Abstract
Aims
Surgical inpatients experience variable glycaemia, with hyper and hypoglycaemia, contributing towards increased morbidity and mortality. Understanding of mechanisms for glycaemic variability and clinician responses remains limited. We aimed to describe patterns of glycaemia in the immediate perioperative period and its drivers, including treatment with glucose‐lowering agents.
Methods
This was a sub‐analysis of participants in the STOIC‐D Surgery trial, a randomised controlled trial of early specialist‐led glycaemic management for inpatients with pre‐admission diabetes or inpatient hyperglycaemia. Patients in this sub‐study underwent at least one procedure during admission. Data from day prior to surgery, day of surgery and day after surgery were collected, including blood glucose levels and factors influencing glycaemia.
Results
518 participants were included, with a mean HbA1c of 52 mmol/mol (6.9%). 295 (57%) patients received insulin during the three perioperative days; 130 (44%) of these insulin‐receiving patients had a consistent increase in their total daily insulin dose across the 3 days. These patients had the highest perioperative glycaemia compared to patients with different insulin titration trends, particularly on the day of and day after surgery. Fasting duration was the only variable identified as being significantly associated with increased postoperative glycaemia. Hypoglycaemia was uncommon, occurring in 3% of all patients for each perioperative day. This was higher for patients receiving perioperatively insulin.
Conclusions
This study characterised perioperative glycaemia trends and suggests that clinicians identify patients who require intensification of insulin treatment in the perioperative period, which needs to be counterbalanced with the risk of hypoglycaemia.
Keywords: inpatient diabetes, perioperative diabetes, type 2 diabetes
What's new?
Our retrospective sub‐study of inpatients undergoing surgery observed that 44% of patients receiving insulin had a consistent increase in total daily insulin dose across the three immediate perioperative days.
Despite an increase in insulin dosage, patients continued to experience hyperglycaemia, which is well established to contribute towards increased inpatient morbidity and mortality.
Future studies should aim to explore individualised strategies for hospital inpatients that balance appropriate up‐titration of insulin to sufficiently treat perioperative hyperglycaemia and risk of hypoglycaemia.
1. INTRODUCTION
Hyperglycaemia and hypoglycaemia in the inpatient setting are associated with increased morbidity and mortality. 1 , 2 , 3 Adverse inpatient outcomes including in‐hospital healthcare‐associated infection, acute kidney injury and stroke have increased risk with hyperglycaemia, independent of pre‐admission diabetes status. 4 Optimising glycaemia in inpatients, regardless of pre‐existing diabetes, represents an intervention to improve health outcomes. 5 , 6 In addition, pre‐existing diabetes is associated with postoperative complications and hospital length of stay. 7
Numerous factors can influence glycaemia in surgical inpatients. These include stress‐induced hyperglycaemia, 8 which can be driven by the underlying disease necessitating a surgical admission, or the physiologic stress of surgery itself. Clinician‐driven factors include glucocorticoid administration, particularly for management of postoperative nausea and vomiting, 9 fasting and dietary modification, and modification of glucose‐lowering agent prescriptions, including insulin. Perioperative hyperglycaemia contributes to increased risks of adverse clinical outcomes, including longer length of stay, hospital complications and mortality. 1
Surgical inpatients are an important cohort for specialised input regarding glycaemic management. Many protocols provide guidance on withholding and substitution of glucose‐lowering agents in the perioperative period, particularly valuable for clinicians without specialist expertise in diabetes management. However, these guidelines do not typically account for the nuance of day‐to‐day variation in glycaemia for surgical inpatients. Indeed, understanding of the degree and type of glycaemic variation in the perioperative period, as well as contributors towards this variation, remains relatively limited.
This study aimed to examine patterns of glycaemia in the immediate perioperative period, and trends in factors influencing glycaemia, particularly treatment with insulin and non‐insulin glucose‐lowering medications. Describing perioperative patterns and drivers of glycaemia in the context of diabetes treatments will enable refinement of perioperative glycaemic management approaches.
