Type 2 diabetes mellitus (T2DM) and heart failure (HF) represent two major global health challenges, with significant epidemiological overlap and a complex pathophysiological interplay [1]. The coexistence of these conditions is associated with worse clinical outcomes compared with either condition alone, exerting synergistic detrimental effects on overall prognosis [2]. Recent epidemiological data indicate that approximately 30–40% of patients with heart failure also have T2DM, while the prevalence of HF among patients with diabetes ranges from 10% to 30%, depending on age and comorbidities [3]. This bidirectional relationship is driven by shared risk factors and intricate mechanistic pathways.

Although glycated hemoglobin (HbA1c) has traditionally served as the cornerstone of glycemic assessment in diabetes management, accumulating evidence suggests that glycemic variability (GV) may provide additional prognostic information beyond mean glucose levels. Glycemic variability refers to the magnitude, frequency, and duration of glucose fluctuations within and between days, representing a dynamic component of glucose homeostasis [4]. A growing body of evidence indicates that increased GV is independently associated with adverse outcomes in diabetic and cardiovascular populations, including microvascular complications, cardiovascular events, and increased risk of mortality [4, 5].

The assessment and management of GV are of particular clinical relevance in patients with concomitant T2DM and HF. Heart failure can influence glucose metabolism through multiple mechanisms, including tissue hypoxia, neurohormonal activation, altered gastrointestinal absorption, and reduced physical activity [6]. Conversely, GV may exacerbate HF through mechanisms such as fluid retention, electrolyte imbalance, and direct myocardial toxicity [7]. These reciprocal interactions may create a vicious cycle in which HF worsens glycemic control, and unstable glycemia further aggravates cardiac dysfunction, ultimately contributing to higher rates of hospitalization and mortality [8].

Despite the significant clinical implications of GV in this high-risk population, current understanding of the factors influencing different GV parameters in patients with T2DM and HF remains limited. Most previous studies have focused on single GV metrics or evaluated diabetic populations without specifically accounting for heart failure status [9, 10]. Furthermore, different GV indices capture distinct dimensions of glucose fluctuations: standard deviation of blood glucose (SDBG) reflects overall dispersion, the coefficient of variation (CV) represents relative variability, the mean amplitude of glycemic excursions (MAGE) captures major glucose swings, and the mean of daily differences (MODD) indicates day-to-day variability [11]. Whether these indices share common determinants or are influenced by distinct factors in patients with HF warrants systematic investigation.

Identifying potentially modifiable factors associated with GV in patients with T2DM and HF may inform targeted interventions aimed at improving glycemic stability and, potentially, clinical outcomes. Therefore, the present study aimed to comprehensively evaluate factors associated with four GV indices (SDBG, CV, MODD, and MAGE) in hospitalized patients with T2DM and HF using multiple linear regression analysis. Elucidating these associations may contribute to the development of personalized glycemic management strategies for this vulnerable population.

This study systematically examined independent factors influencing four GV indices in hospitalized patients with T2DM and HF using multiple linear regression analysis. The results demonstrate that, although significant overlaps exist among the factors associated with different GV metrics, each index also exhibits distinct determinants, reflecting its unique physiological dimensions. These findings enhance current understanding of the complexity of glycemic regulation in this high-risk population and provide a basis for individualized glycemic management strategies.

First, C-peptide level demonstrated a consistent and significant negative association with all four GV indices. This finding strongly supports the central role of endogenous insulin secretory capacity in maintaining glucose homeostasis in patients with T2DM and HF. Preserved $\beta$-cell function allows for timely and adequate insulin secretion in response to dynamic glucose changes, effectively buffering against glycemic fluctuations induced by various stressors, including neurohormonal activation associated with HF [21]. Within the bidirectional pathophysiological interaction between T2DM and HF, both conditions may accelerate $\beta$-cell dysfunction by exacerbating insulin resistance and chronic low-grade inflammation [1]. Consequently, preservation of residual $\beta$-cell function should represent a cornerstone of glycemic management in this patient population.

However, interpretation of C-peptide levels is complex, as they are influenced not only by $\beta$-cell function but also by the degree of insulin resistance. Insulin resistance may lead to compensatory hyperinsulinemia and elevated fasting C-peptide levels [22]. The strong negative associations observed in the present study suggest that, in this specific clinical context, the $\beta$-cell functional component of C-peptide may predominate as a determinant of GV. This finding underscores the critical relevance of preserved insulin secretory capacity for maintaining glycemic stability, even in the presence of concomitant insulin resistance.

