Postoperative Length of Stay in Patients With Stress Hyperglycemia Compared to Patients With Diabetic Hyperglycemia: A Retrospective Cohort Study

Abstract

Background:

Postoperative hospital length of stay (LOS) is longer in patients with diabetes than in patients without diabetes. Stress hyperglycemia (SH) in patients without a history of diabetes has been associated with adverse postoperative outcomes. The effect of SH on postoperative LOS is uncertain. The aim of this study is to compare postoperative LOS in patients with SH to patients with diabetic hyperglycemia (DH) following noncardiac surgery.

Methods:

We carried out a retrospective cohort study of inpatients with at least two glucose measurements ≥180 mg/dL. Two groups were compared. Patients with SH had no preoperative history of diabetes. Patients were considered to have DH if they had an established preoperative diagnosis of diabetes mellitus or a preoperative hemoglobin A1c (HbA1c) ≥6.5%. The primary outcome measure was hospital LOS.

Results:

We included 270 patients with postoperative hyperglycemia—82 in the SH group and 188 in the DH group. In a linear regression analysis, hospital LOS was longer in the SH group than in the DH group (10.4 vs 7.3 days; P = .03). Within the SH group, we found no association between LOS and prompt treatment of hyperglycemia within 12 hours (P = .43), insulin dose per day (P = .89), or overall mean glucose (P = .13).

Conclusions:

Postoperative LOS was even longer in patients with SH than in patients with DH, representing a potential target for quality improvement efforts. We did not, however, find evidence that improved treatment of SH was associated with reduction in LOS.

Keywords

glycemic control hospital length of stay postoperative evaluation and care stress hyperglycemia

Introduction

Postoperative hyperglycemia occurs in 15% to 75% of surgical patients, approximately one third of whom have no history of diabetes mellitus.^1-6 Transient hyperglycemia in patients without diabetes is referred to as stress hyperglycemia (SH). Stress hyperglycemia is thought to be mediated by cytokines and counter-regulatory hormones, including glucagon, cortisol, catecholamines, and growth hormone released in response to physiologic stresses, such as illness or surgery. The result is an increase in hepatic gluconeogenesis, a decrease in glucose uptake by skeletal muscles, and a proinflammatory state.⁷

The distinction between SH and diabetic hyperglycemia (DH) is an important one. Among patients with postoperative hyperglycemia, those with SH suffer worse outcomes than those with DH, including a higher risk of infection and a higher risk of mortality.^1,5 In some studies, postoperative hyperglycemia was associated with adverse outcomes only in patients with SH and not in patients with DH.^3,8 The reason for this apparent paradox is unknown, but several mechanisms have been suggested. One possibility is that a greater degree of physiologic stress is required to produce hyperglycemia in patients with SH than in those with DH.⁶ Alternatively, patients with SH may be less adapted to hyperglycemia than patients with DH.⁹ Finally, hyperglycemia may be less promptly or aggressively treated in patients with SH than it is in patients with DH, who can be anticipated to require treatment.^1-4

Hospital length of stay (LOS) is a commonly studied quality measure that has implications for patient safety and satisfaction as well as healthcare system burden and costs.¹⁰ In the published literature, there is a well-established association between diabetes and longer hospital LOS. Large population studies have shown increased hospital LOS in patients with diabetes compared to patients without diabetes.^11,12 This association appears consistent across a range of surgery types. Diabetes has been associated with an increase in LOS of approximately one day following spinal, head and neck, and a mixed group of noncardiac surgeries.^8,13-15 Furthermore, diabetes has been shown to be an independent predictor of increased postoperative LOS.^16,17 A causal link between diabetes and prolonged LOS is supported by evidence of decreased LOS with improved glycemic control in cardiac surgery patients with diabetes.^18,19 Whether or not SH is associated with prolonged postoperative LOS has not been established.

Based on studies evaluating other postoperative adverse outcomes, longer LOS might be expected in patients with SH than in patients with DH. How the increased LOS that is already known to be associated with diabetes might impact that relationship, however, is uncertain. The aim of this study is to compare postoperative LOS in patients with SH to patients with DH following noncardiac surgery.

Methods

Design, Setting, and Participants

We performed a retrospective cohort study utilizing medical records from a single institution made up of two inpatient facilities, a 718-bed academic medical center and a 91-bed community hospital. The study was approved by the hospitals’ Institutional Review Board. The requirement for informed consent was waived. We included adults at least 19 years of age who underwent a noncardiac inpatient surgical procedure between December 1, 2016 and August 31, 2017 associated with a 72 hour or longer hospital LOS. Procedures outside of the operating room were excluded. Inclusion was not otherwise restricted by procedure type. Planned and unplanned admissions were included. Patients with more than one qualifying hospitalization were eligible for the initial encounter only.

