Perioperative hyperglycemia in patients with and without diabetes impacts as many as 35–50% of all patients undergoing noncardiac surgery1,2,3 and is associated with adverse patient outcomes including increased risk of infection and greater 30-day mortality.4,5,6 Observational data suggest that postoperative hyperglycemia may be more closely associated with adverse outcomes than preoperative hyperglycemia is,4 but the independent contribution of intraoperative hyperglycemia to adverse outcomes is not as well described. Unsurprisingly, intraoperative hyperglycemia is associated with postoperative hyperglycemia,7 but many major anesthesia8,9 and diabetes society10,11 guidelines do not make recommendations on appropriate monitoring or treatment of intraoperative hyperglycemia. Those who do make recommendations focus on patients with diabetes only,12 though as many as 10% of patients without diabetes will have perioperative hyperglycemia.3

Previous quality improvement work has suggested significant intraoperative quality gaps in glycemic management, most notably in intraoperative monitoring. For example, two studies in different settings report that less than 30% of patients at risk of hyperglycemia had any glucose measurement during surgery.13,14 Standardized intraoperative protocols have increased monitoring and reduced intraoperative hyperglycemia, though these protocols are typically based on expert opinion.15,16,17 We undertook a scoping review of the extant literature to describe the evidence to guide intraoperative glycemic management. We aimed to identify evidence knowledge gaps, which may inform the direction of future research on intraoperative anesthetic management of patients at risk of hyperglycemia.

Methods

Study design

A scoping review protocol was developed in accordance with the Joanna Briggs Institute methodology (Electronic Supplementary Material [ESM] eAppendix 1).18 A scoping review method was selected to answer the study question because of the anticipated heterogeneity in study design and outcomes in this topic area. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines informed composition of this manuscript.19

Research question

The population was defined as nonpregnant adults with or without diabetes undergoing noncardiac surgery of any type, duration, urgency, and anesthetic management. We included articles that examined any intraoperative exposure or intervention (e.g., use of dexamethasone, insulin formulation, diabetes medication selection, blood glucose measurement protocol) with a postoperative glycemic outcome, or any clinical outcome.

Data sources and search strategy

The search strategy was developed by a medical librarian (M. V.) for MEDLINE and adapted for other databases (ESM eAppendix 2). The following electronic databases were searched from inception to 14 July 2021: MEDLINE (Ovid), PubMed, PubMed Central, EMBASE, CINAHL, Medline (EBSCO), Cochrane Library, SciVerse Scopus, and Web of Science. A gray literature search (14 July 2021) included Google and Google Scholar,20 hand searching the reference lists of included articles,21 OAISter, review of institutional protocols, and search on ClinicalTrails.gov.

Study selection

All study designs were eligible for inclusion. Studies that were not available in the English language were translated for review and data extraction. Articles that reported on cardiac or obstetrical surgeries were excluded because the interaction between blood glucose and outcomes after cardiac surgery likely differs from those after noncardiac surgeries4 and glycemic targets for pregnant people differ from those of nonpregnant people.22 Studies that included only people younger than 18 yr of age or did not report on an intraoperative strategy or glucose values were excluded. Studies that reported data in duplicate with another citation or published in a predatory journal were excluded.

All identified studies were uploaded into Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia), which automatically removed duplicates. Article titles and abstracts were screened by two independent reviewers for eligibility, and articles without an abstract were screened in full for inclusion criteria. A third reviewer resolved disagreements regarding whether an article met the inclusion criteria for this scoping review.

Data extraction

A data extraction form was created by members of the study team (ESM eAppendix 3) and pilot tested on four articles. The study characteristics; population; surgery characteristics; hospital admission; exposures/interventions in the pre-, intra-, and postoperative periods; comparisons; and clinical and glycemic outcomes were extracted. Data were extracted independently in parallel by two independent study team members and reconciled. Differences were resolved through discussion and consulting the primary article. Remaining disagreements were resolved by a third independent reviewer.

