Background and aims: Substance-induced psychoses (SIP) have the potential to relapse and convert into schizophrenia-spectrum disorders. However, risk factors associated with SIP relapse remain unknown. The aim of this study was to measure the incidence and risk of SIP relapse and associating risk factors.

Design, setting and participants: Population based register study that identified people with first-time SIP between 2006 and 2016 (n = 7320) from Swedish nation-wide registers to examine incidence of relapse and associated risk factors during 2-year follow-up. Participants were censored to death, emigration and the diagnosis of other psychotic illness. Risk factors associated with relapse were studied using multivariable Cox models.

Measurements: SIP was measured via related diagnostic codes (ICD-10: F1x.5) collected from the National Patient Register (NPR). SIP relapse was measured as hospitalization due to SIP during 2-year follow-up also collected from the NPR. Potential risk factors included demographic characteristics, psychiatric comorbidities, sickness absence and disability pension collected from nationwide registers.

Findings: Of the study population (n = 7320), 20.0% (n = 1463) had a SIP relapse during the follow-up (median time 126 days, interquartile range 56-321) and 83.3% had the same type of SIP as their first SIP. Relapse was most common for those whose first SIP was induced by cannabis (25.7%), followed by multi-substance use (23.8%) and (meth)amphetamine (19.7%). Factors associated with SIP relapse were previous substance use disorder [hazard ratio (HR) = 1.37, 95% confidence interval (CI) = 1.20-1.56], younger age (16-29 years, HR = 1.29, 95% CI = 1.05-1.58, versus 50-65), being born abroad (HR = 1.23, 95% CI = 1.07-1.41), attention deficit hyperactivity disorder (HR = 1.21, 95% CI = 1.05-1.39), having had 1-90 days sick leave during the previous year (HR = 1.19, 95% CI = 1.01-1.44), and cannabis- (HR = 2.42, 95% CI = 1.98-2.96), (meth)amphetamine- (HR = 1.49, 95% CI = 1.23-1.81) or multi-substance- (HR = 1.81, 95% CI = 1.52-2.15) induced psychosis compared with alcohol-induced psychosis.

Conclusions: In Sweden, 20% of people with substance-induced psychosis between 2006 and 2016 had a relapse within 2 years follow-up. Cannabis-induced psychosis had the shortest time lapse between episodes. Risk factors for relapse included attention deficit hyperactivity disorder, substance-use disorder, younger age, previous sickness absence and being born outside Sweden.