To the Editors:

Dr. Marshall clarifies 3 points related to their randomized controlled trial (1) that was included in our meta-analysis. First, Dr. Marshall stresses that Euflexxa (Ferring Pharmaceuticals, Inc.) is a non–cross-linked HA with a molecular weight of 2,400–3,600 kD. We did not attempt to characterize all of the diverse viscosupplementation preparations in detail. Rather, we distinguished between the cross-linked hylan and non–cross-linked preparations and pointed out that we classified Euflexxa correctly as a non–cross-linked preparation with an average molecular weight of 3,000 kD.

Second, Dr. Marshall states that they used an ITT approach for the analysis of their trial (1). Definitions of ITT analyses vary considerably (2). Kirchner and Marshall defined the ITT population as randomized patients who received at least 1 injection (1). As stated in the Materials and Methods section in our article, we used a more stringent definition and required that all randomized patients be included in the analysis regardless of the treatment received (2).

Meta-analyses depend on the information provided on included trials, but reports frequently omit methodologic details (3). We thank Dr. Marshall for the additional information on blinding. Since blinding of the physicians evaluating clinical outcomes was not among the prespecified trial characteristics used for stratification of analyses, the results and conclusions of our meta-analysis remain unchanged with this new information becoming available.

Stephan Reichenbach MD*, Sven Trelle MD†, Peter Jüni MD†, * Boston University, Boston, MA, † University of Bern and Bern University Hospital, Bern, Switzerland.