Sexual Abuse and Lifetime Diagnosis of Somatic Disorders:  A Systematic Review and Meta-analysis FREE

Many patients seen for general medical care have a history of sexual abuse. In the United States, national population surveys have found the annual incidence of sexual violence to be 2.5% for women and 0.9% for men.¹ Additionally, it is estimated that 1 in 15 adults has experienced forced sexual intercourse.¹ Retrospective studies have also demonstrated a high prevalence of adult survivors of childhood sexual abuse: 16% of men and 25% of women in the United States.² Estimates of the prevalence of rape range from 7%-21% for adults and 5%-17% for children.^3- 7 These statistics are likely underestimates because the event is frequently underreported.^8- 9

To date, research on the long-term effects of sexual abuse has primarily focused on mental health outcomes. Strong evidence supports a link between childhood sexual abuse and multiple psychiatric sequelae.^10- 11 However, studies investigating the association between sexual abuse and somatic outcomes have been less definitive. Initial inquiries date back to the late 1970s, when the link between sexual abuse and “hysterical seizures” was investigated.¹² Additional studies evaluating the association of sexual abuse with chronic pelvic pain^13- 14 and chronic pain^15- 16 followed in the 1980s. Over time, disorders including fibromyalgia,^17- 19 functional gastrointestinal disorders,^5,20- 21 and obesity²¹ have been assessed for their association with sexual abuse.

Epidemiologic studies demonstrate that such disorders are commonly encountered by physicians across a broad spectrum of medical practice. Fibromyalgia occurs in approximately 5% of patients seen in general practices.^22- 23 Functional gastrointestinal disorders are also highly prevalent, with estimates of 5% to 25% for irritable bowel syndrome in the general population.^24- 25 As such, physicians are commonly faced with the diagnosis and management of these disorders.

Although several studies have attempted to investigate associations between patient history of sexual abuse and development of somatic disorders, to our knowledge no systematic review has been performed to evaluate the strength of these associations. The objective of this study was to systematically assess and summarize the available evidence regarding the association between a history of sexual abuse and lifetime diagnosis of somatic disorders.

The protocol for this systematic review was developed and executed by physicians and researchers in the disciplines of internal medicine, preventive medicine, epidemiology, statistics, and psychology. The methods described by the Cochrane Collaboration were used in the development of the protocol.²⁶ The reporting of results was based on the recommendations of the Meta-analysis of Observational Studies in Epidemiology group.²⁷

A comprehensive search of several databases (from January 1980 to December 2008, all age groups, any language, any population), including MEDLINE, EMBASE, CINAHL, Current Contents, PsycINFO, American College of Physicians Journal Club, Cochrane Controlled Trials Registry, Cochrane Database of Systematic Reviews, and Database of Abstracts and Reviews of Effectiveness, was conducted. The search strategy was designed and performed by an experienced librarian (L.J.P.), with input from the study's principal investigator. Controlled vocabulary supplemented with keywords was used to define the concept areas, sexual abuse, and physical/psychological disorders, as well as to limit epidemiologic studies. The complete strategy is available on request from the author.

Eligible studies were longitudinal case-control and cohort studies. Studies were included regardless of publication status, sample size, length of follow-up, or language of publication.

Sexual abuse was categorized into 2 major groups: rape and all forms of sexual abuse. Quiz Ref IDRape was defined as penetration with a body part or foreign object (intravaginal or anal).²⁸All forms of sexual abuse was designed to be broad to capture the wide variety of definitions used to characterize sexual violence. These other forms included, but were not limited to, noncontact exposure of genitals, threatening sexual violence, and contact involving genitals and mouth.²⁸ If a study did not specify that the form of sexual abuse was rape, then it was classified as all forms of sexual abuse.

