Development of an Ultrasound-Based Score for screening Swallowing Disorders: Diagnostic Accuracy and Cut-Off Points: US-Based Score for Dysphagia Screening.

Abstract

To determine whether specific ultrasound (US) measures and cut-off scores from a new point-based scoring system were able to confirm the findings of the Videofluoroscopic swallow study (VFSS) in diagnosing swallowing disorders (SD). Cross-sectional, single-center study from March 2021 to March 2023. Initial SD screening on the Penetration Aspiration Scale (PAS) and subsequent evaluation on the VFSS were administered to confirm SD. Additionally, an US study examined tongue thickness, tongue movement, hyoid bone displacement, and suprahyoid muscles. A scoring system based on these US measures was developed and cut-off scores for presence/absence of SD were determined from the US and VFSS findings. Single tertiary referral center providing both ambulatory and inpatient care. 119 adults consecutively admitted to the SD unit with suspected SD. Exclusion criteria included total laryngectomy, health conditions preventing standard evaluation (e.g., dyspnoea, musculoskeletal disorders) and prior evaluation in the unit. Not applicable. SD was able to be confirmed on the US-based score. The US cut-off scores were related to the presence or absence of SD based on the findings of the VFSS. SD diagnosed in 75 (63%) based on VFSS; and corroborated with main US features seen during swallowing. The logistic model developed included tongue movement, hyoid bone displacement, geniohyoid thickness at rest and sex (Area Under the Curve = 0.83, 95% CI 0.75-0.91). The ultrasound scoring system developed range from 0 to 5; with 0 indicating no dysphagia and 3-5 indicating SD. The optimal cut-off to rule out SD was ˂1, with a negative likelihood ratio of 0.04 and a negative post-test probability of 6%. To rule in SD, the optimal cut-off was ≥3, with a positive likelihood ratio of 5.1 and a positive post-test probability of 90%. The ultrasound-based scoring system is a promising, easy-to-use assessment tool with good diagnostic accuracy and useful post-test probabilities for SD assessment. Further research is needed before clinical adoption.