2. METHODS
2.1. Participants
This was an analysis of participants in the STOIC‐D Surgery trial, 6 a single‐centre randomised controlled trial of early specialist‐led glycaemic management at the Royal Melbourne Hospital, an adult quaternary metropolitan service. Surgical services provided include general surgery, trauma, orthopaedics, cardiothoracics, plastics and reconstructive, vascular, urology, otolaryngology, ophthalmology and neurosurgery. Patients included in STOIC‐D Surgery were adult surgical inpatients who either had a pre‐admission diagnosis of diabetes (of any type excluding Type 1 diabetes) or newly detected hyperglycaemia during their admission, defined as at least one random blood glucose level ≥ 11.1 mmol/L (200 mg/dL). The trial randomised 1371 inpatients to either receive early electronic proactive specialist‐led diabetes care (intervention), or standard diabetes care (control), during their inpatient stay. Upon randomisation of a participant to the intervention arm, an endocrinology fellow performed an electronic diabetes consult with escalation to the inpatient diabetes team for a proactive consult if certain criteria were met.
Participants randomised to the control arm were managed per usual institutional practice. Diabetes management perioperatively was primarily driven by hospital medical officers of the admitting surgical unit, with escalation to the specialist inpatient diabetes team if required. Anaesthetists were typically responsible for intraoperative glycaemic management, and patients admitted to the intensive care unit were managed by intensivists. A perioperative diabetes management protocol is available for guidance. 10 Insulin is charted through an insulin order set available on the electronic medical record (EMR) Epic (Epic Systems Corporation, Verona, WI, USA). Pre‐morbid non‐insulin glucose‐lowering agent prescriptions were continued if appropriate.
This analysis included only patients who underwent at least one inpatient procedure during their admission, with 38% (518 of 1371) of patients from the STOIC‐D surgery trial included. Procedures which did not require any fasting or changes to diet status (e.g. flexible cystoscopy with local anaesthetic only), or which occurred on the day of admission, were excluded. For patients who underwent multiple procedures during their admission, only the earliest eligible procedure of that admission was considered.
2.2. Data collection
Data from the day prior to (D − 1), day of (D0) and day following surgery (D + 1), across four time categories, were manually collected from the EMR and transcribed into REDCap, an electronic data capture tool. 11 , 12 Time categories were defined by institutional meal service times and were 0000–0759, 0800–1229, 1230–1759 and 1800–2359. Insulin doses by insulin analogue type were captured from the EMR, including both routine and supplemental scale insulin orders. Non‐insulin glucose‐lowering agents were also captured. Glucose‐lowering agents prior to the day of admission were not included unless noted to have been taken in the medical notes. Intravenous insulin infusions were included in data collection, with doses (expressed as units per hour) rounded to the nearest 30‐min interval.
Significant glucocorticoid exposure, which included oral, intramuscular, intravenous or intra‐articular administration, regardless of dose, was recorded. Topical and inhaled glucocorticoids use was excluded. Total glucocorticoid exposure per day was calculated in prednisolone‐equivalent dosage. Parenteral or enteral nutrition received, on any of the three perioperative days, was noted. Fasting status across time categories on each day was recorded. A patient was considered to be fasting if they were fasting for at least 50% of the time period. In the absence of specific documentation, it was presumed that patients were not fasting prior to admission and were unfasted at the next meal following their procedure. Total fasting duration per day was calculated as a sum of all time periods during which the patient had been fasting for at least 50% of the time period. White cell count (WCC) and C‐reactive (CRP) protein were recorded for each of the three perioperative days. When multiple results were available on a single calendar day, the earliest result was used.
Blood glucose levels from the day prior to, day of and day following surgery were extracted from the EMR.
2.3. Data analysis
The method described by Weinberg et al. 13 for excluding repeated measurements from hypoglycaemic or hyperglycaemic events was applied to blood glucose data. Mean blood glucose levels were calculated for each patient, for each day and time category. Insulin dose was calculated as a total daily dose for each patient‐day. The perioperative change in WCC was calculated as the difference between the WCC from the day after surgery and day before surgery, or if there was no WCC from the day before surgery, it was calculated as the difference between the WCC from the day after surgery and the day of surgery.
Descriptive statistical analyses, including t‐test analysis were conducted using Excel (Microsoft Corporation, Redmond, WA, USA), and multivariable linear regression analyses were conducted using R (R Foundation for Statistical Computing, Vienna, Austria).
3. RESULTS
Five hundred and eighteen patients from the STOIC‐D surgery trial were eligible for this perioperative analysis. Baseline characteristics are outlined in Table 1. The cohort mean HbA1c was 52 mmol/mol (6.9%) and for those with pre‐admission diabetes was higher, at 57 mmol/mol (7.4%). Non‐insulin glucose‐lowering medication pre‐admission was prescribed in 264 patients (51%). Insulin pre‐admission was prescribed in 110 patients (21%) with a mean HbA1c of 64 mmol/mol (8.0%). 102 (93%) of these patients who were prescribed pre‐admission insulin also received insulin in the perioperative period. The mean HbA1c of the patients who received perioperative insulin but were not prescribed insulin pre‐admission was 56 mmol/mol (7.3%). The patients who were not prescribed insulin pre‐admission had lower glycaemia during most time categories compared with patients prescribed insulin pre‐admission (Figure S1).
