Association between the triglyceride glucose index and length of hospital stay in patients with heart failure and type 2 diabetes in the intensive care unit: a retrospective cohort study

Background The coexistence of heart failure and diabetes is prevalent, particularly in Intensive Care Units (ICU). However, the relationship between the triglyceride-glucose (TyG) index, heart failure, diabetes, and the length of hospital stay (LHS) in patients with cerebrovascular disease in the ICU remains uncertain. This study aims to investigate the association between the TyG index and LHS in patients with heart failure and diabetes. Methods This retrospective study utilized the Medical Information Mart for Intensive Care (MIMIC)-IV database to analyze patients with diabetes and heart failure. Participants were categorized into quartiles based on the TyG index, and the primary outcome was LHS. The association between the TyG index at ICU admission and LHS was examined through multivariable logistic regression models, restricted cubic spline regression, and subgroup analysis. Results The study included 635 patients with concurrent diabetes and heart failure. The fully adjusted model demonstrated a positive association between the TyG index and LHS. As a tertile variable (Q2 and Q3 vs Q1), the beta (β) values were 0.88 and 2.04, with a 95% confidence interval (95%CI) of -0.68 to 2.44 and 0.33 to 3.74, respectively. As a continuous variable, per 1 unit increment, the β (95% CI) was 1.13 (0.18 to 2.08). The TyG index’s relationship with LHS showed linearity (non-linear p = 0.751). Stratified analyses further confirmed the robustness of this correlation. Conclusion The TyG index exhibited a linearly positive association with the LHS in patients with both heart failure and diabetes. Nevertheless, prospective, randomized, controlled studies are imperative to substantiate and validate the findings presented in this investigation.


Introduction
Diabetes mellitus (DM) encompasses metabolic disorders marked by persistent hyperglycemia and compromised metabolism of carbohydrates, lipids, and proteins due to deficiencies in insulin secretion, action, or both (1).Projections from the International Diabetes Federation anticipate a rise to 578 million adults with diabetes by 2030 and 700 million by 2045 (2,3).Notably, over 90% of global diabetes cases fall under type 2 diabetes mellitus (T2D) (4), which, rather than manifesting in isolation, manifests as a component of a multifaceted metaboliccardiovascular syndrome (5).Epidemiological investigations have firmly established a significant correlation between heart failure and DM (6)(7)(8).Specifically, T2D acts as a cardiovascular risk factor, elevating the likelihood of heart failure and intensifying morbidity and mortality risks in affected individuals (9).Heart failure surpasses myocardial infarction or stroke as the most prevalent cardiovascular complication of diabetes (10).Previous studies suggest that individuals with T2D represent a distinct population with heightened cardiovascular risk, necessitating tailored prevention programs (11).Therefore, assessing the progression in patients with both heart failure and T2D is imperative for optimizing personalized treatment strategies.
The Triglyceride-Glucose (TyG) index, derived from fasting triglyceride and fasting blood glucose, emerges as a novel surrogate measure for insulin resistance, exhibiting superior diagnostic and predictive capabilities for diabetes compared to blood glucose alone (12,13).Several studies associate the TyG index with atherosclerosis, metabolic syndrome, and T2D (14-16), while others explore its relevance in heart failure and severe illnesses (17).Recognizing the shared pathophysiological mechanisms between diabetes and heart failure underscores the significance of managing their synergistic effects (9,10).Nevertheless, the linkage between the TyG index and the length of hospital stay (LHS) in patients with both T2D and heart failure (HF) remains unexplored.
Previous research underscores the importance of LHS as an indicator for assessing postoperative functional recovery, correlating with postoperative complications and resource utilization (18,19).This study endeavors to examine the association between the TyG index and the risk of in-hospital events and LHS in patients with combined T2D and HF admitted to the intensive care unit (ICU).This exploration may assist in identifying high-risk patients, warranting closer monitoring or early intervention.

Method Data source
This retrospective observational study utilized data from the publicly accessible Medical Information Mart for Intensive Care IV (MIMIC-IV) database, which contains comprehensive clinical information from patients treated at Beth Israel Deaconess Hospital in Boston, Massachusetts, USA (20,21).The dataset encompasses demographics, vital signs, test findings, and diagnoses coded with International Classification of Diseases and Ninth Revision (ICD-9) and International Classification of Diseases and Tenth Revision (ICD-10) (22).Access to the MIMIC-IV database was granted to the primary author, Kai Zhang, subsequent to successful completion of the 'Protecting Human Research Participants' examination (ID: 11639604) administered by the National Institutes of Health (NIH).The data within MIMIC-IV had been previously deidentified, obviating the need for a waiver for informed consent (23).Given the public and anonymized nature of the MIMIC database, ethical committee approval was exempted (24).Reporting of this cross-sectional study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.

