https://orcid.org/0009-0002-9457-2511
Figures
Abstract
Objective
The associations of glycemic variability (GV) with the occurrence of delirium and 90-day and 180-day mortality in critically ill patients with sepsis remain unclear. This study aimed to investigate the associations of glycemic variability with the development of delirium and death.
Method
This study employed a retrospective analysis of elderly sepsis patients admitted to the intensive care unit (ICU) for the first time. Patients were categorized into two groups based on their GV: a high-risk group (< 21.399%) and a low-risk group (≥21.39%). The relationship between GV and delirium was assessed via logistic regression and restricted cubic splines. Cox regression was employed to analyze the relationship between GV and patients’ 90-day and 180-day mortality.
Results
This study included 12228 elderly patients who were diagnosed with sepsis. The high-risk group presented a significantly elevated risk of delirium (27.9% vs 22.1%) and higher 90-day (40.0% vs 30.1%) and 180-day (46.0% vs 35.7%) mortality rates (all p < 0.001). The increase in GV was approximately nonlinear with increasing risk of delirium. In addition, high GV was associated with the greatest risk among delirium (OR = 1.122, 95% CI 1.023–1.230), 90-day mortality (HR = 1.149, 95% CI 1.07–1.227) and 180-day mortality (HR = 1.153, 95% CI 1.085–1.224) in elderly sepsis patients.
Citation: Huang J, Xu J, Zhai Y, Liu S, Duan P, Liang J, et al. (2026) Relationship of glycemic variability with delirium and mortality among critically ill elderly patients with sepsis: A retrospective matched cohort study. PLoS One 21(5): e0339707. https://doi.org/10.1371/journal.pone.0339707
Editor: Patricia Rezende do Prado, University of Sao Paulo: Universidade de Sao Paulo, BRAZIL
Received: August 25, 2025; Accepted: December 10, 2025; Published: May 18, 2026
Copyright: © 2026 Huang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data that supports the findings of this study is openly available at the Open Science Framework: https://osf.io/ux9qf, DOI: 10.17605/OSF.IO/FE2A6.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Sepsis represents a life-threatening condition characterized by a dysregulated host response to infection, commonly progressing to multiple organ dysfunction and elevated fatality rates [1]. Beyond its impact on short-term survival, sepsis frequently results in long-term physical and psychological illness, significantly increasing the risk of hospital readmission after discharge [2,3]. Sepsis-associated delirium (SAD) is a common and severe neurological complication characterized by acute brain dysfunction. It presents with altered mental status, impaired attention, cognitive decline, and perceptual disturbances [4]. In elderly patients, delirium frequently arises from intricate interactions among multiple factors. Within the intensive care unit (ICU), this condition is associated with a range of adverse outcomes, including prolonged hospitalization, increased mortality, and reduced quality of life [5,6]. Given its clinical burden, early identification of patients at high risk for delirium is critical. Identifying potential biomarkers associated with SAD is crucial for improving prognosis.
Contemporary research has established glycemic variability (GV) as a significant metric, offering enhanced insight into the pathophysiological consequences of disrupted metabolic homeostasis compared to static glucose measurements. Stress hyperglycemia occurs in nearly 24% of patients admitted to intensive care units and is associated with an increased risk of complications [7,8]. These findings highlight the pivotal role of glucose homeostasis in sepsis patients. Importantly, marked fluctuations in blood glucose appear to exert even more detrimental effects [9]. The guideline advocates the need for research to identify safe and effective strategies for achieving optimal glycemic control, reducing hypoglycemia, and refining glucose management across different patient populations [1]. GV serves as a metric developed by researchers to quantify fluctuations in blood glucose [10]. Current evidence demonstrates that elevated GV is associated with an increased risk of cardiovascular diseases, endothelial injury, insulin resistance, and inflammation [11–13]. Recent studies suggest that substantial glycemic fluctuations may contribute to brain injury through mechanisms such as oxidative stress, mitochondrial dysfunction, and disruption of the blood–brain barrier [14,15]. However, validation of these hypotheses in sepsis remains preliminary. Clinical evidence has centered on the associations between GV and postoperative delirium in individuals undergoing cardiac surgery, demonstrating that higher GV levels are associated with a greater risk of delirium [16,17]. Additionally, elevated GV has been linked to adverse outcomes in various acute conditions, including heart failure, acute pancreatitis, and traumatic brain injury [18–20]. While research on GV in sepsis remains limited, existing evidence indicates that high GV on the first day of sepsis is associated with early adverse outcomes [21]. Moreover, the impact of high GV on mortality appears to be more pronounced in patients with more severe sepsis [22]. These findings indicate that effective glycemic control may play an important role in mitigating the risk of poor outcomes.
