Dynamic change in body fat percentage and the risk of diabetes mellitus: a longitudinal analysis

Abstract

Although higher body fat percentage (BF%) is associated with an increased risk of diabetes mellitus (DM), most evidence is derived from single time-point measurements. We aimed to investigate whether longitudinal changes in BF% are associated with incident DM. We analyzed 11,016 Korean adults from the Korean Genome and Epidemiology Study followed biennially for up to 21 years. Sex-stratified time-varying Cox models were used to assess the associations of repeatedly measured BF% and visit-to-visit change in BF% with incident DM. BF% levels were categorized as Low, Middle, and High using sex-specific quartile cut points. During a mean follow-up of 8.8 years, 1,680 participants developed DM. Higher time-varying BF% was associated with elevated DM risk in both sexes. Each 1-percentage-point increase in BF% between consecutive visits was associated with a higher DM risk in men (HR 1.04; 95% CI 1.02–1.07) and women (HR 1.05; 95% CI 1.03–1.08). Non-obese men with High BF% had a higher DM risk than non-obese men with Middle BF% (HR 1.31; 95% CI 1.00–1.72). Among women with prior High BF%, BF% reductions were associated with lower DM risk relative to no change in BF%. However, when BF% changes were examined as category transitions, High-to-Middle/Low transitions were associated with a higher DM risk than Middle-to-Middle transitions (HR 1.36; 95% CI 1.01–1.84). Longitudinal assessment of BF% may provide complementary information on DM risk beyond body mass index by capturing both current adiposity and changes in adiposity over time.