Second, age emerged as another consistently relevant factor, showing significant positive associations with SDBG, log(CV), and MODD (all p 0.001), although it did not reach statistical significance in the MAGE model (p = 0.050). These findings suggest a progressive decline in glycemic stability with advancing age, potentially attributable to age-related reductions in muscle mass leading to worsening insulin resistance, diminished $\beta$-cell responsiveness, and impaired autonomic nervous system function [23]. Collectively, these mechanisms may predispose older individuals to greater fluctuations in glucose levels. Accordingly, glycemic management in elderly patients with T2DM and HF may require a more cautious and individualized approach, potentially involving less stringent glycemic targets and the preferential use of therapies with a low risk of hypoglycemia to minimize the adverse consequences associated with excessive glycemic fluctuations [24].

This study further revealed that different GV indices have distinct determinants, reflecting their unique physiological interpretations of glucose fluctuations, with important implications for clinical practice. SDBG, which was predominantly influenced by age, diabetes duration, HF severity (as reflected by NT-proBNP), and $\beta$-cell function, underscores the close relationship between the magnitude of absolute glycemic fluctuation and chronic disease progression, cardiac status, and intrinsic insulin secretory capacity.

Variable transformation for CV was performed based on formal residual normality testing (Shapiro-Wilk p 0.001 for untransformed vs. p = 0.223 after logarithmic transformation), thereby ensuring that model assumptions were satisfied and enhancing the reliability of statistical inference. In contrast, MODD, SDBG, and MAGE were analyzed without transformation because their residuals met normality assumptions. This methodological approach ensured that the assumptions underlying regression analysis were adequately met, thereby increasing the robustness of the results. Notably, regression coefficients derived from transformed models must be interpreted on their respective mathematical scales, which may introduce additional complexity but allows for more accurate estimation of effect sizes in clinical interpretations.

Notably, HbA1c was not an independent factor for log(CV) in the multivariate model. Because CV is calculated as SDBG divided by mean glucose, it inherently adjusts for differences in average glycemia and is therefore considered the gold standard for measuring “relative variability” [10]. Our findings indicate that long-term glycemic control, as reflected by HbA1c, and relative glucose fluctuation, as reflected by CV, represent largely independent dimensions [25]. This dissociation may explain why some patients achieve HbA1c targets yet continue to experience marked GV and remain at elevated risk for complications [26].

Furthermore, CV was primarily associated with HF severity, as indicated by NT-proBNP levels, highlighting the direct influence of cardiac dysfunction on glycemic stability. One plausible mechanism involves impaired tissue glucose delivery and altered glucose utilization efficiency secondary to reduced cardiac output. While recent studies have suggested potential links between HF severity, metabolic regulators such as adropin, and insulin sensitivity [27, 28, 29], these mechanisms were not directly examined in the present cohort. Therefore, the observed association between NT-proBNP and GV warrants further investigation to elucidate the underlying pathophysiological pathways.

The divergent associations observed between BMI and different GV indices warrant particular attention. BMI demonstrated a significant negative association with MODD but a positive association with MAGE. This apparent paradox may be explained by distinct physiological mechanisms. Higher BMI may confer greater day-to-day glycemic stability through nutritional reserves and metabolic inertia [30], potentially accounting for the negative association with MODD. In contrast, obesity-related insulin resistance may predispose individuals to larger acute glucose excursions in response to prandial stimuli [31], aligning with the positive association observed with MAGE. These findings align with the so-called “obesity paradox” described in chronic conditions such as HF and underscore the complexity of metabolic regulation in this population [32].

Importantly, sharp glucose fluctuations have been shown to exert acute and severe adverse effects on the vascular system by inducing oxidative stress, exacerbating endothelial dysfunction, and promoting an inflammatory response [33]. These pathophysiological effects align more closely with the glucose “peaks and troughs” captured by MAGE, reinforcing the clinical relevance of index-specific interpretation when assessing GV in patients with concomitant T2DM and HF.

The clinical implications of the present study suggest that glycemic management in patients with T2DM and HF may benefit from a multidimensional assessment approach. For instance, in patients with impaired cardiac function, as indicated by elevated NT-proBNP levels, consideration may be given to cardioprotective therapies known to improve glycemic stability, such as sodium-glucose cotransporter 2 (SGLT2) inhibitors [34], with attention to CV values [9]. Similarly, in patients with an extensive history of cardiovascular disease, therapeutic strategies targeting reductions in MAGE may be warranted. However, these clinical suggestions should be viewed as preliminary and require validation in prospective, adequately powered interventional studies.