TABLE 1.
Baseline characteristics of included patients.
| Total number of inpatients undergoing surgery | 518 |
| Age, years, mean (SD) | 64.6 (16.2) |
| Male sex, n (%) | 329 (64%) |
| Day of admission on which included surgery occurred, median (IQR) | 4 (2–6) |
| HbA1c, %, mean (SD) |
6.9 (1.9) 52 mmol/mol |
| Participants with pre‐admission diabetes, n (%) | 331 (64%) |
| Prescribed insulin and non‐insulin glucose‐lowering agent pre‐admission, n (%) | 79 (24%) |
| Prescribed insulin only pre‐admission, n (%) | 31 (9%) |
| Prescribed non‐insulin glucose‐lowering agent only pre‐admission, n (%) | 184 (56%) |
| Not prescribed any glucose‐lowering agent pre‐admission, n (%) | 37 (11%) |
| Participants in the intervention arm, n (%) | 250 (48%) |
| HbA1c of subgroups, %, mean (SD) | |
| Pre‐admission diabetes |
7.4 (1.7) 57 mmol/mol |
| Prescribed insulin pre‐admission |
8.0 (1.8) 64 mmol/mol |
| Patients with newly detected hyperglycaemia |
6.0 (1.4) 42 mmol/mol |
| Prescribed perioperative insulin and not prescribed insulin pre‐admission |
7.3 (1.8) 56 mmol/mol |
| Participants who received specific treatment during the three perioperative days (D − 1, D0 and D + 1) a | |
| Insulin, n (%) | 295 (57%) |
| Non‐insulin glucose‐lowering agent, n (%) | 135 (26%) |
| Glucocorticoid, n (%) | 215 (42%) |
| Parenteral nutrition, n (%) | 14 (2.7%) |
| Enteral nutrition, n (%) | 89 (17%) |
These groups are non‐exclusive.
Patients with pre‐admission diabetes had higher glycaemia at all timepoints compared with patients with newly detected hyperglycaemia (Figure 1). This was reflective of pre‐admission glycaemia as indicated by HbA1c. There was a general trend of increasing glycaemia from the evening on the day of surgery, which persisted on the day after surgery (Figure 1). For the overall cohort, glycaemia on the evening of surgery until the afternoon after surgery was significantly different compared with the preceding day's corresponding time periods, as evaluated through t‐test statistical comparison (p < 0.05).
FIGURE 1.
Mean patient blood glucose concentration (mmol/L) over the three perioperative days, by time of day, for all patients (n = 503) and the patient subgroups of pre‐admission diabetes (n = 331) and without pre‐admission diabetes (n = 172). * = significant difference (p < 0.05) with t‐test comparison between corresponding value on day prior. D − 1 = day before surgery, D0 = day of surgery, D + 1 = day after surgery.
Insulin medications were used in 295 patients (57%) in the perioperative period either on the day before, day of or day after surgery. 102 (35%) of these patients were prescribed insulin prior to hospital admission. Of the patients who received perioperative insulin, 130 (44%) had their total daily insulin dose increased across the 3 days, with no day‐to‐day down‐titrations (Figure 2). Only 4 (1.4%) patients receiving perioperative insulin had no change to their total daily insulin dose across all three perioperative days (Figure 2). For patients who had their insulin dose increased from day before surgery to day of surgery, the mean increase in total daily dose was 14.7 units. For patients who had their insulin dose increased from day of surgery to day after surgery, the mean increase in total daily dose was 27.4 units.
FIGURE 2.
Trends in total daily dose of insulin in patients over the perioperative period (n = 295).
