Abstract
Background Observational studies link lifestyle factors to diabetes, but confounding limits causal inference. This study employed Mendelian randomization (MR) to investigate the potential causal effects of major dietary, obesity, smoking, and physical activity exposures on diabetes risk. Methods A two-sample MR framework integrated FinnGen and United Kingdom Biobank (UKB) data. Genetic instruments for diet (fruits, vegetables, cheese), smoking (initiation, intensity, maternal), body mass index (BMI), and physical activity came from various consortia (n=64, 949-632, 802). Associations with diabetes odds were assessed using inverse-variance weighted analysis. Results Fruit and cheese intake and physical activity per standard deviation increase causally reduced diabetes risk in both cohorts. Conversely, smoking initiation, maternal smoking around birth, and BMI per standard deviation increase causally increased diabetes risk in both cohorts. Coffee increased diabetes risk only in FinnGen, whereas smoking intensity increased diabetes risk only in UKB. Conclusion This study provides robust evidence that modifiable lifestyle factors may have causal effects on diabetes risk. Fruit, cheese, and physical activity may protect against diabetes, whereas smoking, maternal smoking, and higher BMI appear to increase risk. Findings support public health interventions targeting diet, physical activity, smoking cessation, and healthy weight to combat the global diabetes epidemic.