Association between various blood glucose variability-related indicators during early ICU admission and 28-day mortality in non-diabetic patients with sepsis (2025)

  • Jingyan Zhou1na1,
  • Zhiheng Chen2na1,
  • Hao-Neng Huang2,
  • Chun-Quan Ou1,2 &
  • Xin Li1

volume17, Articlenumber:22 (2025) Cite this article

  • 1 Altmetric

  • Metrics details

Abstract

Background

Various blood glucose (BG) variability-related indexes have been widely used to assess glycemic control and predict glycemic risks, but the association between BG variations and prognosis in non-diabetic patients with sepsis remains unclear.

Methods

The single-center retrospective cohort study included 7,049 non-diabetic adults with sepsis who had at least 3 records of bedside capillary point of care BG testing during the first day after ICU admission from MIMIC-IV database (2008 to 2019). Coefficient of variation and standard deviation of glucose (i.e., GluCV and GluSD), M-value, J-index, high blood glucose index (HBGI), and low blood glucose index (LBGI) were used to describe glucose variability, quality of glycemic control, and glycemic risk of patients with sepsis. The dose-response relationship between BG variability-related indexes and mortality was explored using multivariate logistic regression with restricted cubic spline (RCS) function. If the dose-response curve presented a J-shape with a specific threshold value, a linear threshold function instead of RCS would be employed.

Results

There is a J-shaped relationship between hospital mortality risk and glucose variability-related indexes in ICU patients with sepsis. The mortality risk remained relatively stable below the threshold of these indexes. However, over the threshold, the 28-day mortality risk increased by 2.82% (95% CI: 1.80–3.85%), 1.13% (95% CI: 0.66–1.60%), 1.96% (95% CI: 0.98–2.95%), 1.37% (95% CI: 0.57–2.16%), 11.19% (95% CI: 6.56–15.98%) and 39.04% (95% CI: 29.86–48.81%) for each unit increases in GluCV, GluSD, M-value, J-index, LBGI and HBGI, respectively. The effects of LBGI and HBGI on 7-day and 14-day mortality were more pronounced.

Conclusions

High levels of GluCV, GluSD, M-value, J-index, HBGI, and LBGI on the first day of ICU admission were important risk markers of hospital mortality among non-diabetic patients with sepsis.

Introduction

Sepsis is characterized by dysregulated inflammation triggered by infection and trauma, leading to a systemic inflammatory response syndrome (SIRS) [1]. This condition is estimated to affect more than 30million people worldwide annually. Blood glucose (BG) levels may serve as a significant early indicator of SIRS [2]. Both mean levels and variations of BG were associated with in-hospital mortality of patients with sepsis [3, 4]. Wen et al [5] highlighted that 40% of patients with sepsis exhibit high early BG fluctuations. Research indicates that the incidence of stress hyperglycemia could reach 75% in critically ill patients [6]. Hyperglycemia or relative insulin deficiency during critical illness may confer a predisposition to complications [7]. These findings underscore the critical role of glucose regulation in patients with sepsis [8, 9]. The 2021 Surviving Sepsis Campaign Guidelines (SSCG) encourages research that can safely achieve better glycemic control, identify lower hypoglycemia rates, and ascertain the optimal glycemic control strategies for different types of patients [1].

Though SSCG and American Diabetes Association (ADA) recommended the reference BG control ranges for patients with sepsis initiating insulin therapy, the uncertainties of glycemic management primarily focus on several critical aspects. Firstly, current guidelines for BG management in patients with sepsis mainly focus on patients with diabetes as well as all patients with sepsis. The optimal BG management for non-diabetic patients with sepsis is poorly investigated, which may be different from patients with diabetes. Secondly, determining reliable indicators of BG fluctuations and establishing an appropriate fluctuation range in non-diabetic patients with sepsis is of great clinical significance but remains under-investigated. While GluCV and GluSD have been widely used to indicate the amplitude of glucose variability, they do not adequately reflect the distribution of BG and the risk of hyperglycemia in patients with sepsis [10]. Alternative metrics, such as M-value and J-index, can simultaneously evaluate both the mean values of glucose and their dispersion as parameters [11]. High blood glucose index (HBGI) and low blood glucose index (LBGI) are used to assess the risk of hypoglycemic or hyperglycemic episodes [12]. Although the ideal range of M-value, J-index, HBGI and LBGI has been widely identified in glucose management recommendations for general patients [13], patients with insulinoma [14], patients with diabetes [15], and critically ill patients [16], they have not been incorporated into sepsis guidelines. The above indicators apply to all daily glucose profiles, including frequent self-monitoring of blood glucose (SMBG) measurements, capillary point of care (POC), and continuous glucose monitoring (CGM).

This study aims to evaluate the association between all-cause mortality and six BG variability-related indicators (GluCV, GluSD, J-index, M-value, HBGI and LBGI) measured on the first day of intensive care unit (ICU) admission in non-diabetic patients with sepsis. The findings would assist in prognostic prediction and provide some suggestions for glucose management in this patient population.

