胆囊结石病高危人群预测结果的随访研究和预测准确性评判

Follow-up study on high risk population of gallbladder stone disease and evaluation of the accuracy of prediction model

目的 对胆囊结石病高危人群进行预测结果的随访研究 ,评判新建立的Logistic回归模型预测胆石病高危人群的准确性。方法 预测基地共 4 4 8人 ,病例组 10 1人和对照组 12 7人建立Model 97预测模型。无胆囊结石但不符合对照组标准的作为高危人群随访组。市区随访组 16 2人于 1991年进入预测研究 ,体重指数、血脂、血胰岛素以及B起检查胆囊壁厚和胆囊收缩分数作为预测指标 ,代入预测模型 ,求得成石概率P =0 .2作为判断成石的界值 ,P≥ 0 .2判断为有成石可能。 1997年随访研究 ,以B越复查胆囊 ,检验预测结果。近郊随访组 5 8人于 1997年进入预测研究 ,2 0 0 1年 10月随访研究 ,B超复查胆囊 ,检验预测结果。结果 市区随访组 6年后复查发现 18人新形成胆囊结石 ,14人 (77.8% )预测正确 ,近郊随访组 4年后复查 ,11人新形成胆囊结石 ,9人 (81.8% )预测正确。两个基地合计 2 3例预测正确 ,男性 18例 ,女性 5例 ,准确率为 79.3%。 18人 (6 2 % )为胆固醇结晶 ,11入 (38% )发展成胆石 ,胆石的直径或体积最大径为 5 -2 1mm。结论 预测结果的随访研究表示 ,以年龄、体重指数、血脂、血胰岛素。

Objective To follow up the high risk population of gallbladder stone disease and evaluate the accuracy of prediction model for gallbladder stone disease. Methods 448 subjects was included in this study, 101 in experimental group and 127 in control group. Both groups were used to establish the prediction model. Those without gallbladder stones and excluded in control group constituted follow-up group. In 1991 body mass index (BMI), biochemistry and were examined gallbladder ultrasonographic scanning was performed in 162 subjects of the follow-up group to get their possibility of gallstone formation using the Logistic regression model of prediction with 0.2 as the cut point. In 1997, gallbladder ultrasonography was performed and the accuracy of prediction was evaluated. Fifty-eight subjects in suburb follow-up group were studied on their possibility of gallstone formation in 1997 and followed up with gallbladder ultrasonography in Oct. 2001. Results Eighteen subjects in urban follow-up group were found to form gallbladder stone 6 years later and 14 cases (77.8%) were predicted correctly. Eleven subjects in suburb follow-up group were found to have gallbladder stone 4 years later and 9 cases (81.8%) predicted correctly. A total of 23 cases were predicted correctly (18 male and 5 fenale) with the accuracy of 79.3%. Cholesterol crystals were detected in 18 cases (62%) and gallstones in 11 cases (38%) with the size between 5 mm and 21 mm. Conclusions The population at high risk to gallbladder stone formation could be predicted more accurately and repeatedly by the Logistic regression model including the variables of age, BMI, serum lipid, serum insulin, gallbladder wall thickness and contraction function of gallbladder.