Friedman M A
Stat Med. 1987 Apr-May;6(3):307-14. doi: 10.1002/sim.4780060318.
The pooling of large amounts of clinical data from all available and relevant (diverse) sources in order to achieve greater statistical confidence may be appropriate for most modern and well conducted clinical investigations. However, in order to arrive at specific and meaningful conclusions, such exercises absolutely depend upon the homogeneity (or at least comparability) of the study population and therapy under scrutiny. There are currently many opportunities for pooling studies of cancer treatment. The most attractive disease situations are those with a sufficiently large number of completed randomized studies. Some to be considered are colon cancer, gastric cancer, hepatoma, ovarian cancer and lymphoma.
为了获得更高的统计可信度，从所有可用且相关（多样）的来源汇集大量临床数据，这对于大多数现代且开展良好的临床研究而言可能是合适的。然而，为了得出具体且有意义的结论，此类做法绝对依赖于所审查的研究人群和治疗方法的同质性（或至少可比性）。目前有很多机会对癌症治疗研究进行汇总。最具吸引力的疾病情形是那些有足够数量已完成的随机研究的情况。一些可考虑的疾病包括结肠癌、胃癌、肝癌、卵巢癌和淋巴瘤。
