LAUSR

Objective: To examine the extent to which multiple, sequential complications impacts variation in institutional postoperative mortality rates. Background: Failure to rescue (FTR) has been proposed as an underlying factor in hospital variation in surgical mortality. However, little is currently known about hospital variation in FTR after multiple complications or the contribution of sequential complications to variation. Methods: Retrospective cohort study of 266,101 patients within the Veterans Affairs Surgical Quality Improvement Program (2000–2014) who underwent a subset of high-mortality inpatient general, vascular, or thoracic procedures. The association between number of postoperative complications (0, 1, 2, or ≥3) and 30-day mortality across quintiles of hospital risk-adjusted mortality was evaluated with multivariable, multilevel mixed-effects models. Results: Among patients who had a complication, over half (60.9%) had 1, but those with more than 1 accounted for the majority of the deaths (63.1%). Across hospital quintiles, there were no differences in complications (23.5% very low mortality vs 23.6% very high mortality; trend test P = 0.15). FTR increased significantly (12.0% vs 18.1%; trend test P < 0.001) with an incremental impact as complications accrued (6.7% 1 complication vs 26.1% ≥3, lowest quintile; 11.7% 1 complication vs 33.0% ≥3, highest quintile). However, the risk of FTR associated with increasing complications remained relatively constant across hospital quintiles and was not explained by differences in patients presenting with multiple complications on the index complicated day. Conclusions: FTR occurs predominantly among patients who have more than 1 complication with a dose-response relationship as complications accrue. As this dose-response relationship is observed across hospitals, surgical quality improvement efforts may benefit by shifting focus to broader interventions designed to prevent subsequent complications at all hospitals.