Background: Diabetes is associated with an increased risk of several cancers; however, greater detection of cancer around the time of diabetes diagnosis may partly contribute to this relationship. The goal of the current study was to explore the temporal relationship between diabetes and cancer incidence.
Methods: The authors conducted a retrospective, population-based cohort study of >1 million adults living in Ontario, Canada to evaluate the association between diabetes diagnosis and the incidence of cancer in 3 time periods: within the 10 years before a diabetes diagnosis, within the first 3 months after a diabetes diagnosis, and from 3 months to 10 years after a diabetes diagnosis.
Results: Individuals with diabetes were significantly more likely to have been diagnosed with cancer within the 10 years before a diabetes diagnosis compared with individuals without diabetes (odds ratio, 1.23; 95% confidence interval [95% CI], 1.19-1.27). Cancer incidence also was found to be significantly higher in individuals with diabetes within the 3-month period after a diabetes diagnosis (hazard ratio, 1.62; 95% CI, 1.52-1.74), whereas the risk was not found to be elevated in the later period (hazard ratio, 0.97; 95% CI, 0.95-0.98). Similar trends were noted for individual cancers.
Conclusions: The results demonstrated that individuals with diabetes had a significantly higher risk of most cancers, which was limited to the time periods before and immediately after a diabetes diagnosis. The highest risk period was observed within the first 3 months after a diabetes diagnosis, suggesting a partial role of detection bias in the apparent relationship between diabetes and cancer. Cancer 2016. © 2016 American Cancer Society. Cancer 2016;122:2731-2738. © 2016 American Cancer Society.
Keywords: cancer risk; detection bias; diabetes; hyperinsulinemia; temporal risk.
© 2016 American Cancer Society.