Analysis

Studies were grouped into original research or review methods and compared by study characteristics such as study design and country of origin. When available, we compared surgical characteristics (urgency, discipline, admission type), included population (patients with and/or without diabetes), outcomes and exposures, and pre- or postoperative glycemic management. Where possible, results from studies were converted to an absolute risk reduction (ARR) or absolute risk increase (ARI) or an odds ratio (OR) to allow comparison between studies.

Results

Study characteristics

The search strategy identified 134 studies (Fig. 1). After removing duplicates and title and abstract screening, 76 citations underwent full-text review, and 41 of these met the eligibility criteria. Of these, 24 were original research studies and 17 were reviews, including the Society for Ambulatory Anesthesia (SAMBA) guidelines for perioperative glycemic management (Table 1).23 Most original research studies focused on patients admitted to hospital for elective surgery. There were no studies that compared intraoperative glucose monitoring protocols (Table 2). The most common surgical discipline was neurosurgery, though most studies combined surgical disciplines or did not report the disciplines included (Table 3).

Figure 1
figure 1

PRISMA flow diagram of included and excluded citations in this study

Full size image
Table 1 Select characteristics of the original research articles included in this review, divided into intervention or exposure
Full size table
Table 2 The number of studies included in this review by intervention or exposure and included population
Full size table
Table 3 The surgical and patient characteristics of the population studied in the original research studies included in this review
Full size table

Intraoperative glycemic targets

Eleven studies examined intraoperative glucose targets; one was a randomized trial,24 two were systematic reviews,25,26 and eight were observational studies27,28,29,30,31,32,33,34 (Table 1). Most (but not all29) studies included patients with and without diabetes, though not all studies stratified their results for patients with and without diabetes separately. The surgical types also varied in urgency, duration, and specialty. These studies reported on a range of outcomes using differing lengths of time for ascertainment, including intensive care unit (ICU) admission,24,28,29 postoperative infections,25,27,28,29,31,33,34 cardiovascular events,25,28 surgical site infections,30 anastomotic leaks,32 complications,28 and mortality (Table 4).25,27,29,33 All studies defined the exposure (intraoperative glucose values) categorically rather than continuously and used a range of cut-offs to define hyperglycemia (Table 5).

Table 4 The number of studies included in this review that examined an intraoperative glucose measurement with a clinical outcome. Studies may have reported on multiple different glucose categories and clinical outcomes
Full size table
Table 5 Treatment of intraoperative glucose as a continuous or categorical variable by studies included in this review, stratified by whether glucose was an exposure or the outcome
Full size table

Postoperative infections were variably associated with intraoperative hyperglycemia across observational studies (Table 1). Four studies found that intraoperative glucose measurements more than about 8.0 mmol·L-1 were associated with a greater risk of infection compared with patients with lower measurements. These studies examined liver transplant recipients,27 patients with diabetes undergoing emergent orthopedic surgery,28 patients undergoing elective knee replacement,31 or patients undergoing general, vascular, or urologic procedures.34 Effect estimates of postoperative infections ranged from an ARI of 18.0%27 to an OR of 1.3 (95% confidence interval [CI], 1.0 to 1.7)34 to 4.3 (95% CI, 1.9 to 9.6).28 Only one study of liver transplant recipients focused on surgical site infections specifically and reported an association with intraoperative hyperglycemia (≥ 11.1 mmol·L-1).30 In contrast, two observational studies found no association of intraoperative hyperglycemia (defined as greater than 8.8 mmol·L-1) with infection in patients without diabetes undergoing elective major abdominal procedures29 or patients undergoing any general, vascular, endocrine, or hepatobiliary surgeries.33 None of these studies adjusted for pre- or postoperative glycemic values.

Intraoperative hyperglycemia was not associated with admission to the ICU in a single trial that randomized nearly 400 patients undergoing any major elective surgery to “strict” (4.4–6.1 mmol·L-1) or “loose” (10.0–11.1 mmol·L-1) intraoperative glycemic targets, though the authors did not stratify their results by diabetes status.24 This result was supported by an observational study of patients with diabetes undergoing emergent orthopedic surgeries who had stress hyperglycemia (≥ 7.0 mmol·L-1 fasting) that found no association between hyperglycemia and ICU admission.28 In contrast, an observational study of patients without diabetes undergoing elective major abdominal surgeries found that patients with one or more intraoperative measurements greater than 8.8 mmol·L-1 had an ARI of 8.7% of requiring postoperative ICU admission compared with patients that had intraoperative glucose of 6.9–8.8 mmol·L-1 and an ARR of 9.7% compared with patients with normal measurements (< 7.0 mmol·L-1; P = 0.02).29 Both studies were adjusted for multiple potential confounders, including surgical approach, anesthetic modality, and duration of the procedure.

