Context
Observational studies have reported lower risks of type 2 diabetes with higher vitamin K1 intake, but these studies overlook effect modification due to known diabetes risk factors.

Objective
To identify subgroups that might benefit from vitamin K1 intake, we examined associations between vitamin K1 intake and incident diabetes overall and in subpopulations at risk of diabetes.

Methods
Participants from the prospective cohort, the Danish Diet, Cancer, and Health Study, with no history of diabetes were followed up for diabetes incidence. The association between intake of vitamin K1, estimated from a food frequency questionnaire completed at baseline, and incident diabetes was determined using multivariable-adjusted Cox proportional-hazards models.

Results
In 54 787 Danish residents with a median (interquartile range) age of 56 (52-60) years at baseline, 6700 individuals were diagnosed with diabetes during 20.8 (17.3-21.6) years of follow-up. Vitamin K1 intake was inversely and linearly associated with incident diabetes (P < .0001). Compared to participants with the lowest vitamin K1 intake (median:57 µg/d), participants with the highest intakes (median:191 µg/d) had a 31% lower risk of diabetes (HR; 95% CI, 0.69; 0.64-0.74) after multivariable adjustments. The inverse association between vitamin K1 intake and incident diabetes was present in all subgroups (namely, men and women, ever and never smokers, low and high physical activity groups, and in participants who were normal to overweight and obese), with differences in absolute risk between subgroups.

Conclusion
Higher intake of foods rich in vitamin K1 was associated with a lower risk of diabetes. If the associations observed are causal, our results indicate that more cases of diabetes would be prevented in subgroups at higher risk (men, smokers, participants with obesity, and those with low physical activity).

diet, non–insulin-dependent diabetes, nutrition epidemiology, obesity, phylloquinone, type 2 diabetes mellitus