
==== Front
Ren Fail
Ren Fail
Renal Failure
0886-022X
1525-6049
Taylor & Francis

39230066
10.1080/0886022X.2024.2397555
2397555
Version of Record
Research Article
Acute Kidney Injury
The association between acute kidney injury and dysglycaemia in critically ill patients with and without diabetes mellitus: a retrospective single-center study
C. Zhang et al.
Zhang Chong abcde*
Ning Meng abcde*
Liang Weiru f*
Su Wei abcde
Chen Yi abcde
Guo Tingting abcde
Hu Kun abcde
Peng Wenjin abcde
Liu Yingwu abcde
a The Third Central Clinical College of Tianjin Medical University, Tianjin, China
b Tianjin Key Laboratory of Extracorporeal Life Support for Critical Diseases, The Third Central Hospital of Tianjin, Tianjin, China
c Artificial Cell Engineering Technology Research Center, The Third Central Hospital of Tianjin, Tianjin, China
d Tianjin Institute of Hepatobiliary Disease, The Third Central Hospital of Tianjin, Tianjin, China
e Department of Heart Center, The Third Central Hospital of Tianjin, Tianjin, China
f State Key Laboratory of Experimental Hematology, National Clinical Research Center for Blood Diseases, Haihe Laboratory of Cell Ecosystem, Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China
Supplemental data for this article can be accessed online at https://doi.org/10.1080/0886022X.2024.2397555.

* The first three authors have contributed equally to this study.

CONTACT Wenjin Peng wenjinpeng@hotmail.com
Yingwu Liu liuyingwu3zx@sina.com The Third Central Clinical College of Tianjin Medical University, Tianjin Key Laboratory of Extracorporeal Life Support for Critical Diseases, Artificial Cell Engineering Technology Research Center, Tianjin Institute of Hepatobiliary Disease, Department of Heart Center, The Third Central Hospital of Tianjin, Tianjin, China
4 9 2024
2024
4 9 2024
46 2 239755516 2 2024
20 8 2024
22 8 2024
KnowledgeWorks Global Ltd.30 8 2024
published online in a building issue30 8 2024
© 2024 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
2024
The Author(s)
https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.

Abstract

Background

Critically ill patients in the intensive care unit (ICU) often experience dysglycaemia. However, studies investigating the link between acute kidney injury (AKI) and dysglycaemia, especially in those with and without diabetes mellitus (DM), are limited.

Methods

We used the Medical Information Mart for Intensive Care IV database to investigate the association between AKI within 7 days of admission and subsequent dysglycaemia. The primary outcome was the occurrence of dysglycaemia (both hypoglycemia and hyperglycemia) after 7 days of ICU admission. Logistic regression analyzed the relationship between AKI and dysglycaemia, while a Cox proportional hazards model estimated the long-term mortality risk linked to the AKI combined with dysglycaemia.

Results

A cohort of 20,008 critically ill patients were included. The AKI group demonstrated a higher prevalence of dysglycaemia, compared to the non-AKI group. AKI patients had an increased risk of dysglycaemia (adjusted odds ratio [aOR] 1.53, 95% confidence interval [CI] 1.41–1.65), hypoglycemia (aOR 1.56, 95% CI 1.41–1.73), and hyperglycemia (aOR 1.53, 95% CI 1.41–1.66). In subgroup analysis, compared to DM patients, AKI showed higher risk of dysglycaemia in non-DM patients (aOR: 1.93 vs. 1.33, Pint<0.01). Additionally, the AKI with dysglycaemia group exhibited a higher risk of long-term mortality compared to the non-AKI without dysglycaemia group. Dysglycaemia also mediated the relationship between AKI and long-term mortality.

Conclusion

AKI was associated with a higher risk of dysglycaemia, especially in non-DM patients, and the combination of AKI and dysglycaemia was linked to higher long-term mortality. Further research is needed to develop optimal glycemic control strategies for AKI patients.

Graphical abstract

Keywords

Acute kidney injury
dysglycaemia
hypoglycemia
hyperglycemia
diabetes mellitus
long-term mortality
Beijing Yangtze River Pharmaceutical Development Foundation BYPDF2411209 This study was supported by the Beijing Yangtze River Pharmaceutical Development Foundation [Grant No. BYPDF2411209].
==== Body
pmc1. Introduction

Dysglycaemia, encompassing both hypoglycemia and hyperglycemia, is prevalent in the intensive care unit (ICU) and is associated with increased short- and long-term mortality risks [1–3]. It can be categorized into critical illness-associated dysglycaemia and dysglycaemia in patients with diabetes. One key factor influencing dysglycaemia is impaired renal function. Recent animal studies indicate that acute stress from acute kidney injury (AKI) can significantly affect systemic glucose levels, leading to impaired gluconeogenesis [4] and insulin resistance [5], which might result in hypoglycemia and hyperglycemia. Furthermore, in patients with diabetes mellitus (DM), AKI serves as a risk factor for dysglycaemia [6] and increases the risk of in-hospital mortality compared to patients without DM [7, 8]. Despite these findings, literature assessing the complex relationship between AKI and dysglycaemia in critically ill patients remains scarce, regardless of whether they have critical illness-associated dysglycaemia or preexisting diabetes. Additionally, debates persist regarding whether dysglycaemia directly contributes to higher mortality rates or merely indicates severe illness [9, 10]. Consequently, it is crucial to investigate the role of dysglycaemia in mediating the association between AKI and long-term mortality outcomes.

