VuLogoPsychology Department

Health Psychology Home Page

Papers written by students providing scientific reviews of topics related to health and well being

  HomeWeight LossAlternative Therapy | Supplements | Eating Disorders | Fitness | About this Page |



What is the correlation between sexual abuse and the development of bulimia nervosa?

Chloe Booth

May 4, 2005





Recently, the question of whether there is a direct link between sexual abuse and the onset of bulimia nervosa later in life has received greater amounts of attention in the psychiatric field. Studies conducted to determine whether there is a link between sexual abuse and eating disorders [in general] and the common consensus was that bulimics had the most prevalence of childhood sexual abuse in comparison to participants suffering from other eating disorders. Although this seems to be a trend, it is known that sexual abuse does routinely cause people to develop and suffer from psychiatric disorders later in life, bulimia nervosa being included in this group of disorders. This has prompted several studies to be conducted to determine the correlation, if any, between sexual abuse and onset of bulimia nervosa. Most of the studies that have published reports of their findings, unfortunately, have yielded miserly answers to the question. It is still unknown whether there exists a correlation between sexual abuse and bulimia nervosa and whether this correlation is strictly limited to bulimia or whether it applies to other eating disorders as well.


Psychiatric Disorders That May Be Caused by Sexual Abuse:

         A study conducted at the University of Oxford Department of Psychiatry, headed by Sarah L. Welch and Christopher G. Fairburn, identified three goals for their study to determine if there is a correlation between sexual abuse and development of bulimia nervosa. The three goals were 1) “to determine whether sexual abuse increases the risk of developing bulimia nervosa, 2) to see whether any increase in risk identified was specific to bulimia nervosa or whether it was likely to reflect an increased risk of psychiatric disorder in general and 3) to see whether a clinic group of patients with bulimia nervosa differed from a community group in terms of their histories of sexual abuse, thus indicating whether a bias operates in studies based on clinically detected groups.” (Welch & Fairburn, 1994)A match case control design was used with three case related control comparisons, each complementary to the specific aim being addressed. For goal 1, the comparison was done between fifty subjects with bulimia nervosa (community subjects with bulimia nervosa) and one hundred comparison subjects without an eating or psychiatric disorder (normal comparison subjects). For goal 2, the comparison was done between the same fifty subjects with bulimia nervosa from goal 1 and fifty comparison subjects suffering from some other form of a psychiatric disorder (general psychiatric) comparison subjects. Finally, for goal 3 the comparison was done between the fifty community subjects suffering form bulimia nervosa, used in aims 1 and 2, and fifty patients with bulimia nervosa (clinic cases of bulimia nervosa). Having three aims, which corresponded to observing 3 different types of subjects is a very strong strength of this study. By observing three different types of subjects, Welch and Fairburn were able to answer three different questions that plague the umbrella question of whether there is a correlation between childhood sexual abuse and later onset of bulimia nervosa. By answering other questions that fall under the umbrella question, one can begin to eliminate certain factors that are thought to have effects on the condition being examined as well as lead to discoveries that may allow the researchers to start down the path to finding the answer to the big question.

         Welch and Fairburn were careful to properly identify their definition of sexual abuse, as pertained to their study, properly identify their definition of bulimia nervosa, and attempted to eliminate as many outlying factors when matching comparison groups. The researchers defined sexual abuse as “any sexual experience involving physical contact that was against the subject’s will; it therefore included the subject being touched or being made to touch the abuser in any sexual way, including oral sex and rape.” (Welch & Fairburn, 1994) Community subjects and clinic patients with bulimia nervosa had to meet the DSM-III-R operational definition of bulimia nervosa, the normal subjects were required to have no current or previous eating disorder, as determined through an interview and those subjects suffering from a general psychiatric disorder had to meet axis I DSM-III-R definition of disorder.  The two groups of community comparisons (those suffering from no psychiatric disorder, “normal” and those suffering from psychiatric disorders) were individually matched to those community subjects with bulimia nervosa with regards to age and parental socio-economic class (subjects were matched within 1 year of age difference and then by three sub-sets of parental socio-economic class). Matching subjects by age helped ensure that recall time from the subjects be about the same and by “matching subjects by parental socio-economic class removed an important confounding variable.” The community subjects with bulimia nervosa were matched to the clinic patients with bulimia nervosa solely based on the 1 year age difference criterion. All these steps are very important to eliminating outside factors that could potentially harm the data that the researchers are trying to incur. These precautions are strengths of this study.

