A team from Grass Valley, California, USA, evaluated patient survival after the administration of human albumin.
Two investigators from Hygeia Associates looked at randomized controlled trials comparing albumin therapy with crystalloid therapy, no albumin, or lower doses of albumin.
They conducted computer searches of the MEDLINE and EMBASE databases, the Cochrane Library, and Internet documents. In addition, they performed hand-searching of medical journals; inquiries to investigators and medical directors; and review of reference lists.
The researchers independently extracted data, and measured the relative risk of death.
|Albumin does not significantly affect patient mortality.|
|Annals of Internal Medicine|
Several criteria were used to assess methodologic quality. These were blinding, method of allocation concealment, presence of mortality as a study end-point, and crossover. Small-trial bias was also investigated.
A total of 55 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 found to be 1.11.
Relative risk was lower among trials with blinding (0.73; n = 7), mortality as an end-point (1.00 n = 17), no crossover (1.04; n = 35), and 100 or more patients (0.94; n = 10).
In trials with two or more such attributes, relative risk was further reduced.
Drs Mahlon M. Wilkes and Roberta J. Navickis concluded that, 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, they added.