Evaluating a Regression-Based Approach to Norming the Test of Masticating and Swallowing Solids (ToMaSS): A Comparison with Traditional Stratified Methods

The Test of Masticating and Swallowing Solids (ToMaSS) is a commonly-used adjunctive assessment to quantify solid bolus ingestion. However, published norms are characterised by relatively small sample sizes and wide age brackets, increasing the risk of skew, influence by outliers, and reduced statistical power. Regression-based approaches to norming may mitigate some of these limitations. To evaluate whether a regression-based approach offered a significant advantage over the traditional stratified approach for the refinement of ToMaSS norms. Prospective-retrospective design, with prospectively collected data combined with historical data from published and unpublished laboratory datasets. All data consisted of healthy adults (aged 20 to 80 + years) recruited from the general public. McFadden's Pseudo R² and Root Mean Squared Error (MSE) were calculated and compared for both statistical approaches. Regression-based norms (GAM Gamma and COM-Poisson models) showed a modest improvement in model fit over traditional stratified norms, with McFadden's Pseudo R² increasing by 6% to 43% across outcomes. MSE values followed a similar trend, with regression-based norms reducing prediction error by up to 12%. Findings support that a regression-based approach offered a significant advantage over a traditional stratified approach to constructing ToMaSS norms. Advantages included smaller sample size requirements and individualised estimates across continuous age ranges. This flexibility allows for a more nuanced understanding of individual ToMaSS performance, which may be of particular relevance to clinical populations where changes in the oral phase of swallowing may vary greatly within a short time frame.