Acknowledgements: The authors wish to acknowledge Mary Johnson, Dongying Zhong, Ashley McCorkle, Cady Blackey, and Sara Madison Davenport for their work on the KWHR project.

Introduction

The first study to link obesity with childhood sexual abuse (CSA) was reported by an observant clinician who noted that a high proportion of women enrolled in a weight loss program had a history of CSA (Felitti, 1991). Three (Felitti, 1993; Springs & Friedrich, 1992; Wadden et al., 2006) of four clinic-based studies (Felitti, 1993; Jia, Li, Leserman, Hu, & Drossman, 2004; Springs & Friedrich, 1992; Wadden et al., 2006) also noted an increased prevalence of CSA among obese relative to non-obese female patients. Several population-based, cross-sectional studies have evaluated this association (Aaron & Hughes, 2007; Alvarez, Pavao, Baumrind, & Kimerling, 2007; Brewerton, O'Neil, Dansky, & Kilpatrick, 1999; Cloutier, Martin, & Poole, 2002; Stein & Barrett Connor, 2000), and only one (Stein & Barrett Connor, 2000) of the five studies found no association. Perhaps the most compelling data linking CSA with adult obesity comes from the Adverse Childhood Experiences (ACE) study which addressed the health effects of a range of experiences in childhood in a large retrospective cohort of 13,177 California health maintenance organization members (Felitti et al., 1998). In subsequent analysis of these data which adjusted for psychosocial factors (Williamson, Thompson, Anda, Dietz, & Felitti, 2002), sexual assault in childhood that involved penetration was associated with a 30% increase in the incidence of obesity (95% confidence interval [CI] = 1.2-1.5). Two recent cohort studies provide additional prospective evidence that CSA may be a causal factor in obesity for young adults (Mamun et al., 2007; Noll, Zeller, Trickett, & Putnam, 2007).

Several studies have noted that other forms of abuse were associated with adult obesity. In a small clinic-based study of 239 gastroenterology patients (Jia et al., 2004), physical abuse experienced as a child or as an adult was associated with being overweight or obese (odds ratio [OR] = 1.3; p<0.03). Brewerton (1999) reported that sexual abuse experienced as a child or adult was associated with lifetime morbid obesity. Finally, using data from the ACE study, Felitti et al. (1998) noted that other types of abuse, in addition to CSA, were associated with adult obesity, and as the number of types of abuse experienced increased, the rates of adult obesity likewise increased. In aggregate, these findings suggest that other forms of abuse, including sexual or physical abuse experienced as a child or adult, may be associated with adult obesity. One small study of 104 pregnant women (Bailey & Daughtery, 2007) found that psychological IPV during pregnancy was associated with pre-pregnant body mass index (BMI) as obese (p<0.05). In this study, physical IPV during pregnancy was not associated with an increased risk of obesity. In a large cross-sectional analysis of 2005 BRFSS data, Black & Breiding (2008) reported that lifetime IPV, defined as lifetime threatened, attempted or complete physical or sexual assault by an intimate partner, was associated with 10% increased risk of currently being obese in women yet not in men.

Why might abuse experienced as a child or adult influence obesity? A recent review (Gustafson & Sarwer, 2004) provides evidence for mechanisms. Both anxiety and depression are well documented consequences of gender-based violence including both CSA and IPV (Campbell, 2002; Coker, Weston, Creson, Justice, & Blakeney; Golding, 1999a, 1999b; Nixon, Resick, & Nishith, 2004). Recent cohort studies have noted an increased risk of greater BMI associated with postpartum depression (Herring et al., 2008) and depressive symptoms before age 17 years (Hasler et al., 2005). Obesity may also increase the risk of depression and anxiety. In a large population of female veterans (n=1259), an 80% increase in obesity was observed among women with current symptoms of PTSD (Dobie et al., 2004). In the large HUNTS cohort study, higher BMI at baseline (1984-1986) was positively associated with higher depression scores measured 10 years later (Bjerkeset, Romundstad, Evans, & Gunnell, 2008). A similar pattern was noted for a cohort of females adolescents in which obesity was associated with an increased risk of both major depressive disorders and anxiety disorders (Anderson, Cohen, Naumova, Jacques, & Must, 2007). Finally, the combination of high levels of depressive symptoms and experiencing very stressful life events, such as CSA or IPV, were found to significantly increase the risk of developing a metabolic syndrome which included obesity (Räikkönen, Matthews, & Kuller, 2007).

