Risk of Adverse Surgical Outcomes in Patients with Recent Covid-19 Infection: An Emulated Target Trial
William O’Brien; Kalpana Gupta; Kamal Itani
Background:The American Society of Anesthesiologists published recommendations in December, 2020 on the timing of surgery in recently recovered Covid-19 patients. There are several recent studies about vaccination, infection, and timing of surgery, with conflicting findings. In this study we emulated a trial within a large national Veteran population. Our goal was to assess whether Covid-19 infection before surgery is associated with risk of adverse postoperative outcomes.
Hypothesis:Patients with recent infection in the 30 or 60 days before surgery face increased risk compared with those without infection.
Methods:Study design was a target trial emulation with pseudo-randomization to control (without recent infection), vs 2 mutually exclusive exposures (1st chronological infection in preoperative days 1-30 or 31-60). Eligibility was major surgery during January 1, 2021 – September 30, 2021 reviewed by the Veterans Affairs Surgical Quality Improvement Program.
Time zero for pseudo-randomization was day 60 before surgery. A propensity score model with inverse weighting balanced characteristics across exposure groups. Propensity score covariates were factors known at time zero that would be predictive of exposure, including Covid-19 vaccination status.
The endpoint was any of the following within 30 days after surgery: death, cardiac events, central nervous system outcomes, respiratory outcomes, surgical infection, or thromboembolic events. Weighted logistic regression estimated the odds ratio (OR) of any outcome as a function of exposure group, perioperative factors (since they were not known at time zero), and any covariates with standardized mean difference > 0.2 after weighting.
Results:The overall study population was 29,072 surgeries, and characteristics across exposure groups were well-balanced and typical of VA population studies. Outcomes occurred in 1,337 (4.7%) within the no-infection group, 18 (7.6%) within the preoperative day 1-30 infection group, and 7 (3.2%) within the 31-60d infection group. The OR (95% confidence interval) for postoperative outcomes was 1.49 (0.84-2.46) for patients with infection in preoperative days 1-30 and 0.70 (0.28-1.43) for infection in preoperative days 31-60.
Conclusions:In this large population of Veterans undergoing surgery during the initial 9 months of vaccine availability, we found that recent Covid-19 infection was not associated with a significantly higher risk of adverse outcomes, compared with those not recently infected.