Sleep Disorders Predict Frailty Progression in Type 2 Diabetes Mellitus: Findings from a Cohort of 20,562 Patients

The global incidence and public health burden of type 2 diabetes mellitus (T2DM) have risen steadily over recent decades. T2DM accelerates functional decline and premature aging. Furthermore, T2DM is a major risk factor for frailty. Diabetes and frailty share overlapping pathophysiological mechanisms, including chronic inflammation, insulin resistance, malnutrition, and hormonal dysregulation, while frailty itself may elevate the risk of developing diabetes [1]. Identifying frailty in patients with T2DM is therefore essential. Despite frailty’s clinical importance, the determinants of its development and progression in T2DM remain incompletely understood. Prior studies have implicated certain medication uses as potential contributors to incident frailty in this population [2], but few address risk factors for worsening frailty [3].

Sleep disorders are common in diabetes, with insomnia affecting up to 39% of patients, 4-fold higher than that in the general population [4]. We hypothesized that sleep disorders significantly increase the risk of worsening frailty among patients with T2DM, and assembled a large hospital-based cohort to examine this association. Adults aged ≥40 years with T2DM were retrospectively identified from the National Taiwan University Hospital integrated Medical Database between 2008 and 2016. The study had received approval (No. 201708098RIND). The exposure of interest was physician-diagnosed sleep disorder at baseline, identified using International Classification of Diseases (ICD) codes validated in prior research [5]. The spectrum of sleep disorders identified included insomnia, hypersomnia, parasomnia, sleep related breathing or movement disorders, circadian rhythm disorders, and substance-induced sleep disorders, according to the updated classification of the International Classification of Sleep Disorders. The primary outcome was worsening frailty, defined as an increase of ≥1 Fatigue, Resistance, Ambulation, Illnesses, and Loss of Weight (FRAIL) scale score relative to baseline assessment results, as adopted by others [3]. The five components of FRAIL scale were retrospectively coded from clinical data, diagnoses, procedural details/records, an approach partially adopted by others [6]. The validity of our operationalization approach has been shown to be fair, with results exhibiting good correlation with index-based findings [3]. For multivariate analysis, we adopted four strategies; model 1 included demographic (age, sex) and physical indices (body mass index), comorbidities (hypertension, hyperlipidemia, atrial fibrillation, cerebrovascular disease, myocardial infarction, acute coronary syndrome, heart failure, peripheral vascular disease, chronic kidney disease, chronic liver disease, malignancy, and chronic obstructive pulmonary disease) and medications (glucose-lowering drugs, anti-hypertensive and anti-hyperlipidemic drugs, anti-coagulants, anti-platelet agents, and anti-inflammatory drugs). Model 2 included model 1 covariates and laboratory data (glycemic indices, lipid profile, creatinine and estimated glomerular filtration rate [eGFR]). Models 3 and 4 further accounted for competing risk of mortality and baseline frailty, respectively.

A total of 82,208 adults with T2DM were screened (Fig. 1A), with 20,562 (25.0%) included in final analyses. The mean age was 64.2±10.9 years, with 44.8% female, mean glycosylated hemoglobin 7.1%±1.3%, and a mean eGFR of 73.1±26.5 mL/min/1.73 m². Around 19.2% of patients had a sleep disorder at baseline. After a median 3.94 years (interquartile range, 1.8 to 7.1), 5,698 patients (27.7%) developed worsening frailty. Kaplan-Meier analysis demonstrated a significantly higher cumulative incidence of frailty progression in patients with sleep disorders (log-rank P<0.01) (Fig. 1B). Following multivariable adjustment, this association remained significant (model 1: hazard ratio [HR], 1.13; 95% confidence interval [CI], 1.06 to 1.21; model 2: HR, 1.15; 95% CI, 1.08 to 1.23) (Fig. 1C). The results were unchanged when death (model 3) (Fig. 1C) or baseline frailty (model 4) (Fig. 1C) was considered. Insomnia/hypersomnia alone also conferred elevated risk (HR, 1.14; 95% CI, 1.06 to 1.22). Subgroup analyses (Fig. 1D) showed that the association between sleep disorders and frailty progression was evident among patients of different age strata, sex, and without or with chronic kidney disease, and the adjustment for baseline frailty intensified the association strengths. Glycemic control status did not modify the association between sleep disorders and risk of worsening frailty.