To be eligible, patients must have experienced persistent hyperglycemia (at least 2 glucose readings ≥180 mg/dL) during the surgical hospitalization. Hyperglycemia was managed by the clinical team caring for the patient. A protocol notifying providers of persistent hyperglycemia through the electronic health record was in place during the study. Order sets for the management of hyperglycemia were available but not mandated.

Variables and Data Collection

Digital data extraction from the Epic Clarity database was supplemented with manual chart review using the Epic search functionality. We collected demographic data including age, sex, race, and insurance coverage. We collected American Society of Anesthesiologists (ASA) classification which had been recorded by anesthesiology as a part of routine care. Surgical data included procedure type determined by surgical service line, procedure duration, and anesthesia method. When more than one type of anesthesia was administered, we recorded the most invasive method only (monitored sedation < neuraxial < general). Laboratory and medication data were collected from the preoperative, intraoperative, and postoperative time periods. Point of care and laboratory blood glucose values were both collected and treated as equivalent. We collected all insulin and steroid administration events. We transformed the LOS variable prior to analysis to meet the normality assumption.

Based on the standards of care set forth by the American Diabetes Association, we considered a second glucose reading ≥180 mg/dL at any point during the hospitalization to represent the development of persistent hyperglycemia.²⁰ We defined prompt treatment of hyperglycemia as the administration of any insulin formulation either before or within 12 hours after the development of persistent hyperglycemia. For each patient, we summed the doses (units) of all insulin formulations administered (intravenous and subcutaneous; scheduled or supplemental) over the entire hospitalization and divided by the patient’s LOS to determine insulin dose per day.

Statistical Analysis

Subjects were divided into two groups: hyperglycemic patients with no preoperative history of diabetes mellitus (SH) and hyperglycemic patients with a history of diabetes mellitus (DH). We used the International Classification of Diseases, Tenth Revision (ICD-10) codes, hemoglobin A1c (HbA1c) ≥6.5% during the surgical encounter or the preceding three months,²¹ and documentation of diabetes in clinical notes to identify preoperative history of diabetes. We used the Epic search functionality to query for the term “diabetes” and manually reviewed clinical notes containing that term. Patients with a diagnosis of diabetes documented by a provider were included in the DH group. We did not differentiate between type 1 and type 2 diabetes. Bivariate analysis of categorical variables was performed using the chi-square test. Bivariate analysis of continuous variables was performed using the independent sample t-test or the Wilcoxon rank-sum test. Comparisons of two continuous variables were performed using the Spearman rank correlation.

For our primary analysis, we fit a linear regression model with LOS (transformed) as the dependent variable. Diabetes status (SH or DH), procedure duration, steroid administration on the day of surgery, prompt treatment of hyperglycemia (yes or no), insulin dose per day, and mean glucose over the entire hospitalization were forced into the model. Other variables reaching a significance level of P ≤ .2 on bivariate analysis were included as candidates in the model. A best fit model was developed using backward stepwise selection with P < .05 as the criterion for retention. All analyses were carried out on SAS software version 9.4 (SAS Institute, Cary, North Carolina) using two-tailed tests.

Results

We identified 270 study patients, including 82 (30%) with no history of diabetes who were assigned to the SH group and 188 (70%) with diabetes who were assigned to the DH group. In the SH group, 31 (38%) patients had an HbA1c measurement during the surgical encounter or within the preceding three months with a mean value of 5.8%. In the DH group, 167 (89%) patients had an HbA1c measurement during the surgical encounter or within the preceding three months with a mean value of 7.7%. Table 1 displays baseline characteristics by study group. The groups were similar with regard to age, sex, and race. Patients in the DH group had more severe comorbidity as reflected by higher ASA classification (P = .001). Patients in the SH group underwent surgeries of longer duration (316.9 vs 213.2 minutes; P < .001) and were more likely to receive perioperative steroids (74% vs 60%; P = .02). A second glucose ≥180 occurred prior to the end of surgery in 3 (4%) patients in the SH group and 27 (14%) patients in the DH group. The median time from the end of surgery to a second glucose ≥180 in the remaining patients was 12.1 (interquartile range = 5.7-21.5) hours in the SH group and 16.6 (interquartile range = 4.0-20.4) hours in the DH group.

