Abstract
Introduction:
To improve the detection and management of perioperative hyperglycemia at our tertiary cancer center, we implemented a glycemic control quality improvement initiative. The primary aim was to decrease the percentage of diabetic patients with median postoperative glucose levels >180 mg/dL during hospitalization by 15% within 2 years.
Methods:
A multidisciplinary team standardized preoperative screening, preoperative, intraoperative and postoperative hyperglycemia management. We included all patients undergoing nonemergent inpatient and outpatient operations. . We used a t-test, rank sum, chi-square or Fisher’s exact test to assess differences in outcomes between patients at baseline (BL) (10/2018-4/2019), during the first phase (P1) (10/2019-4/2020), second phase (P2) (5/2020-12/2020) and maintenance phase (M) (1/2021-10/2022).
Results:
The analysis included 9891 BL surgical patients (1470 with diabetes), 8815 P1 patients (1233 with diabetes), 10,401 P2 patients (1531 with diabetes) and 30,410 M patients (4265 with diabetes). The percentage of diabetic patients with median glucose levels >180mg/dL during hospitalization decreased 32% during the initiative (BL, 20.1%; P1, 16.9%; P2, 12.1%; M, 13.7% [P<.001]). We also saw reductions in the percentages of diabetic patients with median glucose levels >180mg/dL intraoperatively (BL, 34.0%; P1, 26.6%; P2, 23.9%; M, 20.3% [P<.001]) and in the post-anesthesia care unit (BL, 36.0%; P1, 30.4%; P2, 28.5%; M, 25.8% [P<.001]). The percentage of patients screened for diabetes by hemoglobin A1C increased during the initiative (BL, 17.5%; P1, 52.5%; P2, 66.8%; M 74.5% [P<.001]).
Conclusion:
Our successful initiative can be replicated in other hospitals to standardize and improve glycemic control among diabetic surgical patients.
Introduction
Perioperative hyperglycemia, defined as a glucose level >180 mg/dL, is associated with adverse postoperative outcomes, including infectious complications, increased length of stay and mortality risk.[1-4] Although intensive glycemic control to levels <108 mg/dL in the critical care setting is detrimental, avoiding hyperglycemia among hospitalized patients remains the standard of care.[5, 6] Thus, the rapid detection and treatment of perioperative hyperglycemia is part of the standard of care for surgery patients.
To improve the detection and management of perioperative hyperglycemia at our tertiary cancer center, we implemented a diabetes screening and glycemic control quality-improvement initiative. The primary aim of our quality-improvement initiative was to decrease the percentage of patients with diabetes whose median postoperative glucose level was >180 mg/dL during hospitalization by 15% over 2 years. Secondary aims were to increase the percentage of patients undergoing diabetes screening before surgery to 30% within 1 year and to increase the percentage of patients with diabetes who have a recent (within 90 days of surgery) hemoglobin A1C result to 95% within 1 year. These aims were chosen as attainable incremental improvements to glycemic control and compliance with screening recommendations.
Methods
Planning the Initiative:
Our quality improvement initiative was known as the Surgical Universal euGlycemic Attainment during Recovery (SUGAR) Initiative. After receiving Quality Improvement Assessment Board approval (QIAB Project 440), we formed a multidisciplinary team of providers, including physicians, nurses, advanced-practice practitioners, clinical pharmacists, and trainees representing the specialties of gynecologic oncology, endocrinology, anesthesiology, internal medicine, and pharmacy. Our team set a standardized diabetic screening recommendation for preoperative patients based on the 2015 United States Preventative Taskforce guidance.[7] The 2015 United States Preventative Taskforce recommends screening patients older than 40 years for diabetes using a hemoglobin A1C if they have a body mass index >35, history of gestational diabetes, and/or ethnic or racial predisposition to diabetes.[7] Institutional policy required that a patient with known diabetes have a hemoglobin A1C within 90 days of surgery.
