This study evaluates the accuracy of different adherence assessment methods in first-episode psychosis (FEP), where long-term adherence is essential for relapse prevention. We compared self-reported adherence, pharmacy refill data, and therapeutic drug monitoring (TDM).

In this one-year follow-up study, 78 FEP patients were assessed for adherence at two and twelve months using the Attitudes towards Neuroleptic Treatment (ANT) scale (self-report), pharmacy refill data (≥1 or ≥2 purchases in four months), and TDM as the reference standard. Statistical analyses included McNemar’s test, sensitivity, specificity, Cohen’s kappa, and Receiver Operating Characteristic (ROC) analysis.

At two months, adherence was 50.0% based on TDM, decreasing to 41.5% at twelve months. At two months, adherence rates were 73.3% for the ANT scale, 84.6% for at least one pharmacy refill, and 55.8% for at least two refills; by twelve months, these were 68.8%, 91.2%, and 52.9%, respectively. The ANT-attitude variable had weak predictive value for adherence (AUC: 0.607 at two months, 0.671 at twelve months). The ANT scale showed high sensitivity but low specificity, leading to adherence overestimation. Pharmacy refill adherence was more reliable, particularly when defined as at least two purchases within four months.

Medication non-adherence is common in FEP. Pharmacy refill data provided a more accurate adherence measure than self-report. Enhancing adherence requires psychoeducation, follow-up, and proactive monitoring. Measuring drug concentrations after hospital discharge could help detect early non-adherence and optimize treatment.