Perioperative glycaemia, particularly on the day of surgery and day after surgery, was highest for those patients who had a consistent increase in their insulin across the three perioperative days (Figure 3). To investigate the potential drivers of this, multivariable linear regression was performed to assess the contribution to glycaemia by inflammation (as reflected by changes in WCC), glucocorticoid exposure, whether surgery was major or minor, fasting duration and other variables (Table 2). Despite the role of inflammation contributing towards perioperative glycaemia, no association was identified between change in WCC and blood glucose change from day of surgery to the day after surgery (odds ratio [OR] 1.03, 95% confidence interval [95% CI] 0.981–1.08, p = 0.233) (Table 2). Furthermore, blood glucose change from day of surgery to the day after surgery was not significantly associated with the presence of glucocorticoids exposure (OR 1.68, 95% CI 0.868–3.24, p = 0.125) or cumulative prednisolone‐equivalent dosage from day prior to and day of surgery (OR 0.999, 95% CI 0.993–1.01, p = 0.733) (Table 2). Mean white cell count on each perioperative day, stratified by insulin trend, is presented in Table S1. Further data regarding glucocorticoid exposure on the day prior to and day of surgery, are illustrated in Table S2. Glucocorticoid exposure, by insulin trend category, is shown in Table S3.
FIGURE 3.
Perioperative mean patient glucose according to time of day in the context of insulin dosing subgroups. Mean patient blood glucose concentration (mmol/L) displayed across the three perioperative days, by time of day, stratified by insulin treatment trend.
TABLE 2.
Multivariable linear regression analysis of blood glucose difference from day of surgery to the day after surgery, with variables which could potentially influence glycaemia. Values expressed to 3 significant figures.
| β coefficient | p | Odds ratio | 95% confidence interval | |
|---|---|---|---|---|
| Any perioperative glucocorticoid exposure | 0.516 | 0.125 | 1.68 | 0.868–3.24 |
| Combined glucocorticoid exposure from day before and day of surgery, in prednisolone‐equivalent dosing | −0.00108 | 0.733 | 0.999 | 0.993–1.01 |
| Change in white cell count | 0.0298 | 0.233 | 1.03 | 0.981–1.08 |
| Major or minor surgery | 0.00599 | 0.983 | 1.01 | 0.579–1.75 |
| Combined fasting duration from day before and day of surgery | 0.0284 | 0.0288 | 1.03 | 1.00–1.06 |
| Age | 0.00603 | 0.509 | 1.01 | 0.988–1.02 |
| Gender | −0.145 | 0.602 | 0.865 | 0.502–1.49 |
| Diabetes without complications | 0.119 | 0.753 | 1.13 | 0.538–2.36 |
| Diabetes with at least one complication | 0.238 | 0.511 | 1.27 | 0.624–2.58 |
| HbA1c | 0.145 | 0.0765 | 1.16 | 0.985–1.36 |
| Haemoglobin | 0.00115 | 0.856 | 1.00 | 0.989–1.01 |
| Creatinine | 0.000757 | 0.435 | 1.00 | 0.999–1.00 |
An association was found between cumulative duration of fasting from day prior to and day of surgery, and blood glucose change from day of surgery to the day after surgery (OR 1.03, 95% CI 1.00–1.06, p = 0.0288). 34% of patients fasted for at least one time period on the day prior to surgery with a mean duration of 14 h, and 98% of patients fasted for at least one time period on the day of surgery with a mean duration of 16 h (Table S2).
No other variables assessed were associated with blood glucose difference between the day of surgery to the day after surgery. There were insufficient consecutive daily CRP data to enable analysis of an association between CRP level and blood glucose change.
Hypoglycaemia (glucose < 4.0 mmol/L) was relatively uncommon, occurring in 3% of all patients on each perioperative day (Table S4). This was higher for patients who received perioperative insulin on the day prior to surgery and day after surgery (4% and 5%, respectively). Of patients who had an increase in their total daily insulin dose between the day before and day of surgery, with either a further increase or the same total daily dose on the day after surgery, 8% had at least one episode of hypoglycaemia on the day after surgery.
With regard to time of day, the highest rates of hypoglycaemia occurred prior to 8 am on the day after surgery, with 13 patients (2.5%) experiencing hypoglycaemia within that period (Table S5). Hypoglycaemia occurred in four patients (0.8%) prior to 8 am on the day before surgery and in four patients (0.8%) prior to 8 am on the day of surgery.
Although 263 patients (79% of patients with diabetes) were prescribed non‐insulin glucose‐lowering medications prior to admission, only 135 (41% of patients with diabetes) were prescribed non‐insulin glucose‐lowering medications across the three perioperative days. Metformin was the most prescribed non‐insulin glucose‐lowering agent in the immediate perioperative period, followed by dipeptidyl peptidase IV inhibitors, and sulphonylureas (Table 3). Non‐insulin glucose‐lowering medications were prescribed less on the day of surgery, compared to the day before and day after surgery (Table 3). Only two patients were prescribed a sodium‐glucose cotransporter‐2 inhibitor on the day prior to surgery, and none were prescribed these on the day of surgery.