For studies in which abuse was not further specified, authors were contacted to provide data for that subset of patients with a history of sexual abuse. If studies included data for sexually abused, nonabused, and physically abused individuals, the latter groups were pooled as controls. This practice was adopted to reduce the confounding effect of physical abuse on the outcome as studies have shown childhood physical abuse prevalence rates ranging between 3.8% and 20.6% in the general population.^29- 31

As per protocol, the outcomes investigated in this study were functional gastrointestinal disorders, chronic headache, psychogenic seizure disorder (also called nonepileptic seizure or pseudoseizure), non-neurocardiogenic syncope, fibromyalgia, obesity, chronic pain syndrome, and chronic pelvic pain syndrome. Eligible studies contained an outcome and a nonoutcome group, designated by self-report or clinical identification.

Teams of reviewers (M.L.P., L.P.C., E.N.G., A.L.S., K.M.C., A.Z.) working independently and in pairs analyzed eligible titles, abstracts, and full-text articles. Data from included studies were extracted in duplicate. Articles in foreign languages (Chinese [1], Czech [1], French [10], German [18], Hungarian [1], Italian [4], Spanish [8], and Turkish [2]) were translated or the author was contacted when available to provide required information.

A description of the type of study and its participants was obtained, including mean age of subjects, sex, and race. If greater than 70% of the study population was of one race, then the study was designated as evaluating that race predominantly. If not further specified, research studies based in Scandinavian countries were assumed to enroll predominantly white participants. The type of sexual abuse (either rape or all forms of sexual abuse) was recorded, along with the age at the abuse. Childhood abuse was defined as abuse occurring at or before age 18 years. Raw data for the exposed, nonexposed, outcome, and nonoutcome groups were obtained when possible. If numbers were not available, then odds ratios (ORs) were recorded, with preference given to adjusted ORs. Authors of eligible articles not presenting raw data or ORs were contacted by e-mail or letters. Authors were requested to provide data required for the analysis that were not printed in the published article.

The quality of each eligible study was assessed in duplicate. Disagreements were resolved by mediation, with input from the principal investigator. The Newcastle-Ottawa Quality Assessment Scale was used for case-control studies and cohort studies.³² This scale consists of 8 questions, with a maximum of 10 possible points for each type of study.

Odds ratios were pooled for dichotomous outcomes from each study with the DerSimonian-Laird random-effects model. The 95% confidence interval (CI) for each outcome was estimated to reflect the uncertainty of point estimates.³³

An OR of 1.0 indicates no association and OR greater than 1.0 indicates increased risk for the referenced outcome. The I² statistic was used to estimate the percentage of total variation across studies because of heterogeneity, rather than chance (ie, the percentage of variability of associations across studies that is not due to chance or random error, but rather due to real differences in study patients, design, or outcome definitions).³⁴ I² values of less than or equal to 25%, 50%, and greater than or equal to 75% represent low, moderate, and high inconsistency, respectively. Statistical analysis was conducted with Comprehensive Meta-Analysis, version 2 (Englewood, New Jersey). P<.05 was set as the threshold for significance.³⁵

A priori hypotheses were formed to explore subgroup interactions and explain inconsistency in the direction and magnitude of associations among studies. These included study design (cohort design vs case-control design), the age at which sexual abuse took place (childhood vs adult), sex of the abused person, and race of the abused subject. To test the hypotheses of a subgroup effect, a test of interaction with a predetermined 2-tailed α of .05 was used.³⁶ Meta-regression was used to assess whether study quality affects the strength of reported associations. In the regression model, the natural logarithm of the OR represents the dependent variable and the quality score represents the independent variable.

A sensitivity analysis was planned to determine whether the type of abuse (rape vs all forms of sexual abuse) affects study conclusions.

To assess the potential effect of publication bias, we planned to inspect funnel plots for asymmetry and use the Duval and Tweedie trim and fill method³⁷ and the Begg and Mazumdar rank correlation test.³⁸

A total of 23 studies met the inclusion criteria and provided data for 4640 subjects (Figure 1). Nineteen of the included studies were case-control design (2778 individuals) and 4 were cohort design (1862 individuals). Sixteen included only female subjects and the remainder evaluated both sexes. Four investigated populations composed of a majority of white subjects, whereas the remaining sample populations were mixed or not specified in terms of race. A total of 11 studies took place in nations outside of the United States. No foreign-language study met inclusion criteria, and no unpublished study met eligibility criteria.

