the obese. However, Amatuzio et al.* found a reduced rate of removal of intravenous glucose in the obese. This conflict may be explained by the latter's use of a smaller sugar load as well as a different method of calculating the specific rate constant. The Influence of Family History Another dimension was added to the diabetes-obesity relationship almost 40 years ago by Allison, who suggested that the duration of the obese state might be of importance.? This matter was studied more exten- sively by Ogilvie in 1935. He found that an impaired carbohydrate 59 tolerance was related to the duration of obesity and not to its degree. Moreover, occasionally during the first five years of obesity there appeared to be a phase of increased glucose tolerance. Recent publications on the “active” and “static” phases of obesity support the above. ? Nevertheless, it is also possible that this relationship is not causal, that the obese who become diabetic carry an inherited susceptibility to diabetes. The data of Tyner who studied “prediabetes” over 30 years ago supports such a contention.®® Ile found that ‘‘prediabetes” was more common in the obese, only when there was a positive family history of diabetes. German?’ investigated this issue by the more stressful technique of the cortisone-glucose tolerance test of Fajans and Conn.?? Obese and nonobese healthy male prisoners with a negative family history for diabetes were studied by both the standard and the cortisone-glucose tolerance tests. The average two-hour precortisone blood sugar level (I"olin-Wu) was 109.5 mg. per 100 ml. for the nonobese and 124.5 mg. for the obese, appar- ently not a significant difference. I'ollowing cortisone priming, the two-hour values were 139.4 mg. and 136.2 mg. per 100 ml. respectively. This reversal in the response was significant and demonstrated that the difference in carbohydrate tolerance between the two groups could be eliminated by cortisone. Recalling that I'ajans and Conn? had noted a marked deterioration in glucose tolerance after cortisonc in a high proportion of subjects with a close family history of diabetes, German’s?’ results do not support the concept that obesity represents a trait accompanying diabetes in which one or more neighboring genes may be responsible for both conditions. The data do suggest that obesity in itself is not necessarily diabetogenic, that it must be engrafted upon the proper genetic soil. It should be mentioned, though, that the average age of the subjects in the above study was only about 33 years. The duration of obesity was probably brief, this tending to decrease the frequency of any abnormality in glucose tolerance. Ou the other hand, a history of being overweight for many years might imply an older subject, and there is some evidence that carbohydrate tolerance decreases with advancing age regardless of body weight. Newburgh and Conn studied four obese glycosuric patients in a respi- ration chamber by indirect calorimetry. ® These subjects were found capable of oxidizing glucose as well as normals. The authors also had spectacular success in restoring normal carbohydrate tolerance to the obese by weight reduction. They speculated that the obese diabetic might repre- sent a particular variety of diabetes caused by an impaired hepatic storage of glycogen. They felt that the obesity was primary and hyperglycemia a secondary phenomenon which might even be “normal” considering its great frequency. The studies of Richardson, ® however, demonstrated that mild diabetes with obesity and diabetes requiring insulin were most likely etiologically related. He found the incidence of diabetes in the relatives of the above 34-621 0 - 74-5 60 two groups to be the same. Furthermore, both types of diabetes occurred in the same (amily. Moreover, the obese female diabetics in this study frequently gave birth to children having a birth weight in excess of 10 pounds. Finally, the incidence of some of the so-called “complications” of diabetes—polyneuritis, retinopathy and cataracts—was similar in the two groups. Murray and Wang likewise compared obese and nonobese classes of diabetic subjects, and their findings support Richardson’s conclusion. The family history of diabetes in the two groups was approximately the same, 26 per cent. There was a tendency to inherit the same type of diabetes, but with notable exceptions. Approximately 12 per cent of obese diabetics had nonobese relatives with diabetes, as compared to 19.2 per cent of the nonobese diabetic group. Approximately 17 per cent of obese patients also bad obese diabetic relatives, whereas 6.6 per cent of the nonobese group had obese diabetic relatives. : I'ajans and Conn? express similar views and feel that obesity, per se, is not to be construed as a prediabetic state. Obesity diminishes carbohy- drate tolerance generally only in those with a family history of diabetes. Approximately 85 per cent of obese diabetic patients whose glucose toler- ance test had returned to normal after weight reduction gave a positive response to the cortisone-glucose tolerance test. Considering now the factor of heredity in obesity, there exist a number of studies supporting such an association. Fellows? studied 294 overweight insurance company cmployees wishing to lose weight. Their parents showed an incidence of obesity ten times that of the normals. Both parents were obese in 24 per cent of the cases. Of additional interest was the frequency of diabetes in the parents of the obese. This was twice that of the control population. Murray and Wang® found additional evidence. A family his- tory of obesity was three times more common among the obese diabetic group. All would agree, though, that definite proof that human obesity is inherited is still lacking. The separation of the genetic from common environmental factors is most difficult. Experimental Production of the Obese-Diabetic State One of the difficulties in analyzing the importance of overeating on the emer- gence of overt diabetes has been the problem of reproducing the phenomenon experimentally. Allen et al. accomplished this to some extent by fattening partially depancreatized dogs. Ranson et al.® observed diabetes in a monkey within a year of being made hyperphagic through destruction of the ventromedial hypothalamic nucleii. Brobeck et al.’ # placed bilateral hypothalamic lesions in nondiabetic partially depancreatized rats. Glucosuria appeared with the hyperphagia and in one animal this became progressively severe. In one rat spontaneous glucosuria was seen after hyperphagia, without pancreatectomy. Dohan and Lukens'® pro- duced permanent diabetes in only one of 27 normal cats stressed with enormous quantities of glucose. Even among partially pancreatectomized cats only two of eight developed diabetes. More recently, Hamilton and Brobeck?® observed diabetes in three of seven 61 hyperphagic obese monkeys with hypothalamic lesions. Of great interest was the appearance of mild hyperglycemia in a 5-year-old normal control monkey who had been caged, with little activity and was moderately obese. IXrook et al." in a study of spontaneous diabetes of dogs finds an interrelation- ship with obesity as close as that in man. Approximately two-thirds of the spou- taneously diabetic dogs were obese. Stated in another perspective, 8 per cent of the obese dogs developed diabetes while the incidence of diabetes among the normal or underweight animals was less than one-tenth of this figure. Meier'" describes a diabetes-like condition in feed-lot sheep. This has caused a loss of animals in feed lots where the rapid method of fattening is practiced. There is hyperglycemia, glucosuria, and finally coma prior to death. Insulin .corrects the situation. A decrease in the carbohydrate content of the diet can prevent the syndrome. The wholesale appearance of severe spontaneous diabetes was recently reported by Schmidt-Nielsen and co-workers.® A North African rodent, the sand rat, when fed the routine commercial laboratory rat chow instead of its usual vegetable diet, overeats and becomes markedly obese. In addition to hyperglycemia, ketonuria and cataracts develop. The beta cells show prominent degranulation. Spontaneous diabetes in another rodent is known, the Chinese hamster. This, however, is un- related to obesity. It is apparent then that in some species, overeating, depending on circumstances, may bring about a permanent diabetic state. There is, also, a very well studied hereditary obese hyperglycemic syndrome in the mouse. Response of the Obese Nondiabetic and Obese Diabetic to Certain Stimuli A number of interesting similarities have been found in the response of the obese nondiabetic and the obese diabetic to some stimuli. By the specific immunochemical method, Yalow and Berson® had demonstrated that the early maturity-onset diabetic subjects had an excessive and more sustained elevation of plasma insulin at one and two hours after glucose administration than did a nondiabetic group. Subsequently, I{aram et al.?® studied ten obese nondiabetic subjects with a negative family history of diabetes using a similar assay technique. They found that nine of these responded to an intravenous load with a prolonged and greater elevation of plasma insulin than did nonobese, nondiabetic subjects. The authors concluded that the obese may be in a compensating phase which requires that an increased amount of insulin be secreted in order to establish glucose homeostasis. Ogilvie’s® observation of many years ago supports this notion. Ie found that 68 per cent of 19 obese, nondiabetic and mostly elderly females had an abnormally high percentage area of pancreatic islet tissue and islands of greater than normal volume. Hyperplasia and hypertrophy of islet tissue are also found in the hereditarily obese hyperglycemic mouse and in the yellow-obese mouse. % Knowledge of the response of the obese nondiabetic to exogenous insulin might prove useful in an analysis of the above phenomenon as well as of the etiology of cbesity. The literature, though, is quite unsettied on this point. I{aram et al.?® and Olsen and Nuetzel® considered the response of the obese to be normal as judged by the percent fall following 0.025 units of insulin per kilogram and 0.05 units of insulin per kilogram intravenously 62 respectively. Godlowski, utilizing an oral glucose load and insulin simul- taneously, also found a normal reaction.?® Arendt and Pattec® and Rabinowitz and Zierler® concluded that over- weight, subjects were resistant to exogenous insulin, The former employed a standard insulin-glucose tolerance test and found the blood sugar of the obese still to be elevated at two and three hours. The latter studied forearm metabolism of control and obese subjects and their responses to small and only locally effective doses of intra-arterial insulin. After an overnight fast under basal conditions, the obese subject exhibited a greater glucose uptake than the normal weight control with appropriate changes in potassium uptake and free fatty acid release suggestive of an endogenous hyperinsulin state. I'ollowing an intra-arterial infusion of insulin, the obese forearm responded significantly less well in respect to glucose and potassium uptakes and free fatty acid release. This decreased reaction to insulin could be a manifestation of tolerance to a chronic load of endogenous insulin which in itsell was adaptive lo unusual quantities of substrate. Insulin resistance disappeared in one subject afer weight loss. It was felt that the insulin resistance was a consequence of obesity and not a primary defect. Himsworth® found that the maturity-onset diabetic patient was insulin-resistant. Such a patient is also {requently obese. How much of this resistance is a feature of the obesity rather than the diabetic state? Increased Cortisol Production in the Obese Another factor to be considered is the increased cortisol production found in the obese.® 47: # As in Cushing's syndrome, highly colored striae may be seen. Unlike Cushing's syndrome, though, a normal hypothalamic- pituitary-adrenocortical feedback is maintained and cortisol production can be inhibited by standard quantities of exogenous glucocorticoids.®: Since the classic studies of Long et al.* the adrenocortical hormone has been known to increase liver glycogen and the blood sugar. This was at first attributed solely to increased gluconeogenesis. Subsequent studies indicated that the adrenal steroids may decrease glucose utilization in a number of tissues. Munck et al. found that physiological quantities of glucocorticoids inhibit the glucose uptake of the rat epididymal fat pad*® and the thymus.? Overcll and co-workers®” noted the same response by the skin. The glucose uptakes of liver and muscle were not impaired.? THE PENALTY OF BEING OVERWEIGHT There remains to be assessed the plight of the overweight, their pen- alty. The recently completed studies by the Society of Actuaries supply the data on the drastic consequences. ¢ These were compiled through an analysis of the cxperience in nearly five million persons accepted for life insurance during 1935 to 1953, and traced to the policy anniversary in 63 1954. Among insured males between ages 15 and 69, the death rate for those 10, 20 and 30 per cent above average weight was 13, 25 and 42 per cent, respectively, in excess of the mortality of standard risks. Women tolerated their avoirdupois somewhat better, the excess mortality for those 10, 20 and 30 per cent overweight being 9, 21 and 30 per cent, respectively. The obese experienced an excess mortality from many disorders, but this was relatively greatest from diabetes. The statistics worsened as the weight increased. Among men 20 per cent or more overweight, the excess mortality from diabetes was 133 per cent greater than that among all persons insured as standard risks. Men 35 per cent overweight had a mortality from diabetes three times that of the standard group. In the case of women 15 per cent or more overweight, the excess mortality was 83 per cent over that among women insured as standard risks. Actually, in absolute terms, the number of obese dying with a diagnosis of diabetes is quite small compared to deaths due to cardiac, cerebral and renal pathology. Hypertension also was associated with an increase in mortality from diabetes among men, but this was not as marked as in the case of obesity. From the same investigation emerged additional evidence that weight reduction has a favorable effect on longevity of overweight individuals. Among overweight men who reduced, mortality declined to that of the standard risks for a period of years following the weight loss. The mortality record was most favorable in the years immediately following the weight reduction. A less detailed study was available for women because the numbers were smaller. However, the data pointed to an improved trend. Similar conclusions were reached in an earlier study by Dublin and Marks.!? TREATMENT Diet with Weight Reduction It is obvious that the treatment of the obese diabetic starts with weight reduction. If weight reduction is achieved, overt diabetes may certainly improve, if not disappear. 36. 81 Twenty-one of 22 obese diabetic patients who lost all of their excess weight subsequently attained a normal oral glucose tolerance test. A long-term follow-up of such a diabetic population would be of great importance. The almost 100 per cent success- ful conversion rate reported above by Newburgh and Conn, while not characteristic of all similar studies is, nevertheless, a good example of what can be accomplished. There are some fascinating and unexplained aspects of this phenom- enon—the improvement of diabetes accompanying weight reduction. A perfect example is supplied by the Newburgh and Conn study. Some patients apparently become normal before all of the excess weight is lost. The two patients showing the earliest return to normal lost 46 and 52 per 64 cent of their excess weight, or 16 and 17 kg. respectively. Yet, another subject lost 80 per cent of the excess or 23 kg. and was still abnormal. With the loss of the remaining 20 per cent, glucose tolerance reverted to normal. In the study of I'etler et al,” a drop in weight of as little as 4 kg. resulted in improvement although the subjects were not studied in a detailed manner. There appears to be little or no information of a factual nature to explain the rate or the ease with which improvement occurs, or why, in any event. Of some practical clinical importance is the appearance of edema and a decline in the progression of weight loss. The patient may become discouraged by the apparent failure to lose weight while dieting faithfully. After careful elimination of other causes of edema, diuretic therapy may be indicated. There is increasing interest, experience and success in the utilization of total starvation or intermittent fasts in the treatment of intractable obesity, in both the diabetic and nondiabetic.!?- 2. 1 When carefully super- vised this approach appears to be safe, although there are some hazards. The study of Drenick and co-workers points out the occurrence of serious hypotension and precipitation of gout in some.!® The serum uric acid remains quite elevated during complete starvation. Considering the impor- tance of weight reduction, especially to the diabetic, and its demonstrated benefits, anorexogenic agents, when not contraindicated, could be added to dietary regimens. While the number of successes may not be large or permanent, such drugs have been known to be helpful in the obesity- diabetes clinic.2f: 5 Drug Therapy Insulin therapy in the obese diabetic generally should be reserved only for the complicated case, with infection or ketoacidosis. Routine insulin use can only increase lipogenesis and weight gain which are already excessive and harmful. One can speculate in a similar manner about the sulfonylureas. These agents stimulate the beta cell to release more insulin, an endogenous increase rather than exogenous as in the first example. IHowever, the temporary use of insulin or tolbutamide is described to great advantage by Bloom in the therapy of 550 new diabetic patients, 85 per cent of whom were older than age 40." In 145 of these subjects, when initial dictary measures were inadequate to control symptoms, insulin or tol- butamide was prescribed briefly, usually for a six-week period. Eventually, all of this group were controlled by dict and ultimately 60 per cent of the total were successfully treated by diet only. Of interest is a recent finding of Grodsky et al.?® that phenformin was capable of reducing the excessive insulin levels after glucose loading in obese diabetic and nondiabetic sub- jects. This would appear to be an attractive feature if the original observa- tion can be confirmed and extended in a larger number of obese diabetic patients. Nevertheless, diet with weight reduction is the foundation of therapy in this segment of the diabetic population. 10. 11. 12. 13. 14. 15. 16. 17. i9. 20. 21. 22, 23. 24. 25. 26. 27. 28. 29. 30. 65 REFERENCES . Adlersberg, D.: Obesity, fat metabolism and diabetes. Diabetes 7: 236, 11158. . Alien, I. M., Stillman, B. and Fitz, R.: Total dietary regulation in the treatment of diabetes. Rockefeller Inst. for Med. Res., New York, Monogr. 11, 1914, jp. 5498. . Allison, R. S.: Carbohydrate tolerance in overweight and obesity. Lancet {: 537, 1927. . Amatuzio, D. S., Rames, E. D. and Nesbitt, S.: The practical application of the rapid intravenous glucose tolerance test in various disease states affecting glucose metabolism. J. Lab. & Clin. Med. 48: 714, 19506. . Arendt, IB. C. and Pattee, C. J.: Studies on obesity. J. Clin. Endocrinol. & Metab. 16: 367, 1956. . Baird, I. M.: Urinary corticosteroid excretion in obese adults. Lancet 2: 1022, 1963. . Balter, A. M.: Glucose tolerance curves in neuropsychiatric patients. Diabetes 10: 100, 1961. . Bartlett, D., Monta, Y. and Munck, A.: Rapid inhibition by cortisol of incorpora- tion of glucose in vivo into the thymus of the rat. Nature 196: 897, 1962. ; Beaudoin, RR. Van Italie, T. B. and Save J.: Carbohydrate metabolism in “aetive’’ and “‘static’”’ human obesity. J. Clin. Nutrition 1: 91, 1953. Berkowitz, D.: Metabolic changes associated with shesity before and after weight reduction. J.A.M.A. 187: 399, 19064. Bloom, A.: Remission in diabetes. Brit. M. J. 2: 731, 1059. Bloom, W. L.: Fasting as an introduction to the treatment of obesity. Metabolism 8: 214, 1959. Brobeck, J. R., Tepperman, J. and Long, C. N. H.: The effect of experimental obesity upon carbohydrate metabolism. Yale J. Biol. & Med. 16: 893, 1943. Dische, S., Stamm, W. P. and Goudie, R. B.: Galactose and glucose tolcrance in normals, the obese, and postgastrectomy patients. J. Clin. Path. 11: 350, 1958. Dohan, F. C. and Lukens, I. D. W.: Experimental diabetes produced by the admin- istration of glucose. Endocrinology 42: 244, 1948. Drenick, E. J., Swendseid, M. I., Blohd, W. H. and Tuttle, S. G.: Prolonged starva- tion as treatment [or severe obesity. J.A.M.A. 187: 100, 1964. Dublin, L. I. and Marks, H. H.: Mortality among insured overweights in recent years. Tr. A. Life Insurance Direct. Am. 35: 235, 1051. 18. Duncan, G. G.: Early clinical picture of diabetes. In Diabetes (R. H. Williams, Ed.). New York, P. B. Hoeber, 1960, p. 370. Duncan, G. G., Jenson, W. IX., Cristofori, F. C. and Schless, G. L.: Intermittent fasts in the correction and control of intractable obesity. Am. J. M. Sec. 245: 515, 1063. Embleton, D.: Glucose tolerance curves in 500 obese cases. Brit. M. J. 2: 739, 1938. Fajans, S. S.: Diagnostic test for diabetes mellitus. /n Diabetes (R. H. Williams, Ed.). New York, P. B. Hoeber, 1960, p. 408. Fajans, S. S. and Conn, J. W.: An approach to the prediction of diabetes mellitus by modifications of the glucose tolerance test with cortisone. Diabetes 3: 296, 1954. Fajans, S. S. and Conn, J. W.: Comments on the cortisone-glucose tolerance test. Diabetes 10: 63, 1961. Fellows, H. H.: Studies of relatively normal obese individuals during and after dietary restrictions. Am. J. M. Sc. 181: 301, 1931. Fetter, F., Durkin, J. I{. and Duncan, G. G.: Dietary versus insulin treatment of the obese diabetic patient. Am. J. M. Sc. 195: 781, 1938. Fineberg, S. I{.: The obesity-diabetes clinic. J.A.M.A. 181: 862, 1962. German, J. L.: The glucose tolerance test after cortisone administration in obese and nonobese men. Diabetes 7: 261, 1958. Godlowski, Z.: Carbohydrate metabolism in obesity. Edinburgh M. J. §3: 574, 1946. Grodsky, G. M., Karam, J. H., Pavlatos, F. C. and Forsham, P.: Reduction by phenformin of excessive insulin levels after glucose loading in obese and diabetic subjects. Metabolism 12: 278, 1963. Hamilton, C. L. and Brobeck, J. R.: Diabetes mellitus in hyperphagic monkeys. Endocrinology 79: 512, 1963. Obesity and Diabetes: The Odd Couple™* EpwiN L. BierMAN, M.p.2 Joun D. BAGDADE, M.».,% AND Daniel PORTE, JR., M.D.S BESITY AND DIABETES are intimately re- O lated, but they make strange bed- fellows. It is well known that adult dia- betics are often obese (1) and, alternatively, obese subjects are likely to have carbo- hydrate intolerance. Despite this very obvious clinical association, the nature of their interrelationship is poorly under- stood. Does long-standing obesity lead to carbohydrate intolerance (which need not /be a necessary or sufficient concomitant of the genetic diabetes mellitus syndrome)? Does “diabetes” cause obesity? Or are they related by some common denominator? Since recent studies have indicated that obesity directly affects serum insulin levels (2, 3), an obvious common denominator that would link obesity and diabetes would relate to the regulation of insulin secre- tion. Studies to define this relationship have shown that basal serum insulin levels are directly related to relative body weight (Fig. 1). Despite this close association, it is likely that a more precise measure of adiposity, such as total-body fat, would correlate even more closely with basal insulin levels. Conceivably, the basal i=- sulin level actually may be the simplest *From the Department of Medicine, University of Washington School of Medicine, and Veterans Administration Hospital, Seattle, Washington. *Supported in part by Public Health Service Grants 5ROIAM-06670; 5K3AM-28,167; and 5TO1- AM-05498. # Professor of Medicine, University of Washing- ton, and Chief, Division of Metabolism and Geron- tology, VA Hospital. *Instructor in Medicine, Uni- versity of Washington, and Clinical Investigator, VA Hospital. * Assistant Professor of Medicine, Univer- sity of Washington, and Chief, Diabetes Unit, VA Hospital. accurate index of adiposity in man. Al- though-other influences undoubtedly con- uribute to the regulation of basal insulin, adiposity, but not diabetes, appears to be its major determinant, since there is no correlation between carbohydrate tolerance and basal insulin. When the insulin responses to oral glu- cose (3) (or other stimuli (4)) are ex- amined, it is apparent that the increase in insulin is directly related to the basal level. Since the degree of obesity predicts the basal insulin level it is not surprising that obesity correlates closely with the insulin response to glucose as well. There- fore, obesity is associated with hyperinsulin- ism both in the basal state and in response to glucose. The effect of diabetes (or any other disorder) on serum insulin and, specifically, the relation between carbohydrate intol- erance and absolute insulin levels can- not be evaluated unless this effect of adiposity on insulin levels is considered and corrected. Since the increment of serum insulin in response to glucose ap- pears to relate directly to the basal level, one can eliminate the effect of the basal level (hence also obesity) by expressing the response as percent change. When this is done the correlation between insulin re- sponse and obesity disappears and one uncovers the progressive impairment of insulin responses associated with increas- ing degrees of carbohydrate intolerance. Since there is a spectrum of carbohydrate tolerance in any population sample with no sharp division between “normal” and “abnormal,” it is almost impossible to ~~ 68 group individuals in any decisive manner. However, at one end of the spectrum, an attempt can be made to characterize a more normal response. The characteristic in- sulin response of normal subjects, whether thin or obese, appears to be approximately a sixfold increase in insulin Jevels during the first 60 min after oral glucose (8). In contrast, the characieristic response of sub- jects with mild carbohydrate intolerance (more than 2 standard deviations above mean “normal”), whether thin or obese, is an early impairment in insulin secre- tion after glucose (Fig. 2). Without the necessity of grouping subjects arbitrarily, a significant inverse correlation is ap- parent between carbohydrate intolerance and insulin responses, expressed as percent increase above basal levels (Fig. 3). Al- though the obese individual with carbo- hydrate intolerance may secrete more in- sulin than his thin counterpart, the response remains relatively impaired and not sufficient to maintain normal carbo- hydrate tolerance (Fig. 4). Therefore, in the presence of carbo- hydrate intolerance, the insulin response to glucose (and to other stimuli of acute insulin secretion) (4, 5) is decreased, and a paradox becomes apparent. Since in- creased plasma insulin levels are associated with obesity and decreased insulin responses with carbohydrate intolerance, why are these two phenomena so frequently found in the same patient? A reasonable postulate is that adiposity, and its associated adipose cell enlargement, results in decreased tissue sensitivity to insulin and impaired glucose uptake, as shown in the human forearm (6) and isolated adipose cells (7). A compensa- tory increase in insulin secretion may he triggered that is adequate to maintain nor- mal carbohydrate tolerance in some in- dividuals. However, in other obese sub- jects, the beta cell may compensate in the basal state but fail to keep up with de- mand, which results in a relative decrease in insulin response and, as a consequence, Fasting Insulin vs. idea! Body Weight (N37) 90 + e472 »< 001 per mi 025 70 © +10 +30 +30 +30 +55 +60 +70 +80 +80 +100 Percent ideal body weight Fic. 1. Corrclation of basal serum immunoreac- tive insulin concentration with cbesity (expressed as percent of ideal body weight) in 37 healthy male volunteers, aged 40-60 (3). carbohydrate intolerance (Fig. 5). This postulate is compatible with the observa- tion in obese groups that the duration, rather than the degree, of obesity (Fig. 6) best correlates with carbohydrate intol- erance (8). It is intriguing that age, which is also associated with carbohydrate intol- erance (8-10), apparently has the same effect as the gene for diabetes on insulin responses (11). With both, carbohydrate intolerance can be related to diminished insulin responses. These observations sug- gest that in general, carbohydrate intol- erance results from a decreased relative insulin response. Since obesity and diabetes can be re- lated via the regulation of plasma insulin levels, what is the role of fatty acid mo- bilization, which also has been thought to be an important factor in this associa- tion? Elevated plasma levels of free fatty acids (FFA) and glycerol, which together are presumably a reflection of accelerated lipolysis, have now been found to occur only when obesity and carbohydrate in- tolerance coexist (12). Basal FFA and glyc- erol levels are directly related to the de- gree of carbohydrate intolerance in only obese, but not thin, subjects and are cor- related with weight only in subjects with mild carbohydrate intolerance. Thus, ac- celerated lipolysis does not appear to 69 Insulin Responses : Percent Increase Above Fasting Levels , Obess Nondiobeics (5) ~ 700 ’ ’ ’ £5 3 Thin Nsstberes 120 90 Time, minutes Fic. 2. Insulin responses (expressed as mean per- cent increments above basal levels) of thin and obese subjects, with “normal” and “abnormal” carbohy- drate tolerance during 3-hr, 100-g, oral glucose tol- erance tests (3). ® Glucose Tolerance va Insulln Resporas during First 60 Minutes offer Glucoss ne37) eo 70 200 300 abo 500 80d Parcant insulin response = 60 minutes. 