TABLE 3.
Number and percentage of patients in the entire cohort who received a non‐insulin glucose‐lowering agent, for each of the perioperative days.
| D − 1 | D0 | D + 1 | |
|---|---|---|---|
| Metformin | 62 (12%) | 46 (9%) | 79 (15%) |
| Sulphonylurea | 19 (4%) | 6 (1%) | 21 (4%) |
| Acarbose | 0 | 0 | 0 |
| Thiazolidinedione | 2 (< 1%) | 0 | 1 (< 1%) |
| DPP4 inhibitor | 32 (6%) | 17 (3%) | 46 (9%) |
| SGLT2 inhibitor | 2 (< 1%) | 0 | 1 (< 1%) |
| GLP1 agonist | 0 | 1 (< 1%) | 2 (< 1%) |
4. CONCLUSION
Through an observational and descriptive sub‐study, we performed an in‐depth assessment of glycaemia in the immediate perioperative period and examined possible contributors, including changes in insulin prescription. The study identified an overall pattern of lower glucose on the morning of surgery compared to the morning prior, likely reflecting pre‐operative fasting, followed by overall increased glycaemia in the evening after surgery and on the day following. We also observed that clinicians are typically able to correctly identify which patients require intensification of glucose‐lowering medication. However, the magnitude of medication intensification is insufficient in the immediate perioperative period, likely due to the fear of iatrogenic hypoglycaemia, which did occur at a marginally higher rate in the insulin intensification group. We observed that longer fasting duration on the day prior to and day of surgery was associated with a greater increase in blood glucose postoperatively.
The cohort exhibited mean perioperative blood glucose concentrations between 8.0 and 10.0 mmol/L at most assessed time categories, similar to previous studies from multiple institutions including our own. 6 , 10 , 14 , 15 Glucose tended to increase from the day prior to the day after an operative procedure, with higher glycaemia particularly from the evening of surgery onwards. However, rates of hypoglycaemia were also highest prior to 8 am on the day after surgery. This could be due to ongoing postoperative nausea or diet limitations which were not accounted for when recommencing glucose‐lowering agents postoperatively for some patients, or rebound hyperglycaemia.
Mean perioperative blood glucose concentrations were largely reflective of pre‐admission glycaemia, with generally higher perioperative glycaemia in patients with higher HbA1c. This occurred despite being in an inpatient environment with more opportunity for monitoring and optimisation of diabetes management than patients would typically receive as an outpatient. For insulin‐naïve patients who were prescribed perioperative insulin, this was likely effective given their mean blood glucose ranged from 9 to 11 mmol/L, although their mean HbA1c was 56 mmol/mol (7.3%; estimated average glucose 9 mmol/L). However, patients who had a consistent increase of insulin prescribed during the three perioperative days continued to experience overall hyperglycaemia in the immediate postoperative period. Many patients had their non‐insulin glucose‐lowering medications appropriately withheld in the perioperative period; this may have contributed to the occurrence of hyperglycaemia.
Although clinicians appear to appropriately identify which patients may benefit from perioperative intensification of glucose‐lowering agents, our results suggest that intensification is insufficient to prevent the observed perioperative hyperglycaemia. A recent study at another centre also found that a larger up‐titration in insulin dose than the typical 10%–20% increase may be required to adequately control inpatient hyperglycaemia. 16
However, patients who experienced perioperative insulin up‐titration did experience higher hypoglycaemia rates than other groups. Furthermore, the cohort of patients who had an increase in their insulin on the day of surgery, but a decrease on the day after surgery, had a hypoglycaemia rate of 6.6% on the day after surgery. This reduction in insulin on the day after surgery is therefore likely driven in part, by clinicians appropriately down‐titrating insulin doses in response to hypoglycaemia on the morning after surgery.
We hypothesise that current practice likely reflects fear of inducing hypoglycaemia, with reluctance to deviate from what is perceived to be a patient's baseline insulin requirement, and challenges in accurately anticipating hyperglycaemia perioperatively. The established association between inpatient hyperglycaemia and adverse events 4 indicates that reaching target glycaemia in the immediate perioperative period, if achievable without hypoglycaemia, would improve inpatient outcomes. It may be that current insulin intensification practice is insufficiently individualised; in those patients correctly identified as benefiting from insulin intensification, dose increases are insufficient in most while being excessive in some. A recent Cochrane review similarly found that hypoglycaemia rates are higher in intensive glycaemic control compared with conventional glycaemic control for patients with diabetes undergoing surgery. 17 This review also concluded that intensive blood glucose control may reduce cardiovascular events, although it does not appear to reduce all‐cause mortality.
