摘要 目的：糖尿病视网膜病变(Diabetic Retinopathy，DR)是糖尿病的严重并发症之一，是世界范围内劳动年龄致盲的主要原因。尽管全视网膜光凝(Panretinal Photocoagulation，PRP)是标准治疗，但PRP治疗的DR仍有很高的进展风险。因此，本研究旨在评估危险因素，并建立预测糖尿病视网膜病变(DR-恶化)在PRP后五年内恶化的模型。方法：纳入被诊断为严重的非增殖性糖尿病视网膜病变或增殖性糖尿病视网膜病变并接受PRP治疗的患者，这些患者被随机分配到对照队列中。多变量Logistic回归分析用于筛选队列中DR恶化的潜在危险因素。然后在包含显著独立危险因素后建立模型，并通过判别和校正进一步对照。结果：共纳入271名患者，56.46%的患者有DR恶化的结局。在观察队列(n=135)中，年龄(OR=0.94，CI0.90~0.98)、基线最佳矫正视力(OR=10.74，95%CI 1.84~62.52)、糖尿病肾病(OR=9.32，95%CI 1.49~58.46)、高脂血症(OR=3.34，95%CI 1.05~10.66)作为独立危险因素纳入预测模型。观察和对照队列的受试者工作特征曲线下面积和校正斜率分别为0.79、0.96(95% CI 0.60~1.31)和0.79、1.00(95% CI 0.66~1.34)。根据预测概率的最佳截断值建立两个风险组，实际发生概率在低危组和高危组分别为34.90%和82.79%(P<0.001)。结论：本研究建议的评估模型可作为临床预测DR恶化的快速风险评估系统，及早识别DR恶化的高危个体，给临床诊治提供参考。
Abstract Objective: diabetic retinopathy (DR) is one of the serious complications of diabetes and a leading cause of blindness at working age worldwide. Although panretinal photocoagulation (PRP) is the standard treatment, Dr treated by PrP still has a high risk of progression. Therefore, this study aimed to evaluate the risk factors and to establish a model for predicting the worsening of diabetic retinopathy (DR worsening) over a period of five years after PrP. Methods: patients diagnosed with severe nonproliferative diabetic retinopathy or proliferative diabetic retinopathy and treated with PRP were included and randomly assigned into a validation cohort. Multivariate logistic regression analysis was used to screen potential risk factors for worsening DR in the cohort. The model was then built after the inclusion of significant independent risk factors and further validated by discrimination and correction. Results: a total of 271 patients were included and 56.46% had an outcome of worsening Dr. In the observational cohort (n = 135), age (odds ratio [or] = 0.94, 95% confidence interval [CI] = 0.90-0.98), baseline best corrected visual acuity (LogMAR) (or = 10.74, 95% CI 1.84-62.52), diabetic nephropathy (or = 9.32, 95% CI 1.49-58.46), and hyperlipidemia (or = 3.34, 95% CI 1.05-10.66) were included as independent risk factors in the predictive model. The area under the receiver operating characteristic curve and adjusted slope were 0.79, 0.96 (95% confidence interval 0.60 to 1.31) and 0.79, 1.00 (95% confidence interval 0.66 to 1.34) for the observational and control cohorts, respectively. Two risk groups were established according to the optimal cutoff value of predicted probability, and the actual probability of occurrence was 34.90% and 82.79% in the low - and high-risk groups, respectively (P < 0.001). Conclusions: This study developed a new model and validated internally to predict the likelihood of worsening Dr over a 5-year period after treatment with PRP. This model may serve as a rapid risk assessment system for clinically predicting Dr exacerbations, identifying high-risk individuals for Dr exacerbations early, and prescribing additional treatments.