The Medical Information Mart for Intensive Care IV (MIMIC-IV) version 2.2 database was used to evaluate the following: (1) the association between AKI and dysglycaemia, including hypoglycemia and hyperglycemia, among critically ill patients, categorized by critical illness-associated dysglycaemia or concurrent DM; (2) the relationship between the co-occurrence of AKI and dysglycaemia and long-term mortality; and (3) the potential mediating effects of dysglycaemia on the association between AKI and long-term mortality outcomes.

2. Materials and methods

2.1. Study patients

This study conducted a retrospective analysis using data from the MIMIC-IV version 2.2 database. Managed by the Massachusetts Institute of Technology, the MIMIC-IV database comprise medical records from the Beth Israel Deaconess Medical Center (Boston, MA, USA) from 2008 to 2019. Authorized access to this database was granted to one of the authors (CZ) under Record ID 51185766. As the database is publicly accessible, this study did not require ethical approval or informed consent. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology statement.

A total of 73,181 critically ill patients aged ≥18 years were included in the initial analysis (Figure 1). The following exclusion criteria were applied: (1) not being the first hospitalization or ICU admission; (2) missing values for glycosylated hemoglobin (HbA1c); (3) experiencing dysglycaemia within the first 7 days of ICU admission; and (4) having end-stage kidney disease or undergoing kidney transplantation at ICU admission. End-stage kidney disease or undergoing kidney transplantation were identified using International Classification of Diseases (ICD) codes. After applying these criteria, 20,008 critically ill patients were included in the final analysis (cohort 1), and 19,295 hospital survivors (cohort 2) were followed up for 1 year. Mortality data for 1 year after discharge were obtained from the United States Social Security Death Index.

Figure 1. Study flow chart. Abbreviations: AKI: acute kidney injury; ICU: intensive care unit; MIMIC-IV: Medical Information Mart for Intensive Care IV.

2.2. Study variables

Data extraction from the MIMIC-IV database was performed using PostgreSQL version 14.5. The study included the following covariates: (1) Demographic characteristics: Age, sex, Sequential Organ Failure Assessment (SOFA) score, and ICU type. (2) Medical history: Presence of DM, hypertension (HTN), acute myocardial infarction (AMI), congestive heart failure (CHF), cerebrovascular disease (CVD), chronic lung disease (CLD), dementia, sepsis, chronic liver failure, and chronic kidney disease (CKD). DM was defined by an HbA1c level of ≥6.5% or a history of DM. (3) Vital signs: Systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR), recorded as mean values recorded on the first day of ICU admission. (4) Laboratory values: Estimated glomerular filtration rate (eGFR) on the first day of ICU admission was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation [11] based on the mean serum creatinine values of the first day, minimum and maximum blood glucose values during the ICU stay, and HbA1c levels during hospitalization. Blood glucose values were obtained from random measurements taken throughout the ICU stay. (5) In-hospital medications: The use of insulin and non-insulin drugs (e.g. glipizide and metformin). Comorbidities were identified using ICD codes, including ICD-9 and ICD-10.

2.3. Definitions of AKI stages and outcomes

The diagnosis of AKI was based on the Kidney Disease: Improving Global Outcomes clinical practice guidelines. These criteria include an increase in serum creatinine of ≥0.3 mg/dL (or ≥26.5 µmol/L) within 48 h, an increase to 1.5 times the baseline level within 7 days or urine output ≤0.5 mL/kg/h for 6 h. The guideline also define the stages of AKI [12]. Since AKI was diagnosed within the first 7 days of ICU admission, the primary outcome was dysglycaemia occurring after the first 7 days. Dysglycaemia was defined as hypoglycemia (at least one random blood glucose level of <70 mg/dL after the first 7 days [13]) or hyperglycemia (at least one random blood glucose level of ≥198 mg/dL after the first 7 days [14]). Dysglycaemia was further classified into the following categories: critical illness-associated hyperglycemia (CIAH, defined as hyperglycemia without DM), hyperglycemia with DM, hypoglycemia without DM, and hypoglycemia with DM. Secondary outcomes included in-hospital mortality, ICU death, 1-year post-discharge mortality among hospital survivors, ICU length of stay (LOS), and hospital LOS.

2.4. Statistical analysis

Patients were categorized into AKI and non-AKI groups. Continuous variables with a normal distribution are presented as mean ± standard deviation, while non-normally distributed variables are described using median values and interquartile ranges (25th–75th percentile). Categorical variables are presented as frequencies and percentages. Differences between groups for continuous variables were assessed using the unpaired Student’s t-test or the Mann–Whitney U-test, and differences for categorical variables were assessed using the chi-square test.