Based on the existing literature which supports a strong association between CSA and obesity and a more limited literature supporting physical abuse and obesity, we hypothesized that currently obese women would be more likely to have experienced IPV. Further, we hypothesized that the rate ratios would be larger for the association between obesity and sexual IPV than for obesity and physical IPV. Finally, based on the association between obesity and depression / anxiety, we hypothesized that the combined effects of abuse and depression / anxiety would be associated with greater rates of current obesity.

Methods

We conducted a cross-sectional analysis of data available from the Kentucky Women's Health Registry (KWHR http://www.mc.uky.edu/kyhealthregistry/). The registry, begun in 2006, is a comprehensive survey open to all women between the ages of 18 and 89 living in Kentucky. The purpose of the registry is to better care for Kentucky women by understanding how risk and protective behavioral factors may differentially affect women's health and to give interested women the chance to participate in medical research. Women can complete the survey online or by using a paper copy which can be mailed in to study staff. The KWHR study was approved by the University of Kentucky Institutional Review Board. Consent for participation in the KWHR and HIPAA waiver were obtained either electronically, if the survey was completed online, or on paper.

As of May 2008, 4915 women, 18 years of age or older, had been recruited and completed their first survey. Reminders were sent to women encouraging them to retake the survey annually with a rate of return from the first to the second year at 63%. Registry participants also received health information updates in the form of newsletters and access to clinical trials of new therapies for conditions that affect them. In this analysis, we focused on women completing their baseline (first) survey for the 2006 and 2007 survey years (as of March 14, 2008). A total of 179 women were excluded due to missing data on height and weight (to define current BMI) and 36 for missing data for more than two types of abuse (childhood sexual or physical abuse, physical IPV or other forced sex), and 310 were missing on confounders included in modeling. The final number for these analyses was 4391.

Three items were used to measure current (past 12 months) and lifetime physical or sexual abuse by an intimate partner (IPV). Physical IPV was addressed with this question: "Has an intimate partner hit, kicked, punched, or otherwise hurt you?" Sexual IPV was assessed with the following item "Has an intimate partner used force (like hitting, holding down, or using a weapon) to make you have sex?" Women were asked to select only one answer from the following response options: yes in the past 12 months, yes in her lifetime, or no. Two dichotomous variables were created for each of the two IPV types. We created two sets of indicator variables for current IPV and lifetime IPV for physical and sexual IPV (four total).

The KWHR included questions to characterize child sexual and physical abuse. The childhood forced sex item read as follows: "When you were a child, did any parent, stepparent, or guardian or any other person make you have sex (any sex act, not just intercourse) by using force or by threatening to harm you or someone close to you?" The childhood physical abuse was assessed with the following question "When you were a child, did any parent, stepparent, or guardian ever hit, kick, punch or otherwise hurt you?" Responses to both items were categorized as ever versus never experienced childhood sexual or physical abuse. Similarly, the next question was used to create a dichotomous indicator variable for non-partner forced sex: "Has anyone other than an intimate partner or anyone else used force (like hitting, holding down, or using a weapon) to make you have sex?"

Because women may experience sexual abuse either as a child, adolescent, or adult, and multiple sexually abusive experiences may increase the risk of current obesity, we created a cumulative sexual abuse score. This score is simply the count of whether women experienced sexual IPV, forced sex by someone other than a partner, and/or childhood sexual abuse. The range for this score is 0-3. We selected sexual abuse because (1) CSA is associated with adult obesity, and (2) women experiencing CSA may be more likely to also experience sexual IPV or forced sex by someone else. Using this cumulative sexual abuse measure, we can evaluate whether co-occurrence of childhood and adult sexual abuse further increase rates of obesity.

Current obesity was defined as a BMI of 30 or more. Subjects were asked their current height and weight and, with these data, current body mass index (BMI) was calculated as weight in kg / height in m².