100 Fic. 3. Correlation -of glucose tolerance (ex- pressed as the integrated area circumscribed by the blood glucose-time curve during 3-hr, 100-g, oral glucose tolerance tests) with the early relative in- sulin response (expressed as the integrated area cir- cumscribed by the percent increment-time curve during the first 60 min after oral glucose). Same sub- jects as in Fig. 1 (3). characterize either uncomplicated obesity or mild diabetes in the nonobese. These recent observations do not support the hypothesis that an abnormal elevation of plasma FFA is a fundamental defect as- sociated with mild carbohydrate intoler- ance. It is possible to construct a tentative © ° Oo GLUCOSE INTOLERANCE = 00 OBESITY SAL RES Fic. 4. A diagrammatic representation of changes in basal insulin levels (central light zone) and acute insulin responses (outer dark zone) with progressive degrees of glucose intolerance (vertically) and obesity (horizontally). (Courtesy of Dr. W. R. Hazzard.) Genetic | on J Bosal Insulin —| Insulin | Resp Secretion to Glucose Insulin | antagonism Adiposity Fic. 5. A postulated mechanism for the interac- tion of obesity and genetic diabetes on the regula- tion of insulin secretion. DURATION OF OBESITY AND GLUCOSE TOLERANCE (Two hours offer 50.gm glucose, po.) 3 3 2 = 3 Blood Sugar Response (oreo omits) 8 40 Years Fic. 6. Correlation of duration of obesity with glucose tolerance (expressed as the integrated area circumscribed by the blood glucose-time curve dur- ing 2-hr, 50-g, oral glucose tolerance tests in 63 obese females without glucosuria, aged 22-65). Cal- culated and plotted from the data of Ogilvie (8). 70 mechanism that relates the effects of obe- sity and mild diabetes on the regulation of plasma insulin to their observed inter- action on lipolysis. Obesity alone, with its associated hyperinsulinism, effectively lim- its fat mobilization. In the thin subject with mild carbohydrate intolerance, in- sulin secretion also appears to be sufficient to limit fat mobilization. Enhanced lipoly- sis ensues only when the tissue insulin antagonism associated with adiposity and the impaired insulin responses character- istic of mild carbohydrate intolerance are combined. In ketoacidosis, the deficiency of insulin response is so severe that pro- found effects on fat mobilization occur without the necessary association with adiposity and tissue antagonism to insulin action. SUMMARY Obesity and mild carbohydrate intol- erance tend to coexist in man. Obesity per se increases basal insulin secretion and insulin responses to glucose, while mild carbohydrate intolerance is associated with impaired insulin responses. Adiposity may cause decreased tissue sensitivity to insulin and reduced glucose uptake, which in turn leads to hyperinsulinism. Prolonged obe- sity may unmask genetic diabetes at an earlier stage to account in part for their frequent association. Abnormal insulin se- cretion and reduced sensitivity to its action also may be responsible for alterations of fat mobilization observed only in the pres- ence of both obesity and mild carbohydrate intolerance. » / { we o a ~ » © REFERENCES . Josun, E. P, L. I. DusLix ano H. H. Marks. Studies in diabetes mellitus. 111. Interpretation of the variations in diabetes incidence. Am. J. Med. Sci. 189: 163, 1935. Karam, J. H, G. M. Grovsky ano P. H. For- sHaM. The relationship of obesity and growth hormone to serum insulin levels. Ann. N.Y. Acad. Sci. 181: 374, 1965. . BAGDADE, J. D., E. L. BiERMAN AND D. PORTE, In. Significance of basal insulin levels in the evalu- ation of the insulin response to glucose in dia- betic and nondiabetic subjects. J. Clin. Invest. 46: 1549, 1967. Hazzaro, W. R., P. M. CrockFOorRD AND R. H. WiLLiams. The insulin response to glucagon: ef- fects of diabetes and obesity on nonglucose- dependent insulin secretion. Diabetes 17: Suppl. 1, 297, 1968. Fajans, S. S., J. C. Fiovp, Jr, 5. Pex, R. F. Knorr, M. JacossonN AND J. W. Conn. Effect of protein meals on plasma insulin in mildly dia- betic patients. Diabetes 17: Suppl. 1, 297, 1968. Rasivowrrz, D., ano K. L. ZierLer. Forearm metabolism in obesity and its response to intra- arterial insulin. J. Clin. Invest. 41: 2173, 1962. . SavLans, L. B,, J. L. KNiTTLE AND J. Hirsch. The role of adipose cell size and adipose tissue insu- lin sensitivity in the carbohydrate intolerance of human obesity. J. Clin. Invest. 47: 153, 1968. Ocivik, R. F. Sugar tolerance in obese subjects. A review of 65 cases. Quart. J. Med. 4: 345, 1935. HavnNEr, N. S.,, M. O. KjeLseerg, F. H. EPSTEIN Anp T. Francis, Jr. Carbohydrate tolerance and diabetes in a total community, Tecumseh, Mich- igan. Diabetes 14: 413, 1965, Anpres, R. Aging, glucose tolerance, and dia- betes—a proposal. In: Endocrines and Aging. Springfield, Ill: Thomas, 1967, chapt. 10, p. 200. . Crockrorp, P. M., R. J. HarBeck Ano R. H. WiLriams. Influence of age on intravenous glu- cose tolerance and serum immunoreactive in- sulin. Lancet 1: 465, 1566. BAGDADE, J. D., D. PORTE, Jr. AND E. L. BIERMAN. The interaction of diabetes and obesity on the regulation of basal lipolysis in man. J. Clin. Invest. 47: 3-a, 1968. 71 Obesity and the Body Image: I. Characteristics of Disturbances in the Body Image of Some Obese Persons BY ALBERT STUNKARD, M.D., AND MYER MENDELSON, M.D. OF THE many behavioral disturbances to which obese persons are subject, only two seem specifically related to their obesity. The first is overeating, the second is a disturbance in body image. This paper attempts to characterize these disturbances in body image, to describe their determi- nants, and to discuss their implications for theory and therapy. It is based upon one- hour interviews with 74 randomly selected obese persons in the general medical and psychiatric clinics of the Hospital of the University of Pennsylvania, plus a back- ground obtained by psychotherapy of about 20 obese persons for periods of from one to ten years. The median age of the sample was 43, with a range from 18 to 70. It was about equally divided between men and women, with medical clinic patients generally of lower class and psychiatric clinic patients of lower middle class. Although the psycho- therapy patients had not been selected for this purpose, their demographic characteris- tics were similar to those of the psychiatric clinic patients. The term “bedy image” refers to “the picture that the person has of the physical appearance of his body”(17). Disturbances in body image may range from “gross depersonalization . . . through distorted thoughts and feelings about the body, to distorted perceptions . . .”(10). The dis- turbances to be described are primarily in the area of feelings. One might expect that Read at the 122nd annual meeting of the Amer- ican Psychiatric Association, Atlantic City, N. J., May 9-13, 1966. Dr. Stunkard is Professor and Dr. Mendelson is Associate Professor, Department of Psychiatry, Uni- versity of Pennsylvania, Philadelphia, Pa. 19104. This work was supported in part Health Service grant MH-03684 tional Institute of Mental Health. by Public from the Na- derogatory feelings about one’s body oc- curred in all obese persons and that they were a central feature of the neurosis of all emotionally disturbed obese persons. Such is not the case. Body image distur- bances do not occur in emotionally healthy obese persons, and we have found them in only a minority of neurotic obese persons. Nature of the Disturbance We have divided the manifestations of body image disturbances rather arbitrarily into three areas: views of the self, self- consciousness in general, and self-conscious- ness in relation to the opposite sex. A dis- turbance in one area usually meant also a disturbance in the other two. Views of the self. A simple and revealing method of assessing a persons body image is to ask him how he views himself in a mirror (3). We have systematically collected data on this topic. An obese woman re- ported, “I call myself a slob and a pig— I look in the mirror and I say ‘youre noth- ing but a big fat pig’.” The experience of looking at oneself in a mirror can become so disturbing to obese persons that it can even interfere with such activities as shopping, which requires look- ing into store windows. One obese man reported, Just looking at myself in a store window makes me feel terrible. It’s gotten so I am very careful not to look by accident. It’s a feeling that people have the right to hate me and hate anyone who looks as fat as me. As soon as I see myself, I feel an uncontrollable burst of hatred. T just look at myself and say “I hate you, you're leathesome !” As we, have indicated, a disturbance in body image is not present in all obese persons, and many view their obesity in a 72 thoroughly realistic manner. As one obese man put it, “You know, I caught a glimpse of myself in the mirror this morning, and - I was surprised at how fat I have become. It made me feel that it’s time to get some of this weight off.” Self-consciousness in general. The private views which some obese persons hold about their bodies are frequently paral- leled by an intense self-consciousness and even misperception of how others view them. It sometimes seems as if they feel that nothing ever happens to them except in some kind of (usually derogatory) re- lationship to their weight. One young man said, “Weight was always a problem. If I missed a note in my music class and someone said, ‘You always mess it up, man,’ then I would think, ‘Maybe it’s because of my weight.”” A 61-year-old housewife said, “I've always been self-conscious—I was al- ways fat. It made me shyer than most people, I was withdrawn and kind of back- ward. I was never pleased with myself at all, never, never. ...” That obesity is not a sufficient cause for such self-consciousness is indicated -by the remarks of a man who was at the time 46 percent overweight; “The cosmetic point of view really bothers me only indirectly— through my wife. She objects to it; I think that she thinks a lot about it.” Self-consciousness in relation to the op- posite sex. Almost all of the subjects with body-image disturbances had serious diffi- culties in relationships with the opposite sex. These difficulties seemed both to arise from, and to reflect, self-consciousness about how others viewed their obesity. One of the most common complaints dealt with humili- ations at parfies. As one woman put it, “Nobody wants to go out with a tub, which was my nickname. By going to a dance all I did was stand against the wall listening to music. Nobody ever asked me to dance. I made believe I didn’t care. But I just thought I was a big nothing.” An attractive young woman who, in other days, would have been considered pleasingly plump, put in even more succinct terms the despair with which recurrent disappointments had brought her to consider the possibility of marriage, “Who would want to marry an elephant?” These disturbances ranged from avoid- ance and inhibitions to hateful devaluation of the opposite sex. For example, a 45-year- old man reported, $ Physical appearances seemed to exclude me from social activities and I resented it. I felt 1 was not physically attractive to women. Now I consider being short and heavy an advantage. I don’t have anything to do with women. I hate their guts. I get such a revulsion when 1 see how women act that I can’t bring myself to go after sex. An obese woman said that her body made her feel embarrassed with men she thought were “normal.” “So I pick up abnormal men. I'll tend to hang out with fairies. That way I don’t feel that I have to compete with real women.” The obese persons who escaped a dis- turbance in the body image had far fewer difficulties in their relations with the op- posite sex. One such man boasted, “I can get a woman any time I want. I'm a good- looking guy, and if you know anything about how women are, you know that I can get one without any trouble.” Factors Influencing the Disturbance The intensity of the body image distur- bances fluctuates widely even over short periods of time. When things are going well and a person with a body image dis- turbance is in good spirits, he may be troubled little or not at all by his disability, although it is rarely far from awareness. Let things go badly, however, let a de- pressive mood ensue, and at once all of the derogatory and unpleasant things in his life become focused upon his obesity, and his body becomes the explanation and the symbol of his unhappiness. A kind of circular relationship obtains between body image disturbances which predispose to es- teem-lowering experiences and depressive moods which in their turn reinforce the disturbed body image. Despite these short-term fluctuations in intensity, body image disturbances persist with remarkably little change over long periods of time and in the face of con- siderable variation in life circumstances. Weight reduction, for example, appears to 73 have little effect upon them, a finding which has been made by a number of obese per- sons to their surprise and dismay. Neither the extent of the weight reduction, as il- lustrated by persons suffering from anorexia nervosa, nor its duration, as exemplified by the subjects described in the second paper of this series(15), seems able to correct the disturbance in body image. And we have not heard reports of spontaneous remission of the disorder. The relatively intractable nature of the disturbance in body image highlights the one form of treatment which has amelio- rated it. In each of five persons suffering from the disturbance who were signifi- cantly benefited by long-term psychother- apy there was a concomitant improvement in the body image disturbance, which has * persisted as long as six years. It is of in- terest that this improvement occurred prior to the control of the obesity and has ap- peared to be a favorable prognostic sign. As one young man said shortly before em- barking on a weight loss of 140 pounds, which he has maintained for five years, “It's strange to see myself in the mirror looking so fat, because I feel so differently about it now. I feel thin.” Origins of the Disturbance Three factors predisposed an obese per- son to the development of a disturbed body image: age of onset of the obesity, presence of emotional disturbance, and negative eval- uation of the obesity by others during the formative years. Three other factors did not seem of importance. They were: gender, extent of overweight, and intelligence. Age of onset. The most surprising aspect of the body image disturbances was the great importance of age of onset. Distur- bances occurred almost exclusively among persons who became obese during child- hood or adolescence. Table 1 shows that TABLE 1 Degree of Body Image Disturbance According to Age of Onset of Qbesity in 74 Persons ONSET CATEGORY ~~ _ SEVERE MiLD MONE Juvenile i: 8 9 Adult 0 10 30 x= 19.44, df = 2, p< .00L not one of 40 subjects with adult-onset obesity showed a severe body image dis- turbance, while such disturbances were found in half of those with onset of obesity in childhood or adolescence, the so-called “juvenile obesity.” Our interviews suggested that adoles- cence was the period during which the disturbed body image was most likely to begin. Again and again subjects reported bitter adolescent experiences which had colored their whole later evaluation of their obesity. By contrast, childhood seemed to have been a time of relative indifference for the elaboration of this disorder. The second paper of this series describes fur- ther studies concerning the period of onset (15). Presence of emotional disturbance. The relatively short interviews of this study fre- quently uncovered evidence of emotional disturbance. Among subjects with adult- onset obesity a wide variety of diagnostic pictures was observed, few if any of which seemed related to their obesity. Among subjects with juvenile obesity, on the other hand, a disturbance in body image was usually a central feature of any emotional disorder. Nine juvenile obese subjects showed no disturbance in body image. No emotional disturbance was detected in six of them, while three manifested well-de- fined neurotic patterns not involving body image disturbance. Emotional disturbance apparently constitutes a necessary but not sufficient cause of a disturbance in body image. Negative evaluation of obesity by sig- nificant others. The almost universal deval- uation of obesity in our society today might make it appear redundant to mention this factor as significant in the development of a disturbed body image. The occasional instance in which an obese youth did not face such censure, however, reminds us of the force of these attitudes. Two obese men in this series, both with onset of obesity in childhood and both from dis- turbed early family environments, escaped body image disturbances. The families in which both men were raised valued large size and looked upon overweight as a sign of strength and health. This favorable fam- ily evaluation was shared by their peers, and 74 both men were much in demand as football players. Despite the fact that they showed clearly defined neurotic patterns at the time of their examinations, neither showed any evidence of a disturbance in body image. Discussion The disturbances in body image of obese persons are different from those which oc- cur in brain-damaged and schizophrenic persons and in normal persons under the influence of drugs, hypnosis, and fatigue (1, 2, 4,6, 7, 11, 12, 16). In the greater emphasis on affective than on cognitive disorder they resemble instead the distur- bances reported among persons suffering from deformities of the face, breasts, and genitals(8, 9, 13). The main feature is a preoccupation with obesity, often to the exclusion of any other personal character- istic. It may make no difference whether the person be talented, wealthy, or intelli- gent; his weight is his overriding concern and he sees the world in terms of body weight. He may divide society into per- sons of differing weights, and his orien- tation toward others may be largely in terms of this division. He envies persons thinner than he and feels contempt for those who are fatter. At the center of this attitude is the appraisal of his own body as grotesque and even loathesome, and the feeling that others view it only with horror and contempt. The more we have learned about obesity, the more we have been impressed with the need for specificity in linking neurosis and obesity. Many obese persons have se- vere neurotic problems. From this obser- vation it has often been inferred that the neurosis is a cause of the obesity. On methodologic grounds alone this is a shaky inference, and the results of recent pop- ulation surveys emphasize its weakness. In one sample, for instance, some degree of emotional disturbance was found in 80 per- cent of the population and obesity in 20 percent(5, 14). In this population, chance alone could ensure a high concordance of neurosis and obesity. Furthermore, it is our clinical impression that the presence of neurosis in an obese person does not explain his obesity, nor is it even necessarily relevant to it. Under- standing the relationship of neurosis to obesity requires careful description of which neurotic features are and which are not specific to obesity. The disturbance in body image constitutes one such specific feature. This concern with specificity is partic- ularly important in assessing the role of psychotherapy in the treatment of obesity. The results of psychotherapy have been disappointing, particularly when it has been used without careful selection of patients and when weight reduction has served as the only criterion of improvement. This disappointment does not deny the frequent usefulness of psychotherapy to obese per- sons suffering from emotional disturbances; it does emphasize the ineffectiveness of treatment of their obesity. In such a situa- tion the delineation of specific indications for psychotherapy should permit our limit- ed psychotherapeutic resources to be in- vested where they have the greatest pros- pects of success. The favorable results of long-term psychotherapy with persons suf- fering from body image disturbances, and the ineffectuality of other treatment, includ- ing prolonged weight reduction, suggests that this disorder constitutes a specific in- dication for psychotherapy. Summary Of the many behavioral disturbances to which obese persons are subject, only two seem specifically related to their obesity. One is overeating, the other is a disturbance in body image. The disturbance in body image is characterized by a feeling that one’s body is grotesque and loathesome and that others view it with hostility and contempt. This feeling is associated with self-consciousness and with impaired social functioning. It arises among emotionally disturbed persons whose obesity began prior to adult life, in families which did not value their obesity. The disturbance is not affected by weight reduction but has been favorably altered by long-term psycho- therapy. REFERENCES 1. Bruch, H.: The Importance of Overweight. New York: W. W. Norton & Co., 1957, pp. 182-187. ® “o no 10. Fedem, P.: Some Variations in Ego Feeling, Int. J. Psychoanal. 7:434-444, 1926. Fisher, S., and Cleveland, S. E.: Body Image and Personality. Princeton, N. J.: D. Van Nostrand Co., 1958. Gerstm J: Psy and Ph enological Aspects of Disorders of Body Im- age, J. Nerv. Ment. Dis. 26:499-512, 1958. Goldblatt, P. B., Moore, M. E., and Stunkard, A. J.: Social Factors in Obesity, J.AM.A. 192:1039-1044, 1965. Klemperer, E.: Changes of Body Image in Hypnoanalysis, J. Clin. Exp. Hypnosis 2: 157-162, 1954. Kolb, L. C.: “Disturbances of Body-Image,” in Arieti, S., ed.: American Handbook of Psy- chiatry, vol. 1. New York: Basic Books, 1959. MacGregor, F. C., Abel, T. M. Bryt, A. Lauer, E., and Weissman, S.: Facial Deformi- ties and Plastic Surgery: A Psychosocial Study. Springfield, Ill.: Charles C Thomas, 1953. Money, J., and Hampson, J. G.: The Evidence of Human Hermaphroditism, Bull. Hopkins Hosp. 97:301-319, 1955. Orbach, J., Traub, A. C., and Olson, R.: Psy- hotnginal 34-621 0 ~ 74-6 11 12. 13. 14. 15. 16. Wn. chophysical Studies of Body Image, Arch. Gen. Psychiat. 12:41-47, 1966. Savage, C.: Variations in Ego Feeling Induced by D-Lysergic Acid Diethylamide (LSD-25), Psychoanal. Rev. 42:1-16, 1955. Schilder, P.: Image and Appearance of the Human Body. London: Kegan Paul, Trench, Trubner & Co., 1935. Schonfeld, W. A.: Gynecomastia in Adoles- cence: Effect on Body Image and Personality Adaptation, Psychosom. Med. 24:379-389, 1962. Srole, L., Langner, T. S., Michael, S. T., Opler, M. K., and Rennie, T. A. C.: Mental Health in the Metropolis: The Midtown Manhattan Study, vol. 1. New York: McGraw-Hill Book Co., 1962. s . Stunkard, A. J., and Burt, V.: Obesity and the Body Image: II. Age of Onset of Disturbances in the Body Image, Amer. J. Psychiat., to be published. Stunkard, A. and Mendelson, M.: Disturb- ances in Body Image of Some Obese Per- sons, J. Amer. Diet. Ass. 38:328-331, 1961. Traub, A. C,, and Orbach, J.: Psychophysi- cal Studies of Body Image, Arch. Gen. Psy- chiat. 11:53-66, 1964. OBESITY AND CORONARY HEART DISEASE William B. Kannel, M.D., M.P.H,, F.A.CP., FA.CC. The Framingham Heart Study Since 1948 the Framingham Study has followed a representative sample of 5209 adult residents in the town by means of a standardized biennial cardiovascular examination, daily surveillance of hospital admissions, death information and information from physicians and other sources outside the clinic. The objective has been to study the epidemiology of cardiovascular disease-—i.e., the circumstances under which it arises, evolves and ‘terminates fatally in the general population. The Framingham Study is designed to determine in what particulars those who go on to develop cardio- vascular diseases differ from those who remain free of the diseases over a longer period of time. By including all cases—those too mild or inapparent and those dying too suddenly as well as those reaching the clinical horizon by appearing inn medical facility—it is possible to gain an undistorted picture of the entire clinical spectrum of the disease. By providing a complete follow-up, the natural history and prognosis of cardiovascular disease is obtained with greater clarity. Since the purpose of the Study is to examine the natural history of cardiovascular disease, no intervention is undertaken, although the findings of the examination are routinely reported to the patient's physician. Thus far 20 years of follow-up are available. Numerous reports have been published on the relation of a variety of living ‘habits and personal attributes to later development of coronary disease, stroke, peripheral vascular disease, and congestive heart failure, among others. The demonstration of geographic variation in coronary incidence suggests that powerful environmental influences exist which evolve from the unbridled application of modern technology.!3 The altered life style which in- cludes the cigarette habit, habits of sloth and gluttony From the Framingham Heart Disease Epid Study, F leading to obesity and lack of physical activity have all been implicated.* Also, dictary differences have been incriminated in coronary atherogenesis and invoked to explain the geographic variation in coronary incidence.5$ The incriminated nutrients are indeed consumed to a lesser extent in low coronary incidence areas. Yet, within the Framingham as well as within other populations there is an imperfect correlation between particular nutrients eaten and serum lipid values or coronary heart disease prevalence. 10 On the other hand, energy balance as reflected by weight change is distinctly related to the major factors which promote accelerated atherogenesis. In Framing- ham, weight change is prominently related to blood pres- sure and lipid values (Figure 1). It is also related to the development of impaired carbohydrate tolerance.’ Thus far, weight pattern is the only major determinant of these powerful con. ributors to coronary incidence which has been identified in population studies. In affluent societies excess calories come largely from saturated fat and refined simple carbohydrate. Also, high saturated fat intakes are invariably accompanied by high cholesterol intakes. Whether the atherogenic effects of obesity result from the nutrient composition of the diet or excess calories per se, it is gratifying to note that as people lose excess weight its atherogenic accompaniments tend to decrease (Figure 1). This suggests that the effect is reversible, and that this is a direct cause and effect relationship. Overweight is associated with a distinct increase in risk of developing a coronary event. There is no critical value of overweight detectable since risk increases in propor- tion to relative weight or skinfold thickness from the lowest to the highest values. The relationship to the devel- NHLI, NIH, Bethesda, Md. opment of angina and sudden death appears to be stronger than that to myocardial infarction. 1.12 This suggests some unique cffect over and above its promotion of ac- celerated atherogenesis by raising the blood pressure and scrum lipids and impairing of carbohydiate tolerance, This is borne out by multivariate analysis which reveals that only a fraction of the increased coronary incidence associated with overweight in men is accounted for by its atherogenic accompaniments.’ 9:25 Not all concur in this observation.'