Background: There has been a theoretical debate as to which retrospective record review method is the most valid, reliable, cost efficient and feasible for detecting adverse events. The aim of the present study was to evaluate the feasibility and capability of two common retrospective record review methods, the "Harvard Medical Practice Study" method and the "Global Trigger Tool" in detecting adverse events in adult orthopaedic inpatients.
Methods: We performed a three-stage structured retrospective record review process in a random sample of 350 orthopaedic admissions during 2009 at a Swedish university hospital. Two teams comprised each of a registered nurse and two physicians were assigned, one to each method. All records were primarily reviewed by registered nurses. Records containing a potential adverse event were forwarded to physicians for review in stage 2. Physicians made an independent review regarding, for example, healthcare causation, preventability and severity. In the third review stage all adverse events that were found with the two methods together were compared and all discrepancies after review stage 2 were analysed. Events that had not been identified by one of the methods in the first two review stages were reviewed by the respective physicians.
Results: Altogether, 160 different adverse events were identified in 105 (30.0%) of the 350 records with both methods combined. The "Harvard Medical Practice Study" method identified 155 of the 160 (96.9%, 95% CI: 92.9-99.0) adverse events in 104 (29.7%) records compared with 137 (85.6%, 95% CI: 79.2-90.7) adverse events in 98 (28.0%) records using the "Global Trigger Tool". Adverse events "causing harm without permanent disability" accounted for most of the observed difference. The overall positive predictive value for criteria and triggers using the "Harvard Medical Practice Study" method and the "Global Trigger Tool" was 40.3% and 30.4%, respectively.
Conclusions: More adverse events were identified using the "Harvard Medical Practice Study" method than using the "Global Trigger Tool". Differences in review methodology, perception of less severe adverse events and context knowledge may explain the observed difference between two expert review teams in the detection of adverse events.