Table 1.

Baseline Characteristics.

Characteristics	SH (n = 82)	DH (n = 188)	P
Demographic information
Age, years, mean (SD)	60.8 (16.4)	62.6 (11.9)	.39
Female	37 (45.1)	95 (50.5)	.41
Race^a			.23
White or Caucasian	73 (89.0)	169 (89.9)
Black or African American	2 (2.4)	11 (5.9)
Insurance coverage^a			.67
Private insurance	27 (32.9)	59 (31.4)
Medicare or Medicaid	46 (56.1)	112 (59.6)
Self-pay	5 (6.1)	6 (3.2)
Smoking status^b			.74
Current	0 (0.0)	2 (1.9)
Former	38 (86.4)	85 (80.2)
Never	6 (13.6)	19 (17.9)
Surgical Characteristics
ASA Classification^a,b,c			.001
II	9 (12.3)	3 (1.7)
III	55 (75.3)	140 (80.0)
IV	9 (12.3)	32 (18.3)
Surgical service^a			.012
General	23 (28.1)	33 (17.6)
Neurosurgery	14 (17.1)	39 (20.7)
Transplant	12 (14.6)	27 (14.4)
Orthopedic	6 (7.3)	22 (11.7)
Surgical Oncology	15 (18.3)	13 (6.9)
Primary anesthesia method^b			.60
General	76 (95.0)	173 (93.5)
Neuraxial^d	2 (2.5)	3 (1.6)
MAC	2 (2.5)	9 (4.9)
Procedure duration, minutes, mean (SD)	316.9 (226.2)	213.2 (139.7)	< .001
Perioperative steroid^e	61 (74.4)	112 (59.6)	.02

Values represent n (%) unless otherwise noted.

Additional categories not displayed.

Missing data for: smoking status, 120 patients; ASA classification, 22 patients; primary anesthesia method, 5 patients.

No patients were classified ASA I, V, or VI.

Neuraxial comprised of spinal or epidural anesthesia.

Any steroid administration on the day of surgery.

Table 2 compares glucometrics and insulinometrics over the entire hospitalization. Patients in the SH group received less frequent glucose monitoring throughout their hospitalizations, averaging 2.1 readings per day compared to 4 readings per day in the DH group (P < .001). We found that patients in the SH group were less likely to receive insulin treatment within 12 hours of a second glucose ≥180 mg/dL (71% vs 88%; P < .001), were less likely to receive basal insulin regimens (34% vs 62%; P < .001), and received less insulin per day (3 units vs 19 units; P < .001) than patients in the DH group. Despite less frequent monitoring and less intense insulin treatment, patients in the SH group had better glucose control than patients in the DH group. Patients in the SH had an overall hospitalization average glucose of 166 mg/dL compared to 179 mg/dL in the DH group (P < .001). Patients in the SH group were also less likely to have any glucose reading < 70 mg/dL (1% vs 12%; P = .003) or any glucose reading >250 mg/dL (38% vs 59%; P = .001).

Table 2.

Glucose and Insulin, Entire Hospitalization.

Metrics	SH (n = 82)	DH (n = 188)	P
Glucose monitoring
Measurement on the day of surgery	65 (79.3)	182 (96.8)	< .001
Measurements per day, mean (SD)	2.11 (1.5)	4.04 (3.2)	< .001
Insulin
Prompt treatment of hyperglycemia^a	58 (70.7)	165 (87.8)	< .001
Insulin regimen			< .001
Basal insulin	27 (34.2)	116 (62.4)
Short-acting insulin only	37 (46.8)	64 (34.4)
No insulin	15 (19.0)	6 (3.2)
Insulin dose per day, units, mean (SD)	3.01 (6.0)	18.5 (58.1)	< .001
Glucose level, mg/dL
Average glucose, mean (SD)	165.6 (26.6)	178.7 (32.7)	< .001
Any glucose <70	1 (1.2)	23 (12.2)	.003
Any glucose ≥250	31 (37.8)	111 (59.0)	.001

Values represent n (%) unless otherwise noted.

Insulin administration before or within 12 hours following a second glucose reading ≥180 mg/dL.

