Time-Varying Risk for Breast Cancer Following Initiation of Glucose-Lowering Therapy in Women with Type 2 Diabetes: Exploring Detection Bias.

Abstract

To explore detection bias in the association between glucose-lowering therapies and breast cancer in a cohort of women with type 2 diabetes. This was a retrospective, population-based cohort study. We identified new users of metformin, sulfonylureas, thiazolidinediones and insulin during the index period of January 1, 2003, to December 31, 2010. The main outcome was incident breast cancer, and patients were followed up from drug exposure index date until death, diagnosis of another type of cancer, termination of medical insurance or December 31, 2010. To explore detection bias, we split follow-up time into 2 discrete time periods of 0 to 3 months and 3 months to 6 years after drug index date. We performed time-varying Cox regression analyses, including duration of cumulative drug exposure and ever/never drug exposure for each glucose-lowering therapy into our model. The reference was no use of the same drug-exposure category. There were 22,169 women with type 2 diabetes, with a mean (SD) age of 53.0 (9.2) years and mean (SD) follow up of 2.2 (1.5) years. Hazard ratios for breast cancer in the first 3 months following initiation of metformin, sulfonylurea or thiazolidinedione were 0.66 (0.43 to 1.02), 0.74 (0.44 to 1.25) and 0.67 (0.38 to 1.18), respectively. In the later period of 3 months to 6 years following drug start, hazard ratios (95% CI) for breast cancer were 1.00 (0.98 to 1.02), 1.01 (0.98 to 1.03) and 0.98 (0.95 to 1.01) for metformin, sulfonylurea and thiazolidinedione cumulative exposure, respectively. Our findings suggest that no detection bias exists for glucose-lowering therapies and breast cancer in this population.