We read with interest the study of Nousbaum and colleagues, which reported a low sensitivity of the Multistix 8 reagent strip in the diagnosis of spontaneous bacterial peritonitis (SBP).1 It is noteworthy that overall prevalence of SBP in this study was only 5.5%, significantly lower than in all previous studies focussing on diagnostic accuracy of different reagent strips, where prevalence of SBP ranged between 7% and 23%. Can this simple statistical bias explain the differences in diagnostic sensitivity of different reagent strips?
Recently, we terminated a similar study comparing Multistix 10 SG and Combur reagent strips for their diagnostic value of SBP in the inpatient setting at 2 tertiary care centers. We found 16 cases of SBP in 194 paracenteses summing up to a prevalence of SBP of 8.2% in inpatients and comparing well with the subset of inpatients in the study of Nousbaum et al. (9%). Sensitivity of Multistix 10 SG was even worse in our setting (31%) and only slightly better for Combur reagent strips (44%). Although at a different level, this finding is in accordance with the study of Campillo et al.2 who reported better performance of Combur reagent strips.
When comparing different parameters between false-negative and true-positive results in our patients with SBP, we found significantly higher mean ascitic protein content in patients with false-negative results than in patients with true-positive results (3.4 versus 2.1 g/dL for Multistix 10 SG and 3.8 versus 2.0 g/dL for Combur reagents strips, respectively). Therefore, we speculate that leukocyte esterase in both test systems may be inhibited by high ascitic protein content thereby decreasing diagnostic sensitivity of SBP.
As stated by Nousbaum et al., reagent strips were designed for detecting urinary tract infection. Because proteinuria above 1 g/L is a rare condition, urinary protein content is unlikely to interfere with detection of leukocytes by these reagent strips. In ascites, however, higher protein concentrations may cause insufficient sensitivity of reagent strips to detect SBP. Thus, we fully agree with Nousbaum et al. that it is premature to replace standard ascitic analyses by reagent strips due to their low sensitivity in detecting SBP.