Purpose Voluntary event reporting systems continue to be the most common method used to identify adverse events in most US hospitals; however, this method fails to capture more than 90% of adverse drug events (ADEs). The purpose of this study is to examine which medication-related triggers have the highest positive predictive values (PPV) for detecting ADEs at a large academic medical centre.

Methods A 1-year, single-centre, retrospective quality improvement study was conducted to assess the PPV of four medication-related triggers: flumazenil, naloxone, glucose <70 mg/dL or dextrose 50%. Retrospective chart review was conducted on a random sample of eligible patients to establish if an ADE occurred and determine its preventability. Assessed triggers were also compared against the hospital’s voluntary event reporting system to determine whether the events were previously reported.

Results A total of 161 triggers were reviewed. PPV values for detection of ADEs were 0.55, 0.58, 0.76 and 0.68 for flumazenil, naloxone, glucose <70 mg/dL and dextrose 50%, respectively. PPV values for detection of preventable ADEs were 0.09, 0.16, 0.32 and 0.34 for flumazenil, naloxone, glucose <70 mg/dL and dextrose 50%, respectively. Of the 107 ADEs identified, three events were reported through the hospital’s voluntary event reporting system (2.8%).

Conclusions Trigger tools successfully detected both preventable and non-preventable ADEs. Events detected using trigger tools are unlikely to be reported through voluntary event reporting systems; therefore, trigger tools can serve as a useful adjunct for adverse event detection.