The Association of Guideline-Directed Prophylaxis With Postoperative Nausea and Vomiting in Adult Patients: A Single-Center, Retrospective Cohort Study.

Abstract

Consensus guidelines for postoperative nausea and vomiting (PONV) prophylaxis recommend a risk-based approach in which the number of antiemetics administered is based on a preoperative estimate of PONV risk. These guidelines have been adapted by the Multicenter Perioperative Outcomes Group (MPOG) to serve as measures of clinician and hospital compliance with guideline-recommended care. However, the impact of this approach on clinical outcomes is not known. We performed a single-center, retrospective study of adult patients undergoing general anesthesia from 2018 to 2021. Risk factors for PONV were defined using MPOG definitions: female sex, history of PONV or motion sickness, nonsmoker, inhaled anesthesia >60 minutes, high-risk procedure (cholecystectomy, laparoscopic, gynecologic), and age <50 years. Adequate prophylaxis was defined using the MPOG PONV-05 metric: at least 2 agents for patients with 1 to 2 risk factors and at least 3 agents for patients with 3+ risk factors. PONV was defined as documented PONV or receipt of rescue antiemetics. To estimate the association between adequate prophylaxis and PONV, we used Bayesian binomial models with overlap propensity score weighting. We included 76,703 cases (43% receiving adequate prophylaxis) with PONV occurring in 19%. In unadjusted and unweighted comparison, adequate prophylaxis was associated with increased incidence of PONV: median odds ratio 1.21 (95% credible interval [1.16-1.25]). However, after propensity score weighting and multivariable adjustment, adequate prophylaxis was associated with reduced relative and absolute risk for PONV: weighted marginal median odds ratio 0.90 [0.84-0.98] and absolute risk reduction (ARR) 1.6% [0.6%-2.6%]. There was evidence for a differential effect of adequate prophylaxis across the guideline-defined risk spectrum, with benefit seen in patients with 1 to 5 risk factors (conditional probabilities of benefit >0.81), but not in those at high predicted risk. Patient-specific, covariate-adjusted ARR was heterogeneous, with a median patient-specific conditional probability of benefit of 0.84 (95% credible interval, 0.73-0.90). Guideline-directed PONV prophylaxis is associated with a modest reduction in PONV, although this effect is small and heterogeneous on the absolute scale. We found evidence for a differential association between adequate prophylaxis and PONV across the guideline-defined risk spectrum, with diminution in patients at very high predicted preoperative risk. While patient-specific benefit was heterogenous, most patients had reasonably high predicted probabilities of absolute benefit from a guideline-directed strategy. Further assessment of these associations in a multicenter setting, with more robust investigation of risk prediction methods will allow for better understanding of the optimal approach to PONV prophylaxis.