Completeness of Reporting of Systematic Reviews of Diagnostic Test Accuracy Based on the PRISMA-DTA Reporting Guideline

• Published in: Clinical chemistry (Baltimore, Md.) Md.), 2019-02, Vol.65 (2), p.291-301
• Main Authors: Salameh, Jean-Paul, McInnes, Matthew D F, Moher, David, Thombs, Brett D, McGrath, Trevor A, Frank, Robert, Dehmoobad Sharifabadi, Anahita, Kraaijpoel, Noémie, Levis, Brooke, Bossuyt, Patrick M
• Format: Article
• Language: English
• Subjects: Accuracy; Checklist; Citation analysis; Completeness; Databases, Factual; Diagnostic systems; Diagnostic Techniques and Procedures - standards; Diagnostic tests; Epidemiology; Guidelines; Guidelines as Topic; Humans; Impact factors; Joint surgery; Literature reviews; Medical diagnosis; Meta-analysis; Quality; Reproducibility of Results; Research Design - standards; Reviews; Studies; Systematic review; Systematic Reviews as Topic; Validity
• ISSN: 0009-9147; 1530-8561
• DOI: 10.1373/clinchem.2018.292987

We evaluated the completeness of reporting of diagnostic test accuracy (DTA) systematic reviews using the recently developed Preferred Reporting Items for Systematic Reviews and MetaAnalyses (PRISMA)-DTA guidelines. MEDLINE was searched for DTA systematic reviews published October 2017 to January 2018. The search time span was modulated to reach the desired sample size of 100 systematic reviews. Reporting on a per-item basis using PRISMA-DTA was evaluated. One hundred reviews were included. Mean reported items were 18.6 of 26 (71%; SD = 1.9) for PRISMA-DTA and 5.5 of 11 (50%; SD = 1.2) for PRISMA-DTA for abstracts. Items in the results were frequently reported. Items related to protocol registration, characteristics of included studies, results synthesis, and definitions used in data extraction were infrequently reported. Infrequently reported items from PRISMA-DTA for abstracts included funding information, strengths and limitations, characteristics of included studies, and assessment of applicability. Reporting completeness was higher in higher impact factor journals (18.9 vs 18.1 items; = 0.04), studies that cited PRISMA (18.9 vs 17.7 items; = 0.003), or used supplementary material (19.1 vs 18.0 items; = 0.004). Variability in reporting was associated with author country ( = 0.04) but not journal ( = 0.6), abstract word count limitations ( = 0.9), PRISMA adoption ( = 0.2), structured abstracts ( = 0.2), study design ( = 0.8), subspecialty area ( = 0.09), or index test ( = 0.5). Abstracts with a higher word count were more informative ( = 0.4; < 0.001). No association with word counts was observed for full-text reports ( = -0.03; = 0.06). Recently published reports of DTA systematic reviews are not fully informative when evaluated against the PRISMA-DTA guidelines. These results should guide knowledge translation strategies, including journal level (e.g., PRISMA-DTA adoption, increased abstract word count, and use of supplementary material) and author level (PRISMA-DTA citation awareness) strategies.