The relationship of frequency of cocaine use to substance and psychiatric disorders in the U.S. general population

Abstract Background In clinical trials of pharmacotherapy for substance use, abstinence is the primary endpoint accepted by regulatory agencies. However, this endpoint could be overly restrictive, impeding efforts to identify effective medications for cocaine use disorder. To examine non-abstinent gradations in cocaine use as potential indicators of improvement, we investigated the relationship of frequency of cocaine use to clinical correlates in national survey data. Methods Lifetime cocaine users (n = 2501) were interviewed in the National Epidemiological Survey on Alcohol and Related Conditions (NESARC) in 2001−2002 and re-interviewed in 2004–2005. Adjusted odds ratios (aORs) indicated associations between heaviest frequency of cocaine use and use of other substances, DSM-IV substance use disorders, psychiatric disorders, and change between 2001−2002 and 2004–2005. The reference category for all aORs was non-users. Results Greater lifetime cocaine use frequency was associated with lifetime cocaine, alcohol, and cannabis dependence (aOR for a linear trend = 2.80, 1.22, 1.22, respectively) and past-year cocaine, alcohol, and cannabis dependence (aOR = 1.78, 1.13, 1.16, respectively). Greater lifetime cocaine use frequency was associated with past-year depressive, panic, and generalized anxiety disorders (aOR = 1.07, 1.09, 1.12, respectively). Among cocaine users in 2001−2002, compared to the reference group using less than monthly, use ≥1x/week and use 1−3 times a month was associated with cocaine use disorder in 2004–2005 (aOR = 2.13 and aOR = 1.67, respectively). Conclusion Gradations in risk for dependence on cocaine, other substances and psychiatric disorders by frequency of cocaine use indicates a promising direction for more sensitive outcome measures of treatment effects on cocaine outcomes than binary indicators (e.g., any use vs. none). Study results add to findings suggesting that non-abstinent measures might be useful indicators of treatment efficacy in clinical trials.