FACTORS IN OBESITY: CURRENT VIEWS

Albert}. Stun kard'

This discussion of factors in obesity will ad-dress three major topics: genetic determinantsof obesity, the distribution of body fat in obe-sity, and the influences of social factors, par-ticularly socioeconomic status (SES).

Any consideration of the role of social fac-tors in the determination of obesity today mustconsider what we know about the role of ge-netic determinants of obesity. This informa-tion is of recent origin and it is necessary toput it into context so that genetic influencesare neither exaggerated nor minimized.

GENETIC DETERMINANTS OF OBESITY

For some time, the existence of numerousforms of genetic obesity in animals and theease with which adiposity can be producedby the selective breeding of farm animals havesuggested that genetic factors may playa simi-larly important role in human obesity, Thestunning advances in our knowledge duringthe past decade have made it dear that geneticfactors do indeed play an important role inhuman obesity.

A previous study, utilizing the classic twinmethod, estimated very high levels of heri-tability (the percentage of variance accountedfor by genetic influences) for the body massindex (BMI-weight (kg)/height2 (rnj). In this

'Professor Emeritus of Psychiatry, University of Penn-sylvania, School of Medicine, Department of Psychiatry,Philadelphia. Pennsylvania. United States of America.

study, heritability was estimated at approxi-mately 80% (1). Even studies of identicaltwins separated at birth, a method that avoidssome of the bias inherent in classic twin stud-ies, estimated heritability at 66% (2). Thesestudies are still widely cited, but there isgrowing consensus that they overestimate theinfluence of heredity.

The results of adoption studies and of com-plex segregation analysis agree on approxi-mately 33% heritability of BMI (3,4), a valuenow viewed as a more reasonable estimatethan that of the twin studies. Genetic influ-ences appear to playa more important rolein determining the distribution of body fat,with a particularly strong influence on thecritical visceral fat depot that is described inthe next section.

The recent entry of molecular genetics intothe study of obesity, with the identification ofmutations of two genes (5-7) that cause obe-sity in mice, promises to greatly advance ourunderstanding of the genetic determinants ofhuman Obesity. Nevertheless, if, as seems likely,the heritability of human obesity is no morethan 33%, then 66% of the variances ·in BMImust be environmental. Thus, although hu-man obesity develops within genetic con-straints, environmental determinants play anenormously important role in its development.

Genetic and environmental determinants ofobesity are not in conflict. It is not a questionof genes or environment or of genes versusenvironment, but of genes and environment;

23

24 Factorsin Obesity: Current Views

neither acts alone to determine the clinicaloutcome. This outcome is determined insteadby the combination of genetic vulnerabilityand adverse environmental events (8). Thiscombination is diagrammed in Figure 1, inwhich the small inner circle represents thosepersons who are genetically predisposed to adisorder. The wedge represents adverse envi-ronmental conditions to which these individu-als may be exposed. The model indicates thatonly those genetically predisposed personswho are exposed to adverse environmentalconditions are clinically affected, as in the caseof obesity.

DISTRIBUTION OF BODY FAT

In the past, various classifications of obesityhave been based upon the character of the pre-dominant adipose tissue, the severity of theobesity, and the age of onset. All of these classi-fication methods are still used to some extent,but they have been superseded to a consider-able degree by the distribution of body fat.

Interest in the distribution of body fat wasaroused in the early 1980s by the finding that

FIGURE 1. Combined effect of genetic pre-disposition to a disorder and exposure to adverseenvironmental conditions on dinical outcomes.

Subject to adverseenvironmental conditions

persons whose fat was located primarily in theupper part of the body suffered far highermortality and morbidity from ischemic heartdisease than did persons whose fat was lo-cated primarily in the lower part of the body(9).Body fat distribution is measured clinicallyby the waist:hip ratio, calculated from thewaist circumference halfway between thelower rib margin and the iliac crest and thehip circumference at the level of the greatertrochanter. Upper body obesity is defined asa waist:hip ratio of more than 1.0 for men and0.8 for women. Risk, however, is directly pro-portional to the size of the waist:hip ratio, in-dependent of gender; the greater mortalityand morbidity of men is a function of theirgreater waist:hip ratio.

