作者: Trevor A McGrath , Matthew D F McInnes , Nick van Es , Mariska M G Leeflang , Daniël A Korevaar
DOI: 10.1373/CLINCHEM.2017.271544
关键词:
摘要: BACKGROUND: We wished to assess the frequency of overinterpretation in systematic reviews diagnostic accuracy studies. METHODS: MEDLINE was searched through PubMed from December 2015 January 2016. Systematic studies English were included if they reported one or more metaanalyses estimates. built and piloted a list 10 items that represent actual abstract and/or full-text conclusion, 9 potential overinterpretation. Two investigators independently used score each review, with disagreements resolved by consensus. RESULTS: 112 reviews. The majority had positive conclusion regarding clinical usefulness investigated test (n = 83; 74%) 74%). Of reviews, 81 (72%) contained at least 1 form abstract, 77 (69%) full-text. This most often “positive not reflecting summary estimates,” 55 (49%) abstracts 56 (50%) full-texts taking high risk bias applicability concerns into account,” 47 (42%) 26 (23%). these 107 (96%) overinterpretation, frequently “nonrecommended statistical methods for metaanalysis performed” 57; 51%). CONCLUSIONS: Most recent present conclusions contain may lead unjustified optimism about performance erroneous decisions recommendations.