Perioperative Mortality, 2010 to 2014: A Retrospective Cohort Study Using the National Anesthesia Clinical Outcomes Registry.

Background: The National Anesthesia Clinical Outcomes Registry collects demographic and outcome data from anesthesia cases, with the goal of improving safety and quality across the specialty. The authors present a preliminary analysis of the National Anesthesia Clinical Outcomes Registry database focusing on the rates of and associations with perioperative mortality (within 48 h of anesthesia induction).
Methods: The authors retrospectively analyzed 2,948,842 cases performed between January 1, 2010, and May 31, 2014. Cases without procedure information and vaginal deliveries were excluded. Mortality and other outcomes were reported by the anesthesia provider. Hierarchical logistic regression was performed on cases with complete information for patient age group, sex, American Society of Anesthesiologists physical status, emergency case status, time of day, and surgery type, controlling for random effects within anesthesia practices.
Results: The final analysis included 2,866,141 cases and 944 deaths (crude mortality rate, 33 per 100,000). Increasing American Society of Anesthesiologists physical status, emergency case status, cases beginning between 4:00 PM and 6:59 AM, and patient age less than 1 yr or greater than or equal to 65 yr were independently associated with higher perioperative mortality. A post hoc subgroup analysis of 279,154 patients limited to 22 elective case types, post hoc models incorporating either more granular estimate of surgical risk or work relative value units, and a post hoc propensity score-matched cohort confirmed the association with time of day.
Conclusions: Several factors were associated with increased perioperative mortality. A case start time after 4:00 PM was associated with an adjusted odds ratio of 1.64 (95% CI, 1.22 to 2.21) for perioperative death, which suggests a potentially modifiable target for perioperative risk reduction. Limitations of this study include nonstandardized mortality reporting and limited ability to adjust for missing data.