A single study including patients undergoing a liver transplant with and without diabetes reported that intraoperative glucose greater than 8.3 mmol·L-1 was associated with an ARI of 13.1% (P < 0.05) in one-year postoperative mortality.27 In contrast, two studies reported no increase in 30-day mortality for patients undergoing abdominal surgery who had intraoperative glucose greater than 7.0 mmol·L-129 or those undergoing unspecified surgery types who had intraoperative glucose greater than 10.0 mmol·L-1 after adjustment for covariates.33

Dexamethasone

Five of six studies,35,36 including three placebo-controlled randomized trials,37,38,39 reported that patients receiving dexamethasone had higher intraoperative or postoperative glucose measurements than patients who did not. Most of these studies used 8 mg of intravenous dexamethasone for postoperative nausea and vomiting36,37,38,39 though one used 10 mg for patients undergoing intracranial surgery.35 This increase in glucose measurements ranged from 2.139 to 3.5 mmol·L-137 in patients with diabetes and from 0.939 to 1.6 mmol·L-137 in patients without. There was no increased risk of hyperglycemia reported in a single observational study that did not stratify outcomes by diabetes status and used a dose range of 4–8 mg of intravenous dexamethasone.40

Anesthetic selection

One cohort study41 and one randomized trial42 compared intraoperative glucose measurements in patients receiving sevoflurane, desflurane, and propofol; however, the trial excluded patients with diabetes and the cohort study was restricted to patients with type 2 diabetes. Despite these differences, sevoflurane and desflurane were associated with a slightly greater rise in intraoperative glucose measurements compared with propofol (0.3–1.0 mmol·L-1) in both studies, though it is not clear whether this increase was clinically meaningful. There were no studies that compared anesthesia modalities and glycemic outcomes.

Insulin formulation and/or dosing

Two randomized trials examined the formulation and dosing regimen of intraoperative insulin delivery. In people with type 2 diabetes undergoing general anesthesia for an elective or emergent operation longer than one hour, those that received intravenous boluses of insulin once per hour based on intraoperative blood sugars spent nearly 15% more operative time in target (5.5–10.0 mmol·L-1) compared with patients who were placed on a continuous intravenous insulin infusion with dextrose 5% that was adjusted every hour based on intraoperative blood sugars.43 In comparison, patients with and without diabetes who discontinued their home medications and were started on a basal bolus insulin regimen and received intravenous insulin during surgery had a lower mean glucose on the first postoperative day compared with patients who continued their home medications and received correction-only subcutaneous insulin.44

Other medications

A retrospective cohort study reported that patients who received heated intraperitoneal chemotherapy using dextrose-containing carrier solutions rather than lactated Ringers had greater prevalence of severe hyperglycemia (> 11.1 mmol·L-1), postoperative infections (ARI, +20.6%; P < 0.01) and increased moderate and severe complications (ARI, +17.6%; P = 0.03).45

Perioperative glycemic management pathways

There were four studies that evaluated the influence of standardized protocols for perioperative glycemic management that contained advice on intraoperative management.15,16,46,47 All protocols reduced hyperglycemia (variably defined), though implementation was not consistently associated with improved clinical outcomes; although the largest of these studies did show a reduction in surgical site infections after implementation (ARR, 3.6%; P < 0.05),16 there was no difference in mortality, readmissions, or length of stay in another.46 All identified protocols recommended monitoring blood glucose intraoperatively every one to two hours but were heterogeneous with respect to whether they were only used for patients with diabetes or for all patients, their target glucose measurements (ranging from less than 7.846,47 to 10.015,16 mmol·L-1) and recommended insulin formulation (intravenous or subcutaneous). In addition, protocols generally had preoperative and postoperative pathways in addition to intraoperative recommendations and variably defined the perioperative period, and so any clinical benefit cannot be directly attributed to intraoperative glycemic management.