Methods

Data source

This retrospective study utilized data from the Medical Information Mart for Intensive Care IV (MIMIC-IV v2.0) database (https://mimic.mit.edu). MIMIC-IV is a single-center longitudinal database with medical records of 315,460 patients who received care at the Beth Israel Deaconess Medical Center from 2008 to 2019. The collection and use of patient information was reviewed and approved by the Institutional Review Board at the Beth Israel Deaconess Medical Center, which granted a waiver of informed consent and approved the data-sharing initiative. The authors of this study have completed the Collaborative Institutional Training Initiative program course (Certification number 49841428).

Patient selection

According to the Sepsis 3.0 definition [17], patients with sepsis were diagnosed based on the following criteria: (1) Sequential Organ Failure Assessment (SOFA) score ≥ 2 ; (2) the presence of infection or suspected infection. Suspected infection was defined as administering antibiotics within 24h before testing or 72h after testing. Patients with diabetes or septic shock were identified by the ICD code of diabetes (e.g.,‘E109’, ‘E119’, ‘E138’) and septic shock (e.g., ‘R6521’).A total of 34,677 patients with sepsis were admitted to the ICU for the first time. The inclusion criteria were as follows: (1) age ≥ 18 years; (2) ICU stay ≥ 48h; (3) at least 3 records of bedside capillary POC BG tests during the first day of ICU admission; and (4) septic patients without diabetes. Ultimately, 7,049 patients with sepsis met these criteria and were included in the study (Fig.1).

Flow chart of patient selection

Full size image

Baseline variables

Baseline variables at admission to ICU were extracted, including demographic characteristics (age, gender, race, and body mass index [BMI]); blood glucose variability-related indicators (mean, GluCV, GluSD, J-index, M-value, HBGI and LBGI); the using of drug injections (insulin, corticosteroids); disease severity scores (SOFA and Charlson Comorbidity Index); the using of treatments (mechanical ventilation[MV], continuous renal replacement therapy[CRRT]).

Blood glucose variability-related indicators

Various methods were used to assess BG status in the first 24h of ICU admission, including:

  • Direct measures of glycemic variability: GluCV and GluSD.

  • Quality of glycemic control metrics: J-index and M-value assess the quality of glycemic control by simultaneously determining the mean values of glucose and their dispersion. The J-index is highly sensitive to both mean glucose levels and glucose variations. The M-value is an index that can eliminate the amplitude correction factor in the short term. Thus, the M-value and J-index can accurately reflect a patient’s BG condition [13].

  • Predictors of glycemic risk: LBGI is the optimal predictor of hypoglycemic risk, offering a rapid assessment to prevent hypoglycemia. Conversely, HBGI indicates an acute risk of hyperglycemia. Significantly, HBGI holds clinical relevance for patients with diabetes and correlates with numerous complications [10]. An additional file shows the detailed definitions and equations of these indicators in more detail [see Additional file 1: Appendix 1].

Outcome

The primary outcome was 28-day all-cause mortality after being admitted to ICU. Secondary outcomes included 7-day and 14-day all-cause mortality.

Statistical analyses

The participants were stratified into survivor and non-survivor groups according to the primary outcome. Continuous variables were expressed as mean (SD) and were compared using a two-sample t-test between two groups. Categorical variables were expressed as frequencies (percentages) and were compared using the Chi-square test. Multivariate logistic regression was conducted to examine the relationship between each of six BG variability-related indexes and mortality, after adjusting for age, gender, race, BMI, mean BG, septic shock, drug injections (insulin, corticosteroids) used or not, treatments (MV, CRRT) used or not, and disease severity scores (i.e., SOFA and Charlson Comorbidity Index).

The dose-response relationship between the BG variability-related indexes and mortality was explored using the restricted cubic spline (RCS) function with three knots at the 25th, 50th, and 75th percentiles of the overall distribution for glucose variability-related indexes. The Wald test was used to test the nonlinearity of the relationship. If the dose-response curve presents a J-shape with a linear relationship over a specific threshold value, a linear threshold function instead of the RCS function would be employed to simplify the main model and make results more interpretable. This approach combined with the recursive algorithm helps determine the threshold and assess the linear impact of BG variability-related indicators above this threshold [18]. The impacts were expressed by excess risk (ER%) of mortality per one unit increase in the level of BG variability-related index. ER% was calculated by [odds ratio (OR)-1]*100%.

In subgroup analyses, the participants were stratified by age (≥ 65 years, < 65 years), gender (male, female), and BMI (< 25kg/m², ≥ 25kg/m²), shock (yes, no), drug (yes, no), treatments (yes, no).

Outliers, defined as values exceeding the 99th percentile or falling below the 1st percentile, were excluded from the dataset [19]. Multiple imputation was performed to reduce selection bias due to missing data, using the random forest technique with the “mice” package. All analyses were performed using R 4.3.3 (http://www.R-project.org, The R Foundation). A two-tailed P-value < 0.05 was considered statistically significant.