? While weight gain is associated with a substantial in- 78 crease in serum cholesterol value, the major lipid aberra- tion observed in the obese is in pre-beta lipoprotein and endogenous triglyceride metabolism. Some believe that its atherogenic potential derives primarily from this metabolic derangement. Risk of myocardial infarction can be shown to be proportional to the antecedent con- centration of each of the major lipids encountered in the blood.’s The lipids cholesterol and triglyceride as well as their lipoprotein vehicles are all related to the rate of development of coronary heart disease. However, regard- less of the associated lipoprotein pattern, risk is pro- portional tc the total serum cholesterol value. In the Framingham Study analysis of the net contribution of cholesterol versus endogenous triglyceride to coronary risk—taking into account the level of cach lipid, and associated risk variables including overweight—suggests that it is primarily cholesterol that determines risk.!s Evidence that triglyceride is atherogenic is less con- vincing. While triglyceride as well as cholesterol is found in atherosclerotic plaques, lesions have not been produced by experimentally raising serum triglyceride values alone. On the other hand, atherosclerosis has been regularly induced in a wide variety of animals, including primates, by raising their serum cholesterol values. Furthermore, while familial type 2 hypercholesterolemia has been con- clusively demonstrated to be associated with premature atherosclerosis, familial type 1 hypertriglyceridemia has not." Type 4 endogenous hypertriglyceridemia is defi- nitely more common in myocardial infarction patients . than in controls, but the accompanying cholesterol values in these patients are also higher. Populations in which high coronary mortality and more severe coronary atherosclerosis have been found characteristically have high cholesterol values. Their triglyceride values, on the other hand, are frequently lower than those found in low coronary areas. Impaired carbohydrate tolerance, also more prevalent in type 4 hyper-pre-beta lipoproteinemia with high endogenous triglyceride values, is often more prevalent in low coronary incidence areas.?6.17 Adult onset, ketoresistent, hyperinsuiinemic diabetes is distinctly more common in obese persons. Such dia- betics appear to develop an excess of coronary heart disease.’ This excess risk is not entirely explained by the associated atherogenic accompaniments of diabetes or the obesity itself.!8.1? Correction of the impaired carbo- hydrate tolerance by conventional drug therapy has not been demonstrated to prevent the cardiovascular se- quelae. In fact, the University Group Diabetes Program suggests that it may do harm.20-2! It would seem that unless concepts of “diabetes control” are broadened to include lipid, blood pressure, and obesity components as well as the impaired glucose tolerance we will not achieve a substantial reduction in cardiovascular sequelae. A prominent feature of such a prophylactic approach is reduction of overweight, which has been shown to improve the impaired carbohydrate tolerance in many persons with diabetes, and to ameliorate the lipid aberration and the hypertension. It would also appear wise to urge diabetics to avoid smoking. Weight Loss The final link in the chain of evidence which implicates obesity in coronary heart disease has yet to be forged; namely, that reducing results in a lowering of coro- nary incidence. Life insurance data suggest that this is 50.22:23.2¢ Factors in reinsurance of obese persons are selective and therefore data are not entirely convincing. However, the fact that weight reduction is associated with improvement in atherogenic traits is encouraging. ~ Obesity in Women While obesity in men makes some unique contribution to coronary incidence, in women it is largely explained by the accompanying atherogenic traits.12.19.25 This is not to say that obesity in women is unimportant as a contribu- tor to coronary incidence, but rather, that in women, it seems entirely mediated through its influence on atherogenic traits. In men there may be an additional unique effect. This difference in the sexes requires con- firmation and explanation. Obesity versus Body Build Assessment of the influence of adiposity on cardio- vascular incidence is complicated by the assertion of anthropologists that relative weight is an imprecise meas- ure of obesity. However, whether judged by relative weight, skin-fold thickness (or weight gain after com- pletion of musculoskeletal growth) coronary incidence is proportional to degree of adiposity.12.19.25 High Blood Pressure and Obesity The relation of adiposity to hypertension is neither simple nor direct. Weight gain on the average is asso- ciated with a distinct and substantial rise in blood pres- sure, but there are numerous exceptions. Normotensive obese persons often develop hypertension after some time lag. Also, lean hypertensives appear to show an in- creased propensity to development of obesity ns well as the converse.2¢ This suggests that some third factor may be related to the development of both obesity and hyper- tension. Clearly more research is needed to elicit the mechanism of blood pressure rise in some obese persons. One thing is clear: this association is not simply a fat arm artifact.26 Obesity and Sudden Death Obese persons sccm to be especially prone to death from a sudden coronary. Incidence of sudden death in- creases in proportion to degree of obesity and, at least in Framingham, the fraction of coronary deaths which are sudden also increases. This suggests that obesity is par- ticularly related to suddenness of coronary fatalities. The mechanism requires elucidation. Preventive hinplications Obesity is the most prevalent of the common metabolic disorders of mankind. It is a scrious condition which adversely affects a variety of organ systems, limits pro- ductivity, causes decades of disability and contributes to premature death. Aside from cardiovascular disease, overweight persons develop an excess of diabetes, diges- tive disturbances, degenerative arthritis, skin ailments, "and post-surgical and obstetric difficulties. The insurance industry has long pointed out the excess of cardiovascular mortality in the obese.2%24.27.28/Yet, despite the wide recognition of the health consequences of obesity, its high prevalence in affluent societies continues unabated. In Framingham more than 15% of men and 20% of women are at least 35% above “ideal weight.” Men are heavier than their counterparts were two decades ago. Most women, though lighter than formerly, are still dis- tinctly overweight, particularly in middle age and beyond. Effective means for preventing and correcting obesity have yet to be devised. Treatment is universally recog- nized as difficult and the efforts of most physicians seem ineffectual. Judging by the number of remedies proposed and discarded, no effective treatment capable of achiev ing a sustained weight loss is available. No field of medicine has been as filled with confusion, faddism or quackery. More information concerning the epidemiol- ogy, physiology, psychology and genetics of obesity is urgently needed. | ” Adiposity seems to be a complex disorder involving psychological, cuitural, genetic, familial and cnviron- mental influences. It tends to run in families. However, families share more than genes, and spouses as well as siblings tend to share a propensity to overweight, a fact that implicates common food patterns. Also, there may be more than one constitutional type, suggesting multiple etiologies. The chief determinants of ordinary exogenous obesity in the general population are not well established. Condi- tioning as well as genetic factors acting early in life seem to play a role. Practices of overfeeding seem to persist from childhood when parents diligently train children to empty their plates rather than eat until they have had enough. Equating leanness with ill-health is a parental attitude which must be dispelled. Most obesity is not attributable to any detectable clin- ical disorder. Simple exogenous obesity appears to derive from faulty eating and exercise habits. Its avoidance would seem to require a change in our ecology away from that which provides a surfeit of rich, high calorie foods in the face of shrinking demands for physical work in daily 79 living. Feeding has changed from a necessity to a form of pleasurable entertainment. The roots of obesity may well lic in outmoded concepts from the past when food sup- plies were intermittent and often inadequate. There is some evidence to suggest that obesity may be a sclf-perpetuating entity by permanently increasing the number of fat cells in adipose tissue depots and by alter- ing their sensitivity to insulin as they become stuffed with triglyceride.2 A blunting of the ability to regulate intake precisely to demand for calories scems to occur in the obese. The metabolic aberrations of obese persons in- clude impaired carbohydrate tolerance, inability to mo- bilize fatty acids adequately, a sluggish free fatty acid response to exogenous insulin, resistance to ketosis and a diminished growth hormone response to hypoglycemia, exercise and starvation.?® These metabolic derangements which accompany obesity could be the chain of events which tend to perpetuate it, an adaptive response to over- feeding, or both. In any cvent, the longer these aberra- tions persist and the more pronounced the obesity, the more resistant is the condition to management. Over- feeding in childhood results in hyperplasia of fat cclls which seems to persist indefinitely. Juvenile obesity tends to persist into adult life and is usually the more severe and resistant form. However, while there may be a wide variation in the tendency of persons to store and retain calories, obesity is still basically a problem of eating too much and exer- cising too little. What makes it insidious is that it occurs imperceptibly by small increments in persons who appear to have blunted their ability to recognize and respond to internal cues of satiety in an ecology which promotes overfeeding. In general, obesity is accompanied by atherogenic alterations in blood lipids, blood pressure and carbo- hydrate tolerance. Whether this is characteristic of all types of obesity or only some particular variety is not known. It is also uncertain whether the nutrient composi- tion of the excess calories is important. Nevertheless, ordinary exogenous obesity as it is encountered in the general population is associated with an excess develop- ment of coronary heart disease, (particularly angina pec- toris and sudden death). To a large extent this seems to - derive from its atherogenic accompaniments, especially in women. In men, there appears to be some additional unique effect; possibly the increased work load imposed on the heart. This may provoke the carlier emergence of angina in persons with a compromised coronary circula- tion. The mechanism by which obesity promotes sudden coronary death requires further investigation. From the foregoing it seems reasonable to expect that correction of overweight should iniprove the exercise tol- erance of persons with established coronary heart disease by reducing their cardiac work load and blood pressure. In persons overweight and predisposed to coronary heart disease weight reduction can improve their coronary pro- file by lowering their blood pressure and cholesterol and improving their carbohydrate tolerance. Whether this will eventuate in a reduced risk of lethal coronary attacks has never been rigorously tested. Nevertheless, it would appear that much would be gained by correcting over- weight in persons highly vulnerable to coronary heart 80 discase. Avoidance of overweight would seem highly desirable for the entire population and preferable to attempts to correct resistant long-standing obesity. Figure 1 CHANGES IN SYSTOLIC BLOOD PRESSURE LEVEL ra CHANGES IN RELATIVE WEIGHT — FRAMINGHAM STU! BIBLIOGRAPHY . Scrimshaw IVs, Trulson M, Tejada C, et al.: Bronte-Stewart B: The epidemiology of ischaemic heart disease. Postgrad. M.J. (London) 35:180, 1959. Puffer RR, Verhoestraete LJ: Mortality from cardiovascular diseases in various countries, with special reference to atherosclerotic heart disease. A preliminary analysis. Bull. World Hlth. Org. 19:315, 1958. Burgess AM, Jr, Fejfar Z, Kagan AR: Arterial hypertension and ischaemic heart discase: comparison in epidemiological studies. 1. The problem of international comparability. Chron. World Hlth. Org. 16:437, 1962. . Kannel WB: Habits and coronary heart disease. The Fram- ingham Heart Study. (PHS Pub. No. 1515) Wash., D. C., U.S. Gov't Printing Office, 1966. . Antonis A, Bersohn I: Serum-triglyceride levels in South African Europeans and Bantu and in ischaemic heart-disease. Lancet 1:998, 1960. Toor M, Kaichalsky A, Agmon J, et al.: Serum-lipids and atherosclerosis among Yemenite immigrants in Israel. Lancet 1:1270, 1957. Keys A, Kimura N, Kusukawa A, et al.: Lessons from serum cholesterol studies in Japan, Hawaii and Los Angeles. Ann. Int. Med. “8:57, 1958. Serum lipo- protein and cholesterol concentrations: Comparison of rural Costa Rican, Guatemalan and United States populations. Circ. 15:805, 1957. Kannel WB, Gordon T: The Framingham diet study: Diet and the regulation of serum cholesterol. Framingham Study, Section 24, Wash., D. C., U.S. Gov't Printing Office, 1970. Keys A (ed): Coronary Heart Disease in Seven Countries. AHA Monograph 29. N. Y., American Heart Association, 1970. . Kannel WB, Pearson G, McNamara PM: Obesity as a force of morbidity and mortality. In Adolescent Nutrition and Growth. Felix P. Heald (ed), Appleton-Century-Crofts, N.Y., pp. 51-71, 1969. Kannel WB, LeBauer EJ, Dawber TR, et al: Relation of body weight to development of coronary heart disease: The Framingham Study. Circ. 35:734, 1967. Keys A, Aravanis C, Blackburn H, et al.: Coronary heart disease: overweight and obesity as risk factors. Ann. Int. Med. 77:15, 1972. . Fredrickson DS, Levy RI, Lees RS: Fat transport in lipo- protcins—An integrated approach to mechanisms and dis- orders. New Engl. J. Med. 276:148, 1967. . Kannel, WB, Garcia MJ, McNamara PM, et al.: Serum lipid precursors of coronary heart disease. Hum. Pathol. 2:129, 1971. 16. 17. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. CHANGES IN SERUM CHOLESTEROL LEVEL WITH CHANGES IN RELATIVE WEIGHT ~— FRAMINGHAM STUDY 30 8 CHANGE IN RELATIVE WEIGHT ° 0 CHANGE IN SERUM CHOLESTEROL — Decreoses Unpublished data NHLI-Interstudy comparison Framing- ham-Puerto Rico and Hawaii. Bennett PH, Burch TA, Miller M: The high prevalence of diabetes in Pima Indians of Arizona, USA. Excerpta Medica Int'l. Cong. Series #221: pp. 33-39, 1970. . Gordon T, Kannel WB: Predisposition to atherosclerosis in the head, heart and legs: The Framingham Study. JAMA 221:661, 1972. . Gordon T, Sorlie P, Kannel WB: Coronary heart discase, atherothrombotic brain infarction, intermittent claudication —A multivariate analysis of some factors related to their incidence. Framingham Study—16 year follow-up. Fram- ingham Study, Section 27, Wash., D. C., U.S. Gov't Printing Office, 1971. University Group Diabetes Program: A study of the effects of hypoglycemic agents on vascular complications in patients with adult esign, hods and baseli results. Diabetes 19 (Suppl.) 747, 1970. University Group Diabetes Program: Effects of hypoglycemic agents on vascular complications in patients with adult-onset diabetes. IV. A preliminary report on phenformin results. JAMA 217:777, 1971. Society of Actuaries: Build and blood pressure study. Chicago, Ii., The Society, 1959. Marks HH: Influence of obesity on morbidity and mortality. Bull. N.Y. Acad. Med. 36:296, 1960. Lew EA: Importance of overweight in life insurance. 11th International Conference of COINTRA: 217-298, 1969. Seltzer CC: Overweight and obesity. The associated cardio- vascular risk. Minn. Med. 52:1265, 1969. Kannel WB, Brand N, Skinner JJ, Jr, et al.: The relation of adiposity to blood pressure and development of hyper- tension: The Framingham Study. Ann. Int. Med. 67:48, 1967. Lew EA: Mortality statistics for life insurance underwriting. JASA 43:274, 1948. Lew EA: Insurance mortality investigations of physical impairments. Am. J. Publ. Health, 44:641, 1954. Heald FP, Mucller PS, Daugela MZ: Glucose and free fatty acid metabolism in obese adolescents. Am. J. Clin. Nutr. 16:256, 1965. Campbell RG, Hashim SA, Van Itallie TB: Studies of food- intake regulation in man: Responses to variations in nutritive density in Jean and obese subjects. New Engl. J. Med. 285:1402, 1971. 81 Social Factors in Obesity Phillip B. Goldblatt, MD, Mary E. Moore, PhD, and Albert J. Stunkard, MD The relationship between obesity and several social factors was investigated among 1,660 adults representa- tive of 2 residential area in midlown Manhattan. An in- verse relationship previously described between obesity and parental socioeconomic status was also found be- tween obesity and one’s own socioeconomic status. Obes- ity was six times more common among women of low status as compared fo those of high status. Furthermore, upwardly mobile females were less obese (12%) than the downwardly mobile (229). Finally, the longer a woman's family had been in this country, the less likely she was to be obese. Similar but less marked trends obtained for the men. Suggestive relationships between ethnic and religious factors and obesity were also found for both sexes. These findings suggest opportunities for more effective weight control measures through programs specially tailored for populations at high risk. his report is an extension of our previous find- ing that social factors play an important role in human obesity.’ Current theories as to the eti- ology of obesity, whether behavioral, biochemical, or physiological, have directed their attention to the individual. We were, therefore, very much in- terested in a finding incidental to our earlier study, which was undertaken to assess the relationship of mental health to obesity in a large, representative, urban population.’ Parental social class, introduced as a controlling variable, showed a high correlation with the prevalence of obesity and was a more powerful predictor of overweight than a number of psychological measures. The present study, under- taken to investigate this relationship further, showed obesity to be related to each of the follow- ing additional social variables: the respondent’s own socioeconomic status, social mobility, and gen- eration in the United States. From the Department of Psychiatry, University of Pennsyl- vania, Philadelphia. Read in part before the Section on Nervous and Mental Diseases during the 114th annual convention of the American Medical Association, New York, June 24, 1965. Reprint requests to 3400 Spruce St, Philadelphia 19104 (Dr. Stunkard). Methods and Materials The data reported here were collected as part of the Midtown Manhattan Study, a comprehensive survey of the epidemiology of mental illness. The details of the sample and the data collection tech- niques have been fully described elsewhere.”® The data in the present analysis were obtained from the Midtown Home Survey of 1,660 adults, con- sisting of 690 males and 970 females between the ages of 20 and 59. One female in the Home Survey was omitted from our study because she was under four feet (122 cm) in height. The population was divided into three weight categories—‘“obese,” “normal,” and “thin”—-based upon the self-re- ported heights and weights as described in our previous paper. The validity of such reports has been attested to by an independent survey. The relationships of these categories to the wide- ly accepted standards for “desirable” weight of the Metropolitan Life Insurance Company® is shown in Table 1. The “desirable” weights are those for which the mortality rates are lowest, so that even the group which we have designated “normal” ex- ceeds their “desirable” weight by about 10%. The means for our “obese” groups were 34% and 44% above the “desirable” weight, indicating a signifi- cant degree of overweight. The corresponding means for the “thin” groups were 13% and 16% below the “desirable” weight. The respondent’s own socioeconomic status (SES) at the time of the interview was rated by a simple score devised by Srole et al® based upon the respondent's occupation, education, weekly in- Table 1.—Relationship of Midtown Home Survey Weight Categories to Standards of ‘Desirable’ Weight® A Over Under (—) Desirable Weight Midtown Weignt: »c EE % Overt) or Under () Desirable Weig Category TE rT Females Thin —13 —16 Normal +11 +9 Obese +44 +34 of Metropolitan Life c for weight. 82 come, and monthly rent. Each of these four varia- bles was subdivided into six categories. In the scoring, each variable was given equal weight. Thus, an individual in the lowest category of each variable (unskilled labor, no schooling, income less . than $49 per week, and monthly rent less than $30) received a score of four. Conversely, an indi- vidual in all of the highest categories (high white collar, graduate school education, income over $300 per week, and monthly rent greater than $200) received a score of 24. (Unmarried working women were rated on the basis of their own occu- pation. Unmarried nonworking women were classi- fied by their fathers’ occupation. Married women, whether working or not, were rated on the basis of their husbands’ occupation.) In order to obtain subgroups of sufficient size to permit control by variables which the analysis showed to be relevant, the population was divided into three socioeconom- ic groups as nearly equal in number as feasible. Individuals with scores of 4 to 10 were designated Jow status; those with scores of 11 to 16 were mid- dle status; and those with 17 to 24 points were high status. In our first paper we used the respondent’s social class of origin as a controlling variable. This mea- sure was employed so as to avoid any reciprocal relationship that might exist between a respond- ent’s present SES and his obesity. Obesity may in part depend on social status but, at the same time, social status may in part depend on obesity. The SES of origin is a measure of important social in- fluences on a respondent which are in no sense a product of his obesity. The social class of origin was based on the education and occupation of the respondent’s father when the respondent was 8 years old. The scores for SES of origin were divided into “low,” “medium” and “high” socioeconomic cate- gories in a manner analogous to that used for the respondent’s own SES. These two sets of scores permitted us to study the relationship of obesity to social mobility by comparing the socioeconomic status of the respondent at the age of 8 with that at the time of the interview. % obese own s.€.5 F][]s €5 oF orien Medium High status 1. Decreasing prevalence of obesity with increasing so- cioeconomic status (SES). Data exclude one female about whom no information on the socioeconomic status of origin was available. Results The present analysis extended in a most dra- matic way our previous finding of the importance in the understanding of human obesity of one socioeconomic variable—socioeconomic status of origin. Every one of the three additional social factors investigated was also strongly related to obesity. Furthermore, each was more strongly re- lated to obesity among women than among men. The analysis of the data for women will be pre- sented first. Obesity Among Women.—OWN SOCIOECONOMIC Srartus.—There was a marked inverse relationship between the prevalence of obesity and the respond- ent’s own SES. Figure 1 shows that the prevalence of obesity among lower SES women was 30%, fall- ing to 16% among the middle SES, and to only 5% in the upper SES. A chi-square test of the relation- ship between socioeconomic status and the three weight categories was significant at the 0.001 level (x* = 1207). Socioeconomic Status OF OrIGIN.—Just as the respondent’s own socioeconomic status was inverse- ly related to her obesity, so also was her socio- economic status of origin (X* = 66.5, P < 0.001). This latter finding, reported in greater detail in our earlier study, is also shown in Fig 1. Note that the relationship between this factor and obesity was nearly as strong as that between the respondent’s own socioeconomic status and her weight category. SociaL MosiLiTy.—The close correspondence be- tween the results for the respondent’s own SES and SES of origin suggested the possibility that these two variables were measuring the same un- derlying dimension. This would be the case in a society in which the vast majority of people lived out their entire lives in the same social class into which they were born. Such was not the case, however, in Midtown Manhattan which showed a high degree of social mobility. Indeed, 44% of the women belonged to a different social class from that of their parents. In other words, many people were classified differently by our two indices of social status, and these two variables did not measure the same underlying dimension. A mea- sure of this discrepancy between the respondent’s own SES and the SES of her origin is given by our index of social mobility. Figure 2 shows the relationship of social mobility to obesity. Of women who remained in the socio- economic status into which they were born 17% were obese, whereas among women who moved down in social status there was a higher preva- lence of obesity (22%), while among those who moved upwards there was a lower prevalence (12%) (X* = 20.5, P < 0.001). Thus, movement among the social classes as well as membership in a social class was predictive of obesity. GENERATION IN THE UNITED StaTES.—The fourth variable, generation in the United States, was also strongly linked to obesity. To assess this variable, 83 Table 2. Percentage of Obese Females by Generation in United States--Controlling the Factor of Socioeconomic Status (SES) Generation i: = - | i ni Ww I — - — Own [3 Own Own Own Own Own Own Own Own Own SES SES SES SES SES SES SES SES SES SES SES Me. High Low Med High low Med High Low Med High = 7 34 19 6 22 © 2 13 % «= 13 58 96 i26 78 23 52 87 35 23 102 *N = Number in sample falling into particular category, eg. 30% of 194 first generation low SES females are obese. Excludes two females about whom no information on generation is available. respondents were divided into one of four groups on the basis of the number of generations their families had been in this country. Generation I consisted of foreign-born immigrants; generation 11, of all those native-born respondents with at least one foreign-born parent; generation I11, of all those who were native-born of native-born parents but had at least one foreign-born grandparent; and generation 1V, of all those who had no foreign-born grandparents and who otherwise met the qualifica- tions for generation 111. Figure 3 shows that the longer a woman's family had been in this country, the less likely she was to be obese. Of first generation respondents 24% were overweight, in contrast to only 5% in the fourth generation (X* == 56.5, P < 0.001). It scemed probable that generation in the United States was closely related to socioeconomic status, and that the longer a family had been in this coun- try, the higher its status was likely to be. The data in Table 2 show that this is indeed the case. Thus, in the first generation, 194 out of 365 respondents were of low SES, while in the fourth generation, 102 out of 140 were of high SES (Xx? = 235.6, P < 0.001). To determine whether this phenomenon 'ac- counted for the finding that obesity was less com- mon the longer a respondent’s family had been in the country, we examined the prevalence of obesity for each SES within each generation. Table 2 clear- ly demonstrates that the inverse relation between obesity and generation was independent of socio- economic status. Of the generation 1 respondents who were of low status 30% were obese, but only 183% of generation IV Jow-status females were overweight. This trend obtained in all the social classes (x? for low SES = 21.5, P < 0.01; x* for middle SES = 18.1, P < 0.01; X* for high SES = 21.5, P < 0.01). Obesity Among Men.—The relationship between social factors and the prevalence of obesity among males paralleled that among the women, but in each instance was less marked. Own Socioeconomic Status.—There was an in- verse relationship between his own SES and obes- ity. Figure 4, however, shows that the eflect was far weaker than in the case of females (X* = 17.4, P. < 0.01). Whereas obesity was six times more common among women of lower socioeconomic status than among those of high status, the corre- sponding ratio among men was only 2:1. Socioeconomic Status oF ORIGIN. —Socioeco- nomic status of origin had an effect upon the prev- alence of overweight among men, although, as was true for women, the effect was weaker than that of the respondent’s own SES. Furthermore, the in- fluence of SES of origin was far weaker among men than among women. Whereas obesity was four times more common among women of lower socio- economic status of origin than among those of high status, the corresponding ratio among males was Jess than 2:1. Social MosiLiTy.—Social mobility was even more common among the men in our sample than among the women, with 47% of the males belong- ing to a different social class than their parents. % obese 30+ 201 10+ Down No change Up Mobility rn 2. Decreasing prevalence of obesity with upward social mobility. Data exclude one female about whom no in- formation on mobility was feasible. Generation in U.S. 3. Decreasing prevalence of obesity with increasing length of time of respondent's family in United States. Data exclude two women about whom no information on generation in United States was available. 84 % obese OWN 5.€.5. EJ] S. €. S. OF ORIGIN Nz235]|M=192 > Medium High status 4. Decreased prevalence of obesity with increased so- cioeconomic status (SES). % obese women ACIMeEN 307 1 N=150 20+ N=370 N-170 10+ | [| Down No change Up ——— Mobility 5. Slight trend toward decreased prevalence of obesity with upward mobility among men as contrasted to sig- nificant trend among women. women AL Imen N=273 N=101 N-86 / 1 I rc Iz Generation in U.S. — 6. Decreasing prevalence of obesity with increasing length of time of respondent’s family in United States. Data exclude two men about whom no information on generation in United States was available. Table 3.