Narrative reviews

There were 15 narrative reviews, including consensus statements, which addressed some aspect of intraoperative glycemic management.13,17,23,48,49,50,51,52,53,54,55,56,57,58,59 Most, including the SAMBA guidelines, recommended measuring a blood glucose at least every two hours during surgeries longer than one to two hours,13,17,49,56,57,58 though some recommended twice hourly.55,59 Similarly, nearly all reviews recommended a glucose target less than 10.0 mmol·L-1. There was variation in recommendations on insulin regimens and formulations. Often these recommendations asked practitioners to consider the type and duration of surgery as well as patient-specific factors when selecting subcutaneous or intravenous insulin. Only one review recommended avoiding dexamethasone and volatile anesthetics17 in patients at risk of hyperglycemia.

Discussion

In this scoping review, we aimed to synthesize the literature examining intraoperative glycemic management to identify gaps requiring additional research, and make suggestions on how researchers can standardize their reporting and methods to best answer the remaining uncertainties in intraoperative glycemic management. The number and characteristics of included narrative reviews reflects the observed heterogeneity in original research studies; we identified nearly as many nonsystematic reviews and expert opinion articles as original research studies. Journals and granting agencies may act as gatekeepers to ensure that future studies analyze and report data in a consistent, usable fashion.

Consistent definitions of exposures and outcomes are needed to better understand the relationship between intraoperative glucose values and adverse clinical outcomes. For example, observational studies examining associations between intraoperative glucose measurements and postoperative outcomes should treat glucose values as continuous data rather than creating categories of “high” and “normal” glucose to facilitate comparison across studies. Current evidence suggests that postoperative hyperglycemia is more associated with adverse outcomes than preoperative hyperglycemia is4 and that identification of patients with hyperglycemia reduces these outcomes;60,61 however, the contribution of intraoperative hyperglycemia to adverse outcomes relative to the influence of pre- or postoperative hyperglycemia is not yet known.62 The literature examining the relationship between diabetes and adverse surgical outcomes is similarly limited by inconsistent definitions of outcomes and exposure.62 Investigators should consider how to adjust or control for pre- and postoperative hyperglycemia to better isolate the association of intraoperative hyperglycemia with adverse outcomes.

Due to potential differences in the inflammatory response and baseline risk of complications between different types of surgeries, investigators should carefully consider restricting the selected surgical population, indication, and urgency to avoid missing important results when combining heterogeneous populations. For example, we identified only two studies that examined day surgery patients, and half of studies combined elective and emergent surgeries or did not specify the urgency. Similarly, 15 of 28 studies combined multiple disciplines or did not report the specific surgical disciplines that were included despite literature suggesting that adverse events differ greatly by procedure type.63 A signal for harm associated with intraoperative hyperglycemia may be missed in studies that combine high- and low-risk procedures.

Similarly, decisions about the study population should be carefully considered. At minimum, investigators should report results separately for patients with and without diabetes and should differentiate patients with type 1 and type 2 diabetes. Perioperative management of patient’s home medication regimens varies between centres64 but may confound studies that examine intraoperative glycemic response to different management strategies, in particular dexamethasone or inhalational anesthetics.

In the interim, the evidence does suggest that intraoperative hyperglycemia is likely a modifiable risk factor for postoperative infection in patients with and without diabetes. Due to study heterogeneity, it is not possible to recommend a specific glucose cut-off for all surgeries and all patients. This heterogeneity is reflected in major society guidelines, which differ in their recommended intraoperative blood glucose targets from less than 10.023 mmol·L-1 to less than 12.012 mmol·L-1. Many societies, including Diabetes Canada and the American Diabetes Association, do not comment on intraoperative targets.10,11 Future studies should examine the safety, feasibility, and impact on infections between different intraoperative glucose targets in populations at high risk of postoperative infection. In light of the available literature, anesthesiologists may consider measuring glucose and treating intraoperative hyperglycemia, while carefully monitoring patients for hypoglycemia throughout the perioperative period, including in the postoperative recovery room.

