Background The directionality and magnitude of the association of cannabis use with elevated anxiety symptoms in the general population is unclear. The aim of this meta-analysis was to investigate the association of cannabis use with the development of elevated anxiety symptoms in the general population.

Methods A ‘random effects’ meta-analysis of prospective longitudinal studies was undertaken in line with Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines. Six databases were systematically searched up until 20 May 2016: PsycINFO, MEDLINE, EMBASE, CINAHL Plus, Social Science Citation Index and System for Information on Grey Literature in Europe (SIGLE). Searching ceased on 20 May 2016. The exposure was cannabis use (or use frequency), measured at baseline and the outcome was anxiety, using diagnosis or cut-off points on standardised scales measuring symptoms.

Results The main analysis (k=10; N=58 538) demonstrated an association of cannabis use with anxiety, with a very small OR of 1.15 (95% CI 1.03 to 1.29). Restricting the analysis to high-quality studies (k=5) decreased the OR considerably (OR=1.04; 95% CI 0.91 to 1.19), as did adjusting for publication bias (OR=1.08; 95% CI 0.94 to 1.23). Studies with a baseline in the last 10 years yielded a lower pooled OR than studies with an earlier baseline and studies set in the Americas yielded a markedly higher pooled OR than European studies and Australasian studies.

Conclusions The findings indicate that cannabis use is no more than a minor risk factor for the development of elevated anxiety symptoms in the general population. They may inform the debate surrounding the legalisation of cannabis.