Background: There is no satisfactory treatment for refractory ascites in patients with cirrhosis. Both peritoneovenous shunts and paracentesis have been used, but there is uncertainty about their relative merits.
Methods: We studied 89 patients with cirrhosis and refractory ascites who were randomly assigned to receive either repeated large-volume paracentesis plus intravenous albumin or a LeVeen peritoneovenous shunt. Patients in the paracentesis group in whom recurrent tense ascites developed during follow-up were treated with paracentesis, and those in the peritoneovenous-shunt group with diuretic agents or by the insertion of a new shunt if there was shunt obstruction.
Results: During the first hospitalization, ascites was removed in all 41 patients in the paracentesis group and in 44 of the 48 patients in the peritoneovenous-shunt group. The mean (+/- SD) duration of hospitalization in the two groups was 11 +/- 5 and 19 +/- 9 days, respectively (P less than 0.01). There were no significant differences in the number of patients who had complications or died. During follow-up, 37 patients in each group were hospitalized again. In the paracentesis group, the number of rehospitalizations for any reason (174 vs. 97 in the peritoneovenous-shunt group) or for ascites (125 vs. 38) was significantly higher, and the median time to a first readmission for any reason (1 +/- 1 vs. 2 +/- 2 months) or for ascites (2 +/- 2 vs. 8 +/- 17 months) was significantly shorter than in the peritoneovenous-shunt group. The total times in the hospital during follow-up, however, were similar in the two groups (48 +/- 49 and 44 +/- 39 days, respectively). Three patients had obstructions of their peritoneovenous shunts during their first hospitalizations, and 15 patients had a total of 20 obstructions during follow-up. Survival was similar in both groups.
Conclusions: The LeVeen shunt and paracentesis are equally effective in relieving refractory ascites. The former may provide better long-term control of ascites, but shunt occlusion is common and survival is not improved.