Application of trigger tools for detecting adverse drug events in older people: A systematic review and meta-analysis

Objective: To identify trigger tools applied to detect adverse drug events (ADEs) in older people and describe their utility and performance. Methods: A systematic review was conducted in the PubMed, Lilacs, and Scopus databases (January 2024). Studies that developed, applied, or validated trigger tools and evaluated their utility and/or performance for detecting ADEs in older people were considered. Direct proportion meta-analyses using the inverse-variance method were performed for prevalence of ADEs and positive predictive value (PPV). Results: Twenty-four studies (25 publications) were included. Twelve trigger tools were identified, of which six were developed for detecting ADEs in older population, four developed for general population and modified for older people, and two developed for general population. No tools for detecting ADEs in older people receiving palliative care or hospitalized in intensive or surgical care units were found. The performance of triggers was presented through PPV (11.5–71%), negative predictive values (83.3%), and sensitivity (30–94.8%). The overall PPV was 33.3% (95%CI: 32.5–34.2%). Triggers with good performance were changes in plasma levels of digoxin, glucose, and potassium; changes in international normalized ratio; abrupt medication stop; hypotension; and constipation. The prevalence of ADEs ranged from 2.8 to 66%, with overall prevalence of ADEs of 20% (95%CI: 19.3–20.8%). Preventability ranged from 8.4 to 94.4%. Metabolic or electrolyte disturbances induced by diuretics, constipation induced by opioids, and falls and delirium induced by benzodiazepines were the most prevalent ADEs. Conclusion: The trigger tools are flexible and easy to apply, and they can contribute to the detection of ADEs, their associated risk factors, the level of harm, and preventability in different health settings. However, there is no consensus on good or poor values of PPV, which indicate the performance of triggers. Furthermore, there is limited evidence regarding the evaluation of performance through negative predictive value, sensitivity, and specificity. PROSPERO: CRD42022379893.