Logistic regression was employed to evaluate the relationships between AKI and dysglycaemia, in-hospital mortality, and ICU death. Multivariable multinomial logistic regression was used to assess the relationship between AKI and dysglycaemia, categorized by DM status. Due to the categorical nature of the dependent variables, relative risk ratios (RRR) were calculated to estimate outcome risks. Dysglycaemia was categorized into non-dysglycaemia, DM with dysglycaemia, and non-DM with dysglycaemia, each further subdivided into hypoglycemia and hyperglycemia. Multiple linear regression analyzed the relationship between AKI and ICU LOS and hospital LOS. A sensitivity analysis using inverse probability treatment weighting (IPTW) models, examined the association between AKI and dysglycaemia, including hypoglycemia and hyperglycemia, after excluding patients meeting any of the following criteria: (1) death within 24 h, (2) death within 48 h, (3) ICU LOS <24 h, (4) ICU LOS >30 days. For discharged patients, Kaplan–Meier survival analysis and Cox proportional hazards models assessed the association between AKI combined with dysglycaemia (non-AKI/non-dysglycaemia, AKI/non-dysglycaemia, non-AKI/dysglycaemia, and AKI/dysglycaemia) and long-term mortality. Mediation analysis, using the Stata command ‘medeff’ and bootstrapping with 1000 replications, was conducted to evaluate the mediating effects of dysglycaemia on long-term mortality associated with AKI. Subgroup analyses estimated the association between AKI and dysglycaemia across various groups, including age, sex, medical history (AMI, HTN, CHF, sepsis, and DM), and insulin therapy status. To control for bias between the AKI and non-AKI groups, covariates that were statistically significant between the two groups were included in the adjusted model. These covariates included age, sex, ICU type, SBP, DBP, HR, SOFA score, eGFR on the first day of ICU admission, and medical history (DM, HTN, AMI, CHF, CVD, CLD, dementia, sepsis, chronic liver failure, and CKD), and in-hospital medication (insulin and non-insulin drugs). Statistical analyses were performed using Stata version 17.0 (StataCorp, College Station, TX, USA), with statistical significance set at a two-tailed p-value of <0.05.

3. Results

3.1. Baseline characteristics

A cohort of 20,008 critically ill patients were analyzed. The AKI group (n = 13,448, 67.2%) comprised a higher proportion of elderly patients, and exhibited elevated SOFA scores, a greater prevalence of comorbidities (including DM, HTN, AMI, CHF, CLD, CKD, chronic liver failure, and sepsis), and a higher rate of insulin treatment in the ICU compared to the non-AKI group (Table 1). The AKI group also demonstrated higher incidence rates of dysglycaemia (40.4 vs. 25.7%), hyperglycemia (35.3 vs. 21.0%), hypoglycemia (19.0 vs. 10.7%), ICU death (4.1 vs. 1.1%), in-hospital mortality (4.6 vs. 1.5%), and longer ICU LOS [2.3 (1.3–4.3) days vs. 1.3 (0.9–2.0) days] and hospital LOS [8.2 (5.3–13.7) days vs. 4.9 (3.1–7.7) days] (Supplementary Table S1).

Table 1. Baseline characteristics between AKI group and non-AKI group.

Categories	Non-AKI group n = 6,560	AKI group n = 13,448	p Value	
Demographic characteristics	 	 	 	
 Age, year	63.9 ± 15.8	67.7 ± 14.0	<0.001	
 Sex, male, n (%)	3650 (55.6)	8127 (60.4)	<0.001	
 SOFA score	2.0 (1.0, 4.0)	4.0 (2.0, 6.0)	<0.001	
Comorbidities, n (%)	 	 	 	
 AMI	1038 (15.8)	2937 (21.8)	<0.001	
 CKD	439 (6.7)	1417 (10.5)	<0.001	
 CLD	1327 (20.2)	3418 (25.4)	<0.001	
 HTN	4193 (63.9)	9930 (73.8)	<0.001	
 DM	2076 (31.6)	5395 (40.1)	<0.001	
 CVD	1556 (23.7)	2540 (18.9)	<0.001	
 Dementia	182 (2.8)	353 (2.6)	0.54	
 CHF	992 (15.1)	4184 (31.1)	<0.001	
 Chronic liver failure	725 (11.1)	2109 (15.7)	<0.001	
 Sepsis	2027 (30.9)	8530 (63.4)	<0.001	
Laboratory values	 	 	 	
 eGFR (day 1), mL/min/1.73m2	86.0 (65.0, 100.0)	74.0 (50.0, 92.0)	<0.001	
 Maximum values for Glua, mg/dL	152.0 (129.0, 189.0)	175.0 (144.0, 244.0)	<0.001	
 Minimum values for Glub, mg/dL	87.0 (77.0, 97.0)	84.0 (73.0, 95.0)	<0.001	
 HbA1c, (%)	5.8 (5.4, 6.3)	5.9 (5.5, 6.6)	<0.001	
ICU type, n (%)	 	 	 	
 CVICU	1532 (23.4)	5364 (39.9)	<0.001	
 CCU	701 (10.7)	1570 (11.7)	 	
 MICU/SICU	3004 (45.8)	4901 (36.4)	 	
 NSICU	876 (13.4)	674 (5.0)	 	
 TSICU	447 (6.8)	939 (7.0)	 	
Vital signs	 	 	 	
 HR, bpm	81.0 (72.0, 91.0)	82.0 (74.0, 92.0)	<0.001	
 SBP, mmHg	121.0 (112.0, 134.0)	118.0 (109.0, 129.0)	<0.001	
 DBP, mmHg	65.0 (58.0, 74.0)	61.0 (55.0, 68.0)	<0.001	
Antihyperglycaemic medications, n (%)	 	 	 	
 Insulin	4012 (61.2)	10,700 (79.6)	<0.001	
 Other medications	439 (6.7)	1106 (8.2)	<0.001	
Abbreviations: AKI: acute kidney injury; AMI: acute myocardial infarction; CCU: coronary care unit; CHF: congestive heart failure; CLD: chronic pulmonary disease; CVD: cerebrovascular disease; CVICU: cardiovascular intensive care unit; DBP: diastolic blood pressure; DM: diabetes mellitus; eGFR: estimated glomerular filtration rate; Glu: glucose; HbA1c: hemoglobin A1c; HR, heart rate; HTN: hypertension; ICU: intensive care unit; MICU/SICU: medical intensive care unit/surgical intensive care unit; NSICU: neurosurgical intensive care unit; SBP: systolic blood pressure; SOFA: Sequential Organ Failure Assessment; TSICU: trauma/surgical intensive care unit.

