Meta-analysis of diagnostic procedures: A brief overview

Abstract

Surveys of the medical literature frequently reveal that studies of diagnostic imaging tests focus primarily on technology development (1).This focus on the measurement of diagnostic accuracy leads to multiple reports of test sensitivity and specificity, often with conflicting results (2).As radiologists are asked to quantify the clinical benefits of the services they provide, research contributing to an understanding of the effect of imaging on patient outcomes is needed (3).This is particularly true for new, emerging, or rapidly evolving technologies.Consequently, many investigators are turning to data synthesis methods, such as decision modeling, and cost-effectiveness analysis (4), to estimate the effects of diagnostic imaging on patient outcome.These data synthesis methods typically use research results from the medical literature in conjunction with a model of test accuracy and therapeutic effectiveness to ascertain the groups of patients for which a given diagnostic technology is advantageous.Not surprisingly, decision models depend critically on credible estimates of the sensitivity and specificity of the imaging technique in question, as well as a reasonable assessment of the variability of those estimates.Meta-analysis of reports concerning diagnostic procedures provides needed information by capitalizing on the large numbers of studies aimed at assessing the accuracy of imaging tests.The technique is especially useful for reducing variability and uncertainty among published reports of diagnostic test accuracy.The purpose of this brief report is to provide an introduction to meta-analytic techniques and to supply pointers to the literature for those interested in applying meta-analysis to diagnostic test evaluation.First,

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 14 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.