Comparing the performance of multiple trigger tools in identifying medication-related hospital readmissions.

Background: Multiple trigger tools have been developed to identify medication-related hospital (re)admissions (MRRs); however, the accuracy of these tools in real-world clinical practice is uncertain. The objective of this study was to compare the accuracy of four different trigger tools (OPERAM, STOPP/START criteria, ADR-tool, and QUADRAT) to identify MRRs compared with clinical adjudication.

Methods: We conducted a secondary analysis of patients readmitted within 30 days to seven departments of a teaching hospital. In the primary study, which involved a retrospective chart review of 1111 readmissions, MRRs and their potential preventability were clinically adjudicated by physicians and pharmacists. In the current study, four trigger tools were applied by a different physician and pharmacist panel. Patients of all ages were included. Trigger tools included both explicit items specifying the event and the associated medication and implicit items requiring clinical knowledge. The accuracy of each trigger tool was assessed by calculating the proportion of clinically adjudicated MRRs each tool identified overall as well as according to explicit and implicit triggers separately. The accuracy of each tool to identify potentially preventable MRRs was also calculated.

Results: Of 1111 readmissions, 181 were adjudicated as medication-related (mean age 69 years, 56% male); 72 (40%) MRRs were potentially preventable. The original OPERAM tool identified 166 (92%) MRRs (62% through explicit triggers). The STOPP/START criteria identified 23 (13%, 7% through explicit triggers), the ADR tool identified 51 (28%, all explicit triggers), and the QUADRAT tool identified 76 (42%; all explicit triggers) MRRs. Of the 72 potentially preventable MRRs, OPERAM identified 59 (82%), STOPP/START identified 18 (25%), ADR identified 20 (28%), and QUADRAT identified 21 (29%).

Conclusion: The original OPERAM tool identified the highest proportion of (preventable) MRRs. However, this tool includes many implicit triggers requiring expert clinical knowledge. Future studies should assess the practicality of implementing this tool in daily practice.