ABSTRACT
BACKGROUND: Spontaneous bacterial peritonitis, a potentially fatal complication of cirrhotic ascites, is diagnosed when the polymorphonuclear leucocyte count in the ascitic fluid is>250/mm. Manual laboratory counting of ascitic polymorphonuclear leucocytes is, however, labour-intensive, costly, results in diagnostic delay and it is not available in all hospitals as part of the 'out-of-hours' service. Thus, a rapid diagnostic screening test for spontaneous bacterial peritonitis would be beneficial in this condition. An exciting new development in the diagnosis of spontaneous bacterial peritonitis is the use of bedside reagent strips; yet, concerns regarding the inherent subjectivity of result reading have prevented the widespread adoption of this technique in clinical practice. OBJECTIVE: To evaluate the combined use of a leucocyte esterase strip together with an objective portable spectrophotometric reading device in the diagnosis of spontaneous bacterial peritonitis when compared with standard manual laboratory polymorphonuclear leucocyte counting. METHODS: Nonselected cirrhotic patients undergoing diagnostic paracentesis had an ascitic sample sent for a conventional polymorphonuclear leucocyte count, Gram stain and culture. In addition, a sample was tested with a bedside Multistix 10SG reagent strip and the result was analysed by the Clinitek Status. The strip test was considered positive if it read anything other than negative (i.e. 'trace', '+1', '+2' or '+3'). RESULTS: The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of the reagent strips to diagnose correctly spontaneous bacterial peritonitis when compared with the manual laboratory polymorphonuclear leucocyte count were 100, 91, 50, 100 and 92%, respectively. CONCLUSIONS: Bedside leucocyte esterase strips, spectrophotometrically read, can reliably exclude spontaneous bacterial peritonitis in patients with cirrhotic ascites. In our series, a negative strip result effectively ruled out this important condition, and suggests that the requirement for manual polymorphonuclear leucocyte counting in this setting could be removed.
