Commentary on meta-analysis

Abstract

I do not know whether this will annoy S. Shapiro but I believe that contributions such as his will actually play a key role in ensuring that meta-analysis survives in epidemiology. I think his critique represents a necessary corrective action that we see in the development of any new method. Perhaps I should spare you my first thousand words and summarise his talk with a picture. I believe what he was essentially saying was that meta-analysis of observational studies gives you about as much guidance as the map does to the poor soul in this picture — a little man on his knees in the desert looking up forlornly at a map with no landmarks, with a dot in the centre labelled, “You are here”. I believe this is what he was saying; that meta-analysis seems to tell us precisely where we are, but we have absolutely no idea whether we are right or wrong, or where to go from there.