Despite this, current evidence regarding the influence of glycemic variability on delirium and outcomes in elderly patients with sepsis remains limited. Given the potential value of GV, this retrospective study aimed to examine the correlation between GV and the incidence of delirium, as well as 90-day and 180-day mortality in this patient population.
2. Materials and methods
2.1. Study design, setting and population
This retrospective observational cohort study employed the Medical Information Mart for Intensive Care-IV (MIMIC-IV-3.1), which is a publicly available and comprehensive database [23]. The database includes anonymized data from 2008-- 2022, sourced from the intensive care units of the Beth Israel Deaconess Medical Center in Boston [24,25]. It comprises detailed demographics, laboratory data, nursing notes, diagnostic records, medication logs, and other critical health information.
This study included only patients who were aged ≥ 65 years, diagnosed according to the Sepsis-3 criteria [26], and were initially admitted to the ICU. We excluded the following patients: (1) whose ICU stay was less than one day; (2) whose blood glucose measurements were less than three times; and (3) who had experienced delirium before ICU admission or were not assessed for delirium during ICU stay (Fig 1). This study followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for reporting [27].
2.2. Data collection
Our researchers utilized Navicat Premium software (version 15.0.12) and structured query language (SQL) to extract data from the MIMIC-IV database. The extracted data included demographic information, vital signs, laboratory indicators, therapy, comorbidities, and disease severity scores. To reduce potential bias, variables exhibiting missing data exceeding 30% were excluded from analysis. Vital signs, laboratory indicators, disease severity scores, and therapy data were extracted from the first measurement within 24 hours after ICU admission. Comorbidities were identified primarily using ICD-9 and ICD-10 codes. The missing data rate did not exceed 30%. For variables with less than 30% missing data, multiple imputations were conducted using the ‘MICE’ package in R software, employing the random forest method.
We computed the Time-Weighted Average Glucose (TWAG) for all study participants by integrating every biochemical glucose measurement obtained during their intensive care unit admission. To address variability in blood glucose sampling intervals, TWAG was utilized as an alternative indicator of mean glycemic levels, enabling individualized patient assessment. The Gi denotes the i-th blood glucose measurement, and ΔTi represents the time interval between the i-th and (i-1)-th measurements. The formula for the TWAG is as follow: [28]
Additionally, the coefficient of variation (CV) is a blood glucose variability-related indicator used to quantify GV, defined as the ratio of the standard deviation (SD) divided by the TWAG and then times 100% [29]. Traditionally, a GV value≤36% has defined the safety threshold [30]. No established standards, however, exist for normal or abnormal GV ranges in critically ill elderly patients with sepsis [31]. Drawing from prior studies of GV in patients with sepsis and heart failure [21,32], we calculated the optimal GV cutoff value of 21.39% that can effectively distinguish the survival risk of patients with 90-day mortality as the outcome (S1 Fig in S1 File). Simultaneously, this is the median value of CV. Based on this, we divided GV into low-risk and high-risk groups.
2.3. Outcomes
The main outcome measure in this study was delirium, which was evaluated by medical staff using the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) [33]. The CAM-ICU evaluates four critical components: (1) acute mental status fluctuations; (2) inattention; (3) altered level of consciousness; and (4) disordered cognition. Delirium is confirmed when concurrent symptoms of both the first and second features, supplemented by either the third or fourth features. For the secondary outcome, we analyzed mortality during 90 and 180 days of follow-up.
2.4. Covariates extraction
The selection of covariates for this study was based on a comprehensive approach that included a review of previous studies, clinical practice experience and expertise, and data availability. Other independent variables analyzed for adjustment included age, gender, race, disease severity scores (Acute Physiology Score III (APS III), Simplified Acute Physiology Score II (SAPS II), Charlson Comorbidity Index (CCI), Glasgow Coma Scale (GCS), Sequential Organ Failure Assessment (SOFA)), laboratory indicators (albumin, blood urea nitrogen, lactate, hemoglobin, platelets, white blood cells), comorbidities (myocardial infarction, congestive heart failure, atrial fibrillation, hypertension, cerebrovascular disease, dementia, diabetes, liver disease, cancer) and treatment (vasoactive agent, ventilation, continuous renal replacement therapy (CRRT), benzodiazepines injection, the use of corticosteroid and insulin). Due to multicollinearity, the Model 3 was not included in SAPS II and CCI.
