Cocaine use and other suspected risk factors for obsessive-compulsive disorder: a prospective study with data from the Epidemiologic Catchment Area surveys

Using prospectively gathered epidemiologic data, we sought to estimate the degree to which the risk of obsessive-compulsive disorder might be elevated among adults actively using cocaine, with and without illicit use of marijuana or other controlled substances. Study subjects were selected in 1980–1984 by taking probability samples of adult household residents at five sites of the Epidemiologic Catchment Area (ECA) Program: New Haven, Connecticut; Baltimore, Maryland; St. Louis, Missouri; Durham-Piedmont, North Carolina; Los Angeles, California. Soon after sampling at baseline, a total of 18 572 participants completed standardized interviews to measure suspected risk factors (including illicit drug use) and to evaluate whether they had met diagnostic criteria for currently or formerly active obsessive-compulsive disorder (OCD), as well as other mental disorders. The interviews were re-administered 1 year later to identify incident cases among 13 306 at-risk participants, 414 being active cocaine users. After sorting these participants into strata defined by age and census tract of household residence and after excluding persons found at baseline to have active or prior OCD, we found 105 incident cases of OCD within 103 of the age-matched and residence-matched strata, which also contained a total of 514 subjects who had not developed OCD. Applying standard epidemiologic strategies presented in prior ECA research reports, we performed conditional multiple logistic regression to estimate the risk of OCD for active cocaine users versus non-users. We also performed unconditional multiple logistic regression to estimate OCD risk for the 414 active cocaine users versus the 12 892 participants not using cocaine. Both of these epidemiologic strategies yielded consistent results: subjects actively using cocaine and also marijuana were found to be at increased risk for OCD. Under the conditional model, the estimated relative risk was 7.2 ( P = 0.03), while the value from unconditional regression was 4.1 ( P = 0.01). Active users of cocaine almost always were active users of marijuana or some other controlled substance, so it was not possible to estimate a relative risk value for subjects using cocaine only. Nonetheless, if replicated, this epidemiologic test of the cocaine-OCD hypothesis warrants attention in laboratory and clinical research, as do other suspected risk factors identified in the study, including sex (being female), employment status (not working for p