Work

Development and pilot temporal validation of a questionnaire-free model for severe obstructive sleep apnea using sex, waist circumference, and TyG index

Year: 2026
Type: article
Study design: Observational Study
Abstract: Brief Summary Current Knowledge/Study Rationale: Polysomnography (PSG) resources remain severely limited in primary care settings across China. Existing Western-derived questionnaire-based screening tools, such as the STOP-Bang questionnaire, are constrained by linguistic and cultural barriers when applied to Chinese populations, creating an urgent need for an objective, low-cost tool to identify patients with severe obstructive sleep apnea (OSA). Study Impact We developed and validated a simple, questionnaire-free prediction model using three routine health-checkup parameters—sex, waist circumference, and the triglyceride-glucose (TyG) index. In a temporal validation cohort, the model achieved a high sensitivity of 87.2% at a low cutoff, supporting its utility as a triage tool in Chinese primary care and health examination centers to reduce missed diagnoses of severe OSA. Purpose To develop and validate a questionnaire-free model for severe obstructive sleep apnea (OSA) using objective parameters from routine checkups, facilitating screening in primary care settings. Methods This retrospective study analyzed 271 snoring participants who underwent polysomnography at a Chinese tertiary hospital (June 2018–June 2024). Participants were categorized into mild-to-moderate ( n = 128) and severe ( n = 143, apnea-hypopnea index ≥ 30 events/h) OSA groups. The candidate predictors were sex, waist circumference, and the triglyceride-glucose (TyG) index. A logistic regression model was developed, internally validated via 1,000 bootstrap resamples, and temporally validated in a 2025 cohort ( n = 82). Performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis (DCA). Results Male sex (odds ratio (OR) = 3.69, 95% CI: 1.421–9.583), waist circumference (OR = 1.039, 95% CI: 0.995–1.085; P = 0.081, 95% CI crossing 1), and TyG index (OR = 1.421, 95% CI: 1.028–1.965) were retained in the model. Male sex and the TyG index were independent predictors of severe OSA, whereas waist circumference showed a borderline, non-significant association and should not be interpreted as a statistically significant independent predictor. In the training set, the model yielded an AUC of 0.751 (95% CI: 0.675–0.827), with an optimism-corrected AUC of 0.739 after bootstrap internal validation. In the temporal validation cohort, the model achieved an AUC of 0.648 (95% CI: 0.525–0.770). At a cutoff of 0.238, sensitivity was 87.2% and negative predictive value was 68.4%. Decision curve analysis demonstrated that the model provided higher net benefit than either the “treat all” or “treat none” strategies across most threshold probabilities (0.07–0.93). Conclusion This simple, questionnaire-free model offers a favorable net clinical benefit and remains highly sensitive at a low cutoff. It may serve as an adjunctive screening aid to help prioritize patients for PSG, particularly in resource-limited primary care and health examination settings. However, given its modest temporal validation AUC (0.648) and low specificity (38.2%), it should not be used alone to rule out severe OSA.
Language: English
Field: Medicine
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7225920670
FWCI:
Cites: 25
Cited by:
Related to: 10