Objective To investigate the effects of diabetes and smoking on occurrence of pancreatic cancer (PC) and their interaction.

Methods Based on the phenotype data of about 500 000 participants in the UK Bio-bank (UKB), the subjects were divided into people with diabetes and non-diabeties, smokers and non-smokers. Multivariate adjusted Logistic regression model was used to analyze the association between smoking, diabetes and PC. The interaction between the two risk factors was assessed by an additive model.

Results A total of 492 637 participants were included, including 1 223 patients with PC and 491 414 patients without PC. After adjusting for age, gender, race, body mass index (BMI) and other confounding factors, it was founded that the risk of PC in people with diabetes (OR=1.999, 95%CI, 1.704 to 2.344) and people with smoking (OR=1.246, 95%CI, 1.109 to 1.400) were significantly higher than those of people with non-diabetes or non-smokers (P < 0.05). People who smoked and had a history of diabetes had a nearly 4-fold increased risk of developing PC compared with those who never smoked and had no history of diabetes (OR=3.561, 95%CI, 2.949 to 4.300, P < 0.001).

Conclusion Diabetes and smoking are independent risk factors for PC, and their interaction increases the risk of PC.