Results

Baseline characteristics of subjects

Table1 presents the baseline characteristics of 7,049 patients (58.45% being male). The average age of the participants was 64.01 years (SD: 17.24 years), ranging from 18 years to 100 years. Notably, all factors except for BMI exhibited significant differences between survivors and non-survivors (P < 0.05). Mean age, mean glucose levels, all BG variability-related indices, disease severity scores, the proportion of using insulin or corticosteroids, the proportion of patients with septic shock and the proportion of receiving MV or CRRT treatments were consistently higher in the non-survivor group except for the proportion of using insulin.

Full size table

Association between blood glucose variability-related indicators and mortality

There is a significant association between each of the six BG variability-related indexes and 28-day mortality (all P values < 0.05). M-value, HBGI and LBGI generally exhibited a positive linear relationship with mortality (P for nonlinearity = 0.872, 0.319, 0.201), while a J-shaped relationship was observed for GluCV, GluSD and J-index (all P values for nonlinearity < 0.05), with mortality risk increasing over a threshold (Fig.2). The thresholds were identified as 21.65%, 22.08 mmol/L, 12.55%, 21.35, 8.42 and 0.11 for GluCV, GluSD, M-value, J-index, HBGI and LBGI, respectively. These indexes had similar dose-response relationships with 14-day mortality (Additional file 1: Fig S1). For 7-day mortality (Additional file 1: Fig S2), M-value and HBGI showed significant non-linear relationships, while GluCV, GluSD, and J-index and LBGI demonstrated linear relationships.

The nonlinear relationship of glucose variability-related indicators associated with 28-day mortality after being admitted to ICU in non-diabetic patients with sepsis (A-F). (A-F): (A) GluCV (B) GluSD (C) M-value (D) J-index (E) HBGI (F) LBGI. Restricted cubic spline curve was used to determine the dose-response relationship. Solid orange line and red line represent the dose-response curve of glucose variability-related indicators and 28-day mortality in septic patients fitted by a restricted cubic spline (RCS) function and a linear threshold function, respectively. The shaded areas indicate 95% confidence intervals. The three vertical dashed lines denote the 5th, 75th, and 95th percentiles of glucose-related indicators. The horizontal dotted lines represent the odds ratios of 1. The boxplot below denotes the distribution of the glucose variability-related indicators, with the box’s left and right sides representing the first and the third quartile, and the line in the middle of the box represents the median. The thresholds for linear threshold function were 21.65%, 22.08 mmol/L, 12.55%, 21.35, 8.42 and 0.11 for GluCV, GluSD, M-value, J-index, HBGI and LBGI, respectively. P-non-linear represents the P value for examining the potential non-linear relationship between glucose variability-related indicators and mortality, with a value less than 0.05 indicating a non-linear relationship and otherwise a linear relationship

Full size image

The dose-response curves, revealed by the RCS function, demonstrated that the mortality risk remained relatively low and changed little with the levels of BG variability-related indexes when these indexes were at a low level. However, when the levels exceeded a certain level, the mortality risk almost increased linearly. Therefore, the linear threshold function was used to simplify modeling and clinical application. Figure2 presents a high degree of similarity between the two models below the 95th percentile of these indicators. However, compared to the linear threshold function, the restricted cubic spline overestimated the effects for extreme BG variability above the 95th percentile. Conservatively, we reported the effect estimates obtained by the linear threshold function. Specifically, when BG variability-related indicators exceeded their thresholds, each unit increases in GluCV, GluSD, M-value, J-index, LBGI and HBGI were associated with an increase in 28-day mortality by 2.82% (95% CI: 1.80–3.85%), 1.13% (95% CI: 0.66–1.60%), 1.96% (95% CI:0.98–2.95%), 1.37% (95% CI: 0.57–2.16%), 11.19% (95% CI: 6.56–15.98%) and 39.04% (95% CI: 29.86–48.81%), respectively. Furthermore, HBGI and LBGI had greater effects on 14-day and 7-day mortality than on 28-day mortality, indicating more intensive effects on prognosis in the shorter term (Table2).

Full size table

Subgroup analysis

All BG variability-related indicators were positively associated with 28-day mortality risk in all subgroups stratified by age, gender, except that the stratified effects of J-index were not significant in females or individuals younger than 65 years. J-index had significantly greater impacts on mortality for the elderly (i.e., ≥ 65 years) than the young. Additionally, LBGI had a greater effect on mortality for male or septic shock patients (Fig.3). The forest plots depicted glucose variability-related 14-day mortality risk (Additional file 1: Fig S3) and 7-day mortality risk (Additional file 1: Fig S4) among subgroups in non-diabetic patients with sepsis. The effects of GluSD, J-index and HBGI on 14-day mortality are greater in older adults than in younger people. J-index had significantly greater impacts on 7-day mortality for the individuals receiving CRRT or MV than the others.