Because patients with and without diabetes who receive dexamethasone have increases in their blood glucose,37,39 anesthesiologists may consider monitoring for and treating hyperglycemia in patients at greatest risk of hyperglycemia or adverse outcomes from hyperglycemia who receive dexamethasone. Although intraoperative glucose may increase with dexamethasone administration, a randomized placebo-controlled trial found that patients with and without diabetes who received dexamethasone did not have a greater incidence of surgical site infection compared with those who received placebo.65 Alternatives to dexamethasone could be considered, when possible, for patients at greatest risk of hyperglycemia or postoperative infections regardless of diabetes status; although prediction models are not well established, this risk could be estimated based on patient characteristics (e.g., hemoglobin A1c,66 use of immunosuppression medications67), surgical factors (e.g., surgical approach68), and the severity of the consequences of infection (e.g., procedures involving implanted prosthetic materials). Similarly, the inhaled anesthetics sevoflurane and desflurane may raise intraoperative glucose values, and anesthesiologists could consider blood glucose monitoring in patients at greatest risk of hyperglycemia who receive these agents. It is not known whether intraoperative treatment of hyperglycemia reduces any potential adverse effects of using these medications, and the association of other anesthetic modalities with hyperglycemia has not been studied. There is an ongoing, randomized study comparing intraoperative glucose measurements and clinical outcomes for patients receiving total inhalational anesthesia and those receiving total intravenous anesthesia, which may address these questions.69

The optimal intraoperative management of patients with identified hyperglycemia is not clear based on the available evidence. No studies compared outcomes between patients who underwent different intraoperative monitoring regimens though most studies measured and most reviews recommended measuring a blood glucose value every one to two hours in surgeries greater than one hour, in keeping with guideline recommendations from the UK.12 Although between 5 and 10% of patients without diabetes experience intraoperative and postoperative hyperglycemia,1,3 it is not clear whether all patients without diabetes should undergo intraoperative glucose monitoring and at what interval. This likely would depend on surgery duration, although this was inconsistently reported across studies. Rigorous studies that compare the safety and efficacy of the route of intraoperative insulin dosing (e.g., subcutaneous vs intravenous) and the regimen (e.g., bolus vs continuous infusion) are needed to guide practice. At present, the SAMBA recommendations suggest using bolus-dosed subcutaneous rapid-acting insulin for surgeries less than four hours long with anticipated hemodynamic stability.49 Although there is a risk of publication bias, formalized protocols to direct intraoperative glycemic management probably reduce intraoperative hyperglycemia and may improve clinical outcomes compared with a lack of standard guidelines. To reduce variation between anesthesiologists, organizations should consider standard perioperative glycemic protocols that define intraoperative glycemic targets, suggest insulin dosing protocols, and clarify communication about results and treatment between healthcare providers.70

An important limitation of this review may be that the search strategy was too specific in focusing only on studies that included at least one intraoperative exposure or intervention, thus restricting the available evidence. Because of this inclusion criteria, our strategy excluded the NICE-SUGAR study, which examined “conventional” (less than 10.0 mmol·L-1) vs “intensive” (4.5–6.0 mmol·L-1) glucose targets in critically ill patients without examining intraoperative values.71 We focused on studies that included intraoperative glycemic management with the aim of informing the care delivered specifically by the anesthesia team during the surgical procedure rather than using pre- or postoperative management to direct intraoperative care. In addition, the low number of included citations examining each exposure and/or outcome makes drawing conclusions from the available literature difficult.

Conclusion

Altogether, there are multiple important research gaps in intraoperative glycemic management, most notably a lack of clear glycemic targets and optimal management strategy for intraoperative hyperglycemia. Heterogeneity within and between studies has limited our ability to compare or interpret their results, and investigators should carefully define their outcomes, surgical covariates, patient factors, and pre- and postoperative glycemic management to improve the generalizability of study data. Our synthesis of the available literature should guide future research rather than act as a definitive set of recommendations.