The adverse drug reaction risk in older persons (ADRROP) prediction scale: derivation and prospective validation of an ADR risk assessment tool in older multi-morbid patients

Abstract

Background: Adverse drug reactions (ADRs) cause serious morbidity and mortality in multi-morbid older adults. Reliable ADR risk prediction would improve patient safety in this at-risk population. We aimed to derive and validate a new predictive tool for assessing ADR Risk in Older People (acronym ADRROP). Design: We combined four databases describing 2217 older people hospitalized with acute illness in order to determine risk factor variables significantly associated with ADRs. We identified the independent ADR risk factors from 1687 patients (derivation cohort) and used them to construct the ADRROP scale. We prospectively validated ADRROP using data from 530 patients (validation cohort). We applied area under the curve (AUC) analysis to test ADRROP’s ADR predictive power. We also compared ADRROP’s performance to the GerontoNet ADR risk scale. Results: Eight independent ADR risk factors were identified in the derivation patient cohort: female gender, age > 70 years, estimated GFR < 30 ml/min/1.73 m2, assistance required for ≥ 1 daily activity, ≥ 4 co-morbidities, liver disease, presence and number of STOPP criteria-defined potentially inappropriate medications, and ≥ 1 fall in the previous year. The ADRROP predictive scale constructed from these combined variables ranged from 0 to 27. The derivation cohort AUC value was 0.623 (95% CI 0.598–0.665); the validation cohort AUC was 0.592 (95% CI 0.532–0.652). Applying the GerontoNet ADR risk scale to the combined cohorts yielded an AUC of 0.566 (95% CI 0.537–0.596). Conclusions and relevance: Neither ADRROP nor the GerontoNet ADR risk scale predicted ADRs to a high level in hospitalized older people with multi-morbidity.