Forest plot depicting glucose variability-related 28-day mortality risk in non-diabetic patients with sepsis (A-F). (A-F): (A) GluCV, (B) GluSD, (C) M-value, (D) J-index, (E) HBGI, (F) LBGI. A multivariate logistic regression was used to estimate the effect of glucose variability-related indicators, after adjustment for age, gender, race, BMI, mean BG, SOFA, Charlson Comorbidity Index, septic shock, drug (insulin and corticosteroids) and treatment (mechanical ventilation [MV], continuous renal replacement therapy [CRRT]). The blue squares denote the estimated excess risk and the blue lines present the corresponding 95% CIs. A P value for Interaction of less than 0.05 shows significant interaction between the subgroup variables and BG variability-related indicators, indicating the significant differences in excess risk between subgroups

Full size image

Discussion

This study comprehensively investigated the J-shaped relationship between mortality and various BG variability-related indicators measured in the first day of ICU admission for non-diabetic patients with sepsis. Relatively safe ranges for GluCV, GluSD, M-value, J-index, HBGI, and LBGI were identified. The mortality risk almost increased linearly when the levels exceeded thresholds. The thresholds for linear threshold function were 21.65%, 22.08 mmol/L, 12.55%, 21.35, 8.42 and 0.11 for GluCV, GluSD, M-value, J-index, HBGI and LBGI. HBGI and LBGI indicate more intensive effects on prognosis in the shorter term. These findings suggest that, in addition to absolute glucose levels, the variability and extremity of glucose fluctuations are important factors affecting patients’ prognosis. This study aligns with previous research that has emphasized the detrimental effects of glucose variability on non-diabetic patients with sepsis, and it goes further by exploring the ideal ranges and quantifying the effects in the context of sepsis.

The influence of concurrent diabetes on the prognosis of patients with sepsis is inconclusive. Krinsley et al [20] suggested that in critically ill patients, well-controlled diabetic group may benefit from increasing glucose fluctuations than non-diabetic group. However, a meta-analysis from Jiang et al [21] indicated that diabetic status did not obviously affect the outcome of patients with sepsis, and a few studies have failed to demonstrate a consistent association with long-term variations and adverse outcomes in patients with diabetes [22, 23]. Generally, the glycogen utilization and storage capacity of individuals without diabetes is normal and undamaged comparing to patients with diabetes. To date, few studies have examined the association between short-term fluctuations of BG and the prognosis of patients without diabetes in ICU. Our study indicates that, for non-diabetic patients with sepsis in ICU, 24-hour BG fluctuations were significantly associated with an increase in mortality. This may suggest that acute fluctuations in BG play a critical role in the prognosis for patients without diabetes under critical conditions who are experiencing a depletion of the normal glycogen storage capacity. Patients with sepsis may suffer from admission hyperglycemia by a mechanism of abnormal activation in various cytokines [24], vascular endothelium, and the coagulation system [25]. For critically ill patients without diabetes, enhancing the ability to respond to short-term BG fluctuations is needed in order to maintain the homeostasis of glucose. The target BG control range and optimal BG monitoring window may vary for patients with and without diabetes, which warrants further studies especially on the causal impact of glycemic variability and the underlying mechanisms. According to ADA [26], patients with diabetes require high-quality hospital care, including “Best practice” protocols, diabetes self-management education, medical nutrition therapy, insulin and noninsulin therapy. Combining the previous research with our research, we speculated that glucose control status in patients may largely determine the relationship between glucose fluctuation and mortality of patients, which indicated the necessity of an optimized treatment plan.

In terms of age difference, we also found the susceptibility of older patients to the impacts of glucose fluctuations. Possible explanations include low immunity, reduced organ reserve function, and complex comorbidities in the elderly [27, 28]. Current knowledge on ICU mortality response to the number of organ failures and the degree of organ dysfunction [29], which mainly contributed to abnormal glucose status in older patients [30]. As estrogen exerts great influence on insulin sensitivity, insulin secretion, and total glucose utilization, resulting in glucose homeostasis among women. This may explain the sexual difference in this study. Meanwhile, lactate formation during sepsis may lead to glycemic variability, while shock status is associated with hyperlactatemia [31]. It is hypothesized that in the case of poor glycemic control, patients with septic shock will have greater BG excursion than general patients with sepsis. Thus, it is necessary to address controlling BG fluctuations within safe range through appropriate treatment and dosage of drug injections.

GluCV and GluSD are straightforward and commonly used indicators of BG fluctuations. Both can be used to represent the prognosis of many diseases, including severe acute pancreatitis [32], chronic kidney disease (CKD) [33], acute coronary syndrome [34], and sepsis [3, 5, 35]. Gao et al [33] reported that in diabetes mellitus patients with CKD, the second (24-31%) and third quartiles (31-39%) of GluCV were associated with a significantly lower in-hospital mortality risk compared to the first quartile (below 24%). This study indicated that the mortality risk tended to increase once the level of GluCV exceeded 21.65% in non-diabetic patients with sepsis. This discrepancy may be due to different organ function statuses and intervention measures for patients with different diseases, leading to varying organ tolerance for BG fluctuations. This study also identified a threshold of 22.08 mmol/L for GluSD, which has not been determined in previous studies. It is confirmed that high glucose variability, as measured by GluCV and GluSD, above the thresholds was associated with increased mortality in non-diabetic patients with sepsis.

