论文部分内容阅读
目的探讨适合我国农村人口的糖尿病风险评估及预测模型。方法检索国外各种糖尿病预测模型,用ROC曲线分析验证各种预测模型在1461例江苏省高邮农村地区非确诊糖尿病人群中的适用性。结果共有9个临床模型(仅利用简单的病史资料和体格检查的指标)和5个生化模型(在临床模型的基础上加上生化或血液检测指标)纳入验证试验。在9个临床模型中,DESIR(女)模型的ROC曲线下面积(AUC)最大,但其灵敏度较差,仅为52.72%。Danish模型的AUC较大(0.640),灵敏度最高(68.04%)。在生化模型中,ARIC(临床指标+空腹血糖)模型的AUC最大(0.814),其灵敏度和特异度分别为69.07%和78.85%。PROCAM模型的AUC次之(0.809)。结论 Danish模型可以利用简单的问诊和体格检查给出糖尿病风险评估,适用于经济状况较差、不具备大规模生化检测条件的地区。对于条件较好的地区,ARIC模型和PROCAM模型可以进一步提高糖尿病风险的预测效能。
Objective To explore the diabetes risk assessment and prediction model suitable for rural population in China. METHODS: A variety of foreign diabetes prediction models were retrieved and validated by ROC curve analysis in 1461 unrelated diabetic patients in rural areas of Gaoyou, Jiangsu Province. Results A total of 9 clinical models (using simple history data and physical examination indicators) and 5 biochemical models (based on the clinical model plus biochemical or blood test indicators) were included in the validation test. In 9 clinical models, the area under the ROC curve (AUC) of the DESIR (female) model was the largest, but its sensitivity was poor, only 52.72%. The Danish model has a larger AUC (0.640) and the highest sensitivity (68.04%). In the biochemical model, ARIC (clinical indicators + fasting blood glucose) model AUC maximum (0.814), the sensitivity and specificity were 69.07% and 78.85%. The AUC of the PROCAM model was second (0.809). Conclusions The Danish model provides a simple assessment of diabetes risk using a simple interrogation and physical examination. It is suitable for areas with poor economic status and large-scale biochemical tests. For regions with better conditions, ARIC models and PROCAM models can further improve the predictive power of diabetes risk.