CAN STUDY SELECTION, VARIABLE MANAGEMENT AND TIME PERIOD INFLUENCE OBSERVED EFFECT SIZES IN SYSTEMATIC REVIEWS OF HYALURONAN/HYLAN PRODUCTS?

Authors: N Bellamy, J Campbell, T Gee1

References: Osteoarthritis and Cartilage, Volume 13, Supplement A, page S71. 2005

Abstract
Aim of Study: To examine potential explanations for discordance in measures of clinical benefit in four recently reported systematic reviews of hyaluronan/hylan (HA) products: Lo et al. Dec 03 (JAMA), Wang et al. Mar 04 (JBJS), Arrich et al. Apr 05 (CMAJ), and Bellamy et al. Apr 05 (The Cochrane Library). Methods: Secondary analyses of a pooled HA class dataset containing 21 different HA studies were performed to investigate the potential influence of study selection, variable selection and time point management on the observed clinical benefit. Results: In the pooled analyses of HA class efficacy, weighted mean differences were different for different variables at the same time point and the same variable at different time points. Combinatorial meta-analyses indicated that inclusion of different studies influenced the observed clinical benefit. Conclusions: These observations suggest that estimation of the therapeutic response to HA products is influenced by methodology, and is dependent on factors including study inclusion criteria, variable selection and time point management. Heterogeneity in the effect of the HA class of products on different outcomes, at different time points, argue against designating only a single variable and time point to characterise each component study in meta-analyses. Discordance in HA meta-analyses may, therefore, be, in part, attributable to differences in study inclusion and variable and time point management. Meta-analyses which attempt to identify and minimize these influences may provide more meaningful assessments of clinical benefit.