Body weight and obesity in adults and self-reported abuse in childhood

Abstract

BACKGROUND: Little is known about childhood factors and adult obesity. A previous study found a strong association between childhood neglect and obesity in young adults.

OBJECTIVE: To estimate associations between self-reported abuse in childhood (sexual, verbal, fear of physical abuse and physical) adult body weight, and risk of obesity.

DESIGN: Retrospective cohort study with surveys during 1995–1997.

PATIENTS: A total of 13 177 members of California health maintenance organization aged 19–92 y.

MEASUREMENTS: Body weight measured during clinical examination, followed by mailed survey to recall experiences during first 18 y of life. Estimates adjusted for adult demographic factors and health practices, and characteristics of the childhood household.

RESULTS: Some 66% of participants reported one or more type of abuse. Physical abuse and verbal abuse were most strongly associated with body weight and obesity. Compared with no physical abuse (55%), being ‘often hit and injured’ (2.5%) had a 4.0 kg (95% confidence interval: 2.4–5.6 kg) higher weight and a 1.4 (1.2–1.6) relative risk (RR) of body mass index (BMI)≥30. Compared with no verbal abuse (53%), being ‘often verbally abused’ (9.5%) had an RR of 1.9 (1.3–2.7) for BMI≥40. The abuse associations were not mutually independent, however, because the abuse types strongly co-occurred. Obesity risk increased with number and severity of each type of abuse. The population attributable fraction for ‘any mention’ of abuse (67%) was 8% (3.4–12.3%) for BMI≥30 and 17.3% (−1.0–32.4%) for BMI≥40.

CONCLUSIONS: Abuse in childhood is associated with adult obesity. If causal, preventing child abuse may modestly decrease adult obesity. Treatment of obese adults abused as children may benefit from identification of mechanisms that lead to maintenance of adult obesity.

Introduction

The etiology of adult obesity is poorly understood. Although about 50% of variation in body weight is inherited, obesity etiology is complex involving interactions among multiple genes, environmental factors and behaviors.1 One view is that obesity results from an inability to balance food intake with physical activity in a culture that aggressively promotes food consumption and sedentary living.2,3 Although family environment is considered a key factor in the development of child obesity,4 few studies have examined the impact of childhood factors on adult obesity. Most studies have used clinical samples lacking normal-weight controls.5,6

A notable exception is the work of Lissau and Sørensen, who carried out a 10 y follow-up of a random sample of 9–10-y-old children in Copenhagen.7 At baseline the children's teachers and school nurses reported their impressions of the students' family structure, parental support and general hygiene; at follow-up the heights and weights of the participants were re-measured. After controlling for body weight in childhood, children characterized as ‘dirty and neglected’ had 10 times the risk of becoming obese adults as those with ‘average’ childhood hygiene. The authors noted that in an earlier study on heritability of adult obesity,8 genetic factors approximately doubled the risk of adult obesity.7

Lissau's and Sørensen's study, however, may not apply to other aspects of child maltreatment, such as child abuse, nor to obesity risk in older adults. The purpose of this study was to investigate relationships between abuse in childhood and adult body weight and risk of obesity. We studied adults (mean age 56 y) enrolled in the Kaiser Permanente health maintenance organization (HMO) in San Diego, California.

Methods

Study population

Data are from the Adverse Childhood Experiences (ACE) Study, whose objective is to assess the impact of childhood experiences on adult health behaviors and outcomes.

Annually, more than 45 000 adult members of the Kaiser Permanente HMO undergo a standardized medical examination at the organization's health appraisal clinic. Approximately 80% of members are evaluated at that clinic, primarily for preventive health assessments rather than symptom/illness-based care. The ACE study sampled all adult members aged ≥19 y examined at the clinic during two time periods: August 1995–March 1996 (wave 1) and June 1997–October 1997 (wave 2). One week after their clinic visit, participants were mailed a questionnaire about their childhood experiences and current health behaviors.

Prior publications from the ACE Study9,10,11 included only respondents to the wave 1 survey, which had a response rate of 70% (9508/13 494); the wave 2 survey had a response rate of 65% (8667/13 330). Thus, the response rate for the two waves combined was 68% (18 175/26 824). Because 754 persons responded to both waves, the unduplicated number of respondents was 17 421. All questions appeared on both survey waves.

