中文摘要：

目的：探讨能够预测非肥胖者是否发生非酒精性脂肪肝（Nonalcohohc fatty liver disease，NAFLD）的临床指标。方法：从广州社区人群中选取体重指数〈25且年龄、性别相匹配的NAFLD和非NAFLD个体分别为38和82例，测量其身高、体重、腰围、臀围及空腹血糖、甘油三脂、胆固醇、低密度脂蛋白、高密度脂蛋白、HBsAg和空腹胰岛素，计算体重指数、腰臀比、腰围身高比和HOMA胰岛素抵抗指数。先采用t检验和x^2检验对上述临床指标进行分析，对两组间存在显著差异者进行Logisde回归以发现独立的预测指标，再针对各预测指标进行受试者工作特征（reciever operating charactefistic，ROC）曲线分析判断各指标的预测准确度，并确定最佳的预测截断值。结果：NAFLD和非NAFLD的体重、腰围、臀围、体重指数、腰臀比、腰围身高比及空腹血糖、甘油三脂、低密度脂蛋白、胰岛素、HOMA胰岛素抵抗指数均有显著差异，但仅腰围、低密度脂蛋白和HOMA胰岛素抵抗指数进入Logistic回归方程，且其ROC曲线下面积均大于0．5（分别是0．821，0．665和0、722）。以腰围的预测准确度最高，且在80．5cm处敏感性和特异性之和最大。结论：腰围是预测非肥胖者是否发生NAFLD的合适指标，80．5cm为其最佳预测截断值。

英文摘要：

Objective： To discover predictors for occurrence of non- alcoholic fatty liver disease（NAFLD） in non- obese individuals. Methods： 38 NAFLD and 82 non - NAFLD individuals with body mass index less than 25, who were age and gender- matched, were recruited from Guangzhou commtmities. Height, body weight, waist circumference, hip circumference, fasting blood glucose, triglyceride, cholesterol, low density lipoprotein, high density lipoprotein, insulin and HBsAg were measured, and body mass index, waist hip ratio, waist height ratio and HOMA insulin resistance index were calculated. Parameters mentioned above were first tested with t test and x2 test, and those with significant difference between NAFLD and non - NAFKD were further analyzed by logistic regression. Reciever operating characteristic （ROC） curve of each independent predictor found by logistic regression was dravm to judge the accuuracy of prediction, and the optimal cut - off value was decided with maxi- mum stun of sensitivity and specificity. Results： Body weight, waist circumference, hip circumference, body mass index, waist hip ratio, waist height ratio, fasting blood glucose, triglyeefide, low density lipoprotein, insulin and HOMA resistance index of NAFLD were significantly different from those of non - NAFLD, but only waist circumference, low density lipoprotein and HOMA insulin resistance index entered logistic regression equation with area under ROC crave more than 0.5 （0.821,0.665 and 0.722 respectively）. Predicting accuracy of waist circumference was the best and the sum of specificity and sensitivity reached peak at 80.5cm. Conclusions： Waist circumference is a suitable parameter to predict occurrence of NAFLD in non- obese individuals, and 80.5cm is its optimal cut - off value.