Modern Biostatistical Methods for Evidence-Based Global Health Research
Modern Biostatistical Methods for Evidence-Based Global Health Research
复制标题
循证全球健康研究的现代生物统计方法
DOI：
10.1007/978-3-031-11012-2_8
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Musekiwa A
中科院分区：
文献类型：
--
作者：
Musekiwa A
单变量效应量的元分析方法是众所周知的和发展起来的。然而，在医学研究中越来越多地测量和报告多种结果，这可能导致估计多种效应量。估计的效应量可能是相关的，因为它们是从相同的研究中测量的。此外，结果通常是纵向测量的，导致随着时间的推移反复估计多个效应量。因此，由于在同一研究中同一效应随时间的重复估计，估计的效应量可以在研究中横断面和序列上相关。这导致纵向多重效应大小。本章提出了统计荟萃分析的方法，将多个纵向研究的汇总数据与多个效应大小相结合。对HIV研究进行纵向荟萃分析，在开始分配治疗后的第4、8、12、16、20、24、32、40和48周，评估一些抗逆转录病毒药物在改善病毒载量抑制和增加CD4计数方面的作用。
Meta-analysis methods for univariate effect sizes are well-known and developed. However, multiple outcomes are increasingly being measured and reported in medical research studies, which may lead to multiple effect sizes being estimated. The estimated effect sizes could be correlated because they are measured from the same studies. Additionally, the outcomes are often measured longitudinally, resulting in multiple effect sizes estimated repeatedly over time. Thus, the estimated effect sizes could be correlated within studies both cross-sectionally and serially due to the repeated estimation of the same effect over time in the same study. This results into longitudinal multiple effect sizes. This chapter proposes methods for statistical meta-analysis combining summary data from more than one longitudinal study with multiple effect sizes. The proposed methods are illustrated by an analysis of an example involving longitudinal meta-analysis of HIV studies assessing the effect of some antiretroviral drugs in improving viral load suppression and increasing CD4 count at weeks 4, 8, 12, 16, 20, 24, 32, 40, and 48 after start of treatment assignment.