Diagnostic Accuracy of the Eating Assessment Tool-10 (EAT-10) in Screening Dysphagia: A Systematic Review and Meta-Analysis.

Only seven studies were included, two rated high quality and five medium; several did not say whether the person reading the VFSS or FEES images knew the EAT-10 result, which tends to make a screening test look more accurate than it is. The studies disagreed a great deal at a cutoff of 3 (an inconsistency measure of 92.8%), so a single summary figure hides a wide spread; the same cutoff produced sensitivities from 0.49 in Parkinson's disease to 0.98 in a mixed high-risk group. The EAT-10 is a self-report questionnaire, so it depends on patients noticing their own difficulty. People with reduced throat sensation, such as those with Parkinson's disease, may answer that they have no problem while still aspirating. People with major psychiatric or cognitive disorders who could not complete the questionnaire were excluded, so the results do not tell you how the tool behaves in them. Version and language matter. The authors note that a version that scores higher will appear more sensitive and less specific, so a cutoff validated in one language may not transfer to another. VFSS and FEES themselves can miss aspiration that does not happen during the examination, which would make the questionnaire's measured accuracy lower than its true accuracy. The suggested changes to the questionnaire, such as removing items 2, 7 and 10, come from a separate analysis of the tool's structure and were not tested in this review.

Three funding sources are listed, all from 2021 and all in China: the Shandong Traditional Chinese Medicine Science and Technology Project, the Weifang soft science research project, and the School Innovation and Entrepreneurship Training Program of Weifang Medical University. The paper gives no competing-interest statement and says nothing about whether the funders had any role in the review.

Do not read a normal EAT-10 as ruling dysphagia or aspiration out. The pooled sensitivity was 0.89 at a cutoff of 2 and 0.85 at a cutoff of 3, so some people with swallowing problems will score below the cutoff, and one included study of Parkinson's disease found a sensitivity of only 0.49 at a cutoff of 3.