Medical tests are essential for patient care. Evidence-based assessment of the relative accuracy of competing diagnostic tests informs clinical and policy decision making. This thesis addresses questions centred on assessing the reliability and transparency of evidence from systematic reviews and meta-analyses of comparative test accuracy, including validity of meta-analytic methods. Case studies were used to highlight key methodological issues, and provided rationale and context for the thesis. Published systematic reviews of multiple tests were identified and used to provide a descriptive survey of recent practice. Availability of comparative accuracy studies and differences between meta-analyses of direct (head-to-head) and indirect (between-study) comparisons were assessed. Comparative meta-analysis methods were reviewed and those deemed statistically robust were empirically evaluated. Using simulation, performance of hierarchical methods for meta-analysis of a single test was investigated in challenging scenarios (e.g. few studies or sparse data) and implications for test comparisons were considered. Poor statistical methods and incomplete reporting threatens the reliability of comparative reviews. Differences exist between direct and indirect comparisons but direct comparisons were seldom feasible because comparative studies were unavailable. Furthermore, inappropriate use of meta-analytic methods generated misleading results and conclusions. Therefore, recommendations for use of valid methods and a reporting checklist were developed.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:687529 |
| Date | January 2016 |
| Creators | Takwoingi, Yemisi |
| Publisher | University of Birmingham |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://etheses.bham.ac.uk//id/eprint/6759/ |
Page generated in 0.0192 seconds