SSCG [1] recommends that BG in patients with sepsis should be controlled between 144 and 180mg/dl. When BG decreases to 80mg/dl, the use of peripheral glucose decreases, and hepatic gluconeogenesis blockade is removed. When BG decreases below 65mg/dl, glucagon secretion increases to maintain glucose homeostasis [36]. This highlights the importance of monitoring both BG fluctuations and average BG levels. M-value and J-index can integrate the information of mean glucose levels and glucose variability in a single measurement. A previous study identified an ideal reference range for general patients’ daily glucose detection of 10 to 20 for J-index [37] and M-value [38], which is slightly differed from the thresholds of 12.55 for M-value and 21.35 for J-index that we identified in non-diabetic patients with sepsis in the first day of ICU admission. Our findings suggest that special attention is needed to the physical condition of patients as well as glucose monitoring window time. Relying on one single glucose variability-related indicator was not recommended.

LBGI and HGBI have been used in CGM systems, which can predict 40–60% of severe hypoglycemic or hyperglycemic risk [39]. LBGI provides a rapid risk assessment, potentially preventing severe hypoglycemic episodes. HBGI is clinically correlated with the risk of oxidative stress, which may relate to complications. The physiological response of hyperglycemia is viewed as an adaptive immune response that promotes cellular glucose uptake [40]. Both hypoglycemia and hyperglycemia produce oxidative and inflammatory stress, which can lead to functional organ failure such as irreversible brain injury [41], cardiac rate and rhythm disturbances, and exacerbating preexisting atherosclerosis [42]. Research indicates that LBGI lower than 2.5 or HBGI below 10 indicates a low risk of severe hypoglycemia, and LBGI above 5 or HBGI above 20 indicates a high risk of severe hypoglycemia in children with Type 1 diabetes mellitus [43]. BG less than 70mg/dl [44] or more than 200mg/dl [25] at admission were associated with poor 30-day survival probability in septic patients without diabetes. This study identified a low risk of mortality for HBGI below 8.42 and LBGI below 0.11, while exceeding these levels, each one-unit increase in HBGI and LBGI was associated with an 11.19% and 39.04% increase in mortality risk, respectively. These findings emphasize the critical need to tightly control HBGI and LBGI within 24h after admission in non-diabetic patients with sepsis.

Techniques used to put BG variability-related indicators into transformation are evolving continuously, strengthening the necessity of individualized management in glucose control and glycemic variability. For example, SocialDiabetes app (SDA) allows delivering BG data to patients’ phones and adding real-time patient information of BG intervention factors, while their BG progress are monitored by professional practitioner care teams employed by the app[45]. After uploading the raw BG measurement data, the app can calculate the value of BG variability-related indicators automatically. The popularity of these mHealth apps provides valuable insights into real-world data, increasing the clinical applicability of these BG variability-related indicators. By providing SSCG and ADA guidelines on BG management in patients with sepsis, these apps may track BG dynamics, set threshold of glycemic risk episodes, and help patients with BG management. While a growing body of research demonstrate strong associations between glucose fluctuations and the prognosis of sepsis, non-pharmacological and pharmacological treatments are essential to control BG in patients with sepsis. The popularity of real-time or flash continuous glucose monitoring provides possibilities for the implementation of BG monitoring. Future interventional studies are necessary to confirm the efficacy of active control of glycemic variation within a specific level in reducing mortality risk for patients with sepsis.

The results of our study should be interpreted in light of several limitations. Firstly, due to the limited BG records in MIMIC database, eligible data conducted in our study were fewer than RCTs. Secondly, in this single-center retrospective study, patients undergo physician-dependent rather than protocol-based intensive glucose monitoring. The findings need to be further validated in prospective multi-center cohort studies. Thirdly, identifying the optimal ranges for these indicators is a complex issue. The glycemic alarm state should not be only described by any mathematical indices. The threshold values being identified through statistical modeling in this study indicated that the mortality risk increased linearly if the indicators surpass these values. Finally, the mechanism of glucose metabolism in patients with sepsis is complex and undetermined. Although this cohort study demonstrated strong associations between glycemic variability and mortality, we cannot determine the causal role of glycemic variability in the prognosis of sepsis, which warrants further interventional studies. Further studies, considering additional prognostic outcomes and incorporating clinical practice and patient characteristics, are needed to establish a recommended level of acceptable BG variation in non-diabetic patients with sepsis.

Conclusions

This study found that early BG variability, BG quality control, and glycemic risk at ICU were significantly associated with mortality risk in non-diabetic patients with sepsis. The findings emphasize the importance of monitoring and analyzing patients’ BG variability from multiple perspectives to aid clinical diagnosis and treatment decisions.

