August 27, 1997

Meta-analysis of Prevention of Nonvertebral Fractures by Alendronate-Reply

Author Affiliations

Merck Research Laboratories Rahway, NJ
Roche Global Development Palo Alto, Calif

JAMA. 1997;278(8):631-632. doi:10.1001/jama.1997.03550080041023

To the Editor.  —Dr Ross has expressed several concerns about the meta-analysis that we believe are not justified. Our article states that data were derived from all studies of the treatment of osteoporosis with oral alendronate that were of at least 2 years' duration and were completed in 1995 when the analysis was performed—no eligible studies were excluded. Alendronate, a proprietary drug, was not marketed at the time the meta-analysis was conducted; therefore we are certain there were no other data. Ross states that there is no evidence that a review protocol was written a priori. As we reported, the decision to conduct a meta-analysis of nonvertebral fracture was post hoc (originally undertaken in response to regulatory agencies); however, the conservative analytical ground rules were defined a priori (before the examination of fracture data). We did not restrict the analysis to patients receiving the optimal dose or to a high-risk subpopulation that may