This quality improvement intervention was applied across the Division of Surgery to all surgical specialties. The specialties included in our analysis are Breast Surgery, Colorectal Surgery, General Gynecology, Gynecologic Oncology, Head and Neck Surgery, Neurosurgery, Orthopedic Oncology Surgery, Plastic Surgery, Surgical Oncology (including divisions of Endocrine Surgery, Gastrointestinal Surgery, Hepatobiliary Surgery, Pancreatic Surgery, Sarcoma Surgery and Melanoma Surgery), Thoracic Surgery, Vascular Surgery and Urology. Before surgery, patients were triaged by the surgical teams to the preoperative anesthesia clinic or the endocrinology clinic for glycemic optimization depending on their level of glycemic control. The endocrinology clinic has the ability to more aggressively intervene for patients with severe or refractory hyperglycemia due to expertise and additional resources such as diabetic educators. The final version of the referral algorithm, which was adapted according to feedback from key stakeholders during Plan-Do-Study-Act cycles throughout the initiative, is shown in Figure 1. Patients with the following were referred directly to endocrinology and not managed as part of this quality improvement initiative: Type 1 diabetes mellitus, current use of an insulin pump, history of a total pancreatectomy or planned to undergo a pancreatectomy, current use of U-500 insulin or on >100 units of insulin daily and A1C>8.5%, history of diabetic ketoacidosis, current use of systemic steroids and glucose >200mg/dL or new onset hyperglycemia in a patient with history of immune-checkpoint inhibition. In addition, those patients without a prior diagnosis of diabetes who were found to have significant hyperglycemia as defined by a hemoglobin A1C≥10% or a random glucose ≥300mg/dL were also referred directly to endocrinology. If patients did not meet these criteria and were planned for surgery in less than one month, they were referred to our preoperative anesthesia clinic for optimization prior to surgery. Specifically, we defined these criteria as a hemoglobin A1C between 6.5% and 9.9% and, if available, a fasting glucose >150mg/dL or a random glucose between 200mg/dL and 299mg/dL. If surgery was planned for a month or more in the future, patients with or without known diabetes who met the following criteria were referred to general internal medicine for optimization prior to surgery: a hemoglobin A1C between 6.5% and 9.9% or a fasting glucose >125mg/dL or a random glucose between 200mg/dL and 299mg/dL. This triaging system allowed us to utilize our endocrinology specialists for patients who required intensive intervention due to poorly controlled hyperglycemia without overwhelming those resources. Feedback on the feasibility of these referral algorithms was obtained from the endocrinology and preoperative anesthesia clinic to the surgical teams when deviations were noted.
Figure 1.
Algorithm for the preoperative management of patients with and without diabetes.
We also set glycemic control goals for surgical patients with diabetes to be <180mg/dL. Preoperative glucose levels were determined using capillary blood samples and hyperglycemia was treated with subcutaneous insulin. Intraoperative monitoring and correction of hyperglycemia of patients with diabetes was done every 1 to 2 hours. In the post-anesthesia care unit (PACU), glucose levels were measured for all patients who qualified for glucose monitoring prior to surgery. Patients discharged on the day of the procedure had no additional interventions. Patients admitted after surgery were given basal-bolus insulin therapy, which included basal insulin glargine (0.1 unit/kg every 12 hours), insulin lispro (0.06 units/kg prior to meals), and the use of a correctional insulin sliding scale for meals with hold parameters to improve hyperglycemia management. This basal-bolus insulin order set was imbedded into the postoperative care order sets for all surgical services to facilitate use by the clinical teams for patients with diabetes. After surgery, patients who had two or more glucose values of 180 mg/dL by postoperative day two received an endocrinology consultation to improve glycemic control.
In terms of safety and monitoring of adverse events, we tracked episodes of hypoglycemia as well as diabetic ketoacidosis. During phase one, we identified that the basal bolus order set did not provide guidance on how to use the order set with patients who had not resumed a regular diet. The order set was updated to instruct nurses to hold all insulin except for the basal insulin glargine if a patient was not ordered for a regular diet. The order set also had hold parameters to not administer insulin if the point of care glucose prior to administration was below 140mg/dL for insulin glargine or 120mg/dL for insulin lispro with meals. In addition to clarifying the order set instructions, we provided education to the inpatient nursing teams related to appropriate insulin administration and monitoring of glucose.
