临床研究中最小临床意义变化值确定方法

Methods to determine minimal clinically important difference

目前确定最小临床意义变化值的方法有效标法、分布法、专家意见法和文献分析法。效标法可对所确定的最小临床意义变化值给出专业解释,但没有考虑测量误差,不同效标会产生不同的最小临床意义变化值,而且通常情况下很难找到合适的效标。分布法考虑了测量误差,易于实施,但不同样本得出的结果可能不同,无法为所确定的最小可测变化值给出专业解释,对得出的最小可测变化值缺乏公认的判断准则。专家意见法和文献分析法则分别基于专家意见和已发表文献,只能作为辅助。在确定最小临床意义变化值时,最好能综合使用效标法、分布法和专家意见法。

There are 4 methods to determine a minimal clinically important difference （ MCID ） currently, namely anchor-based method, distribution-based method, opinion-based method, and literature review. The anchor-based method offers the clinical significance of MCID, but it does not consider the measurement error. The anchor-based method is limited due to difficulty to obtained a suitable anchor in most cases, and variation of MCID with the anchor. The distribution-based method takes measurement error into account, and is easily implemented. The distribution-based method can not provide the clinical explanation of MCID, and the MCID from different samples might differ from each other. There is no golden criterion to judge MCID by distribution-based method so far. The opinion-based method and literature review are based on the opinions of experts and the published literature, respectively. They are usually regarded as a supplement to anchor-baseds method and distribution-based method. It is suggested that anchor-based method, distribution-based method, and opinion-based method should be used together when determining a MCID.