Seven of the studies investigated sexual abuse occurring only in childhood. None of the studies examined abuse only in adulthood. Seven included data for both childhood and adulthood abuse. The remaining studies did not specify the age at which abuse occurred. The included case-control studies ascertained sexual abuse exposure by using self-reporting in questionnaires (11 studies) and interviews (8 studies). Cohort studies confirmed histories of sexual abuse by using legal documentation (3 studies) and interviews (1 study).

Four studies provided data for individuals who had experienced rape. The remaining studies collected data from persons who reported exposure to all forms of sexual abuse. Nineteen of the 23 studies provided definitions used to assess sexual abuse. These included the following: threatening sexual attacks, sexual coercion, unwanted exposure of perpetrator's genitalia, unwanted touching of abused individual's genitals or other sexual parts, and unwanted or forced touching of perpetrator's genitals or other sexual parts. The details of the baseline characteristics of each study are listed in Table 1.

Each study was assessed for quality according to the population and sampling methods, description of exposure and outcomes, and statistical adjustment of the data (Table 2 and Table 3). Only 5 of 23 studies matched abuse survivors and controls by 2 factors or adjusted for 2 or more confounders simultaneously in their multiple regression models. Among case-control studies, only 13 of 19 studies obtained controls from the same population as cases. Furthermore, only 10 of 19 case-control studies were considered to use rigorous selection criteria of cases (during defined period, throughout defined area, all cases or random/consecutive cases in a defined group).

Quiz Ref IDThere was a significant association between a history of sexual abuse and a lifetime diagnosis of functional gastrointestinal disorders (OR, 2.43; 95% CI, 1.36-4.31; I² = 82%; 5 studies), nonspecific chronic pain (OR, 2.20; 95% CI, 1.54-3.15; I² = not available; 1 study), psychogenic seizures (OR, 2.96; 95% CI, 1.12-4.69; I² = 0%; 3 studies), and chronic pelvic pain (OR, 2.73; 95% CI, 1.73-4.30; I² = 40%; 10 studies).

There was no statistically significant association between a history of sexual abuse and a lifetime diagnosis of fibromyalgia (OR 1.61; 95% CI, 0.85-3.07; I² = 0%; 4 studies), obesity (OR, 1.47; 95% CI, 0.88-2.46; I² = 71%; 2 studies), or headache (OR, 1.49; 95% CI, 0.96-2.31; I² = not available; 1 study). We found no longitudinal studies that assessed the outcome of syncope. The results of meta-analyses are depicted in Figure 2.

We found no significant subgroup × strength of association interactions according to the age of abuse (childhood vs adulthood). Only for the outcome of obesity did we find a significant interaction suggesting higher association in studies with case-control design compared with cohort design. This interaction (design × strength of association) was not found for other outcomes. There were insufficient data to conduct subgroup analyses based on the sex or the race of the abused individual. Subgroup analyses are summarized in Table 4. Meta-regression did not reveal a statistically significant association between the quality score and the effect size (P > .05 for all outcomes).

To determine the effect of sexual abuse category on the estimated association, we conducted a sensitivity analysis, including only those studies that defined sexual abuse as rape (Figure 3). A significant association was found between a history of rape and a lifetime diagnosis of fibromyalgia (OR, 3.35; 95% CI, 1.51-7.46; I² = not available; 2 studies), functional gastrointestinal disorders (OR, 4.01; 95% CI, 1.88-8.57; I² = not available; 1 study), and chronic pelvic pain (OR, 3.27; 95% CI, 1.02-10.53; I² = not available; 1 study). There were insufficient data to conduct this sensitivity analysis for the other outcomes.

Inspection of funnel plots and statistical tests for publication bias did not show an obvious effect of publication bias. However, these analyses were underpowered because of the small number of included studies.

We performed a comprehensive systematic review of the literature and meta-analysis to assess the association between sexual abuse and the lifetime prevalence of several somatic disorders. This review found that sexual abuse was associated with a lifetime diagnosis of nonspecific chronic pain, functional gastrointestinal disorders, psychogenic seizures, and chronic pelvic pain. There was no statistically significant association between a history of sexual abuse and the lifetime diagnosis of fibromyalgia, obesity, or headache. No data were available to assess the outcome of syncope. When analysis was restricted to studies in which sexual abuse was defined as rape, significant associations were observed between rape and the lifetime diagnosis of fibromyalgia, chronic pelvic pain, and functional gastrointestinal disorders.

