Charles E. Schmidt College of Medicine, Florida Atlantic University, Boca Raton, FL.
Charles E. Schmidt College of Medicine, Florida Atlantic University, Boca Raton, FL.
Ann Epidemiol. 2021 Jun;58:38-41. doi: 10.1016/j.annepidem.2021.02.013. Epub 2021 Feb 26.
Meta-analyses of observational studies reduce the role of chance but also introduce bias because the individual component studies are not randomized. Further, it is plausible that the bias may be different in case-control and cohort studies. We explored these issues in meta-analyses of observational studies of Opioid Use Disorder (OUD).
From a systematic literature review of 152 published meta-analyses, 11 fulfilled the initial inclusion criteria of observational studies of OUD. Of these, 9 were meta-analyses of case-control studies and 2 were meta-analyses of cohort studies but only 4 (3 case-control and 1 cohort) targeted more than one specific chromosomal location.
The meta-analyses of the 3 case-control studies, which included 13 individual studies, identified 12 different single nucleotide polymorphisms on 6 different genes on 5 different chromosomes. None was the same as the gene on Chromosome 15 identified from the meta-analysis of the cohort studies.
These data, from genetic studies, suggest biases are different in meta-analyses of case-control and cohort studies, perhaps due to greater selection bias in case-control studies. These observations have potential importance in the application of meta-analyses to many common and serious diseases, as well as genomics and precision medicine, including OUD.
观察性研究的荟萃分析减少了偶然性的作用，但也引入了偏倚，因为单个的组成研究不是随机的。此外，在病例对照研究和队列研究中，偏倚可能不同，这是有道理的。我们在观察性研究的荟萃分析中探讨了这些问题，这些研究涉及阿片类药物使用障碍（OUD）。
从对 152 篇已发表荟萃分析的系统文献回顾中，有 11 项符合 OUD 观察性研究的初始纳入标准。其中，9 项是病例对照研究的荟萃分析，2 项是队列研究的荟萃分析，但只有 4 项（3 项病例对照和 1 项队列）针对一个以上的特定染色体位置。
这 3 项病例对照研究的荟萃分析包括 13 项独立研究，确定了 5 条不同染色体上 6 个不同基因的 12 个不同单核苷酸多态性。没有一个与来自队列研究的荟萃分析中确定的 15 号染色体上的基因相同。
这些来自遗传学研究的数据表明，病例对照研究和队列研究的荟萃分析中的偏倚是不同的，这可能是由于病例对照研究中的选择偏倚更大。这些观察结果在荟萃分析应用于许多常见和严重疾病以及基因组学和精准医学方面具有潜在的重要性，包括 OUD。
