病例代理者提供的信息对研究结果的影响及校正方法

Study on the Effectiveness of Surrogate Information on Results of Case control Study

本文应用肝癌病人及代理者所提供的患者过去的暴露信息分别与对照组作病例对照研究分析，并对应用这两者的资料所得出的危险度（ＯＲ）作了比较。结果表明：应用病例本身提供的信息和其代理者提供的信息所计算出来的ＯＲ值有较大的差别。在所调查的１４个因素中，来源于这两种信息的ＯＲ值的总的差值为１７．４８，平均值为１．２５。采用Ｋｒｕｅｇｅｒ提出的二值变量的校正方法以及在此基础上我们建立的等级变量的校正方法，应用４８０例恶性肿瘤患者及其代理者的信息对肝癌病人的代理者的信息进行校正。校正后的代理者信息再与对照组作ＯＲ分析。结果显示来源于校正后代理者信息的ＯＲ值与来源于病例本身的信息的ＯＲ值之间的差异大大缩小。总的差值为７．５４，平均差值为０．５４。提示：在一些通过代理者获得信息的病例对照研究中。

In order to assess the effectiveness of surrogate information on the results of case control study, the exposure information of cases obtained from cases and their surrogates were analyzed in a case control study of hapotocarcinoma. 60 inpatients with hapotocarcinoma and their relatives(surrogate) were selected from several hospitals in Guangzhou. 480 inpatients with other malignant tumor and their relatives wete also selected from the same hospitals, All the cases and surrogates were visited in hospital under the condition that cases and surrogates can not communicate each other . The same questionaire about the past exposure of cases was applied in the study. 60 controls were selected in the same area. The surrogate variables were corrected by means of an extimated reclassification matrix based on the respones from 480 cases of other malignant tumors and their surrogates. 14 variables in study were used to calculate OR. Three kinds of OR were calculated. OR1 was calculated from the data obtained from 60 cases of hapotocarcinoma and 60 controls, OR2 from the data obtained from 60 cases' surrogates and 60 controls, OR3 from corrected surrogates of 60 cases and controls. The results showed that the total difference of 14 variables between the OR1 and OR2was 17.48 (averiage 1.25).The total difference of 14 variables between OR1 and OR3 was 7.54 (Avariage 0.54). It suggeste that the information dias caused by surrogates can be reduced by the correction.