| Abstract: | BACKGROUND: The association between metformin and colorectal cancer (CRC) has rarely been investigated in Asian populations. METHODS: This retrospective cohort study included patients with newly diagnosed type 2 diabetes during 1999-2005, recruited from Taiwan's National Health Insurance database. A total of 169,601 patients (original cohort: 153,270 ever-users and 16,331 never-users of metformin) and a subgroup of 1:1 propensity-score-matched pairs of 16,331 ever-users and 16,331 never-users (matched cohort) were followed up to 31 December 2011. Cox regression was constructed with the inverse probability of treatment-weighting, using propensity scores, and was used to estimate hazard ratios (HRs). RESULTS: In the original cohort, the incidence of CRC was 242.9 and 480.9 per 100,000 person-years, respectively, in ever- and never-users. The overall HR [0.50, 95% confidence interval (CI): 0.45-0.56] suggested a significantly lower risk in metformin users, while compared with never-users, the HR (95% CI) for the first (<27.1 months), second (27.1-58.1 months) and third (>58.1 months) tertiles of cumulative duration of metformin therapy was 0.86 (0.76-0.98), 0.51 (0.45-0.59) and 0.26 (0.23-0.30), respectively. Analyses in the matched cohort showed similar findings with an overall HR of 0.62 (0.53-0.74), and a tertile analysis HR of 1.02 (0.81-1.28), 0.70 (0.56-0.89) and 0.32 (0.23-0.43), respectively. Re-analyses using more stringent diagnoses of CRC and cumulative duration as a continuous variable have consistently supported a protective effect with metformin use. CONCLUSION: Metformin is associated with a lower frequency of CRC. |