         By observing the three sub-set groups, three different questions were answered by the research. The data that compared the community subjects with bulimia nervosa to the normal comparison subjects showed (goal1), with respects to all types of sexual abuse, that a history of sexual abuse including physical contact was more prevalent among the community subjects with bulimia nervosa than the normal comparison group. However, the difference with respect to sexual abuse before the age of twelve was not of statistical significance. The data that compares the community subjects with bulimia nervosa and the comparison subjects with general psychiatric disorders (goal 2) showed that there is no significant statistical difference, with respect to history of sexual abuse, indicating that sexual abuse is just as likely to cause bulimia nervosa to develop in someone as it is to cause any other array of general psychiatric disorders. Finally, the data that compared the community subjects with bulimia nervosa and the clinic patients with bulimia nervosa showed that there was no significant statistical bias, with respect to a history of sexual abuse.

         Researchers in the same field, H.G.Pope and J.I.Hudson, identified five defects with the Welch and Fairburn study. They described the five problems as: “1) lack of control groups or use of inadequate control groups (such as “supernormal” control subjects with no personal or family history of psychiatric disorder or control groups unmatched by sex), 2) unsatisfactory definitions of sexual abuse, 3) poor evaluation of sexual abuse (sometimes by questionnaire alone, sometimes by using different methods for the study subjects and control subjects), 4) lack of attention to the timing of the abuse in relation to the onset of the eating disorder, and 5) absence of comparisons with control groups with other psychiatric disorders in order to examine specificity.” (Welch & Fairburn, 1994) According to Pope and Hudson, each of these defects could yield misleading figures when analyzing data from the study.

         Welch and Fairburn state in the conclusions of their research report that, although more research has to be done, at the time they feel comfortable stating they find there to be no correlation between childhood sexual abuse and later onset of bulimia nervosa.


Relationship between Childhood Sexual Abuse and Development of Eating Disorders:

Karen van Gerko, Mari Laura Hughes, Michelle Hamill, Glenn Waller collaborated on a study at St. George’s Mental Health NHS Trust in London, England to determine whether there was a link between reported cases of childhood sexual abuse (CSA) and eating-disordered behaviors. These researchers decided to examine the relationship between CSA and disordered eating behaviors as opposed to eating disorders themselves because, although many links have been proposed between CSA and the onset of eating disorders later in life, no previous studies had been able to confirm this. They decided to take one step back in the hope that the answer to this question would be able to help them and fellow researchers in the future answer the larger question regarding the correlation CSA and eating’s disorders. This in and of itself is a strength because they are not jumping to conclusions that people who exemplify these disordered eating behaviors have an eating disorder but are looking to examine and hopefully explain why these people exhibit these behaviors.

         The study was done in a cross-sectional format, with a population of eating disordered women. The 299 women reported past experiences of CSA and current conditions of disordered eating. The subjects were then interviewed to make sure they met DSM-IV criteria for eating disorders. Of all the subjects, sixty-two had anorexia nervosa, restrictive subtype, sixty-one had anorexia nervosa, binge/purge subtype, ninety-two had bulimia nervosa, twenty-seven had EDNOS-binge eating disorder, twenty-nine had EDNOS a-typical anorexia nervosa, ten had EDNOS a-typical bulimia nervosa and 18 had other forms of EDNOS. To determine whether the women had experiences CSA, subjects were asked a set of questions that would allude to anything from an abusive past. In this study, childhood sexual abuse was defined as “any sexual experience with an adult (not necessarily a parent or relative) that occurred before the age of 14. Included in this definition were: experiences of witnessing the sexual abuse of another family member; traumatic sexual experiences; and being fearful of sexual mistreatment when parents were under the influence of alcohol.” (van Gerko, Hughes, Hammill, & Waller, 2005) All of the women completed the Eating Disorders Examination-Questionnaire, a standardized set of questions, the responses to which measure the women’s eating attitudes and behaviors. This research approach is a good one because it is uniform and impersonal, i.e., all women are being asked the exact same question and they only have to “answer to” a sheet of paper. This is in contrast to an approach where an individual therapist is asking the questions. In that’s case, the questions may not be uniform and the subjects may not be willing to provide the necessary information if they are uncomfortable with the therapist to whom they are speaking. The questionnaire is also very thorough in that it measures for restraint, eating concern, body shape concern and body weight concern. It also addresses key behaviors of eating disorders, e.g. laxative abuse, excessive exercise, and vomiting for weight control…