Note: aThe maximum values for Glu during hospitalization; bthe minimum values for Glu during hospitalization.

3.2. Association between AKI and the outcomes

Logistic regression analysis revealed that higher AKI stages were associated with an increased risk of dysglycaemia, hypoglycemia, and hyperglycemia (Figure 2). Specifically, AKI was associated with a greater risk of dysglycaemia (adjusted odds ratio [aOR] 1.53, 95% confidence interval [CI] 1.41–1.65), hypoglycemia (aOR 1.56, 95% CI 1.41–1.73), hyperglycemia (aOR 1.53, 95% CI 1.41–1.66), ICU death (aOR 2.29, 95% CI 1.75–3.00), in-hospital mortality (aOR 1.59, 95% CI 1.26–2.02), longer ICU LOS (coefficient 1.17, 95% CI 1.03–1.31), and longer hospital LOS (coefficient 2.76, 95% CI 2.45–3.06) (Table 2). Furthermore, in the sensitivity analysis using the IPTW model, AKI was associated with an increased risk of dysglycaemia (Figure 3(A)), hypoglycemia (Figure 3(B)), and hyperglycemia (Figure 3(C)) even after excluding specific patients, such as those who died within 24 or 48 h, patients with ICU LOS <24 h or longer than 30 days.

Figure 2. Risk for dysglycaemia among by AKI stages. Abbreviations: AKI: acute kidney injury. Note: Non-AKI group was selected as reference in the multivariable logistic regression.

Figure 3. The association between AKI and dysglycaemia in logistic regression with sensitivity analysis and IPTW models. (A) The association between AKI and dysglycaemia; (B) the association between AKI and hypoglycemia; (C) the association between AKI and hyperglycemia. Abbreviations: AKI: acute kidney injury; CI: confidence interval; ICU: intensive care unit; IPTW: inverse probability treatment weighting; LOS: length of stay; OR: odds ratio.

Table 2. The association of AKI with outcomes.

Categories	Crude model	Adjusted model	
OR/Coefficient and 95%CI	p Value	OR/ Coefficient and 95%CI	p Value	
Dysglycaemia	1.96 (1.84–2.10)	<0.001	1.53 (1.41–1.65)	<0.001	
Hypoglycemia	1.95 (1.79–2.14)	<0.001	1.56 (1.41–1.73)	<0.001	
Hyperglycemia	2.05 (1.91–2.20)	<0.001	1.53 (1.41–1.66)	<0.001	
ICU death	3.94 (3.07–5.06)	<0.001	2.29 (1.75–3.00)	<0.001	
In-hospital mortality	3.05 (2.47–3.77)	<0.001	1.59 (1.26–2.02)	<0.001	
ICU LOSa (days)	2.30 (2.16–2.44)	<0.001	1.17 (1.03–1.31)	<0.001	
Hospital LOSa (days)	5.14 (4.85–5.44)	<0.001	2.76 (2.45–3.06)	<0.001	
Abbreviations: AKI: acute kidney injury; CI: confidence interval; ICU: intensive care unit; LOS: length of stay; OR: odds ratio.

Note: aThe ICU LOS and hospital LOS were continuous variables with multivariable linear regression. Covariates were described in the part of ‘Statistical Analysis’.

In the multivariable multinomial logistic regression analysis, AKI was associated with an increased risk of hyperglycemia compared to non-hyperglycemia. Specifically, AKI increased the risk of hyperglycemia in patients with DM (RRR 1.15, 95% CI 1.05–1.26) and CIAH (RRR 1.67, 95% CI 1.49–1.86). Regarding hypoglycemia, AKI was associated with an increased risk of DM with hypoglycemia (RRR 1.18, 95% CI 1.05–1.32) and non-DM with hypoglycemia (RRR 1.35, 95% CI 1.20–1.53) compared to non-hypoglycemia. Additionally, AKI was associated with an increased risk of dysglycaemia, including DM with dysglycaemia (RRR 1.24, 95% CI 1.13–1.37) and non-DM with dysglycaemia (RRR 1.94, 95% CI 1.72–2.18) (Table 3).

Table 3. Multinomial logistic regression for the association between AKI and DM status with dysglycaemia.

