Article révisé par les pairs
Résumé : Background: Few effective treatments are available for patients with cirrhosis and refractory ascites. New treatment modalities are needed for these patients. Aim: To synthesise the available evidence on the efficacy and safety of automated low-flow ascites pump therapy in patients with cirrhosis and refractory ascites. Methods: Electronic databases were searched for trials evaluating automated low-flow ascites pump therapy in patients with refractory ascites. Results: Nine studies were included. Eight were case series, one was a randomised controlled trial. Pooled estimate rates were 0.62 (95% CI = 0.49-0.74) for the absence of requirement of large volume paracentesis (LVP) after pump insertion, 0.30 (95% CI = 0.17-0.47) for acute kidney injury, 0.27 (95% CI = 0.13-0.49) for bacterial peritonitis and 0.20 (95% CI = 0.09-0.37) for urinary tract infection. There was high heterogeneity between studies which was often reduced or eliminated in sensitivity analyses by excluding studies of patients with a mean or median model for end-stage liver disease (MELD) score > 15. Results of sensitivity analyses were similar to those of overall analyses. Mean increase in serum creatinine level after pump insertion was 23 µmol/L (95% CI = 10-35) with no heterogeneity between studies. The pooled estimate rate for pump-related side effects was 0.77 (95% CI = 0.64-0.87) with low heterogeneity between studies. Conclusion: This meta-analysis demonstrates that most patients treated with automated low-flow ascites pump therapy do not require LVP after pump insertion. Acute kidney injury occurs in 30% of patients and creatinine levels increase by a mean of 23 µmol/L after pump insertion. Bacterial peritonitis and urinary tract infection occur in 27% and 20% of patients respectively.