Across 22 Longitudinal Studies, Cannabis Use Tracked With Higher Odds of Later Depression

When researchers combined results from 22 longitudinal studies that adjusted for baseline depression, cannabis users had higher odds of later depression than non‑users. The pooled odds ratio was 1.29, which translates to roughly 29 percent higher odds on average. Most included samples were adolescents and many were based in the United States. Risk of bias was rated medium overall, with the biggest concern being how cannabis exposure was measured. Funnel plot and Egger testing did not flag publication bias. The signal is consistent across cohorts but modest in size and based on observational data.

Systematic review and meta-analysis of longitudinal human studies that measured cannabis use, controlled for depression at baseline, and assessed later depression outcomes. The team screened 1,599 titles across databases and included 22 studies in the meta-analysis. Random effects models were used, along with multilevel meta-regression to explore moderators. Eleven studies were from the United States and most participants were under 18 at enrollment. Risk of bias was assessed and rated medium overall, with exposure measurement as a key weakness. Publication bias was evaluated using a funnel plot and Egger's Sandwich test, which did not suggest bias.

Debates about cannabis and mental health often cite cross-sectional snapshots that cannot separate cause from correlation. This project aggregated only longitudinal studies that accounted for baseline depression, offering a clearer look at whether cannabis use is followed by higher odds of later depression in real-world cohorts during a period of shifting laws and rising potency.

A modest association that repeats across many cohorts can still matter at the population level. Here, the effect size is small to moderate and the studies are observational, so residual confounding remains on the table. Exposure was often captured as a simple yes or no rather than frequency, potency, or product type. That limits what can be said about dose, trajectory, or risk in the current market where concentrates and high-THC products are common. The design choice to adjust for baseline depression supports a temporal link, but it does not establish causation or rule out shared risk factors such as family history, early adversity, or co-use of alcohol and nicotine.

Observational evidence only. Even with baseline adjustment, unmeasured confounding can explain part of the signal. Cannabis exposure was inconsistently measured, often as any use versus none, with little information on frequency, potency, product type, route, or duration. Most participants were adolescents, which narrows generalizability to older adults. Heterogeneity in how depression was assessed across cohorts can affect comparability. Funding sources and conflicts were not reported in the abstract. There is a timeline inconsistency in the provided materials, which cite a 2014 publication but describe a search window extending to March 2023; if this reflects an updated analysis that retained the original author list and title, the pooled estimate may differ from the 2014-only dataset.