Data availability

The datasets analyzed during the current study are available in the MIMIC-IV v2.0 database, https://mimic.mit.edu.The R codes and dataset of this research were deposited in a public repository (https://github.com/9rockchen/the-additional-code-of-paper).

Abbreviations

BG:

Blood glucose

HBGI:

High blood glucose index

LBGI:

Low blood glucose index

RCS:

Restricted cubic spline

SIRS:

Systemic inflammatory response syndrome

GluCV :

Coefficient of glucose variation

GluSD :

Standard deviation of glucose

MIMIC-IV:

Medical Information Mart for Intensive Care IV

SOFA:

Sequential Organ Failure Assessment

WBC:

White blood cell count

BMI:

Body mass index

ER:

Excess risk

OR:

Odds ratio

CKD:

Chronic kidney disease

SSCG:

Surviving Sepsis Campaign Guidelines

RCTS:

Randomized controlled trials

References

  1. Evans L, Rhodes A, Alhazzani W, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. 2021;47(11):1181–247.

    Article PubMed PubMed Central Google Scholar

  2. HyunSoo Oh, EK Bae, SY Lim, et al. Temporal changes in physiological parameters of systemic inflammatory response syndrome during the three days prior to a diagnosis of sepsis: a case–control study. J Clin Nurs. 2016;25:3176–3188.

    Article Google Scholar

  3. Naeem A. Ali, James M. O’Brien Jr, Kathleen Dungan, et al. Glucose variability and mortality in patients with sepsis. Crit Care Med. 2008;36(8):2316–21.

  4. Moritoki Egi, R Bellomo, F.J.F.I.C.M., et al. Variability of blood glucose concentration and short-term mortality in critically ill patients. Anesthesiology. 2006;105:244–52.

  5. Wen–Cheng Chao, Chien-Hua Tseng, Chieh-Liang Wu, et al. Higher glycemic variability within the first day of ICU admission is associated with increased 30-day mortality in ICU patients with sepsis. Annals of Intensive Care. 2020;10(1):17.

  6. Zhengyu Jiang, Lulong Bo, Zhenhua Xu, et al. An explainable machine learning algorithm for risk factor analysis of in-hospital mortality in sepsis survivors with ICU readmission. Comput Methods Programs Biomed. 2021;204:106060.

  7. Greet Van Den B, P Wouters, F Weekers, et al. Intensive insulin therapy in critically ill patients. N Engl J Med. 2001;345:1359-67.

  8. Leonidia Leonidou A, Mouzaki M, Michalaki AL, DeLastic V, Kyriazopoulou HP, Bassaris, Gogos CA. Cytokine production and hospital mortality in patients with sepsis-induced stress hyperglycemia. J Infect. 2007;55:340–6.

    Article PubMed Google Scholar

  9. Hou N, Li M, He L, Xie B, Wang L, Zhang R, Yu Y, Sun X, Pan Z, Wang K. Predicting 30-days mortality for MIMIC-III patients with sepsis-3: a machine learning approach using XGboost. J Translational Med. 2020;18:462.

    Article CAS Google Scholar

  10. Service FJ. Glucose variability. Diabetes. 2013;62:1398–404.

  11. Mathur MB. The M-Value: a simple sensitivity analysis for Bias due to Missing Data in Treatment Effect estimates. Am J Epidemiol. 2023;192(4):612–20.

    Article PubMed Google Scholar

  12. Muñoz OM, Gómez AM, Maira G-J, L-V FabiánM, Ruiz-Morales ÁJ. The different methods of assessing glycemic variability, quality of glycemic control and glycemic risk cannot be interpreted as equivalent in clinical practice. Diabetes Metabolic Syndrome: Clin Res Reviews. 2018;12(4):555–61.

  13. McCall AL, Kovatchev BP. The median is not the only message: a clinician’s perspective on Mathematical Analysis of Glycemic variability and modeling in Diabetes Mellitus. J Diabetes Sci Technol. 2009;3(1):3–11.

    Article PubMed PubMed Central Google Scholar

  14. Gu W, Liu Y, Liu H, et al. Characteristics of glucose metabolism indexes and continuous glucose monitoring system (CGMS) in patients with insulinoma. Diabetol Metab Syndr. 2017;9:17.

    Article PubMed PubMed Central Google Scholar

  15. Heise T, Mari A, DeVries JH, et al. Effects of subcutaneous tirzepatide versus placebo or semaglutide on pancreatic islet function and insulin sensitivity in adults with type 2 diabetes: a multicentre, randomised, double-blind, parallel-arm, phase 1 clinical trial. Lancet Diabetes Endocrinol. 2022;10(6):418–29.

    Article PubMed CAS Google Scholar

  16. Qian W-Q, Jin Z-C, Cai Y, Kong X-R, Ji M-S. The effects of real-time continuous glucose monitoring on oxidative stress and mortality in critically ill patients. Zhonghua Nei Ke Za Zhi. 2013;52:30–3.

    PubMed CAS Google Scholar

  17. Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus definitions for Sepsis and septic shock (Sepsis-3). JAMA. 2016;315(8);801-810.