2.5. Statistical analysis
Patients were stratified into two cohorts based on GV: a low-risk group (GV < 21.39%) and a high-risk group (GV ≥ 21.39%). The normality of continuous variables was assessed via the Shapiro-Wilk test. Normally distributed data are presented as mean ± standard deviation, whereas non-normally distributed data are expressed as median with interquartile range (IQR). Categorical variables were presented as numbers and percentages. The Wilcoxon rank-sum test and the chi-square test were used to assess group differences. Univariate and multivariate logistic regression were used to assess the association between CV and the risk of delirium. Prior to multivariate logistic regression, the variance inflation factors (VIFs) were used to check for multicollinearity between variables. Restricted cubic splines (RCSs) were plotted to illustrate the association between CV and delirium. Cox proportional hazard regression was used to examine the associations between CV and 90-day mortality and 180-day mortality. Results were reported as hazard ratios (HRs) or odds ratios (ORs) with 95% confidence intervals (95% CIs). By plotting the Kaplan–Meier (K–M) survival curves, the relationship between the CV and the survival probability of elderly patients with sepsis in the ICU was investigated. The log-rank test was used to compare the curves between the high-risk groups and the low-risk groups. Statistical analyses were performed using R version 4.4.1. Subgroup analyses were conducted according to the prespecified subgroups including sex, race, myocardial infarction, atrial fibrillation, cerebrovascular disease, dementia, diabetes, vasoactive agents and ventilation.
2.6. Sensitivity analysis
To evaluate the robustness of our findings, propensity score matching (PSM) was used to mitigate potential estimation bias by adjusting for confounding factors in observational datasets. Analyses were adjusted for factors including sex, race, APS III, GCS, albumin, blood urea nitrogen, lactate, and myocardial infarction, congestive heart failure, atrial fibrillation, hypertension, cerebrovascular disease, dementia, diabetes, liver disease, cancer and vasoactive agent, ventilation, CRRT, benzodiazepines, corticosteroid, insulin, enteral nutrition, parenteral nutrition, GCS, APS III, SAPS II, SOFA, CCI. Optimal matching employing logistic regression calculated patient propensity scores through a 1:1 matching protocol. Following matching, standardized mean differences (SMDs) were computed to assess intergroup balance regarding baseline characteristics.
2.7. Ethics statement
Given that MIMIC-IV is a publicly available, de-identified dataset that typically does not require additional institutional ethical approval for secondary analysis. After completing the necessary training and assessments, our researchers have secured a licence (record ID: 13578377) to use the database.
3. Results
3.1. Sociodemographic and clinical characteristics
This study included 12228 elderly patients who were diagnosed with sepsis. Their baseline characteristics are detailed in Table 1. The median age of the cohort was 76.86 years (IQR: 70.83–83.76 years), with 5,583 patients (45.7%) being female and 8209 (67.1%) being white. Patients classified as high-risk (GV ≥ 21.39%) exhibited prolonged hospitalizations, extended ICU admissions, greater disease severity, and increased comorbidity frequency relative to their low-risk counterparts (GV < 21.39%). These high-risk patients also presented a significantly greater risk of adverse outcomes, including delirium and mortality (all p < 0.001).
3.2. GV and the risk of delirium
Among elderly sepsis patients, delirium prevalence reached 25.0% overall. A significantly higher rate was observed in high-risk individuals (27.9%) compared to their low-risk counterparts (22.1%, p < 0.001). We developed three multivariate logistic regression models to investigate the association between GV and delirium incidence. (Table 2). As a continuous variable, each 1% increase in GV was associated with a 1.390-fold increased likelihood of delirium occurrence (95% CI: 1.120–1.734). The results show that delirium risk was significantly amplified in the high-risk patients (OR: 1.122, 95% CI: 1.023–1.230). Furthermore, the restricted cubic spline curves in Fig 2 visually illustrate the linear relationship between CV and delirium risk. Before adjustment, the CV exhibited a nonlinear relationship with delirium (p < 0.001), whereas after adjustment, it also showed a nonlinear relationship with delirium (p = 0.0226), indicating that the risk of delirium increased with rising CV values above 0.23.
A for model 1 (unadjusted), B for model 3 (adjust for gender, race, bun, lactate, albumin, myocardial infarction, congestive heart failure, atrial fibrillation, hypertension, cerebrovascular disease, dementia, diabetes, liver disease, cancer, vasoactive agent, ventilation, CRRT, benzodiazepines, corticosteroid, insulin, GCS, APS III, SOFA). Abbreviation: GV glycemic variability, CV coefficient of variation, CRRT continuous renal replacement therapy, GCS Glasgow Coma Scale, APS III Acute Physiology Score III, SOFA Sequential Organ Failure Assessment.