Phase one (October 2019-April 2020):
We held educational sessions for each department within the Division of Surgery, the Operating Room and PACU teams and the inpatient nursing staff to explain the rationale behind the changes to the institutional diabetic screening and glycemic control protocols as well as the new responsibilities of the clinical teams. We monitored compliance and shared monthly outcomes reports with all surgical and anesthesia departments. The reports included the total surgical volume by department; the number of patients with diabetes; the percentage of patients with a hemoglobin A1C result and what percentage > or < 6.5; and the number of patients with a median glucose value > or < 180 mg/dL during the intraoperative, PACU, or postoperative time periods.
Phase two (May 2020-December 2020):
The second phase included additional educational and targeted training efforts for the departments that were outliers for compliance and hyperglycemia. We also embedded the postoperative basal-bolus insulin order set into departmental order sets to encourage its use. Interim results were reviewed with departments frequently to ensure the teams’ ongoing support and engagement. During this time, because of the restrictions of the global pandemic, all education and meetings were conducted virtually. We also modified the preoperative referral algorithm to reflect feedback from the surgical teams and endocrinology consultants. These modifications were to raise the hemoglobin A1c value that requires referral to endocrinology as the preoperative anesthesia clinic became more accustomed to our goals of perioperative glycemic control; to specify random glucose values as criteria for referral to endocrinology; and to include the specific examples of patients that require referral to endocrinology such as those with hyperglycemia and a history of immune-checkpoint inhibition or a history of diabetic ketoacidosis.
Maintenance phase (1/2021-10/2022):
Beginning 1/2021, we monitored compliance of the initiative interventions and provided education to new providers as needed. Information related to the SUGAR initiative was included in orientation information for new trainees and summaries of the initiative were provided as reminders to surgical teams yearly or as needed if compliance was noted to decrease. For the purpose of data analysis, the end of these period was chosen as 10/2022. However, these efforts to maintain this quality improvement initiative have continued past that date.
Assessment:
All patients undergoing scheduled inpatient or outpatient surgery were included in this initiative. Emergency surgical patients and patients with Type 1 diabetes (including patients planned to undergo a pancreatectomy resulting in Type 1 diabetes) were excluded. We collected baseline data from October 2018 to April 2019, phase one data from October 2019 to April 2020 phase two data from May 2020 to December 2020 and maintenance phase data from January 2021 to October 2022. Simple summary statistics were used to describe glucose values by cohort and by procedure time and by cohort. We also calculated the median glucose value per procedure for each patient for each time period (preoperatively, intraoperatively, in PACU and during inpatient admission after discharge from PACU) and whether the median value was above 180mg/dL for each procedure. The average values were compared by t-test or rank sum test and proportion above 180md/gL using chi-squared test or Fisher’s exact test with a statistical significance threshold of P < .05. All statistical analysis were performed using Stata/MP v17.0 (College Station, TX). This research was in part supported by the National Institutes of Health through M.D. Anderson's Cancer Center Support Grant CA016672.
Results
The baseline period included 9891 patients, the first phase included 8815 patients, the second phase included 10,401 patients and the maintenance phase 30,410. Demographic and clinical characteristics are shown in Table 1. Within the Division of Surgery, outcomes between specialties were compared and no differences noted. Compared with the baseline group, the phase one, phase two and maintenance groups had similar percentages of patients with diabetes [14.9% (n=1470, 14.0% (n=1233) 14.7% (n=1531) and 14.0% (n=4265) [P=.08], respectively]. The percentage of patients screened for hemoglobin A1C increased during the initiative (baseline, 17.5% (n=1726); first phase, 52.5% (n=4625); second phase, 66.8% (n=6946) and maintenance phase 74.5% (n=22656) [P<.001]). The percentage of patients with diabetes who had a hemoglobin A1C test within 90 days of surgery also increased during the initiative (baseline, 62.6% (n=920); first phase, 88.7% (n=1093); second phase, 89.6% (n=1372) and maintenance phase 91.5% (n=3902) [P<.001]).