  18. Xiao Q, Cai B, Yin A, Huo H, Lan K, Zhou G, Shen L, He B. L-shaped association of serum 25-hydroxyvitamin D concentrations with cardiovascular and all-cause mortality in individuals with osteoarthritis: results from the NHANES database prospective cohort study. BMC Med. 2022;20:308.

  19. Kijpaisalratana N, Sanglertsinlapachai D, Techaratsami S, Musikatavorn K, Saoraya J. Machine learning algorithms for early sepsis detection in the emergency department: a retrospective study. Int J Med Informatics. 2022;160:104689.

  20. James S, Krinsley M, Egi A, et al. Diabetic status and the relation of the three domains of glycemic control to mortality in critically ill patients: an international multicenter cohort study. Crit Care. 2013;17:R37.

    Article Google Scholar

  21. Jiang L, Cheng M. Impact of diabetes mellitus on outcomes of patients with sepsis: an updated systematic review and meta-analysis. Diabetol Metab Syndr. 2022;14(1):39.

  22. El Malahi A, Van Elsen M, Charleer S, et al. Relationship between Time in Range, Glycemic Variability, HbA1c, and complications in adults with type 1 diabetes Mellitus. J Clin Endocrinol Metabolism. 2022;107(2):e570–81.

    Article Google Scholar

  23. Sebastian-Valles F, Martínez-Alfonso J, Arranz Martin JA, et al. Time above range and no coefficient of variation is associated with diabetic retinopathy in individuals with type 1 diabetes and glycated hemoglobin within target. Acta Diabetologica. 2024.

  24. Hotchkiss RS, Monneret G, Payen D. Immunosuppression in sepsis: a novel understanding of the disorder and a new therapeutic approach. Lancet Infect Dis. 2013;13(3):260–8.

    Article PubMed PubMed Central CAS Google Scholar

  25. LA. Van Vught, MA. Wiewel, PM.C. Klein Klouwenberg, et al. Admission hyperglycemia in critically ill Sepsis patients: Association with outcome and Host Response. Crit Care Med. 2016;44(7):1338–46.

    Article PubMed Google Scholar

  26. American Diabetes Association. Diabetes care in the hospital: Standards of Medical Care in Diabetes—2018. Diabetes Care. 2018;41(Supplement_1):S144-S151.

  27. Müller L, Di Benedetto S, Pawelec G. The Immune System and Its Dysregulation with Aging. Subcellular biochemistry. 2019;91:21–43.

  28. Carbajal-Guerrero J, Cayuela-Dominguez A, Fernandez-Garcia E, Aldabo-Pallas T, Marquez-Vacaro JA, Ortiz-Leyba C, Garnacho-Montero J. Epidemiology and long-term outcome of sepsis in elderly patients. Med Intensiva. 2014;38(1):21–32.

    Article PubMed CAS Google Scholar

  29. JC. Marshall, DJ Cook, NV. Christou, et al. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med. 1995;23(10):1638–1652.

  30. Mitra B, Roman C, Charters KE, O’Reilly G, Gantner D, Cameron PA. Lactate, bicarbonate and anion gap for evaluation of patients presenting with sepsis to the emergency department: a prospective cohort study. Emerg Med Australasia. 2019;32(1):20–4.

    Article Google Scholar

  31. Levy B. Lactate and shock state: the metabolic view. Curr Opin Crit Care. 2006;12:315–21.

    Article PubMed Google Scholar

  32. Zhang H, Li L, Wu J, Xu W, Wu J. Enteral Nutrition Preparations for blood glucose variability and prognosis for severe Acute Pancreatitis with stress hyperglycemia. Altern Ther Health Med. 2023;29(1):163–9.

    PubMed Google Scholar

  33. Gao M, Zhong Z, Yue Y, Liu F. Correlation between glycaemic variability and prognosis in diabetic patients with CKD. Endokrynologia Polska. 2022;73(6):947–53.

    Article PubMed CAS Google Scholar

  34. Arvia C. Conventional insulinvsinsulin infusion therapy in acute coronary syndrome diabetic patients. World J Diabetes. 2014;5(4):562–568.

  35. Lu Z, Tao G, Sun X, et al. Association of Blood Glucose Level and glycemic variability with mortality in Sepsis patients during ICU hospitalization. Front Public Health. 2022;10:857368.

  36. Hoffman RP. Sympathetic mechanisms of hypoglycemic counterregulation. Curr Diabetes Rev. 2007;3:185–93.

    Article PubMed CAS Google Scholar

  37. Wójcicki JM. J-Index. A New Proposition of the Assessment of current glucose control in Diabetic patients. Horm Metab Res. 1995;27:41–2.

  38. Wójcick JM. Mathematical descriptions of the Glucose Control in Diabetes Therapy. Analysis of the Schlichtkrull M-Value. Horm Metab Res. 1995;27(1):1–5.