Strategies to improve perioperative glycaemia have included perioperative protocols for diabetes management, and proactive input from specialist inpatient diabetes teams, as was undertaken in the intervention arm of the STOIC‐D surgery trial. Given our findings, guidelines which recommend the degree of insulin dose change perioperatively or specific blood glucose targets for certain patient populations could be useful. Additionally, incorporating insulin infusions into perioperative protocols may play a role in attenuating glycaemic variability, while also limiting hypoglycaemia risk. 18 , 19 For instance, after cardiac surgery, insulin infusions with concurrent variable glucose infusions have been suggested to reduce mortality and morbidity. 20 In addition, the emergence of machine learning techniques to pre‐empt changes in glycaemia and provide advice on appropriate medication adjustment may be a promising assistive tool. 21 , 22
Strengths of this analysis include a relatively large sample size of 518 patients, whose characteristics were readily available from the EMR. It explored additional contributors to perioperative glycaemia including inflammatory status, glucocorticoid exposure and fasting. Our institution is equipped with networked blood glucose meters, which allow automatic integration of blood glucose data into the electronic medical record, guaranteeing glucose data quality. Similarly, the use of a well‐characterised clinical trial cohort ensures high‐quality clinical data.
The main limitation is in identifying the independent contribution of glucose‐lowering medication modification to glycaemia, limited by the observational study design. The multivariable analysis did not identify a significant contribution to perioperative glycaemia from glucocorticoid administration or inflammation. The variable timing of onset of postoperative inflammation and the extent to which it is reflected in serum white cell counts may have contributed towards this. Additionally, the lack of association found may be influenced by the administration of glucose‐lowering agents.
Our analysis did identify that cumulative fasting duration on the day before and day of surgery was associated with greater odds of increased glucose from day of surgery to the day after surgery (OR 1.03). This is perhaps counterintuitive, as we may typically expect that increased duration of fasting would lower glycaemia. This could possibly be explained by fasting from the day prior to and day of surgery, contributing towards lower blood glucose levels on the day of surgery compared to the day after surgery, therefore being perceived as a greater immediate postoperatively increase. Longer duration of fasting may also lead to greater likelihood of glucose‐lowering agents being withheld, particularly non‐insulin agents. Furthermore, a patient needing to fast for longer could be reflective of increased severity of patient disease and therefore contribute towards increased inflammation and subsequent hyperglycaemia.
The generalisability of our study may be limited by being from a single site only, in a clinical trial population. Furthermore, patients who underwent surgery on the second day of their admission may have not spent a full calendar day as an inpatient prior to their surgery. We may therefore lack complete data for these patients, who would also have had less opportunity for clinician optimisation of their glycaemia preoperatively. In retrospective data quality checks, the method of glucose‐lowering data acquisition may have excluded a small proportion of intraoperative insulin administered by anaesthetists with non‐standard order formats.
Future studies could aim to further characterise the interplay of drivers of perioperative hyperglycaemia, including those other than modification of glucose‐lowering medications. This could facilitate identification of specific cohorts at higher risk of hyperglycaemia in the perioperative period, supporting development of targeted strategies to prevent hyperglycaemia. In addition, studies could explore clinician preferences towards modifying glucose‐lowering agents in the perioperative period, including the extent to which clinicians aim for euglycaemia, and attitudes towards implementation of machine learning and predictive algorithms. Evidence generation for individualised perioperative insulin intensification protocols would further assist in optimising anticipatory actions to prevent hyperglycaemia.
In conclusion, this perioperative glucose analysis of the STOIC‐D surgery trial characterised patterns of glycaemia on the day prior, day of and day after surgery, as well as trends in modification of insulin doses across those 3 days. Our results suggest that there is opportunity for further intensification of glycaemic management strategies in the potentially volatile perioperative period, and that achieving this through employment of proactive strategies could improve outcomes for surgical inpatients with diabetes and hyperglycaemia, if these can be implemented without increasing hypoglycaemia.
FUNDING INFORMATION
This investigator‐initiated study was conducted with the support of the Rowe Family Foundation Perpetual Grant and the Royal Melbourne Hospital Home Lottery Victor Hurley grant. RDB was supported by the Australian Government Research Training Program Scholarship. The funders had no role in the study design, recruitment, data collection, analysis, interpretation or writing of the report.
