6% vs 80.4%,

grade cut off ≥ 2; 45.7% vs 63%).21 Likewise, Kim et al. showed a statistically significant difference in sensitivity between Multistix and Uriscan strips (50% vs 67%).20 In the present series, we compared all validity scores among the three reagent strips, at the lowest cut off ≥ 1, we confirmed that Multistix strip gave the lowest sensitivity (80% vs 90%). In our series, the prevalence of SBP was not lower or higher (12%) than standard series.1–3 Clinical suspicion level plays important role on the prevalence and PPV of SBP positive cases in our study. The PPV by strip test was excellent in symptomatic patients (> 95.5–100%) whereas in BVD-523 asymptomatic patients with a low suspicion for SBP, as expected, the PPV was much poorer (20–22%) (data not shown). It has been speculated in the past that prior antibiotic use may result in false negative and positive results.22 However, none of our patients with false negative result reported history of recent or current antibiotic uses. Our previous study showed that the smaller number of PMN cells in the specimens (close to 250 PMN/mm3) may contribute to a false negative result.13 However, in the present series, we observed that all false negative specimens from the strip test always had PMN higher than 1000 PMN/mm3

by manual count. In summary, our false negative rates from Aution Mutistix, and Combur strips were quite significant (10%, 20% and 10%, respectively). Although there were some discrepancies in the range of PMN number between automated reading and manual reading, the specimen number 11 was the only one that this discrepancy resulted in Palbociclib concentration the different interpretation and caused false positive. If we lowered our cut off value for SBP diagnosis by the automated system to 200 cells/mm3, Bcl-w we would not have any false negative case. It has been reported that the agreement between manual and automated cerebrospinal fluid cell counts was suboptimal if the PMN count was less than 200

cells/mm3.23 In addition, there was a certain lower limit for PMN detection by each automated system. For instance, based on coefficient of variation, the Iris iQ200 automated microscopy analyzer Body Fluids Module (Iris Diagnostics, Chatsworth, CA), has a limitation at 35 PMN cells/mm3.24 Therefore, further study on the threshold for clinical diagnosis of SBP form ascitic fluid analysis by each automated cell count is required. At the critical threshold of manual PMN count to diagnose SBP (250 > PMN < 1000) cells/mm3 (n= 12), there were another five specimens (specimen number 2, 17, 20, 22, 23) that had a significant discrepancy of cell count by the two methods but this was not affecting SBP diagnosis interpretation by the automated cell count (Table 2). A better correlation between the two methods was demonstrated by a study from Rome.25 Riggio et al. reported the limits of agreement of the two methods were +124 cells/mm3[95% confidence interval (CI): +145 to +103] and −108 cells/mm3 (95% CI: −87 to −129).