 	Crude model	Adjusted model	
RRR and 95% CI	p Value	RRR and 95% CI	p Value	
DM status combined with dysglycaemia	 	 	 	 	
DM with dysglycaemia vs. non-dysglycaemia	 	 	 	 	
 Non-AKI	Ref.	 	Ref.	 	
 AKI	1.87 (1.74–2.02)	<0.001	1.24 (1.13–1.37)	<0.001	
Non-DM with dysglycaemia vs. non-dysglycaemia	 	 	 	 	
 Non-AKI	Ref.	 	Ref.	 	
 AKI	2.16 (1.95–2.39)	<0.001	1.94 (1.72–2.18)	<0.001	
DM status combined with hypoglycemia	 	 	 	 	
DM with hypoglycemia vs. non-hypoglycemia	 	 	 	 	
 Non-AKI	Ref	 	Ref	 	
 AKI	1.94 (1.76–2.13)	<0.001	1.18 (1.05–1.32)	0.004	
Non-DM with hypoglycemia vs. non-hypoglycemia	 	 	 	 	
 Non-AKI	Ref	 	Ref	 	
 AKI	1.35 (1.21–1.49)	<0.001	1.35 (1.20–1.53)	<0.001	
DM status combined with hyperglycemia	 	 	 	 	
DM with hyperglycemia vs. non-hyperglycemia	 	 	 	 	
 Non-AKI	Ref	 	Ref	 	
 AKI	1.79 (1.67–1.92)	<0.001	1.15 (1.05–1.26)	0.003	
CIAH vs. non-hyperglycemia	 	 	 	 	
 Non-AKI	Ref	 	Ref	 	
 AKI	2.15 (1.96–2.37)	<0.001	1.67 (1.49–1.86)	<0.001	
Abbreviations: AKI: acute kidney injury; CIAH: critical illness-associated hyperglycemia; CI: confidence interval; DM: diabetes mellitus; RRR: relative risk ratio.

Note: RRR is chosen to estimate the risk of outcomes in multinomial logistic regression. The outcomes were categorized into non-dysglycaemia (including hypoglycemia and hyperglycemia), DM with dysglycaemia (including hypoglycemia and hyperglycemia), and non-DM with dysglycaemia (including hypoglycemia and CIAH). The non-dysglycaemia, non-hypoglycemia, and non-hyperglycemia were chosen as the reference outcomes in multinomial logistic regression respectively. Covariates were described in the part of ‘Statistical Analysis’.

3.3. Subgroup analysis

In cohort 1, the relationships between AKI and various subgroups were examined, including age, sex, medical history (AMI, HTN, CHF, sepsis, and DM), and insulin therapy status. For dysglycaemia, AKI exhibited significant interaction effects with sepsis (non-sepsis: aOR 1.18, 95% CI 1.06–1.32; sepsis: aOR 2.05, 95% CI 1.83–2.30; Pint<0.001), DM (non-DM: aOR 1.93, 95% CI 1.71–2.18; DM: aOR 1.33, 95% CI 1.18–1.51; Pint<0.001), and insulin therapy (no insulin therapy: OR 1.81, 95% CI 1.56–2.10; insulin therapy: OR 1.40, 95% CI 1.28–1.54; Pint=0.001) (Figure 4(A)). In terms of hyperglycemia, AKI demonstrated interaction effects with sepsis (non-sepsis: OR 1.13, 95% CI 1.01–1.27; sepsis: OR 2.10, 95% CI 1.87–2.37; Pint<0.001), DM (non-DM: OR 2.26, 95% CI 1.96–2.61; DM: OR 1.31, 95% CI 1.16–1.48; Pint<0.001), and insulin therapy (no insulin therapy: OR 2.05, 95% CI 1.72–2.44; insulin therapy: OR 1.40, 95% CI 1.28–1.54; Pint<0.001) (Figure 4(B)). For hypoglycemia, AKI demonstrated interaction effects with sepsis (non-sepsis: OR 1.21, 95% CI 1.04–1.41; sepsis: OR 2.00, 95% CI 1.73–2.31; Pint<0.001) (Figure 4(C)).

Figure 4. Subgroup analysis. (A) Subgroup analysis for dysglycaemia; (B) subgroup analysis for hyperglycemia; (C) subgroup analysis for hypoglycemia. The statistical significance was defined as p for interaction <0.05. Abbreviations: AKI: acute kidney injury; AMI: acute myocardial infarction; CHF: congestive heart failure; CI: confidence interval; DM: diabetes mellitus; HTN: hypertension; OR: odds ratio.

3.4. Survival analysis for AKI combined with dysglycaemia

In cohort 2, 19,295 hospital survivors were followed for 1 year, during which 1,640 patients experienced post-discharge mortality. The Cox proportional hazards models revealed a progressively increasing risk of 1-year post-discharge mortality, ranging from non-AKI/non-dysglycaemia to AKI/dysglycaemia (Table 4). Kaplan–Meier survival analysis corroborated these findings, indicating that the AKI/dysglycaemia group exhibited the highest 1-year post-discharge mortality risk compared to the non-AKI/non-dysglycaemia group (Supplementary Figure S1).

Table 4. The association between AKI combined with dysglycaemia and 1-year post-discharge mortality.

