Table 9Summary of selected previously published overviews of systematic review of test accuracy

CharacteristicsIrwig, 199432Whiting, 200533Dinnes, 200534Mallet, 200631Moher, 200735Willis, 201111Current Project
Number of included SRs11114189 (133 used statistical synthesis methods)89 (25 assessed in detail)23 (diagnostic/prognostic SRs among 300 SRs identified)236760
Selection criteriaAll inclusive; meta-analysis of test accuracy as primary focus.All inclusiveAll inclusiveCancer diagnosis; included SRs regardless of the use of quantitative synthesis methods. Screening tests + tests for risk factors were excluded; computer decision tools were also excluded.All inclusiveAll inclusive; reviews had to have searched ≥2 databases, stated search terms and inclusion criteria, and used a statistical method to summarize test accuracy.All inclusive; reviews had to have used a statistical method to summarize test accuracy. Excluded HTAs, Cochrane reviews and AHRQ EPC reports.
Years coveredJan 1990 – Dec 19911995 – 2001Up to 20021990 – 2003November 2004Up to 20081966 – 2009
Databases searchedMEDLINE, experts, bibliographies of retrieved papersDAREDARE (update of the search used in Whiting et al.33)MEDLINE, Embase, MEDION, Cancerlit, HTA, DARE, Cochrane Database of Systematic ReviewsMEDLINEMEDLINE, Embase, CINAHL, Cochrane Library, PsychInfo, Global health, HMIC, AMEDMEDLINE, bibliographies of papers and relevant reviews
Items extractedLiterature review methods; data extraction and presentation; statistical analysis methods.Quality assessment methods.Systematic review methods; statistical analysis and reporting; trends over time in statistical method use.Objectives and setting of the SRs; participant characteristics. In the 25 studies assessed in detail: quality assessment methods; whether meta-analysis was performed; reporting of results; availability of data for reanalysis.Basic reporting characteristics of reviews (bibliometric features, outcomes considered, whether any quantitative method was used).Statistical and quality assessment methods; settings of test use; trends over time in statistical method use.Literature search and study selection methods; quality assessment methods; statistical analyses and reporting methods; trends over time in multiple aspects of the review process, statistical analysis and reporting; availability of data for re-analysis.
Main findings2 of 11 studies reported the complete search strategy; all studies analyzed sensitivity and specificity, 2 studies used the sROC method and 2 did not provide a summary estimate; 6 of 11 studies discussed variability in reference standards; 7 studies reported comparisons between 2 or more index tests.49% of SRs had not conducted quality assessment; in most cases information on quality was incorporated in narrative synthesis; 13% of reviews used quality as an inclusion criterion.70% of SRs used quantitative methods; 52% used MEDLINE as the only source; 69% performed quality assessment; median number of studies=18; 68% of SRs do not report tests for heterogeneity (58% of those using statistical analyses); naïve pooling has decreased over time.75% of SRs stated inclusion criteria, 40% reported details of study design, 17% reported on the clinical setting, 17% reported on disease severity, 49% reported on tumor stage. Of the 25 reviews assessed in detail, 56% reported sensitivity, specificity, and sample sizes for individual studies. Of the 89 reviews, 61% attempted to formally synthesize results of the studies and 32% reported formal assessments of study quality.No SRs were updates of previous reviews; harms were considered in 54% and costs in 35% of the reviews were this information was considered relevant; median number of included studies=39; quantitative synthesis was performed in 48% of SRs.27% of SRs used advanced statistical methods (BREM or hsROC); between 2006 and 2008 QUADAS was used in 40% of the studies; imaging tests are the most commonly assessed test category; 80% of tests are normally used in specialist settings.As detailed in this report.

Studies are listed chronologically, based on the dates covered by their searches.

AHRQ = Agency for Healthcare Research and Quality; AMED = Allied and Complementary Medicine database; BREM = bivariate random effects meta-analysis; CINAHL = Cumulative Index to Nursing and Allied Health Literature; DARE = Database of Reviews of Effects; EPC = Evidence-based Practice Center; HMIC = Health Management Information Consortium database; hsROC = hierarchical summary receiver operating characteristic meta-analysis method; HTAs = health technology assessments; QUADAS = Quality Assessment of Diagnostic Accuracy Studies; SR = systematic review; sROC = summary receiver operating characteristic method

From: Discussion

Cover of Comprehensive Overview of Methods and Reporting of Meta-Analyses of Test Accuracy
Comprehensive Overview of Methods and Reporting of Meta-Analyses of Test Accuracy [Internet].
Dahabreh IJ, Chung M, Kitsios GD, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.