Abstract
OBJECTIVE: Pancreatic cancer diagnoses are frequently preceded by a new diabetes diagnosis. Screening individuals newly diagnosed with diabetes could enable the earlier detection of pancreatic cancer. We sought to estimate the risk of pancreatic cancer by age, sex, race and time since diabetes diagnosis. DESIGN: Claims-based cohort study. SETTING: Johns Hopkins Medicine conducted this deidentified claims-based cohort study using the Optum Labs Data Warehouse. PARTICIPANTS: Insurance enrollees from 1/2008-9/2018 were identified as non-diabetic or newly diagnosed diabetic. Our risk set included 4 732 313 individuals (424 129 newly diabetic) in 5 844 934 enrolment periods. PRIMARY OUTCOME MEASURES: Time to pancreatic cancer. Diabetes and cancer were defined using International Classification of Diseases (ICD)-9/10 codes. RESULTS: Individuals with newly diagnosed diabetes were at an increased HR of pancreatic cancer, but this effect waned over time. The HR of pancreatic cancer following a diabetes diagnosis was higher in younger individuals and varied by race (lower HR in non-White individuals) (p<0.01, main effects and interactions). Thus, the probability of pancreatic cancer following a diabetes diagnosis was dependent on age, race and sex. For example, the 1-year probability of pancreatic cancer in a White male aged 75 was 0.45% (95%CI 0.41% to 0.49%) if they were newly diagnosed with diabetes and 0.090% (95%CI 0.084% to 0.096%) if they were free of diabetes. In contrast, the risk was lower at 0.15% (new-diabetic, 95% CI 0.13% to 0.16%) and 0.022% (diabetes free, 95%CI 0.020% to 0.023%) at age 55. The HR of pancreatic cancer for individuals with newly diagnosed diabetes compared with those free of diabetes was highest in the month following diagnosis (HR=14.7 and 9.6 for a 55 and 75 year old White male, respectively) but decreased in the following months, with a HR of 7.8 and 5.8 at 3 months, 5.6 and 4.1 at 6 months, and 3.9 and 2.8 at 1 year (p<0.01). CONCLUSIONS: Consideration of the age-race-sex specific probability of pancreatic cancer and time since diabetes diagnosis is necessary when evaluating the risk of pancreatic cancer following a diabetes diagnosis.