Alderson · Psychological medicine 2017 · retrospective population-based cohort study · n=3486

Risk of transition to schizophrenia following first admission with substance-induced psychotic disorder: a population-based longitudinal cohort study.

Level 3 - non-randomized controlled study

Retrospective population-based longitudinal cohort study

PubMed 28464965 · doi:10.1017/S0033291717001118 · record verified 2026-08-26

What was done

Researchers used Scottish Morbidity Record data to follow 3,486 patients from their first psychiatric hospital admission for substance-induced psychotic disorder (SIPD; ICD-10 F10–F19.5) between January 1997 and July 2012. Patients were followed for up to 15.5 years (range: 1 day to 15.5 years) until a first diagnosis of schizophrenia (ICD-10 F20) or study end to assess transition rates, time to conversion, and associated risk factors.

What was found

The 15.5-year cumulative hazard rate for a subsequent schizophrenia diagnosis was 17.3% (standard error = 0.007). Hazard rates were similar across cannabis-, stimulant-, opiate-, and polydrug-induced psychoses. While the mean time to transition across the cohort was approximately 13 years, more than 50% of transitions occurred within 2 years and over 80% within 5 years of initial SIPD diagnosis. Male sex, younger age, and an initial hospital stay longer than 2 weeks were identified as risk factors for transition.

Why it matters

These findings show that nearly one in six patients hospitalized for substance-induced psychosis are eventually diagnosed with schizophrenia, highlighting the need for prolonged clinical monitoring—particularly during the first two to five years post-admission.

Limits

The study is restricted to hospitalized cases recorded in Scottish psychiatric registries, excluding individuals managed purely in outpatient or community settings. Findings rely on administrative ICD-10 diagnostic coding without standardized clinical re-interviews, and specific substance dosages, frequency of ongoing drug use, and potential unmeasured confounders were not detailed in the abstract.

Cited by