         In order to test their first hypothesis, van Gerko et al. used a chi-squared test, which is a reliable measure. To test their second hypothesis, they used a MANCOVA test, which compares the EDE-Q scores of women who did or did not report CSA (as age and BMI varied across groups, these two variables were used a covariates—which is another good precaution that was taken into account to ensure valid data).

         In total, 28.8% of the women reported having a past of CSA. This response was low among restrictive anorexics and binge eating disorder groups and fairly higher among the atypical anorexics, however the chi-squared test did not produce a result that could be considered statistically significant (x2=11.8,P=0.066). However to test hypothesis one, the correlation between women who report cases of CSA and those who do/don’t engage in some type of purging activity is what needed to be measured. In this instance, only17.7% of women who did not engage in purging behaviors reported cases of CSA while 33.8% of women who did engage in purging behaviors reported cases of CSA. Here, the chi-squared test did show statistical significance. The scores form the EDE-Q tests also indicated that women who reported having a past of CSA had higher levels of objective bingeing, vomiting, laxative abuse and diuretic abuse (when age and BMI were not used as covariates, the number did not change significantly, meaning they do not influence group differences for eating pathology).

         The outcome of this research clearly indicates that there is a direct correlation between a past of CSA and bulimic symptoms. The correlation is only shown to be true for eating behavior’s that are non-restrictive and non-purging i.e. those who purge had more incidences of CSA. The research does not provide for a direct causal link between CSA and eating disorders, it suggests that CSA can steer one’s eating pathology toward bulimic tendencies.

         One weakness of this study was that the subjects were adults. Similar studies that would have been carried out with adolescents may yield more indicative and porbable results since eating disorders are much more typical of adolescents than adults. Testing adolescents would also allow researchers to observe and document any sudden or new changes in the subject’s eating behavior and weight, things much less commonly characteristic of adults. Unlike other studies mentioned in this paper, this study does not take into account the nature of the abuse and how the nature of the abuse can dictate the type of eating behavior that becomes disordered.

            I believe this study definitely deserves credit because it did not attempt to answer the big picture question in one go but rather tried to answer one part of the puzzle that would be beneficial to them and other researchers who are working to find the answer to the big picture question.


Determining whether Sexual Abuse was a factor in the development of AN/BN:

            Glenn Waller, of the Psychiatry Department at the University of Manchester, conducted a study that examines the links between childhood or adulthood sexual abuse and subsequent development of bulimia nervosa or anorexia nervosa. His two goals for his research: 1) to test his hypothesis that CSA would be less prevalent among people with non-purging behaviors and 2) “to determine the dimensional association of reported sexual abuses with a broad range of eating characteristics.” (Waller, 2001)

         The subjects consisted of sixty-seven women, all of whom met the DSM-III-R requirement for anorexia nervosa or bulimia nervosa, it being very important to set a baseline that all subjects can be measured from. The subjects were also undergoing individual cognitive-behavior treatment. These women were then divided into four groups: women with restrictive anorexia nervosa, women with bulimic subtype of anorexia nervosa, women with bulimia nervosa who had a history of anorexia nervosa and women with bulimia nervosa who had no history of anorexia nervosa. Waller states that it was important to examine these four different groups because previous research has indicated that the different disorders reflect important psychological differences. This is a strength of Waller’s study in that he takes into account the differences in the eating disorders and realizes that each of them are characterized by different and distinct eating histories as well as distinct eating behaviors. By examining more than one type of eating disorder, one can see whether the “control” (history of sexual abuse) contributes to onset bulimia nervosa, or other eating’s disorders as well. To determine prevalence of sexual abuse, all women were interviewed by a therapist, and one-third of them were also asked to complete a questionnaire about their history of sexual abuse. By having all women subject to an interview eliminates the outlying factor that is information obtained in different ways may yield different results; this ensured that all women were given the exact same opportunity to answer the questions, which gives rise to a higher statistical power. Having one-third of the women, randomly chosen, answer questions about their sexual abuse history in the form of a questionnaire allowed Waller to determine whether there is a difference in the information obtained through interview as compared with the information obtained through questionnaire. “A x2 analysis (x2=0.001, d.f. = 1, P=0.98) showed that there was no association between reported occurrence of sexual abuse and the method of examination.” (Waller, 2001)

