Special comments--problems in systematic review for diagnostic accuracy of high-sensitivity C-reactive protein assay in neonatal infection

摘要

We conducted a systematic review for the diagnostic accuracy of high-sensitivity C-reactive protein assay for neonatal infection. In total, 558 relevant published reports were found in a search of MEDLINE and Igaku-Chuo-Shi. With our inclusion and exclusion criteria, we finally selected 30 primary studies for meta-analysis. We critically appraised the quality of selected clinical studies and then extracted sensitivity and specificity data from each article for meta-analysis. There was only one article that fulfilled our three major standards for quality assessment(rank A; high-quality), while 19 studies which fulfilled only one criterion were assigned to rank C(low-quality). Most studies neither took careful consideration of the blindness against the results of the index test nor described this methodological standard. Lack of qualified primary studies may have had substantial influence on the summary estimates. Insufficient number of primary studies for meta-analysis and their unsatisfactory quality would be improved by the addition of supplemental results retrieved from raw data, which the authors of clinical studies did not present in their articles.