Study population
This investigation encompassed individuals diagnosed with HF and T2D identified through ICD-9 and ICD-10 codes (https:// icd.who.int/browse10/2019/en).Supplementary Material provides specific diagnostic codes.Exclusions comprised patients under 18 years old at admission, those without HF and T2D diagnoses, and those with repeated admissions, resulting in a total exclusion of 311,025 individuals from an initial pool of 315,460 ICU-admitted patients in MIMIC-IV.An additional 3,800 individuals lacking triglyceride (TG) and glucose data on the first admission day were excluded.The final study cohort comprised 635 patients categorized into three groups based on the Terciles of the TyG index on the first day of ICU stay (see Figure 1).

Data collection
The study utilized PostgresSQL (version 13.7.2) and Navicate Premium (version 16) for information extraction through a running Structured Query Language (SQL) (25,26).Data extraction from the MIMIC-IV database encompassed demographics, lab results, hourly vital signs, comorbidities, medications, surgical procedures, and ICU stay specifics, facilitated by pgAdmin PostgreSQL tools (version 1.22.1).

Expose and outcome measures
Baseline TyG index was calculated based on fasting TGs and FBG values obtained at admission as previously reported: ln [fasting TGs (mg/dL) * FBG (mg/dL)/2] (29).The participants were divided into three groups according to the Tierce level of the TyG index.The primary outcome indicator was the patient's LHS.

Statistical analyses
The data were initially categorized into continuous and categorical variables.Continuous variables were expressed as mean (SD) or median (IQR) and compared using appropriate statistical tests, such as the student t-test or nonparametric tests.Categorical variables were presented as frequencies and percentages (%) and compared using the Pearson chi-square test or Fisher's exact test.Baseline characteristics were grouped according to Tertiles of the TyG index.
To assess the relationship between the TyG index and LHS, multivariable logistic regression analyses were conducted.The models progressively adjusted for covariates: Model 1 included no adjusted covariates.Model 2 included demographic variables (gender, age, marital status, body mass index).Flowchart detailing the selection process for patients included in this analysis.

Baseline characteristics of selected participants
Table 1 presents the baseline characteristics of 635 patients diagnosed with both diabetes and congestive heart failure (CHF).Stratified into tertiles based on the TyG index (Q1 ≤ 8.984, Q2 8.984-9.625,Q3 ≥ 9.625), the entire patient cohort had a mean age of 71.0 ± 12.3 years.High TyG index individuals, regardless of gender, exhibited characteristics such as youthfulness, elevated BMI, increased heart and respiratory rates, reduced Spo2, heightened WBC, lowered Chlorine, diminished HDL, elevated LDL, increased LD, higher APSIII and SOFA scores, and extended LHS compared to their low TyG index counterparts.

Association between TyG index and length of hospital stay
Table 2 displays results from a multivariable logistic regression analysis exploring the association between TyG index and LHS in patients with HF and diabetes.The study investigates b and corresponding 95% confidence intervals (CIs) across varying TyG index to length of hospital stay within this patient cohort.

Dose-response relationships
Employing a logistic regression model with a cubic spline function, the study illustrated a statistically significant linear association between TyG index and LHS (p=0.751) after adjusting for confounding factors, as depicted in Figure 2.

Subgroup analysis
Subgroup analyses, stratifying participants by age, gender, BMI, hyperlipidemia, CKD, and hypertension, consistently showed a significant association between TyG index and LHS across various subgroups.The Forrest plot in Figure 3 highlights the independent and  consistent link between TyG index and LHS.Statistically significant associations were not observed in certain subgroups (P > 0.05).

