Statistical issues in the interpretation of diabetes as a risk factor for dementia

摘要

We read with great interest the recently published article by Cheng et al. that updated the previous systematic reviews2 of diabetes as a risk factor of dementia. The authors conclude that subjects with diabetes had significantly higher risk for Alzheimer disease (AD) (relative risk (RR) 1.46 (95% confidence interval (CI), 1.20-1.77), vascular dementia (VD) (RR 1.51 (95% CI, 2.08-2.96)), any type of dementia (ATD) (RR 1.51 (95% CI, 1.31-1.74)) and mild cognitive impairment (RR 1.21 (95% CI, 1.02-1.45)) relative to non-diabetic subjects. However, certain issues remain unresolved in the article to come to a firm conclusion. The authors have included 19 studies in the pooled analysis and have found the heterogeneity'(AD; % = 47.3, P < 0.001) (VD % = 6.3, P = 0.71) (ATD x2 = 28.9, P = 0.001) in the included studies. A test for heterogeneity examines the inconsistency across studies. Since meta-analysis brings together studies that are diverse both clinically and methodologically, heterogeneity is expected.