Routinely available clinical variables predict swallowing status after stroke

Photo by Vitaly Gariev / Unsplash

medRxiv Published April 30, 2026 Medically reviewed April 30, 2026 Study authors: Okamoto, K.; Irie, K.; Hoyano, K.; Matsushita, I. DOI ↗ By Dr. Ji-eun Park, MD · Brain, Mind & Pain

Key Takeaway

Consider using routinely available clinical variables to estimate swallowing status after stroke, but external validation is needed.

This retrospective observational study included 126 patients with first-ever stroke who underwent a videofluoroscopic swallowing study or fiberoptic endoscopic evaluation of swallowing during hospitalization. The authors developed prediction models using routinely available clinical variables (age, sex, stroke type, Japan Coma Scale, modified Rankin Scale, Functional Independence Measure motor and cognition, serum albumin level, body mass index, time from stroke onset) to estimate swallowing status based on Food Intake LEVEL Scale (FILS) thresholds of ≥3 and ≥7.

At the index assessment, 112 (88.9%) patients had FILS ≥3 and 73 (57.9%) had FILS ≥7. The area under the receiver operating characteristic curve (AUC) was 0.759 for FILS ≥3 and 0.821 for FILS ≥7. Decision curve analysis suggested greater clinical utility for the FILS ≥7 model than for the FILS ≥3 model.

The authors note that the models were internally validated only, and external validation is recommended before clinical application. Adverse events were not reported.

These findings suggest that routinely available clinical variables may help estimate swallowing status after acute stroke when dysphagia-specific assessments are unavailable or incomplete. However, clinicians should interpret the results cautiously given the lack of external validation.

Study Details

Sample sizen = 126

EvidenceLevel 5

PublishedApr 2026

View Original Abstract ↓

Objectives: To develop and internally validate models that estimate swallowing status after acute stroke by using routinely available clinical variables. Methods: This retrospective observational study included patients with first-ever stroke who underwent a videofluoroscopic swallowing study or fiberoptic endoscopic evaluation of swallowing during hospitalization. Two binary outcomes were defined at the index assessment: Food Intake LEVEL Scale (FILS) [≥] 3 and FILS [≥] 7. The candidate predictors included age, sex, stroke type, Japan Coma Scale, modified Rankin Scale, Functional Independence Measure motor and cognition, serum albumin level, body mass index, and time from stroke onset. Performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration, and decision curve analysis (DCA). Stability selection was used to examine the predictor consistency. Results: Among 126 patients, 112 (88.9%) and 73 (57.9%) patients had FILS [≥] 3 and FILS [≥] 7 at the index assessment, respectively. The out-of-fold AUCs were 0.759 and 0.821 for FILS [≥] 3 and FILS [≥] 7, respectively. DCA suggested greater clinical utility for the FILS [≥] 7 model than for the FILS [≥] 3 model. Stability selection showed that cognitive and functional measures were consistently important for FILS [≥] 3, whereas albumin and time from onset additionally contributed to FILS [≥] 7. Conclusions: Routinely available clinical variables may help estimate swallowing status after acute stroke when dysphagia-specific assessments are unavailable or incomplete. The FILS [≥] 7 model showed the most favorable potential clinical utility and should be externally validated.