Hospital costs associated with intraoperative hypotension among non-cardiac surgical patients in the US: a simulation model.

Recent studies indicate intraoperative hypotension, common in non-cardiac surgical patients, is associated with myocardial injury, acute kidney injury, and mortality. This study extends on these findings by quantifying the association between intraoperative hypotension and hospital expenditures in the US. Monte Carlo simulations (10,000 trial per simulation) based on current epidemiological and cost outcomes literature were developed for both acute kidney injury (AKI) and myocardial injury in non-cardiac surgery (MINS). For AKI, three models with different epidemiological assumptions (two models based on observational studies and one model based on a randomized control trial [RCT]) estimate the marginal probability of AKI conditional on intraoperative hypotension status. Similar models are also developed for MINS (except for the RCT case). Marginal probabilities of AKI and MINS sequelae (myocardial infarction, congestive heart failure, stroke, cardiac catheterization, and percutaneous coronary intervention) are multiplied by marginal cost estimates for each outcome to evaluate costs associated with intraoperative hypotension. The unadjusted (adjusted) model found hypotension control lowers the absolute probability of AKI by 2.2% (0.7%). Multiplying these probabilities by the marginal cost of AKI, the unadjusted (adjusted) AKI model estimated a cost reduction of $272 [95% CI = $223-$321] ($86 [95% CI = $47-$127]) per patient. The AKI model based on relative risks from the RCT had a mean cost reduction estimate of $281 (95% CI = -$346-$750). The unadjusted (adjusted) MINS model yielded a cost reduction of $186 [95% CI = $73-$393] ($33 [95% CI = $10-$77]) per patient. The model results suggest improved intraoperative hypotension control in a hospital with an annual volume of 10,000 non-cardiac surgical patients is associated with mean cost reductions ranging from $1.2-$4.6 million per year. Since the magnitude of the RCT mean estimate is similar to the unadjusted observational model, the institutional costs are likely at the upper end of this range.