Test meta-analyses for stabilityBMJ 1998; 317 doi: https://doi.org/10.1136/bmj.317.7152.206b (Published 18 July 1998) Cite this as: BMJ 1998;317:206
- C David Naylo, Professor of medicine,
- George Davey Smith, Professor of clinical epidemiology.
- Sunnybrook Health Science Centre, University of Toronto, Ontario, Canada, M4N 3M5
- Department of Social Medicine, University of Bristol, Bristol BS8 2PR
EDITOR–An editorial on meta-analysis that one of us wrote1 noted in passing two difficulties with a paper by Le Lorier et al that showed apparent discrepancies between results of meta-analyses and of individual trials.2 Le Lorier and Gregoire have written to counter those criticisms,3 but we think that they have missed the salient methodological issues.
Le Lorier et al defined a negative trial as “one in which the treatment resulted in an equal or worse …