Predicting the prognosis of primary and substance-associated psychoses using urine drug screens: A 5-year retrospective longitudinal study using medical records

Substance use is frequently implicated in psychosis. However, the prognostic assumptions guiding differential treatment decisions between substance-induced and primary/affective psychoses have received inconsistent support. Substance-induced psychosis research has often been limited by aggregating substances or using diagnosis as both a grouping variable and an outcome (i.e., diagnostic conversion). The current study examined medical records for 1379 patients for 5 years following their first presentation with psychosis after January 2013 to a regional medical center. Clinically relevant outcomes were compared across groups defined by the substances detected in patients' urine at their initial presentation. Initial drug screen results predicted the initial encounter duration, the likelihood of a future presentation with psychosis and a negative drug screen, and the total hospital days associated with psychosis over 5 years, but not patients' number of psychiatric hospitalizations. Cocaine alone and cannabis combined with stimulants exhibited the courses expected of substance-induced psychosis, including quick remission, infrequent recurrence with negative drug screens, and low healthcare utilization. Amphetamines and cannabis alone exhibited improved prognoses compared to those with negative initial screens, but more chronic courses than expected of substance-induced psychosis. Depressants exhibited similar courses to those with negative initial screens. Other polysubstance combinations displayed divergent outcomes. The chronicity of psychosis associated with certain substances suggests further examination of their course with protracted abstinence, response to psychosis-focused treatments, and the potentially diverse mechanisms by which use precipitates psychosis. Additionally, the dissimilar courses observed for distinct polysubstance combinations emphasize the importance of disambiguating these groups in future research.