-—Weight Categories by Respondent's own 2 Socioeconomic Status (SES) % of Each Weight Category in Each SES te es Low SES iddle S| WRI Darren Aires pie ae er np Es Categories Males Females Males Females Males Females Thin 10% $% 9%. 19% 12% 37% Normal 59 61 64 65 73 58 Obese 31 30 27 16 15 5 N (100%)* 215 329 240 315 235 32% “Number of individuals in sample falling into particular category. Figure 5 shows the prevalence of obesity among those men who moved downward, stayed in the same class, or moved upward. Once again the same trend obtained as for females; in this instance, however, the chi-square did not reach a level of statistical significance. GENERATION IN THE UNITED STATES.—Among the men as among the women the percentage of obese respondents decreased as the number of genera- tions in the United States increased. Figure 6 re- veals that obesity was three times more common among the males in the first generation as compared to those in the fourth generation (X°=18.7, P< 0.001). There was, however, no sharp drop in the percentage of obese between generation II and III, as was the case among females. As among women, these findings resulted even when socioeconomic status was held constant. The Thin Category.—-At the beginning of the analysis of the weight categories, we expected that “thinness” would behave as though it were the op- posite of obesity. We found, however, a striking difference in this regard between men and women. With increasing status, women moved from the “obese” to the “thin” category, whereas men moved from the “obese” to the “normal.” Thus, thinness did operate as the opposite of obesity for women, but not for men. Table 3 shows that there were four times as many thin respondents among women of high status as there were among those of low status. Among men, however, about 10% were thin in all classes, and it was the percentage of normal- weight respondents that increased with increasing status. Comment The most important finding of this study was the remarkable consistency with which social fac- tors correlated with body weight. Such a strong correlation, appearing in all the factors investi- gated, is highly significant. The only other attempt to study obesity as a social phenomenon, that by Pflanz® in Germany, has reported similar findings. It is now apparent that obesity can no longer be viewed as simply an abnormal characteristic of the individual. It must also be viewed as one of the possible, and not too infrequent, normal responses of persons in certain subgroups of society to the perceived expectations of their social milieu. Although not being obese, indeed, being thin, seems culturally desirable for the women of Mid- town, almost one out of three lower class females was obese. Such a high percentage implies that in this subgroup of Midtown society, overweight is common enough that it need not be viewed as abnormal. In the Midtown society we do not have to look far to see the image of the slim, attractive female as portrayed throughout the popular culture. Mo- tion picture stars, television personalities, women in advertisements, fashion models, and, indeed, the fashionable clothes themselves, all reflect the defi- nition of the beautiful female as the one who is thin. How does such an ideal of beauty exert its impact on the body weight of persons in the differ- ent elements of society? At least two mechanisms seem plausible. First, a selection process may operate so that in any status-conferring situation, such as a promo- tion at work or marriage to a higher status male, thinner women may be preferentially selected over their competitors. Second, an acculturation phenomenon may be operating. For example, an individual's adult weight can be seen to be partly a product of social influences operating in his childhood. That this occurred in Midtown is demonstrated by our find- ing of a marked relationship between obesity and the SES of one’s origin. A similar process may also be operating throughout Lfe. Thus a female who acquires upper socioeconomic status for desirable attributes other than thinness will perceive that among her new peers more emphasis is placed upon being slim than was true in her old environment. She is likely, therefore, to make a greater effort to lose weight than she might previously have done. The lesser importance of social factors as related to body weight among men, as contrasted to wom- en, may arise from a lesser importance that society attaches to the physical appearance of men, as well as from a different definition of culturally desirable weight for them. In advertisements, for example, men are as likely to be exhorted to avoid being “97- Ib weaklings,” as to avoid being obese. It is, per- haps, not surprising that the normal weight cate- gory correlates so highly with upper socioeconomic status. The extent to which a respondent has adopted the Midtown values about body weight apparently depends upon at least two factors: first, on the length of his family’s exposure to these values (as measured by number of generations in the United States) and second, the amount of pressure to conform to these values, which is a function of his proximity to the upper classes where the values are most strongly exemplified. Although generation and socioeconomic status are related to each other, it has been shown in this study that each makes an independent contri- bution to the prevalence of obesity. It is unfortu- nate that the size of the sample precluded more precise estimates of their relative contributions. It is obvious that there were important differ- ences among our respondents besides those of class and generation. The Midtown sample included nine ethnic groups (British, Russian-Polish-Lithuanian, German-Austrian, Irish, Puerto Rican, Italian, Hungarian, Czech, and fourth generation Ameri- can) as well as many religions and sects. We dis- covered several relationships between ethnic and religious backgrounds and obesity. These were so intertwined with each other and with other social factors, such as generation and socioeconomic status, that we were unable to control all of the 85 relevant factors simultaneously. Nevertheless, some of the data are worth describing briefly. For example, only 9% of female respondents of British descent were obese, whereas 27% of those of Ttalian extraction were in this weight category. These differences diminished when social class was the control. Thus, for example, when only the up- per classes of both ethnic groups were contrasted, the prevalence of obesity was 10% for the British and 20% for the Italian. Such differences can be related to what is known about the traditional diets and social implications of eating of these two ethnic groups. Joffe,” for example, has reported that first generation Italian mothers regard obesity in their children as protection against tuberculosis. Childs" found the basic diet of Italian-Americans to have a high fat content. Finally, a recent study in a small Pennsylvania town inhabited almost entirely by Italian-Americans revealed that the diet had a greater proportion of fat than that of the average American diet and that the prevalence of obesity was also significantly above average. Another example of such a phenomenon may be found among our data for Americans of eastern European extraction. Joffe notes that the Czechs love food and are less Americanized than the Poles as far as cooking habits are concerned. Among the Czechs, there is a great deal of visiting on Sundays during which time large quantities of food are consumed. Refusing a second or even a third help- ing of food is considered impolite. Our data reflect the results of these customs. Of the lower-class Czechs 419% were obese as compared with only 18% of the lower-class Polish-Russian-Lithuanians that were obese. We also found differences among respondents of different religions. Lutherans, for example, were more often obese (24%) than Episcopalians (3%), but any statement made about the Lutherans and Episcopalians reflects also the difference between respondents of German and of British extraction. Unfortunately, we did not have enough cases to sift out the effects of religion per se. What has been reported in this study about obesity in Midtown in 1954 is not necessarily ap- plicable in toto to any other country, or any other urban area, or even to the Midtcwn of today. In- deed, Pflanz found an increased incidence of obesity among upwardly mobile German men and a de- creased incidence among German women; in con- trast, a decreased incidence among the upwardly mobile of both sexes was found in the present study. It thus appears that the same social mech- anisms which discourage obesity among the socially mobile in this country may encourage it among German men. Even though Pflanz’s specific find- ings differed from ours, his conclusion was the same: human obesity must be understood in part as a social phenomenon. Many of the present theories about human obes- ity were formulated by psychiatrists on the basis of their treatment of middle and of upper class 86 women, for whom obesity was a severe social liabil- ity. In other segments of society, however, obesity appears to be by no means such a handicap. Future researchers will have to explore the ways in which some respondents in all classes develop the attitude that a slim appearance is very important. Studies will be needed to determine the reasons why certain subgroups have a higher incidence of this belief than others, the mechanisms by which this belief is inculcated, and the ages at which it appears with differing frequencies in differing social classes. Future theories will have to take into account the differing implications of overweight for the different social classes. It seems quite possible that the lack of success in the control and treatment of obesity stems from the fact that until now physicians have thought of obesity as always being abnormal. This is certainly not true for persons in the lower socioeconomic population. Obesity may always be unhealthy, but it is not always abnormal. Unfortunately, our weight control programs have directed their appeals in a nonspecific way to rich and poor alike. The present study reveals an un- expected opportunity for increasing the selectivity of public health measures for the control of obesity. Would it not be more effective to initiate programs tailored specifically for subgroups of society where obesity is most common? The success of a similar approach has been demonstrated by Johnson et al,’ who studied the epidemiology of polio vaccine ac- ceptance in Dade County, Fla. They pointed up the importance of ethnic background and social class as an index of commonly held beliefs, shared feel- ings, group values and attitudes, and social partici- pation. Utilizing such information in work with a high-risk population (lower class, Spanish-speaking residents of Dade County), they significantly in- creased the percentage of respondents who took polio vaccine over the percentage who had done so in previous campaigns. The present study shows the feasibility of iden- tifying obese populations at high risk. Such identi- fication has generally been a prerequisite for effective public health programs. Recognition of the significance of social factors in obesity may lay the foundation for our first effective public health prograra for the control of obesity. ‘This investigation was supported by a Public Health Service re- search grant from the National Institute of Mental Health. The data used in this study are from the Midtown Manhattan Study of the Department of Psychiatry, Cornell University Med- ical College and the New York Hospital, both in New York. Per- mission lo use the data was granted by Alexander H. Leighton. Director of the Midtown Manhattan: Study, but the contents of this paper represent independent work by the authors. Technical assistance was rendered by several other members of the Midtown Group, especially Leo Srole, PhD, Stanley T. Michael, MD, and Thomas Langner, MD. Financial support for the Midtown Study was provided by the National Institute of Mental Health, the Milbank Memorial Fund, the Grant Foundation, the Rockefeller Brothers Fund, and the Corporation Trust, all in New York. References 1. Moore, ME; Stunkard, A; and Srole, L.: Obesity, Social Class, and Mental Illness, JAMA 181:962-966 (Sept 15) 1962. 2. Srole, I., et al: Mental Health in the Metropolis: Midtown Manhattan Study, New York: McGraw-Hill Book Co., Inc, vol 1, 1962. 3. Langner, T'S. and Michael, ST.: Life Stress and Mental Health: Midtown Manhattan Study, New York: The Free Press of Glencoe, Inc., vol 2, 1963. 4. Perry, L., and Learnard, B.: Obesity and Mental Health, JAMA 183:807-808 (March 2) 1963. 5. New Weight Standards for Men and Women, Statist Bull Metrop Life Insur Co 40:2-3 (Nov-Dec) 1959. 6. Pflanz, M.: Medizinische-soziologische Aspekte der Fett- sucht, Psyche 16:575-591. 1962-1963. 7. Joffe, N.F.: Food Habits of Selected Subcultures in United States, Bull Nat Res Council 108:97-103 (Oct) 1943. 8. Childs, A.: Some Dietary Studies of Poles, Mexicans, Ital- ians, and Negroes, Child Health Buil 9:84.91, 1933. 9. Stout, C, et al: Unusually Low Incidence of Death From Myocardial Infarction: Study of Italian American Community in Pennsylvania, JAMA 188:845-849 (June 8) 1964. 10. Johnson, AL, et al: Epidemiology of Polio Vaccine Ac- ceptance — Social and Psychological Analysis, monograph No. 3, Florida State Board of Health, 1962. 87 Influence of Social Class on Obesity and Thinness in Children Albert Stunkard, MD; Eugene d’Aquili, MD; Sonja Fox; and Ross D. L. Filion, PhD Several social factors have been closely linked to obesity and thinness in adults. This study, based on 3,344 measurements of triceps skin-fold thickness found similar relationships in white urban children. Obesity was far more prevalent in the lower-class girls than in those of the upper class—nine times as prevalent by age 6. Similar though less striking differences were found between boys of upper and lower socioeconomic status. The pattern of thinness among girls was similar to that previously reported in women, with significantly more thinness in the upper-class group. Among boys, as among men, there were no such differences. The remarkably early onset of class-linked differences in prevalence of obesity underlines the importance of attempts to prevent the disorder in childhood. These attempts should be directed particularly toward those at high risk because of their lower socioeconomic status. his report continues our as- sessment of the influence of social factors on obesity in man. In carlier studies, carried out in New York City, we demonstrated a strong inverse relationship between socioeconomic status and obesity. Obesity was six times more prevalent among women of lower than among women of upper sociocconomic status." Correlation between paren- tal socioeconomic status and preva- lence of obesity was nearly as strong, indicating that socioeconomic status was cause as well as correlate (Fig 1). From the Department of Ps sity of Pennsylvania and the cral Hospital, Philadelphia. Dr. Stunkard is now on sabbatical lea t the Center for Advanced Study in the Behavioral Sciences, Stanford, Calif. : Reprint requests to 3400 Spruce St, Phila- delphia 19104 (Dr. Stunkard). We also demonstrated significant in- verse relationships between social mability and obesity, and between number of generations in this coun- try and obesity. In addition, several cthnic and religious variables ap- peared related to the prevalence of obesity.“' Subsequently, we found similar results in a study in London.* In all these investigations, the rela- tionship between social factors and prevalence of obesity among men paralleled that in women but in each instance was less marked. The present study was designed to establish the age at which the influ- ence of socioeconomic status on body weight becomes apparent. We also wanted to delineate the subsequent evolution of the relationship between socioeconomic status and obesity and thinness. Materials and Methods To assess the prevalence of obesity we measured the skin-fold thick- nesses of 3.344 white school children in three Eastern cities. The 11 schools in the study were chosen so as to pro- vide a population of both upper- and lower-class children. The respondent's socioeconomic status was determined on the basis of the father’s occupa- tion, according to Intermediate Occu- pational Classification for Males, a 1950 publication of the Burcau of the Census. The respondents were 5 to 18 years old. We decided on the use of the triceps skin-fold thickness as the best index of obesity for a large field study on the basis of Seltzer and Mayer's ex- tensive work with this measure, ”" as well as the view of Dugdale et al* that it is the best anthropometric measure of adiposity. Furthermore, Shephard et al have presented evi- dence that the triceps skin-fold pro- vides an especially accurate assess- ment of obesity in children and adolescents. To avoid interobserver error, all measurements were made by the same observer (S.F.), using the Lange skin-fold calipers. Reliability coefficient of the measurements was 0.93. : Since there is no generally accepted criterion for obesity in children, we chose two criteria that had heen uti- lized in other studies and that seemed % Obese 88 30 own ses {_] [] ses of Origin re N=291 20 eT) N= N=315 ge? 10 TY y 1] N=315 N=3: 0 Vy Low Medium High (Status) Fig 1.--Decreasing prevalence of obesity with increasing socioeconomic status (SES) among women in an Eastern city. reasonable. The first criterion was the values for skin-fold thickness re- ported by Seltzer and Mayer in their study of Boston school children. They defined as ohese those children whose skin-fold thickness exceeded one standard deviation from the mean for their age and sex. Table 1 shows the minimum triceps skin-fold thickness indicating obesity according to Scit- zer and Mayer.’ Malina has criticized the Seltzer- Mayer criterion as inapplicable to other populations and, indeed, the standard deviation for some age groups in our population differed from that of Seltzer and Mayer. Ac- cordingly, we subjected our data to a second criterion of obesity. We also defined as obese the 10% of cach sex _in the total population that had the thickest skin-folds; and we used the minimum skin-fold thickness of this group to define obesity within each age group. Hampton et al effectively used a similar percentile criterion to define obesity and leanness.” In fact, . according to Dr. Joseph Brozek, the percentile criterion is favored by many physical anthropologists, in part, at least, because it has the ad- vantage of showing that obesity in- creases with age, a trend that is ob- vious in the raw data. Furthermore, by using percentiles we were able to define thinness in children, for whom no such standard is now available. We defined as thin the 10% of cach sex with the thinnest skin-folds and ana- lyzed the data as for the obese group. These empirically derived values for obesity were 23 mm for girls and 18 mm for boys. For thinness they were girls, 8 mm, and boys, 6 mm. In the course of studying the 3,344 white children we measured also the skin-fold thicknesses of 1903 black and Puerto Rican children. Since blacks, Puerto Ricans, and whites have different distributions of skin- fold thickness, it was not possible to analyze all 5247 respondents as a single population. Hampton et al had also found significant differences in anthropometric measurements among teen-agers of different racial origins and cautioned against using the same standards for different races. Furthermore, the small num- ber of upper-class blacks and Puerto Ricans made it impossible to run sep- arate analyses of blacks and Puerto Ricans relating socioeconomic status to obesity. For these reasons the anal- vsis reported here was confined to the 3,344 white respondents. Of these, 2310 were classified in the upper socioeconomic status (occupational categories I and II by 1950 Burcau of Census listing); 857 were classified as of lower socioeconomic status (occu- pational categories III and IV); the remaining 167 could not be clearly classified. Table 2 shows the number of respondents at each age according to socioeconomic status and sex. Results We found marked differences in the prevalence of obesity between the up- per- and lower-class children. More- over, these differences were apparent by age 6. Obesity in Girls.—Figure 2 shows the relationship between socioeconom- ic status and prevalence of obesity for girls, using the Seltzer-Mayer crite- rion. At age 6, 29% of the lower-class girls were obese as compared with only 3% of the upper-class girls. This class-linked difference continued through age 18, but fell to a minimum at age 12, when 19% of lower-class and 9% of upper-class girls were obese. Table 3 shows the four-fold contingeney table relating high and lower social class obesity or its ab- sence. When we applied the percentile eri- terion, we also demonstrated the marked difference in the prevalence of obesity between social classes (X* = 70.838, P < 0.001). At age 6, the lower socioeconomic group contained 8% obese girls, while the upper-class group had no obese girls at either age 6 or 7. This difference was maintained until age 18, as with the Selizer- Mayer criterion. In addition, the per- centile criterion demonstrated an in- crease in the prevalence of obesity as a function of increasing age in both socioeconomic groups. Figure 3 shows further that the slopes for the upper and lower classes differ, with a greater yedrly increment in the per- centage of obese in the lower class. Obesity is not only more prevalent among poor girls, but this greater prevalence is established earlier and increases at a more rapid rate than among upper-class girls. Ohesity in Boys.—Lower-class boys showed a greater prevalence of ohe- sity than did those of the upper class, although here the differences were not as striking as among the girls. Figure 4 shows the data for boys as analyzed by the Seltzer-Mayer crite- rion. At age 6, a marked difference between the two socioeconomic groups is already established, with 40% of the lower socioeconomic group classified as obese, compared with 25% of the upper-class group. Unlike the pattern among the girls, however, the difference between the boys is not continuous to age 18. Note the rever- sal at age 12, when the upper-class group has a greater percentage of obese. But by age 14 the lower-class group again shows a greater preva- lence of obesity, and this difference is maintained until age 18. These data are summarized in Table 4. Figure 5 shows the data for boys analyzed by the percentile criterion. Although the profile differs from that in Fig 4, where the Seltzer-Mayer cri- terion was utilized, the basic trend is 89 White Girls a Classes 1 & 41 £ ° 504 © Classes lit & tv 404 204 % Obesity (Seltzer-Mayer Cri 8 rr TTT, 6 8 10 12 14 16 18 Age (Years) Fig 2.—Sociveconomic status and obesity by social class among girls (Seltzer-Mayer criterion). Lower class girls show far higher prevalence than upper class girls, especially during younger years. Fig 4.—-Socivecconomic status and obesity by social class Jor boys (Seltzer-Mayer criterion). Lower class boys show greater prevalence of obesity than upper class boys, but differences are less striking than among girls. White Boys o Classes! & ll 504 eo Classes! &iv Ty % Obesity (Seltzer-Mayer Criterion) 8 8 10 12 14 16 18 Age (Years) White Girls 35 304" aClassesi & ll Ez g . 5 © Classes fl & V { Cc 254 2 § 8 xf \ = > ® 8 15 # 10 ‘ 54 0 fp rp—p—— 6 8 10 2? u 16 8 Age (Years) Fig 3.—Sociocconomic status and obesity by social class (percentile criterion). Apparent increase in prevalence of obesity with age probubly reflects physiclogical facts. Fig 5.—Sociocconomic status and obesity by social class for boys (percentile eriterien). Write Boys 35 4 4 Classes 1 & II 30 @ Classes ili & IV z 2 2 SG 251 > 8 & 20 ge 2 2 & 15 4 & 10 4 5 Bieler p————————— 6 8 10 12 4 16 18 Age (Years) similar (X*=40.439, /’<0.001). Once again a significant differenced he- tween social classes is apparent by age 6. This inverse relationship be- tween social status and obesity is maintained, except for the previously noted reversal at age 12, Our earlier studies had shown a positive correlation between socio economic status and prevalence of thinness among women. We found four times as many thin women among those of high status as among | "obese child magnifies the over-all those of low status. In the present study, applying the percentile crite- rion, we found a similar pattern among girls. Figure 6 shows that there was more Jeanness among girls of upper socioeconomic status. At age 6, 15% of the upper-class girls were thin as compared to only 4% of the lower-class. This difference continued until age 12, at which point the two groups converged and showed de- creasing prevalences of thinness. Table 5 shows the relationship be- tween high and low socioeconomic status and leanness or its absence. Thinness in Boys.—Our carlier studies had shown no association be- Lwecen socioeconomic status and lean- ness among men. About 10% of cach group was lean. The data on boys sim- ilarly failed to show such an associa- tion. Figure 7 demonstrates this ab- sence of any clear trend. Comment During the past ten years, we have learned a great deal about obesity in the United States, and the results have been a surprise. Our conception of the nature of obesity, based in large part on the results of treating ~ members of the upper and middle classes, has been shaken by the dis- covery that obesity is largely a prob- lem of the lower classes. It now ap- pears that socioeconomic status and related social factors have more to do with determining whether a person will be obese than does individual psychopathology." The implications of these findings are far reaching. For one, they suggest that we need not be constrained by current psy- chodynamic formulations of obesity and their pessimistic outlook for treatment, when dealing with the group most afflicted with the dis- order. Instead, an educational ap- proach that recognizes the impor- 90 tance of social factors and is designed to influence the values and Jif of large groups, may he more : Pro- priate and more effective than con- ventional tech- niques. The study reported here extends previous work and defines a discrete and partienlarly vulnerable group within the high-risk population-the children of the poor. Not only is child- hood obesity a major problem in its own right, but the prognosis for the psychotherapeutic problem for obese children liccome obese adults. The most authoritative estimate is that 85% of obese children follow this course.” Furthermore, the odds against an obese child hecoming a normal-weight “adult, which are more than 4:1 at age 12, rise to 28:1 if weight has not been reduced by the end of adolescence." Recent research into juvenile-onset obesity in laboratory animals adds to these actuarial cautions a possible ex- planation and a cause for further con- cern. In rats, the cellularity of adipose tissue, and consequently its lipid stor- age capacity, is determined very early--probably during the first three wecks of life-and primarily by the animal's level of food intake.’ Over- nutrition during this critical period leads to marked increase in the cellu- larity of adipose tissue, increased hody size, and obesity; undernutrition has the opposite effect. Nor do changes in diet after infancy have any effect on the number of adipose cells. Caloric restriction reduces weight solely by reducing the lipid content of these cells, often to an ab- normally low level. The depleted cells then remain ready to return lo their initial levels of adiposity whenever sufficient lipids are available. By eon- trast, adult-onset obesity is produced by cellular hypertrophy, and weight loss returns the adipose tissue cells to a more normal size. Although data of comparable pre- cision for man are lacking, available information strongly suggests a sim- ilar pattern.” And the mechanism described offers a convincing biologic explanation for the remarkable tend- eney of juvenile-onset obesity to persist into adult life. We still do not know precisely what period in human development corresponds to the crit- ical first three weeks in the rat's life. However, the fact that such a large percentage of lower- children is obese hy age 6 suggests that hyper- cellular adipose tissue accounts for at least part of this increased incidence of obesity. While we know that childhood obe- sity tends te persist into adulthood, we do not know what proportion of obese adults has juvenile-onset obesi- ty. Even if the contribution is not greater than the one third that has been suggested, however, juvenile- onset obesity remains a serious prob- lem. For all the pathologic correlates and sequelae of obesity are more prevalent and more severe in adults with juvenile-onset obesity, from dia- betes and atherosclerosis to emotional disturbance. Furthermore, the juve- nile-onset obese have special prob- lems. Any psychopathology is likely to be related to obesity, and fully half of persons with juvenile-onset obesity suffer from body-image disturb- ances” In persons with adult-onset obesity, on the other hand, psycho pathology is usually coincidental and disturbance in body image rare. One aspect of these findings de- serves special note. The prevalence of thinness among the lower-class chil- dren was very low despite the poverty in which they lived. This finding sur- prised us, coming as it did in the midst of reports of widespread hun- ger among the poor, and its signifi- cance is unclear. Perhaps skin-fold calipers failed to detect evidence of undernutrition. However, they have proved adequate to the task of assess- ment of the undernutrition of ano- rexia nervosa. Perhaps the level of poverty associatied with frank under- nutrition is lower than that of the children we studied. But we sought out the children with the lowest level of socioeconomic status that could be found in New York, Philadelphia, and Wilmington. It may be that white children, at least, in these cities do not suffer from undernutrition. Conclusion.