The ACE Study was approved by the institutional review boards of the Southern California Permanente Medical Group and of Emory University, and by the Office of Protection from Research Risks, National Institutes of Health.

Study variables

Questions about childhood experiences on the mailed questionnaire pertained to the respondent's first 18 y of life. Data were collected on four types of child abuse: sexual, verbal, fear of physical and physical. For each type of abuse we defined three or four categories, with one category classified as severe.

Abuse variables

Four questions from Wyatt12 were adapted to assess contact sexual abuse. The questions were introduced with the statement, ‘Some people, while they are growing up in their first 18 y of life, had a sexual experience with an adult or someone at least 5 y older than themselves. These experiences may have involved a relative, family friend or stranger. During the first 18 y of life, did an adult, relative, family friend, or stranger ever… (1) Touch or fondle your body in a sexual way?, (2) Have you touched their body in a sexual way?, (3) Attempt to have any type of sexual intercourse with you (oral, anal or vaginal)?, (4) Actually have any type of sexual intercourse with you (oral, anal or vaginal)?’

The categories of sexual abuse were none, touched only, attempted intercourse and intercourse (the last defined as severe).

Questions on verbal, fear of physical and physical abuse were adapted from the Conflict Tactics Scale,13 in which response categories are never, once or twice, sometimes, often, or very often. ‘Once or twice’ and ‘sometimes’, were combined as ‘sometimes’, ‘often’ and ‘very often’ were combined as ‘often’.

Verbal abuse was assessed by asking, ‘How often did a parent, step-parent, or adult living in your home swear at you, insult you, or put you down?’ Categories were none, sometimes and often (severe).

Fear of physical abuse was assessed by asking, ‘(1) How often did a parent, step-parent, or adult living in your home act in a way that made you afraid that you might be physically hurt? and (2) How often did a parent, step-parent, or adult living in your home threaten to hit you or throw something at you, but didn't do it?’ Categories were none, sometimes and often (severe).

Physical abuse was assessed by asking, ‘(1) How often did a parent, step-parent, or adult living in your home push, grab, slap or throw something at you? and (2) How often did a parent, step-parent, or adult living in your home hit you so hard that you had marks or were injured?’ Categories were none; hit, not injured; sometimes hit and injured, and often hit and injured (the last two categories defined severe).

We also created four dichotomous abuse variables: any mention of abuse; mention of one or more types of severe abuse; mention of all four types of abuse, regardless of severity; and mention of all four types of abuse, with all types severe.

Weight and height

During the clinic examination, height to the nearest inch and weight to the nearest pound were measured while enrollee wore an examination gown without shoes. Body mass index (BMI) was computed by dividing weight in kg by height in m2. We defined obesity as BMI≥30, and severe obesity as BMI≥40.14

Covariates

Age was coded as: 19–34, 35–49, 50–64, or ≥65 y. Ethnicity was white, black, Hispanic, Asian or other. We used four dichotomous variables (violence against mother, alcohol/drug abuse, mental illness, household member in prison) to assess level of dysfunction in the childhood household.15 Smoking status was coded as never, former or current, and alcohol consumption as none, 0.1–0.5, 0.6–1.0 and ≥1.0 drinks per day. Recreational physical activity was coded as none, 1–15, 16–30, 31–60, or ≥60 min per week. Education was classified as less than high school, high school graduate, some college, or college graduate, and employment status as unemployed, employed full-time (≥35 h/week), employed part-time (1–34 h/week), retired, or homemaker/student. Among women, number of births was coded as 0, 1, 2, 3, 4 or ≥5.

Exclusions

We excluded 796 respondents who were pregnant at the time of the mail surveys and 128 who had missing values for weight and height. Additional exclusions included: 1190 for the abuse variables; 424 for characteristics of the childhood household environment; 184 for age, race or sex; and 1522 for smoking status, physical activity, alcohol, education, employment or live births. These exclusions left 13 177 or 76% of the eligible sample.

Observations with missing values for the abuse variables were re-coded in two ways: ‘not abused’ or ‘severely abused’. The analysis was repeated but results were very similar regardless of whether missing values were excluded or re-coded.

