组合聚集数据与个体数据的多水平Meta分析模型

目的 探讨组合聚集数据和个体数据的多水平Meta分析模型及其在医学领域的应用。方法 以组合文献检索的马来酸罗格列酮临床试验数据和国产药罗格列酮钠临床试验数据获得“罗格列酮类药物相对未治疗或其它药物治疗糖尿病26周糖化血红蛋白（HbA1c）减少量”为例构造组合聚集数据和个体数据的多水平Meta分析模型。结果 基线HbA1c含量为9％，每天用药剂量为4mg时用罗格列酮类治疗26周后相对未治疗或其它药物治疗使HbA1c含量平均下降0．464％，95％可信区间为（0．168％，0．760％）。结论 当可以获得个别研究的个体数据时，组合聚集数据和个体数据的多水平Meta分析模型能充分利用有限的数据资源，可能使Meta分析的结果更准确。

The Multi-level Meta Analysis Model Combining Individual Level Data with Aggregative Level Data

Objective To explore the multi-level Meta analysis model combining individual level data with aggregative level data and its application in medicine.Methods The difference in respect to "decreased hemoglobin A(1c)" between treatment with roglizatone and treatment without roglizatone was obtained by combining individual level data from clinical trial of roglizatone natrium with aggregative level data from literatures.This endeavor was regarded as an example to construct the multi-level Meta analysis model combining individual level data with aggregative level data.Results The baseline hemoglobin A(1c) was 9% and the dose of drug was 4 mg per day.The average difference in "decreased hemoglobin A(1c)" between treatment with and without roglizatone was 0.464%,with a 95%CI from 0.168% to 0.760%.Conclusion In case that individual data are available,the multi-level Meta analysis model combining individual level data with aggregative level data can utilize the data resources adequately and the results obtained should be more accurate.