不同肥胖指标对糖尿病风险预测效果比较的性别差异

Gender-based comparison of different measures of body indexes in predicting risk of diabetes

目的比较不同肥胖指标对糖尿病风险的预测价值。方法基于深圳市宝安区2009年慢性病流行病学调查数据,对受检者的体质指数（BMI）,腰围（WC）,腰臀比（WHR）和腰围与身高的比值（WHt R）进行标准化变换,通过比较Logistic回归分析的OR值和应用受试者工作曲线（ROC）下的面积大小,评价各肥胖指标对糖尿病的风险预测能力。结果在调整可能的混杂因素后,男性中各指标对糖尿病的风险预测能力相当,OR值于1.46-1.71之间,ROC曲线下面积于0.659-0.689之间,但综合筛选时WHR是唯一进入Logistic方程的指标,OR为1.39（95%CI：1.12-1.73）;女性中WHt R的OR值最大为2.08（95%CI：1.57-2.75）,WC次之为2.03（95%CI：1.53-2.68）,而BMI的OR值最小,仅为1.25（95%CI：1.02-1.52）,ROC曲线下面积结果类似,与BMI相比差异有统计学意义,依次为0.866（95%CI：0.824-0.907）,0.866（95%CI：0.826-0.907）和0.857（95%CI：0.816-0.898）,综合筛选时WHt R是唯一进入logistic方程的指标,OR为2.08（95%CI：1.57-2.75）。结论 WC／WHR／WHt R预测糖尿病风险优于BMI,男性WHR和女性WHt R相对重要。

Objective To compare values of different body indexes for predictng diabetes risk. Methods Based onchronic disease epidemiology survey data in 2009 in Baoan district of Shenzhen,after standardized transformation of BMI,waistcircumference（WC）,waist/hip ratio（WHR）and waist/height ratio（WHt R）,the ORs in Logistic regression model and areaunder the ROC curve were used to evaluate the power of predicting risk of diabetes. Results After adjusting for potentialconfounders,in males,the ORs and areas under ROC curve were similar and ranged from 1.46- 1.71 and 0.659- 0.689,respectively. Meantime,WHR was the only one enter the logistic regression model and OR was 1.39（95% CI：1.12- 1.73）.Whilst in females,the ORs of diabetes for WHt R was 2.08（95%CI：1.57-2.75）,WC 2.03（95%CI：1.53-2.68）and BMI 1.25（95%CI：1.02-1.52）. Similarly,the areas under ROC for WHt R,WC and BMI were significantly different with 0.866（95%CI：0.824-0.907）,0.866（95%CI：0.826-0.907）and 0.857（95%CI： 0.816-0.898）. Meantime,WHt R was the only one enter thelogistic regression model and OR was 2.08（95%CI：1.57-2.75）. Conclusion WC/WHR/WHt R have stronger power than BMIfor predicting risk of diabetes,especially WHR for males and WHt R for females.