We analyzed data from all episodes of SBP (n = 216) during a 7-year period that occurred in a nonselected series of 167 patients with cirrhosis. Low-risk episodes (urea <11 mmol/L and bilirubin <68 ¦Ìmol/L) were not treated with albumin, whereas high-risk episodes (urea >11 mmol/L and/or bilirubin >68 ¦Ìmol/L) were or were not given albumin at the discretion of the attending physician.
Sixty-four episodes of SBP (29.6 % ) were low risk and not treated with albumin, whereas 152 (70.4 % ) were high risk; 73 of these (48 % ) were treated with albumin and 79 (52 % ) were not. Renal failure before SBP resolution was less frequent after low-risk episodes than high-risk episodes (4.7 % versus 25.6 % ; P = .001), in-hospital mortality was lower (3.1 % versus 38.2 % ; P < .001), and the 3-month probability of survival was higher (93 % versus 53 % ; P < .001). In an analysis of only the high-risk group, patients who received albumin had lower in-hospital mortality than those not treated with albumin (28.8 % versus 46.8 % ; P = .02) and a greater 3-month probability of survival (62 % versus 45 % ; P = .01).
Albumin therapy increases survival of patients who have high-risk episodes of SBP, although it does not seem to be necessary for patients with low risk of death.