Finally we hypothesized that abuse may influence obesity through depressive or anxiety symptoms. We therefore used questions there were included in the KWHR to measure lifetime symptoms of PTSD and depression. To reduce respondent burden, we reduced the number of items used to measure these constructs. We constructed a short depression scale based on the Zung Depression Scale (Zung, 1965) and psychometric analyses reported by Passik (Passik et al., 2000). The following four items were used to define having depressive symptoms: "Has there been a period of at least two straight weeks when you have felt down, depressed, or hopeless?", "Has there been a period of at least two straight weeks when you have felt little interest or pleasure in doing things?", "Has there been a period of at least two weeks straight when you had trouble concentrating on things?", and "Have you had thoughts that you would be better off dead or of hurting yourself in some way?" The mutually exclusive response options were yes- in the past 12 months, yes-in my lifetime, and no. We created measures of current and lifetime depressive symptoms by summing the four items for lifetime symptoms (yes in the past 12 months or yes in my lifetime to any of the four questions) and for current symptoms (Yes in the past 12 months for any of the four questions). The four items had good internal consistency as indicated by the Cronbach's alpha of 0.82 for both lifetime and current depressive symptoms. We used 4 items from the 7-item, Short Screening Scale for DSM-IV Posttraumatic Stress Disorder (Breslau, Peterson, Kessler, & Schultz, 1999). A score was created by summing the yes responses to the questions described below. Women were first asked whether they had ever been exposed to a traumatic event in which they experienced or witnessed actual or threatened death or serious injury that involved intense fear or horror for that experience. Those who answered yes were asked the following questions. "After this experience did you have more trouble than usual falling asleep or staying asleep?", "Did you begin to feel more isolated or distant from other people?", "Did you avoid being reminded of this experience by staying away from certain places, people or activities?", and "Did you begin to feel there was no point in planning for the future?" Cronbach's alpha indicating scale internal consistency was 0.86 for PTSD symptoms.

To address potential confounders / effect modifiers, the KWHR included the following sets of questions relevant to this analysis. Demographic factors included were current age (continuous variable), current marital status, race /ethnicity (dichotomous), education (ordinal variable), and current health insurance coverage (categorical variable). Data to characterize smoking status (dichotomous variable), duration of smoking (age at first and last smoking), and pack years of smoking based on years of smoking and average number of packs of cigarette usually smoked per day (continuous variables) were analyzed.

Statistical Analysis

All analyses were completed using SAS 9.1. Stratified analysis was used to describe the demographic variables associated with current obesity. The purpose of this analysis was to determine potential confounders for subsequent multivariate analyses. Given the large sample size, only differences in IPV or obesity of p<0.0001 will be noted as meaningfully different. Chi square and associated p values are presented. Proc Genmod (log link function and poisson distribution) were used to calculate adjusted prevalence rate ratios for abuse by timing (past 12 months, childhood or as an adult) and type (e.g. physical or sexual), and obesity adjusting for sociodemographic factors. Lastly, to explore whether the number of types of abuse experienced influenced current obesity rates differently by lifetime symptoms of PTSD or depression, we conducted stratified multivariate regression in which the cumulative sexual abuse score was included in models as the independent variable, current obesity was the dependent variable, and PTSD / depressive symptoms was the stratifying variable.

Results

Women who were currently obese were less likely than non-obese women to be college educated, to have private health insurance, were more likely to have greater smoking pack years, more symptoms of depression or anxiety (PTSD), and to currently perceive more stress (table 1).

Women ever experiencing IPV were less likely than non-abused women to have a college education, to be currently married, and to have private health insurance. Abused women had greater smoking pack years, more symptoms of depression or anxiety (PTSD), and were more likely to currently perceive more stress (table 1).

We used the definition of a confounder as a factor associated with both the exposure (here IPV) and outcome (here current obesity). Further, the factor could not be in the potentially causal pathway (Rothman & Greenland, 1998) such as depressive or PTSD symptoms or current stress. We selected the following as confounders and included these in subsequent multivariate analyses: education, private health insurance, and pack years of smoking.

Current IPV was not associated with current obesity (table 2). Lifetime sexual IPV (adjusted rate ratio [aRR] = 1.25; 95% CI = 1.06-1.47) was associated with an increase in prevalence of current obesity. CSA was associated with a 38% increase in prevalence of current obesity while child physical abuse was associated with a 26% increase, after adjusting for potential confounders. Because sexual abuse appears to be the more important factor for obesity, we calculated obesity rates for those experiencing sexual abuse independent of timing or relationship with the perpetrator. Women who have experienced sexual abuse as an adult or child were 32% more likely to be currently obese. Physical abuse without sexual abuse, independent of timing of the physical abuse, was not associated with current obesity. Compared with those never experiencing sexual abuse, those experiencing more than one type of sexual abuse in a woman's lifetime appeared to have increased rates of current obesity from 23% for one form of sexual abuse (aRR = 1.23; 95% CI = 1.06-1.42) to 49% for all three types of sexual abuse (aRR = 1.49; 95% CI = 1.02-2.16; not presented in table 2).

