Suicide attempts in US adults with lifetime DSM-5 eating disorders

BACKGROUND: Rates of suicide are increasing in the US. Although psychiatric disorders are associated with suicide risk, there is a dearth of epidemiological research on the relationship between suicide attempts (SAs) and eating disorders (EDs). The study therefore aimed to examine prevalence and correlates of SAs in DSM-5 EDs—anorexia nervosa (AN), bulimia nervosa (BN), and binge eating disorder (BED)—in a nationally representative sample of US adults. In addition, prevalence and correlates of SAs were examined in the two subtypes of AN—restricting (AN-R) and binge/purge (AN-BP) types. METHODS: The study included 36,171 respondents in the Third National Epidemiological Survey on Alcohol and Related Conditions (NESARC-III) who completed structured diagnostic interviews (AUDADIS-5) and answered questions regarding SA histories and psychosocial impairment associated with EDs. We evaluated lifetime prevalence of SA, psychosocial impairment, clinical profiles, and psychiatric comorbidity in adults with EDs with and without SA histories, and temporal relationships between age onset of SA and EDs. RESULTS: Prevalence estimates of suicide attempts were 24.9% (for AN), 15.7% (for AN-R), 44.1% (for AN-BP), 31.4% (for BN), and 22.9% (for BED). Relative to respondents without specific EDs, adjusted odds ratios (AORs) of SAs were significantly greater in all EDs: AN = 5.40 (95% confidence intervals [CIs] = 3.80–7.67), AN-R = 3.16 (95% CIs = 1.82–5.42), AN-BP = 12.09 (95% CIs = 6.29–23.24), BN = 6.33 (95% CIs = 3.39–11.81), and BED = 4.83 (95% CIs = 3.54–6.60). Among those with SA history, mean age at first SA and number of SAs were not significantly different across the specific EDs. SA was associated with significantly earlier ED onset in BN and BED, longer duration of AN but shorter duration of BN, greater psychosocial impairment in AN and BN, and with significantly increased risk for psychiatric disorder comorbidity across EDs. Onset of BED was significantly more likely to precede SA (71.2%) but onsets of AN (50.4%) and BN (47.6%) were not. CONCLUSIONS: US adults with lifetime DSM-5 EDs have significantly elevated risk of SA history. Even after adjusting for sociodemographic factors, those with lifetime EDs had a roughly 5-to-6-fold risk of SAs relative to those without specific EDs; the AN binge/purge type had an especially elevated risk of SAs. SA history was associated with distinctively different clinical profiles including greater risk for psychosocial impairment and psychiatric comorbidity. These findings highlight the importance of improving screening for EDs and for suicide histories. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s12916-019-1352-3) contains supplementary material, which is available to authorized users.