3.3. GV and mortality
We subsequently examined the associations between GV and 90-day and 180-day mortality among the patients. The high-risk cohort exhibited a markedly elevated 90-day mortality rate compared to low-risk patients (40.0% vs 30.1%, p < 0.001). Similarly, the high-risk cohort exhibited a markedly elevated 180-day mortality rate compared to low-risk patients (46.0% vs 35.7%, p < 0.001) (Table 3). CV was independently associated with increased 90-day mortality (HR 1.296, 95% CI: 1.146–1.465) and 180-day mortality (HR 1.317, 95% CI: 1.174–1.477). Cox regression analysis revealed hazard ratios of 1.149 (95% CI: 1.077–1.227) for 90-day mortality and 1.153 (95% CI: 1.085–1.224) for 180-day mortality in the high-risk group. Furthermore, the application of Kaplan–Meier survival analysis integrated with log-rank test demonstrated statistically significant disparities in mortality at 90-day and 180-day across different GV groups (p < 0.001) (Fig 3). Additionally, within both delirious and non-delirious cohorts, significant differences in mortality rates at 90-day and 180-day were observed (p < 0.001).
A, B and C for 90-day mortality rate; D, E and F for 180-day mortality rate; A and D for elderly sepsis patients; B and E for elderly sepsis patients with delirium; C and F for elderly sepsis patients without delirium.
3.4. Subgroup analysis and sensitivity analysis
To investigate the relationship between GV and delirium across different subgroups, analyses were performed by stratifying the study population based on sex, race (white/others), myocardial infarction, heart failure, cerebrovascular disease, dementia, diabetes, vasoactive agent, and ventilation. The findings revealed no interaction effects across all eight subgroups (Fig 4). Notably, a significant interaction was observed for gender (p for interaction = 0.03). PSM was employed to achieve a better balance among covariates across groups, and Supplementary Fig 1 illustrates the corresponding changes in matching (S2 Fig in S1 File). Within the PSM-matching cohort, the high-risk group maintained substantially prolonged durations of both ICU care and hospitalization relative to low-risk counterparts (S1 Table in S1 File). Furthermore, after PSM, CV persisted as an independent predictor of delirium among elderly sepsis patients (S2 Table in S1 File), and the association between CV and mortality risk at both 90 and 180 days remained robust (S3 Table in S1 File).
4. Discussion
Delirium, characterized by acute brain dysfunction, is typified by decreased cognitive ability, reduced attention span, and altered levels of consciousness [4]. SAD is a specific form of delirium occurring during sepsis, and it is notably difficult to control once it manifests [4,34]. The incidence of delirium is intricately linked to the prognosis of sepsis patients, with severe cases potentially exacerbating illness severity, prolonging hospitalization, and increasing mortality [35]. In this study, we examined the relationship between the GV cutoff of 21.39% and the incidence of delirium and mortality in elderly patients with sepsis. This cutoff is closely aligned with the median GV values across diverse ICU and sepsis cohorts [21]. Additionally, previous studies have shown that a GV level around 22% is linked to adverse neurological outcomes, further supporting the relevance of this threshold [36]. Our findings show a significant nonlinear correlation between higher GV and an increased incidence of delirium. Further analysis revealed that the high-risk group exhibited significantly higher mortality rates during both the 90-day and 180-day follow-up periods compared to the low-risk group. In addition to its role in assessing the risk of diabetes complications, these results highlight GV may serve as a clinically meaningful biomarker for identifying elderly sepsis patients at higher risk of delirium and long-term mortality.
The pathophysiological pathways linking GV to delirium in sepsis remain incompletely understood. The systemic inflammatory response triggered by sepsis has been shown to affect brain function, leading to neuroinflammation and worsening of brain dysfunction, which in turn results in delirium [14,15]. Glucose dysregulation is thought to play an important role in this process. Hyperglycemia can intensify glucose metabolic activity, contributing to elevated reactive oxygen species generation and subsequent neuroinflammatory responses [37]. Animal studies have shown that hyperglycemia is associated with barrier disruption, leading to neuronal dysfunction and incidence of delirium [38]. Severe hypoglycemia may reduce adenosine triphosphate levels, causing neuronal hyperpolarization and resulting in delirium [39]. In addition, abnormal glucose regulation can disrupt neuronal networks, contributing to the development of delirium [6,40]. Therefore, high GV may be an important risk indicator for delirium episodes in patients with sepsis.
GV is an emerging metric that quantifies fluctuations in blood glucose levels, and is a powerful tool for evaluating the relationship between metabolic status and cognitive function [10,41,42]. In young individuals, abnormal GV harms their cognition [43]. During cardiac surgery, elevated GV has been shown to increase the risk of postoperative delirium in elderly patients [40]. Moreover, several studies have established the significant prognostic value of GV in cardiovascular diseases, underscoring its importance in clinical assessment [44–47]. Nevertheless, the relationship between GV and the occurrence of delirium in sepsis patients remains unexplored. This study is the first to assess the connection between delirium occurrence and GV in ICU elderly patients, providing novel insights into this association.