Although the waist:hip ratio is still themost widely used clinical measure of bodyfat distribution, a major refinement has beenintroduced by imaging techniques that haveshown that essentially all of the risk of up-per body obesity is conveyed by the visceralfat depot, within the abdominal wall (9). Thisfinding has greatly expanded our under-standing of the complications of obesity andhas provided a rationale for the metaboliccascade that mediates many of the steroidhormones, which increases free fatty flux andleads to decreased hepatic insulin clearanceas well as hyperinsulinemia, hyperlipidemia,hypertension, and ultimately, cardiovas-cular disease.

SOCIAL DETERMINANTS OF OBESITY

The fact that genetic influences account foronly one-third of the variance in body weightmeans that the environment exerts a profoundinfluence. One measure of the extent of thisinfluence is the dramatic increase (33%) of theprevalence of obesity in the United States dur-ing the past decade (10). Unfortunately, ourunderstanding of these important environ-mental determinants is limited and system-atic studies of the topic are rare. Most of thesestudies focus on socioeconomic status.

Stunkard 25

An influential report in 1965 delineated theinfluence of SES on obesity. In a stratifiedsample of 1660 persons in the Midtown Man-hattan Study, as shown in Figure 2, obesity wassix times more prevalent among women oflower SES than among those of upper SES (11).These results were important because of thestrength of the relationship between currentSES and obesity, but the study went further. Itmeasured parental SESwhen the respondentswere eight years old, the so-called "SES of ori-gin," and assessed its relationship to the preva-lence of obesity. As shown in Figure 2, SES oforigin was related to the prevalence of obe-sity aImost as strongly as was the respondents'own SES. The respondents' obesity couldhardly have influenced their SES of origin,strongly suggesting that SES of origin was adeterminant of obesity in this population atleast. Note, however, that the prevalence ofobesity in SES of origin was lower than cur-rent SESamong persons of lower SES and washigher among persons of upper SES. Thesedifferences indicate that, in addition to SESinfluencing obesity, obesity also influencesSES, in that it leads to a decrease in SES.Thus,the prevalence of obesity was nearly twice ashigh (22%) among women who fell in socialclass as it was (12%) among those who rose insocial class. The relationship of social factorsand obesity was far less important among men

FIGURE2. Prevalence of obesity amongwomen, by own socioeconomic status and

socioeconomic status of origin.

)0

~:...20~~C 10

_Own socioeconomic status= Socioeconomic status of origin

o

Socioeconomic status

Source: Goldblatt PB. Moore ME. Stunkard AI. Social factors inobesity. lAMA 1965;192:1039-1044.

than among women. A recent prospectivestudy had provided powerful support for theview that obesity has a deleterious influenceon social functioning: women who had beenobese in adolescence suffered significant so-cial disability in adult life (12).

The Midtown Manhattan Study also re-vealed another social factor related to theprevalence of obesity-acculturation to theu.s. The prevalence of obesity decreased in amonotonic manner over four levels of increas-ing acculturation-foreign-born, second gen-eration in the United States, one parent sec-ond generation and, finally, third and latergenerations.

These findings have been confirmed in nofewer than 54 studies in developed countries,which found a strong inverse relation betweenSES and obesity among women (13). Further-more, two prospective longitudinal studiesfrom England have provided strong confirma-tion that SES is a determinant of obesity. Theyhave shown that both girls and boys born intoa lower SES were more overweight as adultsthan those born into a higher SES (14,15).

The simplest explanation for these findingsis that the relation of SES and obesity is bi-directional; SES determines the prevalence ofobesity and obesity leads to a decline in SES.There is, however, a complication. Anotherfactor or factors may influence both SES andobesity (16). An example of such a commonfactor is heredity. As we have noted, geneticfactors influence obesity. Less well-known isthe fact that genetic factors may also influenceSES. Thus, studies of Danish adoptees haverevealed that the SES of biological parentsinfluences the SES of their offspring, eventhough they have had no personal contactwith them (17). A path analysis helps to ex-plain this surprising finding. It shows that theinfluence ofSES on obesity is mediated via theintelligence quotient (IQ). Biological parentsinfluence the IQ of their children and IQ, inturn, influences obesity: the higher the IQ, thelower the prevalence of obesity. It is notewor-thy that this influence was present even whenthe SES of the adoptee was controlled. It ap-

26 Factors in Obesity:Current Views

pears that the relationship between SES andobesity is indeed complex (16).