Categories	Crude model	Adjusted model	
HR and 95% CI	p Value	HR and 95% CI	p Value	
AKI combined with dysglycaemia	 	 	 	 	
 Non-AKI/non-dysglycaemia	Ref.	 	Ref.	 	
 AKI/non-dysglycaemia	1.33 (1.15–1.55)	<0.001	1.15 (0.98–1.34)	0.086	
 Non-AKI/dysglycaemia	1.80 (1.48–2.20)	<0.001	1.66 (1.36–2.04)	<0.001	
 AKI/dysglycaemia	2.65 (2.30–3.07)	<0.001	1.74 (1.48–2.05)	<0.001	
AKI combined with hypoglycemia	 	 	 	 	
 Non-AKI/non-hypoglycemia	Ref.	 	Ref.	 	
 AKI/non-hypoglycemia	1.43 (1.27–1.62)	<0.001	1.13 (0.99–1.29)	0.071	
 Non-AKI/hypoglycemia	1.22 (0.91–1.63)	0.179	1.14 (0.85–1.52)	0.381	
 AKI/hypoglycemia	2.18 (1.88–2.53)	<0.001	1.35 (1.14–1.60)	<0.001	
AKI combined with hyperglycemia	 	 	 	 	
 Non-AKI/non-hyperglycemia	Ref.	 	Ref.	 	
 AKI/non-hyperglycemia	1.44 (1.24–1.66)	<0.001	1.28 (1.10–1.48)	0.001	
 Non-AKI/hyperglycemia	1.94 (1.58–2.38)	<0.001	1.60 (1.29–1.98)	<0.001	
 AKI/hyperglycemia	2.62 (2.27–3.02)	<0.001	1.65 (1.40–1.95)	<0.001	
Abbreviations: AKI: acute kidney injury; CI: confidence interval; HR: hazard ratio.

Note: Covariates were described in the part of ‘Statistical Analysis’.

Furthermore, dysglycaemia contributed to 35.8% of the increased risk of long-term mortality associated with AKI. Within this subset, hypoglycemia and hyperglycemia were responsible for 16.4 and 32.4% of the increased risk, ­respectively (Figure 5).

Figure 5. Mediation analysis. (A) Mediating effect of dysglycaemia and long-term mortality associated with AKI; (B) mediating effect of hypoglycemia and long-term mortality associated with AKI; (C) mediating effect of hyperglycemia and long-term mortality associated with AKI. Path A, the effect of AKI level on dysglycaemia; path B, the effect of dysglycaemia on long-term mortality; path C, the total effect of AKI on long-term mortality; path C’, the direct effect of AKI on long-term mortality after controlling dysglycaemia. Abbreviations: ACME: average causal mediating effect; ADE: average direct effect; AKI: acute kidney injury; TE: total effect.

4. Discussion

This study demonstrated that AKI increases the risk of subsequent dysglycemia, encompassing both hyperglycemia and hypoglycemia, particularly in patients without DM. AKI combined with dysglycaemia is associated with the highest risk of long-term mortality. Dysglycaemia proves to be a valuable factor for risk stratification in critically ill patients with AKI. Moreover, mediation reveals that dysglycaemia plays a mediating role in the relationship between AKI and long-term mortality.

Previous studies have indicated that critically ill patients with dysglycaemia are at an increased risk of developing AKI [5]. However, research on the impact of AKI on subsequent dysglycemia remains limited [15]. Furthermore, the relationship between AKI and dysglycaemia in patients with and without diabetes remains unclear. An observational study by Basi et al. involving 90 patients with acute renal failure, found an increased risk of insulin resistance and hyperglycemia, which were associated with higher in-hospital mortality [16]. Our study corroborates these findings by demonstrating that the association between AKI and subsequent hyperglycemia is stronger in patients without DM compared to those with DM [16]. Similarly, Hung et al. studied 65,151 patients with AKI and found that AKI significantly increased the risk of hypoglycemia within 90 days after discharge [17]. Carreira et al. examined 478 patients with diabetes, including 239 with AKI, and identified AKI as a significant risk factor for hypoglycemia during hospitalization [6]. Our findings also indicate that the association between AKI and risk of subsequent dysglycemia is stronger in non-DM patients with AKI compared to those with DM. This is potentially due to several pathophysiological mechanisms. First, AKI often leads to insulin resistance, which can result from impaired kidney metabolic function and severe renal tissue damage, thereby increasing the risk of hyperglycemia [5, 16]. Second, critically ill patients with hyperglycemia frequently receive insulin treatment to maintain target glycaemic levels (80–110 mg/dL) during ICU stays, which can heighten the risk of hypoglycemia [5]. Our subgroup analysis indicated that AKI increased the risk of hypoglycemia, especially in patients receiving insulin treatment during hospitalization, reinforcing this finding. Third, the renal cortex contributes to 15–30% of total body gluconeogenesis and plays a role in insulin metabolism and clearance. AKI can impair renal gluconeogenesis and disrupt insulin metabolism, increasing the risk of hypoglycemia [5].