A noteworthy finding in the study was the presence of elevated mean HbA1c levels, accompanied by relatively low SDBG values. We hypothesize that this finding may be attributable to the standardized inpatient environment, which minimizes lifestyle-related glucose triggers, and the use of intermittent capillary blood glucose monitoring. Compared with continuous glucose monitoring (CGM), intermittent capillary measurements may underestimate the true amplitude and frequency of acute glycemic excursions. Consequently, the GV indices reported in this study should be interpreted as reflecting glycemic stability under controlled hospitalized conditions rather than glycemic patterns in free-living ambulatory settings.

Several important limitations of this study should be acknowledged. First, the single-center and retrospective design introduces an inherent risk of selection bias and limits the generalizability of the findings. Additionally, the reliance on electronic medical records precluded systematic collection of several potentially relevant confounding factors, including detailed glucose-lowering management regimens (e.g., specific insulin protocols and use of SGLT2 inhibitors or glucagon-like peptide-1 (GLP-1) receptor agonists), dietary intake, and physical activity levels. Consequently, the observational and cross-sectional nature of the analysis precludes causal inference regarding the relationships between identified factors and GV outcomes.

Additionally, although the sample size of 150 patients was sufficient for exploratory analyses, it remains relatively modest for multivariable linear regression models with multiple predictors, potentially affecting model stability and the precision of estimates. Glycemic variability was assessed using capillary blood glucose measurements rather than CGM, which is considered the gold standard for evaluating GV, especially for indices such as MAGE [35]. Although the standardized seven-point capillary glucose monitoring protocol employed in this study represents a pragmatic and widely applicable approach in the inpatient setting, it likely underestimates the true magnitude of acute glucose fluctuations. Future studies incorporating CGM technology and larger, multicenter cohorts will be essential to validate and extend these findings.

Glycemic variability in hospitalized patients with T2DM and HF is influenced by multiple clinical factors, with C-peptide level, age, diabetes duration, HF severity (as reflected by NT-proBNP), and overall glycemic control representing the primary influencing factors. Different GV indices exhibit unique spectra of associated factors, reflecting the diverse dimensions of glycemic variability they represent. Accordingly, clinical practice may potentially benefit from a multidimensional assessment of GV. For example, greater attention to CV may be particularly relevant in patients with elevated NT-proBNP levels, while strategies aimed at reducing MAGE may be considered in patients with a history of cardiovascular disease. However, these implications should be interpreted with caution, as they are derived from an exploratory analysis, and their clinical utility requires validation in prospective intervention studies.

• C-peptide levels demonstrated a significant negative association with all indices of glycemic variability, underscoring that preserved $\beta$-cell function is critical for maintaining glucose homeostasis in patients with T2DM and HF, and should be a central consideration in clinical management.

• N-terminal pro-B-type natriuretic peptide (NT-proBNP), a biomarker of HF severity, was an independent positive predictor of most GV indices, revealing that cardiac dysfunction itself constitutes a clinically measurable and relevant pathophysiological driver of dysglycemia.

• The finding that BMI was negatively associated with MODD but positively associated with MAGE highlights that glycemic variability is a multidimensional concept, with distinct components, such as day-to-day stability and acute glucose excursions, being influenced by distinct physiological mechanisms.

• Taken together, these findings suggest that glycemic management in patients with T2DM and HF should extend beyond exclusive reliance on HbA1c and incorporate a multidimensional assessment of GV to support more individualized therapeutic strategies, such as prioritizing CV in patients with advanced HF and targeting MAGE reduction in those with established cardiovascular disease.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

YJ and JBM designed the research study. YJ performed the research. JBM analyzed the data. YJ and JBM drafted this article. Both authors contributed to the important editorial changes in the manuscript. Both authors read and approved the final manuscript. Both authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study was approved by the Ethics Committee of People’s Hospital of Baishan (approval number: KY2025023). All procedures followed the ethical principles outlined in the Declaration of Helsinki. Informed consent was waived by the Ethics Committee of People’s Hospital of Baishan due to the retrospective study design and the anonymization of all patient data.

Not applicable.

This research received no external funding.

The authors declare no conflicts of interest.