In studies evaluating obesity, case-control studies demonstrated a greater association compared with cohort studies. This observation may be due to type I error or chance. Alternatively, this observation may reflect recall bias in studies of case-control design.

To our knowledge, this is the first attempt to summarize the available data on the association between patient history of sexual abuse and somatic outcomes. Quiz Ref IDPrevious systematic reviews empirically evaluating childhood sexual abuse outcomes have revealed increased psychiatric outcomes, behavior problems, victim-perpetrator cycle, and adult maladjustment.^10- 11,58 Our systematic review builds on this work by expanding it to include somatic sequelae.

Significant strengths of this study include the exhaustive and reproducible search strategy, attempts to decrease bias by performing selection, review and extraction of data in pairs of reviewers, and communication with authors of original studies to ascertain unpublished or incomplete data. Furthermore, efforts were made to evaluate foreign-language and unpublished studies. Subgroup and sensitivity analyses were conducted to explain heterogeneity.

The principal limitation of this study was the use of evidence susceptible to bias. No studies met all of the Newcastle-Ottawa criteria for quality. Only 2 of 23 studies met at least 8 of the 10 maximum allowed points of the criteria. Analysis with meta-regression did not demonstrate an association between study quality (total number of points for each study) and effect size. This lack of association is consistent with previous literature that revealed that summary scores of clinical trials' quality do not correlate with treatment effect.⁵⁹ We did not detect publication bias; however, our analysis was underpowered because of the small number of included studies.⁶⁰ Also, unexplained heterogeneity was demonstrated in analyses of the outcomes of functional gastrointestinal orders and obesity. Imprecision caused by the small number of events could have masked a real association between sexual abuse and the outcomes of obesity, headache, and fibromyalgia. In addition, the associations reported in this review could be overstated because of the use of ORs to estimate the risk of a common event.⁶¹ Therefore, the overall level of evidence found by this review may be reduced by methodological limitations, heterogeneity, and imprecision.⁶²

Also, our results may not generalize to men. Sixteen of the 23 studies in this review were restricted to female subjects, whereas no study evaluated male subjects exclusively. These findings may reflect the higher prevalence rates of sexual abuse against females; however, sexual abuse is known to affect males. Further research investigating males would be worthwhile to determine whether similar sequelae also tend to affect male individuals or whether this sex is prone to develop different sets of outcomes.

Another potential limitation of this study is recall bias. Although 3 of the 4 cohort studies obtained data on sexual abuse exposure from medical or legal documentation, all case-control studies relied on self-reporting from either questionnaires or interviews. However, use of questionnaires or interviews to ascertain exposure may be problematic. Previous studies have found notable variability in the percentage of documented survivors of childhood sexual abuse (62%-81%) who recall the abuse as adults.^63- 65 These statistics indicate that questionnaire respondents or interview subjects likely underreport childhood sexual abuse. The effect of this underreporting would be the inclusion of sexual abuse survivors in control groups. This in turn may diminish the effect size of the association between sexual abuse and somatic outcomes.

As an additional point, research shows that emotional, physical, and sexual abuse tend to coexist.^29- 30 However, only a minority of studies ascertained whether subjects had been exposed to 1 or multiple categories of abuse. To better define outcomes specifically associated with sexual abuse, we recommend that research studies work to better clarify the type of abuse to which patients have been exposed.