Furthermore, both CIAH and DM with hyperglycemia are associated with increased short- and long-term mortality, likely due to the role of hyperglycemia in promoting inflammation and endothelial dysfunction, which can exacerbate critical illness [3, 18]. Conversely, hypoglycemia might reflect the severity of the illness, as it often results from increased glucose consumption in macrophage-rich tissues such as the gut and liver, potentially worsening the patient’s condition and increasing mortality risk [19]. Considering dysglycaemia is a known risk factor for mortality, it might play a crucial role in risk stratification for patients with AKI. Our analysis examined the impact of AKI combined with dysglycaemia on long-term mortality and found that the AKI/dysglycaemia group had the highest associated risk. Mediation analysis further showed that dysglycaemia mediates the relationship between AKI and long-term mortality, underscoring its significant effect on the prognosis of patients with AKI.

Additionally, multinomial logistic regression and subgroup analyses revealed that the associations between AKI and risk of CIAH and hypoglycemia were stronger than the associations between AKI and the risk hyperglycemia and hypoglycemia in patients with DM. Our previous study indicated that the stress hyperglycemia ratio is associated with an increased mortality risk [3], a finding corroborated by other studies [20–22]. This increased risk might be partly due to endothelial damage and oxidative stress caused by hyperglycemia. In patients without DM, this could represent a novel stress response, whereas patients with DM might have developed adaptive mechanisms to mitigate oxidative stress over time [21]. Furthermore, AKI patients with sepsis demonstrated a higher risk of hyperglycemia, likely due to the severe kidney damage often associated with sepsis. While dysglycaemia is common in critically ill patients, the management of glycaemic control in patients with AKI remains underexplored. An optimal strategy might involve balancing hypoglycemia and hyperglycemia to stabilize blood glucose levels, which could improve both short-term and long-term outcomes for patients with AKI. For AKI patients with DM, maintaining blood glucose within a narrow range is crucial to prevent hyperglycemia and hypoglycemia. This requires frequent glucose monitoring and adjustments in insulin therapy. Continuous glucose monitoring systems might be particularly useful for real-time adjustments. Due to reduced insulin clearance in AKI, lower insulin doses might be necessary to avoid hypoglycemia. Collaboration with endocrinologists and nephrologists can help tailor insulin regimens based on renal function. In patients without DM, CIAH is common in the ICU and can worsen AKI outcomes. Management strategies might include intermittent insulin therapy or low-dose continuous infusions to keep glucose levels within safe limits. Non-DM patients are also at risk for hypoglycemia, especially with aggressive insulin therapy. Frequent monitoring and conservative insulin use, possibly with a higher glucose target range, can help minimize this risk. Tailoring blood glucose management to diabetic status can reduce mortality risks associated with dysglycaemia in patients with AKI. Further research is needed to refine these strategies and develop evidence-based protocols for clinical practice.

Using a large cohort, this study identified an association between AKI and subsequent dysglycemia in critically ill patients, highlighting the need for improved treatment strategies. These findings support the need for further treatment strategies. However, several limitations must be addressed. First, the MIMIC-IV database is a single-center observational study, necessitating validation with multicenter and ethnically diverse prospective studies. Second, the database does not include post-discharge blood glucose data, leaving the incidence of dysglycaemia after discharge unknown. Third, optimal glycaemic control strategies for AKI patients with or without DM should be investigated in future prospective studies. Fourth, using eGFR on the first day of ICU admission as the baseline may misidentify the true baseline eGFR, as many patients may present with AKI upon ICU admission. Fifth, since AKI was diagnosed within the first 7 days of ICU admission, we lack information on the impact of AKI before admission or after 7 days on dysglycaemia, as well as its influence on dysglycaemia risk within the first 7 days.

5. Conclusion

In this single-center retrospective study, AKI within 7 days of ICU admission was associated with an increased risk of subsequent dysglycaemia, particularly among patients without DM. AKI combined with dysglycaemia might serve as a valuable tool for long-term mortality risk stratification in critically ill patients. Furthermore, dysglycaemia mediates the relationship between AKI and long-term mortality. Future glycaemic control strategies in patients with AKI should aim to balance the risks of hypoglycemia and hyperglycemia, considering the patient’s diabetic status.

Supplementary Material

STROBE Checklist.docx

Supplementary Material.docx

Acknowledgments

We acknowledge the dedicated work of all the authors that implemented the intervention and evaluation components of the study. Moreover, we thank Bullet Edits Limited for the linguistic editing and proofreading of the manuscript.

Ethical approval

The research was conducted in accordance with the principles outlined in the Helsinki Declaration. The utilization of the MIMIC-IV database received approval from both the review committee at Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. As the data is readily available within the MIMIC-IV database, there was no requirement for ethical approval statement or obtaining informed consent for this study.

