Most Cannabis Use Came From People With Recent Mental Illness in a Large U.S. Survey

Researchers analyzed 43,070 U.S. adults from the 2001 to 2002 National Epidemiologic Survey on Alcohol and Related Conditions, a nationally representative household survey. Psychiatric diagnoses were assessed with the Alcohol Use Disorders and Associated Disabilities Interview Schedule-IV, covering DSM-IV Axis I and II conditions over the past 12 months. Outcomes included cannabis use frequency (at least weekly, or less than weekly) and DSM-IV cannabis use disorders. Logistic models adjusted for sociodemographic factors and other substance use disorders. The team also estimated the share of total cannabis consumption attributable to people with and without mental illness using self-reported frequency and daily dose.

Public health debates often treat cannabis use as broadly distributed across the population. This analysis suggests that, at least in 2001 to 2002, use and cannabis use disorders clustered among people with recent mental illness, with implications for how researchers characterize risk and where services are typically directed.

The concentration of cannabis use and cannabis use disorders within psychiatric populations challenges the idea that cannabis exposure is evenly spread across adults. The pattern aligns with several possible explanations, including shared vulnerability factors, self-medication, or environmental stressors that travel with both mental illness and substance use. The study reports strong associations, not causes, and it highlights diagnostic groups where cannabis involvement was most common in this dataset, such as bipolar disorder and certain personality disorders.

Cross-sectional design cannot establish which came first, mental illness or cannabis involvement. Diagnoses relied on a structured interview (AUDADIS-IV), which can misclassify some Axis I and II conditions. Cannabis measures came from self-reported frequency and typical daily dose, with no verification of product potency, THC-to-CBD ratios, or mode of use. The consumption estimate (83%) depends on self-report–based calculations rather than direct measurement. Data were collected in 2001 to 2002, before major shifts in legalization, product potency, and market dynamics, which may limit relevance to current patterns. Residual confounding is possible despite adjustment for demographics and other substance use disorders.