Background & aims: In retrospective studies an indisputable causal relationship between hyperglycemia and postoperative infections cannot be entirely disclaimed. We aimed investigate whether the time trends of blood glucose levels in the perioperative period could be a determinant of surgery-related infections.
Methods: Adult patients without diabetes who were candidates for elective major abdominal operation were prospectively enrolled in a longitudinal, observational multicenter study. The blood glucose level was measured every 6 h for 3 days. We calculated the association between blood glucose (BG) levels and the risk of occurrence of surgery-related infections using a joint regression modeling for longitudinal and time-to-event outcomes which accounts for the effect of other risk factors.
Results: Between January 2016 and November 2017, we obtained 6078 BG measures distributed on different time-points in 452 patients. There was a nearly 3-fold increased risk of having hyperglycemia, defined as BG ≥ 125 mg/dL, if the BG level at admission was >100 mg/dL (OR = 2.986, P < 0.001).The hazard of infection for each 10 mg/dL increase of BG levels over time was marginal (HR = 1.065, P = 0.045). The calculated risk of having an infection was 9.6% for BG going from 110 mg/dL during surgery to 84 mg/dL at the end of day 3, 10.5% for BG decreasing from 140 to 114, 11.8% for BG decreasing from 180 to 154 and 24.5% for BG increasing from 80 to 145, 24.7% for BG increasing from 110 to 175, and 25.4% for BG increasing from 140 to 205.
Conclusions: The time trends of BG - as opposed to the absolute concentration -are major determinants of the risk of postoperative infections.
Keywords: Hypergycemia; Infection; Joint regression model; Outcomes; Surgery; Trend.
Copyright © 2018 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.