Fat preference, not sweet, tracked worse outcomes

Dietary research usually measures what people report eating, which is burdensome and error-prone; how much people say they like particular foods is far quicker to capture and may reveal stable preference patterns. If liking patterns track disease risk, they could help target health promotion at the people whose tastes pull them towards higher-risk diets. It also lets researchers ask a specific question that gets muddled in intake data: is it the fat, the sweetness, or the resulting body weight that carries the risk? This study set out to separate those strands in a very large cohort.

The analysis included 182,181 participants from UK Biobank who rated foods on a 9-point hedonic scale, with scores of 6-9 classed as liking and 1-5 as disliking. Ratings were grouped into high fat savoury (HFSa), high fat sweet (HFSw), vegetable, low fat savoury (LFSa) and fruit-based low fat sweet (LFSw) patterns. Cox proportional hazards models estimated hazard ratios with 95% confidence intervals for outcomes including all-cause mortality, type 2 diabetes, cardiovascular disease and several cancers, adjusted for relevant confounders. Models were then re-run with body mass index added to test whether adiposity explained any associations.

Compared with disliking, liking high fat savoury foods was associated with higher risk of type 2 diabetes (HR 1.14; 95% CI 1.09-1.19) and lung cancer (HR 1.10; 95% CI 1.01-1.20). Liking high fat sweet foods was associated with higher all-cause mortality and cancer risk, with hazard ratios in the range 1.03-1.07 (CI range 1.00-1.14). In the other direction, liking low fat savoury foods was associated with lower risk of all-cause mortality, type 2 diabetes, cardiovascular disease, all-cause cancer and colorectal cancer (HR range 0.85-0.97), and liking low fat sweet foods with lower risk of diabetes, cardiovascular disease and colorectal cancer (HR range 0.91-0.96). Sweet preference alone was not associated with increased risk, and most associations were not attenuated after adjusting for BMI.

The findings complicate the popular framing that a sweet tooth is the main dietary villain: here the adverse signals clustered around higher-fat preference patterns instead. That most associations persisted after BMI adjustment argues against body weight being the sole pathway, though BMI is a crude measure of adiposity and adjustment can be incomplete. Because liking is not intake, the results leave open how much of the risk reflects what people actually eat versus other habits that travel with those preferences. Effect sizes are modest and several intervals brush against no effect, so replication in cohorts with measured diet is needed.

This is a reason to think that the fat character of the foods someone gravitates towards may carry more information about long-term risk than sweetness alone. It does not mean sweet foods are harmless or that liking a food determines your health; these are population-level associations with small hazard ratios. The more practical insight is that taste preferences are patterned and measurable, which is why researchers see them as a route to targeting dietary advice. Nothing here identifies an amount or a food to change.