Abstract P107: High-throughput Screening Detects Thiazide-induced Prescribing Cascades.

Abstract

Introduction: Thiazide diuretics are effective but can cause adverse effects that prompt the use of additional therapy, i.e., a prescribing cascade. We aimed to identify potential thiazide-induced prescribing cascades using high throughput sequence symmetry analysis. Methods: Using claims from 5% (2011-15) and 15% (2016-20) samples of Medicare fee-for-service beneficiaries we identified new thiazide users aged ≥66 years who had continuous enrollment for ≥360 days pre- and ≥180 days post-thiazide initiation. We screened for the initiation of 446 other ‘marker’ drug classes (based on WHO Anatomical Therapeutic Classification level 4 codes) within ±90 days of thiazide initiation, generating sequence ratios (SRs) representing the proportion of thiazide initiators starting the marker class after, compared to before, the thiazide. Adjusted SRs (aSRs), accounting for prescribing trends over time, were calculated with 95% CIs >1 suggesting a prescribing cascade signal. Results: We identified unique 189,971 thiazide initiators (mean ± SD age 76.2 ± 7.4 years, 66% women, 90% with hypertension). Among 446 ‘marker’ classes analyzed, we identified 21 statistically significant prescribing cascade signals (Figure). The 2 strongest prescribing cascade signals were carbohydrates and electrolyte solutions. Conclusion: Among 446 classes, we identified 21 possible prescribing cascades in our Medicare population, which may impact patient care outcomes. Further investigation is required to determine the biological plausibility of these signals.