A recent Letter to the Editor, published in JACC: Advances, raises important concerns about how cannabis use is classified in epidemiologic research. Drs. Jamie Corroon and Igor Grant from the Center for Medicinal Cannabis Research (CMCR), along with Drs. Ryan Bradley and Gretchen Bandoli from the Herbert Wertheim School of Public Health and Human Longevity Science, and Dr. Matthew A. Allison from the Department of Family Medicine at UC San Diego, critiqued a retrospective analysis that reported a sixfold increased risk of myocardial infarction among cannabis users under 50.

The authors argue that the exposure in this study—labeled as “cannabis use”—was based entirely on diagnostic codes from the International Classification of Diseases (ICD), such as F12.1 (cannabis abuse), F12.9 (unspecified cannabis use), and F12.90 (unspecified, uncomplicated cannabis use). These codes reflect clinically diagnosed cannabis use and Cannabis Use Disorder, not general "cannabis use". As such, the collapsing of these diagnoses into a single term conflates general cannabis use with diagnosed cannabis use and abuse, despite substantial differences between these classifications. In addition, "non-users", as defined in the study, are more accurately described as patients without a cannabis-related diagnosis.

In addition, the authors highlight that patients diagnosed with a cannabis use ICD code may differ systematically from those without such diagnoses in ways that affect cardiovascular risk. These include higher rates of psychiatric conditions, substance use disorders, and other social determinants of health. The study in question failed to adjust for several of these confounders—including anxiety, schizophrenia, bipolar disorder, tobacco use, and psychotropic medications—despite their relevance.

The letter calls for more rigorous classification of cannabis exposure in health data, including stratification by ICD code subtype, and urges researchers to conduct quantitative bias analyses to assess the potential impact of misclassification and residual confounding.

Read the full letter here.

Citation: Corroon J, Bradley R, Grant I, Allison MA, Bandoli G. Limitations of Diagnostic Codes in Assessing Cannabis-Related Cardiovascular Risk. JACC: Advances. 2025 Aug;4(8):101960. https://tr.ee/IjHDOa