We read with interest the article by Nousbaum et al.1 In this multicenter French study, the authors report their results with Multistix 8SG reagent strip for the diagnosis of spontaneous bacterial peritonitis (SBP) in more than 2000 paracentesis procedures, performed in more than 1000 patients. This is the largest group of patients ever studied for this technique. Their conclusions focus on the low sensitivity of the dipsticks for the diagnosis of SBP found in the current study (45.3%), making the dipsticks an inaccurate test. The sensitivity for asymptomatic patients was even worse (16.7%). However, the negative predictive value (NPV) and specificity were high, ranging between 98% and 100%.
The results are really different from the findings of all other studies on the same issue that reported sensitivity greater than 64.7%, with NPV and specificity of barely 100% in 72 to 245 paracenteses per study.2–6
Vanbiervliet et al.2 published the first article in 2002, and reported sensitivity, specificity, NPV, and positive predictive value (PPV) of 100% in 72 paracenteses. The reagent strip used was the Multistix 8SG; tests were considered positive when leukocyte stearase became 1, 2, or 3+. Zero or trace detections were considered negative for diagnosis of SBP. The studies by Butani et al.3 and Sapey et al.4 used Multistix 10SG with the same criteria for positives and negatives and reached sensitivity of 83%-100%, specificity of 92.5%-100%, PPV of 75%-100%, and NPV of 96%-100%. Sapey et al.5 published another analysis of dipstick testing using Multistix 10SG and Nephur-Test with sensitivity of 64.7% and 88.2%, specificity of 99.6% and 99.6%, PPV of 91.7% and 93.8%, and NPV of 97.4% and 99.1%, respectively.
The differences between previous studies and the current study may arise from the interpretation of the dipsticks. The study by Nousbaum et al.1 considered the dipsticks positive when they became 2 and 3+, and negative when they indicated 0, trace, and 1+. These cutoffs were based on the manufacturer description for tests of urine. Thevenot et al.6 used the same criteria and had good results. However, the other papers on the rapid diagnosis of SBP used cutoffs for positive with results of 1 to 3+, and negative for 0 and trace.2–5 The study by Butani et al.3 used the receiver operating characteristic curve to determine the best cutoff for diagnosis.
SBP is a highly lethal disease, and its diagnosis has to be made with highly sensitive devices. We believe that in order to increase sensitivity of the dipstick test, the results of Nousbaum et al.1 should be better evaluated using the receiver operating characteristic curve to establish the best sensitivity with the least loss of specificity and NPV.