Increasing evidence suggests that elderly patients with sepsis are more susceptible to the effects of GV [21]. The incidence of delirium is is known to be associated with multiple factors, including systemic inflammation, comorbidities, the use of sedatives and vasopressors, elevated blood urea nitrogen levels, reduced albumin, hypernatremia, and hypercapnia [6]. GV is influenced by hormones such as insulin and glucocorticoids [48]. However, glucocorticoid use is not significantly associated with the occurrence of delirium in postoperative patients [49]. Nevertheless, the use of insulin or glucocorticoids may still result in episodes of hyperglycemia or hypoglycemia, both contributing to the risk of delirium [14]. Benzodiazepines have been shown to be associated with the development of delirium, and for patients at risk of delirium, dexmedetomidine is generally preferred for sedation [50,51]. Furthermore, we noted that nutritional support strategies have an impact on glycemic levels. Therefore, we conducted a sensitivity analysis, which showed that glycemic variability remained associated with the occurrence of delirium. Although continuous enteral nutrition may help stabilize glucose levels, it can still be influenced by circadian rhythms [52]. Moreover, the production of lactate during sepsis may lead to glycemic fluctuations [21]. Despite significant differences in precipitating factors of delirium across patient populations, our investigation demonstrated variation in how glycemic variability influences delirium development by sex, with women over 65 being at a higher risk of delirium. Another study also indicates that women over the age of 85 in intensive care units may be at a greater risk of developing delirium [53]. However, this result analysis needs to be cautious.
GV has demonstrated utility as a prognostic indicator across a variety of conditions, including acute pancreatitis, cerebrovascular disease, cardiovascular disease, and sepsis [20,47]. In critically ill patients, corticosteroid administration and excessive enteral nutrition can aggravate glycemic fluctuations [54]. A previous study revealed that in non-diabetic sepsis patients, increased GV on the first day of ICU admission is associated with increased in-hospital mortality [21]. Moreover, the impact of high GV on mortality in sepsis patients appears to increase with disease severity [22]. A meta-analysis has shown that elevated acute GV potentially serves as a predictor for mortality risk in patients with sepsis [31]. In our study, we found that a smaller GV was associated with relatively lower mortality among elderly patients with sepsis, suggesting that a high GV in the ICU may be associated with a poor prognosis even after ICU discharge. Recent research indicated that assessments such as the SOFA score, acute physiology score, and comorbidity score can predict the prognosis of sepsis patients [55, 56]. Higher scores generally indicate a poorer prognosis. However, severity of illness scales primarily focus on physiological parameters, GV can provide a more comprehensive metabolic assessment [57], which helps medical staff improve the prognosis of sepsis patients.
Notably, this study has several limitations. First, as a retrospective observational study, this research is inevitably subject to inherent biases and unmeasured confounding factors, which may influence the results to some extent. It is crucial to emphasize that the design of this study can only for the identification of associations, not causal relationships. Second, because this analysis relied on a single-center simulated database, patient inclusion was based exclusively on ICD codes without detailed verification of individual clinical data, which may introduce selection bias. Third, the use of data-driven medians as cutoff values for binary classification could potentially lead to bias. This cutoff was derived from a single-center cohort and may limit the generalizability of the results. Fourth, the exclusion of patients with fewer than three blood glucose measurements, less than one day of ICU stay or who were not evaluated for delirium may have contributed to selection bias. These patients may represent either more severely or less severely ill cases, as suggested by their lower monitoring frequency and shorter ICU stays. Therefore, future studies should aim to validate these findings in larger, more diverse cohorts and evaluate whether interventions targeting blood glucose fluctuations can directly reduce the incidence of delirium and mortality.
5. Conclusion
Our study indicates that higher GV is significantly associated with an increased risk of delirium and higher mortality rates at 90 and 180 days, suggesting that continuous glucose monitoring is helpful to improve the prognostic evaluation of these individuals. In addition, this study emphasizes the importance of stable control of blood glucose levels to improve the prognosis of elderly patients with sepsis.
Supporting information
S1 File. S1 Fig. Determination of the optimal cutpoint for glycemic variability.
S2 Fig. Standardized mean differences before and after propensity score matching. S1 Table. Baseline characteristics of elderly patients with sepsis in the ICU after propensity score matching. S2 Table. The association between glycemic variability and delirium after propensity score matching. S3 Table. The association between glycemic variability and mortality after propensity score matching.
