Table 3.

Mediation analysis investigating association of history of cannabis use at presentation with clinical outcomes mediated by number of unique antipsychotics prescribed at 5-year follow-up (n=926)

Outcomes at 5-year follow-up	Natural direct effect (95% CI, p value)	Natural indirect effect (95% CI, p value)	Total effect (95% CI, p value)	Percentage of total effect mediated by number of unique antipsychotics
Number of admissions to hospital: incidence rate ratio	1.37 (1.12 to 1.68, p=0.002)	1.09 (1.01 to 1.18, p=0.03)	1.50 (1.21 to 1.87, p<0.001)	21.3
Compulsory hospital admission: OR	1.39 (0.68 to 2.83, p=0.37)	1.27 (1.03 to 1.58, p=0.03)	1.76 (0.81 to 3.84, p=0.15)	42.3
Number of days spent in hospital: β coefficient (days)	17.0 (−1.5 to 35.4, p=0.07)	17.9 (2.4 to 33.4, p=0.02)	34.8 (11.6 to 58.1, p=0.003)	51.4

Natural direct effect: cannabis use → outcome.

Natural indirect effect: cannabis use → number of unique antipsychotics → outcome.

Total effect: combined natural direct and natural indirect effect.