Clustering method and stimulus presentation order shift sweet liking phenotype distributions

Abstract

Sweet liking is a phenotype with an intuitive connection to nutrition and health, but phenotyping methods and association with health outcomes reported in the literature are highly variable. Recently, a “Rapid Method” of phenotyping based on 1 M sucrose solution liking was proposed, but validation of this method in new subject groups is needed. This study compared sweet liking phenotyping methods against 3 reliability criteria. Participants (n = 92, aged 18–55 years) rated their liking of both a 5-concentration series of sucrose solutions and a stand-alone 1 M sucrose solution. Participants were classified as sweet likers, sweet dislikers, or inverted U-shape using 7 hierarchical clustering analysis (HCA) linkage methods, the Rapid Method, and a method based on the slope of the liking-sucrose concentration function (“Slope Method”). The proportion of sweet likers, defined as members of the cluster with the highest mean 1 M sucrose liking, varied widely from 1 % to 48 % depending on HCA method; in comparison, the Rapid and Slope methods classified 39 % and 51 % of participants as sweet likers, respectively. Liking for sucrose solutions decreased systematically across presentations (p = 0.00046); 63 % of participants who rated the 1 M solution first met the Rapid Method cut-off for sweet likers compared to 39 % of participants who rated it last. The Slope Method achieved the highest intragroup correlations while reproducing the commonly recognized phenotypes, but performance of the Rapid Method is strong enough to warrant its continued use when convenience is of paramount importance. Replication in larger cohorts is needed to validate these results.