Preventable in-hospital deaths due to adverse events: a retrospective case record review study

Abstract Background There is no single parameter that describes the level of patient safety in hospitals. Among options, retrospective case record review (RCRR) is an established method for analysing in-hospital deaths to determine the percentage that were preventable. The assessment of preventability can vary considerably between reviewers. Current literature indicates that 0.5% − 8.4% of in-hospital deaths are preventable. This descriptive study reports the magnitude of preventable in-hospital deaths using RCRR. Methods All in-hospital deaths ( n = 1081) from one year (2014) at Oslo University Hospital were, throughout eight years, reviewed by the same physician using a Mortality Analysis form derived from The Institute for Healthcare Improvement reviews of 50 consecutive deaths. A random subset of 20% of the cases were initially also reviewed by a patient safety specialist. The prevalence of adverse events and the rate of deaths that could have been prevented were evaluated with descriptive statistics. Results Out of 1078 reviewable in-hospital deaths 32 (3%) were assessed to have 50% or more probability of being prevented. Twelve of these (37.5%) had been reported to the hospital reporting system. In 19 of the cases an adverse event was considered the direct cause of death, most often during or after surgical treatment. In the random subset assessed by two reviewers, 12 (6%) and 11 (5.5%) cases were found preventable. Overall interrater agreement was fair (Kappa 0.4). The proportion of preventable deaths found by the single physician reviewer in the remaining 2014-cohort significantly decreased throughout an eight year-long review time from 5.5% in the first to 1.0% in the last of four review periods ( p < 0.001). Conclusions This study from a large university hospital assessed 3% of in-hospital deaths from one year as preventable. Using a standardized RCRR-form, the proportion was in accordance with rates in comparable studies. The variation in assessment of preventability seen over an extended review time suggests the use of RCRR to evaluate preventable deaths for benchmarking purposes might reflect reviewer differences more than hospital differences.