Discussion
In this retrospective observational cohort study, we present the inaugural investigation into the association between the TyG index and the duration of hospitalization in critically ill patients with coexisting HF and T2D within a United States (US) cohort.Following meticulous adjustment for numerous potential confounding variables, our analysis reveals a positive correlation between the TyG index and the LHS.Employing smooth curve fitting techniques reinforces the identification of a linear relationship between the TyG index and LHS.The robustness of these findings is underscored by stratified and sensitivity analyses.
The TyG index, serving as a novel surrogate marker for insulin resistance, has been extensively utilized in predicting the risk and severity of various diseases (31).Numerous clinical investigations on the association between the TyG index and patient outcomes across diverse diseases have been conducted.Meta-analysis shows that the TyG index as a valuable marker to assess the risk of HF incidence (32), ischemic stroke (33), coronary artery disease (34), cardiovascular diseases and mortality in the general population (35) and atrial fibrillation (36).A study focusing on the general American population demonstrated that an elevated TyG index signifies more severe insulin resistance, displaying a non-linear association with LHS due to all-cause and cardiovascular diseases (37).In the context of chronic kidney disease (CKD), another study found a non-linear relationship between the TyG index and CKD, with a higher TyG index correlating with an increased risk of CKD (38).In the stroke population, a study identified the TyG index as a potential predictor for hospital and intensive care unit (ICU) mortality in critically ill stroke patients (39) Significantly, prior research on the TyG index and its connection to diabetes or HF has primarily focused on individual diseases.A Chinese study revealed an elevated risk of incident T2D with increasing TyG index among rural Chinese individuals, suggesting its potential as an important indicator for identifying individuals at high risk of T2D (40).Another study found that patients with a discordantly high TyG index faced a significantly greater risk of cardiovascular events irrespective of diabetes status (41).Furthermore, an independent association between an elevated TyG index and poor prognosis in patients with acute decompensated HF was identified (17).HF, a prevalent and severe cardiovascular consequence of DM, has garnered substantial global attention (4).Approximately 40% of HF patients are affected by DM, resulting in a two-to five-fold elevation in the risk of HF compared to normoglycemic individuals (42).Previous investigations emphasize the unfavorable clinical prognosis for diabetic patients with HF, displaying higher mortality rates than their normoglycemic counterparts (43)(44)(45).Despite the well-established association between diabetes and HF, there is limited data supporting the relationship between the TyG index and the LHS in patients with both HF and T2D.Through multivariate regression analysis and subgroup examinations, our study establishes the independent relevance of the TyG index to the LHS in patients with concurrent T2D and HF.Considering the TyG index as a straightforward method for evaluating insulin resistance, our findings affirm that insulin resistance is independently associated with the LHS.
This study utilized smooth curve fitting and restricted cubic spline analysis to evaluate the linear relationship between the TyG index and the duration of hospitalization in patients with concomitant HF and diabetes, diverging from previous investigations.Subsequent subgroup analyses were conducted to ensure the consistency of the primary findings.The observed association demonstrated a linear pattern, indicating that higher TyG index values are correlated with an increased LHS.These findings carry significant implications for clinical practice and patient care.In our investigation, elevated TyG levels were associated with prolonged in-hospital stays among patients with both T2D and HF within a specific range.This suggests that TyG may serve as a valuable tool for risk stratification and management in this high-risk patient population.Addressing the heightened risk associated with high TyG levels requires a comprehensive approach to risk management.These results underscore the importance of considering the TyG index in clinical practice due to its simplicity, reliability, and convenience, setting it apart from other commonly used methods for disease assessment (46).Researchers have increasingly explored the TyG index, identifying it as a reproducible, reliable, cost-effective, and valid surrogate marker of insulin resistance (IR) (47).Previous studies have demonstrated that the TyG index is highly sensitive (96.5%) and specific (85.0%) for detecting IR compared to the hyperinsulinemic-euglycemic clamp technique (48).Moreover, the TyG index has shown superior performance to Homeostatic model assessment insulin resistance index (HOMA-IR) (49), making it widely applicable in clinical practice, given the availability of glucose and triglyceride tests in all clinical laboratories.Additionally, there may be differences between early onset T2D and late onset T2D, which requires further discussion.One study reviewed the health consequences of earlyonset T2DM and compared the results with those of late-onset T2DM, based on matched cohorts in the UK Biobank and reveal that early-onset T2D is more challenging in terms of glycemic control, with a higher risk of severe complications (50).The TyG index may provide an earlier clinical evaluation and provide strategies for future diabetes intervention.Furthermore, from the perspective of medical conditions and promotion, other insulin resistance markers such as HOMA-IR need more structural resources, and access to healthcare.The TyG index can be used as substitute for other tests in clinical settings where these tests are not feasible, and it has been demonstrated that the TyG index has high sensitivity and specificity compared with HOMA-IR and Clamp (48,51).The TyG index can be easily calculated by TG and FPG without additional cost and has a better performance for the prediction of IR.
Several potential mechanisms elucidate the correlation between the TyG index and LHS.The association with adverse hospitalization duration may be attributed to insulin resistance reducing glucose bioavailability, favoring a shift to fatty acid metabolism, thereby increasing myocardial oxygen consumption and compromising myocardial compensatory capacity (52,53).Additionally, glycolipid metabolism disorder can induce reactive oxygen species, mitochondrial dysfunction, endoplasmic reticulum stress, impaired cardiac calcium signaling, systemic low-grade inflammation, and inappropriate activation of the reninangiotensin system, exacerbating heart failure progression (54)(55)(56).Furthermore, glycosylation end product deposition may increase diastolic left ventricular stiffness and inactivate nitric oxide, a protective factor for vascular endothelium (57).Lastly, a vicious cycle between congestive heart failure and insulin resistance may intensify cardiac function deterioration (58).
Our study boasts several notable strengths.Firstly, it pioneers the examination of the TyG index's role in patients with both T2D and HF in the Intensive Care Unit (ICU).Emphasizing the necessity of a comprehensive risk management approach for this patient cohort, our study positions the TyG index as a valuable tool in assessing the LHS.Additionally, we employ restricted cubic splines to explore nonlinear relationships between the TyG index and outcomes.To address potential confounding variables, logistic regression analysis, incorporating multiple models, and subgroup analyses with appropriate categorizations were conducted.
Despite these strengths, our study has acknowledged limitations.Primarily, being a single-center retrospective study limits the definitive establishment of causality.We recommend future validation through multicenter prospective studies.Furthermore, drug use may affect the patient's condition, but due to database limitations, it may not provide sufficient information on drug use.We plan to investigate the usage of patients more in future research on self-built databases.The baseline-only availability of data impedes the assessment of TyG level changes during the follow-up period, though initial levels likely accurately reflect the TyG index at hospitalization commencement.Then, due to the sample size, there may not be statistically significant differences in some subgroups.However, the overall effect value meets the requirements, and we will explore it in a larger sample size population in the future.Lastly, our findings should not be indiscriminately extrapolated to other countries or ICU facilities, given the study's confinement to the United States and a single ICU institution.Nevertheless, the considerable sample size enhances the credibility and representativeness of our findings.