https://doi.org/10.1371/journal.pone.0339707.s001
(ZIP)
References
- 1. Evans L, Rhodes A, Alhazzani W, Antonelli M, Coopersmith CM, French C, et al. Surviving sepsis campaign: International Guidelines for management of sepsis and septic shock 2021. Crit Care Med. 2021;49(11):e1063–143. pmid:34605781
- 2. Pandolfi F, Brun-Buisson C, Guillemot D, Watier L. Care pathways of sepsis survivors: Sequelae, mortality and use of healthcare services in France, 2015-2018. Crit Care. 2023;27(1):438. pmid:37950254
- 3. Fleischmann-Struzek C, Rose N, Freytag A, Spoden M, Prescott HC, Schettler A, et al. Epidemiology and costs of postsepsis morbidity, nursing care dependency, and mortality in Germany, 2013 to 2017. JAMA Netw Open. 2021;4(11):e2134290. pmid:34767025
- 4. Atterton B, Paulino MC, Povoa P, Martin-Loeches I. Sepsis associated delirium. Medicina (Kaunas). 2020;56(5):240. pmid:32443606
- 5. Bellelli G, Brathwaite JS, Mazzola P. Delirium: A marker of vulnerability in older people. Front Aging Neurosci. 2021;13:626127. pmid:33994990
- 6. Inouye SK, Westendorp RGJ, Saczynski JS. Delirium in elderly people. Lancet. 2014;383(9920):911–22. pmid:23992774
- 7. Ali Abdelhamid Y, Kar P, Finnis ME, Phillips LK, Plummer MP, Shaw JE, et al. Stress hyperglycaemia in critically ill patients and the subsequent risk of diabetes: a systematic review and meta-analysis. Crit Care. 2016;20(1):301. pmid:27677709
- 8. Rivas AM, Nugent K. Hyperglycemia, insulin, and insulin resistance in sepsis. Am J Med Sci. 2021;361(3):297–302. pmid:33500122
- 9. van Herpt TTW, van Rosmalen F, Hulsewé HPMG, van der Horst-Schrivers ANA, Driessen M, Jetten R, et al. Hyperglycemia and glucose variability are associated with worse survival in mechanically ventilated COVID-19 patients: The prospective Maastricht Intensive Care Covid Cohort. Diabetol Metab Syndr. 2023;15(1):253. pmid:38057908
- 10. Service FJ. Glucose variability. Diabetes. 2013;62(5):1398–404. pmid:23613565
- 11. Echouffo-Tcheugui JB, Zhao S, Brock G, Matsouaka RA, Kline D, Joseph JJ. Visit-to-visit glycemic variability and risks of cardiovascular events and all-cause mortality: The ALLHAT study. Diabetes Care. 2019;42(3):486–93. pmid:30659073
- 12. Shivyari FT, Pakniat H, Nooshabadi MR, Rostami S, Haghighian HK, Shiri-Shahsavari MR. Examining the oleoylethanolamide supplement effects on glycemic status, oxidative stress, inflammation, and anti-mullerian hormone in polycystic ovary syndrome. J Ovarian Res. 2024;17(1):111. pmid:38778429
- 13. Pouryousefi E, Javadi M, Hashemipour S, Nooshabadi MR, Haghighian HK. Improved glycemic status, insulin resistance and inflammation after receiving oral oleoylethanolamide supplement in people with prediabetes: A randomized controlled trial. Diabetol Metab Syndr. 2022;14(1):77. pmid:35659064
- 14. Tauber SC, Djukic M, Gossner J, Eiffert H, Brück W, Nau R. Sepsis-associated encephalopathy and septic encephalitis: An update. Expert Rev Anti Infect Ther. 2021;19(2):215–31. pmid:32808580
- 15. Gao Q, Hernandes MS. Sepsis-associated encephalopathy and blood-brain barrier dysfunction. Inflammation. 2021;44(6):2143–50. pmid:34291398
- 16. Choi H, Park CS, Huh J, Koo J, Jeon J, Kim E, et al. Intraoperative glycemic variability and mean glucose are predictors for postoperative delirium after cardiac surgery: A retrospective cohort study. Clin Interv Aging. 2022;17:79–95. pmid:35153478
- 17. Lin Y-J, Lin L-Y, Peng Y-C, Zhang H-R, Chen L-W, Huang X-Z, et al. Association between glucose variability and postoperative delirium in acute aortic dissection patients: An observational study. J Cardiothorac Surg. 2021;16(1):82. pmid:33858463
- 18. Hsu J-C, Yang Y-Y, Chuang S-L, Lin L-Y. Long-term glycemic variability predicts adverse outcomes in diabetic heart failure with preserved ejection fraction. J Clin Endocrinol Metab. 2025;110(7):1929–37. pmid:39383298
- 19. Qi L, Geng X, Feng R, Wu S, Fu T, Li N, et al. Association of glycemic variability and prognosis in patients with traumatic brain injury: A retrospective study from the MIMIC-IV database. Diabetes Res Clin Pract. 2024;217:111869. pmid:39332533
