Predicting discontinuation of biologic therapy caused by adverse events in psoriasis patients—A Danish nationwide cohort study

Abstract

Abstract Background Some patients treated with biologic therapy for psoriasis experience adverse events causing treatment discontinuation. Objectives To identify predictors of adverse events resulting in discontinuation of biologics for psoriasis. Methods This cohort study involved data from the Danish nationwide registry DERMBIO. A classification algorithm based on gradient‐boosted decision trees combined with a mixed‐effects model was applied to data on various drugs to predict treatment discontinuation caused by adverse events and to identify important predictors. Results This study comprised 4876 treatment series, including 3078 (63.1%) that were discontinued. With adverse events causing 19.6% of these, it was the second most frequent reason for discontinuation following ineffectiveness. The proportion of treatments discontinued due to adverse events was 20.0% [16.7%, 23.9%; 95% confidence interval (CI)] among patients that had discontinued their previous treatment series due to adverse events, and 11.6% [10.3%, 13.1%; 95% CI] among treatment series where the previous treatment was discontinued due to other reasons (p < 0.001). Conclusions The proportion of treatment series discontinued due to adverse events among patients who had discontinued their previous treatment due to adverse events was 72.4% larger compared to those who discontinued their previous treatment for other reasons.

Keywords

• adverse events
• biologic treatment
• machine learning
• prediction models
• psoriasis