Society of Cardiovascular Anesthesiologists

April 2002 Newsletter

Patient Survival after Human Albumin Administration

A Meta-Analysis of Randomized Controlled Trials. Annals of Internal Medicine, 2001;135:149:64

Reviewer: Christos Koutentis, MB, ChB
Attending Anesthesiologist
VAMC - West Palm Beach, FL

A new meta-analysis was performed to address the question of whether albumin was actually associated with excess mortality in comparison to crystalloid. This new meta-analysis was designed to specifically address criticisms concerning the validity of an earlier meta-analysis performed by the Cochrane Group that indicated that albumin was associated with a significantly higher mortality (BJM 1998;317:235-40). The results of the Cochrane meta-analysis has been challenged because of the omission of relevant trials, the combination of disparate and heterogeneous trials, and inadequate assessment of the methodologic quality of outcomes. Further, the results of the Cochrane meta-analysis was inconsistent with the relatively few adverse reactions have been reported with albumin despite the widespread use over an extended period.

The new meta-analysis used data from all randomized trials with information on mortality. From a total of 78 trials, mortality data was available in 55 trials involving 3504 patients where there were 525 deaths. The primary outcome measure was relative risk of death. Seven (13%) of these trials were blinded, 12 (38%) were considered to have adequate randomized group allocation, and 17 (31%) had mortality as an end point. Patients in the control group crossed over into the albumin group in 8 trials (15%).

The relative risk of death associated with albumin for all trials was calculated to be 1.11 (95% CI, 0.95 to 1.28), indicating that albumin therapy had no significant effect on overall mortality. There was no significant between trial homogeneity in relative risk. It was noted that relative risk of death associated with albumin was 38% lower in blinded rather than unblinded trials, 33% lower in trials with mortality as an end point, and 27% lower in trials where patients in the control group were not crossed over into the albumin group. Further, no evidence emerged that albumin significantly affected mortality either across the board, or within any of the subcategories of indications for volume expansion.

Four methodological weaknesses were noted which systematically reduced the accuracy of estimation of relative risk. These were: blinding, mortality as an end point, no crossover, and sample size. Since these studies were not designed with the intention of following mortality or following up long enough to observe effects on mortality, there was a danger of bias in interpreting the results. There was also a pattern of crossover in that the most seriously ill patients were often switched to albumin therapy, but not the other way around. Since many of these trials involved small numbers of patients, an overestimate in either direction was easily manifested.

Some of the reasons for the discrepancy between this new meta-analysis and the one performed by the Cochrane Group included differences in the selection criteria used to qualify studies for inclusion in the meta-analysis. For example, the Cochrane meta-analysis included a trial involving 53 neonates where mortality was greater in the albumin group, but excluded a trial involving 100 neonates by the same group of investigators where mortality was greater in the non-albumin treatment group. The lack of blinding, absence of mortality end-point, and crossover may have also contributed to increased mortality associated with albumin in the Cochrane meta-analysis. Finally, this new meta-analysis included a greater number of trials involving a large number of patients, a factor that may have decreased the influence of small-trial bias.

A potential problem with both meta-analyses was that mortality in itself was a relatively insensitive endpoint since mortality rates are relatively low even for acutely ill patients in the setting of contemporary medical care. Moreover, both trials had relatively broad inclusion criteria, disparate indications for volume expansion, fluid management regimes, and patient populations. Both studies attempted to address heterogeneity by calculating the relative risk of death in subcategories of patients based on clinical indications for therapy, but the more recent meta-analysis failed to detect a significant effect of albumin in any specific patient group. It was also noteworthy that the fluid management regimes in the older trials did not necessarily reflect current management strategies. In some trials, fluids were administered to attain hemodynamic, physiologic, or biochemical targets while in other trials, patients received fixed dosages or volumes of fluid. Finally, concomitant administration of blood products made calculating the actual dosages or contribution of albumin difficult to estimate.

In summary, the current meta-analysis suggests that fears concerning the inherent safety of albumin may be less well founded than previously thought. Although better designed and powered trials may resolve or answer continuing uncertainties regarding the safety of albumin administration, the results of meta-analysis suggests that it would be exceedingly difficult to design such a trial.

Table of Contents:

Archived Newsletter Articles