—Before we can in- stitute effective measures to prevent or treat a disorder, it is helpful to de- fine the population at high risk. This all-important step has now been taken for obesity. The lower socio- economic class is the one with by far the greatest prevalence of obesity. Some of the preventive and therapeu- tic measures that this finding sug- 91 : - White Girls White Boys 35 4 . 35 4 a Classes 1 & 1 ¢ Classes 1 & Il _ 30] eCasesmaw © 304 eClassesmaw 5 - § : 5 25 4 > 25 4 ° 2 z= € g 20 < 20 4 : f § 15 3 154 3 - 2 *® 10 4 10 54 5 4 0 0 food TTY 6 8 10. 12 14 16 8 6 10 12 14 16 18 Age (Years) Age (Years) Fig 6.—Socioeconomic status and thinness by social class for girls (percentile criterion). Upper class girls show greater prevalence of leanness until age 12 when condition essentially disappears for both groups. Table 1.—Obesity Standards in White Americans According to Seltzer and Mayer Minimum Triceps SkinFeld Siskasss Indicating Obesity mm imam — 12 a gests have already been described.’ We have now taken a second step and pinpointed a discrete population at particularly high risk—the children of the poor. As early as age 6, the prevalence of obesity is far higher among the lower classes, particularly among girls, than it is among those of 34-621 O ~ 74 = 7 higher socioeconomic status. Further- more, application of the percentile criterion, as illustrated in Fig 3, dem- onstrates that obesity is not only more prevalent in poor girls, but this prevalence is established carlier and increases at a more rapid rate than among upper-class girls. Fig 7.—Sociocconomic status and thinness by social class for boys (percentile criterion). Lack of association apparent. Table 2.—Number of Respondents in Each Age Group by Socioeco- nomic Status (SES) and Sex Upper SES Lower SES Pr ey 5 These findings help define our task, and they should encourage us in the fight against obesity. For the remark- ably early age at which obesity be- gins among so many of the poor be- speaks faulty nutritional practices by parents. We do not yet know to what extent these faulty nutritional prac- Table 3.—Distribution of Obesity by Socioeconomic Status (SES) (Girls)* Upper SES __ Lpwer SES __ 93 Nonobese sar *X? = 81.367, P <0.001. Table 4.—Distribution of Obesity by Socioeconomic Status (SES) (Boys)* Upper SES Lower SES Obese 187 00 ~ Nonobese 1,269 *X? = 37.210, P <0.001. Table 5.—Distribution of Thinness by Socioeconomic (SES) Status* 92 nutrition among the poor—nutritional misinformation or economic depriva- tion—is sorely needed. We have re- cently embarked upon such a project. One final note—Despite the poverty in which our lower-class children lived, they were no more likely to be thin than were the upper-class children, This investigation was supported in part by a research grant MH-15383-03 from the National + Institute of Mental Health. References 1. Goldblatt PB, Moore ME, Stunkard AJ: So- cial factors in obesity. JAMA 192:1039-1044, 1965. 2. Moore ME, Stunkard A, Srole L: Obesity, so- cial class, and mental illness. JAMA 181:962-966, 1962. 3 Stunkard Al: Environment and stesiy; Re- t in our Upper SES ___ Lower SES Thin 88 28 Nonthin 700 387 *X? = 6.078, P <0.02. tices result from lack of information or from lack of appropriate food. An cffective program of obesity control among poor children requires that we distinguish between these two causes. Research that will enable us to deter- mine more precisely what causes poor cent regu- lation of food intake in man. Fed Proc n: 1367- 1373, 1968. 4. ‘Stunkard AJ: Obesity, in Freedman AM, Kaplan HI (eds), Comprehensive Textbook of Psy- chiatry. Baltimore, Williams & Wilkins Co, 1967, Pp 1059-1062. 5. Silverstone JT, Gordon RP, Stunkard AJ: Sacial factors in obesity in London. Practitioner 202:682-G88, 1969. 6. Seltzer CC, Goldman RF, Mayer J: The tri- cops skinfold as a predictive measure of body density and body fat in obese adolescent girls. Pediatrics 136:212-218, 1965. 7. Seltzer CC, Mayer J: A simple criterion of obesity. Postgrad Mcd 38:A101-107, 1965 8. Mayer J: Some aspects of the problem of regulation of food intake and obesity. New Eng J jd 274: pr 1966. 9. Seltzer CC, Mayer J: Greater reliability of the triceps skinfold over the subscapular skinfold as an index of obesity. Amer J Clin Nutr 20:950- 453, 1967. 10. Dugdale AE, Chen ST, Hewitt G: Patterns of growth and nutrition in childhood. Amer J Clin Nutr 23:1280-1287, 1970. 11. Shephard RJ, Jones G, Ishii, ct al: Factors affecting body density and thickness of sub- cutancous fal. Amer J Clin Nutr 22:1175-1189, 1969. 2. Malina RM: Patterns of development of skinfolds of negro and white Philadelphia chil- dren. Human Biology 38:89-103, 1965. 13. Hampton MC, Hueneman RL, Shapiro LR, ct al: A longitudinal study of gross body com- position and body conformation and their associ- ation with food and activity in a Ween-age popu- lation. Amer J Clin Nutr 19:422-435, 1966. 14. Holland J, Masling J, Copley D: Mental ill- ness in lower class normal, obese and hyperobese women. Psychosom Med 32:351-357, 1970. 15. Abraham S, Nordsieck M: Relationship of excess weight in children and adults. Public Health Rep 75:263-213, 1970. 16. Stunkard AJ, Burt V: Obesity and the body image: 11. Age at onset of disturbances in the body image. Amer J Psychiat 123:1443-1447, 1967. 17. Knittle JL, Hirsch J: Effect of early nutri- tion on the development of rat epididymal fat Cellularity and metabolism. J Clin Invest 91-2098, 1968. 18. Salans LB, Knittle JL, Hirsch J: Role of adipose cell size and adipose tissue insulin sensi- tivity in the carbohydrate intolerance of human obesity. J Clin Invest 47:153-165, 1968. 19. Hirsch J, Knittle JL: Cellularity of obese and nonobese human adipose tissue. Fed Proc 29:1516-1521, 1970. 20. Stunkard AJ, Mendelson M: Obesity and body image: 1. Characteristics and disturbances in the body i e of some obese persons. Amer J Psychiat 123:1296-1300, 1967. 93 Effects of Weight Reduction on Obesity STUDIES OF LIPID AND CARBOHYDRATE METABOLISM IN NORMAL AND HYPERLIPOPROTEINEMIC SUBJECTS JerroLD OLEFsKY, GERALD M. REAVEN, and Jorn W. FARQUHAR From the Department of Medicine, Stanford University School of Medicine, Stanford, California 94305 and Veterans Administration Hospital, Palo Alto, California 94304 ABsTrAcT Considerable controversy exists over the purported role of obesity in causing hyperglycemia, hyperlipemia, hyperinsulinemia, and insulin resistance; and the potential beneficial effects of weight reduction remain incompletely defined. Hypertriglyceridemia is one of the metabolic abnormalities proposed to accom- pany obesity, and in order to help explain the mecha- nisms leading to this abnormality we have proposed the following sequential hypothesis: insulin resistance => hyperinsulinemia — accelerated hepatic triglyceride (TG) production => elevated plasma TG concentrations. To test this hypothesis and to gain insight into both the possible role of obesity in causing the above metabolic abnormalities and the potential benefit of weight reduc- tion we studied the effects of weight loss on various aspects of carbohydrate and lipid metabolism in a group of 36 normal and hyperlipoproteinemic subjects. Only weak to absent correlations (7 =003 — 0.46). were noted between obesity and the metabolic variables mea- sured. This points out that in our study group obesity cannot be the sole, or even the major, cause of these abnormalities in the first place. Further, we have ob- served marked decreases after weight reduction in fast- ing plasma TG (mean value: pre-weight reduction, 319 mg/100 ml; post-weight reduction, 180 mg/100 ml) and cholesterol (mean values: pre-weight reduction, 282 mg/100 ml; post-weight reduction, 223 mg/100 ml) levels, with a direct relationship between the magnitude of the fall in plasn« lipid values and the height of the initial plasma TG level. We have also noted significant Dr. Olefsky is a Research ccd Education Associate, Veterans Administration. Dr. Reaven is a Medical Investig. ., Ver. n3 Admin- istration. Received for publication 16 March 1%. «2 n revised form 31 August 1973. decreases after weight reduction in the insulin and glucose responses during the oral glucose tolerance test (37% decrease and 129, decrease, respectively). In- sulin and glucose responses to liquid food before and after weight reduction were also measured and the overall post-weight reduction decrease in insulin re- sponse was 48% while the glucose response was rela- tively unchanged. In a subgroup of patients we studied both the degree of cellular insulin resistance and the: rate of hepatic very low density (VLDL) TG produc- tion before and after weight reduction. These subjects demonstrated significant decreases after weight reduc- tion in both degree of insulin resistance (33% decrease) and VLDL-TG production rates (40% decrease). Thus, weight reduction has lowered each of the antecedent variables (insulin resistance, hyperinsulinemia, and VLDL-TG production) that according to the above hypothesis lead to hypertriglyceridemia, and we believe the overall scheme is greatly strengthened. Furthermore, the consistent decreases in plasma TG and cholesterol levels seen in all subjects lead us to conclude that weight reduction is an important therapeutic modality for pa- tients with endogenous hypertriglyceridemia. INTRODUCTION It has been suggested that obesity has an important causal role in the development of hyperglycemia (1), hyperlipemia (2, 3), hyperinsulinemia (4), and insulin resistance (5). However, considerable controversy still exists over the role of obesity in causing the above meta- bolic abnormalities (6-10), and the mechanisms in- volved remain quite unclear. Furthermore, although it it commonly accepted that weight loss by obese subjects ameliorates these abnormalities (11-12) there are sur- prisingly little data to support this widespread belief. Cob : Ficure 1 : repr jon of a hyperinsulinemia and subsequent hypertriglyceridemia. Most of the reports to date have involved only a few patients (13-18), and have not examined a wide spec- trum of variables (14, 15, 18, 19). Furthermore, even if weight loss does lead to a decrease in hyperglycemia, hyperlipemia, hyperinsulinemia, and insulin r ial h th y is relating insulin resistance to tion, then the hypothesis would be strengthened, whereas, if one observed a dissociation in the direction of the pre- dicted changes, the argument would be weakened. To test our hypothesis in this manner, and to gain clinical h ible beneficial effects of weight loss, this does not necessarily mean that obesity was the sole cause of these metabolic abnormalities in the first place. For these reasons, and because hyperglycemia (20) and hyperlipemia (21, 22) have been proposed as inde- pendent risk factors for the development of coronary heart disease, it seemed important to evaluate the rela- tionship between obesity and various aspects of carbo- hydrate and lipid metabolism in a large group of normal and hyperlipoproteinemic subjects with varying degrees of obesity; and then to study the effects of weight loss on these variables in the same group of patients. Plasma triglyceride (TG)® concentration is one of the metabolic variables we have measured. To help explain the mechanism leading to hypertriglyceridemia, we have previously proposed a sequential hypothesis (Fig. 1) that states that tissue resistance to insulin- mediated glucose uptake is a common underlying finding in most patients with endogenous hypertriglyceridemia, and that in an effort to maintain glucose homeostasis, insulin-resistant subjects secrete increased amounts of insulin (23, 24). We have also suggested that this hy- perinsulinemia may act upon the liver to accelerate very low density lipoprotein (VLDL) TG production rate, which in turn leads to endogenous hypertriglyceridemia (10, 24, 25). The term “endogenous hypertriglyceri- demia” corresponds to the primary hyperlipoprotein- emias recently defined as types IIb, III, and IV (26). A way to help validate a three-part sequential hypothe- sis such as this would be to perturb the system and mea- sure the effects of this perturbation on each step of the sequence. If all the changes were in the predicted direc- * Abbreviations used in this paper: SSPG, steady state plasma glucose levels (degree of insulin resistance); TG, triglyceride; VLDL, very low density lipoprotein. insight into the p we have studied the metabolic effects of weight reduc- tion in 36 patients. METHODS Subjects 36 subjects selected from Stanford’s Nutrition and Metabolism Clinic underwent a mean weight loss of 10.9 kg (range 9.1-14.2 kg). The only criteria for selection was the subject's acquisition of more than 10 kg in body weight since age 20 and the subjects willingness to undertake a weight reduction program. Tables I and II list the clinical characteristics and metabolic data of our study group before and after weight reduction. None of our patients were massively obese and the degree of obesity of these patients is similar to the degree of moderate obesity commonly encountered by a physician. By weight history all of them had acquired the bulk of their excess poundage in adult- hood. Percentage adiposity was determined according to the anthropometric technique of Steinkamp et al. (27). With this method the mean percentage adiposity of normal adults of this age group =SD is 27%=*7% for men and 31%*10% for women. The mean pre-weight reduction per- cent adiposity of our male subjects was 32.0% with a range of 23.5%-48.1% and the mean percent adiposity of our female subjects was 40.9% with a range of 36.5%- 48.8%. Relative weight was determined by dividing a patient’s weight by his “ideal weight” as determined accord- ing to the Metropolitan Life Tables. The mean pre-weight reduction relative weight of our study group was 1.21 with a range of 1.00-1.76. We do not wish to imply that a per- cent adiposity of 27% for men or 31% for women is “ideal”: it is merely average, and we consider the “aver- age” American adult to be somewhat overweight. Other than chemical diabetes, as defined by an abnormal oral glucose tolerance test without fasting hyperglycemia (28), no subject had any disease state or was ingesting any drug known to affect carbohydrate or lipid metabolism. According to the lipoprotein classification system pub- lished by the World Health Organization (26), 4 of our patients ‘were type IIa, 6 were type IIb, 3 were type III, 17 were type IV and 6 were normal. In this study, nor- TasLE | Clinical Characteristics and Melabolic Data in 36 Patients before Weight Reduction Li precgn Glucose* Insulin* Fasting Fasting Insulin Insulin electro- response response TG cholesterol ~~ VLDL-TG resistance "espanse Patient Body Relative phoretic during during concen- concen- production SSPG to no. Age Sex weight weight Adiposity pattern OGTT OGTT tration tration rate level formula kg % area U area U mg/100 ml mg/100 ml mg/kg/h mg/ 100 ml area U 1 33 M 91 1.17 25.31 wv 381 157 212 245 — — —_ 2 59 M 87.1 1.12 23.45 Iv 414 267 306 — 11.90 — — 3 53 M 118.9 1.57 48.10 1b 587 376 228 254 — — we 4 51 M 96.3 1.20 40.30 Iv 425 250 357 242 —_ 307 357 5 58 M 76 1.00 26.63 wv 496 250 279 324 —_ — — 6 50 F 75.7 1.36 40.85 HI 356 391 906 408 -— — — 7 26 F 70.4 1.06 36.47 N 542 223 88 203 10.30 210 174 8 36 F 81.5 1.31 42.40 1b 399 260 268 342 —_— — — 9 27 M 823 1.15 24.14 lib 424 299 159 274 11.60 -_ - 10 39 M 100 1.36 28.30 11 522 426 862 571 — — — 11 25 M 90.6 1.31 31.88 1v 393 350 214 253 16.90 257 240 12 55 M 88.4 1.16 29.08 II 493 408 147 343 — 300 258 13 45 M 74 1.02 29.76 1b 326 308 181 295 — — —_ 14 43 M 89.4 1.02 27.91 v 464 341 204 220 -_ —_ — 15 61 M 106.8 1.18 39.47 wv 469 507 352 326 23.90 270 510 16 53 M 104 1.17 33.79 N 398 327 136 198 10.60 210 291 17 61 F 71.7 1.11 34.02 Ib 476 352 379 349 17.60 226 213 18 49 M 94.2 1.07 28.95 1ib 345 156 177 286 — — —_ 19 48 M 82.5 1.13 38.60 1 310 145 130 343 — — — 20 46 M 84.9 1.04 24.02 wv 361 156 308 278 —_— — — 21 41 M 79.4 1.09 28.19 1v 433 330 171 260 —_— — — 22 42 F 70.5 1.16 44.10 1v 517 574 582 231 — —_ — 23 58 M 78.2 1.09 30.30 1b 361 189 254 276 — — — 24 54 M 94.4 1.25 29.68 1v 568 357 349 206 18.60 276 258 23 44 M 96.1 1.46 41.30 wv 414 672 563 251 16.00 — —_ 26 45 F 85 1.35 41.90 Iv 459 367 108 228 10.80 107 213 27 46 M 97.5 1.27 —-— v 345 487 479 269 20.70 —_ — 28 48 M 115.6 1.30 34.20 wv 429 479 816 248 26.40 —_ - 29 50 M 97.5 1.27 30.00 wv 583 303 625 —_ — — — 30 25 M 87.0 1.20 — 11 382 687 326 429 —_ — — 31 24 M 125.7 1.76 39.80 N 420 364 92 210 — 304 — 32 41 M 86.5 1.07 26.14 wv 537 538 707 269 —_— — — 33 41 F 102.8 1.60 48.82 N 315 309 9 201 — — — 34 41 F 74 1.10 38.94 II 267 157 143 283 _— — — 35 59 M 81.4 1.15 —_— mv 410 183 175 242 -— 137 — 36 69 M 82.5 1.11 33.61 N 606 193 134 231 9.64 — —_— * Area under the plasma response curve during the 3-h oral glucose tolerance test (COGTT). G6 96 TasLe II Clinical Characteristics and Melabolic Data in 36 Patients after Weight Reduction Glucose* Insulin* VLDL-TG Insulin response response Fasting Fasting produc- resistance Insulin Patient Body Relative during during TG cholesterol tion SSPG to response no. ‘weight weight Adiposity OGTT OGTT concentration concentration rate level formula kg % area U area U mg/100 ml mg/100 ml mg/kg/h mg/100 mi area U 1 78.8 0.94 20.93 349 87 73 199 —_— —_— p— 2 74.0 0.85 15.79 329 105 139 — 2.50 — — 3 109.2 1.44 40.69 448 336 142 246 -_— —— -_— 4 83.2 1.12 36.01 392 123 157 198 -— 134 108 5 63.5 0.83 21.72 412 159 132 312 ee] -_— — 6 66.6 1.20 30.17 324 185 483 280 — -_ — 7 60.0 0.91 18.29 503 195 65 181 7.26 61 108 8 7.4 1.17 38.70 466 207 134 200 — — -— 9 73.1 0.99 23.35 424 177 81 181 12.50 -— — 10 88.4 1.20 29.27 324 277 230 155 -— — — 11 78.9 1.14 30.95 352 240 269 243 11.59 220 159 12 77.4 1.02 28.61 366 169 185 276 — 145 132 13 67.4 0.89 26.33 287 226 108 279 —_— _— — 14 78.2 0.89 23.46 318 169 168 168 -— — — 15 91.7 1.08 36.79 400 140 222 255 14.71 279 243 16 91.5 1.03 30.05 391 281 128 187 8.70 121 219 17 68.1 0.97 31.29 426 213 260 284 9.23 175 81 18 84.5 0.97 25.14 366 264 145 204 Ed - -_— 19 71.4 0.98 33.15 377 13 9 240 — -— 20 75.6 0.88 23.35 428 105 203 201 _ -— -— 21 70.1 0.96 25.41 335 266 120 269 -_— -_— — 22 61.3 1.05 38.62 480 337 355 255 -_— -_— — 23 68.6 0.95 27.63 319 143 187 259 _— —_ — 24 83.8 1.10 30.45 47 153 138 176 13.32 144 150 25 82.9 1.26 35.90 361 389 328 199 15.50 — — 26 75.3 1.17 38.66 472 429 103 263 12.22 - 162 27 83.6 1.08 — -_— — 126 196 20.30 — — 28 101.5 1.13 30.80 363 331 253 187 - 6.20 — - 29 83.5 1.07 31.60 486 288 392 — -— -— — 30 7.7 1.04 — 357 139 233 253 — -_— — 31 111.5 1.60 32.91 402 144 90 162 -— 261 — 32 77.4 0.87 22.31 383 305 3n 230 — — — 33 93.6 1.46 44.12 361 288 64 200 — foe 34 62.3 0.93 34.17 261 147 102 274 — _— — 35 72.2 1.00 — 382 142 106 186 -— 77 - 36 na 0.92 28.49 562 150 93 183 7.60 - — * Area under the plasma response curve during the 3-hr oral glucose tolerance test (OGTT). mality was defined as a plasma TG level less than 150 mg/ 100 ml and a plasma cholesterol level less than 250 mg/100 ml, and a classification of Type III was confirmed by ultracentrifugation. Subjects were studied and managed either entirely on an outpatient basis, or on a combined outpatient and inpatient basis. The outpatient study is de- scribed first. - Experimental protocol After a careful dietary history and collection of anthro- pometric measurements, 23 of the 36 subjects were placed on a solid food weight-maintaining diet designed to mimic the typical diet consumed in this country (29, 30) (ap- proximately 15% protein, 43% carbohydrate, and 42% fat with a polyunsaturated: saturated ratio of 0.21). At the end of 1 wk of this weight maintenance period, an oral glucose tolerance test was performed by administering 40 g glucose/m® body surface area. Plasma glucose and insulin were determined at 0, 30, 60, 120; and 180 min. After an overnight fast, plasma TG and cholesterol were measured on at least two separate occasions during the 2nd wk of the weight maintenance period. Plasma lipoprotein electro- phoresis was performed on one of these two samples. Once these data were collected, subjects were placed on a hypo- caloric diet containing 600-1,600 kcal/day. The percentage by calories of carbohydrate, protein, and fat was not changed, only the total amounts consumed. Furthermore, the degree of physical activity was not appreciably changed in any of these patients. After approximately 11 kg of weight had been lost (the period of weight loss was quite variable: the range was from 2 to 10 mo, with a mean of 4.3 mo), each subject was placed again on his original weight-maintaining diet for 2 wk. During this 2-wk period no subject's weight changed by more than 3%, and during the last 10 days of this period no subject’s weight changed by more than 1.5%. At the end of the 2nd wk, the anthro- pometric measurements, oral glucose tolerance test, fasting TG and cholesterol determinations (again, on at least two separate occasions), and plasma lipoprotein electrophoresis were repeated. This study design attempted to isolate weight loss as the only independent variable. An additional 13 patients were studied in a similar manner, except that the pre- and post-weight reduction studies were performed while the subjects were hospitalized in the metabolic ward setting of the Stanford General Clini- cal Research Center. Pre-weight reduction degree of obesity and amount of weight lost were comparable with the out- patient group. During the hospitalization, these subjects consumed a weight maintenance liquid formula diet con- taining 35 kcal/kg body weight/day. The caloric break- down was: 43% carbohydrate, 15% protein, and 42% fat with a polyunsaturated: saturated ratio of 0.21. This diet was also designed to closely approximate the content of a diet typical for the United States (29, 30). The formula provided 251 mg cholesterol for every 1021 cal and was consumed daily in four equal feedings. Each feeding was ingested evenly and slowly over a 30-min period. Oral glu- cose tolerance test, fasting plasma TG and cholesterol measurements, and plasma lipoprotein electrophoresis were performed as was done in the outpatient subjects. Addi- tional studies were performed on these inpatient subjects as follows: Studies of glucose and insulin responses to formula. On two separate occasions during each hospitalization, plasma was sampled for insulin and glucose just before the 11 a.m. feeding, and again at 12, 1 and 2 p.m. The data are expressed as the mean values of these two studies for each subject. VILDL-TG production rate studics. This was done ac- cording to a previously reported method (25, 31), which is briefly summarized as follows: [*H]glycerol is injected intravenously and becomes incorporated into plasma VLDL- TG molecules. The specific activity decay curve of the endogenously labeled VLDL-TG is measured, and from this curve, the fractional loss rate of the labeled VLDL-TG is calculated. If one assumes plasma volume to be 4.5% of body weight, then the VLDL-TG pool size can be calcu- lated according to the following formula: VLDL-TG pool size = 4.5% X kg body weight X VLDL-TG concentration. Finally, the product of the fractional loss rate and VLDL- TG pool size equals the VLDL-TG removal rate (mg/ kg/h). Since the VLDL-TG concentration is in a steady state throughout the study, the VLDL-TG removal rate is equal to the VLDL-TG production rate (25, 31). Estimate of insulin resistance (or impedance). The method for estimating this variable has been previously reported (23), and the insulin resistance measured by this technique should not be confused with the type of insulin resistance sometimes found in insulin-treated diabetics who develop excessive amounts of anti-insulin antibodies. For this reason we have proposed the term impedance to de- scribe this type of inculin-resistant state and we will use these two terms interchangeably in this report. This study is performed by simultaneously infusing constant amounts of crystalline pork insulin (50 mU/min), glucose (6 mg/ kg/min), epinephrine (6 wg/min), and propranolol (0.08 mg/min). The infusion is begun 5 min after a 5-mg loading dose of propranolol, and is constantly administered via a Harvard pump (Harvard Apparatus Co. Inc, Millis, Mass.) over a period of 150 min. Steady state plasma con- centrations of insulin and glucose are achieved within 90 min after the start of the infusion, and we then measure these levels every 10 min for an additional 60 min. Insulin resistance is expressed as the mean of the seven steady state plasma glucose levels (SSPG). This approach is based upon the known ability of epinephrine and propranolol to suppress endogenous insulin secretion (32). Confirmation of this action under these particular experimental conditions has been obtained by finding no rise in plasma endogenous insulin levels during extreme hyperglycemia after an in- fusion of glucose, epinephrine, and propranolol (23). Dur- ing this study, comparable steady state plasma levels are achieved in all subjects. (Mean steady state plasma insulin level, 102 uU/ml; coefficient of variation, 10%.) Thus, we are able to measure the ability of closely similar circulating levels of exogenous insulin to promote disposal of com- parable glucose loads in a variety of subjects. If one as- sumes that endogenous glucose production is inhibited dur- ing this steady state period, then the glucose uptake rate should be equal to the glucose infusion rate. We have verified this assumption by directly measuring the irre- versible glucose loss rate (23), and have found it to be identical to the rate of glucose infusion. Thus, since in- sulin concentration and glucose uptake (infusion rate) are the same for all subjects at the steady state, the height of the SSPG response is a direct reflection of a subject’s overall efficiency of insulin-mediated glucose up- take. It should be pointed out that the degree of insulin resistance measured by this technique is a result of the combined efficiencies of glucose uptake of each tissue ac- tively transporting glucose. Thus, this estimate does not provide information as to the relative degrees of insulin resistance of any particular tissues. Analytical methods Samples for plasma glucose were collected in potassium oxalate sodium fluoride tubes and measured by the Techni- con AutoAnalyzer (Technicon Instruments Corp. Tarry- town, N. Y.), by the ferricyanide method of Hoffman (33). Plasma insulin was measured according to the double- antibody immunoprecipitation techniques of Hales and Randle (34). Cholesterol was determined according to the N-24 AutoAnalyzer method (Technicon Instrument Corp.) (35) on a Folch extract of plasma, and plasma TG was measured by to the chromatropic acid method of Carlson and Wadstrom (36) after acid hydrolysis with a dilute solution of H.SO.. Lipoprotein electrophoresis was per- formed according to the agarose gel method of Noble (37). Statistical methods T tests were performed by the two-tailed, paired Stu- dent’s t test (38). Simple correlation coefficients were cal- culated by the product-moment correlation method (38). Partial correlation coefficients were calculated by the use of multiple regression analysis (38). For correlation analy- sis the insulin and glucose responses to both formula and oral glugose were expressed as total area under the re- sponse curve. RESULTS The correlation coefficient between relative weight and percent adiposity was 0.63 for men and 0.81 for women. While these correlations are obviously highly significant, the deviations from perfect correlations indicates that these two estimates are measuring somewhat different things. Consequently, in an effort to understand the re- lationships among obesity, insulin resistance, and plasma levels of insulin, glucose, TG, and cholesterol, we at- tempted to quantify, within the entire study group, the impact of obesity as estimated either by percent adi- posity or relative weight on each of the metabolic 98 TasLE III Product- Moment Correlation Coefficients (r) between Percent :1diposily and Several Metabolic Variables in Men and Women before Weight Loss* Insulin Glucose Fasting area area VLDL-TG \ Fasting Insulin insulin during during Fasting production cholesterol resistance level OGTT OGTT TG level rate level Men 0.36 (7) 0.39 25)t 0.19 (25) 0.10 (25) —0.03 (25) 0.43 (9) —0.17 (23) Women — —-0.03 (8) 0.30 (8) —-0.36 (8) —0.0+¢ (8) —_ —0.36 (8) Product-moment r between relative weight and several metabolic variables in the entire group before weight loss 0.46 (10) 0.10 (36) 0.33 (36)1 0.03 (36) 0.12 (36) 0.34 (13) —0.05 (34) OGTT, oral glucose tolerance test. * After each correlation coefficient is the number of observations in parentheses. tP < 0.05. variables measured. These data are shown in Table III. The subjects were divided by sex, and the preweight reduction product-moment correlation coefficients be- tween percent adiposity and the above variables were calculated (Table ITI, top). These correlations ranged from 0.03 to 0.43. In addition, by using multiple re- gression analysis, the partial correlation coefficient was calculated between percent adiposity and each variable while holding the effect of all the other variables con- stant. None of the partial correlation coefficients ex- ceeded 0.22. Therefore, percent adiposity never ac- counted for more than 18% of the variation in any variable. When a similar analysis was performed with relative weight as the measure of obesity essentially identical relationships were obtained (Table III, bot- tom). The product-moment correlation coefficients be- tween relative weight and the various metabolic vari- ables ranged from 0.03 to 0.46, and none of the partial correlation coefficients between relative weight and any .of the metabolic variables exceeded 0.25. Thus signifi- cant correlations did not exist in most instances and in the remaining instances only modest correlations were present. This indicates that while obesity, estimated as either percent adiposity or relative weight, has a modest influence on some of these variables it cannot be im- plicated as the sole determinant of any of them. Table IV presents the data for the cross-correlations among the metabolic variables themselves in the entire group. As predicted by our hypothesis (Fig. 1) a highly significant correlation is noted between fasting TG level and both fasting insulin level and insulin response. Next, we looked at the effects of weight loss on the metabolic variables in question. Fig. 2 summarizes the effects of weight reduction on plasma TG and choles- terol levels, The mean cholesterol and TG levels for each patient were determined before and after weight loss from a minimum of two samples at each stage, and TaBLE IV Product-Moment Correlation Coefficients among the Indicated Metabolic Variables in the Entire Group of Subjects before Weight Reduction Insulin Glucose Fasting area area Fasting insulin during during TG level OGTT OGTT level Fasting insulin level —_ —_— - - Insulin area during OGTT 0.50 (36)1 —_ — - Glucose area during OGTT —0.10 (36) 0.23 (36) —_ — Fasting TG level 0.43 (36)* 0.52 (36)§ 0.21 (36) _ Fasting cholesterol level —0.08 (34) 0.20 (34) —0.06 (34) 0.48 (34)1 OGTT, oral glucose tolerance test. *P <0.01. 