The relationship between SES and obesityamong women in developed societies is notfound regularly among men or children.Among these groups, a significant percentagemanifest either no relationship between SESand obesity or a direct relationship: the higherthe SES the greater the prevalence of obesity.

Whe~ we turn from developed to develop-ing societies, there is a complete reversal inthe relationship between SES and obesity. Inevery developing society that has been stud-ied, there is a direct and often very strong re-lationship between SES and the prevalence ofobesity in men, women, and children (13).

What is the reason for this striking differ-ence in the relationship between SESand obe-sity in developed and developing societies?Four factors may mediate the inverse relation-ship between SES and obesity among womenin developed societies, particularly factors thatcontrol obesity among women of upper SES.

One influence, and probably the most im-portant one, constraining the prevalence ofobesity among women of upper SES in devel-oped societies is dieting and dietary restraint.Women of higher SES diet more often than dowomen of lower SES, have greater access toresources that facilitate dieting, have greaterknowledge of nutrition, and are more commit-ted to the view that slimness is desirable and,therefore, are more motivated to achieve it.

A second direct influence on the control ofobesity among upper SES women is theirgreater physical activity, derived from theirgreater leisure time and greater opportunityfor recreational exercise. Interestingly, patho-logically increased physical activity character-izes the young women of upper SES whosedieting leads to anorexia nervosa.

A third factor mediating the inverse rela-tionship between SES and obesity amongwomen is social mobility. As noted in the Mid-town Manhattan Study, the prevalence of obe-sity is nearly twice as high among women whofall in social class than it is among women whorise in it (11). One of the longitudinal studies

in Britain confirmed and extended this find-ing. It showed that obesity was significantlyless prevalent (5%) among women who rosein social status than among those (11%) whoremained in their social class of origin (14).As is the case with SES itself, social mobilityplays a far less important part in determiningthe prevalence in obesity among men.

A fourth factor influencing the relationshipbetween SES and obesity is heredity. As wehave seen, studies of Danish adoptees haverevealed a significant influence of the SES ofbiological parents upon the prevalence of obe-sity in their children, with whom they havehad no personal contact. This influence ap-pears to be genetically transmitted via IQ.

The strong direct relationship between SESand obesity in developing societies has a morestraightforward rationale than does the in-verse relationship in developed societies. Thelow prevalence of obesity in developing so-cieties appears to be due to a lack of food,coupled with cultural values favoring fat bod-ies. Obesity may be a sign of health and wealth Iin these societies, the opposite of its signifi-cance in developed societies. In such societ-ies, the biological propensity to store fat is as-sociated with, and may even influence culturalevolution that selects "fatness" as a valuedtrait Among most of the 58 traditional culturesfor which information is available, "plump-ness" is viewed as an ideal of feminine beautyand a symbol of prestige (18). In circumstancesof relative deprivation, members of the upperSES may have access to sufficient food to be-come fat; members of lower SES do not.

Because of the enormous importance of en-vironmental factors in determining obesity, itis unfortunate that our information about fac-tors influencing the prevalence of obesity islargely limited to SES. The effort to controlobesity will require a far better understand-ing of the social factors that promote it than isnow available. There is, therefore, an urgentneed to explore other social factors in addi-tion to SES. A few recent examples illustratethe direction in which such study may profit-ably be conducted.

Stunkard 27

Among adolescent girls in the United States,the inverse relationship between SES andobesity, which had been so firmly establishedamong women, is absent among African-American adolescent girls. Evidently thesegirls were not affected by the messages onbody shape that prevail in Ll.S.culture. Analy-sis of the recent Hispanic Health and Nutri-tion Examination Survey (HANES) has re-vealed a similar lack of relationship betweenSES and obesity among certain groups: PuertoRican women and Mexican-Americans andCuban-Americans of both genders (19). Thisstudy also suggests social determinants dif-ferent from those that have been the subjectof most of the research to date (Kahn LK, per-sonal communication).

Along with these essentially negative find-ings, an interesting positive finding hasemerged from the Hispanic HANES analysis.Acculturation to the American lifestyle byMexican-American men is associated with anincreased prevalence of obesity.

FUTURE DIRECTION

In the future, epidemiologic studies of obe-sity will benefit from measurement and con-trol of genetic influences. A measure of geneticinfluence may be obtained through assessmentof parental obesity, but a more precise measureawaits the detection of genetic markers.