Finally, we explored whether sexual abuse might affect current obesity differently depending on whether the woman had symptoms of anxiety or depression (table 3). Current obesity rates were highest among women reporting two or more types of sexual abuse (p for trend <0.0001 among all women). This pattern held for those with no lifetime PTSD or depressive symptoms (p for trend = 0.006) as well as for those with lifetime symptoms of PTSD or depression (p for trend = 0.02) and depressive symptoms (p=0.02). This pattern was not statistically significant among the 896 women with PTSD symptoms (p=0.22). The Breslow-Day test for homogeneity was not significant indicating no interaction between symptoms and sexual abuse and rates of current obesity. (p=0.43).

Discussion

This analysis adds to the literature by finding that in addition to CSA, sexual IPV was associated with current obesity. Strengths of this analysis include the large sample size and the more "population-based" study design. Our finding of a 32% increased risk of current obesity associated with lifetime sexual abuse was very consistent with data recently reported from the California Women's Health Survey of 1115 adult women which also found a 30% increase in lifetime obesity with childhood physical or sexual abuse (95% CI = 1.1-1.4) (Alvarez et al., 2007) and that reported by Black and Breiding (2008). Based on these data, it appears that sexual abuse differentially increases risk of current obesity.

There are limitations to this analysis. The abuse questions included in the KWHR do not characterize the frequency, severity, or duration of the abuse experienced. Had this data been available, we could have investigated a dose response association between these attributes of IPV experienced and rates of current obesity. Because we have no data on the timing of IPV or forced sex by someone other than a partner, we cannot establish the correct temporal sequence from the abuse experienced to current obesity. However, we are measuring current obesity and few women (3%) reported current IPV, thus we have more confidence that IPV preceded current obesity. Furthermore, we know that childhood abuse precedes current obesity. We do not have data to characterize the patterns by which adverse childhood experiences cluster with each other or with adult abuse experiences and lack the ability to adjust for these patterns in investigating abuse and obesity.

Implications

Obesity is a significant health threat to women in the USA and worldwide. In this Kentucky-based sample, almost one-third of women were currently obese. Lifetime sexual abuse experienced as an adult or child was associated with a 32% increase in current obesity. As evidenced in this report and substantiated by others, both IPV and obesity are common health threats for women. Continuing efforts to prevent child and partner physical and sexual abuse and to prevent the mental health consequences of abuse on victims can have important implications for improving women's health.

References

Aaron, D. J., & Hughes, T. L. (2007). Association of childhood sexual abuse with obesity in a community sample of lesbians. Obesity, 15, 1023-1028.

Alvarez, J., Pavao, J., Baumrind, N., & Kimerling, R. (2007). The relationship between child abuse and adult obesity among California women. American Journal of Preventive Medicine, 33, 28-33.

Anderson, S. E., Cohen, P., Naumova, E. N., Jacques, P. F., & Must, A. (2007). Adolescent obesity and risk for subsequent major depressive disorder and anxiety disorder: prospective evidence. Psychosomatic Medicine, 69, 740-747.

Bjerkeset, O., Romundstad, P., Evans, J., & Gunnell, D. (2008). Association of adult body mass index and height with anxiety, depression, and suicide in the general population: the HUNT study. American Journal of Epidemiology,167, 193-202.

Black, M.C., Breiding M.J. (2008). Adverse health conditions and health risk behaviors associated with intimate partner violence – United States, 2005. Morbidity and Mortality Weekly, 57,113-117.

Breslau, N., Peterson, E. L., Kessler, R. C., & Schultz, L. R. (1999). Short screening scale for DSM-IV posttraumatic stress disorder. American Journal of Psychiatry, 156, 908-911.

Brewerton, T. D., O'Neil, P. M., Dansky, B. S., & Kilpatrick, D. K. (1999). Links between morbid obesity, victimization, PTSD, major depression & bulimia in a national sample of women (Abstract). Obesity Research, 7, 56S.

Campbell, J. C. (2002). Health consequences of intimate partner violence. Lancet, 359(9314), 1331-1336.

Cloutier, S., Martin, S. L., & Poole, C. (2002). Sexual assault among North Carolina women: prevalence and health risk factors. Journal of Epidemiology and Community Health, 56, 265-71.