CONFLICT OF INTEREST STATEMENT
SF contributes to the advisory panel for Viatris Inc. and Pfizer Inc. SF contributed to the speaker's bureau for Novo Nordisk, Astra Zeneca, and the Boehringer Ingelheim and Eli Lilly Alliance. No other authors report any conflict of interest.
ETHICS STATEMENT
The protocol for the STOIC‐D surgery trial was approved by the ethics committee of The Royal Melbourne Hospital (Melbourne, Australia; HREC/61095/mH‐2020) and was registered with the Australian New Zealand Clinical Trials Registry (trial number ACTRN12620001303932).
Supporting information
Table S1. Mean white cell count (WCC), for each of the perioperative days, by insulin trend category.
Table S2. Glucocorticoid exposure and duration of fasting, for day prior to and day of surgery.
Table S3. Number and percentage of patients who received glucocorticoids, for each of the perioperative days.
Table S4. Number and percentage of patients who had at least one blood glucose reading less than 4.0 mmol/L, for each of the perioperative days. Percentages not expressed for patients who did not have any change to their insulin dose, due to low absolute numbers in this category.
Table S5. Number and percentage of patients who had at least one blood glucose reading less than 4.0 mmol/L, for each of the perioperative days, by time of day.
Figure S1. Mean patient blood glucose level (mmol/L) across the three perioperative days, by time of day, stratified patients who were on insulin pre‐admission (n = 110) and patients who received perioperative insulin but were not on insulin pre‐admission (n = 193). D − 1 = day before surgery, D0 = day of surgery, D + 1 = day after surgery.
REFERENCES
- 1. Frisch A, Chandra P, Smiley D, et al. Prevalence and clinical outcome of hyperglycemia in the perioperative period in noncardiac surgery. Diabetes Care. 2010;33(8):1783‐1788. doi: 10.2337/dc10-0304 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Turchin A, Matheny ME, Shubina M, Scanlon SV, Greenwood B, Pendergrass ML. Hypoglycemia and clinical outcomes in patients with diabetes hospitalized in the general ward. Diabetes Care. 2009;32(7):1153‐1157. doi: 10.2337/dc08-2127 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM, Kitabchi AE. Hyperglycemia: an independent marker of in‐hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978‐982. doi: 10.1210/jcem.87.3.8341 [DOI] [PubMed] [Google Scholar]
- 4. Barmanray RD, Kyi M, Worth LJ, et al. Hyperglycemia in hospital: an independent marker of infection, acute kidney injury, and stroke for hospital inpatients. J Clin Endocrinol Metab. 2024;109(11):e2048‐e2056. doi: 10.1210/clinem/dgae051 [DOI] [PubMed] [Google Scholar]
- 5. Kyi M, Colman PG, Wraight PR, et al. Early intervention for diabetes in medical and surgical inpatients decreases hyperglycemia and hospital‐acquired infections: a cluster randomized trial. Diabetes Care. 2019;42:832‐840. doi: 10.2337/dc18-2342 [DOI] [PubMed] [Google Scholar]
- 6. Barmanray RD, Kyi M, Colman PG, et al. The specialist treatment of inpatients: caring for diabetes in surgery (STOIC‐D surgery) trial: a randomized controlled trial of early intervention with an electronic specialist‐led model of diabetes care. Diabetes Care. 2024;47(6):948‐955. doi: 10.2337/dc23-1905 [DOI] [PubMed] [Google Scholar]
- 7. Yong PH, Weinberg L, Torkamani N, et al. The presence of diabetes and higher HbA1c are independently associated with adverse outcomes after surgery. Diabetes Care. 2018;41(6):1172‐1179. doi: 10.2337/dc17-2304 [DOI] [PubMed] [Google Scholar]
- 8. Mifsud S, Schembri EL, Gruppetta M. Stress‐induced hyperglycaemia. Br J Hosp Med. 2018;79(11):634‐639. doi: 10.12968/hmed.2018.79.11.634 [DOI] [PubMed] [Google Scholar]
- 9. Nazar CE, Lacassie HJ, López RA, Muñoz HR. Dexamethasone for postoperative nausea and vomiting prophylaxis: effect on glycaemia in obese patients with impaired glucose tolerance. Eur J Anaesthesiol. 2009;26(4):318‐321. doi: 10.1097/EJA.0b013e328319c09b [DOI] [PubMed] [Google Scholar]
