Lifetime Cannabis Use Not Linked to Cognitive Decline or Dementia in Older Adults: BMJ Mental Health Study

Breakthrough Findings from Oxford-Led Research on Cannabis and Aging Brains

Researchers at the University of Oxford have delivered compelling new evidence challenging long-held assumptions about cannabis's impact on cognitive health in later life. Published in the prestigious journal BMJ Mental Health, the study titled "Cannabis use, cognitive function and dementia risk in older adults: observational and genetic analyses" analyzed vast datasets from the UK Biobank and the US Million Veteran Program. These findings suggest that lifetime cannabis use does not accelerate cognitive decline or elevate dementia risk among older adults, marking a significant moment in academic inquiry into substance use and brain aging.

The investigation, led by DPhil candidate Saba Ishrat from Oxford's Department of Psychiatry, with senior oversight from Dr. Anya Topiwala of the Nuffield Department of Population Health, underscores the power of large-scale, longitudinal research in higher education settings. Universities like Oxford are at the forefront, leveraging massive cohorts to disentangle correlation from causation in complex public health questions.

Context: Rising Cannabis Use Among Older Adults and Research Gaps

As cannabis legalization expands globally—particularly for medical purposes—use among those over 50 has surged. In the UK, surveys indicate a doubling of recreational and therapeutic consumption in this demographic over the past decade, driven by conditions like chronic pain, insomnia, and anxiety. Yet, prior studies on cannabis and cognition yielded mixed results: some linked adolescent or heavy use to memory deficits, while others found no lasting effects in adults.

This Oxford study addresses a critical gap by focusing exclusively on older adults (aged 40+ at baseline), where age-related cognitive changes naturally occur. Traditional research often suffered from small samples, short follow-ups, or confounding factors like tobacco co-use. By harnessing university-led biobanks, the team could examine lifetime patterns prospectively, providing robust insights for clinicians, policymakers, and fellow academics.

Study Methodology: Power of Big Data in Academic Cohorts

The research employed a multi-pronged approach, combining observational epidemiology with advanced genetics. Key elements included:

• UK Biobank (UKB): A landmark resource from the University of Manchester and collaborators, comprising genetic, lifestyle, and health data from over 500,000 UK volunteers recruited between 2006 and 2010. Here, up to 18,975 lifetime cannabis users (self-reported 'ever used,' categorized by frequency: 1-10 times low; 11-100+ high) were compared to 60,598 non-users.
• Million Veteran Program (MVP): A US Department of Veterans Affairs initiative partnering with universities like Yale, involving over 900,000 veterans. The study focused on 12,222 with cannabis use disorder (CanUD) diagnoses via electronic health records (EHR).
• Cognitive assessments in UKB spanned five domains: numeric memory (recalling sequences), fluid intelligence (problem-solving), trail making (executive function), symbol digit substitution (processing speed), and pairs matching (visual memory). Tests were repeated ~7 years apart (2014-2021).
• Mendelian randomization (MR): Using genome-wide association study (GWAS) variants as proxies for cannabis exposure to infer causality, avoiding reverse causation biases common in observational data.

Covariates adjusted for age, sex, socioeconomic status (SES), education, comorbidities (e.g., depression, cardiovascular disease), and other substances. Statistical rigor included linear mixed-effects models for longitudinal change, Cox proportional hazards for dementia incidence, and inverse-variance weighted (IVW) MR with sensitivity checks.

Baseline Cognitive Performance: Surprising Advantages for Users

At study entry, cannabis users outperformed non-users on select tests. Numeric memory scores were higher by a standardized beta of 0.07 (95% CI 0.06-0.09, p<0.001), and fluid intelligence by 0.12 (95% CI 0.10-0.13, p<0.001). No differences emerged in other domains after false discovery rate (FDR) correction.

Lead author Saba Ishrat emphasized: "Although cannabis users performed slightly better on some cognitive tests at baseline, this should not be interpreted as cannabis improving cognition. The differences we observed are more likely explained by underlying demographic, educational and socioeconomic factors." Indeed, UKB users tended to be younger, male, higher-educated—hallmarks of volunteer bias in academic cohorts.