Mean hospital LOS was 10.4 days (standard deviation [SD] = 9.7) in the SH group compared to 7.3 days (SD = 7.0) in the DH group. In the adjusted analysis of our primary outcome (Table 3), neither prompt treatment of hyperglycemia (P = .15) nor insulin dose per day (P = .13) were independent predictors of LOS. Patients in the SH group had longer average LOS than patients in the DH group (adjusted P = .03) after adjustment for all other predictors, including prompt treatment of hyperglycemia, insulin dose per day, and average glucose over the hospitalization. Other independent predictors of LOS were procedure duration, the insulin regimen administered on postoperative day 2, surgical service, and discharge disposition.

Table 3.

Linear Regression, Hospital Length of Stay.

Final Model Independent Variables	β coefficient	P
SH group	0.278	.03
Prompt treatment of hyperglycemia^a	0.240	.15
Insulin regimen on postoperative day 2		.03
Any continuous IV insulin	(reference)
Any long-acting insulin	-0.120
Short-acting insulin only	-0.256
No insulin	0.278
Insulin dose per day, units	-0.002	.13
Average glucose over hospitalization	-0.003	.10
Surgical service		< .001
General	(reference)
Neurosurgery	-0.381
Transplant	0.406
Orthopedics	-0.651
Surgical oncology	0.314
Other	-0.170
Procedure duration (hours)	0.099	<.001
Perioperative steroid^b	0.020	.87
Discharge disposition		< .001
Nursing facility	(reference)
Home	-0.841
Expired	-0.161

N = 268.

Insulin administration before or within 12 hours following a second glucose reading ≥180 mg/dL.

Any steroid administration on the day of surgery.

We performed an exploratory subgroup analysis of the 82 patients in our SH group. Within the SH group, LOS in the 58 patients who received prompt insulin treatment by our definition was longer at 10.9 days (SD = 11.1) that in the 24 who did not receive prompt treatment, who had an LOS of 8.6 days (SD = 4.8). This difference was in the opposite direction than might be expected but failed to reach statistical significance (P = .43). Within the SH subgroup, we found no evidence of a correlation between LOS and either insulin dose per day (P = .89) or mean glucose (P = .13).

Discussion

In patients undergoing noncardiac, inpatient procedures who developed persistent hyperglycemia, we found that the hospital LOS was longer for patients with SH than for patients for patients with DH. This is despite a lower average glucose in patients with SH. We also found that patients with SH were not treated as promptly in response to persistent hyperglycemia and received lower insulin doses. After adjustment for procedure duration, mean glucose, prompt treatment of persistent hyperglycemia, and insulin dose per day, SH remained an independent predictor of longer LOS.

Previous research has conclusively demonstrated that postoperative hyperglycemia is associated with longer hospital LOS. This effect appears to be consistent in patients undergoing cardiac surgery^18,19 and a variety of noncardiac surgeries.^8,22-24 None of these studies, however, directly compared LOS in patients with SH to patients with DH. Previous studies of intensive care unit patients and mixed medical and surgical patients have found longer LOS in patients with SH than in patients with DH.^1,25 We were able to confirm that association in a population of patients undergoing noncardiac surgery and show that SH is an independent risk factor for longer LOS in patients with postoperative hyperglycemia.

The mechanism through which SH results in a higher risk for adverse postoperative outcomes than DH has not been clarified. A 2019 study by Thompson et al⁴ found that there was no difference in the inflammatory state, as measured by C-reactive protein and interlueken-6 levels, of patients with SH compared to patients with DH following vascular surgery. Some evidence in critical care patients with SH suggests that the degree of deviation from preadmission glucose values is an important factor.^9,26 A similar relationship could be postulated for postoperative patients. Mannion et al,²⁷ in their 2021 study of patients undergoing major joint replacement, found that insulin treatment of patients with SH reduced the risk of infection. Our findings contrast somewhat with those results but are in alignment with a 2022 publication by Chen et al,⁶ who found that insulin treatment of SH within 4 hours of a glucose level >180 mg/dL did not reduce the risk of postoperative complications. In our study, we found that the increased LOS in patients with SH compared to DH could not be attributed to less aggressive insulin treatment or worse glucose control.

