I read with interest the article by Nousbaum et al.1 about the diagnostic accuracy of the Multistix 8 SG reagent strip (Bayer HealthCare) for the diagnosis of spontaneous bacterial peritonitis (SBP). A test sensitivity of 45.3% is disappointing for an infectious complication with a still high mortality. The test was read by an investigator and a nurse; when there was discordance between the 2 readers, the highest value was registered. If we need to increase the sensitivity, it may be better to register the lowest value because the majority of the patients were free of SBP. Because concordance was good but not perfect, to recalculate statistics in this other way may substantially change the sensitivity and specificity.
It would be interesting to know the sensitivity and specificity found in the study if a different cutoff were considered. In our study using Aution sticks (Menarini Diagnostics),2 the sensitivity increased to 89%-96% when the cutoff passed from ≥3 to ≥2 on a scale of 0-4, with only a slight decrease in specificity. Contrary to what the authors state in their discussion, only patients with cirrhosis were included in our study, and in only 1 of 228 cases did the patient have peritoneal carcinomatosis, but it was correctly classified by the reagent strip as a true negative. In 2 of 3 studies comparing Multistix with other reagent strips,1 the sensitivity was clearly lower with Multistix. We can conclude that Multistix is not enough sensitive and may be not used to diagnose SBP. Results with other strips are encouraging, although every dipstick should be tested and compared with others before its use is recommended.
I agree with the authors that reagent strips should not systematically replace standard fluid analyses. However, they can be very useful when analyses are not quickly available or are unavailable. A clearly positive test is highly indicative of SBP and is an indication of antibiotic therapy; a clearly negative test excludes SBP with confidence because the negative predictive value has ranged from 96%-100% in all the studies reported until now. In those cases with intermediate results, a conventional polymorphonuclear leukocyte count is mandatory if it is available, and if it is not, empirical antibiotic therapy should be initiated.
Finally, Nousbaum et al.1 reported a low prevalence of SBP in asymptomatic outpatients (0.57%). To the best of our knowledge, 5 studies have prospectively studied SBP prevalence in outpatients with asymptomatic cirrhosis undergoing large-volume paracentesis.3–7 Only 1 of 834 cases was diagnosed with SBP. In view of this figure, I think that ascitic fluid analysis may be not necessary in this clinical context.