Hi All,
I came across, a few years ago (hence I've forgotten), a paper that reported stability in meta-analysis was only achieved after 3-4 trials. By stability it was something along the lines of after 3-4 trials the estimated effect size stopped flipping between significance/non-significance (something like that).
They also referenced a paper that suggested something like 1,000 patients in the meta-analysis before we got to stability.
Ring any bells? Apologies for my vagueness!
Best wishes
Jon
|