Limitations

We acknowledge that a number of limitations to this study should be considered. First, the retrospective design of this study allows for the possibility of unmeasured confounders. Although we adjusted for ASA classification, anesthesia method, and procedure duration, we likely failed to completely adjust for the physiologic stress of surgery. We lack a comparison group of patients without postoperative hyperglycemia, which would allow for a more complete evaluation of surgical stress and hospital LOS. We could not account for the probable existence of unrecognized postoperative hyperglycemia. This and our exclusion of patients with hospital LOS less than 72 hours may have resulted in selection bias of uncertain impact. We lack data on critical care status and on the use of enteral or parenteral nutrition, which are potential sources of bias. We acknowledge that there was likely some misclassification of patients with undiagnosed diabetes in our SH group. To minimize misclassification, we used three diabetes diagnosis criteria: billing codes, HbA1c, and documentation. This is similar to the three preadmission criteria used in a published study of undiagnosed diabetes in hospitalized patients: billing codes, HbA1c, and home anti-hyperglycemic medication.²⁸ Any misclassification may have biased our analyses toward the null. We analyzed patients with type 1 diabetes and type 2 diabetes as a single group and did not have information about their preoperative treatment regimens. We combined a wide variety of surgical procedures in our analyses. The 12 hour period we used to define prompt treatment of hyperglycemia was chosen by the authors and has not been validated. Our subgroup analysis of patients with 82 patients with SH is exploratory and had low power to show an association between prompt treatment of hyperglycemia and hospital LOS. Finally, use of data from a single institution limits generalizability.

Future Directions

Our finding that postoperative LOS is longer in patients with SH than in patients with DH is novel and needs to be confirmed in other settings and populations. If that finding were confirmed, it would add to the already strong argument that postoperative SH should be screened for in a systematic fashion. Of great importance is the further study of insulin’s role in patients with SH, particularly given the disparities in the published literature.^6,27 Because the role of insulin and the importance of glucose control are uncertain, more research investigating the mechanisms by which postoperative SH confers increased risk and how that risk might be mitigated is needed.

Conclusions

Postoperative LOS is known to be prolonged in patients with a history of diabetes. We found that postoperative LOS was even longer in patients with SH than in patients with DH. Because postoperative SH occurs commonly and because LOS has importance to both patients and healthcare systems,¹⁰ this finding has implications for future quality improvement efforts. The process measures that might be included in those quality improvement efforts are less clear, as we found no evidence that either earlier or more intense insulin treatment resulted in shorter LOS.

Footnotes

Abbreviations

ASA, American Society of Anesthesiologists; DH, diabetic hyperglycemia; HbA1c, hemoglobin A1c; LOS, length of stay; MAC, monitored anesthesia care; SD, standard deviation; SH, stress hyperglycemia.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jason Shiffermiller

References

Umpierrez

Isaacs

Bazargan

You

Thaler

Kitabchi

. 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.

Kwon

Thompson

Dellinger

Yanez

Farrohki

Flum

. Importance of perioperative glycemic control in general surgery: a report from the Surgical Care and Outcomes Assessment Program. Ann Surg. 2013;257(1):8-14. doi:10.1097/SLA.0b013e31827b6bbc.

Kotagal

Symons

Hirsch

, et al. Perioperative hyperglycemia and risk of adverse events among patients with and without diabetes. Ann Surg. 2015;261(1):97-103. doi:10.1097/SLA.0000000000000688.

Thompson

Khor

Thornblade

Flum

Sobel

. The paradox of hyperglycemia and surgical outcomes in patients with and without diabetes. Surg Infect (Larchmt). 2019;20(4):338-339. doi:10.1089/sur.2018.270.

Tarbunou

Smith

Kruse

Vogel

. Outcomes associated with hyperglycemia after abdominal aortic aneurysm repair. J Vasc Surg. 2019;69(3):763.e3-773.e3. doi:10.1016/j.jvs.2018.05.240.

Chen

Nassereldine

Cook

Thornblade

Dellinger

Flum

. Paradoxical association of hyperglycemia and surgical complications among patients with and without diabetes. JAMA Surg. 2022;157(9):765-770. doi:10.1001/jamasurg.2021.5561.

Vedantam

Poman

Motwani

Asif

Patel

Anne

. Stress-induced hyperglycemia: consequences and management. Cureus. 2022;14(7):e26714. doi:10.7759/cureus.26714.

Frisch

Chandra

Smiley

, 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.

Roberts

Krinsley

Preiser

, et al. The glycemic ratio is strongly and independently associated with mortality in the critically ill. J Diabetes Sci Technol. 2022;12:19322968221124114. doi:10.1177/19322968221124114.

10.

Tipton

Leas

Mull

, et al. Interventions To Decrease Hospital Length of Stay. AHRQ Comparative Effectiveness Technical Briefs. Rockville, MD: Agency for Healthcare Research and Quality; 2021.

11.

Fingar

Reid

Fingar

Reid

. Diabetes-Related Inpatient Stays, 2018. Rockville, MD: Agency for Healthcare Research and Quality; 2006.

