Abstract
We investigated associations between food liking and risk of health outcomes in a prospective cohort and explored if associations were potentially explained by body mass index (BMI). We included 182,181 participants from UK Biobank. Food liking was measured using a 9-point hedonic scale, covering liking (points 6-9)/disliking (points 1-5) for high fat savoury (HFSa), high fat sweet (HFSw), vegetable, low fat savoury (LFSa), and fruit-based, low fat sweet (LFSw) foods. Cox proportional hazard models assessed associations with health outcomes, as hazard ratios (HR) with 95% confidence intervals (CI), adjusted for relevant confounders. Compared to disliking, liking HFSa was associated with higher risks of type-2-diabetes (T2D) (HR 1.14; 95% CI: 1.09-1.19) and lung cancer (HR 1.10; 95% CI: 1.01-1.20). Liking HFSw was associated with higher risks of all-cause mortality, and cancer (HR range 1.03-1.07; 95% CI range: 1.00-1.14). Liking LFSa was associated with lower risks of all-cause mortality, T2D, cardiovascular disease (CVD), all-cause cancer, and colorectal cancer (HR range 0.85-0.97; 95% CI range: 0.79-0.99). Liking LFSw was associated with lower risks of T2D, CVD, and colorectal cancer (HR range 0.91-0.96; 95% CI range: 0.86-1.00). Associations between food liking and health outcomes differed between food groups, with most appearing not to be attenuated after adjusting for BMI. The key findings were that associations with adverse health outcomes were strongest for higher-fat containing food preference patterns, while sweet preference alone was not associated with increased risk. Our study highlighted the potential role of food liking in designing and targeting health promotion interventions.</p>