Data analysis

We used Poisson regression to examine dependence among the four types of child abuse. Linear regression was used to estimate mean differences in body weight (kg) between those exposed and unexposed to abuse. We used logistic regression to estimate relative risks from abuse for BMI ≥30 and ≥40. Because odds ratios from logistic regression may overestimate relative risks, we used predictive margins16 to estimate relative risks from predicted values produced by logistic regression. Standard normal confidence intervals for log-relative risks were calculated using bootstrap methods with 200 replications,17 then transformed back to the original relative risk scale.

In regression analyses we estimated differences in body weight and relative risks from three models with increasing levels of control for covariates. Model 1 included categories for a single type of abuse and was adjusted for survey wave (linear regression models were also adjusted for height and height2), sex, age and ethnicity. Model 2 was additionally adjusted for household characteristics during childhood (violence against mother, alcohol/drug abuse, mental illness, member in prison) and adult characteristics (smoking status, physical activity, alcohol consumption, education, employment status, and number of births (women)). For visual clarity we report the 95% confidence intervals only for model 2.

For completeness we show results for model 3, in which we further adjusted for the other three types of abuse. This approach may underestimate the impact of each type of abuse if the four types of abuse co-occur.

Population attributable fractions (PAF)18 and exposed attributable fractions (EAF) were estimated for each of the four dichotomous abuse variables. Predictive margins from logistic regression models were used to estimate adjusted PAF and EAF.19 We used bootstrap methods to calculate 95% confidence intervals for attributable fractions. The PAF is an estimate of the proportion of disease cases (eg BMI≥30) in the population that would be prevented by eliminating the exposure (eg child abuse). The EAF is an estimate of the proportion of cases, among those exposed, that would be prevented if they were not exposed. The assumption underlying attributable fractions is that exposure causes the outcome.

Interaction between the abuse variables and survey wave, sex, age group (≤55 vs >55 y), and ethnic group (white, black, Hispanic, Asian, other) was assessed by comparing models with and without the interaction terms. For linear regression we used the multiple partial F-test, and for logistic regression we used the chi-square test of difference in log-likelihoods. We found no evidence of interaction by survey wave, sex, age, or ethnic group.

We used SAS statistical software20 for the Poisson (GENMOD), linear (REG), and logistic (LOGISTIC) regression analyses; GAUSS21 was used for predictive margin and bootstrap estimation.

Results

Sample characteristics

Mean age of participants was 55.7 y, 51% were women, nearly one-quarter were ethnic minorities and over 40% were college educated (Table 1). Among women, the modal number of births was two. Mean BMI of participants was 27.4; 25% had a BMI≥30, and 2.4% had a BMI≥40. Nine percent of participants currently smoked, almost four-fifths spent ≤15 min/week in recreational activity, and nearly 60% consumed alcohol. During childhood, over 20% witnessed violence against their mother, 28% witnessed alcohol or drug abuse in their households, 20% had a mentally ill household member, and 5% had a household member in prison.

Table 1 Characteristics of the Adverse Childhood Experiences (ACE) Study samplea

Distribution of abuse types

Two-thirds of participants reported one or more type of childhood abuse (Table 2). The most common type was verbal (47.3%), followed by physical (44.5%), fear of physical abuse (42.7%) and sexual (21.7%). Single types of abuse occurred much less than expected if the abuse types were independent; 4702 participants were expected to report only one type of abuse, but only about half this number (2486) did so. Similarly, 258 participants were expected to report all four types of abuse, but nearly five times this number (1264) did so. The hypothesis of independence among abuse types was strongly rejected (χ2=11 278, d.f.=11, P<0.0001).

Table 2 Distribution of types of abuse by number of types reported in the Adverse Childhood Experiences (ACE) Study

For each type of abuse, the frequency of severe abuse rose with increasing number of types of abuse reported. For example, among those exposed to only physical abuse, 99 (18%) reported being injured, while among those reporting all four types of abuse, 836 (66%) reported being injured. This pattern was repeated for sexual, verbal and fear of physical abuse.