To our knowledge, there is no validated theory to explain the association between sexual abuse and chronic somatic disorders. Sexual abuse, however, may be an early, inciting environmental factor in a multistep process that culminates in physical dysfunction. Furthermore, neuroendocrine maladaptation may be important to this progression. Animal studies have revealed that early life stress is associated with dysregulation of the hypothalamic pituitary axis.^66- 67 Hypothalamic-pituitary axis dysregulation has also been measured in women with a history of childhood sexual abuse who experience depression and anxiety. Similar findings of hypothalamic-pituitary axis dysregulation were observed in combat veterans with posttraumatic stress disorder.^68- 69 Evidence also suggests that neuroendocrine changes caused by stress, including dysregulation of corticotropin-releasing factor, may result in neurologic and behavioral changes.^70- 72 Similar dysregulation of corticotropin-releasing factor has been demonstrated in laboratory animals with colonic dysmotility.^72- 74 However, the role of hypothalamic-pituitary axis dysregulation in chronic pelvic pain and fibromyalgia remains controversial.^46,75- 77 We speculate that the neuroendocrine system mediates the connection between sexual abuse and the development of somatic dysfunction.

Building greater awareness of the association between sexual abuse and somatic disorders may lead to improved health care delivery and outcomes for sexual abuse survivors. Quiz Ref IDAs a group, survivors of abuse have higher medical care use and incur greater costs compared with the general patient population.^21,78- 81 Analysis of expenditures demonstrates that costs are primarily the result of increased use of primary care, specialty medicine, and pharmacy and laboratory services.^82- 83 Higher medical use may also expose these patients to greater risk without clearly defined benefits, including increased abdominal and pelvic surgeries, adverse effects of medications, and chronic opioid use and dependence.^84- 89

Despite evidence of high health care use among sexual abuse survivors, physicians remain largely unaware of this aspect of their patients' medical history. Only 5% of sexual abuse survivors report a history of abuse to their physician.⁵ However, heightened awareness of these specific health associations may prompt earlier recognition and improve care for sexual abuse survivors. Quiz Ref IDOne systematic review found that disclosure of childhood sexual abuse memories during psychotherapy may improve posttraumatic stress disorder symptoms, decrease depression, and alleviate anxiety.⁷⁸ Also, cognitive behavioral therapy and cognitive processing therapy have been found effective in treating posttraumatic stress disorder in survivors of sexual abuse.^90- 91 We hypothesize that similar therapeutic approaches may be feasible for sexual abuse survivors with somatic sequelae.⁹² Given evidence of sexual abuse prevalence and related physical and mental health sequelae, we urge physicians to more routinely conduct inquiries about sexual abuse in patients with the identified somatic syndromes. Disclosure of abuse in the clinic setting may allow for earlier consultation with mental health professionals. Prompt recognition of the physical and psychological sequelae of sexual abuse may halt unnecessary medical escalation and provide care better suited to promote recovery.

This review highlights the need for future research to better define the nature and scope of mental and physical health outcomes in sexual abuse survivors. As with previous reviews of sexual abuse and psychiatric sequelae, we were unable to more fully ascertain specific mediators of chronic somatic illness, such as duration of abuse, sex of the victim, and socioeconomic factors (eg, family income, single-parent upbringing, parental alcohol use). To more precisely define the temporal relationship between sexual abuse and somatic outcomes, further studies will be necessary. We recommend a prospective cohort study to evaluate somatic outcomes in survivors of abuse, with focused attention on the potential mediators and modifiers of the associations. This design would also be predicted to better address concerns of potential bias in the literature. Additionally, the premise that disclosure of abuse during psychotherapy improves psychiatric symptoms opens the door to clinical trials to assess the utility of abuse disclosure in the treatment of somatic sequelae. Finally, the success of cognitive behavioral therapy treatment of posttraumatic stress disorder for sexual abuse survivors points to a possible role for this modality for abuse survivors with somatic sequelae.

Sexual abuse remains prevalent and survivors are commonly encountered in general medical practice. Increasingly, it has been shown that survivors of sexual abuse face a spectrum of often challenging health concerns, resulting in greater health care use and cost and significant morbidity. This systematic review and meta-analysis demonstrates an association between sexual abuse and lifetime diagnosis of functional gastrointestinal disorders, chronic nonspecific pain, psychogenic nonepileptic seizure disorder, and chronic pelvic pain. Recognition of this association may have important clinical implications for patients coping with these disorders and their clinicians. Future research should be conducted to better define the epidemiology of this association and improve clinical outcomes for sexual abuse survivors with somatic sequelae.