1 P < 0.005. § P < 0.001. 99 A 8. before after before after 300 -r— P<0.000! le P<0.000! § = e 200 8 £ = 100 Figure 2 Plasma TG (A) and cholesterol (B) concen- trations in 36 patients before and after weight reduction. Data are given as means=SEM. the data are expressed as the means for the entire group. The decreases in both plasma TG (from 319 mg/100 ml to 180 mg/100 ml) and plasma cholesterol levels (from 282 mg/100 ml to 223 mg/100 ml) are both considerable in magnitude and highly significant (P < 0.0001). Fur- thermore, the TG levels of 34 out of 36 subjects and the cholesterol levels of 33 out of 36 subjects fell after weight loss. To illustrate the magnitude of the post-weight re- duction fall in plasma TG and cholesterol in relation to the pre-weight reduction plasma TG level, the patients were divided into quartiles on the basis of their pre- weight reduction plasma TG concentrations (Table V). The average amount of weight loss was the same in all four groups. Going from the lowest to highest quartile the mean decreases in plasma TG level after weight loss were 19 mg/100 ml, 43 mg/100 ml, 140 mg/100 ml, and 347 mg/100 ml. The mean post-weight reduction fall in cholesterol levels showed the same trend with decreases of 24 mg/100 ml, 46 mg/100 ml, 71 mg/100 ml, and 101 mg/100 ml, going from the lowest to highest pre-weight reduction TG quartiles. Thus, the greater the pre-weight reduction plasma TG level, the greater the decrease in plasma TG and cholesterol concentration in both per- centage and absolute terms. The mean plasma insulin and glucose responses dur- ing the oral glucose tolerance test before and after weight reduction are shown in Fig. 3. The mean insulin values after weight loss are lower at every time interval (3B) and these differences are statistically significant at 30, 60, 120, and 180 min (an overall 37% decrease in area under the curve). Although not as impressive as the lowering of insulin responses after weight loss, the mean glucose responses are also lower after weight loss (3A), and these differences reach statistical significance at each time point (an overall 12% decrease in area under the curve). Thus, the insulin decrement was approximately three times the glucose decrement. After weight loss, the insulin response to oral glucose decreased in 34 out of 36 subjects and glucose response decreased in 31 out of 36 subjects. Additionally, the plasma glucose and insulin responses to the 11 a.m. liquid food ingestion were studied in nine inpatient subjects before and after weight loss. These data are presented in Fig. 4, and it can be seen that a post-weight reduction fall in insulin response is quite evident with significant differences noted at each time point (4B). Furthermore, the insulin response to formula TaBLE V Decrease in plasma TG and cholesterol levels in subjects divided into quartiles Mean +SE plasma TG levels* Mean =:SE plasma cholesterol levels Plasma TG range Absolute Percentage Absolute Percentage per quartile Before After decrease decrease Before After decrease decrease 1st quartile 11548 96+9 19 17 239416 21513 24 10 (71-147) 2nd quartile 18248 13920 43 24 26912 22315 46 17 (159-228) 3rd quartile 312413 172+13 140 45 303424 232416 n 23 (254-357) 4th quartile 658-60 31135 347 53 324435 223417 101 31 * The 36 subjects are divided into quartiles (9 in each quartile) on the basis of their plasma TG concentrations before weight reduction. Of the nine subjects in the 1st quartile, six were classified as normal, and three had Type Ila hyperlipoproteinemia. Of the nine subjects in the 2nd quartile, one was classified as Type 11a, three as Type IIb, and five as Type IV, Of the nine subjects in the 3rd quartile, three were classified as Type IIb, one as Type III, and five as Type IV. Of the nine subjects in the 4th quartile, two were classified as Type III and seven as Type IV. 100 A. Glucose responses B. Insulin responses 200 _ 150 iso § E 2 1 S < g I 100 P<00I P + LS VLOL-TG Production Rate mi S wo Pe Ficure 6 VLDL-TG production rates in 13 patients before and after weight reduction. The lines connect each patient’s before and after value. Data are given as means=SEM. subjects after weight loss. Fig. 5B compares the mean steady state plasma insulin levels during these infusion studies for the group before and after weight loss. No significant differences exist between these two mean values, and thus changes in SSPG levels cannot be at- tributed to changes in concomitant steady state plasma insulin values. According to our previously outlined se- quential hypothesis, the decreases in glucose and insulin responses to oral glucose that we have shown in the TasLE VI Product-Moment Correlation Coeflicients between Degree of Insulin Resistance and Insulin Response, Insulin Response and VLDL-TG Production Rate, and VLDL-TG Production Rate and Fasting TG Concentration Insulin response during Fasting TG OGTT concentration Degree of insulin resistance 0.57 — P <0.05 VLDL-TG 0.65 0.82 Production rate P < 0.005 P 20 lipoproteins (43) then 20% of the mean post- weight reduction total plasma TG decrement would give a reasonable estimate of the cholesterol decrement at- tributable to decreases in plasma VLDL lipoproteins (ice., 0.20 X 139 mg/100 ml = 28 mg/100 ml). How- ever, the mean post-weight reduction drop in total cho- lesterol was 59 mg/100 ml, and thus it is unlikely that this fall can be accounted for solely by the fall in Se > 20 lipoproteins. This strongly suggests that the post-weight reduction decrease in cholesterol also oc- curred in one or more of the other lipoprotein classes. We are not sure how to reconcile our results with those of Wilson and Lees, but perhaps the answer may be found in the fact that they studied a relatively small number of patients and did not collect the post-weight reduction data during a period of weight stability. Thus, at the time of the post-weight reduction studies, it is pos- sible that their patients were in a state of negative ca- loric balance, and this injects an additional variable that might have an important effect on plasma lipopro- tein dynamics. For these reasons, and because of the impressive decreases in plasma TG concentrations we observed in response to modest amounts of weight loss we believe it is reasonable to consider weight reduction as an appropriate therapeutic measure in patients with endogenous hypertriglyceridemia. We have previously proposed an hypothesis to help explain one of the mechanisms causing endogenous hypertriglyceridemia (primary Types IIb, III, and IV) in man. We have suggested that a basic underlying ab- normality is impedance or resistance to insulin-mediated glucose uptake at the cellular level (23). It was our postulate that in order to maintain glucose homeostasis, insulin-resistant individuals secrete increased amounts of insulin that accelerate the hepatic VLDL-TG pro- duction rate, which in turn leads to hypertriglyceridemia (10, 24, 25). The significant correlations noted in this study between degree of insulin resistance and insulin response, insulin response and VLDL-TG production rate, and VLDL-TG production rate and fasting TG concentration in the inpatient subgroup, and between group support each step of this scheme (Fig. 1). To help confirm this sequential hypothesis, we have turned insulin response and TG concentration in the entire to a principle often used to establish causal relationships in complex multivariable systems. We have perturbed the system, thus allowing greater insight than is allowed in static situations. Also, by focusing on intraindividual comparisons this study design avoids the confounding effect of excessive interindividual variation when seek- ing relationships between biologic variables. After the perturbation of weight loss, we observed decreases in both plasma TG concentration and the glucose and in- sulin responses to oral glucose. Our inpatient studies were designed to uncover the mechanisms behind these changes, and these inpatient studies revealed marked de- creases in both degree of insulin resistance and VLDL- TG production rates after weight loss. According to the sequence outlined in Fig. 1, the decrease in degree of insulin resistance leads to decreases in insulin response to both oral glucose and food, and this lowering of plasma insulin levels results in a lowering of the rate of hepatic VLDL-TG production with a subsequent de- crease in plasma TG concentration. Furthermore, we found a marked post-weight reduction lowering of in- sulin response to liquid food containing protein, carbo- hydrate, and fat in the face of no appreciable fall in glu- cose response to liquid food. This observation provides evidence that post-weight reduction decreases in insulin response, but not necessarily glucose response, can be expected after food ingestion in the free-living state. This finding, coupled with the marked fall in plasma TG concentration after weight loss, further underscores the important role of an individual's daily postprandial plasma insulin levels in regulating TG metabolism. While these findings are simply associations, and by themselves do not prove causality, we believe that since weight reduction lowered all three of the supposed ante- cedent variables, which according to our hypothesis, lead to increased plasma TG concentration (i.e., insulin resistance, plasma insulin concentration, and VLDL-TG production rate) the overall validity of the hypothesis is strengthened. We believe this hypothesis applies to the great majority of patients with endogenous hypertri- glyceridemia. However, it should be pointed out that increased insulin levels are not present in all cases of hypertriglyceridemia. For example, many ketosis-prone diabetic patients with severe insulin deficiency develop increased TG levels, which obviously cannot be related to hyperinsulinemia. Lastly, one may speculate as to why an average weight loss of only 11 kg leads to the rather sizable metabolic changes we have observed. In a study such as ours, which compares each individual to himself, the effects of excessive interindividual biologic variability tend to dampen out. This gives greater leverage to the variable allowed to change (in this study weight alone) and its effects are easier to detect. Thus, we believe that obesity 104 is one of a variety of factors that influence the metabolic variables measured, and that this influence is amplified by our study design. The mechanisms by which weight loss initiates the changes we have observed remain un- clear. Acquired weight gain has been associated with enlarged adipocytes, and Salans, Knittle, and Hirsch (44) have shown enlarged adipocytes to be insulin-re- sistant. One might reason then that post-weight re- duction decreases in insulin and glucose responses are simply due to the decrease in mass of insulin-resistant adipose tissue. However, Bjorntorp, Krotkiewski, Lars- son, and Solmo-Szucs, and Bjorntorp, Berchtold, Holm, and Larsson have pointed out that adipose tissue ap- pears to account for less than 59% of an individuals total glucose consumption (45, 46), and thus, modest changes in the amount of adipose tissue present are not likely to appreciably affect the body’s overall insulin and glucose economy. It is obvious, then, that further in- formation is needed before a mechanism can be sug- gested whereby modest decreases in weight can initiate profound changes in the metabolic sequence we have described. Whatever the mechanism, these studies clearly dem- onstrate the potent beneficial effects of modest amounts of weight reduction in patients with varying degrees of obesity on some of the known metabolic cardiovascular risk factors. ACKNOWLEDGMENTS We wish to thank Miss Janet Wagner and Miss Phyllis Crapo for their invaluable technical assistance. This work was supported by the following grants and contracts from the National Institutes of Health: Gen- eral Clinical Research Centers Branch RR-70, National Heart and Lung Institutes, HL 08506, HL 71- 2161, HL 14174, Training Grant AM 05021, and Career Develop- ' ment Award K3-HE 6003. REFERENCES Smith, M., and R. Levine. 1964. Obesity and diabetes. Med. Clin. North Am. 48: 1387. Albrink, M., and J. Meigs. 1965. The relationship be- tween serum triglyceride and skinfold thickness in obese subjects. Ann. N. ¥Y. Acad. Sci. 131: 673. Harlan, W., A. Oberman, R. Mitchell, and A. Gray- bill. 1967. Constitutional and environmental factors re- lated to serum lipid and lipoprotein levels. Ann. Intern. Med. 66: 540. Bagdade, J. D., E. L. Bierman, and D. Porte, Jr. 1967. The significance of basal insulin levels in the evalua- tion of the insulin response to glucose in diabetic and nondiabetic subjects. J. Clin. Invest. 46: 1549. Rabinowitz, D., and K. L. Zierler. 1962. Forearm metabolism in obesity and its response to intra-arterial insulin. J. Clin. Invest. 41: 2173. Stunkard, A. J., and S. A. Blumenthal. 1972. Glucose tolerance and obesity. Metab. (Clin. Exp.). 21: 599. Drenick, E. J.,, A. S. Brickman, and E. M. Gold. 1972. Dissociation of the obesity-hyperinsulinism relationship Np od > @m N oa © © o _- > a N » ® —- = ou I “following dietary restriction ‘and hyperalimentation. im. J. Clin. Nutr. 25: 746. York, D. A, J. Steinke, and G. Bray. 1972. Hyperin- sulinemia and insulin resistance in genetically obese rats. Metab. (Clin. Exp.). 21: 277. Keys, A.,, C. Aravanis, H. Blackburn, F. S. P. Van Buchem, R. Buzina, B. S. Djordjevic, F. Fidanza, M. J. Karvonen, A. Menotti, V. Puddu, and H. Taylor. 1972. Coronary heart disease: overweight and obesity as risk factors. Ann. Intern. Med. 77: 15. . Reaven, G. M., R. L. Lerner, M. P. Stern, and J. W. Farquhar. 1967. Role of insulin in endogenous hyper- triglyceridemia. J. Clin. Invest. 46: 1756. . National Heart and Lung Institute. 1971. The dietary management of hyperlipoproteinemia. A handbook for physicians. U. S. Government Printing Office, Wash- ington, D. C. . Committee on Food and Nutrition, American Diabetes Association. 1971. Principles of nutrition and dietary recommendations for patients with diabetes mellitus. Diabetes. 20: 633. . Berkowitz, D. 1964. Metabolic changes associated with obesity before and after weight reduction. J. Am. Med. Assoc. 187: 399. . Kinsell, L., and G. Schlierf. 1965. Alimentary and non- alimentary hyperglyceridemia. Ann. N. VY. Acad. Sci. 131: 603. 5. Levy, R, M. Bonnell, and N. Ernst. 1971. Dietary management of hyperlipoproteinemia. J. Am. Diet As- soc. 58: 406. Kalkhoff, R., H. Kim, J. Cerletty, and C. Ferrou. 1971. Metabolic effects of weight loss in obese subjects. Dia- betes. 20: 83. Kanittle, J., and F. Ginsberg-Fellner. 1972. Effect of weight reduction on in vitro adipose tissue lipolysis and cellularity in obese adolescents and adults. Diabetes. 21: 754. Wilson, D., and R. Lees. 1972. Metabolic relationships among the plasma lipoproteins. J. Clin. Invest. 51: 1051. Kosaka, K., R. Hagura, R. Odagiri, F. Saita, and T. Kuzaya. 1972. Effect of weight changes on serum in- sulin response in subjects with normal oral glucose tolerance. J. Clin. Endocrinol. Metab. 35: 655. . Ostrander, L., T. Francis, N. Hayner, M. Kjelsberg, and F. Epstein. 1965. The relationship of cardiovascular disease to hyperglycemia. Ann. Intern. Med. 62: 1188. . Kannel, W., W. Castelli, T. Gordon, and P. McNamara. 1971. Serum cholesterol, lipoproteins and the risk of coronary heart disease: the Framingham study. Ann. Intern. Med. 74: 1. 5 Carlson, L., and L. Bottinger. 1972. Ischemic heart disease in relation to fasting values of plasma triglyce- rides and cholesterol. Lancet. 1: . Shen, S-W., G. Reaven, and J. Farquhar. 1970. Com- parison of "impedance to insulin-mediated glucose up- take in normal subjects and in subjects with latent diabetes. J. Clin. Invest. 49: 2151, Olefsky, J., J. Farquhar, and G. Reaven. 1972. Cause of endogenous hypertriglyceridemia in man. Clin. Res. 20: 552 (Abstr.). . Reaven, G., D. Hill, R. Gross, and J. Farquhar. 1965. Kinetics of triglyceride turnover of very low density lipoproteins of human plasma. J. Clin. Invest. 44: 1826. Beaumont, J., L. Carlson, G. Cooper, Z. Fejfar, D. Fredrickson, and T. Strasser. 1970. Classification of hyperlipidaemias and hyperlipoproteinaemies. Bull. W. H. 0.43: 891. 105 Steinkamp, R., N. Cohen, W. Goffey, T. McKey, G. Bron, W. Siri, T. Sargent, and E. Isaacs. 1965. Mea- sure of body fat and related factors in normal adults. IL J. Chromic. Dis. 18: 1291. 28. Committee on Statistics of the American Diabetes As- 29. 30. 31. 33. 34. 35. sociation. 1969. Standardization of the oral glucose tolerance test. Diabetes. 18: 299. Food for Us All. The Yearbook of Agriculture. 1969. 2 S. Government Printing Office, Washington, D. C. Agricultural Research Service, United States Depart- ment of Agriculture. 1962. Diet and some Health Ct istics of 123 B and Professional Men. U. S. Government Printing Office, Washington, D. C. 3. Farquhar, J., R. Gross, R. Wagner, and G. Reaven. 1965. Validation of an incompletely coupled, non-re- cycling catenary model of turnover of hepatic and plasma triglyceride in man. J. Lipid Res. 6: 119. Porte, D. 1967. A receptor mechanism for the inhibition of insulin release by epinephrine in man. J. Clin. Invest. 46: 86. Hoffman, W. 1937. A rapid photoelectric method for determination of glucose in blood and urine. J. Biol. Chem. 120: 51. Hales, C., and P. Randle. 1963. Immunoassay of insulin with insulin antibody precipitate. Biochem. J. 88: 137. N-24, AutoAnalyzer Manual. 1964. Technicon Instru- ments, Tarrytown, N. Y. 36. Carlson, L., and L. Wadstrom. 1959. Determination 37. of giycerides in blood serum. Clin. Chim. Acta. 4: 197. Noble, R. 1968. Electrophoretic separation of plasma lipoproteins in agarose gel. J. Lipid Res. 9: 693. 38. Armitage, P. 1971. Statistical Methods in Medical Re- 39. 41. 42. 45. on search. Blackwell Scientific Publications Ltd, Oxford, England. 99, 147, 302. ’ Abrams, M. E,, R. J. Jarrett, H. Keen, D. R. Bayns, and J. N. Crossley. 1969. Oral glucose tolerance and related factors in a normal population sample, Br. Med. J. 1: 599. Bagdade, J. D., E. I. Bierman, and D. Porte, Jr. 1971. Influence of obesity on the relationship between insulin and triglyceride levels in endogenous hypertriglyceri- demia. Diabetes. 20: 664. Dunn, J. P.,, and C. Moses. 1965. Correlation of serum lipids with uric acid and blood sugar in normal males.” Metab. (Clin. Exp.). 14: 788. Ford, S., R. C. Bozian, and H. Knowles, Jr. 1968. Interactions of obesity and glucose and insulin levels in hypertriglyceridemia. Am. J. Clin. Nutr. 21: 904. Fredrickson, D., R. Levy, and R. Lees. transport in lipoproteins. An integrated approach to mechanisms and disorders. N. Engl. J. Med. 276: 34. Salans, L. B, J. L. Knittle, and J. Hirsch. 1968. The role of adipose cell size and adipose tissue insulin sensitivity in the carbohydrate intolerance of human obesity. J. Clin. Invest. 47: 153. Bjorntérp, P., M. Krotkiewski, B. Larsson, and Z. Somlo-Sziics. 1970. Effects of feeding states on lipid radioactivity in liver, muscle and adipose tissue after injection of labelled glucose in the rat. Acta Physiol. Scand. 80: 29. . Bjorntorp, P., P. Berchtold, J. Holm, and B. Larsson. 1971. The glucose uptake of human adipose tissue in obesity. Eur. J. Clin. Invest. 1: 480. 1967. Fat - 107 New Therapies for the Eating Disorders Behavior Modification of Obesity and Anorexia Nervosa Albert Stunkard, MD, Philadelphia New approaches to psychotherapy, which appear more effective than traditional ones in modifying several kinds of disturbed behavior, have recently been applied to the eating disorders. Patients with both anorexia nervosa and obesity have responded to behavior modi- fication, and experience with obesity has already been sufficient to permit the development and description of relatively specific behav- ioral programs. These programs have been used to compare behav- ior modification in a systematic manner with a variety of alternate treatment methods. Every one of eight such studies has reported re- sults favoring behavior modification, an unusual example of unanim- ity in this heterogeneous and complex disorder. Furthermore, some new experimental designs developed in these studies are making a significant contribution to the study of psychotherapy and the eluci- dation of its effective components. N RECENT years new and distinctive forms of psychotherapy have commanded increasing attention as evidence mounts that they are more effective than tra- ditional techniques in a variety of disorders. These new treatments, known as behavior modification, behavior therapy, and experimental analysis of behavior, comprise a heterogenous series of techniques, bound together by the efforts of their proponents to apply the findings and methods of experimental psychology to disorders of hu- man behavior. The most consistent evidence of their effec- tiveness comes from the treatment of obesity where their superiority over other methods has been demonstrated in nine diflerent studies. This essay reviews these studies, describes the application of behavior modification tech- niques to obesity, and outlines some of the increasingly sophisticated experimental designs which are elucidating the effective clements of psychotherapy. It seems partic- ularly important that psychiatrists know of these develop- T Accepted for publication Aug 19, 1971. From the Department of Psychiatry, University of Pennsylvania, and the Philadelphia General Hospital, Philadelphia, and the Center for Ad- vanced Study in the Behavioral Sciences, Stanford, Calif. Reprint requests to Department of Psychiatry, University of Pennsylva- nia, Philadelphia 19104 (Dr. Stunkard). 34-621 0 - 74-8 ments, for behavior modification has grown up largely outside their purview and, at times, in the face of their op- position. The extent of the leadership exercised by psy- chologists and social workers is nowhere better illustrated than in the “medical” problem of obesity: physicians par- ticipated in only one of nine studies described here. The distinctive characteristic of the various methods of behavior modification is the belief that behavior disorders of the most divergent types are learned responses and that modern theories of learning have much to teach us re- garding both the acquisition and extinction of these re- sponses.’ Furthermore, proponents of behavior meodi- fication have been distinguished by their explicit statements of methods and goals, and their willingness to put their results on the line for comparison with other forms of treatment. Illustrative of these methodological concerns, behavior therapists have been among the first to recognize the power, as a dependent variable in psy- chotherapy research, of weight change in pounds, and they have turned to the treatment of obesity in order to utilize this measure. It is ironic that psychiatry, so sorely in need of measures to evaluate therapeutic success and failure, has taken so long to recognize the sensivity, relia- bility, and validity of weight change as such a measure. Anorexia Nervosa The characteristics and potential of behavior modification can perhaps best be introduced, as they were to me, by the description of a behavioral approach to anorexia nervosa. Three case reports had suggested that behavioral techniques could modify the under- cating of anorexic patients and produce weight gain.** Not until behavioral analysis, however, revealed an unexpectedly prominent feature of the disorder was it possible to apply an effective behav- ioral technology to a series of patients.>* This feature was hyper- _ activity, and its use in therapy illustrates the simplicity and power of behavior modification in this chronic, often intractable, disorder. Pedometer measurements of patients hospitalized for treat- ment of anorexia nervosa revealed that they walked an average of 108 6.8 miles per day, as compared with a mean daily value of 4.9 miles for women of normal weight living at home. This observation sug- gested that opportunity for ph activity might serve reinforcement for increased food intake or, as to specify, weight gain. Accordingly, the patient's access to physi cal activity was made contingent upon weight gain. Specificall she was permitted a six-hour unrestricted period outside the hos- pital on any day that her morning weight was at least 0.5 1b above her previous morning's weight. No comment was made concerning the level of activity or food consumption, thus avoiding direct con- {rontation over eating. Less than one week after this regrimen was instituted, the first three patients responded with a rapid and consistent increase in body weight. They averaged gains of 4 Ib per week during six weeks of hospitalization. These results rank with the best re- ported in the medical literature, including series of patients treated with far more aggressive measures (hed rest, tube feed- ing, very large doses of chlorpromazine and insulin).