Type 2 diabetes (T2D) is the dominant form of diabetes worldwide and poses substantial clinical and socioeconomic burdens. Increasing evidence links disturbances in sleep to impaired glucose metabolism and higher T2D risk via mechanisms such as reduced insulin sensitivity and low-grade inflammation. Prior research often modeled sleep–diabetes associations linearly, which may obscure complex dose–response relationships and critical thresholds. This study aimed to assess the association between sleep quality and prevalence of type 2 diabetes among community patients with chronic diseases and to investigate whether a nonlinear relationship exists using restricted cubic spline regression.
Participants were recruited from community hospitals in Jiangsu and Shanghai between 5 January and 30 March 2025. Inclusion criteria required age ≥18 years and presence of chronic disease(s) such as hypertension, cardiovascular disease, or COPD. Exclusion criteria included study withdrawal, incomplete key information (age, sex, height, weight), severe mental disorders, severe visual or hearing impairment, and central nervous system diseases (for example stroke or Parkinson’s disease). All participants provided written informed consent.
Sleep quality was measured using the Chinese-validated Pittsburgh Sleep Quality Index (PSQI). The PSQI comprises seven components—sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, subjective sleep quality, use of sleep medication, and daytime dysfunction—each scored 0–3 and combined into a global score. The instrument’s internal consistency in this context was reported as Cronbach’s α = 0.84. Demographic and clinical baseline data, including age, sex, and BMI-derived metrics, were collected via a standardized questionnaire. Supporting data are provided in the article’s supporting information.
The authors conducted descriptive analyses and logistic regression to examine associations between PSQI scores and prevalent T2D. To explore potential nonlinear dose–response relationships, restricted cubic spline (RCS) regression was applied. Models adjusted for relevant covariates as described in the source. The RCS approach allowed identification of inflection points in the continuous PSQI–T2D association rather than imposing a linear fit.
The analysis included 902 valid samples. Among these, 565 participants (62.64%) were cases of type 2 diabetes. The mean global PSQI score for the sample was 11.11. Additional descriptive tables and figures were provided in the article’s supporting materials.
In all reported regression models, higher PSQI scores were positively associated with prevalence of T2D. The RCS regression confirmed a nonlinear association and identified two inflection points at PSQI scores of 12 and 13. Specifically, when the PSQI score exceeded 12 the odds ratio (OR) for prevalent T2D increased with higher PSQI scores, indicating a stronger association between worse sleep quality and diabetes prevalence. When the PSQI reached 13 the OR curve stabilised, indicating the rate of increase in risk slowed beyond this point. These RCS-derived thresholds suggest a threshold-range effect rather than a simple linear escalation in risk across the full PSQI spectrum.
The findings indicate that among community-dwelling patients with chronic diseases, poor sleep quality as measured by the PSQI is associated with higher prevalence of T2D, and that the relationship is nonlinear with threshold behavior around PSQI 12–13. The authors propose that community health practitioners consider routine sleep quality assessment for patients with chronic conditions and that targeted sleep improvement interventions might be particularly relevant for individuals in the higher-risk range (reported as PSQI≥12).
These results extend prior population-based literature by focusing on a chronically ill community cohort—an important group given their elevated baseline metabolic vulnerability and common coexisting sleep disturbances due to symptoms or medications.
Strengths reported by the authors include the application of RCS modeling to detect nonlinear associations and the focus on a clinically important subgroup (patients with chronic diseases). Limitations identified in the source include the cross-sectional design, which precludes causal inference, and potential biases inherent to questionnaire-based assessments. Additional methodological details and supporting data are available in the article’s supporting information.
In this cross-sectional community sample of patients with chronic diseases, worse sleep quality was positively associated with prevalent type 2 diabetes, and the association exhibited a nonlinear, threshold-range pattern with inflection points at PSQI 12 and 13. The authors suggest incorporating sleep quality assessment into community chronic-disease care and considering targeted interventions for patients with PSQI≥12 as a potential strategy linked to lower T2D prevalence. Funding, data availability, and competing-interest statements were reported in the source article.