[A systematic review for the diagnostic accuracy o serum C-reactive protein measurement in neonatal infants with infection].

Abstract

We presented an example of systematic review regarding diagnostic test accuracy. We compared the diagnostic accuracy of serum C-reactive protein (CRP) concentrations measured by the highly sensitive methods for neonatal infection with that by the conventional-sensitivity methods in the literatures available. Our results with the summary ROC curve indicated that the high sensitivity CRP is likely to provide a better diagnostic accuracy than the conventional methods, even though there was a heterogeneity in the specificity among studies. In the reviewing process of the literatures, we found several critical problems, such as the publishing bias and insufficient quality of primary articles. These problems may have had a substantial influence on the summary estimates. Systematic reviewing of the related articles uncovered the defects in the primary studies, requiring more prudent study designs for future clinical trials.