Abuse and adult body weight

All types of abuse were associated with increased weight in adulthood (Table 3). After adjustment for all covariates (model 2), often hit and injured was associated with the largest increase in body weight (4.0 kg). The other types of abuse were associated with weight increases about half as large. Because the types of abuse are highly correlated, each abuse association decreased in magnitude after adjustment for the three other types of abuse (model 3).

Table 3 Mean differences in adult body weight associated with type of childhood abuse in the adverse childhood experiences (ACE)a

Abuse and adult obesity

After adjustment for all covariates (Table 4, model 2), the risk of BMI≥30 increased most strongly for being often hit and injured (+39%). The other types of abuse were associated with risk increases of less than 30%. When each type of abuse was adjusted for the other three types, the relative risks decreased because of correlation among the abuse types (model 3).

Table 4 Relative risk of BMI≥30 and BMI≥40 associated with type of childhood abuse in the Adverse Childhood Experiences (ACE) Studya

Risk of BMI≥40 was more strongly related to abuse than risk of BMI≥30. Being often verbally abused had the largest increase in risk of 88%. Being often hit and injured increased the risk by 71%, intercourse and attempted intercourse increased risk by 42 and 37%, while often being in fear of physical abuse increased risk by 34%. Adjustment for the other three types of abuse (model 3) reduced the relative risk for often fearing physical abuse by about 45%, but had less impact on the other three types of abuse.

The risks of BMI≥30 and ≥40 both increased with increasing number of severe types of abuse (Figure 1). Exposure to all four types of severe abuse (1%) had RR for BMI≥30 and ≥40 of 1.46 (1.16–1.85) and 2.54 (1.21–5.35). There was no evidence that risk of obesity increased with increasing number of non-severe types of abuse (data not shown).

Figure 1
figure1

Relative risks of having a BMI≥30 (white bars) or a BMI≥40 (grey bars) by number of severe types of severe abuse; the referent group is those reporting no types of severe abuse. Relative risks are adjusted for all covariates including the number of types of abuse that occurred at non-severe levels. Tests for linear trend were statistically significant for BMI≥30 (P<0.001) and for BMI≥40 (P=0.037).

Attributable fractions for adult obesity

The four dichotomous abuse variables were associated with increases of about 15–25% in the risk of BMI≥30 (Table 5). The largest population attributable fraction (PAF) of BMI≥30, 8%, was for abuse defined as ‘any mention’. The largest EAF of BMI≥30, 19.4%, was for abuse defined as ‘four types-severe’. The confidence interval around the EAF, however, was wide.

Table 5 Childhood abuse relative risks (RR), population attributable fraction (PAF%) and exposed attributable fraction (EAF%) for adult BMI≥30 and ≥40, using dichotomous definitions of childhood abuse: Adverse Childhood Experiences (ACE) Studya

The four definitions of abuse were associated with increases of about 20–90% in risk of BMI≥40. The largest PAF of BMI≥40 was 17.3% for ‘any mention’ of abuse, and the largest EAF of BMI≥40 was 48.3% for ‘four types-severe’. The confidence intervals around the attributable fractions were wide, however.

Discussion

We found that adults who reported sexual and verbal abuse, fear of physical abuse, and physical abuse in childhood were, on average, 0.6–4.0 kg heavier than adults who did not report abuse in childhood. These weight increases were translated into higher risks of adult obesity. Increases in risk of BMI≥30 ranged from 6 to 39%, and increases in risk of BMI≥40 ranged from 6 to 88%. Frequent verbal abuse and frequent physical abuse with injury were most strongly associated with increased risk of obesity in adulthood. In addition, the risk of obesity increased with the number of types of severe abuse.

