Glucosamine and chondroitin for treatment of osteoarthritis - A systematicquality assessment and meta-analysis

Context Glucosamine and chondroitin preparations are widely touted in the l ay press as remedies for osteoarthritis (OA), but uncertainty about their e fficacy exists among the medical community. Objective To evaluate benefit of glucosamine and chondroitin preparations f or OA symptoms using meta-analysis combined with systematic quality assessm ent of clinical trials of these preparations in knee and/or hip OA. Data Sources We searched for human clinical trials in MEDLINE (1966 to June 1999) and the Cochrane Controlled Trials Register using the terms osteoart hritis, osteoarthrosis, degenerative arthritis, glucosamine, chondroitin, a nd glycosaminoglycans. We also manually searched review articles, manuscrip ts, and supplements from rheumatology and OA journals and sought unpublishe d data by contacting content experts, study authors, and manufacturers of g lucosamine or chondroitin. Study Selection Studies were included if they were published or unpublished double-blind, randomized, placebo-controlled trials of 4 or more weeks' du ration that tested glucosamine or chondroitin for knee or hip OA and report ed extractable data on the effect of treatment on symptoms. Fifteen of 37 s tudies were included in the analysis. Data Extraction Reviewers performed data extraction and scored each trial u sing a quality assessment instrument. We computed an effect size from the i ntergroup difference in mean outcome values at trial end, divided by the SD of the outcome value in the placebo group (0.2, small effect; 0.5, moderat e; 0.8, large), and applied a correction factor to reduce bias. We tested f or trial heterogeneity and publication bias and stratified for trial qualit y and size. We pooled effect sizes using a random effects model. Data Synthesis Quality scores ranged from 12.3% to 55.4% of the maximum, wi th a mean (SD) of 35.5% (12%). Only 1 study described adequate allocation c oncealment and 2 reported an intent-to-treat analysis. Most were supported or performed by a manufacturer. Funnel plots showed significant asymmetry ( P less than or equal to.01) compatible with publication bias. Tests for het erogeneity were nonsignificant after removing 1 outlier trial. The aggregat ed effect sizes were 0.44 (95% confidence interval [CI], 0.24-0.64) for glu cosamine and 0.78 (95% CI, 0.60-0.95) for chondroitin, but they were dimini shed when only high-quality or large trials were considered. The effect siz es were relatively consistent for pain and functional outcomes. Conclusions Trials of glucosamine and chondroitin preparations for OA sympt oms demonstrate moderate to large effects, but quality issues and likely pu blication bias suggest that these effects are exaggerated. Nevertheless, so me degree of efficacy appears probable for these preparations.