Vlad Steven C, LaValley Michael P, McAlindon Timothy E, Felson David T
Boston University Medical Center, Boston, Massachusetts 02118, USA.
Arthritis Rheum. 2007 Jul;56(7):2267-77. doi: 10.1002/art.22728.
Investigators in trials of glucosamine report a range of estimates for efficacy, making conclusions difficult. We undertook this study to identify factors that explain heterogeneity in trials of glucosamine.
We searched for reports of trial results in Ovid Medline, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews, and proceedings of scientific conferences. We selected reports of randomized, double-blind, placebo-controlled trials of glucosamine for pain from osteoarthritis of the knee or hip. We extracted data regarding features of design, subjects, and markers of industry involvement, including industry funding, whether a drug was supplied by industry, industry participation, and industry-affiliated authorship. We examined which factors best accounted for differences in the effect sizes of studies grouped by these characteristics, and we examined changes in I(2), a measure of heterogeneity.
Fifteen trials met our inclusion criteria. The summary effect size was 0.35 (95% confidence interval 0.14, 0.56). I(2) was 0.80. Except for allocation concealment, no feature of study design explained this substantial heterogeneity. Summary effect sizes ranged from 0.05 to 0.16 in trials without industry involvement, but the range was 0.47-0.55 in trials with industry involvement. The effect size was 0.06 for trials using glucosamine hydrochloride and 0.44 for trials using glucosamine sulfate. Trials using Rottapharm products had an effect size of 0.55, compared with 0.11 for the rest.
Heterogeneity among trials of glucosamine is larger than would be expected by chance. Glucosamine hydrochloride is not effective. Among trials with industry involvement, effect sizes were consistently higher. Potential explanations include different glucosamine preparations, inadequate allocation concealment, and industry bias.
氨基葡萄糖试验的研究者报告了一系列疗效评估结果，这使得得出结论变得困难。我们开展这项研究以确定能解释氨基葡萄糖试验异质性的因素。
我们在Ovid Medline、Cochrane对照试验中心注册库、Cochrane系统评价数据库以及科学会议论文集中搜索试验结果报告。我们选择了关于氨基葡萄糖用于膝或髋骨关节炎疼痛的随机、双盲、安慰剂对照试验的报告。我们提取了有关设计特征、受试者以及行业参与标志的数据，包括行业资助、药物是否由行业提供、行业参与情况以及行业附属作者身份。我们研究了哪些因素最能解释按这些特征分组的研究效应大小的差异，并研究了异质性度量指标I²的变化。
15项试验符合我们的纳入标准。汇总效应大小为0.35（95%置信区间0.14, 0.56）。I²为0.80。除分配隐藏外，研究设计的任何特征都无法解释这种显著的异质性。在无行业参与的试验中，汇总效应大小范围为0.05至0.16，但在有行业参与的试验中，范围为0.47 - 0.55。使用盐酸氨基葡萄糖的试验效应大小为0.06，使用硫酸氨基葡萄糖的试验效应大小为0.44。使用罗塔法姆产品的试验效应大小为0.55，其余试验为0.11。
氨基葡萄糖试验之间的异质性大于偶然预期。盐酸氨基葡萄糖无效。在有行业参与的试验中，效应大小始终较高。可能的解释包括不同的氨基葡萄糖制剂、分配隐藏不足以及行业偏差。
