Continued versus discontinued cannabis use in patients with psychosis: a systematic review and meta-analysis.

Background: Although the link between cannabis use and development of psychosis is well established, less is known about the effect of continued versus discontinued cannabis use after the onset of psychosis. We aimed to summarise available evidence focusing on the relationship between continued and discontinued cannabis use after onset of psychosis and its relapse. Methods: In this systematic review and meta-analysis, we searched MEDLINE for articles published in any language from the database inception date up until April 21, 2015 that included a sample of patients with a pre-existing psychotic disorder with a follow-up duration of at least 6 months. We used a combination of search terms for describing cannabis, the outcome of interest (relapse of psychosis), and the study population. We excluded studies if continued cannabis use or discontinued cannabis use could not be established. We compared relapse outcomes between those who continued (CC) or discontinued (DC) cannabis use or were non-users (NC). We used summary data (individual patient data were not sought out) to estimate Cohen's d, which was entered into random effects models (REM) to compare CC with NC, CC with DC, and DC with NC. Meta-regression and sensitivity analyses were used to address the issue of heterogeneity. Findings: Of 1903 citations identified, 24 studies (16 565 participants) met the inclusion criteria. Independent of the stage of illness, continued cannabis users had a greater increase in relapse of psychosis than did both non-users (dCC-NC=0.36, 95% CI 0.22-0.50, p<0.0001) and discontinued users (dCC-DC=0.28, 0.12-0.44, p=0.0005), as well as longer hospital admissions than non-users (dCC-NC=0.36, 0.13 to 0.58, p=0.02). By contrast, cannabis discontinuation was not associated with relapse (dDC-NC=0.02, -0.12 to 0.15; p=0.82). Meta-regression suggested greater effects of continued cannabis use than discontinued use on relapse (dCC-NC=0.36 vs dDC-NC=0.02, p=0.04), positive symptoms (dCC-NC=0.15 vs dDC-NC=-0.30, p=0.05) and level of functioning (dCC-NC=0.04 vs dDC-NC=-0.49, p=0.008) but not on negative symptoms (dCC-NC=-0.09 vs dDC-NC=-0.31, p=0.41). Interpretation: Continued cannabis use after onset of psychosis predicts adverse outcome, including higher relapse rates, longer hospital admissions, and more severe positive symptoms than for individuals who discontinue cannabis use and those who are non-users. These findings point to reductions in cannabis use as a crucial interventional target to improve outcome in patients with psychosis. Funding: UK National Institute of Health Research.