Conclusions
Our study establishes the TyG index as a significant predictor of prognosis in critically ill patients with both HF and T2D, showing an approximately linear relationship with the LHS.In this high-risk group, TyG emerges as a valuable tool for risk categorization and management.To augment validity and coherence, future research should adopt prospective, randomized, controlled study designs to confirm these results and elucidate the mechanisms linking TyG to the LHS in HF and T2D patients.
Model 3 added vital signs and comorbidities.Model 4 further included medications and laboratory indicators.Model 5 additionally incorporated severity of illness scores at admission.Trend tests utilized logistic regression with the TyG index categorized into three groups.Variables with a variance inflation factor exceeding 5 were excluded to prevent multicollinearity.Potential nonlinear correlations between TyG index levels and outcomes were explored using restricted cubic splines.Stratified regression models and likelihood ratio tests identified modifications and interactions in subgroups based on factors such as age, gender, BMI, hyperlipidemia, chronic kidney disease, and hypertension.Missing values for laboratory indicators in the MIMIC-IV database were addressed by calculating the percentage of missing values for

FIGURE 2
FIGURE 2Dose-response relationship between TyG index with length of hospital stay in ICU.

FIGURE 3
FIGURE 3Stratified analyses of the association between TyG index with length of hospital stay in ICU.
(30)tation was performed using a random forest-based method for variables with less than 20% missing values, while variables with over 20% missing values were categorized.Non-normally distributed continuous variables were analyzed after conversion into categorical variables based on the normal reference range from the MIMIC-IV database.All statistical analyses were conducted using R, version 4.1.1,andFreeStatistics software, version 1.7(30).A significance level of P < 0.05 (two-sided) was considered statistically significant.Reporting adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for this cross-sectional study.

TABLE 1
Characteristics of the study population (N = 635).

TABLE 2
Multivariable logistic regression to assess the association of TyG index with length of hospital stay in ICU.