Table 1:
Demographic and clinical characteristics by cohort
| Baseline | First Phase | Second Phase | Maintenance Phase | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | N | % | N | % | N | % | N | % | P-value |
| Age | |||||||||
| N | 9891 | 8815 | 10401 | 30410 | <0.001 | ||||
| Mean (SD) | 58.9 (14.3) | 59.0 (14.3) | 58.6 (14.2) | 59.2 (14.2) | |||||
| Median (Min-Max) | 60.8 (4.4 - 99.0) | 60.9 (4.3 - 98.0) | 60.5 (5.3 - 98.2) | 61.2 (1.9 - 100.4) | |||||
| Race | |||||||||
| Asian | 470 | 4.8 | 435 | 5.0 | 486 | 4.7 | 1638 | 5.4 | <0.001 |
| African American | 810 | 8.2 | 779 | 8.9 | 976 | 9.5 | 2729 | 9.1 | |
| Other | 835 | 8.5 | 637 | 7.3 | 756 | 7.3 | 2123 | 7.1 | |
| White/Caucasian | 7712 | 78.0 | 6887 | 78.1 | 8081 | 77.7 | 23635 | 77.7 | |
| Unknown | 64 | 77 | 102 | 285 | |||||
| Ethnicity | 0.009 | ||||||||
| Hispanic or Latino | 1310 | 13.5 | 1154 | 13.4 | 1469 | 14.4 | 4322 | 14.5 | |
| Not Hispanic or Latino | 8382 | 86.5 | 7467 | 86.6 | 8716 | 85.6 | 25423 | 85.6 | |
| Unknown | 199 | 194 | 216 | 665 | |||||
| Gender | <0.001 | ||||||||
| Female | 5403 | 54.6 | 4783 | 54.3 | 6030 | 58.0 | 17731 | 58.3 | |
| Male | 4487 | 45.4 | 4032 | 45.7 | 4371 | 42.0 | 12679 | 41.7 | |
| Unknown | 1 | 0.01 | 0 | 0.00 | 0 | 0.00 | 0 | 0.00 | |
| Diabetes? | 0.085 | ||||||||
| No | 8421 | 85.1 | 7582 | 86.0 | 8870 | 85.3 | 26145 | 86.0 | |
| Yes | 1470 | 14.9 | 1233 | 14.0 | 1531 | 14.7 | 4265 | 14.0 | |
| A1C Obtained Prior to Surgery? | <0.001 | ||||||||
| No | 8165 | 82.6 | 4190 | 47.5 | 3455 | 33.2 | 7754 | 25.5 | |
| Yes | 1726 | 17.4 | 4625 | 52.5 | 6946 | 66.8 | 22656 | 74.5 | |
| A1C >6.5% for Diabetic Patients? | 0.06 | ||||||||
| No | 426 | 46.3 | 493 | 45.1 | 598 | 43.6 | 1857 | 47.6 | |
| Yes | 494 | 53.7 | 600 | 54.9 | 773 | 56.4 | 2046 | 52.4 | |
The percentage of patients with diabetes and postoperative median glucose values >180mg/dL during hospitalization decreased over time (baseline, 20.1% (n=141) ; first phase, 16.9% (n=90) ; second phase, 12.1% (n=78); and maintenance phase 13.7% (n=248) [P<.001]; Figure 2). The percentage of patients with diabetes with intraoperative median glucose values of >180mg/dL also decreased (baseline, 34.0% (n=181); first phase, 26.6% (n=140) ; second phase, 23.9% (n=164); and maintenance phase 20.3% (n=464) [P<.001]), as did the percentage of patients with median PACU glucose values >180mg/dL (baseline, 36.0% (n=385); first phase, 30.4% (n=292); second phase, 28.5% (n=334); and maintenance phase 25.8% (n=900) [P<.001)]. (Table 2)
Figure 2.
Percentage of patients with hyperglycemia over time by phase of care
Table 2.