- 20. Lin L, Liang Z. Association between glycemic variability and all-cause mortality in patients with acute pancreatitis in the intensive care unit: A retrospective analysis. Dig Dis Sci. 2025;70(6):2194–203. pmid:40163289
- 21. Zhou J, Chen Z, Huang H-N, Ou C-Q, Li X. Association between various blood glucose variability-related indicators during early ICU admission and 28-day mortality in non-diabetic patients with sepsis. Diabetol Metab Syndr. 2025;17(1):22. pmid:39828689
- 22. Lu Z, Tao G, Sun X, Zhang Y, Jiang M, Liu Y, et al. Association of blood glucose level and glycemic variability with mortality in sepsis patients during ICU hospitalization. Front Public Health. 2022;10:857368. pmid:35570924
- 23. Johnson AEW, Bulgarelli L, Shen L, Gayles A, Shammout A, Horng S, et al. MIMIC-IV, a freely accessible electronic health record dataset. Scientific Data. 2023;10(1):1. pmid:36596836
- 24. Xu Y, Zheng X, Li Y, Ye X, Cheng H, Wang H, et al. Exploring patient medication adherence and data mining methods in clinical big data: A contemporary review. J Evid Based Med. 2023;16(3):342–75. pmid:37718729
- 25. Wu WT, Li YJ, Feng AZ, Li L, Huang T, Xu AD. Data mining in clinical big data: the frequently used databases, steps, and methodological models. Mil Med Res. 2021;8(1):44. pmid:34380547
- 26. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA. 2016;315(8):801–10. pmid:26903338
- 27. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. Lancet. 2007;370(9596):1453–7. pmid:18064739
- 28. Badawi O, Yeung SY, Rosenfeld BA. Evaluation of glycemic control metrics for intensive care unit populations. Am J Med Qual. 2009;24(4):310–20. pmid:19509216
- 29. Battelino T, Danne T, Bergenstal RM, Amiel SA, Beck R, Biester T, et al. Clinical targets for continuous glucose monitoring data interpretation: Recommendations from the international consensus on time in range. Diabetes Care. 2019;42(8):1593–603. pmid:31177185
- 30. ElSayed NA, Aleppo G, Aroda VR, Bannuru RR, Brown FM, Bruemmer D. Summary of revisions: Standards of care in diabetes-2023. Diabetes Care. 2023;46(Suppl 1):S5-9. pmid:36507641
- 31. Li X, Zhang D, Chen Y, Ye W, Wu S, Lou L, et al. Acute glycemic variability and risk of mortality in patients with sepsis: A meta-analysis. Diabetol Metab Syndr. 2022;14(1):59. pmid:35461267
- 32. Chun K-H, Oh J, Lee CJ, Park JJ, Lee SE, Kim M-S, et al. In-hospital glycemic variability and all-cause mortality among patients hospitalized for acute heart failure. Cardiovasc Diabetol. 2022;21(1):291. pmid:36575485
- 33. Ely EW, Inouye SK, Bernard GR, Gordon S, Francis J, May L, et al. Delirium in mechanically ventilated patients: Validity and reliability of the confusion assessment method for the intensive care unit (CAM-ICU). JAMA. 2001;286(21):2703–10. pmid:11730446
- 34. Lei W, Ren Z, Su J, Zheng X, Gao L, Xu Y, et al. Immunological risk factors for sepsis-associated delirium and mortality in ICU patients. Front Immunol. 2022;13:940779. pmid:36203605
- 35. Liu SB, Wu HY, Duan ML, Yang RL, Ji CH, Liu JJ, et al. Delirium in the ICU: How much do we know? A narrative review. Ann Med. 2024;56(1):2405072. pmid:39308447
- 36. Cai W, Li Y, Guo K, Wu X, Chen C, Lin X. Association of glycemic variability with death and severe consciousness disturbance among critically ill patients with cerebrovascular disease: Analysis of the MIMIC-IV database. Cardiovasc Diabetol. 2023;22(1):315. pmid:37974159
- 37. Bogush M, Heldt NA, Persidsky Y. Blood brain barrier injury in diabetes: Unrecognized effects on brain and cognition. J Neuroimmune Pharmacol. 2017;12(4):593–601. pmid:28555373
- 38. Rom S, Zuluaga-Ramirez V, Gajghate S, Seliga A, Winfield M, Heldt NA, et al. Hyperglycemia-driven neuroinflammation compromises BBB leading to memory loss in both diabetes mellitus (DM) Type 1 and Type 2 mouse models. Mol Neurobiol. 2019;56(3):1883–96. pmid:29974394
- 39. Sanford AM, Flaherty JH. Do nutrients play a role in delirium?. Curr Opin Clin Nutr Metab Care. 2014;17(1):45–50. pmid:24296414
