The predictive value of family income to poverty ratio and diabetic retinopathy in adults aged 20 years and above in the United States: a cross-sectional study based on NHANES 1999-2020.

INTRODUCTION: The prevalence of diabetes and its complications is an important global health problem. Diabetic retinopathy (DR), a common and specific microvascular complication of diabetes, has seriously affected socioeconomic factors and individual quality of life. This cross-sectional study utilized data from NHANES 1999-2020 to investigate the association between family income to poverty ratio (PIR) and DR prevalence among U.S. adults aged 20 years and above. METHODS: Data from the National Health and Nutrition Examination Survey (NHANES) in 1999-2020, which included 39,210 participants, were utilized for this study. They were divided into two groups according to PIR < 5 or PIR ≥ 5. The baseline characteristics were described to show the distribution of each characteristic. Multiple linear regression and curve fitting were used to study the linear and nonlinear correlations between PIR and DR prevalence. Subgroup analysis and interaction tests were performed to test the stability of the relationship between PIR and DR prevalence. ROC analysis was performed to evaluate predictive performance. RESULTS: Among 39,210 adults aged 20 years and above, 31,907 (81.37%) were in Group 1 (PIR < 5), and 7,303 (18.63%) were in Group 2 (PIR ≥ 5). Compared with the prevalence of DR in Group 1 (2.7%), the prevalence of DR in Group 2 (1.5%) was lower. After adjusting for all covariates, a significant negative correlation was found between PIR and the prevalence of DR (OR: 0.91, 95% CI: [0.86 ~ 0.97], p = 0.002). Non-Hispanic Whites showing significant benefit (OR: 0.87, 95% CI: [0.80 ~ 0.96], p = 0.005, P for interaction = 0.017). After adjusting for age, the ROC analysis demonstrated a discriminative ability of the model with AUC of 0.747(95% CI: [0.74 ~ 0.76]). CONCLUSION: The results of this study indicate that an increase in PIR is correlated with a decrease in the likelihood of DR occurrence, suggesting its potential application value as an indicator for predicting the likelihood of DR occurrence. More screening and attention should be provided to low-income populations to reduce preventable social burdens.