We also estimated the fraction of adult obesity cases that were attributable to childhood abuse. In our study population, 8% of cases of BMI≥30 and 17% of cases of BMI≥40 were attributable to abuse in childhood. Surprisingly, we were unable to identify any other published estimates of the fraction of adult obesity attributable to child abuse, or attributable to any other exposure. In one study of interest an estimated 17% of the cases of bulimia in US women was attributable to childhood sexual abuse.22

We found a weaker association between adult obesity and child abuse than the association found by Lissau and Sørensen between adult obesity and child neglect.7 The impact of childhood exposures on adult body weight may weaken over time; Lissau and Sørensen studied young adults aged 20 y, while mean age in our study was 56 y. Their study was prospective, while our retrospective study required participants to recall childhood experiences that often occurred decades in the past. Thus, our findings may be attenuated by failure to recall abuse in childhood by memory lapses, or from repression of those memories. Furthermore, mechanisms by which childhood neglect results in adult obesity may be more powerful than mechanisms of abuse. In the child maltreatment literature, neglect is defined as acts of omission including failure to provide for basic biological needs, abandonment, or lack of supervision; abuse is defined as acts of commission by intentionally inflicted behaviors that can harm a child.23 It is likely, however, that neglect and abuse are not independent and also tend to co-occur.

Our findings may also be attenuated because we adjusted for participants' adult characteristics. These characteristics (smoking status, physical activity, alcohol consumption, education, employment status, and number of births (women)) are potential intervening variables because they occurred after the childhood abuse, and may have been at least partially ‘caused’ by abuse in childhood. Thus by adjusting for these variables we may have ‘over-controlled’ in our analysis, and the estimates presented in model 2 may be biased downard.

Lissau and Sørensen suggested that neglect ‘may cause a psychological state that affects energy balance by altering behavior (overeating and physical inactivity) or hormone balances, influencing fat storage (corticosteriods, catecholamines, or insulin).’7 Felitti suggested that compulsive eating by patients who were sexually abused in childhood was an attempt to manage the dysphoria related to negative childhood experiences.24 These mechanisms could also account for maintenance of adult obesity, even if maltreatment in childhood did not initially cause obesity.

Childhood abuse, especially sexual abuse, has been documented among obese persons seeking weight loss.25 Among our sample reporting any abuse (67%), the fraction of obesity cases attributable to abuse was 11% for BMI≥30 and 23% for BMI≥40. This suggests that for a substantial minority of obese adults abused as children, their obesity may result from childhood experiences.

Causality, however, cannot be established from a single observational study. Causality can only be established using external judgements about the plausibility of a study's findings, the putative role of bias in the findings, and the replication of the findings over repeated studies by independent investigators. Thus, our use of attributable fractions has limited value if exposure to childhood abuse does not cause adult obesity. In this study we did not know when obesity first occurred. If obese participants were also obese as children26 they may have been abused because they were obese.27 If so, control for childhood obesity would have reduced the associations in our study. Childhood weights and heights, however, were not collected. Control for childhood obesity, however, may be inappropriate if the first occurrence of obesity is not well established, and may inappropriately attenuate the relationship between child abuse and adult obesity. It is noteworthy that Lissau and Sørensen7 controlled for child obesity in their analysis but found no impact on the association between childhood neglect and adult obesity.

It is possible that obese adults are more likely than normal-weight adults to falsely report abuse or they may be more sensitive reporters of actual abuse. We were unable to identify any studies to support either of these possibilities, however. Only 51% of all available adult members of Kaiser Permanente were included in our study sample. Thus it is possible that extremely obese persons and persons most seriously abused in childhood were under-represented in our study. If non-response occurred more frequently in those who were both obese and abused then we have underestimated the association of childhood abuse with adult obesity.

In theory, prospective studies would avoid these potential biases. In practice, however, prospective studies are difficult to conduct properly because of misclassification of abuse. Studies that use legal records to classify children as abused may misclassify cases that never reach the criminal justice system as ‘not abused’, thus biasing results toward the null.28 Parental or caretaker reports would be unlikely alternatives, given the social and legal implications of self-identification as a child abuser. In contrast, studies of neglect can use more objective reports of teachers and health workers for exposure classification.