  39. BORIS P, KOVATCHEV WLC, MARC BRETON, KENNETH BRAYMAN. Quantifying temporal glucose variability in diabetes via continuous glucose monitoring: Mathematical methods and clinical application. Diabetes Technol Ther. 2005;5:849–62.

    Google Scholar

  40. Losser M-R, Damoisel C, Payen D. Bench-to-bedside review: glucose and stress conditions in the intensive care unit. Crit Care. 2010;14:231.

    Article PubMed PubMed Central Google Scholar

  41. Rolf Gruetter, Kân Ugurbil, ER Seaquist. Steady-state cerebral glucose concentrations and transport in the human brain. J Neurochem. 1998;70(1):397–408.

    Article Google Scholar

  42. Fisher BM, Gillen G, Hepburn DA, Dargie HJ, Frier BM. Cardiac responses to acute insulin-induced hypoglycemia in humans. Am J Physiol. 2004;258(6 PT 2):H1775–1779.

    Google Scholar

  43. Peña NV, Torres M, Cardona JAC, Iniesta R. Impact of Telemedicine Assessment on Glycemic variability in children with type 1 diabetes Mellitus. Diabetes Technol Ther. 2013;15(2):136–42.

    Article PubMed Google Scholar

  44. Xiaoyuan Wei, Yu Min, Jiangchuan Yu, et al. Admission blood glucose is Associated with the 30-Days mortality in septic patients: a retrospective cohort study. Front Med. 2021;8:757061.

  45. Vehi J, Regincos Isern J, Parcerisas A, Calm R, Contreras I. Impact of use frequency of a Mobile Diabetes Management App on blood glucose control: evaluation study. JMIR Mhealth Uhealth. 2019;7(3):e11933.

    Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

The authors express their gratitude to the participants and investigators of the study. The authors also appreciate the MIMIC-IV database for releasing the statistics.

Funding

This study was funded by the National key research and development program intergovernmental key projects (2023YFE0114300); The Joint Funds of the Natural Science Foundation of China(No. U24A20652); National Science Foundation of China (No. 82272246); Basic and Applied Basic Research Foundation of Guangdong Province (No.2024A1515012697); Science and Technology Program of Guangzhou, China (No. 202206010044).

Author information

Author notes

  1. Jingyan Zhou and Zhiheng Chen contributed equally to this work.

Authors and Affiliations

  1. Department of Emergency Medicine, Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences), Southern Medical University, Guangdong, 510006, China

    Jingyan Zhou,Chun-Quan Ou&Xin Li

  2. State Key Laboratory of Organ Failure Research, Department of Biostatistics, Guangdong Provincial Key Laboratory of Tropical Disease Research, School of Public Health, Southern Medical University, Guangzhou, 510515, Guangdong, China

    Zhiheng Chen,Hao-Neng Huang&Chun-Quan Ou

Authors

  1. Jingyan Zhou

    View author publications

    You can also search for this author in PubMedGoogle Scholar

  2. Zhiheng Chen

    View author publications

    You can also search for this author in PubMedGoogle Scholar

  3. Hao-Neng Huang

    View author publications

    You can also search for this author in PubMedGoogle Scholar

  4. Chun-Quan Ou

    View author publications

    You can also search for this author in PubMedGoogle Scholar

  5. Xin Li

    View author publications

    You can also search for this author in PubMedGoogle Scholar

Contributions

JYZ and ZHC designed the study, analyzed and interpreted the data, and drafted the manuscript. HNH revised the manuscript.XL and CQO conceived and designed the study and revised the manuscript. The authors read and approved the final manuscript.

Corresponding authors

Correspondence to Chun-Quan Ou or Xin Li.

Ethics declarations

Ethics approval and consent to participate

Our analysis used publicly available MIMIC-IV database summary statistics. No new data were collected, and no new ethical approval was required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Association between various blood glucose variability-related indicators during early ICU admission and 28-day mortality in non-diabetic patients with sepsis (4)

Cite this article

Zhou, J., Chen, Z., Huang, HN. et al. 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 17, 22 (2025). https://doi.org/10.1186/s13098-025-01580-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13098-025-01580-4

Keywords

  • MIMIC-IV
  • Sepsis
  • Mortality
  • Glucose variability
  • Quality of glycemic control
  • Glycemic risk
Association between various blood glucose variability-related indicators during early ICU admission and 28-day mortality in non-diabetic patients with sepsis (2025)
Top Articles
Latest Posts
Recommended Articles
Article information

Author: Duncan Muller

Last Updated:

Views: 6034

Rating: 4.9 / 5 (59 voted)

Reviews: 90% of readers found this page helpful

Author information

Name: Duncan Muller

Birthday: 1997-01-13

Address: Apt. 505 914 Phillip Crossroad, O'Konborough, NV 62411

Phone: +8555305800947

Job: Construction Agent

Hobby: Shopping, Table tennis, Snowboarding, Rafting, Motor sports, Homebrewing, Taxidermy

Introduction: My name is Duncan Muller, I am a enchanting, good, gentle, modern, tasty, nice, elegant person who loves writing and wants to share my knowledge and understanding with you.