Leeflang Mariska M G, Reitsma Johannes B
1Department Clinical Epidemiology, Biostatistics and Bioinformatics, Amsterdam Public Health, Amsterdam UMC, University of Amsterdam, Amsterdam, The Netherlands.
2Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands.
Diagn Progn Res. 2018 Sep 10;2:17. doi: 10.1186/s41512-018-0039-0. eCollection 2018.
While most relevant clinical questions are comparative, most diagnostic test accuracy studies focus on the accuracy of only one test. If we combine these single-test evaluations in a systematic review that aims to compare the accuracy of two or more tests to indicate the most accurate one, the resulting comparative accuracy estimates may be biased.
Systematic reviews comparing the accuracy of two tests should only include studies that evaluate both tests in the same patients and against the same reference standard. However, these studies are not always available. And even if available, they may still be biased. For example because they included a specific patient group that would not have been tested with two or more tests in actual practice. Combining comparative and non-comparative studies in a comparative accuracy meta-analysis requires novel statistical approaches.
In order to improve decision-making about the use of test in practice, better designed and reported primary diagnostic studies are needed. Meta-analytic and network-type approaches available for therapeutic questions need to be extended to comparative diagnostic accuracy questions.
虽然大多数相关临床问题是比较性的，但大多数诊断试验准确性研究仅关注一项试验的准确性。如果我们在一项旨在比较两项或更多项试验的准确性以指出最准确试验的系统评价中合并这些单项试验评估，那么由此得出的比较准确性估计可能会有偏差。
比较两项试验准确性的系统评价应仅纳入在同一患者中针对同一参考标准评估这两项试验的研究。然而，这些研究并非总是可得。即便可得，它们仍可能存在偏差。例如，因为它们纳入了在实际临床中不会接受两项或更多项试验检测的特定患者群体。在比较准确性荟萃分析中合并比较性和非比较性研究需要新颖的统计方法。
为了改善实践中关于试验使用的决策，需要设计和报告更优的原发性诊断研究。可用于治疗性问题的荟萃分析和网络型方法需要扩展至比较诊断准确性问题。