         Thirty-two women (48%) reported having been sexually abused in their past. In examining the data produced by the interviews, a clear pattern emerged; fifteen of the twenty women who had bulimia nervosa with a history of anorexia reported abuse while only one of the fifteen women who had anorexia nervosa, restriction subtype, reported abuse. Women with anorexia nervosa, bulimic subtype and women with bulimia nervosa and no history of anorexia were evenly divided, with about half of the women reporting some type of unwanted sexual experience. “A x2 analysis (x2= 16.29, d.f. = 3, P<0.001) confirmed that there was a strong, significant association between reported sexual abuse and diagnosis. Considering the two bulimic groups only, there was a weak association between reported abuse and diagnosis, which failed to achieve significance

(x2=2.08, d.f.=1, P<0.15).” (Waller, 2001)

         A strength of Waller’s investigation is that he took into more consideration the nature of the sexual abuse, when it occurred and other details that may play a significant role in determining the eating disorder that occurs. He found that most of the sexual abuse frequently involved someone the victim knew, either a relative or close friend. He suggests that this may be an important factor in the onset of bulimia nervosa as many of the women who reported having been sexually abused by someone close to them did not disclose it at the time of the abuse because they were afraid that they would be held responsible for the demise of the family or because they thought that it was actually their own fault. This may be an important factor as it appears to indicate that women who are abused by someone they know are less likely to disclose it at the time, hence they are the ones more likely to develop an eating disorder later in life as opposed to a woman who was abused by someone unknown to them and hence felt more comfortable reporting it and didn’t have to live with the as much of the heavy burden of having had this experience as the other women. Waller found that age at the time of abuse was not a critical factor.

         A weakness of Waller’s study is that it did not include a comparison group of non-clinical patients; he only used subjects who had an eating disorder and then looked at their sexual abuse past. In order to correct this, he should have added a non-clinical group, asked them the same questions about their sexual abuse past, in the same format, and then compare results from the clinical and non-clinical groups. This would allow him to see whether there is a correlation between having been sexually abused and the development of an eating disorder as opposed to examining people with eating disorders and identifying which of them had been sexually abused. Another weakness of Waller’s study is that he used a very small number of subjects in his research. This approach is problematic because it did not allow him to draw meaningful analysis of factors such as age of victim at time of abuse, nature of abuse and identity of abuser, all which have the potential to affect any type of psychiatric disorder, including eating disorders. Small subject groups are rarely representative of the entire population of people that they are trying to represent and they do not yield a high enough statistical power to allow one to claim valid results. Further research would need to be done to address the links between different types of abuse and age at time of abuse and the nature of the disorder that eventually develops.

          One conclusion that Waller comes to is “unwanted sexual experience does not predispose to eating disorders per Se, but that a history of sexual abuse determines the nature of any eating disorder that develops due to other actors Women who develop restricting anorexia initially for a complex set of reasons may be influenced to use bingeing and vomiting (and thus to progress to anorexia of the bulimic subtype (AN -F) or bulimia nervosa (BN +)) if they have a history of abuse. Other women may develop bulimia directly (BN-), as an impulsive' means of coping with the emotional difficulties that arise from the abuse and associated factors.” (Waller, 2001)

         “The clinical implication of these findings is clear. Bulimic women are likely to have a history of sexual abuse that is relevant to the disorder they have developed. Within a cognitive framework, the abuse is related to feelings of guilt and low self-esteem.” (Waller, 2001)


Childhood Sexual Abuse and onset of AN/BN in Australian Women:

            Lena Sanci, MBBS, PhD, FRACGP; Carolyn Coffey, BSc, Grad Dip Epi; Craig Olsson, PhD; Sophie Reid, PhD; John B. Carlin, PhD; George Patton, MBBS, MD, FRANZCP all worked together to conduct a prospective study to determine if there was a correlation between childhood sexual abuse (CSA) and later onset of anorexia nervosa or bulimia nervosa in women. The fact that they did a prospective study is a huge strength for their findings as this allowed them to keep track of the group of people over time (in this case, a decade) and see whether an eating disorder ever developed. The ability to document on the progression or the emergence of a disorder over time is much more reliable than just having someone come in and say I have an eating disorder and I was/was not abused as a child. This study followed 999 women, from the ages of fourteen to twenty-four, in Victoria, Australia. The study focuses on the relationship between sexual abuse before the age of sixteen and the onset of anorexic/bulimic symptoms in later adolescent life. The study took place in “8 waves”—wave 1: was the first group of young women, wave 2: second group of young women added 6 months later, waves 3-6: each woman was reviewed at four 6-month intervals, wave 7: follow-up around age 20-21 wave 8: second follow-up age 24-25. Subsequent follow-ups allowed the researchers to see, over a period of time, whether an eating disorder ever developed.

         Criterion for anorexia nervosa and bulimia nervosa were those dictated by the DSM-IV-MR, which set a standard for everyone included in the subject group. Childhood sexual abuse was reported retrospectively during the 8th wave check-up, around age 24. It is strength that everyone was to report any childhood sexual abuse at the same because having people report at different times may alter some reports. I also find it to be a strength that subjects ere asked to report sexual abuse at the end of the study as opposed to the beginning because I believe that it is very possible that if people were asked to report them at the beginning, or even at the middle, some people may have thought to themselves that they should have an eating disorder because they were abused and hence induce one when they would not have ever developed one on their own. To measure childhood sexual abuse, subjects were asked the same six questions,

“Before you were 16, did any adult or older person involve you in any unwanted incidents like: (i) inviting or requesting you to do something sexual; (ii) kissing or hugging you in a sexual way”; (iii) touching or fondling your private parts; (iv) showing their sex organs to you; (v) making them touch you in a sexual way; (vi) attempting or having sexual intercourse?” The response set was “never,” “once,” and “more than once.” These items were reduced to 3 measures, all with categories of “no report,” “one report,” and “two or more reports”: (1) “sexual abuse without physical contact,” classified according to the individual’s most severe response to questions i and iv; (2) “sexual abuse with physical contact,” classified according to the individual’s most severe response to questions ii, iii, v, and vi; and (3) “any sexual abuse,” classified according to the individual’s most severe response to all abuse questions with categories.” (Sanci, Coffey, Olsson, Reid, Patton, & Carlin, 2008)The researchers also took into account dieting behavior during waves 3-6 using the Adolescent Dieting Scale as well as symptoms of anxiety and depression during the same waves using the revised Clinical Interview Schedule. The researchers finally used multiple imputation to account for participants that missed certain follow-ups or who “disappeared” from the study all together. I think all of these features that the researchers took to ensure producing valid and accurate data is impressive. This study has a huge strength in all the precautions that were taken in order to produce valid results.

         The results of the study showed that 12.1% of the women reported having one report of childhood sexual abuse and 8.2% of the women reported having two or more reports of childhood sexual abuse. Ninety-five female participants reported one episode and sixty-nine reported two or more episodes of CSA without physical contact. Ninety-six females reported one episode and seventy reported two or more episodes of CSA with physical contact. In waves 4-6, thirty-five women were identified as new cases of bulimia nervosa and thirty–two were identified as new cases of anorexia nervosa. Study also produced results that showed that women who reported having two or more episodes of CSA before the age of 16 had more than five times higher chance of developing bulimia nervosa and females reporting 2 or more episodes of CSA were more than 4 times as likely to make the transition to purging than those with no CSA. (Sanci, Coffey, Olsson, Reid, Patton, & Carlin, 2008)This staggering fact was still the result after dieting factors and depressive/anxiety factors were taken into account, adding to the reliability of the data.

