2 | Page
Contents
10.1 Introduction ....................................................................................................................................................... 4
10.1.1 Aims of meta-analysis for DTA reviews....................................................................................................... 4
10.1.2 When not to use a meta-analysis in a review ................................................................................................ 5
10.1.3 How does meta-analysis of diagnostic test accuracy differ from meta-analysis of interventions? ................ 5
10.1.4 Questions which can be addressed in DTA analyses .................................................................................... 6
10.1.4.1 What is the accuracy of a test? ............................................................................................................ 6
10.1.4.2 How does the accuracy vary with clinical and methodological characteristics? ................................. 6
10.1.4.3 How does the accuracy of two or more tests compare? ....................................................................... 6
10.1.5 Planning the analysis .................................................................................................................................... 7
10.1.6 Writing the analysis section of the protocol .................................................................................................. 7
10.2 Key concepts ..................................................................................................................................................... 8
10.2.1 Disease status ................................................................................................................................................ 8
10.2.2 Types of test data .......................................................................................................................................... 9
10.2.3 Analysis of a primary test accuracy study ..................................................................................................... 9
10.2.3.1 Sensitivity and Specificity ................................................................................................................. 10
10.2.3.2 Predictive values ............................................................................................................................... 10
10.2.3.3 Likelihood ratios ............................................................................................................................... 10
10.2.3.4 Diagnostic odds ratios ....................................................................................................................... 11
10.2.4 Positivity thresholds .................................................................................................................................... 11
10.2.5 ROC curves ................................................................................................................................................. 13
10.2.6 Relationships between ROC curves, diagnostic odds ratios and Q* ........................................................... 14
10.3 Graphical and tabular presentation .................................................................................................................. 15
10.3.1 Summary ROC plots ................................................................................................................................... 15
10.3.2 Linked ROC plots ....................................................................................................................................... 16
10.3.3 Coupled forest plots .................................................................................................................................... 16
10.3.4 Example 1: Anti-CCP for the diagnosis of rheumatoid arthritis - Descriptive Plots. .................................. 16
10.3.5 Tables of results .......................................................................................................................................... 18
10.4 Meta-analytical summaries .............................................................................................................................. 18
10.4.1 Should I estimate a SROC curve or a summary point? ............................................................................... 18
10.4.2 Meta-analytical methods not routinely used in Cochrane Reviews ............................................................. 19
10.4.3 Heterogeneity .............................................................................................................................................. 20
10.5 Model fitting .................................................................................................................................................... 20
10.5.1 Moses-Littenberg SROC curves (RevMan) ................................................................................................ 20
10.5.1.1 Properties of the curve ...................................................................................................................... 21
10.5.1.2 Choice of weights .............................................................................................................................. 22
10.5.2 Hierarchical models .................................................................................................................................... 22
10.5.2.1 Bivariate model ................................................................................................................................. 24
10.5.2.2 Example 1 continued: Anti-CCP for the diagnosis of rheumatoid arthritis. ..................................... 25
10.5.2.3 The Rutter and Gatsonis HSROC model ........................................................................................... 26
10.5.2.4 Example 2: Rheumatoid Factor as a marker for Rheumatoid Arthritis. ........................................... 28
10.5.3 Investigating heterogeneity ......................................................................................................................... 29
10.5.3.1 Heterogeneity and Regression Analysis using the Bivariate model .................................................. 29
10.5.3.2 Example 1 (cont).: Investigation of heterogeneity in diagnostic performance of anti-CCP ............. 31
10.5.3.3 Heterogeneity and Regression Analysis using the Rutter and Gatsonis HSROC model ................... 32
10.5.3.4 Criteria for model selection ............................................................................................................... 34
10.5.3.5 Example 2 (cont.): Investigating heterogeneity in diagnostic accuracy of Rheumatoid Factor ........ 35
10.5.4 Comparing Index Tests ............................................................................................................................... 36
10.5.4.1 Test comparisons based on all available studies ................................................................................ 36
10.5.4.2 Test comparisons using the Bivariate model ..................................................................................... 36
10.5.4.3 Example 3: CT versus MRI for the diagnosis of coronary artery disease .......................................... 38
10.5.4.4 Test comparisons using the Rutter and Gatsonis HSROC model ...................................................... 39
10.5.4.5 Test comparison based on studies that directly compare tests ........................................................... 40
10.5.4.6 Example 3 (cont.): CT versus MRI for the diagnosis of coronary artery disease .............................. 41
10.5.5 Computer software ...................................................................................................................................... 42
10.5.6 Approaches to analysis with small numbers of studies ............................................................................... 43
10.6 Special topics ................................................................................................................................................... 44
10.6.1 Sensitivity analysis ..................................................................................................................................... 44
10.6.2 Investigating and handling verification bias. .............................................................................................. 46
10.6.3 Investigating and handling publication bias ................................................................................................ 46
10.6.4 Developments in meta-analysis for DTA reviews ...................................................................................... 47