Coker, A. L., Weston, R., Creson, D. L., Justice, B., & Blakeney, P. PTSD symptoms among men and women survivors of intimate partner violence: the role of risk and protective factors. Violence & Victims, 20, 625-643.

Dobie, D. J., Kivlahan, D. R., Maynard, C., Bush, K. R., Davis, T. M., & Bradley, K. A. (2004). Posttraumatic stress disorder in female veterans. Associations with self-reported health problems and functional impairment. Archives of Internal Medicine, 164, 394-4000.

Felitti, V. J. (1991). Long-term medical consequences of incest, rape, and molestation. Southern Medical Journal, 84, 328-331.

Felitti, V. J. (1993). Childhood sexual abuse, depression, and family dysfunction in adult obese patients: a case control study. Southern Medical Journal, 86, 732-736.

Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., et al. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. American Journal of Preventive Medicine,141, 245-258.

Golding, J. M. (1999a). Intimate partner violence as a risk factor for mental disorders: A meta-analysis. Journal of Family Violence, 14, 99-132.

Golding, J. M. (1999b). Sexual assault history and long-term physical health problems: evidence from clinic and population epidemiology. Current Directions in Psychological Science, 8, 191-194.

Gustafson, T. B., & Sarwer, D. B. (2004). Childhood sexual abuse and obesity. Obesity Reviews, 5, 129-135.

Hasler, G., Pine, D. S., Kleinbaum, D. G., Gamma, A., Luckenbaugh, D., Ajdacic, V., et al. (2005). Depressive symptoms during childhood and adult obesity: the Zurich Cohort Study. Molecular Psychiatry, 10, 842-850.

Herring, S. J., Rich-Edwards, J. W., Oken, E., Rifas-Shiman, S. L., Kleinman, K. P., & Gillman, M. W. (2008). Association of postpartum depression with weight retention 1 year after childbirth. Obesity, 16, 1296-1301.

Jia, H., Li, J. Z., Leserman, J., Hu, Y., & Drossman, D. A. (2004). Relationship of abuse history and other risk factors with obesity among female gastrointestinal patients. Digestive Diseases and Sciences 49, 872–877.

Mamun, A. A., Lawlor, D. A., O'Callaghan, M. J., Bor, W., Williams, G. M., & Najman, J. M. (2007). Does childhood sexual abuse predict young adult's BMI? A birth cohort study. Obesity, 15, 2103-2110.

Nixon, R. D., Resick, P. A., & Nishith, P. (2004). An exploration of comorbid depression among female victims of intimate partner violence with posttraumatic stress disorder. Journal of Affective Disorders, 82, 315-320.

Noll, J. G., Zeller, M. H., Trickett, P. K., & Putnam, F. W. (2007). Obesity risk for female victims of childhood sexual abuse: a prospective study. Pediatrics,120, e61-67.

Passik, S. D., Lundberg, J. C., Rosenfeld, B., Kirsh, K. L., Donaghy, K., Theobald, D., et al. (2000). Factor Analysis of the Zung Self-rating depression scale in a large ambulatory oncology sample. Psychosomatics, 41,121–127.

Räikkönen, K., Matthews, K. A., & Kuller, L. H. (2007). Depressive symptoms and stressful life events predict metabolic syndrome among middle-aged women: a comparison of World Health Organization, Adult Treatment Panel III, and International Diabetes Foundation definitions. Diabetes Care, 30, 872-877.

Rothman, K. J., & Greenland, S. (1998). Modern Epidemiology (2nd ed.). Philadelphia, PA: Lippincott-Raven.

Springs, F. E., & Friedrich, W. N. (1992). Health risk beahviors and medical sequelae of childhood sexual abuse. Mayo Clinic Proceedings, 67, 527-532.

Stein, M. B. & Barrett-Connor, E. (2000). Sexual assault and physical health: findings from a population-based study of older adults. Psychosomatics Medicine, 62, 838-843.

Wadden, T. A., Butryn, M. L., Sarwer, D. B., Fabricatore, A. N., Crerand, C. E., Lipschutz, P. E., et al. (2006). Comparison of psychosocial status in treatment- seeking women with class III vs. class I-II obesity. Surgery for Obesity and Related Diseases, 2, 138-145.

Williamson, D. F., Thompson, T. J., Anda, R. F., Dietz, W. H., & Felitti, V. (2002). Body weight and obesity in adults and self-reported abuse in childhood. International Journal of Obesity and Related Metabolic Disorders, 26,1075-1082.

Zung, W. W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70.