Purpose: The aim of this study was to explore the association between diabetic retinopathy (DR) and the risk of 6-year death, as well as the association between visual impairment (VI) and the risk of 6-year death in a rural Chinese population of age >= 30 years. Methods: This was a population-based cohort study. In 2006-2007, 6,830 subjects aged >= 30 years were recruited from 13 villages in Northern China through clustered randomization. In 2012-2013, a 6-year follow-up was further done. Six different proportional hazards models, with different confounders adjusted, were used to explore the association between baseline DR and risk of death. Results: 5,570 subjects were included in this study by our inclusion and exclusion criteria. Four hundred and ten (7.36%) subjects died by follow-up. The median ages of the dead subjects and survived subjects were 67 (interquartile range [IQR]: 58-72) years and 52 (IQR: 42-58) years (Z = 21.979, p < 0.001). Male accounted for 62.20 and 44.92% among the dead and survived subjects (chi(2) = 45.591, p < 0.001). Besides, compared with those survived, the dead were found to be with lower education (chi(2) = 109.981, p < 0.001), lower marriage rate (chi(2) = 101.341, p < 0.001), lower income (chi(2) = 123.763, p < 0.001), higher proportion of smoking (chi(2) = 8.869, p = 0.003), higher systolic blood pressure (Z = 10.411, p < 0.001), lower body mass index (Z = -3.302, p = 0.001), larger spherical equivalent error (Z = 4.248, p < 0.001), lower intraocular pressure (Z = -4.912, p < 0.001), smaller anterior chamber depth (Z = -9.186, p < 0.001), larger length thickness (Z = 11.069, p < 0.001), higher fast blood glucose level (Z = 5.650, p < 0.001), higher total cholesterols (Z = 2.015, p = 0.044), higher low-density lipoprotein (Z = 2.024, p = 0.043), and higher proportion of drug usage (chi(2) = 56.108, p < 0.001). Besides, the dead subjects were more likely to be with VI, glaucoma, cataract, age-related macular degeneration, diabetes, and DR. Hundred and forty-eight subjects were diagnosed with DR at baseline, and 33 (22.30%) of them were dead before follow-up. By adjusting all relative confounders in a proportional hazards model, DR was found to be a risk factor of 6-year death, the hazard ratio was 1.739 (95% confidence intervals: 1.080, 2.803). Another 5 different statistical models with different confounders adjusted also revealed a statistically significant association between DR and 6-year death. The association between VI and 6-year death was not statistically significant. Conclusions: DR increased the risk of 6-year death in a rural Chinese population aged >= 30 years, while VI did not. (c) 2020 S. Karger AG, Basel