【摘要】：缺失數(shù)據(jù)是臨床試驗(yàn)中常見但又不可避免的一個(gè)問題。缺失數(shù)據(jù)不僅會(huì)降低試驗(yàn)的把握度,還會(huì)給試驗(yàn)結(jié)果帶來(lái)偏倚。因此,一方面可以在統(tǒng)計(jì)分析中采用合適的缺失數(shù)據(jù)處理方法,另一方面要特別注意盡可能預(yù)防缺失數(shù)據(jù)的產(chǎn)生。其中,缺失數(shù)據(jù)的預(yù)防應(yīng)當(dāng)是第一位的。從數(shù)據(jù)的角度來(lái)講,首先,應(yīng)在方案設(shè)計(jì)、數(shù)據(jù)采集和數(shù)據(jù)核查的各個(gè)階段,采取合理措施提高受試者的依從性,減少不必要的數(shù)據(jù)缺失;其次,對(duì)于確認(rèn)發(fā)生的數(shù)據(jù)缺失,應(yīng)詳細(xì)記錄缺失數(shù)據(jù)產(chǎn)生的原因,這對(duì)于判定數(shù)據(jù)缺失的機(jī)制和選擇合適的缺失數(shù)據(jù)處理方法 (例如,前一次觀察數(shù)據(jù)向后結(jié)轉(zhuǎn)、多重填補(bǔ)和重復(fù)測(cè)量數(shù)據(jù)混合效應(yīng)模型等)具有非常重要的作用。
[Abstract]:Missing data is a common but unavoidable problem in clinical trials. Missing data will not only reduce the degree of understanding of the test, but also bring bias to the test results. Therefore, on the one hand, appropriate missing data processing methods can be adopted in statistical analysis, on the other hand, special attention should be paid to prevent the generation of missing data as much as possible. Among them, the prevention of missing data should be the first place. From the data point of view, firstly, reasonable measures should be taken to improve the subjects' compliance and reduce unnecessary data loss at all stages of project design, data acquisition and data verification. The causes of missing data should be recorded in detail, which is useful for determining the mechanisms for missing data and for selecting appropriate missing data processing methods (for example, the previous observation data is carried forward backwards, The mixed effect model of multi-filling and repeated measurement data is very important.
【作者單位】： 第四軍醫(yī)大學(xué)衛(wèi)生統(tǒng)計(jì)學(xué)教研室;默沙東研發(fā)(中國(guó))有限公司;
【分類號(hào)】：R969.4
，

本文編號(hào)：2146964