Key points are not available for this paper at this time.
ObjectivePrevious retrospective studies have established a relationship between postoperative hypoglycemia and adverse outcomes after cardiac surgery, but none have accounted for the cause of hypoglycemia.MethodsA retrospective review was performed of patients who underwent cardiac surgery at a single institution between 2016 and 2021. Patients were categorized as hypoglycemic if they had 1 or more postoperative blood glucose measurement less than 70 mg/dL and normoglycemic otherwise. Hypoglycemia was subcategorized as noniatrogenic (underlying liver failure, adrenal insufficiency, sepsis, or shock) or iatrogenic (insulin infusion continued while nil per os or infusion protocol violated) via manual chart review. Baseline characteristics were compared between groups using Pearson χ2, analysis of variance, and Kruskal-Wallis testing, and outcomes were compared using multivariable logistic regression.ResultsIn total, 5373 patients and 183,346 glucose measurements were included. Hypoglycemia occurred in 5% (267) of patients, of whom 63% (169) were iatrogenic and 37% (98) were noniatrogenic. In a multivariate analysis adjusting for age, sex, case urgency, pre-existing diabetes, and bypass time, both iatrogenic and noniatrogenic hypoglycemia were associated with greater odds of renal failure, prolonged ventilation, and prolonged intensive care unit length of stay relative to normoglycemia, but the magnitude was substantially lower in iatrogenic hypoglycemia. Patients with noniatrogenic hypoglycemia had 68.6 times greater odds of mortality relative to patients who were normoglycemic (odds ratio, 68.6; confidence interval, 39.5-119), but patients with iatrogenic hypoglycemia had no increased odds of mortality (odds ratio, 1.45; confidence interval, 0.77-2.73).ConclusionsWhen excluding patients with conditions known to cause hypoglycemia from the analysis, the morbidity and mortality of iatrogenic hypoglycemia from tight postoperative glycemic control is dramatically attenuated.Graphical abstract
Rando et al. (Thu,) studied this question.