Factors associated with probable sarcopenic dysphagia and development of a clinical classification model in older ICU patients: a cross-sectional study.

OBJECTIVE

To analyze factors associated with probable sarcopenic dysphagia in older ICU patients, develop a clinical classification model, and provide evidence for early bedside identification and targeted intervention.

METHODS

A cross-sectional study was conducted with 160 older patients admitted to a tertiary ICU from January to December 2025 via convenience sampling. Patients were grouped using the Ishii sarcopenia screening score and the Modified Water Swallowing Test (MWST). Data collected included general patient characteristics, ultrasound measurements of swallowing-related muscles, gastric motility ultrasound parameters, functional status indicators, and nutritional risk screening results. Univariate analysis and binary logistic regression were applied to identify factors associated with the outcome. A classification model was constructed and visualized as a nomogram using R software.

RESULTS

The prevalence of probable sarcopenic dysphagia among older ICU patients was 65.62%. Multivariable logistic regression showed that NRS 2002 score ≥ 3, severe genioglossus muscle thinning, presence of comorbidities, abnormal vocalization, and voluntary cough ability score ≤ 2 were independently associated with higher odds of probable sarcopenic dysphagia. The apparent area under the receiver operating characteristic curve (AUC) was 0.837 (95% CI: 0.775-0.900), and the bias-corrected AUC after bootstrap internal validation was 0.812.

CONCLUSION

Probable sarcopenic dysphagia is highly prevalent in older ICU patients and is associated with nutritional status, swallowing muscle morphology, comorbidity burden, vocal function, and airway protective capacity. The classification model demonstrates good discrimination, calibration, and clinical utility, and may serve as a bedside tool for risk stratification. Multicenter external validation is required before clinical implementation.