| Abstract: | ObjectiveTo compare the risk of hearing loss with regard to metformin exposure. Study DesignRetrospective cohort. SettingTaiwan's National Health Insurance database. MethodsWe enrolled 292,071 ever users and 18,200 never users of metformin with new-onset diabetes mellitus from 1999 to 2005 and followed them for hearing loss from January 1, 2006, to December 31, 2011. Hazard ratios (HRs) weighted by propensity score were estimated. ResultsHearing loss was newly diagnosed in 10,085 ever users and 1072 never users. Their respective incidence rates (per 100,000 person-years) were 738.09 and 1366.83. The HR comparing ever-to-never users was 0.534 (95% confidence interval [CI]: 0.501-0.569]. The HR (95% CI) for the first (<27.07 months), second (27.07-59.13 months), and third (>59.13 months) tertiles of cumulative duration of metformin therapy were 0.912 (0.852-0.975), 0.544 (0.508-0.582), and 0.275 (0.255-0.295), respectively; and were 0.900 (0.841-0.962), 0.531 (0.496-0.569), and 0.293 (0.273-0.315), respectively, for the first (<796.70 g), second (796.70-2020.15 g), and third (>2020.15 g) tertiles of cumulative dose. The magnitude of risk reduction became more remarkable in corresponding to the increasing tertiles of the defined daily dose prescribed. Subtype analyses suggested that the risk reduction was more significant for sensorineural than conductive hearing loss. Findings derived from a propensity score-matched cohort did not substantially change the conclusions, and the risk reduction for mixed hearing loss was not statistically significant in the matched cohort as significantly observed in the unmatched cohort. ConclusionThe risk of hearing loss is reduced in a dose-response pattern in patients who use metformin. |