Our study has several strengths. First, study participants were unaware that their body weight, measured at a routine clinical exam, would be related to their later reports of childhood experiences. Because participants did not know study intent, we believe there was no systematic over-reporting of abuse among obese participants. Second, the study's large sample size allowed control for 14 separate covariates, each specified as a categorical variable. This approach minimizes residual confounding. Third, in terms of study variables the participants were similar to those in the broader general population. For example, in our study the mean BMI was 27.4, and the prevalence of BMI ≥30 and ≥40 was 24.7 and 2.4%, respectively. In US adults aged ≥20 y these estimates are 26.3, 22.5 and 2.8%.29,14 In our study 22% had a history of sexual abuse in childhood, compared with 27% in US adults.30 A history of physical abuse with injury was reported by 21% of our study participants, the same as for women, but below the 31% for men in the population of Ontario, Canada.31

We found strong evidence that different types of child abuse co-occur, with nearly three-quarters of the abused reporting two or more types of abuse. The strong co-occurrence of different types of abuse in this and other studies32 questions the appropriateness of studying only single types of abuse, and of adjusting the effects of individual types for each other.

In summary, we found that abuse in childhood is related to increased body weight and risk of obesity in middle age. The proportion of cases of obesity that could be prevented with elimination of childhood abuse was modest, however. Still, given the multi-factorial etiology of obesity it is unlikely that any exposure, be it genetic, behavioral or environmental, will account for a large proportion of cases in the population. Our study suggests that clinical management of obese adults who were abused as children may benefit from understanding the mechanisms by which child abuse leads to maintenance of adult obesity. This understanding may be achieved by increased availability of psychological assistance for such patients.

References

  1. 1

    Comuzzie AG, Allison DB . The search for obesity genes Science 1998 280: 1374–1377.

    CAS  Article  Google Scholar 

  2. 2

    Hill JO, Peters JC . Environmental contributions to the obesity epidemic Science 1998 280: 1371–1374.

    CAS  Article  Google Scholar 

  3. 3

    Stamler J . Epidemic obesity in the United States Arch Intern Med 1993 153: 1040–1044.

    CAS  Article  Google Scholar 

  4. 4

    Dietz WH . Prevention of childhood obesity Pediatr Clin N Am 1986 33: 823–833.

    Article  Google Scholar 

  5. 5

    Felitti VJ, Williams SA . Long-term follow-up and analysis of more than 100 patients who each lost more than 100 pounds Permanente J 1998 2: 17–21.

    Google Scholar 

  6. 6

    Maddi SR, Khoshaba DM, Persico M, Bleecker F, VanArsdall G . Psychosocial correlates of psychopathology in a national sample of the morbidly obese Obes Surg 1997 7: 397–404.

    CAS  Article  Google Scholar 

  7. 7

    Lissau I, Sørensen TIA . Parental neglect during childhood and increased risk of obesity in young adulthood Lancet 1994 343: 324–327.

    CAS  Article  Google Scholar 

  8. 8

    Stunkard AJ, Sørensen TIA, Hanis C, Teasdale TW, Chakraborty R, Schull WJ, Schulsinger F . An adoption study of human obesity New Engl J Med 1986 314: 193–198.

    CAS  Article  Google Scholar 

  9. 9

    Felitti VJ, Anda RF, Nordenberg D, Williamson DF, Spitz AM, Edwards V, Koss MP, Marks JS . Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults Am J Prev Med 1998 14: 245–258.

    CAS  Article  Google Scholar 

  10. 10

    Dietz PM, Spitz AM, Anda RF, Williamson DF, McMahon PM, Santelli JS, Nordenberg D, Felitti VJ, Kendrick JS . Unintended pregnancy among adult women exposed to abuse or household dysfunction during their childhood JAMA 1999 282: 1359–1364.

    CAS  Article  Google Scholar 

  11. 11

    Anda RF, Croft JB, Felitti VJ, Nordenberg D, Giles WH, Williamson DF, Giovino GA . Adverse childhood experiences and smoking during adolescence and adulthood JAMA 1999 282: 1652–1658.

    CAS  Article  Google Scholar 

  12. 12

    Wyatt GE . The sexual abuse of Afro-American and White-American women in childhood Child Abuse Negl 1985 9: 507–519.

    CAS  Article  Google Scholar 

  13. 13

    Straus M, Gelles RJ . Physical violence in American families: risk factors and adaptations to violence in 8,145 families Transaction Press: New Brunswick, NJ 1990.

    Google Scholar 

  14. 14

    Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL . Overweight and obesity in the United States: prevalence and trends, 1960–1994 Int J Obes Relat Metab Disord 1998 22: 39–47.

