Frequently Asked Questions

Yes, current cannabis use was associated with a higher risk of early psychiatric rehospitalization, but not clearly with higher readmission risk across the full year after discharge. In adjusted analyses, cannabis use at the index admission was not associated with increased 12-month readmission risk (HR=1.6; 95% CI, 1.002.5). However, sensitivity analyses showed a stronger short-term association, with a 2.8-fold increase in readmissions during the first month and a significant association over the first 3 months (HR=1.9; 95% CI, 1.13.3), which was no longer significant by 6 months (HR=1.3; 95% CI, 0.82.2).

Cannabis use was very common. In this sample of 200 young adults hospitalized for bipolar or psychotic disorders, 58% had used cannabis during their lifetime and 43% were current cannabis users at the time of index hospitalization. Among current cannabis users, 67% used cannabis daily.

Substance exposure was broad in this population. Excluding tobacco and alcohol, 58% of patients had used a drug during their lifetime and 44% were using a drug at the time of index hospitalization. After cannabis, the next most common substances were cocaine (17% lifetime, 6% current) and ecstasy (15% lifetime, 5% current), and 13% of the sample had current multiple substance use.

More than half were tobacco smokers (53%), and half used alcohol (n=99), although most alcohol use was less than monthly.

Among the 187 patients included in the Cox regression analysis, 95 were readmitted to full-time psychiatric hospitalization within 1 year after discharge from the index hospitalization. The median time to readmission was 95 days, with an interquartile range of 15 to 349 days.

Absent or conflicted family relationships and prior psychiatric hospitalization were independently associated with readmission. In the adjusted Cox model, absent or conflicted family relationship had an HR of 2.2 (95% CI, 1.43.2), and psychiatric hospitalization history had an HR of 2.1 (95% CI, 1.43.2). The discussion also notes that other associated early-readmission risk factors included nonpsychiatric comorbidity.

The study suggests poorer adherence among cannabis users, but the data were incomplete. During the index hospitalization comparison, good adherence was reported more often in nonusers than users (52.7% vs 42.2%), while poor adherence was more common in current cannabis users (17.2% vs 9.5%). However, adherence data at readmission were often missing, and the authors stated that treatment adherence could not be included in the statistical analyses.

Substance use disorder appeared more common than specialized addiction treatment. In the full sample, 37% had a documented substance use disorder diagnosis, but only 12% were receiving specialized addiction care. The authors also noted substantial missing information in the medical records for both substance use disorder diagnosis (24%) and addiction care (33%).

Current cannabis use meant cannabis use reported at the time of the index hospital admission, regardless of frequency, quantity, or whether the patient had a current substance use disorder diagnosis. The study treated this as cannabis use before hospitalization based on the admission assessment documented in the medical record.

This was a historical cohort study based on electronic medical records from a single public psychiatric hospital in Bordeaux, France. It included the first 200 consecutive young adults aged 18 to 29 years who were hospitalized full-time for bipolar or psychotic disorders between January 1, 2017, and January 27, 2019, with follow-up through May 31, 2023.

Because the study was observational and based on chart review, it can identify associations but cannot establish that cannabis use caused clinical worsening or readmission. The authors also noted that missing data prevented inclusion of all relevant covariates in the analyses.

The main limitations were small sample size, missing data, possible underestimation of substance use, and single-center design. The authors stated that the limited number of patients may have reduced statistical power, especially for detecting a significant association over the full year after discharge. Substance use prevalence may have been underestimated because it depended on clinician documentation and patient self-report, and urinary toxicology screening was performed in only 30% of patients.

Additional limitations included the heterogeneous definition of current cannabis users, inability to analyze results by specific psychiatric diagnosis because of sample size, poor reporting of treatment adherence and specialized addiction treatment, and limited generalizability because the study came from one hospital.