Substance-induced psychotic disorders, more colloquially referred to as drug-induced psychoses, may represent up to one-quarter of first hospitalizations for psychosis. These persons are often excluded from early psychosis studies, which limits evidence on their prevalence, clinical course, and outcomes.
Many people with substance-induced psychoses will later transition to a diagnosis of schizophrenia, but estimates vary widely between early psychosis services and population-based registers. A recent review found that 21% of people with first-episode substance-induced psychosis later received a diagnosis of schizophrenia or schizoaffective disorder.
Murrie and colleagues synthesized the results of longitudinal observations studies of transition from substance-induced psychosis to schizophrenia, using transition from other brief and atypical psychoses as a comparison group. They also aimed to investigate potential moderators of transition risk. The authors searched PsycINFO, MEDLINE, and Embase for studies from 1980-2018 that reported follow-up diagnoses in people with substance-induced psychosis, brief psychosis, atypical psychosis, schizophreniform psychosis, and psychosis NOS.
Inclusion criteria were: 1) a baseline diagnosis of substance-induced, brief, atypical, NOS, or schizophreniform psychosis, 2) a follow-up diagnosis with a minimum follow-up period of 6 months, and 3) the number of people with schizophrenia at the follow-up assessment. Studies defining psychosis with symptom scales or self-report were excluded.
The primary outcome was the proportion of people with a follow-up diagnosis of schizophrenia. Potential moderating variables included service setting, country, urban versus rural, age, sex, diagnostic system, diagnostic method, dropout rate, symptom severity scale scores, global assessment of function, cohorts limited to first-episode psychosis, year of follow-up, and use of toxicology to establish follow-up.
The authors first performed a meta-analysis of substance-induced psychoses compared with brief and atypical psychoses. Secondly, they performed separate meta-analyses of substance type (eg, stimulants, hallucinogens, cannabis) as a separate subgroup. Subgroup analyses for study-level data were also performed to investigate potential moderators of the primary outcome.