Patient survival after human albumin administration. A meta-analysis of randomized, controlled trials

To test the hypothesis that albumin administration is not associated with excess mortality.

Computer searches of the MEDLINE and EMBASE databases, the Cochrane Library, and Internet documents; hand searching of medical journals; inquiries to investigators and medical directors; and review of reference lists.

Randomized, controlled trials comparing albumin therapy with crystalloid therapy, no albumin, or lower doses of albumin.

Two investigators independently extracted data. The primary end point was relative risk for death. Criteria used to assess methodologic quality were blinding, method of allocation concealment, presence of mortality as a study end point, and crossover. Small-trial bias was also investigated.

Fifty-five trials involving surgery or trauma, burns, hypoalbuminemia, high-risk neonates, ascites, and other indications were included. Albumin administration did not significantly affect mortality in any category of indications. For all trials, the relative risk for death was 1.11 (95% CI, 0.95 to 1.28). Relative risk was lower among trials with blinding (0.73 [CI, 0.48 to 1.12]; n = 7), mortality as an end point (1.00 [CI, 0.84 to 1.18]; n = 17), no crossover (1.04 [CI, 0.89 to 1.22]; n = 35), and 100 or more patients (0.94 [CI, 0.77 to 1.14]; n = 10). In trials with two or more such attributes, relative risk was further reduced.

Overall, no effect of albumin on mortality was detected; any such effect may therefore be small. This finding supports the safety of albumin. The influence of methodologic quality on relative risk for death suggests the need for further well-designed clinical trials.