Therapeutic Paracentesis With and Without

Intravenous Albumin in Cirrhosis

Clinical Scenario: Patients with ascites are generally recommended to receive IV albumin prior to or immediately after large volume paracentesis because of the risk of hepatorenal syndrome.

Clinical Question: In patients with tense ascites, does giving albumin reduce the incidence of complications enough to recommend its use?

Article: Randomized Comparative Study of Therapeutic Paracentesis With and Without Intravenous Albumin in Cirrosis. Gines et al. Gastroenterology 1988;94:1493-1502.

Study Design: Randomized controlled clinical trial.

52 patients given 40g albumin after being tapped 4-6L repeated until ascites completely drained. 53 patients tapped without albumin. Patients followed

24 patients from each group had plasma renin activity and plasma aldosterone levels measured prior to and 48 hours post tap.

16 patients from each group had inulin clearance and free water clearance measured before and 48 hours after treatment.

Patients also monitored for renal impairment (>50% increase in BUN or Cr) or decrease in serum Na by>5mEq/L.

All patients discharged with diuretics. Average length of stay 11 days in both groups.

1) Are the Results Valid?

Inclusion criteria: Cirrhosis with tense ascites, no liver CA, no encephalopathy, no GI bleed, no infection at admission, bili <10, platelets >40k, Cr <3, Urinary Na<10.

Patients were randomized blindly and are similar (see Table 1)

Follow-up was complete and those patients who did not complete follow-up were

mentioned in the article. Follow up periods were similar (29 vs 34 wks)

Physicians wee not blinded to treatment group, i.e. no sham infusion was given, but measurements not subjective. It is not certain whether the two groups were treated equally following the paracentesis.

The age and sex of the participants were similar, as were the number of patients with peripheral edema (which is thought to be a protective effect by some) The control group had better renal indices with respect to GFR and Free water clearance as compared to the treatment group. It is important to remember that patients with renal failure but Cr <3 were evenly divided into the 2 groups, and not randomized with the larger groups. This was done presumably to not make one group more likely to have renal failure by being worse (and possibly closer to failure per the study criteria) prior to the trial.

2) Assessment of Results:

See Table 3:

1 patient in treatment group developed hyponatremia, none developed renal impairment

9 patients in control group had hyponatremia, 6 developed renal insufficiency.

ARR for hyponatremia= .15, RRR=.89 ARR for renal impairment= .11, RRR=1

NNT for hyponatremia= 6.66 NNT for renal impairment=9

Results of the laboratory tests are provided on Table 5. Readmission rates were similar between the two groups (Figure 2). Figure 3 indicates the outcomes The top graph shows survival after paracentesis, with effectively no difference between the two groups. The bottom graph is most useful for the lines C and D. This indicates that those with either hyponatremia, renal failure or both did more poorly than controls, and as bad as those with other complications. This indicates that any setback, be it tested in this study, or consequences of the liver disease not affected by albumin (encephalopathy, GI bleed, etc) are indicative of poorer outcomes.

3) Will this help the patient in question?

This study applies to our general population in that it includes a general population for liver failure, and not one specific cause of liver failure. The substitute endpoints seem reasonably valid given that the way that we follow hyponatremia and renal failure is by serum assessments. Their definition of failure is a little suspect and may not be critical enough.

One has to decide if the surrogate endpoints that are affected by administration of albumin are really worth watching out for and balanced by the cost of the medication. Perhaps a better risk stratification method than those which allowed admission to this trial could be assumed if all the data from the study were here, i.e., did the people who went on to renal failure have something else going on that could be assessed, then the data might be useful. The numbers needed to treat in either case are low, but this study doesn’t seem to fully answer our questions, because of the assumed definitions of a bad outcome by lab testing and other design flaws. However, giving albumin seems to be a good idea if the person looking at the study feels that these endpoints are valid enough to draw a conclusion.