Summary statistics of glycemic outcomes throughout the phases of care
| Baseline | First Phase | Second Phase | Maintenance phase | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Pre-Operative | N | % | N | % | N | % | N | % | p-value |
| Average glucose | 0.006 | ||||||||
| Mean (SD) | 138.1 ( 44.9) | 136.9 ( 45.5) | 133.4 ( 41.0) | 136.6 ( 39.3) | |||||
| Median (Min-Max) | 128.0 ( 57.0 - 477.0) | 126.0 ( 62.0 - 344.0) | 125.0 ( 41.0 - 437.5) | 130.0 ( 34.0 - 351.0) | |||||
| Median glucose > 180 | 0.217 | ||||||||
| No | 965 | 85.9 | 798 | 86.3 | 1043 | 88.2 | 2926 | 87.7 | |
| Yes | 159 | 14.1 | 127 | 13.7 | 139 | 11.8 | 410 | 12.3 | |
| Intra-Operative | |||||||||
| Average glucose | 0.000 | ||||||||
| Mean (SD) | 165.4 ( 45.5) | 158.4 ( 46.5) | 155.0 ( 40.4) | 152.2 ( 36.9) | |||||
| Median (Min-Max) | 163.7 ( 71.0 - 422.0) | 153.5 ( 49.0 - 334.0) | 152.0 ( 56.0 - 337.5) | 150.8 ( 38.0 - 299.0) | |||||
| Median glucose > 180 | 0.000 | ||||||||
| No | 352 | 66.0 | 386 | 73.4 | 523 | 76.1 | 1825 | 79.7 | |
| Yes | 181 | 34.0 | 140 | 26.6 | 164 | 23.9 | 464 | 20.3 | |
| PACU | |||||||||
| Average glucose | 0.000 | ||||||||
| Mean (SD) | 166.9 ( 48.6) | 161.3 ( 46.8) | 157.5 ( 44.1) | 155.3 ( 41.0) | |||||
| Median (Min-Max) | 162.0 ( 60.0 - 367.7) | 159.0 ( 44.0 - 339.0) | 154.8 ( 44.8 - 339.0) | 154.0 ( 67.0 - 399.3) | |||||
| Median glucose > 180 | 0.000 | ||||||||
| No | 685 | 64.0 | 668 | 69.6 | 838 | 71.5 | 2595 | 74.3 | |
| Yes | 385 | 36.0 | 292 | 30.4 | 334 | 28.5 | 900 | 25.7 | |
| Post-Operative | |||||||||
| Average glucose | 0.000 | ||||||||
| Mean (SD) | 155.6 ( 37.3) | 152.0 ( 34.7) | 146.4 ( 33.8) | 148.3 ( 31.9) | |||||
| Median (Min-Max) | 149.5 ( 79.9 - 298.7) | 148.3 ( 89.6 - 285.3) | 139.9 ( 73.0 - 331.6) | 143.8 ( 67.0 - 356.8) | |||||
| Median glucose > 180 | 0.000 | ||||||||
| No | 559 | 79.9 | 442 | 83.1 | 567 | 87.91 | 1565 | 86.3 | |
| Yes | 141 | 20.1 | 90 | 16.9 | 78 | 12.09 | 248 | 13.7 | |
In terms of safety, during phase one, two patients experienced transient hypoglycemia after receiving prandial insulin when not consuming a regular diet. No episodes of hypoglycemia occurred during phase two following the updated instructions in the order set and additional education. No episodes of diabetic ketoacidosis occurred among patients receiving the SUGAR protocol after surgery.
