Knapp Guido, Biggerstaff Brad J, Hartung Joachim
Department of Statistics, University of Dortmund, 44221 Dortmund, Germany.
Biom J. 2006 Apr;48(2):271-85. doi: 10.1002/bimj.200510175.
In a random-effects meta-analysis, a new confidence interval for the heterogeneity parameter is proposed. With this interval, the amount of heterogeneity in a meta-analysis can be assessed so that it can be judged whether the pooling of the estimates is meaningful. Through suitable corrections of the lower bound, based on the treatment effect measure of interest, the resulting interval yields satisfactory results with respect to the predefined confidence coefficient. Lower and upper bound of the interval can be used for one-sided hypothesis testing on the amount of the underlying between-trial variability.
在随机效应荟萃分析中，提出了一种用于异质性参数的新置信区间。利用该区间，可以评估荟萃分析中的异质性程度，从而判断估计值的合并是否有意义。基于感兴趣的治疗效应测量，通过对下限进行适当校正，所得区间在预定义的置信系数方面产生了令人满意的结果。该区间的下限和上限可用于对试验间潜在变异性的大小进行单侧假设检验。
