Model fits.

BackgroundAssociations between intraoperative hypotension (IOH) and various postoperative outcomes were shown in retrospective trials using a variety of different definitions of IOH. This complicates the comparability of these trials and makes clinical application difficult. Information about the best performing definitions of IOH regarding 30-day mortality, hospital length of stay (hLOS), and postanesthesia care unit length of stay (PACU-LOS) is missing.MethodsA retrospective cohort trial was conducted using data from patients undergoing noncardiothoracic surgery. We split the obtained dataset into two subsets. First, we used one subset to choose the best fitting definitions of IOH for the outcomes 30-day mortality, hLOS, and PACU-LOS. The other subset was used to independently assess the performance of the chosen definitions of IOH.ResultsThe final cohort consisted of 65,454 patients. In the shaping subset, nearly all tested definitions of IOH showed associations with the three outcomes, where the risk of adverse outcomes often increased continuously with decreasing MAP. The best fitting definitions were relative time with a MAP (mean arterial pressure) of for 30-day mortality, lowest MAP for one minute for hLOS, and lowest MAP for one cumulative minute for PACU-LOS. Testing these three definitions of IOH in the independent second subset confirmed the associations of IOH with 30-day mortality, hLOS, and PACU-LOS.ConclusionsUsing a data-driven approach, we identified the best fitting definitions of IOH for 30-day mortality, hLOS, and PACU-LOS. Our results demonstrate the need for careful selection of IOH definitions.Clinical trial number: n/a, EC #2245/2020