We performed a meta-analysis to analyze the associations of depression with pre-diabetes (PreDM), undiagnosed diabetes (UDM), and previously diagnosed diabetes (PDM), and whether the association was affected by important study characteristics. We searched relevant articles published in PubMed and EMBASE up to August, 2015. Studies reporting cross-sectional associations of depression with PreDM, UDM, or PDM compared with normal glucose metabolism (NGM) were included. Odds ratios (ORs) were pooled with random-effect and fixed-effect models. Subgroup analyses by sex, study mean age, different degrees of adjustment, publication year, quality score, and depression assessment scales were also performed. Twenty studies were eligible and included in current analysis. Summary estimates showed that compared with NGM individuals, prevalence of depression was moderately increased in PreDM (random-effect odds ratio (OR) 1.11, 95 % confidence interval (CI) 1.03-1.19) and UDM (OR 1.27, 95 % CI 1.02-1.59), and markedly increased in PDM (OR 1.80, 95 % CI 1.40-2.31). Subgroup analyses showed that the positive association remained only among studies with mean age <60 years old but not among those with mean age ≥60 years old. Summary estimates of ORs with cardiovascular disease adjustment substantially attenuated the association. Our findings suggested that risk of prevalent depression was gradually increased with the deterioration of glucose metabolism among younger age groups but not among older age groups. Comorbid cardiovascular diseases might be an important intermediate factor underlying the association between depression and diabetes.
Keywords: Depression; Meta-analysis; Pre-diabetes; Undiagnosed diabetes.