BACKGROUND: High dairy/milk intake has been associated with a low risk of type 2 diabetes observationally, but whether this represents a causal association is unknown.

OBJECTIVE: We tested the hypothesis that high milk intake is associated with a low risk of type 2 diabetes and of overweight-obesity, observationally and genetically.

DESIGN: In 97,811 individuals from the Danish general population, we examined the risk of incident type 2 diabetes and of overweight-obesity by milk intake observationally and by LCT-13910 C/T genotype [polymorphism (rs4988235) upstream from the lactase (LCT) gene], where TT and TC genotypes are associated with lactase persistence and CC with nonpersistence.

RESULTS: Observationally for any compared with no milk intake, the HR for type 2 diabetes was 1.10 (95% CI: 0.98, 1.24; P = 0.11), whereas the OR for overweight-obesity was 1.06 (1.02, 1.09; P = 0.002). Median milk intake was 5 glasses/wk (IQR: 0-10) for lactase TT/TC persistence and 3 (0-7) for CC nonpersistence. Genetically for lactase TT/TC persistence compared with CC nonpersistence, the OR was 0.96 (0.86, 1.08; P = 0.50) for type 2 diabetes and 1.06 (1.00, 1.12; P = 0.04) for overweight-obesity. In a stratified analysis for type 2 diabetes, corresponding values in those with and without milk intake were 0.88 (0.76, 1.03; P = 0.11) and 1.35 (1.07, 1.70; P = 0.01) (P-interaction: 0.002), whereas no gene-milk interaction on overweight-obesity was found. For a 1-glass/wk higher milk intake, the genetic risk ratio for type 2 diabetes was 0.99 (0.93, 1.06), and the corresponding observational risk was 1.01 (1.00, 1.01). For overweight-obesity, the corresponding values were 1.01 (1.00, 1.02) genetically and 1.00 (1.00, 1.01) observationally.

CONCLUSIONS: High milk intake is not associated with a low risk of type 2 diabetes or overweight-obesity, observationally or genetically via lactase persistence. The higher risk of type 2 diabetes in lactase-persistent individuals without milk intake likely is explained by collider stratification bias.