         The results of this study clearly indicate that there is a correlation between CSA and the development of bulimia nervosa. The reason to believe that this statement is correct, based on this study alone, is that the ate of new cases of bulimia nervosa still remained the same even after the number was adjusted for anxiety/depressive/dieting factors as well as background and family measures. This study further supported the claims made by Wonderlich et al, that specifically examined the hypothesis that CSA does contribute to the onset of bulimia nervosa, whose findings showed that there is a definite correlation between the two. The good things about this study is that it was a population-based cohort study, it took into account the different types/gave definitions to sexual abuse and it used “proportional hazard models that allowed them to examine the influence of time-dependent covariates.” (Sanci, Coffey, Olsson, Reid, Patton, & Carlin, 2008)

         Some weaknesses, however, include: the fact that reports of sexual abuse before puberty and after puberty were taken, which means that they could not draw proper inferences about whether the age at the time of sexual abuse plays a role in the development of an eating disorder. Researchers were also unable to determine validity or accuracy of the participants’ recollections of the abuse. Although a weakness, it would take extensive background checks to determine this. I, personally think, for the most part, that you could take the recollection to be pretty accurate as most of them occurred at an age where people can actively recall what truly happened, as well as the fact that I would give people the benefit of the doubt that they would not want to exaggerate the extent of their abuse past. The final weakness of this study, as described in the report itself, was that partial syndromes (people who had signs of bulimia nervosa but it did meet all the criteria described in the DSM-IV-MR) were counted as full-cases of bulimia nervosa. (Sanci, Coffey, Olsson, Reid, Patton, & Carlin, 2008)The problem with this is that some of those people who exhibited certain symptoms of bulimia nervosa may never have developed the final symptoms that would have qualified them to be diagnosed with bulimia nervosa. It is true that some may have gone on to develop full-blow bulimia nervosa, but the actually number that truly ended up developing the disease is unknown.

         This study goes on to support the notion that CSA has a definitive link to conditions that result in emotional distress. It further supports previous studies that have produced results that indicate that CSA does cause the incidence of bulimia nervosa to increase as one ages.



         I have only examined four studies that have been conducted to address whether there is a correlation between sexual abuse and the development of bulimia nervosa later in life. After reading several other studies and then carefully examining these four studies, I have concluded that there really is no authoritative answer as to whether these two entities are correlated. Of these four studies, none of them takes into account all the factors that each of them do, which always leads to questionable data. Also, I believe it is very difficult to come out with an incontrovertible statement linking an eating disorder one hundred percent to a single other factor. There will always be researchers who believe that there are other variables that should be taken into considerations. In addition, it is very difficult for a simple, although complexly thought out, questionnaire or interview with a trained therapist to provide researchers the information necessary to solve the mystery as to why someone behaves in the way they do. I believe the study carried out by van Gerko et al. offers the best answer out of the four I reviewed because it attempted to focus on disordered eating behaviors as opposed to someone who is diagnosed with bulimia nervosa. There may be other contributing factors when examining someone with bulimia nervosa that get ignored or that can’t be controlled, so one cannot know if their answers are truly reliable. I am inclined to believe that someone with a past of CSA is more likely to engage in purging behaviors as purging is seen as the symbolism of rejecting or getting rid of something bad; people who were sexually abused want to purge and cleanse themselves of that traumatic experience. In this instance, although many probably disagree, Freudian Theory may have been onto something when saying that girls who had had issues or who were thought to have had some experience with engaging in oral sex with their father developed bulimia as a way to purge themselves of the act. It is clear, from all the studies that are available to be examined and critiqued, that further research and study must happen before an irrefutable answer can be given to this question.








          I.    Sanci, Lena; Coffey, Carolyn; Olsson, Craig; Reid, Sophie; Patton, George; Carlin, John B: “Childhood Sexual Abuse and Eating Disorders in Females”, Archives of Pediatrics & Adolescent Medicine, Vol 162, 2008

        II.    Waller, Glenn; “Sexual abuse as a factor in eating disorders”, British Journal of Psychiatry, Vol. 159, 2001

       III.    Welch, Sarah L; Fairburn, Christopher G; “Sexual abuse and bulimia nervosa; three integrated case control comparisons”, American Journal of Psychiatry, Vol 151, 1994

      IV.    van Gerko, Karen; Hughes, Mari Laura; Hammill, Michelle; Waller, Glenn; “Reported childhood sexual abuse and eating-disordered  cognitions and behaviors”, Child Abuse & Neglect: The International Journal, 2005




Psychology Department

The Health Psychology Home Page is produced and maintained by David Schlundt, PhD.


VuLogoVanderbilt Homepage

Return to the Health Psychology Home Page
Send E-mail comments or questions to Dr. Schlundt