A new study published by BMJ Mental Health and conducted by researchers from Yale University School of Medicine and the University of Oxford found no evidence that marijuana use is associated with faster cognitive decline or a higher risk of dementia in older adults.
Researchers examined data from two large cohorts, the UK Biobank and the U.S. Million Veteran Program, to explore how marijuana use may relate to cognitive performance and dementia risk later in life. The study looked at multiple areas of cognition, including memory, reasoning and processing speed, while also using Mendelian randomization to test whether any apparent associations might be causal.
In the UK Biobank portion of the study, marijuana users showed slightly better performance at baseline on tests of numeric memory and fluid intelligence. But those differences did not hold up when researchers looked at how cognition changed over time. According to the study, “no significant differences were observed in longitudinal cognitive change.” In the Million Veteran Program, researchers also found that cannabis use disorder was not significantly associated with incident all-cause dementia.
The study’s conclusion was direct, stating that “cannabis use was not linked to longitudinal cognitive decline or dementia risk.” Researchers added that, “within the limits of these cohorts, we found no evidence that cannabis use contributes substantially to cognitive ageing or dementia risk in older adults.”
The findings are notable given the growing use of marijuana among older adults for both medical and recreational purposes. Researchers said previous studies have often been limited by small sample sizes or cross-sectional designs, making it harder to determine whether marijuana itself plays a meaningful role in long-term cognitive outcomes.
Importantly, the authors said the modestly better baseline scores seen among some users were likely not caused by marijuana itself. Instead, they pointed to factors such as education, socioeconomic status and other confounding variables as more likely explanations.
