Multivariate Meta-anaylsis of Diagnostic Tests

描述（由申请人提供）：比较有效性研究（CER）从根本上依赖于对临床结果的准确评估。评估工具的数量不断增加，以及成本的迅速上升，越来越需要通过荟萃分析对临床实践中的诊断测试进行科学严格的比较。与随机临床试验的传统荟萃分析相比，诊断测试的荟萃分析提出了许多额外的统计挑战。特别是，诊断准确性无法通过单一测量来充分概括；通常使用配对测量，例如最常用的敏感性 (Se) 和特异性 (Sp)，或者阳性和阴性预测值；并且配对的测量中的任何一个通常都是相关的。此外，这些诊断准确性测量可能取决于疾病患病率，并且在许多研究中，参考标准也受到测量误差的影响。 针对PAR-10-168，该提案的总体目标是开发尖端的诊断测试多元元分析方法，并将其集成到公开可用、易于使用的R软件包元中，以增强一致性、适用性和通用性。在本提案中，我们将重点关注：（1）开发多变量方法和软件，以有效检测和调整诊断测试荟萃分析中的发表偏倚和其他样本量影响； (2)开发网络荟萃分析框架和软件以同时比较多个诊断测试。我们建议通过实际数据应用和模拟对这些方法的优点和缺点进行实证评估。所提出的统计方法将广泛适用于比较诊断测试的荟萃分析。它将通过促进诊断来改善公共卫生 各种癌症、心血管、传染病等疾病。这两个目标的完成将直接有利于 CER 计划，通过使用 WinBUGS 和 R 统计语言在用户友好的软件中实现最先进的方法，这些语言将免费向公众开放。

DESCRIPTION (provided by applicant): Comparative effectiveness research (CER) relies fundamentally on accurate assessment of clinical outcomes. The growing number of assessment instruments, as well as the rapid escalation in the cost has generated the increasing need for scientifically rigorous comparisons of the diagnostic tests in clinical practic via meta-analysis. Meta-analysis of diagnostic tests presents many additional statistical challenges compared to traditional meta-analysis of randomized clinical trials. In particular, diagnostic accuracy cannot be adequately summarized by a single measure; paired measures are typically used, for example, most popularly sensitivity (Se) and specificity (Sp), or alternatively positive and negative predictive values; and either of the paired measures is typically correlated. Furthermore, those diagnostic accuracy measures may depend on disease prevalence and in many studies, the reference standard is also subject to measurement error. In response to PAR-10-168, the overall goal of this proposal is to develop cutting-edge multivariate meta- analysis methods of diagnostic tests, and to integrate them into publicly available, easy-to-use R software package meta to enhance the consistency, applicability, and generalizability. In this proposal, we will focus on: (1) developing multivariate methods and software to efficiently detect and adjust for publication bias and other sample size effects in meta-analysis of diagnostic tests; and (2) developing network meta-analysis framework and software to simultaneously compare multiple diagnostic tests. We propose to perform empirical assessment of the strengths and weaknesses of these methods through real data applications and simulations. The proposed statistical methodology will be broadly applicable to the meta-analysis comparing diagnostic tests. It will improve public health by facilitating the diagnosis of various cancers, cardiovascular, infectious and other diseases. Completion of these two aims will directly benefit the CER program by providing state-of-the art methods implemented in user-friendly software using the WinBUGS and R statistical languages that will be made freely available to the public.