No Evidence of Accelerated Longitudinal Cognitive Decline

Over time, no accelerated decline linked to cannabis. Longitudinal betas hovered near zero (e.g., fluid intelligence: beta=0.00155, 95% CI -0.00082 to 0.00393, p=0.20). Frequency analyses (low vs. high use) and sex-stratified models confirmed consistency. This aligns with prior university-led longitudinals, like Danish cohorts tracking IQ over 44 years, finding minimal impact.

Dementia Risk Assessment in Veteran Populations

In MVP's real-world EHR data, CanUD showed no heightened dementia hazard: HR=1.11 (95% CI 0.97-1.26, p=0.12) for European ancestry; HR=0.97 (95% CI 0.79-1.20, p=0.80) African ancestry. Follow-up averaged 4.3 years, capturing incident cases via ICD codes.

These results from Yale-Oxford collaboration highlight EHR's value in higher ed research for scaling beyond self-reports.

Mendelian Randomization: Ruling Out Causality

MR provided genetic 'natural experiments.' No bidirectional causal links: cannabis variants predicted neither cognition (e.g., CanUD-numeric memory beta=-0.0076, p=0.78) nor dementia (OR=1.01, p=0.71). Reverse MR (cognition to cannabis) similarly null. This technique, pioneered in university genetics labs, strengthens claims against direct neurotoxicity in aging brains. For the full peer-reviewed paper, explore the BMJ Mental Health publication.

Expert Perspectives from Oxford Academics

Dr. Anya Topiwala, Honorary Consultant Psychiatrist at Oxford Health NHS Foundation Trust, stated: “Across large UK and US cohorts, and using genetic approaches... we found no evidence that cannabis use was associated with accelerated cognitive decline or increased dementia risk in older adults. However, this should not be interpreted as cannabis being risk-free.” She calls for studies on high-potency products amid rising medical prescriptions.

Professor Klaus P. Ebmeier, University of Oxford, contextualizes within neuroimaging trends: prior work showed corpus callosum changes, but non-causal.

Visit the Oxford press release for deeper quotes.

Limitations and Methodological Considerations

Honest science acknowledges bounds:

• UKB's healthy volunteer bias: users less likely heavy consumers.
• MVP skews to severe CanUD, veterans (mostly male).
• Self-report recall bias; no THC/CBD potency data amid evolving strains.
• Relatively short follow-up for dementia (needs 10+ years).
• Residual confounding (e.g., lifestyle unmeasured).

Future higher ed projects must prioritize detailed exposure histories via wearables or assays.

Implications for Higher Education and Research Careers

This study exemplifies interdisciplinary higher ed excellence: psychiatry, epidemiology, genomics converging at Oxford and Yale. It opens doors for PhD/postdoc roles in biobank analytics, MR methodology, and aging neuroscience. Universities increasingly fund such work, with grants from Wellcome Trust, NIH mirroring UKB/MVP scales.

For academics, it highlights ethical reporting: balanced views countering stigma while noting risks like psychosis vulnerability in youth.

Broader Public Health and Policy Ramifications

No cognitive harm signal reassures moderate older users, but cautions persist: respiratory issues from smoking, drug interactions. UK NICE guidelines may evolve, informing GP training at medical schools.

Comparisons: CU Anschutz research (2026) linked moderate use to larger CB1-rich brain volumes (hippocampus, amygdala), better memory—potentially neuroprotective via endocannabinoid modulation.

Future Directions in University-Led Cannabis Research

Next frontiers: potency-stratified trials, neuroimaging (fMRI during tasks), diverse ancestries. Oxford's Big Data Institute eyes AI for predicting trajectories. Collaborative EU-US consortia could track legalization cohorts.

Actionable for researchers: master R/Python for mixed models, PLINK for GWAS—skills boosting employability in academia.

Conclusion: Reshaping Narratives on Cannabis and Cognitive Aging

Oxford's BMJ Mental Health study reframes lifetime cannabis use as non-contributory to cognitive woes in seniors, empowering evidence-based discourse. As universities drive discovery, staying informed equips educators, clinicians, and policymakers. Ongoing vigilance ensures science evolves with society.

Photo by Elsa Olofsson on Unsplash