Impact of hospital volume on patient safety indicators and failure to rescue following open aortic aneurysm repair

AbstractObjectiveFailure to rescue (FTR), a patient safety indicator (PSI) defined, codified, and adjudicated by the Agency for Healthcare Research and Quality, is classified as a preventable inpatient death following major complications. FTR has been reported to be a significant driver of postoperative mortality after open abdominal aortic aneurysm (OAAA) repair. The association between hospital volume (HV) and mortality is well known; however, the mechanisms responsible for these improved outcomes and relative contribution to observed interhospital variation is poorly understood. Similarly, HV influence on specific complications predictive of FTR is unknown; therefore, we sought to determine how HV influences risk and contributes to interhospital variation in PSI events leading to FTR and/or in-hospital mortality after OAAA repair. MethodsThe Vizient database (174 academic/nonacademic hospitals) was queried for all OAAA repairs (elective, n = 2827; nonelective, n = 1622) completed from 2012 to 2014. The primary endpoint was combined FTR and/or in-hospital 30-day mortality. Risk-adjusted rates of complications, Agency for Healthcare Research and Quality-designated PSIs, and FTR were determined. Additional modeling identified PSIs associated with FTR, whereas HV effects on PSIs and FTR were evaluated using mixed-effect models accounting for interhospital variation. Proportion of variation attributable to HV was estimated by contrasting hospital random effect variances in the presence/absence of volume effects. ResultsThe combined overall FTR/in-hospital 30-day mortality rate was 9.3% (n = 414). For elective and nonelective cases, the overall FTR and 30-day mortality rates were: FTR, 1.6%, 4.9%; and 30-day in-hospital mortality, 3.4%, 17.5%, respectively. HV significantly influenced FTR/30-day in-hospital mortality ( P < .0001). FTR/30-day mortality odds for hospitals with 3-year volumes of 50, 100, 150, and 200 cases were 1.4, 2.0, 2.7, and 3.0 times lower, respectively, than hospitals performing ≤25 cases/3 years. The proportion of interhospital variation attributed to HV was greatest for FTR/30-day mortality (62%). Procedural volume accounted for 41% and 38% of interhospital variation in postoperative bleeding and myocardial infarction, respectively. Preoperative predictors of FTR included coagulopathy, arrhythmia (nonelective cases); congestive heart failure, obesity (elective cases); and age, neurological disease, hypertension, and valvular disease (al