There is an urgent need to move beyond thetraditional environmental measure of socio-economic status as a determinant of the preva-lence of obesity. One useful measure has beenthe disaggregation of the three traditionalcomponents of SES--income, occupation, andeducation-relating each of them individuallyto the prevalence of obesity. Among the threecomponents, years of education is the easiestto measure and has the greatest cross-culturalrelevance. It may also be the most powerful.

New measures of social influences shouldbe sought and Latin America provides an ex-cellent opportunity for such a search. Accul-turation is a promising candidate. Other mea-

sures undoubtedly will present themselves tothe inquisitive investigator; they should beexplored with vigor and dispatch.

REFERENCES

1. 5tunkard AJ, Foch IT, Hrubec Z. A twin study ofhuman Obesity. lAMA 1986;256:51-54.

2. 5tunkard AJ, Harris JR. Pedersen NL, McCleam GE.The body-mass index of twins who have been rearedapart. N Engl I Med 1990;322:1483-1487.

3. Volglec Gp, Serensen TIA, Stunkard AJ, SrinivasanMR, Rao DC. Influences of genes and shared familyenvironment on adult body mass index assessed inan adoption study by a comprehensive path modelI11t I Olles 1995;19:40-45.

4. Bouchard C, ed. T~ genetic» of obrsity. Boca Raton:CRC Press; 1994.

5. Michaud EJ, Bultman 5J, Stubbs LJ, Woychik RP. Theembryonic lethality of homozygous lethal yellowmice (Ay / Ay) is associated with the disruption of anovel RNA-binding protein. Genes Deu 1993;7:1203-1213.

6. Michaud EJ, Bultman 5J, KJebig Ml, van Nugt MJ,Stubbs LJ, Russell LB, et al. A molecular model forthe genetic and phenotypiC characteristics of themouse lethal yellow (Ay) mutation. Proc Natl AcadSci USA 1994;91:2562-2.566.

7. Zhang Y, Proenca R, Maffei M, Barone M, Leopold L,Friedman J. Positional cloning of the mouse ooesegene and its human homologue. Nature 1994;972:425-432-

8. 5tunkard AJ. Genetic contributions to human bodyobesity. In: McHugh PR, McKusick VA, eds, Genes,brain, and behavior. New York: Raven Press; 1990:205-219.

9. Sjostrom L. Impacts of body weight, body com-position, and adipose tissue distribution on morbid-ity and mortality. In: Stunkard AJ, Wadden TA, eds.Obesity: theDry and therapy. New York: Raven Press;1993:13--41.

10. Kuczmarski RJ, Flegal KM, Campbell SM, JohnsonCL. Increasing prevalence of overweight among USadults. lAMA 1994;272:205-211.

11. Goldblatt PB, Moore ME, Stunkard AJ. Social factorsin obesity. lAMA 1965;192:1039-1044.

12. Gortmaker SL, Must A, Perrin JM, Sobol AM, DietzWH. Social economic consequences of overweight inadolescence and young adulthood. N Eng! J Med1993;329:1008-1012.

13. Sobal J, Stunkard AJ. Socioeconomic status and obe-sity: a review of the literature. Psychol Bull 1989;105:260-275.

14. Braddon FE, Rogers B, Wadsworth ME, Davies J.Onset of obesity in a 36 year birth cohort study. BrMed 11986;293:299-303.

28 Factorsin Obesity:CurrentViews

15. Power C, Moynihan C. Social class changes andweight-for-height between childhood and earlyadulthood.Jnt / Obes 1988;12:445-453.

16. 5tunkard AJ, Serensen TIA. Obesity and socioeco-nomic status-a complex relation. N Engl , M~d1993;329:1036-1037.

17. Teasdale TW,Serensen TIA. Educational attainmentand social dass in adoptees: genetic and envirorunen-tal contributions., BiosocSci 1983;15:509-518.

18. Brown PJ, KOMer M. An anthropologica1 perspec-tive on obesity. Ann NY Aaui Sci 1987;499:29-46.

19_ Maurer ICR. Plan and operation of the HispanicHealth and Nutrition Examination Survey, 1982-84.United States Department of Health and Human Ser-vices, Centers for Disease Control and Prevention,National Center for Health Statistics. Villli He/llth 511111985;1:19. (PHS) 85-1321.