Fig. 1.

(A) Study participant selection algorithm. (B) Kaplan-Meier event-free analyses according to sleep disorder presence or not. (C) Results of multivariate Cox proportional hazard regression analyses, including models 1 to 3. (D) Subgroup analysis results. DM, diabetes mellitus; NTUH, National Taiwan University Hospital; P, population; PY, person-year; HR, hazard ratio; CI, confidence interval; eGFR, estimated glomerular filtration rate.

Previous research on sleep and frailty in diabetes has been limited to cross-sectional designs, which preclude causal inferences [7]. Our study adds to this literature by providing longitudinal evidence from a large, well-characterized cohort, using a broad definition of sleep disorders that extends beyond sleep duration or subjective quality. Importantly, our results demonstrate that the effect is not confined to older adults but is relevant across the adult diabetes mellitus (DM) population, providing a glimpse into the degree of risk elevation. The mechanisms linking sleep disorders to frailty progression are likely multifactorial. Sleep disturbances can exacerbate cardiovascular morbidity, depression, and poor nutritional status, all established contributors to frailty [7]. Sleep disorders are also associated with sarcopenia [8], which represents a core pathway to functional decline. Additionally, sleep disruption may accelerate cognitive impairment, a recognized risk factor for frailty propagation. These overlapping biological and behavioral mechanisms provide a plausible basis for the associations observed.

Our subgroup findings (Fig. 1D) warrant further discussion. A previous study disclosed that predictors of impaired physical function in patients with DM changed with age; poor sleep quality correlated with decreasing walking distance only in those age <58 years but not older ones [9]. The stronger association in middle-aged adults is consistent with these data that sleep quality predicts physical decline more prominently at younger ages, whereas in older individuals, other factors such as sarcopenia or nutritional compromise may overshadow the impact of sleep disruption. Similarly, the absence of an association in patients with impaired kidney function may reflect the dominant influence of uremic toxins and chronic inflammation on frailty in this population [10]. These observations highlight the heterogeneous impact of sleep disorders and suggest that targeted screening strategies may be most beneficial in specific patient groups.

The clinical implications of our findings are substantial. Sleep hygiene and disorder management are increasingly recognized as integral components of chronic disease care. Non-pharmacological interventions, such as bedtime relaxation techniques, exercise, and structured sleep education, have shown promise in improving sleep quality. Other strategies, like bedroom environment re-arrangement or sleep pattern adaptation, require more evidence to support. Optimizing sleep through personalized interventions may also improve glycemic control stability, reduce hypoglycemia risk, and ultimately lead to lower long-term likelihood of worsening frailty.

Our study has several strengths, including the large sample size, longitudinal design, and comprehensive adjustment for confounding variables. Nevertheless, limitations should be acknowledged. First, frailty was assessed solely using the FRAIL scale only, though this tool has been validated in the diabetes population [3]. Second, we lacked direct measures of muscle mass and nutritional status. Third, residual confounding and selection bias cannot be excluded due to the retrospective design. Enrollees were younger and had a higher prevalence of comorbidities than the excluded ones, but four out of the five FRAIL components of the former group at baseline did not differ from those of the excluded patients, except illness. Our findings are more applicable to those with T2DM and high comorbidity burden, receiving hospital-based care. Fourth, we did not document sleep-managing medications in this study.

In conclusion, in this large, longitudinal cohort of patients with T2DM, baseline sleep disorders were independently associated with a higher risk of frailty progression. Incorporating routine sleep assessment into diabetes care and developing targeted interventions to improve sleep quality may offer new opportunities to favorably alter frailty courses in this vulnerable population.

The Institutional Review Board of National Taiwan University Hospital has approved the parent protocol of the main study (No. 201708098RIND). Informed consent was waived due to the anonymized nature of retrospective data retrieval from the Integrated Medical Database.