A comparison of diagnostic efficacies among different reagent strips and automated cell count in spontaneous bacterial peritonitis

Abstract

Background: Currently, decision to give antibiotics in spontaneous bacterial peritonitis (SBP) suspected patient depends mainly on the result of manual cell count, which requires significant waiting period. Recently, many reports on the efficacies of reagent strips and a few reports of automated cell count are available but there has been no direct comparison study. Aims: This prospective study was to assess the diagnostic efficacies of different reagent strips (Aution, Multistix, Combur) and automated cell count. Methods and Results: A total of 250 paracenteses were performed. There were 40 specimens obtained from patients with clinical suspicion for SBP, the rest were obtained from non SBP suspected patients. Thirty specimens from 250 samples (12%) were diagnosed as SBP by manual cell count. Automated system provided higher value for SBP diagnosis in all parameters (sensitivity, specificity, PPV, NPV, and accuracy; 87.5-99.1%) whereas the strip tests provided lower number in all parameters (80-98.6%). Multistix provided the lowest sensitivity (80%). The false negative rates by Aution, Multistix, Combur tests and automated cell count were 10%, 20%, 10% and 3.3%, respectively. By lowering the cut off for SBP diagnosis with the automated system to 200 cells/mm 3, there was no false negative. Conclusions: Comparing to reagent strips, automated cell count is a better screening tool for SBP diagnosis because it provides higher validity scores and a lower false negative rate. However, the discrepancy of cell count reading may occur, we suggest using a lower cut off for SBP diagnosis by the automated system. © 2010 Journal of Gastroenterology and Hepatology Foundation and Blackwell Publishing Asia Pty Ltd.