Letters Section Editor: Robert M. Golub, MD, Senior Editor.
In Reply: We appreciate Dr Cummings' suggestion to use the matched risk ratio to present the overall treatment effect from this study design. As there were no bilateral fractures in our study, the calculations are correct as stated in his letter. We have not found an adjustment to the CI for the clustered design, but any adjustment would be minimal since the intraclass correlation is typically very small in studies of this type.
We fully support the use of such estimates of effect sizes as suggested by Cummings. We hope others will use this matched risk ratio in future meta-analyses of hip protector efficacy, and we will include this calculation of the overall treatment effect in any future publications that emerge from our trials using this design.
Kiel DP, Birge SJ, Magaziner J, Zimmerman S, Brown KM, Barton BA. Hip Protectors and Hip Fracture—Reply. JAMA. 2007;298(18):2139. doi:10.1001/jama.298.18.2139-b