We read with interest the study by Nousbaum et al. on the use of leukocyte reagent esterase strips for the diagnosis of spontaneous bacterial peritonitis.1 It is by far the biggest and the only multicenter study of a total of 16 prospective studies conducted (all published as full papers). Since the reports by Vanbiervliet et al.2 and Castellote et al.3, almost all strips (UriScan, Multistix 8SG and 10SG, Aution, Combur, Nephur) have come under scrutiny in small, single-center or 2-center studies, which reported sensitivities between 50% and 100% and specificity ranging between 81% and 100%.4, 5
We would like to draw attention to a couple of points. First, the number of included patients is slightly different (n = 1041) than the number found (n = 1069) in abstract presentations of the same study 2 years ago. That seems to have raised the initially reported sensitivity of the Multistix 8SG from 43.6% to 45.3%.6, 7
Second, in the current abstract of the study, the authors report that 2 reactive strips were tested independently in patients undergoing paracenteses. It is clear that the investigators did not use different brands of dipsticks to check the total of 2123 ascitic fluid samples. We assume this statement means that every sample was double tested on the spot (with 2 Multistix 8SG).
If that is the case, then we wonder if the 2 dipstick readers were each reading the dipstick they had each immersed in the sample, or was there eventually obligatory cross-reading.
Operator subjectivity and interobserver variability is extremely important for dipstick tests, and we are not sure if the reported concordance (K = 0.805) between the readers refers to a total of 2123 or in fact 4246 readings.