Background: Spontaneous bacterial peritonitis (SBP) can be diagnosed via leucocyte esterase reagent strips, although diagnostic performances vary.
Aim: To perform critical review of literature on the use of reagent strips in SBP.
Methods: Nineteen studies were analysed (Medline search), comparing reagent strips in cirrhotic ascites vs. cytobacteriological methods. Diagnostic grades (G) were: GO = 0 leucocytes/mm3; G1 = 15; G2 = 70; G3 = 125; G4 = 500 for Multistix, GO = 0; G1 = 25; G2 = 75; G3 = 500 for Nephur, Combur, UriScan, and GO = 0; G1 = 25; G2 = 75; G3 = 250; G4 = 500 for Aution.
Results: Medians per study were: 75 patients (range: 31-1041), 136 ascites (47-2123), 17 SBP (5-117). For Multistix (12 studies), the sensitivities fell within the ranges 64.7-100% (G > or = 1), 45.7-83% (G > or = 2) and 45.3-89% (G > or = 3). For Nephur (n = 2), Combur (n = 6), UriScan (n = 1), sensitivities ranged 80.4-100% (G > or = 1), 63-100% (G > or = 2) and 67.7-97% (G > or = 3). For Aution (n = 3), sensitivities ranged 93-96% (G > or = 2) and 89% (G > or = 3). Nephur, Combur, UriScan displayed higher sensitivities than Multistix. However, in larger studies, sensitivities dramatically fell at 45.3% for Multistix (G > or = 3) if ascites polymorphonuclear count <1000/mm3 and 22.2% for bacterascites or 16.7-25% for asymptomatic patients.
Conclusion: Use of reagent strips for the diagnosis of SBP cannot be recommended, in view of low sensitivity and a high risk of false negatives, especially in patients with SBP and low polymorphonuclear count.