Abstract
BACKGROUND: The beneficial effect of pioglitazone on dementia requires confirmation. METHODS: The database of Taiwan's National Health Insurance was used to enroll a propensity score-matched-pair cohort of patients who had ever used pioglitazone and patients who had never used pioglitazone from Taiwanese patients with newly diagnosed diabetes mellitus during 1999⁻2008. The patients were to be alive on 1 January 2009 and were followed up for dementia until 31 December 2011. Hazard ratios were estimated using the Cox proportional hazards model. RESULTS: There were 11,011 never users and 11,011 ever users of pioglitazone, with respective numbers of incident dementia of 123 and 91. The overall hazard ratio was 0.716 (95% confidence interval: 0.545⁻0.940) for ever users versus never users. The hazard ratios for the first (<11.0 months), second (11.0⁻19.6 months) and third (>19.6 months) tertiles of cumulative duration were 0.806 (0.544⁻1.193), 0.654 (0.430⁻0.994) and 0.694 (0.469⁻1.026), respectively. When cumulative duration was treated as a continuous variable, the hazard ratio was 0.987 (0.976⁻0.998). In subgroup analyses, the beneficial effect was mainly observed in patients who had not been treated with metformin. Among metformin ever users, the hazard ratio for dementia for pioglitazone ever users versus never users was 0.802 (0.580⁻1.109); and was 0.494 (0.284⁻0.857) among never users of metformin. No interaction between pioglitazone and major risk factors of dementia (i.e., stroke, hypoglycemia, head injury and Parkinson's disease) was observed. CONCLUSIONS: Pioglitazone use is associated with a lower risk of dementia, especially when it is used in never users of metformin and has been used for more than 20 months.