Abstract
INTRODUCTION: While type 2 diabetes (T2DM) has become a major health issue in the North American and Caribbean region, the effects of weight change on incident T2DM, conditional on either initial or attained weight, are poorly addressed. Therefore, we aimed to assess the impact of 3-year weight change on incident T2DM over 6 years among US individuals. METHODS: A total of 8377 participants aged 45-64 years (4601 women), free of T2DM or cancer at baseline from the Atherosclerosis Risk in Communities (ARIC) study were included. Weight measurements were taken at baseline (visit 1, 1987-89) and approximately 3 years later (visit 2, 1990-92). Participants were categorised based on their weight change ratio into ≥ 5% weight loss, stable (±5%), and ≥ 5% weight gain. Cox proportional hazards models, adjusting for known diabetes risk factors, were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of incident T2DM, with stable weight (±5%) as the reference category. RESULTS: During a median follow-up period of 6 years, participants were classified into three categories: 361 persons remained stable (±5%), 47 with ≥ 5% loss, and 135 with ≥ 5% gain. In multivariable analysis, after adjustment with initial weight, ≥ 5% weight gain and loss were significantly associated with higher [HR (95% CI): 1.68 (1.36-2.06), p-value < 0.0001] and lower [0.73 (0.53-1.00), p-value = 0.05] risks of incident T2DM, respectively. When adjusted for attained weight, weight gain ≥ 5% remained a significant risk factor for T2DM [1.51 (1.21-1.88)]; however, weight loss ≥ 5% lost statistical significance [0.84 (0.60-1.17), p-value = 0.31]. CONCLUSIONS: We found a robust association between weight gain and incident T2DM; however, the beneficial impact of weight loss was significantly attenuated after considering the attained weight.