™ The effectiveness of this approach led us to apply it, with modi- fication, in three additional patients, all of whom responded with similarly gratifying weight gains. An instructive example was a 17-year-old girl admitted in a state of profound inanition and weighing 22.7 kg (50 1b) (height, 149.9 cm [4 ft 11 in]). The pa- tient's behavioral repertoire was so limited that a search for a po- tential reinforcer was at first unsuccessful. Soon after chlor- promazine treatment was begun, however, the patient began to complain of its sedative effects. Her complaints suggested a new reinforcement contingency. We prescribed deereases in chlorpromazine dosage proportional to the amount of weight gained on the previous day: on any day that followed a loss or no change in weight, the patient received 400 mg; a 0.25-1b gain resulted in a decrease to 300 mg; a 0.5-1b gain resulted in a decrease to 200 mg; a 0.75-1b gain resulted in a decrease to 100 mg; and a 1-1b gain resulted in a decrease to no drug. This patient averaged a gain of 6 Ib per week, despite the consequent radical decrease in chlorpromazine dosage! Moreover, the results made it clear that the therapeutic efficacy of the behav joral approach did not depend upon hyperactivity or any other specific symptom. Rather, any suitable contingency may serve as the reinforcer in behavioral therapy. This finding was particularly indicative to me of the effectiveness of this form of treatment. Obesity—First Applications Spurred by the surprising success of a simple behavioral tech- nique in the treatment of a condition as stubborn as anorexia ner- vosa and by encouraging case reports of the effectiveness of be- havioral modification in obesity,”* I attempted to apply its principles to the treatment of two obese persons. Taking the ano- rexia nervosa paradigm as my model, I decided to make a small number of reinforcements contingent upon weight changes mea- sured at first once and later as often as four times a day. In the case of one patient, a cooperalive roommate offered to take over such unpleasant chores as doing dishes or taking out the garbage on the day following one during which the patient lost weight. The second patient's wife, herself a psychologist, agreed to make sex- ual intercourse contingent upon her husband having lost weight that day. - After an initial loss of 20 Ib, the weights of both patients stabi- lized and they eventually stopped treatment. Paradoxically, it ap- peared that obesity posed a greater challenge to behavior modi- fication than did the more stubborn anorexia nervosa. Other attempts to apply behavioral therapy to obesity also suggest that the task is more complicated than experience with anorexia ner- vosa had indicated. Most of these attempts were based on a 1962 paper by Ferster et al*' who presented a detailed behavioral anal- ysis of eating and means of control. ‘The Behavioral Program Perhaps it would be well, at this point, to describe a typical bes havioral program for the treatment of obesity. Since our program: is derived, as were most others, from Ferster et al,” and is sim- ilar to the other programs, I will describe it in some detail. Four principles are involved. = Description of the Behavior To Be Controlled. ~The patients were asked to keep daily records of the amount, time, and circum- stances of their eating. The immediate results of this time-con- suming and inconvenient procedure were grumbling and com- plaints. But eventually each patient reluctantly acknowledged that keeping these records had proved very helpful, particularly in increasing his awareness of how much he ate, the speed with which he ate, and the large variety of environmental and psycho- logical situations associated with eating. For example, after two weeks of record-keeping a 30-year-old housewife reported that, for the first time in her life, she recognized that anger stimulated ber eating. Accordingly, whenever she began to get angry, she left the kitchen and wrote down how she felt, thereby decreasing her anger and aborting her eating. Modification and Control of the Discriminatory Stimuli Governing Eating.—Most of the patients reported that their eating took place in a wide variety of places and at many different times during the day. It was postulated that these times and places had become so- called discriminatory stimuli signaling eating. The concept of a discriminatory stimulus derives from the animal laboratory, where such stimuli as the flashing of a light or sounding of a tone may signal to an animal that pressing a lever will produce food pellets or other reward. Since the reinforcer never occurs without the discriminatory stimulus, in the language of learning theory, the stimuli come to “control” various forms of behavior. In an ef- fort to decrease the potency of the discriminatory stimuli that controlled their eating, patients were encouraged to confine cat- ing, including snacking, to one place. In order not to disrupt do- mestie routines, this place was usually the kitchen. Further ef- Torts to control discriminatory stimuli included using distinctive table settings, perhaps an unusually colored place mat and napkin. In addition, patients were encouraged to make eating a pure ex- perience, unaccompanied by other activity such as reading, watch- ing television, or arguing with their families. Development of Techniques To Centrol the Act of Eating.—Specific techniques were utilized to help patients decrease their speed of cating, to become aware of all the components of the cating pro- cess, and to gain control over these components. Exercises in- cluded counting each mouthful of food eaten during a meal, plac- ing utensils on the plate after every third mouthful until that mouthful was chewed and swallowed, and introducing a two-min- ute interruption of the meal. Prompt Reinforcement of Behaviors That Delay or Control Eat- ing.— A reinforcement schedule, using a point system, was devised for control of eating behavior. Exercise of the suggested control procedures during a meal earned a certain number of points. These points were converted into money, which was brought to the next meeting and donated to the group. At the beginning of the program, the groups decided how the money should be used and they chose highly altruistic uses. Each week one group do- nated its savings to the Salvation Army and another to a needy {friend of one of the members, a widow with 14 children. It has, in the past, been fairly casy to assess any outpatient treatment for obesity because the results have been so uniformly poor and the treatments themselves so obviously inadequate. (In- patient treatment, with its potential for greater control of the pa- tient has of course been more successful in weight reduction. Its usefulness has been limited, however, by the almost invariable re- gaining of weight after discharge.*) I have summarized my own and my colleagues’ results with outpatient treatment quite sim- 109 ply: “Most obese persons will not stay in treatment for obesity. Of those who stay in treatment, most will not lose weight, and of those who do lose weight most will regain it. Attrition rates vary between 20%: and 80%. Only 25% of those who enter treat- ment lose as much as 20 1h; only 5% as much as 40 1h. Against this background, the results obtained by Ferster et al, whose subjects averaged weight losses of only 10 1b, must be considered poor. Against this same background, moreover, the significance of a re- port on “Behavioral Control of Overeating” is at once apparent. For in this report, Stuart, using a treatment program based on Ferster et al's, described the best results yet obtained in the out- patient treatment of obesity. A Landmark in the Treatment of Obesity Stuart’s results are summarized in Fig 1 and show the weight losses, over a one-year period, of eight patients who re- mained in treatment (the initial study group had included ten pa- tients). Three, or 30% of the original sample, lost more than 40 1b and six lost more than 30 1b. These results are the hest ever re- ported for outpatient treatment of obesity, and they constitute a landmark in our understanding of this disorder. Even the absence of a control group does not vitiate the significance of the study. Certain features of the report deserve attention: First, the ex- penditure of time was not exorbitant. In fact, the study took no longer than a number of others which achieved far poorer results. At the beginning of the treatment program, patients were seen in 30-minute sessions held three times each week for a total of 12 to 15 sessions. Thereafter, treatment sessions were scheduled as needed, usually at two-week intervals, for the next three months. Subsequently, there were monthly sessions and finally “main- tenance” sessions were provided as needed. The total number of sessions during the year varied from 16 to 41. The specific behavioral techniques applied by Stuart are similar to those used in other studies and described previously. One im- portant feature of this study was that the regimen specified a rigid set of "how to do it” instructions for cach of the first 12 in- terviews. Within this framework, however, there was great oppor- tunity for the exercise of creativity by both the therapist and the patient. For mple, Stuart noted that for patients suffering from a “behavioral depression,” eating may be the only readily available reward or reinforcement. For these individuals, the therapist must cultivate a reservoir of positively reinforcing re- sponses. Two patients in the series were helped to develop such re- Spo ~an interestin caged birds and in growing African violets respectively. In contrast to the other studies described here, all of which used group therapy, Stuart treated his patients individ- ually. Use of No-Treatment Controls In 1969, Harris reported a well-controlled study which utilized behavioral techniques to control eating in mildly overweight col- lege students.” Two treatment groups, of three male and five fe- male students cach, were compared with a control group of eight students. In order not to discourage them, and thereby bias the re- sults, the controls were told that they could not enter treatment at once because of a conflict in schedules, but that they would receive treatment later. Treatment sessions were held twice weekly for the first two months and then on a more irregular basis for the second two months. The results are illustrated in Fig 2. The mean weight loss for the experimental group was 10.5 Ib as compared with a weight gain of 3.6 Ib for the control group, a difference that was very highly significant (P< .001). Although the results in the treatment group are clearly far su- perior to those in the no-treatment control group, they are not as good as others reported in the literature, when judged by the eri- teria I have mentioned: only 21% of Harris’ subjects lost 20 1b and none lost as much as 40 Ib. A major reason for these results was Fig 1. —Weight profiles of eight women undergoing behavior therapy for obesity. 225 Patient 1 180 Patient 2 195 Patient 3 1901 Patient 4 220 75+ 190 185+ 215 170 185 180 210 165 180 175 4 205 § 501 2115 $ 70 3 200 & 1557 8 mo 2 165- 195 150 165 160 190 1454 160 155-1 185 = 140-] 155 150 180 135 pry pp— 150 145+ 0 3 6 9 R 0-3 69.1” 6 12 ? 3 £5 =» Months Months Months Months 200 Potiont 5 180 Patient - 6 220 Patient 7 2104 Pofient 8 195 175 ’ 215 205 ; 190 170 210: 200 . 185 165 205 195+ < > 2 T 180: 2 160 3 200 2 1904 Fa & 1554 3 195 £ 185+ 170 150 * 190 180 165 1454 185. 1754 160 140 . 180 1704 15 . °% 3 6 is RR ps 12 vos Months Months Months Months 110 that her subjects were less obese than those studied by other in- vestigators, An interesting and perhaps significant aspeet of the study is the note that BE (the experimenter) lost 27 1b. As Harris continues in the quaint language of the psychological literature, “the modeling effect of E, who went from fat to moderate with the pretest Ss and from moderate to thin with the Ss in the final study, was com- mented upon by many of the subjects.” Harris noted that several variables may have contributed to the outcome besides the planned experimental procedures, She suggests that "a much more controlled study, in which various techniques and com- binations of procedures are isolated, would be necessary to dis- cover their differential effects.” Precisely such an investigation was carried out by Woller- sheim.* Her work is representative of a small but extremely valu- able group of studies that are elucidating the effective components of psychotherapy. Introduction of Alternate-Treatment Controls Wollersheim’s elegant study attempted to disentangle the con- tributions of various techniques by establishing four experimental conditions: (1) “focal” (behavioral) treatment, 20 patients; (2) non- specific therapy, 20 patients; (3) social pressure, 20 patients; and (4) control groups of persons promised treatment but not yet re- ceiving it, 19 patients. The study thus contained three treatment groups (1, 2, and 3) and three control groups (2, 3, and 4) for the behavior modification group. Subjects were, again, mildly (10%) overweight female col- lege students. Four therapists treated groups of five subjects un- der each of the three treatment conditions. A course of treatment consisted of ten sessions extending over a three-month period. Wollersheim's findings are illustrated in Fig 3. They showed that at the end of treatment and at eight weeks follow-up, the fo- cal group's results were superior not only to the no-treatment con- trol’s, but also to those of the other two treatment groups. The “social pressure” group had participated in 20-minute sessions based upon those of TOPS (Take Off Mounds Sensibly).” The ses- sions included a weigh-in, verbal praise for weight loss, encour- agement for failure to lose weight, and the wearing of such TOPS artifacts as a star for weight loss, a sign in the form of a pig for weight gain, and a sign reading “Turtle” for no change in weight. The purpose of this technique is to foster a high positive ex- pectation for losing weight and to develop and use special pres- sure to help subjects reduce. The purpose of a “nonspecific therapy” group was to control for the effects of group treatment that resulted from such nonspecific factors as increased attention, “faith,” expectation of relief, and presentation of a treatment rationale and meaningful “ritual.” The rationale presented to the subjects in this group was that they needed to develop insight into the “real and not readily recogniz- able underlying reasons” for their behavior and to discover the “unconscious motives” underlying their “personality make-up.” Each subject was told that as she obtained insight and better un- derstanding of the "real motives and forces” operating within her personality, she would find it casier to accomplish her goals and lose weight. In cach group, not only weight loss but also responses to an eat- ing-patlerns questionnaire were used as dependent variables. Here, 100, the focal therapy group changed more than the other three, reaching statistically significant results on three of six fac- tors: “emotional and uncontrolled overeating,” “cating in iso- lation,” and “between meal eating.” More Alternate- Treatment Controls: Therapy Without a Therapist Hagen, following up the work of Wollersheim, accepted her finding that behavior modification was the most effective method for the treatment of obesity." He turned his attention to a further refinement—determining whether the results obtained were due only to the specific behavioral techniques used or were dependent also upon the interpersonal influence of the therapist. Hagen con- structed an experimental design similar to Wollersheim's. The various treatment groups were compared with each other and with a control group promised treatment but not yet receiving it: (1) group (behavioral) therapy, 18 subjects; (2) bibliotherapy (use of a written manual), 18 subjects; (3) group and bibliotherapy combined, 1R subjects; and (4) the group promised treatment bat not yet receiving it, 35 subjects. The 90 subjects in his study were also mildly (107) overweight female college students. They were randomly assigned to one of four experimental groups in such a way that the groups were com- parable in relation to the obesity of their members. Three thera- pists treated six subjects each in the group therapy and combined therapy conditions. Ten treatment sessions were held over a three-month period. The results of treatment are shown in Fig 4. The greatest weight Joss occurred in the group and bibliotherapy combined group, which lost an average of 15 Ib during treatment and re- gained 2 1b during the four-weck follow-up period. The difference in weight loss between this group and the other two treatment groups was, however, not statistically significant. There was a sig- nificant difference (P< 01) between its results and those of the no-treatment group. Wollersheim's cating-patterns questionnaire also ¢licited results from all three treatment groups that were similar to those from her “focal” group, but demonstrated no change in the control group. Hagen’s work showed that it is possible to treat obesity effec- tively by using a written manual that embodies behavioral ther- apy principles. Moreover, this treatment is apparently as effective as onc that utilizes therapists. These results further confirm the effectiveness of behavioral principles in the treatment of obesity. A Study With a Crossover Design Stuart has recently completed a study that used the patient as his own control in what is known as a crossover design.” The re- sults are summarized in Fig 5. The subjects were divided into two cohorts which contained three moderately obese women each. Dur- ing a preliminary five-week period they kept careful records of their weight and food intake. Group 1 then received twice weekly treatment sessions of about 40 minutes, for 15 weeks. Meanwhile, group 2 was given diet-planning materials and an exercise pro- gram (both of which were also offered to group 1). Group 1 lost an average of 15 1b while the control group gained an average of 4 1b. - At the end of the 15 weeks, group 1 continued with the program on its own, while group 2 received 15 weeks of the same treatment group 1 had received (group therapy plus diet planning and ex- ercise). Under these circumstances, group 1 lost an additional 9 Ib, but at a slightly lower rate than when it was under active treat- ment. Group 2, on the other hand, which had gained weight during the preceding 15 weeks, lost 15 1b. Both groups continued to lose in the subsequent 12 weeks without further treatment. Treatment of Severe Obesity Another dimension of the behavioral approach to obesity was provided by a study of severely obese patients (78% overweight). Twa cohorts of eight and seven patients respectively were seen in weekly group therapy sessions lasting about two hours, for three months. The therapists were a male experimental psychologist with a strong background in learning theory but little clinical ex- perience, and a female research technician with no previous ex- perience in therapy. Two control groups received supportive psy- chotherapy, instruction about dieting and nutrition, and, upon 111 Control, ee, —0 ame - § ¢ z s 8 Experimental Pre 2% Months Post Time of Meosurement Fig 2.—Mean weights for experimental and control subjects. O----0 sinciol Pressure -2: Pre Post FU Time of Assessment Fig 3.—Mean weight loss of the focal (be- havioral) treatment group. the two alterna. tive treatment control groups. and the no- treatment control group. Treatment lasted three months and follow-up (FU) occurred Pounds 2 Time of Assessment Fig 4. —Mean weight loss of the three eight weeks later. demand, appetite suppressants. Their male and female therapy team consisted of an internist just completing a psychiatric res- idency, who had had extensive experience in the treatment of obesity, and a research nurse. The results of treatment of the two cohorts are summarized in the Table. The control group's losses are comparable to those to be found in the medical literature—none lost 40 1b and 24% lost more than 20 Ib. By contrast, 13% of the behavior modification group lost more than 40 Ib and 53% lost more than 20 Ib. Although the differences between the behavior modification and control groups for weight Josses over 20 and 40 Ib are not statistically significant, the difference for weight losses over 30 1b is (P = 015 by Fisher exact probability test). Furthermore, those who had lost weight during treatment continued to lose weight during the following year. The weight losses for each subject are plotted in Fig 6. Two findings should be noted: First, in each cohort the median weight loss for the behavior modification group was greater than that for the control group—24 1b versus 18 Ib for the first cohort: 13 versus 11 for the second. Sccond, there is far greater variability of re- sults in the behavior modification groups (F = 4.38, P < .005). Be- cause of this variability the difference in weight loss between the behavior modification and control groups did not reach statistical significance. Testing the Limits of the Behavioral Approach Recently three further studies, on strikingly dissimilar popu- lations, have demonstrated the remarkably wide range of circum- stances in which behavior modification may be effective in the treatment groups and of the no-treatment control group. The former lost significantly more weight than the latter, but there was no significant difference in weight loss be- tween the three treatment groups. Treat- ment lasted three months and follow-up oc- curred one month later. Both Group 1: Group 1: Both Groups: Treatment Maintenance Groups: Self- Group 2: Group 2: Follow monitoring | “Self-Control” Treatment up 195 }= 38 2 $180 NLC s $ 17 5165 8s fe 155 150+ THN T T — NT 1234 10 15 20 25 30 3 40 52 Weeks ome Group 1(N=3) Group 2(N=3) Fig 5.—Weight changes in two groups of women undergoing be: havior therapy for overeating. 112 treatment of ohes Results of Treatment: 9, of Groups Losing Specified Amounts of Weight o P! Bg Sp! first population con: long-term schizophrenic pa- Behavior Modification Control Therapy tients in a Veterans Adminis- Groups (No. 15) Groups (No. : 17) Average 5 ab oF rm Ne pie i mie ei tm mn Medical tration Hospital," the second Alter One-Year After One-Year Literature of a heterogencous group of Weight Loss Treatment Follow-up Treatment Follow-up End of Treatment Dutch men and women,” and |~ More than 4016 137% 33, Soy You “TRY the third of ohese women in a “More than 30 Ib 3, nT eT Tey i general hospital.“ Asin three |” Morethan201b 53% 53% 24%, 27% 759, of the foregoing studies, fol- low-up data on the first two of these showed significantly greater weight loss among behavior modification sub- jects than among those who received other forms of treat- ment. The first of these studies was carried out to explore the management of the obesity which so frequently develops among schizophrenic patients during prolonged hospi- talization® Seven men, matched for degree of over- weight, were assigned to each of three groups—behavior modification, group therapy, and diet only. Treatment was carried out over a six-week period with a four-week fol- Jowup. Behavior modification consisted of a penalty sched- ule involving forfeiture of 0 4 8 12 4 8 12 part of $5 weckly allowance for failure to lose weight dur- ing the previous week. Weight loss carried no re- ward other than assuring the Fig 6, Above and below.—Weight changes of patients in the two cohorts. Dotted lines represented in- patient of his regular allot- terpolated data based upon weights obtained during follow-up. Note greater weight loss of behavior modi- ment. Group therapy sessions fication groups and greater variability of this weight loss as compared with that of the control group. were held once cach week for one hour, and consisted of weighings with encouraging comments for weight losses +10 Behavior Group 2 and discussion of reasons for N=7) gains and losses. Both treat- ment groups lost signifi- cantly more weight than the dict only group during the six-week treatment period. But the behavior madi- fication subjects lost signifi- cantly more weight during the four-week follow-up than did the group therapy sub- jects. 4 In the second study the Dutch investigators chose as their subjects university stu- dents and others recruited by a variety of means, including advertisements in women's magazines.* Two behavior maodifieation groups of 15 subjeets cach were compared ° 4 8 12 4 8 12 with 15 subjects in a tradi- Weeks of Treatment Behavior Group 1 (N=8) Weight Loss or Gain (Pounds) Weeks of Treatment Control Group 2 (N=7) weight Loss or Gam (Pounds) 113 tional diet-therapy group and 15 no-treatment controls. The pro- gram consisted of a four-week treatment period with follow-ups at six weeks and five months, At the end of four weeks, all three treatment groups had lost more weight than these who had re- ceived no treatment, and one of the behavior modification groups had lost more than the other two treatment groups. At the five- month follow-up, both behavior madifiecation groups, one of which had continued to Jose weight and the other of which had main- tained its weight loss, had performed significantly more effee- tively than the diet therapy group, which had regained most of the weight it had lost. The results of the third study, reported only briefly, again fa- vored behavior modification. Eleven obese women treated with this modality Jost more weight than did nine bbese women treated by conventional group psychotherapy (15 vs 6.6 Ih). Furthermore, in an interesting parallel to the results reported by Penick et al, this 8.4-Ib difference in weight loss was not statistically signifi- cant, largely because of the very great variability in the behavior mudification group (W. Shipman, written communication, May 15, 1971). Comments and Conclusions The studies we have reviewed are an impressive ex- ample of the kind of contribution which behavior thera- pists are making to the scientific study of psychotherapy. These workers have introduced experimental designs of a sophistication and power unprecedented in psychotherapy research: they may well be laying the foundations of a truly cumulative science of psychotherapy. : Two further aspects of behavior modification may be less well appreciated by psychiatrists: the possibility of exercising great creativity on the part of both patient and therapist and the encouragement of patients to assume an unusually high degree of responsibility for their own treatment. Creativity in Bchavior Modification.—An unflattering view of hehavior modification sees in it mechanistic ma- nipulations to produce trivial changes in overt behavior at the expense of the patient's inner freedom and uniquely human qualities. Such polemics may cite the examples of systematic desensitization of phobias or the conditioned aversion therapy of alcoholism, treatments based upon re- spondent (pavlovian) conditioning. Whatever the merit of the criticism of these treatments, it is hardly relevant to behavioral measures, such as those reviewed here, which utilize primarily operant (skinnerian) conditioning. A closer look at these measures may help to clarify this point. A behavioral analysis begins with a careful study of the environmental variables that control the patient’s symp- tomatic behavior. These variables may be divided for con- venience into antecedent events, which may precipitate the behavior, and consequent events, which may help to reward and maintain it. For example, understanding a particular episode of hinge-eating requires precise infor- mation about what was happening to the patient just be- fore the breakdown in eating control occurred. Similarly, it is important to ascertain the immediate cffects upon the patient's feclings, on her husband's behavior, and, in fact, upon any significant aspect in her life. Stuart has re- cently outlined a useful schema for a behavioral analysis and we have described carlier the kind of therapeutic pro- gram which can be constructed out of these analyses of specific behaviors, 1 was persuaded of the value of such a method of pro- ceeding as 1 considered the degree to which it specified techniques for the treatment of obesity which 1 had evolved empirically over nearly 20 years of treatment of obese persons. It contrasted sharply with the recommen- dations for treatment derived from psychodynamic theo- rics. Such theories, for example, cautioned against eonsid- ering with the patient his specific acts of overeating and consequent weight gains. For they saw in these acts merely symptoms of an underlying conflict. Any small benefits which might accrue from concern with these symptoms, it was taught, were likely to be far outweighed by distraction of paticnt and therapist from their primary task of resolution of the pathogenic conflicts. Such resolu- tion, and only such resolution, could lead to a lasting cure. Quite early, expericnces with patients convinced me that such a course was countertherapeutic, and I began to discuss their eating behavior with my patients, focusing in ever greater detail upon specific acts of overeating. 1 had been content to explain these departures from ac- ecpted teaching as necessary adaptations to the special circumstances of the cating disorders. So, when I encoun- tered an approach which prescribed what 1 was already doing, its appeal was immediate. This approach, morcover, went further to suggest potent new measures of great creativity. What is the source of this creativity? Central to a behavioral analysis is the scarch by patient and therapist for solutions to problems which arc at the same time both relatively modest and potentially soluble. By limiting therapeutic concern to discrete, clearly speci- fied behaviors, this approach reduces the potentially limit- less field of therapeutic encounter and permits patient and therapist to concentrate their efforts on a more lim- ited number of variables than is possible with traditional therapies. By focusing a great deal of attention upon rela- tively small problems the probability of solving them is greatly increased. And, the experience of success, even in small matters, encourages the patient to continue the pro- cess of defining manageable problems and secking solu- tions to them. It is hard to do justice to the remarkable. creativity evoked in the course of this endeavor. The few examples described in this review can only suggest the kind of innovative measures which, in remarkable scope and diversity, have been applied to the treatment of obe- sity. Responsibility in Behavior Modification.