    CAS  Article  Google Scholar 

  15. 15

    Schoenborn CA . Exposure to alcoholism in the family: United States, 1988 Advance Data From Vital and Health Statistics, no. 205. US Department of Health and Human Services publication (PHS) 95-1880 National Center for Health Statistics: Hyattsville, MD 1995.

    Google Scholar 

  16. 16

    Graubard BI, Korn EL . Predictive margins with survey data Biometrics 1999 55: 652–659.

    CAS  Article  Google Scholar 

  17. 17

    Efron B, Tibshirani RJ . An introduction to the bootstrap Chapman and Hall: New York 1993.

    Google Scholar 

  18. 18

    Rockhill B, Newman B, Weinberg C . Use and misuse of population attributable fractions Am J Pub Health 1998 88: 15–19.

    CAS  Article  Google Scholar 

  19. 19

    Greenland S, Drescher K . Maximum likelihood estimation of the attributable fraction from logistic models Biometrics 1993 40: 865–872.

    Article  Google Scholar 

  20. 20

    SAS Institute. SAS version 8.00 for Windows SAS Institute: Cary, NC 2000.

  21. 21

    Aptech Systems. GAUSS Mathematical and statistical system Aptech Systems Inc.: Maple Valley, WA 1996.

  22. 22

    Wonderlich SA, Wilsnack RW, Wilsnack SC, Harris TR . Childhood sexual abuse and bulimic behavior in a nationally representative sample Am J Public Health 1996 86: 1082–1086.

    CAS  Article  Google Scholar 

  23. 23

    Lutzker JR, Boyle CL . Preventing physical and sexual abuse In Innovative strategies for preventing psychological problems Glenwick D, Jason L (Eds) Springer: New York 2002.

    Google Scholar 

  24. 24

    Felitti VJ . Childhood sexual abuse, depression, and family dysfunction in adult obese patients: a case control study South Med J 1993 86: 732–736.

    CAS  Article  Google Scholar 

  25. 25

    King TK, Clark MM, Pera V . History of sexual abuse and obesity treatment outcome Addict Behav 1996 21: 283–290.

    CAS  Article  Google Scholar 

  26. 26

    Serdula MK, Ivery D, Coates RJ, Freedman DS, Williamson DF, Byers T . Do obese children become obese adults? A review of the literature Prev Med 1993 22: 167–177.

    CAS  Article  Google Scholar 

  27. 27

    Wadden TA, Stunkard AJ . Social and psychological consequences of obesity Ann Intern Med 1985 103 (6 pt 2): 1062–1067.

    CAS  Article  Google Scholar 

  28. 28

    Widom CS, Weiler BL, Cottler LB . Childhood victimization and drug abuse: a comparison of prospective and retrospective findings J Consult Clin Psychol 1999 67: 867–880.

    CAS  Article  Google Scholar 

  29. 29

    Kuczmarski RJ, Flegal KM, Campbell SM, Johnson CL . Increasing prevalence of overweight among US adults: the National Health and Nutrition Examination Surveys, 1960 to 1991 JAMA 1994 272: 205–211.

    CAS  Article  Google Scholar 

  30. 30

    Finkelhor D, Hotaling G, Lewis IA, Smith C . Sexual abuse in a national survey of adult men and women Child Abuse Negl 1990 14: 19–28.

    CAS  Article  Google Scholar 

  31. 31

    MacMillan HL, Fleming JE, Trocme N, Boyle MH, Wong M, Racine YA, Beardslee WR, Offord DR . Prevalence of child physical and sexual abuse in the community: results from the Ontario Health Supplement JAMA 1997 278: 131–135.

    CAS  Article  Google Scholar 

  32. 32

    Ney PG, Fung T, Wickett AR . The worst combinations of child abuse and neglect Child Abuse Neglect 1994 18: 705–714.

    CAS  Article  Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to DF Williamson.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Williamson, D., Thompson, T., Anda, R. et al. Body weight and obesity in adults and self-reported abuse in childhood. Int J Obes 26, 1075–1082 (2002). https://doi.org/10.1038/sj.ijo.0802038

Download citation

Keywords

  • adult obesity
  • attributable fraction
  • body mass index
  • child abuse
  • relative risk

Further reading

Search