Transition of Substance-Induced, Brief, and Atypical Psychoses to Schizophrenia

A Systematic Review and Meta-analysis

Benjamin Murrie; Julia Lappin; Matthew Large; Grant Sara


Schizophr Bull. 2020;46(3):505-516. 

In This Article

Abstract and Introduction


Some people who experience substance-induced psychosis later develop an enduring psychotic disorder such as schizophrenia. This study examines the proportion of people with substance-induced psychoses who transition to schizophrenia, compares this to other brief and atypical psychoses, and examines moderators of this risk. A search of MEDLINE, PsychINFO, and Embase identified 50 eligible studies, providing 79 estimates of transition to schizophrenia among 40 783 people, including 25 studies providing 43 substance-specific estimates in 34 244 people. The pooled proportion of transition from substance-induced psychosis to schizophrenia was 25% (95% CI 18%–35%), compared with 36% (95% CI 30%–43%) for brief, atypical and not otherwise specified psychoses. Type of substance was the primary predictor of transition from drug-induced psychosis to schizophrenia, with highest rates associated with cannabis (6 studies, 34%, CI 25%–46%), hallucinogens (3 studies, 26%, CI 14%–43%) and amphetamines (5 studies, 22%, CI 14%–34%). Lower rates were reported for opioid (12%), alcohol (10%) and sedative (9%) induced psychoses. Transition rates were slightly lower in older cohorts but were not affected by sex, country of the study, hospital or community location, urban or rural setting, diagnostic methods, or duration of follow-up. Substance-induced psychoses associated with cannabis, hallucinogens, and amphetamines have a substantial risk of transition to schizophrenia and should be a focus for assertive psychiatric intervention.


Substance-induced psychotic disorders, sometimes called drug-induced psychoses, are brief psychotic syndromes triggered by substance use and persisting for days or weeks after substance intoxication has resolved.[1] They are common disorders: estimates of their incidence range from 1.5[2] to 6.5[3] per 100 000 person-years, similar to estimated incidence rates for affective psychoses and bipolar disorder (4.6 and 6.1 episodes per 100 000, respectively).[4] Up to 25% of first hospital admissions for psychosis may include a diagnosis of substance-induced psychosis.[5] In high-risk populations, such as amphetamine users, their prevalence may exceed 40%.[6] Despite this, debate continues about the overlap of substance-induced psychoses with other brief and atypical psychoses, and the validity and reliablility of their diagnostic criteria.[7,8] People with substance-induced psychoses are often excluded from studies of early psychosis,[9] limiting the evidence on prevalence, course, and outcomes that is required to guide the management and treatment of these conditions.[10]

A significant proportion of people with substance-induced psychosis later transition to a diagnosis of schizophrenia. Estimates of this proportion vary widely. Studies of treatment cohorts from early psychosis services have reported probabilities of transition as high as 44%[11] and 66%.[12] Some of these studies found that the probability of transition to schizophrenia was highest in people with cannabis-[12] or amphetamine-induced psychoses.[13] However, estimates derived from treatment cohorts may be increased because people with enduring disorders may be more likely to remain in contact with services. Early psychosis services may also be more likely to see young people who have high rates of substance use, increasing the rate of apparent transition by chance.[14]

Population-based registers may provide a more accurate estimate of the probability of transition than studies of treatment cohorts because of better follow-up and more representative sampling. Studies of national register data from Sweden, Denmark, and Finland have reported proportions of transition from substance-induced psychosis to schizophrenia ranging from 6% to 17%.[15–17] However, these lower proportions may also reflect the different diagnostic mix captured by registry data compared to clinical cohorts. In several registry studies alcohol-induced psychosis was the most common subtype of substance-induced psychosis, and had a lower probability of transition to schizophrenia.[16,17] Estimates of transition might also vary for other reasons including differences in study design, patient populations, and health care settings.

A meta-analytically derived estimate of transition from substance-induced psychosis to schizophrenia has been provided by a recent review of transition in first-episode psychosis.[18] This study found that 21% of people with first-episode substance-induced psychosis received a later diagnosis of schizophrenia or schizoaffective disorder, based on 10 studies and 164 subjects. The broader focus of that review meant that it could not examine whether substance type or other factors predicted transition to schizophrenia in substance-induced psychoses.

The primary aim of the current study was to synthesize the results of longitudinal observational studies of transition from substance-induced psychosis to schizophrenia. Studies of transition from other brief and atypical psychoses were also examined as a comparison group. These were included to reflect the complex and heterogeneous nature of presentations to early psychosis and other clinical services, and because many people with these diagnoses also transition to later diagnosis of schizophrenia.[10,18] We hypothesized that substance-induced psychosis would be associated with the same risk for transition to a later diagnosis of schizophrenia as is observed in other brief and atypical psychoses, based on the findings of the clinical follow-up and register studies described above.

The secondary aim of the study was to examine potential moderators of the risk for transition to schizophrenia. Several studies have found that cannabis-associated psychoses have a greater risk of transition to schizophrenia than other substance-related psychoses.[16,17,19] Other potential moderators of prognosis in early psychosis include male gender,[20,21] urban location,[21–23] age at onset,[24] duration of untreated psychosis,[25,26] symptom profile[27] and the ongoing use of cannabis or other substances following the index psychosis episode.[28] Methodological issues such as diagnostic criteria,[29] diagnostic methods, follow-up periods, or completeness of follow-up could also potentially influence study findings.