- 40. van Keulen K, Knol W, Belitser SV, van der Linden PD, Heerdink ER, Egberts TCG, et al. Diabetes and glucose dysregulation and transition to delirium in ICU patients. Crit Care Med. 2018;46(9):1444–9. pmid:29957712
- 41. Suh S, Kim JH. Glycemic variability: How do we measure it and why is it important?. Diabetes Metab J. 2015;39(4):273–82. pmid:26301188
- 42. DeVries JH. Glucose variability: Where it is important and how to measure it. Diabetes. 2013;62(5):1405–8. pmid:23613566
- 43. He J, Li S, Liu F, Zheng H, Yan X, Xie Y, et al. Glycemic control is related to cognitive dysfunction in Chinese children with type 1 diabetes mellitus. J Diabetes. 2018;10(12):948–57. pmid:29671962
- 44. Zhou Z, Zhang H, Gu Y, Zhang K, Ouyang C. Relationship between glycemic variability and the incidence of postoperative atrial fibrillation following cardiac Surgery: A retrospective study from MIMIC-IV database. Diabetes Res Clin Pract. 2025;219:111978. pmid:39736333
- 45. Chen Y, Yang Z, Liu Y, Gue Y, Zhong Z, Chen T, et al. Prognostic value of glycaemic variability for mortality in critically ill atrial fibrillation patients and mortality prediction model using machine learning. Cardiovasc Diabetol. 2024;23(1):426. pmid:39593120
- 46. He H-M, Zheng S-W, Xie Y-Y, Wang Z, Jiao S-Q, Yang F-R, et al. Simultaneous assessment of stress hyperglycemia ratio and glycemic variability to predict mortality in patients with coronary artery disease: A retrospective cohort study from the MIMIC-IV database. Cardiovasc Diabetol. 2024;23(1):61. pmid:38336720
- 47. Shuai W-L, Zhang H-J, Wang N, Zhang H-C, Zeng Q-T, Wang R, et al. Association of glycemic variability with short and long-term mortality among critically ill patients with heart failure: Analysis of the MIMIC-IV database. Diabetes Res Clin Pract. 2025;221:112009. pmid:39870182
- 48. Wu Z, Liu J, Zhang D, Kang K, Zuo X, Xu Q, et al. Expert consensus on the glycemic management of critically ill patients. J Intensive Med. 2022;2(3):131–45. pmid:36789019
- 49. Liu W, Wang Y, Wang J, Shi J, Pan J, Wang D. Effects of glucocorticoids on postoperative delirium in adult patients undergoing cardiac surgery: A systematic review and meta-analysis. Clin Ther. 2021;43(9):1608–21. pmid:34548175
- 50. Burry LD, Cheng W, Williamson DR, Adhikari NK, Egerod I, Kanji S, et al. Pharmacological and non-pharmacological interventions to prevent delirium in critically ill patients: A systematic review and network meta-analysis. Intensive Care Med. 2021;47(9):943–60. pmid:34379152
- 51. Mazeraud A, Righy C, Bouchereau E, Benghanem S, Bozza FA, Sharshar T. Septic-associated encephalopathy: A comprehensive review. Neurotherapeutics. 2020;17(2):392–403. pmid:32378026
- 52.
Hiemstra FW, Stenvers DJ, Kalsbeek A, de Jonge E, van Westerloo DJ, Kervezee L. Daily variation in blood glucose levels during continuous enteral nutrition in patients on the intensive care unit: A retrospective observational study. EBioMedicine. 2024;104:105169. https://doi.org/10.1016/j.ebiom.2024.105169 38821022.
- 53. Serpytis P, Navickas P, Navickas A, Serpytis R, Navickas G, Glaveckaite S. Age- and gender-related peculiarities of patients with delirium in the cardiac intensive care unit. Kardiol Pol. 2017;75(10):1041–50. pmid:28715077
- 54. Silveira LM, Basile-Filho A, Nicolini EA, Dessotte CAM, Aguiar GCS, Stabile AM. Glycaemic variability in patients with severe sepsis or septic shock admitted to an Intensive Care Unit. Intensive Crit Care Nurs. 2017;41:98–103. pmid:28318952
- 55. Jouffroy R, Parfait PA, Gilbert B, Tourtier JP, Bloch-Laine E, Ecollan P, et al. Relationship between prehospital modified Charlson Comorbidity Index and septic shock 30-day mortality. Am J Emerg Med. 2022;60:128–33. pmid:35961123
- 56. Hwang SY, Kim IK, Jeong D, Park JE, Lee GT, Yoo J. Prognostic performance of sequential organ failure assessment, acute physiology and chronic health evaluation III, and simplified acute physiology score II scores in patients with suspected infection according to intensive care unit type. J Clin Med. 2023;12(19):6402. pmid:37835046
- 57. Pellathy TP, Pinsky MR, Hravnak M. Intensive care unit scoring systems. Crit Care Nurse. 2021;41(4):54–64. pmid:34333619