论文部分内容阅读
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，应用此校正方法可缩小信息偏倚
In this paper, patients with past history of liver cancer patients and agents provided information on exposure in the past were compared with the control group case-control study and analysis of the use of these two data derived from the risk (OR) were compared. The results show that there is a big difference between the OR calculated by using the information provided by the case itself and the information provided by the agent. Among the 14 factors investigated, the total difference of OR values derived from these two kinds of information was 17.48 with an average of 1.25. Using the method of Krueger’s method of correcting binary variables and the calibration method we established based on this, the information of 480 malignant tumor patients and their agents was used to correct the information of the agents of liver cancer patients. The corrected agency information was compared with the control group for OR analysis. The results show that the difference between the OR value derived from the corrected surrogate information and the OR value derived from the case itself is greatly reduced. The total difference is 7.54 and the average difference is 0.54. Hint: In some case-control studies that obtained information through agents, applying this method of correction can narrow the information bias