Attenuation of the association between sugar-sweetened beverages and diabetes risk by adiposity adjustment: a secondary analysis of national health survey data

Yi Jing, Thang S. Han, Majid M. Alkhalaf, Michael E. J. Lean

While weight gain and obesity are the dominant factors, dietary sugar and specifically sugar-sweetened beverages (SSB) has been implicated in causing type 2 diabetes (T2DM). We assessed how much of the apparent effect of SSB is explained by adiposity, but not captured by adjustment for BMI, which is a poor index of body fat. We examined data from 5187 adults (mean age 50.8 years, SD = 16.4, 172 (3.3%) T2DM), from the Scottish Health Survey 2003 and 2008-2010 databases. Logistic regression was used to assess the association between SSB consumption and T2DM (non-insulin treated) and its attenuation (reduction in odds ratios, ORs), after entering published anthropometric indices of adiposity into the regression model, adjusted for age, sex, social class, education, smoking, alcohol consumption and physical activity. Compared with low SSB categories ("less often/never", once/week or 1-3 times/month), the OR without adiposity adjustment for having T2DM in high SSB consumers (2-3, 4-5, ≥ 6/day) was 2.56 (95% CI 1.12-5.83; p = 0.026). That OR was marginally changed by adjusting for BMI (+ 4.3%), WC (+ 5.5%) or total body fat (- 4.3%), but greatly attenuated by adjusting for estimated %body fat (- 23.4%). These indices had similar influences on the associations between SSB and T2DM combining known T2DM patients with unknown HbA1c > 6.5%, > 48 mmol/mol. Associations between SSB and T2DM are attenuated more markedly by adjustment with estimated %body fat than with BMI, indicating an adiposity effect not captured using BMI. Future research should employ best available estimates of adiposity.