- 10. Qi QYD, Kyi M, Pemberton E, Colman PG, Fourlanos S. The pro‐Diab Melbourne perioperative study: a structured pre‐admission perioperative diabetes management plan to improve medication usage in elective surgery. Diabet Med. 2022;39(7):e14838. doi: 10.1111/dme.14838 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: building an international community of software platform partners. J Biomed Inform. 2019;95:103208. doi: 10.1016/j.jbi.2019.103208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap) – a metadata‐driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377‐381. doi: 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Weinberg ME, Bacchetti P, Rushakoff RJ. Frequently repeated glucose measurements overestimate the incidence of inpatient hypoglycemia and severe hyperglycemia. J Diabetes Sci Technol. 2010;4(3):577‐582. doi: 10.1177/193229681000400311 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Cheuk N, Worth LJ, Tatoulis J, Skillington P, Kyi M, Fourlanos S. The relationship between diabetes and surgical site infection following coronary artery bypass graft surgery in current‐era models of care. J Hosp Infect. 2021;116:47‐52. doi: 10.1016/j.jhin.2021.07.009 [DOI] [PubMed] [Google Scholar]
- 15. Aragón‐Sánchez J, Lázaro‐Martínez JL. Impact of perioperative glycaemia and glycated haemoglobin on the outcomes of the surgical treatment of diabetic foot osteomyelitis. Diabetes Res Clin Pract. 2011;94(3):e83‐e85. doi: 10.1016/j.diabres.2011.09.009 [DOI] [PubMed] [Google Scholar]
- 16. Kanbour S, Zale AD, Shim JY, Abusamaan MS, Mathioudakis N. Association between daily insulin dose adjustments and glycemic control in noncritically ill hospitalized hyperglycemic patients: a retrospective cohort study. Endocr Pract. 2025;31(5):557‐563. doi: 10.1016/j.eprac.2025.01.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Bellon F, Solà I, Gimenez‐Perez G, et al. Perioperative glycaemic control for people with diabetes undergoing surgery. Cochrane Database Syst Rev. 2023;2023(8):CD007315. doi: 10.1002/14651858.CD007315.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Hui ML, Kumar A, Adams GG. Protocol‐directed insulin infusion sliding scales improve perioperative hyperglycaemia in critical care. Perioperative Medicine. 2012;1(1):7. doi: 10.1186/2047-0525-1-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Ekström E, Fagher K, Tingstedt B, et al. Hyperglycemia and insulin infusion in pancreatoduodenectomy: a prospective cohort study on feasibility and impact on complications. Int J Surg. 2023;109(12):3770‐3777. doi: 10.1097/JS9.0000000000000714 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Duncan AE, Sessler DI, Sato H, et al. Hyperinsulinemic normoglycemia during cardiac surgery reduces a composite of 30‐day mortality and serious in‐hospital complications: a randomized clinical trial. Anesthesiology. 2018;128(6):1125‐1139. doi: 10.1097/ALN.0000000000002156 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Wright AP, Wright AT, McCoy AB, Sittig DF. The use of sequential pattern mining to predict next prescribed medications. J Biomed Inform. 2015;53:73‐80. doi: 10.1016/j.jbi.2014.09.003 [DOI] [PubMed] [Google Scholar]
- 22. Fujihara K, Matsubayashi Y, Yamada MH, et al. Machine learning approach to decision making for insulin initiation in Japanese patients with type 2 diabetes (JDDM 58): model development and validation study. JMIR Med Inform. 2021;9(1):22148. doi: 10.2196/22148 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Mean white cell count (WCC), for each of the perioperative days, by insulin trend category.
Table S2. Glucocorticoid exposure and duration of fasting, for day prior to and day of surgery.
Table S3. Number and percentage of patients who received glucocorticoids, for each of the perioperative days.
Table S4. Number and percentage of patients who had at least one blood glucose reading less than 4.0 mmol/L, for each of the perioperative days. Percentages not expressed for patients who did not have any change to their insulin dose, due to low absolute numbers in this category.
Table S5. Number and percentage of patients who had at least one blood glucose reading less than 4.0 mmol/L, for each of the perioperative days, by time of day.
Figure S1. Mean patient blood glucose level (mmol/L) across the three perioperative days, by time of day, stratified patients who were on insulin pre‐admission (n = 110) and patients who received perioperative insulin but were not on insulin pre‐admission (n = 193). D − 1 = day before surgery, D0 = day of surgery, D + 1 = day after surgery.