Author contributions

CZ, MN, and WL contributed to the study design, data analysis, and writing the initial version of the manuscript. WS, YC, TG, and KH contributed to the critical revision of the manuscript for intellectual content. WP and YL are the corresponding authors of the manuscript, and contributed to the study design, study supervision and revisions of the manuscript. All authors read and approved the final manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets analyzed during the current study are available from the corresponding author, upon reasonable request.
==== Refs
References

1 Finfer S, Liu B, Chittock DR, et al. Hypoglycemia and risk of death in critically ill patients. N Engl J Med. 2012;367 (12 ):1108–1118. doi: 10.1056/NEJMoa1204942.22992074
2 Li L, Zhao M, Zhang Z, et al. Prognostic significance of the stress hyperglycemia ratio in critically ill patients. Cardiovasc Diabetol. 2023;22 (1 ):275. doi: 10.1186/s12933-023-02005-0.37833697
3 Zhang C, Shen HC, Liang WR, et al. Relationship between stress hyperglycemia ratio and allcause mortality in critically ill patients: results from the MIMIC-IV database. Front Endocrinol . 2023;14 :1111026. doi: 10.3389/fendo.2023.1111026.
4 Legouis D, Ricksten SE, Faivre A, et al. Altered proximal tubular cell glucose metabolism during acute kidney injury is associated with mortality. Nat Metab. 2020;2 (8 ):732–743. doi: 10.1038/s42255-020-0238-1.32694833
5 Fiaccadori E, Sabatino A, Morabito S, et al. Hyper/hypoglycemia and acute kidney injury in critically ill patients. Clin Nutr. 2016;35 (2 ):317–321. doi: 10.1016/j.clnu.2015.04.006.25912231
6 Carreira A, Castro P, Mira F, et al. Acute kidney injury: a strong risk factor for hypoglycaemia in hospitalized ­patients with type 2 diabetes. Acta Diabetol. 2023;60 (9 ):1179–1185. doi: 10.1007/s00592-023-02112-0.37173530
7 Sechterberger MK, Bosman RJ, Oudemans-van Straaten HM, et al. The effect of diabetes mellitus on the association between measures of glycaemic control and ICU mortality: a retrospective cohort study. Crit Care. 2013;17 (2 ):R52. doi: 10.1186/cc12572.23510051
8 Deane AM, Horowitz M. Dysglycaemia in the critically ill – significance and management. Diabetes Obes Metab. 2013;15 (9 ):792–801. doi: 10.1111/dom.12078.23368662
9 Zaccardi F, Ling S, Lawson C, et al. Severe hypoglycaemia and absolute risk of cause-specific mortality in ­individuals with type 2 diabetes: a UK primary care observational study. Diabetologia. 2020;63 (10 ):2129–2139. doi: 10.1007/s00125-020-05223-3.32728893
10 Heller SR, Peyrot M, Oates SK, et al. Hypoglycemia in patient with type 2 diabetes treated with insulin: it can happen. BMJ Open Diabetes Res Care. 2020;8 (1 ):e001194. doi: 10.1136/bmjdrc-2020-001194.
11 Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150 (9 ):604–612. doi: 10.7326/0003-4819-150-9-200905050-00006.19414839
12 Kellum J, Lameire N, Aspelin P, et al. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl. 2012;2 :1–138.
13 American Diabetes Association Professional Practice Committee. 6. Glycemic targets: standards of medical care in diabetes-2022. Diabetes Care. 2022;45 (Suppl 1 ):S83–S96. doi: 10.2337/dc22-S006.34964868
14 Plummer MP, Bellomo R, Cousins CE, et al. Dysglycaemia in the critically ill and the interaction of chronic and acute glycaemia with mortality. Intensive Care Med. 2014;40 (7 ):973–980. doi: 10.1007/s00134-014-3287-7.24760120
15 Copur S, Demiray A, Basile C, et al. Endocrinological disorders in acute kidney injury: an often overlooked field of clinical research. J Nephrol. 2023;36 (3 ):885–893. doi: 10.1007/s40620-022-01554-z.36652168
16 Basi S, Pupim LB, Simmons EM, et al. Insulin resistance in critically ill patients with acute renal failure. Am J Physiol Renal Physiol. 2005;289 (2 ):F259–264. doi: 10.1152/ajprenal.00002.2005.15840772
17 Hung AM, Siew ED, Wilson OD, et al. Risk of hypoglycemia following hospital discharge in patients with diabetes and acute kidney injury. Diabetes Care. 2018;41 (3 ):503–512. doi: 10.2337/dc17-1237.29326106
18 Liu J, Zhou Y, Huang H, et al. Impact of stress hyperglycemia ratio on mortality in patients with critical acute myocardial infarction: insight from American MIMIC-IV and the Chinese CIN-II study. Cardiovasc Diabetol. 2023;22 (1 ):281. doi: 10.1186/s12933-023-02012-1.37865764
19 Guo JY, Chou RH, Kuo CS, et al. The paradox of the glycemic gap: does relative hypoglycemia exist in critically ill patients? Clin Nutr. 2021;40 (7 ):4654–4661. doi: 10.1016/j.clnu.2021.06.004.34229272
20 Kerby JD, Griffin RL, MacLennan P, et al. Stress-induced hyperglycemia, not diabetic hyperglycemia, is associated with higher mortality in trauma. Ann Surg. 2012;256 (3 ):446–452. doi: 10.1097/SLA.0b013e3182654549.22868366
21 Ren X, Zheng S, Han T, et al. Hyperglycemia and outcomes of medical intensive care unit patients with and without a history of diabetes mellitus in a Chinese population. Intensive Care Med. 2017;43 (1 ):144–145. doi: 10.1007/s00134-016-4598-7.27761592
22 Egi M, Bellomo R, Stachowski E, et al. Blood glucose concentration and outcome of critical illness: the impact of diabetes. Crit Care Med. 2008;36 (8 ):2249–2255. doi: 10.1097/CCM.0b013e318181039a.18664780