12.

Comino

Harris

Islam

, et al. Impact of diabetes on hospital admission and length of stay among a general population aged 45 year or more: a record linkage study. BMC Health Serv Res. 2015;15:12. doi:10.1186/s12913-014-0666-2.

13.

Guzman

Iatridis

Skovrlj

, et al. Outcomes and complications of diabetes mellitus on patients undergoing degenerative lumbar spine surgery. Spine (Phila Pa. 1976). 2014;39(19):1596-1604. doi:10.1097/BRS.0000000000000482.

14.

Guzman

Skovrlj

Shin

, et al. The impact of diabetes mellitus on patients undergoing degenerative cervical spine surgery. Spine (Phila Pa. 1976). 2014;39(20):1656-1665. doi:10.1097/BRS.0000000000000498.

15.

Raikundalia

Fang

Spinazzi

, et al. Impact of diabetes mellitus on head and neck cancer patients undergoing surgery. Otolaryngol Head Neck Surg. 2016;154(2):294-299. doi:10.1177/0194599815607852.

16.

Yong

Weinberg

Torkamani

, et al. The presence of diabetes and higher HbA(1c) are independently associated with adverse outcomes after surgery. Diabetes Care. 2018;41(6):1172-1179. doi:10.2337/dc17-2304.

17.

Worley

Buza

Jalai

, et al. Diabetes as an independent predictor for extended length of hospital stay and increased adverse post-operative events in patients treated surgically for cervical spondylotic myelopathy. Int J Spine Surg. 2017;11(2):10. doi:10.14444/4010.

18.

Lazar

Chipkin

Fitzgerald

Bao

Cabral

Apstein

. Tight glycemic control in diabetic coronary artery bypass graft patients improves perioperative outcomes and decreases recurrent ischemic events. Circulation. 2004;109(12):1497-1502. doi:10.1161/01.CIR.0000121747.71054.

19.

Furnary

. Eliminating the diabetic disadvantage: the Portland Diabetic Project. Semin Thorac Cardiovasc Surg. 2006;18(4):302-308. doi:10.1053/j.semtcvs.2006.04.005.

20.

ElSayed

Aleppo

Aroda

, et al. 16. Diabetes care in the hospital: standards of care in diabetes-2023. Diabetes Care. 2023;46(suppl. 1):S267-S278. doi:10.2337/dc23-S016.

21.

American Diabetes Association Professional Practice Committee. 2. Classification and diagnosis of diabetes: standards of medical care in diabetes-2022. Diabetes Care. 2022;45(suppl. 1):S17-S38. doi:10.2337/dc22-S002.

22.

Ramos

Khalpey

Lipsitz

, et al. Relationship of perioperative hyperglycemia and postoperative infections in patients who undergo general and vascular surgery. Ann Surg. 2008;248(4):585-591. doi:10.1097/SLA.0b013e31818990d1.

23.

Nair

Neradilek

Newman

Horibe

. Association between acute phase perioperative glucose parameters and postoperative outcomes in diabetic and nondiabetic patients undergoing noncardiac surgery. Am J Surg. 2019;218(2):302-310. doi:10.1016/j.amjsurg.2018.10.024.

24.

Davis

Fayfman

Reyes-Umpierrez

, et al. Stress hyperglycemia in general surgery: why should we care. J Diabetes Complications. 2018;32(3):305-309. doi:10.1016/j.jdiacomp.2017.11.010.

25.

Mamtani

Kulkarni

Bihari

, et al. Degree of hyperglycemia independently associates with hospital mortality and length of stay in critically ill, nondiabetic patients: results from the ANZICS CORE binational registry. J Crit Care. 2020;55:149-156. doi:10.1016/j.jcrc.2019.11.003.

26.

Krinsley

Roberts

Brownlee

, et al. Case-control investigation of previously undiagnosed diabetes in the critically ill. J Endocr Soc. 2022;7(2):bvac180. doi:10.1210/jendso/bvac180.

27.

Mannion

Rather

Manifold

, et al. Postoperative hyperglycemia in patients with and without diabetes after major joint replacement: the impact of an enhanced glucose management program. JB JS Open Access. 2021;6(3):172. doi:10.2106/JBJS.OA.20.00172.

28.

Ren

Abusamaan

Mathioudakis

. Validation of diagnostic coding for diabetes mellitus in hospitalized patients. Endocr Pract. 2022;28(5):458-464. doi:10.1016/j.eprac.2022.01.014.