The fifteen most common antipsychotic monotherapies were analyzed on drug and dose category levels. Time periods of specific dose categories were compared with time periods of nonuse of any antipsychotics for relapse risk. In sensitivity analyses, the reference was the most commonly used drug and dose category, being oral olanzapine >1.6 DDDs/day in this cohort (most common in terms of both the number of users and person-years of use). In addition, sensitivity analyses where doses of each specific antipsychotic were compared with standard dose of that antipsychotic.
To reduce chances of confounding by indication, the main analyses were conducted in a within-individual design where all comparisons are conducted within the same individual, ie, each individual acts as his or her own control in a stratified Cox model. Time was reset to zero after each relapse. The impact of time-invariant factors, such as genetics and initial severity of the illness, are eliminated by the design, and we adjusted for time-varying factors, which were time since cohort entry, temporal order of specific antipsychotics, and use of antidepressants, mood stabilizers, benzodiazepines, and related drugs. Sensitivity analyses were conducted by censoring the first 30 days from all dose categories, as the full therapeutic effect is usually not reached immediately, especially when the dose has to be titrated up slowly. In these analyses, censoring was equally applied to nonuse periods. Sensitivity analyses also included analyses stratified by baseline age (≤45 vs >45 years), schizoaffective diagnoses (where schizoaffective disorder was defined as ICD-10 F25 and schizophrenia F20), and among first-episode patients.
As only persons with variation in exposure and outcome event (relapse) contribute to the within-individual comparisons, also traditional between-individual Cox models including all patients were conducted. Between-individual models were adjusted for age, sex, temporal order of treatment, previous number of psychiatric rehospitalizations, calendar year, use of antidepressants, benzodiazepines, and related drugs, anticholinergic antiparkinson drugs and statins, diagnosis of cardiovascular disease, diabetes, cancer, asthma/COPD, substance abuse, suicide attempt, liver disease, and renal disease. The results are reported as adjusted HRs (aHRs) with 95% confidence intervals (CIs). The significance of differences between aHRs was tested by comparing two hazard ratios from their betas and standard errors as computed by a Cox model, and tested by Student t test for dependent samples (R-package survcomp, function hr.comp, version 1.44).
Permissions for this study were granted by pertinent institutional authorities at the National Institute for Health and Welfare of Finland, The Social Insurance Institution of Finland, and Statistics Finland. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Schizophr Bull. 2022;48(4):774-784. © 2022 Oxford University Press