—The particular relevance of focusing upon environmental variables in the treatment of obesity is emphasized by Schachter’s recent demonstration of the extent to which the eating behavior of many obese persons is under environmental control.” He found, for example, that obese persons were far more influenced than non-obese persons by such “external” fac- tors as palatability, time of day, and availability of food, and far less influenced by such “internal” factors as hun- ger, measured by self-report, and by length of time after cating. = The high degree of environmental control of the food intake of obese persons may help to explain the failures of both routine medical management and of traditional in- 114 sight therapy. Obese persons often adapt easily to general medicine's authoritarian-physician-dependent-patient re- lationship and lose some weight to please the doctor. Fail- ure to deal with the environmental bles which play such an important part in the patient's cating, however, leaves him vulnerable to their influence, and sooner or later he breaks his diet. This transgression strikes at the special qualities of this kind of doctor-patient relation- ship, which then begins to lose its potency. Overeating re- curs and a vicious cycle ensues. The obese person who enters psychiatric treatment fre- quently fares little better. Insight therapy, with its focus on inner drives, motives, and conflicts, all too often ig- nores environmental factors in the control of food intake as thoroughly us does general medical treatment. Further, by holding out hope for an eventual solution to obesity through the resolution of conflict, it can foster magical ex- pectations which distract the patient from more mundane concerns of greater therapeutic potential. In contrast to this neglect by traditional therapies of the environmental influences to which the obese person seems so vulnerable, behavior modification helps him to focus his attention upon them. Not only is he encouraged to observe and make a detailed record of these influences on his eating behavior, but he is also shown how to use this information to plan and carry out tasks to help him gain control over this behavior. The difference between behav- ior modification and more traditional therapies is partic- ularly notable in the extent of the demands cach makes upon the obese patient in the interval between visits to the therapist. In contrast to the limited demands of tradi- tional therapies, behavior modification makes it possible for the patient to invest a great deal of hard work in his treatment. The apparent dependency of successful out- come upon the amount of work invested encourages pa- tients to assume an unusually high degree of responsi- bility for their own treatment. This increase in the opportunity for the exercise of personal responsibility may prove to be onc of behavior modification’s major con- tributions to treatment in psychiatry. In conclusion, both greater weight loss during treat- ment and superior maintenance of weight loss after treat- ment indicate that behavior modification is more efTective than previous methods of treatment for obesity. Further, the experimental designs developed to assess these treat- ments constitute a significant advance in the study of psy- chotherapy. (Since submission of this review, a compre- hensive description of a behavioral approach to obesity has been published by Stuart and Davis.*) This study was supported in part by grant MH-15383 from the National Institute of Mental Health. References 1. Eysenck HJ: Editorial. Behav Res Ther 1:1-2, 1963. 2 Bachrach AJ, Erwin W, Mohr JP: The control of eating be- havior in un anorexic by operant cond ing techniques, in Ull- mann LP, Krasner L (eds): Case Si ¢ in Behavior Modi- Sication. New York, Holt Rinehart & Winston, 1965, pp 153-163. 3. Hallsten EA Jr: Adolescent anorexia nervosa treated by de- sensitiation. Behav Res Ther 3:87-91, 1965. 4. Leitenberg H, Agras WS, Thomson LE: A sequential analysis of the effect of selective positive reinforcement in modifying ano- rexia nervosa. Behar Res Ther 6:211-218, 1965, 5. Blinder BJ, Freeman DMA, Ringold A restoration in anorexia nervosa. Clin Res 1 6. Blinder BJ, Freeman DMA, Stank i: He Tr therapy of anorexia nervosa: Effectiveness of a as a reinforcer of weight gain. Amer J Psychiat 126: T9008 1970. 7. Williams E: Anorexia nervosa, a somatic disorder. Brit Med J z 190-195, 1958. 8. Dally P, i W: A new treatment of anorexia. Brit Med J 1:1770- 1773, 196 9. Russell GH, i AG: An analysis of weight gain in tients with anorexia nervosa treated with high calorie diets. Sri 23:449-461, 1 10. Crisp A “Clinical and therapeutic aspects of anorexia. J Psychosom a 18, 11. Crisp AH: A treatment regime for anorexia nervosa. Brit J Psychiat Tia: RITE 2, 1966. 12. Dally P, Surgant wv: Treatment and outcome of anorexia. Brit Med J 2:293-295 13. Groen JJ, Li Soledano Z: Educative treatment of tients and parents in anorexia nervosa. Brit J Psychiat 112:6 i 678, 1966. 14. Browning CH, Miller SI: Anorexia nervosa: AS] on prog- nosis and management. Amer J Psychiat 124:1125- 15. Erickson MA: The utilization of patient he a) in in or notherapy of obesity. Amer J Clin Hypn 3:112-116, 1960. 16. Thorpe JG, Schmidt E, Brown “Br, et al: Aversion relief therapy: A new method for general application. Behar Kes Ther 2:71-82, 1964. 17. Meyer V, Crisp AH: Avera therapy in two cases of obe- sity. Behav Res Ther 2:14: 3-147, 1964. 1s. Cautela J: Covert sensitization. Psychol Rep 20:459-468, a Sad weight 19. Kennedy WA, Foreyt J: Control of cating hehavior in an these patient by avoidance conditioning. Psycho ol Rep 22:571-576, 20. Wolpe J: The Practise of Behavior Therapy. Oxford, En- gland, Pergamon Press, 1969, ) 204, 205, 216, and 217. 21. Ferster CB, Nurnher; Ker 1, Levitt EB: The control of eat- ing. J Mathetics 1:87-100, 1 32. Penick SB, Filion R, Hor S, et al: Behavior motificgtion d in the treatment of obesity. Psychosom Med 33:49-55, 1971 23. Maccuish AC, Munro JF, Duncun LJP: Follow-up study of refractory obesity treated by fasting Brit Med .J 1:91-200, 19G8. 24. Swanson DW, Dinello “ollow-up of patients starved for obe Psychosom "Med 3 a, 1970. 2 Stunkard AJ: The management of obesity. New York J Med , 1958, 26. Stunkard AJ, McLaren-Hume M: The results of treatment for obesity. Arch Intern Med 103: 79-85, 1959. 27. Stuart RB: pohievined control of overeating. Behav Res ‘Ther 5:357-365, 196 2%. Harris MB: Sr. directed program for weight control: A pi- Jot study. J Abnorm Psychol 74:263-270, 1969. 29. Wollersheim JP: The effectiveness of group therapy based upon learning principles in the treatment of overweight women. J Abnorm Psychol 16:462-474, 1970. 30. Stunkard AJ, Fox S, an H: The management of obesity: Patient self-help "and medical treatment. Arch Intern Med 125:1067-1072, 1970. 31. Hagen RL: Group Therapy Versus Bibliotherapy in Weight Reduction, thesis. University of Illinois, Champaign, 1969. 32, Stuart RB: A three-dimensional nro gram for the treatment sity. Behav Res Ther 9:177-186, Harmatz MG, La puc P: Behavior 15 ean of overeatin; psychiatric population, J Consult Clin Psychol 32: 583-581 Seman JG: Vermagerings-Therapieen, thesis. Psycho- ch Laboratorium van de Universiteit van Amsterdam, Am- sterdam, 1969. 35. Shipman W: Behavior Therapy With Obese Dieters, annual report of the Institute for Psychosomatic and Psychiatric Re- search and Training. Chicago, Michael Reese Hospital and Medi- cal Center, 1970, pp 70-71. 36. Stuart RB: Trick or Treatment: How and When Psy- chotherapy Fails. Champaign, Ill, Research Press, 1971, pp 183- 194. Schachter S: Obesity and eating. Science 161:751-756, 1968. Stuart RB, Davis HB: Slim Chance in a Fat World: Behav: ioral Control of Obesity. Champaign, Ill, Research Press, 1971. 115 TOPICAL PAPERS: Treatment of Eating Disorders A Therapeutic Coalition for Obesity : Behavior Modification and Patient Self-Help BY LEONARD S. LEVITZ, PH.D, AND ALBERT J. STUNKARD, M.D. The effectiveness of a self-help organization for the obese was significantly increased by behavior modification techniques. Sixteen chapters of TOPS (Take Off Pounds Sensibly), with a total of 234 members, received one of four treatments: behavior modification conducted by a professional therapist, behavior modification conducted by the TOPS leader, nutrition education conducted by the TOPS leader, and continuation of the usual TOPS program. During the three-month treatment period, be- havior modification produced significantly lower attrition rates and significantly greater weight losses than did the alternate treatment methods. At nine-month follow-up, the differences among treatments were even greater. SELF-HELP ORGANIZATIONS for weight control are a large and growing vehicle for the control of obesity. Unfortu- nately, their effectiveness is limited; most members lose very little weight (1). Behavior modification seems to be more effective in treating obesity than any traditional method, but it has not been widely applied. We designed this study to see if self-help and behavior modification could be bined. Behavior modi ion, we r d. might provide self-help organizations with a more effec- tive treatment. At the same time, the self-help organiza- tions could provide a means for the widespread appli- A preliminary version of this paper was read at the 126th annual meet- ing of the American Psychiatric Association, Honolulu, Hawaii, May 7-11, 1973. Dr. Levitz is Associate in Psychology, Department of Psychiatry, Uni- versity of Pennsylvania, Philadelphia, Pa. 19174. Dr. Stunkard is Pro- fessor and Chairman, Department of Psychiatry, Stanford University School of Medicine, Stanford, Calif. 94305. This work was supported in part by Public Health Service grant MH- 15383 from the National Institute of Mental Health. cation of behavior modification. We asked ourselves four questions: 1. Can behavior modification be used in self-help groups for the obese? 2. If so, can it produce greater weight loss and/or lower attrition rates than other treatment approaches? 3. Can self-help group leaders who are given special training carry out a behavior modification program as successfully as professional therapists? 4. Are there any predictors of treatment success? METHOD TOPS (Take Off Pounds Sensibly) is a 25-year-old self-help organization for the obese. It has 320,000 mem- bers in more than 12,000 chapters nationwide; most are women. The 16 chapters in our study were selected from 21 chapters in the Philadelphia area whose weight reduc- tion records we had monitored for the previous four years. Two of the other five chapters had disbanded, two had fewer than six members, and one chapter declined to participate. At the beginning of the study the 16 chapters had 318 members. We deleted from data analysis the records of 20 members who were less than ten percent above their ideal weight; they had already reached their ideal weight or had joined TOPS in order to lose only a small amount of weight. Sixty-four members joined TOPS between the time the study was announced and its start. Interviews re- vealed that knowledge of the impending study infl d at least some of these persons to join TOPS. Accord- ingly, we analyzed the data from this group separately. The main sample thus consisted of 234 subjects. Each of the 16 chapters was assigned to one of four treatment conditions, with four chapters in each condi- ~ Topical Papers is a new section of the Journal. The papers in this section are grouped around a given topic. Publication here does not, however, imply that the Editors consider this material to constitute a comprehensive analysis of the topic. TABLE 1 Subjects’ Characteristics at the Start of Treatment 116 Current Percent Previous Length of Number Weight (1b.) Overweight Weight Loss (Ib.) Age (Years) Membership (Months) Condition of Subjects Mean £ S.D. Mean + S.D. Mean £S.D. Mean + S.D. Mean + S.D Professional behavior modification 73 180.80 + 15.03 404 £238 10.80 + 8.76 45.6 + 10.6 38.24 27.6 Chapter leaders Behavior modification 54 181.31 £11.27 445+ 19.0 11.004+ 9.95 48.7 8.7 379+ 205 Nutrition education 55 180.40 + 14.42 4254 25.1 11.79% 11.95 433 + 148 3534274 TOPS conirol 52 177.50 + 15.71 420+ 22.1 11.31 + 9.50 45.1 £116 336+£249 tion. Active treatment lasted 12 weeks. Behavior Modification Conducted by a Professional Therapist Three psychiatric residents and one graduate student in clinical psychology, all men, conducted these groups. At each weekly chapter meeting the therapist introduced two or three behavior modification techniques. Group members were asked to keep detailed records of thei food intake. Each time a subject ate, she was asked to record the time, place, and duration of eating; the type, amount, and caloric value of the food consumed; and any associated emotions. At cach meeting the therapist taught techniques designed to change any problem habits identified from the food intake records. Among the tech- niques, which are described more fully elsewhere (2), were: 1) introducing changes in the act of eating, includ- ing slowing down the pace and leaving food at the end of a meal; 2) developing control over the stimuli signaling eating, including learning to eat at specific times and in a very limited number of places and removing excess food from the environment; 3) planning food intake well in ad- vance of eating; 4) responding to boredom, fatigue, and emotional states with activities that do not involve eating; and 5) instituting group and individual rewards for be- havior change and weight loss. In addition, a special ef- fort was made to help subjects increase physical activity in their daily lives. Behavior Modification by the TOPS Chapter Leaders After two preliminary training sessions, each leader and coleader attended 12 weekly training sessions con- ducted by the investigators. Concurrently, the leaders in- troduced the previously described behavior modification program to their members at the weekly chapter meet- ings. Nutrition Training TOPS leaders also attended a training program sched- uled precisely as above. They were taught general prin- ciples of nutrition and in turn taught these principles to their members. The program was designed to control for nonspecific training and therapeutic effects. Attendance and weight records in each of these experimental condi- tions were given to the investigators each week. Control Group In the fourth condition the usual TOPS program was continued. This program, which has been described ex- tensively elsewhere (3), includes a weigh-in, an announce- ment of weight gains and losses, rewards and sometimes punishments, group singing, and a general discussion of weight-related topics. Each week the chapters’ atten- dance and weight records were mailed to the investiga- tors. Aside from this data collection, the four chapters in this condition had no contact with the investigators and received no information from them. Following the 12-week active treatment period, the therapist visits and training sessions were discontinued. For the nine-month follow-up period, leaders mailed weekly records of attendance and weights to the investi- gators. The homogeneity of TOPS membership made it pos- sible to match the subjects in the different experimental conditions very closely. Table | shows the remarkable match in age, current weight, percent overweight, length of membership in TOPS, and previous weight loss in TOPS. The average subject was a 45-year-old woman who had been a member of TOPS for three years and who had lost 5.0 kg. (11 Ib.) during her membership. She currently weighed 81.6 kg. (180 Ib.), 42 percent above her ideal weight. RESULTS Attrition rate The first major therapeutic problem in obesity is the number of patients who drop out of treatment; attrition rates ranging from 20 percent to as high as 80 percent are reported (4, 5). Figure 1 presents the attrition rates of each experimental condition during the active-treatment and follow-up periods. During the three months of active {reatment fewer TOPS members dropped out of the two behavior modifi- cation groups than out of the nutrition education and control groups. At 12 months this difference had become striking. Only 38 and 41 percent had dropped out of the 117 FIGURE 1! Rates of Attrition in the Four Treatment Groups @— Behavior modification by professionals o——o Behavior modification by TOPS leaders & ---a Nutrition education & - -a TOPS control 8 SUBJECTS STILL IN TOPS (PERCENT) DURATION IN MONTHS TABLE 2 Mean Weight Change During 12-Week Treatment Period Loss or Gain Pounds Kilograms Condition (Mean +S.D.) (Mean) Professional behavior modification 4.24 £247 =1.92 Chapter leaders Behavior modification -1.90 + 1.88 -0.06 Nutrition education -0.25 + 1.83 -0.11 TOPS control +0.71 + 1.89 +0.32 behavior modification groups, compared with 55 and 67 percent for the nutrition education and control groups re- spectively (x* = 12.35, p < .01). Weight Loss During Treatment What happened to the subjects who remained in treat- ment? It must be remembered that their lower dropout rates seriously bias the results against the behavior modi- fication groups. Previous work has shown that poor weight-losers drop out at a more rapid rate than do those who lose greater amounts (1). Decreasing the attrition rates thus means retaining the less successful members. Despite this bias, groups using behavior modification lost significantly more weight than those in the control conditions (F = 10.7, p < .001). The chapters in which behavior modification was introduced by a professional therapist lost a mean of 1.92 kg. (4.2 Ib.), significantly more (p < .001) than the weight loss in both the nutri- tion education condition (0.11 kg. or 0.2 Ib.) and the FIGURE 2 Mean Weight Change During Treatment and Follow-Up +304 e—s Behavior modification by professionals o—o Behavior modification by TOPS leaders #&- 4 Nutrition education 204 &~--5 TOPS control i . . / ’ £ +104 w o z I 5-0 - I < 5 z "107 < o = ~2.0+ -3.0 T T + a 0 3 6 9 12 Treatment Follow-Up MONTHS TOPS control condition, in which subjects actually gained 0.32 kg. (0.7 Ib.). Chapters in which behavior modification was introduced by professionals lost signifi- cantly more weight (p < .05) than those taught the same program by the TOPS chapter leaders (-0.86 kg. [1.9 b.]). Consider now the three conditions in which leadership was provided by the TOPS chapter leaders. The behavior modification program produced significantly greater weight loss (p < .05) than did the continuation of the usual TOPS program. The difference between the behav- ior modification and nutrition education programs was not statistically significant, although the results favored the behavior modification program. The two control con- ditions did not differ significantly from each other. Behavior modification thus kept more persons in TOPS during and after treatment. It also produced greater weight losses, despite the bias against weight loss produced by the lower attrition rates. Weight Loss During Follow-Up Figure 2 shows the mean weight changes of subjects who stayed in each experimental condition for 12 months. Subjects in the behavior modification groups led by professionals not only maintained their higher weight loss for one year but even increased it slightly. This final mean weight loss of 2.63 kg. (5.8 Ib.) was significantly higher (p < .001) than that of any of the other conditions. It also represented half the weight lost by these subjects 118 TABLE 3 Weight Change at the End of One Year, Given in Percents* Professional Chapter Leaders Behavior Behavior Nutrition TOPS - Weight Change Modification Modification Education Control Gained five pounds 15 21 46 37 Lost five pounds 54 24 15 17 Gained ten pounds 5 6 23 pil Lost ten pounds 24 9 8 10 Gained 20 pounds 0 0 8 0 Lost 20 pounds 15 0 0 0 * The percents given here for weight changes are cumulative, in that the greater losses and gains are included in the lesser categories above them. in their three previous years in TOPS. The initial weight loss of subjects in the behavior modi- fication program conducted by TOPS group leaders was not maintained during follow-up, and the subjects’ weights returned to their pretreatment levels. However, these subjects did better than the control and nutrition education groups. Subjects in these two conditions ac- tually gained 1.27 kg. (2.8 Ib.) and 1.81 kg. (4.0 Ib.) during the follow-up period. In addition to the mean weight changes, we also exam- ined the percentage of subjects gaining and losing certain amounts of weight. Table 3 shows the percentage of sub- jects in each experimental condition who attained three levels of weight change at the end of 12 months. The two behavior modification programs produced significantly greater rates of weight loss, and significantly lower rates of weight gain, at three criterion levels: 2.3 kg. (5 Ib.) (x? = 39.69, p<.001), 4.5 kg. (10 Ib.) (x* = 12.85, p < .05), and 9.1 kg. (20 Ib.) (x* = 18.38, p < .01). In the professional-led behavior modification program, over 50 percent of the subjects lost more than five pounds, while only 15 percent gained that much weight. By contrast, in the TOPS control condition, 37 percent gained more than five pounds while only 17 percent of the subjects lost that amount of weight. Behavior modification by TOPS lead- ers was similar to the nutrition education condition in its rate of weight loss, but resulted in a far lower rate of weight gain. A 9.1 kg. (20 Ib.) criterion has often been used as a single measure of effectiveness. Only subjects in the pro- fessional-led behavior modification program achieved this degree of weight loss. The subjects who lost more than 9.1 kg. shared a striking similarity: all eight had been members of TOPS for more than two years and, during that time, all had gained weight! This criterion seemed an effective predictor of success in treatment. The mean weight loss of all subjects who met it was 8.35 kg. (18.4 1b.), and 88 percent of them lost more than 9.1 kg. These results contrast dramatically with the seven sub- jects who had been in TOPS for the same length of time but who had lost weight during that time. None of these lost more than 9.1 kg. and their mean weight loss was only 0.54 kg. (1.2 1b.). New Members We analyzed separately the results of the small sample (N = 64) of new members. During the active treatment period the behavior modification program again pro- duced lower attrition rates than did the control condi- tions: 11 percent for each of the behavioral conditions, 32 percent for nutrition education, and 20 percent for the TOPS control. Mean weight losses during treatment followed the same pattern as that of longer-term members. The two behavior modification conditions achieved mean weight losses of 4.54 kg. (10 Ib.) and 3.46 kg. (7.6 Ib.) respec- tively. Subjects receiving nutrition education lost an aver- age of 3.23 kg. (7.1 Ib.), and those in the TOPS control condition 1.36 kg. (3.0 Ib.). Because of the small sample size, differences among the groups did not reach statisti- cal significance. During the follow-up period, the dropout rates showed a different pattern from those of old members. Subjects in the three active treatment conditions dropped out at a higher rate than those in the TOPS control condition. Three months after termination of treatment, the attri- tion rates in the behavior modification conditions were 57 and 58 percent; in the nutrition education condition it was 52 percent. By contrast, the TOPS control showed a dropout rate of only 27 percent. At one year, 78 and 89 percent dropped out of the behavior modification groups, 73 percent out of the nutrition education condition, but only 53 percent from the TOPS control groups. Because of the small initial sample and the extremely high drop- out rates in the active treatment conditions, not enough subjects remained in the treatment groups at one year to permit a comparison of final mean weight loss. COMMENTS The introduction of behavior modification techniques substantially improved the effectiveness of patient self- help groups on two interacting and critical measures of success. The program resulted in far lower dropout rates and in greater weight losses for those who remained in treatment, both during the treatment period and at fol- low-up. The overall ineffectiveness of self-help groups for the obese was recently demonstrated in a two-year study of 485 TOPS members (1). Attrition rates were 47 percent in one year and 70 percent in two. Furthermore, attrition was not a result of successful weight reduction; members who dropped out were those who had lost less weight: “Although a small percentage of persons joining TOPS are able to lose substantial amounts of weight and to maintain the weight loss, for the vast majority of mem- bers, TOPS is a relatively ineffective method of weight control.” The present investigation reaffirmed this con- clusion; the four TOPS chapters that served as controls gained weight during the year of study. The behavior modification program helped TOPS 119 members to lose weight. But the amount of weight lost was less than in other studies using similar treatment techniques (6-8). At least four factors may account for this difference. 1. TOPS members are older, less well-educated, and of lower socioeconomic status than were the subjects of pre- vious behavior modification programs. Each of these fac- tors may unfavorably influence the effectiveness of be- havior modification. t 2. The TOPS subjects had already lost weight (mean = 5.0 kg.), whereas other behavior modification studies used subjects new to treatment. A high percentage of obese persons lose weight at the outset of any weight re- duction program. 3. Members who had reached or approximated their ideal weight before intérvention were excluded from the data analysis. We thus started with a sample selected for their inability to lose weight. 4. Probably most important, each TOPS. group was considerably larger than those in earlier studies, which had rarely exceeded eight members. By contrast, our chapters averaged 19 members. This larger size made it impossible to individualize treatment for each member. In addition to these theoretical reasons, a recent study provides further empirical evidence that the performance of TOPS members is poorer than that of the subjects in earlier behavior modification research. A behavior modi- fication program of individualized treatment for a small selected sample of TOPS members produced dropout rates and weight losses quite similar to ours (9). Professional therapists achieved significantly greater weight losses in their subjects than TOPS chapter lead- ers, even though both used the same behavioral program. Equally important, subjects in the professional-led (but not TOPS leader-led) groups maintained their weight loss after the end of treatment. In one year, these subjects increased by one-half the weight loss they had attained in three previous years of TOPS membership. Moreover, only in the professional-led groups were there members who lost as much as 20 pounds. We believe this is the clearest demonstration to date of the greater effectiveness -of professional over non- professional therapeutic intervention. We have earlier shown that it is possible for a nonprofessional, carefully selected for successful weight reduction and leadership qualities, to promote substantial weight losses in a behav- ior modification group (10). But within the context of TOPS, a professional therapist was able to produce greater weight losses than were the TOPS group leaders using any of a variety of treatment approaches. Neither investigators of obesity nor investigators of be- havior, modification have been able to predict an individ- ual’s response to treatment. The identification of even a small group who can predictably lose weight is thus of considerable theoretical importance. Further, the favor- able response of persons whose motivation, unreinforced by weight loss, was high enough to stay in TOPS for two years is of practical importance. TOPS may be particu- larly useful in scréening large numbers of obese persons to identify those with a good prognosis for weight loss. This would permit the more effective deployment of lim- ited professional treatment resources. REFERENCES - Garb J, Stunkard A: A further assessment of the effectiveness of TOPS in the control of obesity, in Proceedings of the Fogarty In- ternational Center Conference on Obesity. Edited by Bray G. Washington, DC, US Government Printing Office (in press) Stuart RB, Davis B: Slim Chance in a Fat World: Behavioral Con- trol of Obesity. Champaign, Ill, Research Press, 1972 Stunkard A, Levine H, Fox S: The management of obesity: patient self-help and medical treatment. Arch Intern Med 125:1067- 1072, 1970 Stunkard A, McLaren-Hume M: The results of treatment of obe- sity. Arch Intern Med 103:79- 85, 1959 Seaton DA, Rose K: Defaulters from a weight reduction clinic. J Chronic Dis 18:1007-1011, 1965 Stuart RB: Behavioral control of overeating. Behav Res Ther 5:357-365, 1967 Penick SB, Filion R, Fox S, et al: Behavior modification in the treatment of obesity. Psychosom Med 33:49-55, 1971 . Stunkard AJ: New therapies for the eating disorders: behavior modification of obesity and anorexia nervosa. Arch Gen Psychiatry 26:391-398, 1972 . Hall SM: Self-control and therapist control in the behavioral treat- ment of overweight women. Behav Res Ther 10:59-68, 1972 10. Jordan HA, Levitz LS: Behavior modification in a self-help group. J Am Diet Assoc 62:27-29, 1973 $l ® No wv a © ptieion