My recent research efforts have concerned metaresearch on orthopaedic RCTs. We have argued many aspects in those studies and in their quality. One major aspect is the sample size which still remains very small. In our recent paper we investigated orthopaedic RCTs published in 2016 and 2017 and we reported:
The median numbers of patients in the study groups were 37 (range, 6 to 675; interquartile range [IQR], 26 to 56) and 37 (range, 6 to 850; IQR, 25 to 58).
So, overall median sample was 74 patients. That is not very large. All this relates to short discussion I had in Twitter:
EBM Analytics makes an excellent point there. It´s not only small sample size. It´s the fact that we don´t make the most of it. Covariate adjustment is very rare in orthopaedics. In addition to larger trials, we need also to increase awareness about more appropriate methods when analyzing our (small) trials.