Does insulin and not reduced blood glucose levels decrease postoperative infections?

Editor, Perioperative hyperglycemia in surgical patients with type 2 diabetes mellitus is an independent risk factor for adverse postoperative cardiovascular complications, prolonged hospital stay and poorer outcome.1 Poor postoperative blood glucose control is independently associated with increased infections in retrospective studies and may increase length of hospital stay (LOHS).2–4 We recently compared retrospectively perioperative blood glucose levels and insulin administration in a control cohort with liberal blood glucose management and afterwards a cohort of patients subjected to a moderately tight perioperative diabetes mellitus protocol (PDM-P). The medical records of 238 consecutive diabetic patients undergoing elective low-to-moderate risk surgery were evaluated. The study was approved by the local Human Subjects Committee of the VU University Medical Centre (Amsterdam, the Netherlands; chair Professor Dr. J. Rauwerda), on 26th November 2008. No protocol number was assigned. The Committee waived informed consent. The control cohort received liberal diabetes mellitus care (n = 118), whereas the PDM-P intervention cohort was treated according to the following protocol (n = 120). On the day of surgery, oral diabetic medication or short-acting insulin preparations were stopped, whereas long-acting insulin dosing was reduced by 50%. All patients received a standard infusion regimen starting at 7:00 a.m. (glucose 5% with 4 I.U. short acting human insulin Actrapid at 83 ml/h). In the PDM-P intervention group preoperative, intraoperative and postoperative blood glucose were measured and patients received short-acting human insulin [Actrapid intravenously (i.v.) Novo Nordisk] if a blood glucose level was above 10 mmol/l. The protocol was applied on the general ward by trained nurses, who dealt with the specific safety issues associated with the administration of short-acting insulin, clinical symptoms of hypoglycemia, and the emergency procedures in case of hypoglycemia (100 ml 5% glucose i.v.). The protocol was continued during the first three postoperative days. Hypoglycemia was defined as a blood glucose less than 2.5 mmol/l. Medical records were reviewed for the 30-day incidence of wound infections and LOHS. There were no significant differences in patient characteristics between the control and PDM-P intervention cohort with respect to age, sex, insulin-dependent diabetes mellitus, median ASA classification, surgical category, preoperative HbA1c, potassium and fasting blood glucose values. The BMI of PDMP-P patients was slightly higher when compared to control (30 ± 6 vs. 28 ± 6 kg/m2; P = 0.02). In the PDM-P intervention group we found a reduced LOHS (8 ± 1 vs. 5 ± 5 days; P = 0.035) and lower incidence of postoperative wound infections (15 vs. 7%; P = 0.03). We observed no differences in perioperative blood glucose levels between groups (Fig. 1, panel a), whereas the frequency of insulin administration was higher in the PDM-P group when compared to controls [15 vs. 0%; P = 0.03 (preoperative); 16 vs. 4%; n.s. (intraoperative); 49 vs. 15%; P = 0.001 (postanesthesia care unit and 36 vs. 14%; P = 0.04 (ward); panel b]. There was no difference in the incidence of hyperglycemia greater than 10 mmol/l among groups (panel c). Moreover, PDM-P patients received a larger overall insulin dose (10.9 ± 14.2 vs. 3.3 ± 7.9 I.U).Fig. 1: No captions available.The unexpected finding in our study was that, despite more insulin administration in the PDM-P intervention group, the beneficial postoperative effects of our moderately tight insulin regimen could not be attributed to lower perioperative blood glucose levels. However, the PDMP-P group experienced less wound infection and a reduced length of LOHS when compared with the control group. Our findings support the suggestion that insulin exerts protective benefits for postoperative outcome by other mechanisms. Indeed, besides normalisation of blood glucose, insulin ensures normalisation of dyslipidemia and promotes immunologic effects, such as suppression of excessive inflammation and improvement and stabilisation of macrophage function.5,6 An alternative explanation could be that the implementation of a standard operating procedure results in earlier diagnostic interventions and specific treatments contributing to a decrease in LOHS and wound infections. Our observation has many limitations related to the study design and related to the absence of adjustments of measured risk factors (obesity) and unmeasured factors (tobacco use, operative time, tissue perfusion, prophylactic antibiotic use, postoperative mobilisation and nutrition). A large randomised controlled trial is needed to test formally the direct anti-infectious effect of insulin in a surgical population with increased preoperative blood glucose levels.