Meta-analysis of grip strength: most common, but superfluous variable in comparative NSAID trials
Gøtzsche, P.C.
Danish Medical Bulletin 36(5): 493-495
1989
ISSN/ISBN: 0907-8916 PMID: 2680316 Document Number: 340038
A meta-analysis was done on grip strength, the most common variable in comparative NSAID trials. The median grip strength was 133 mm Hg (SD 64). The drugs were better than placebo (11.9 mm Hg, confidence interval (CI): 8.9-14.9), but the full dose was not significantly superior to half the dose (3.5 mm Hg, CI: -1.4 to +8.4 mm Hg). The difference between new drugs and control drugs in 70 crossover trials was only 1.4 mm Hg (CI: -0.1 to +2.9). Significant results in favour of the new drug were claimed more often than expected (p less than 0.01), suggesting biased data analysis. Grip strength is a superfluous variable in comparative NSAID trials: even the drug effect over placebo was less than what most rheumatologists considered of relevance.