Work

Self-report and objective measures of swallowing capture different constructs in community-dwelling older people: a cross-classification study across two surveys

Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background: Self-report questionnaires such as the Eating Assessment Tool are widely recommended for community dysphagia screening, but evidence for them comes largely from clinical populations, and the abbreviated version was validated against its parent questionnaire. Whether they measure the same thing as an observed swallowing task is unresolved. We examined agreement between two self-report instruments and an objective water swallow test, tested whether cognitive impairment accounts for discordance, and assessed whether objective measures carry information the questionnaires do not. Methods: Cross-classification analysis of two consecutive community surveys in eastern China: 2024 (2,886 participants, 121 positive tests) derived a model; 2025 (2,765 participants, 230 positive tests) validated it. The comparator was a positive 30 mL water swallow test, grade II or above. Because the comparator is a bedside observation, not an instrumental reference standard, agreement was reported as positive and negative percent agreement and Cohen's kappa. Agreement was examined by cognitive stratum in both surveys, and discrimination was compared with DeLong tests. Results: Agreement between the instruments was slight in both surveys (kappa 0.022 and 0.101; positive percent agreement 15.7% and 27.0%). Agreement did not fall with cognitive impairment: it was higher in the impaired group in 2025 (kappa 0.280 versus 0.082), driven by a higher prevalence of positive tests (15.2% versus 7.9%) rather than closer classification, but this was not reproduced in 2024 (-0.011 versus 0.025). A five-variable model discriminated the comparator outcome better than the questionnaire (0.661 versus 0.595, P = 0.010), but tongue pressure, calf circumference and cognitive score added nothing to age and grip strength (0.659; P = 0.971). Calibration remained acceptable when prevalence rose from 4.2% to 8.3%. Conclusions: A self-report dysphagia questionnaire and an objective swallow test classify the same older people differently, and cognitive impairment does not explain this. They should be treated as measuring different constructs rather than as interchangeable screens, and pathways should state which construct they intend to capture. Clinical trial number: Not applicable.
Institution: Zhejiang University
Language: English
Open Access status: green
OpenAlex ID: W7236860134
Cites: 30
Cited by:
Related to: 10