Long-term outcomes of childhood sexual abuse: an umbrella review

Summary Background Although many meta-analyses have examined the association between childhood sexual abuse and subsequent outcomes, the scope, validity, and quality of this evidence has not been comprehensively assessed. We aimed to systematically review existing meta-analyses on a wide range of long-term psychiatric, psychosocial, and physical health outcomes of childhood sexual abuse, and evaluate the quality of the literature. Methods In this umbrella review, we searched four databases (PsycINFO, PubMed, Cumulative Index to Nursing and Allied Health Literature, and Global Health) from inception to Dec 31, 2018, to identify meta-analyses of observational studies that examined the association between childhood sexual abuse (before 18 years of age) and long-term consequences (after 18 years). We compared odds ratios (ORs) across different outcomes. We also examined measures of quality, including heterogeneity between studies and evidence for publication bias. This study is registered with PROSPERO, CRD42016049701. Findings We identified 19 meta-analyses that included 559 primary studies, covering 28 outcomes in 4 089 547 participants. Childhood sexual abuse was associated with 26 of 28 specific outcomes: specifically, six of eight adult psychiatric diagnoses (ORs ranged from 2·2 [95% CI 1·8–2·8] to 3·3 [2·2–4·8]), all studied negative psychosocial outcomes (ORs ranged from 1·2 [1·1–1·4] to 3·4 [2·3–4·8]), and all physical health conditions (ORs ranged from 1·4 [1·3–1·6] to 1·9 [1·4–2·8]). Strongest psychiatric associations with childhood sexual abuse were reported for conversion disorder (OR 3·3 [95% CI 2·2–4·8]), borderline personality disorder (2·9 [2·5–3·3]), anxiety (2·7 [2·5–2·8]), and depression (2·7 [2·4–3·0]). The systematic reviews for two psychiatric outcomes (post-traumatic stress disorder and schizophrenia) and one psychosocial outcome (substance misuse) met high quality standards. Quality was low for meta-analyses on borderline personality disorder and anxiety, and moderate for conversion disorder. Assuming causality, population attributable risk fractions for outcomes ranged from 1·7% (95% CI 0·7–3·3) for unprotected sexual intercourse to 14·4% (8·8–19·9) for conversion disorder. Interpretation Although childhood sexual abuse was associated with a wide range of psychosocial and health outcomes, systematic reviews on only two psychiatric disorders (post-traumatic stress disorder and schizophrenia) and one psychosocial outcome (substance misuse) were of a high quality. Whether services should prioritise interventions that mitigate developing certain psychiatric disorders following childhood abuse requires further review. Higher-quality meta-analyses for specific outcomes and more empirical studies on the developmental pathways from childhood sexual abuse to later outcomes are necessary. Funding Wellcome Trust.