Background
Blood glucose levels rise during and after surgery due to surgical stress. Surgery causes a stress response that results in a release of catabolic hormones and the inhibition of insulin. Moreover, surgical stress influences pancreatic beta-cell function, which results in lower plasma insulin levels. Taken together, this relative hypoinsulinaemia, insulin resistance and excessive catabolism from the action of counter-regulatory hormones make surgical patients at high risk for hyperglycaemia, even non-diabetic individuals (16).
Several observational studies (17–20) showed that hyperglycaemia is associated with an increased risk of SSI and therefore an increased risk of morbidity, mortality and higher health care costs in both diabetic and non-diabetic patients and in different types of surgery. Conflicting results have been reported regarding the different treatment options to control hyperglycaemia in diabetic and non-diabetic patients, the optimal target levels of blood glucose and the ideal timing for glucose control (intra- and/or postoperative). Moreover, some studies targeting relatively low perioperative glucose levels have highlighted the risk of adverse effects associated with intensive protocols as they may cause hypoglycaemia (21–24).
Several organizations have issued recommendations regarding perioperative blood glucose control (). While most recommendations focus on the diabetic patient only, those issued by SHEA/IDSA (25) and the American College of Physicians (26) apply to all surgical patients. They recommend either target levels between 140–200 mg/dL (7.8–11.1 mmol/L) or upper limits of 180 mg/dL (10mmol/L) or 198 mg/dL (11mmol/L). Due to the risk of hypoglycaemia, targeting lower levels should be avoided (26, 27).
Recommendations on perioperative blood glucose control according to available guidelines.
Following an in-depth analysis of the sources and strength of the evidence in current guidelines, the GDG members decided to conduct a systematic review to assess the impact of perioperative blood glucose levels on the risk of SSI and to determine the optimal perioperative target levels in diabetic and non-diabetic surgical patients to prevent SSI.
Summary of the evidence
The purpose of the evidence review (web Appendix 15) was to evaluate whether the use of protocols for intensive perioperative blood glucose control is more effective in reducing the risk of SSI than conventional protocols with less stringent blood glucose target levels. The population studied were patients of all ages, both diabetic and non-diabetic, and undergoing several types of surgical procedures. The primary outcome was the occurrence of SSI and SSI-attributable mortality. A total of 15 RCTs (1–15) including a total of 2836 patients and comparing intensive perioperative blood glucose protocols vs. conventional protocols with less stringent blood glucose target levels were identified. Eight studies were performed in adult patients undergoing cardiac surgery (1, 2, 4, 6, 9–11, 15), 6 in patients undergoing abdominal or major non-cardiac surgery (3, 5, 7, 12–14), and one other study in patients undergoing emergency cerebral aneurysm clipping (8). No study was available in a paediatric population. In 2 studies (4, 7), glucose control was performed intraoperatively only. Eight studies (1, 2, 6, 8, 9, 11, 13, 15) investigated intra- and postoperative glucose control and 5 studies (3, 5, 10, 12, 14) focused on postoperative glucose control.
None of the studies had SSI as their primary outcome. Most studies had a combined outcome of postoperative complications. The definition of SSI was also different in most studies.
There was substantial heterogeneity among the selected studies in the population, notably regarding the timing when intensive blood glucose protocols were applied in the perioperative period and intensive blood glucose target levels. For this reason, separate meta-analyses were performed to evaluate intensive protocols vs. conventional protocols in different settings (that is, in diabetic, non-diabetic and a mixed population) with intraoperative, intra- and postoperative glucose control, and in trials with intensive blood glucose upper limit target levels of ≤110 mg/dL (6.1 mmol/L) and 110–150 mg/dL (6.1–8.3 mmol/L) (web Appendix 15).
Overall, there is low quality evidence that a protocol with more strict blood glucose target levels has a significant benefit in reducing SSI rates when compared to a conventional protocol (OR: 0.43; 95% CI; 0.29–0.64). In addition, there was no evidence in meta-regression analyses that the effect of intensive blood glucose control differed between studies of diabetic and non-diabetic patients (P=0.590). There was evidence that the effect was smaller in studies that used intensive blood glucose control intraoperatively only (OR: 0.88; 95% CI: 0.45–1.74) compared to studies that used controls postoperatively or both intra- and postoperatively (OR: 0.47; 95% CI: 0.25–0.55; P=0.049 for the difference between ORs). Among the intensive protocols, the effect was similar in studies with upper limit target blood glucose levels of ≤110 mg/dL (6.1 mmol/L) and 110–150 mg/dL (6.1–8.3 mmol/L) (P=0.328).
Data from the available evidence showed no difference in the risk of postoperative death and stroke with the use of an intensive protocol compared to a conventional protocol (OR: 0.74; 95% CI: 0.45–1.23 and OR: 1.37; 95% CI: 0.26–7.20, respectively). The study by Ghandi and colleagues was the only one that reported more strokes and deaths in the intensive group (11). This study had comparable 24-hour achieved blood glucose levels in the intensive care unit in both groups, although they were significantly lower in the intensive group intraoperatively and at baseline. Other studies showed equal or even less strokes and/or deaths in the intensive group, but these findings were not significant. In meta-regression analyses, there is no evidence for a difference in risk between studies with an upper limit target blood glucose level of ≤110 mg/dL (6.1 mmol/L) and an upper limit level of 110–150 mg/dL (6.1–8.3 mmol/L) (P=0.484 for mortality and P=0.511 for stroke).
Meta-analysis of hypoglycaemic events in the 8 RCTs with a blood glucose upper limit target level of ≤110 mg/dL (6.1 mmol/L) showed an increased risk of hypoglycaemic events with the use of an intensive protocol over a conventional protocol (OR: 4.18; 95% CI: 1.79–9.79). However, 2 of the 8 studies included in this analysis had no hypoglycaemic events (4, 13) and only 3 studies (3, 7, 14) found significantly more hypoglycaemic events with the use of the intensive protocol. A meta-analysis of 4 studies (1, 2, 6, 10) showed an increased risk of hypoglycaemic events with the use of a strict protocol with a blood glucose upper limit target level of 110–150 mg/dL (6.1–8.3 mmol/L) compared to a conventional protocol (OR: 9.87; 95% CI: 1.41–69.20). Two studies included in this analysis had no hypoglycaemic events (1, 2). Among the studies that could not be included in the meta-analysis due to missing data, 2 reported significantly more hypoglycaemic events in the intensive group (8, 12), while no difference in risk was observed in another study (9). Overall, there is an increased risk for hypoglycaemic events with the use of either intensive protocol for blood glucose control (OR: 5.55; 95% CI: 2.58–11.96). In meta-regression analyses, there is no evidence for a difference in the risk of hypoglycaemia between studies with a target blood glucose level of ≤110 mg/dL (6.1 mmol/L) and an upper limit target level of 110–150 mg/dL (6.1–8.3 mmol/L) (P=0.413).
The GDG underlined that there are many observational studies showing a reduction in SSI with intensive blood glucose control in non-diabetic populations. However, after discussion, the GDG agreed not to include the data from the observational studies.