Discussion
This multidisciplinary quality-improvement project standardized screening for hyperglycemia before surgery and the management of patients with diabetes in the preoperative, intraoperative, and postoperative settings. The percentage of patients with diabetes who had a median postoperative glucose value of >180 mg/dL during hospitalization decreased by 32% in the second phase compared to baseline, exceeding our primary aim. We also screened 74.5% of patients undergoing surgery for undiagnosed diabetes or hyperglycemia, exceeding one of our secondary aims. In addition, we decreased the percentage of patients with diabetes who had a median glucose value >180 mg/dL by 40% intraoperatively and by 28% in the PACU. Although we increased the percentage of patients with diabetes with a hemoglobin A1C result within 90 days of surgery to 91.5%, this was short of our goal of 95%. Through this initiative, we improved identification of patients at risk of hyperglycemia and improved overall glycemic control for diabetic patients undergoing surgery at a high-volume cancer center. Avoiding perioperative hyperglycemia has been shown to decrease the risk of infectious complications, length of stay and need for intensive care and is recommended for optimal surgical care.[1-4, 8]
Our initiative’s strengths include its reproducibility, its applicability to all surgical specialties, and a framework that allows sustained, long-term change. By developing order sets to standardize postoperative insulin administration and imbedding these orders in postoperative order sets used for most patients, clinical teams were able to consistently follow evidence-based recommendations for management of postoperative patients with diabetes. This reduction in variability for postoperative glycemic management was critical to the success of our initiative. Another important strength was the focus on screening for patients with previously unrecognized diabetes or poorly controlled diabetes. By identifying these patients prior to surgery, surgical teams were able to avoid postoperative diabetic ketoacidosis and extreme hyperglycemia. We made incremental improvements in compliance with screening and glycemic control from the first to the second phase and have since continued these practices. Because we collected data retrospectively, we were unable to account for any confounders in our assessment of the changes in glycemic control observed during the initiative. However, we assume that the changes we observed during our initiative were directly related to the initiative, as no other programs targeting glycemic control were in place at the time. Finally, although avoiding hyperglycemia is associated with a decrease in infectious complications, we could not track postoperative infectious outcomes as part of this initiative. Another limitation is that we could not determine compliance rates with the different aspects of the intervention due to the volume of patients who received the interventions and limitations in reporting capabilities. The project team did perform random quality checks to assess if teams were using the order sets and found high compliance. Deviations in following the care algorithms were also requested to be submitted to our safety event reporting system.
In conclusion, by avoiding hyperglycemia, we believe we have decreased our patients’ risks of infectious complications following surgery and improved the quality of their care. This successfully implemented multidisciplinary perioperative algorithm to improve screening and glycemic management in patients with diabetes can be used as a model for other organizations.
Synopsis:
Glycemic control can be safely standardized and optimized in the peri-operative setting.
Acknowledgements:
We thank Laura L. Russell, scientific editor, Research Medical Library, for editing this article
This research was in part supported by the National Institutes of Health through M.D. Anderson's Cancer Center Support Grant CA016672.
Footnotes
Disclaimers: None
Prior presentations: A summary or an earlier analysis of this work has been presented at the American College of Surgery Quality Forum in 2021 and at the Institute for Healthcare Improvement Conference in 2021.
References:
- 1.Omar AS, et al. , Association of time in blood glucose range with outcomes following cardiac surgery. BMC anesthesiology, 2015. 15(1): p. 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ouattara A, et al. , Poor intraoperative blood glucose control is associated with a worsened hospital outcome after cardiac surgery in diabetic patients. The Journal of the American Society of Anesthesiologists, 2005. 103(4): p. 687–694. [DOI] [PubMed] [Google Scholar]
- 3.Shohat N, et al. , Increased postoperative glucose variability is associated with adverse outcomes following orthopaedic surgery. Bone Joint J, 2018. 100(8): p. 1125–1132. [DOI] [PubMed] [Google Scholar]
- 4.Kiran RP, et al. , The clinical significance of an elevated postoperative glucose value in nondiabetic patients after colorectal surgery: evidence for the need for tight glucose control? Annals of surgery, 2013. 258(4): p. 599–605. [DOI] [PubMed] [Google Scholar]
- 5.Van den Berghe G, et al. , Intensive Insulin Therapy in Critically Ill Patients. New England Journal of Medicine, 2001. 345(19): p. 1359–1367. [DOI] [PubMed] [Google Scholar]
- 6.Investigators, T.N.-S.S., Intensive versus Conventional Glucose Control in Critically Ill Patients. New England Journal of Medicine, 2009. 360(13): p. 1283–1297. [DOI] [PubMed] [Google Scholar]
- 7.United Stated Preventative Services Task Force, Abnormal blood glucose and type 2 diabetes mellitus: screening. October 2015. https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/screening-for-abnormal-blood-glucose-and-type-2-diabetes-october-2015. Accessed 08/24/2022. [Google Scholar]
- 8.Al-Niaimi AN, et al. , Intensive postoperative glucose control reduces the surgical site infection rates in gynecologic oncology patients. Gynecologic Oncology, 2015. 136(1): p. 71–76. [DOI] [PMC free article] [PubMed] [Google Scholar]