Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Patient Survival after Human Albumin Administration A Meta-Analysis of Randomized, Controlled Trials.

Wilkes MM, Navick RJ.

Ann Int Med 2001; 135:149-164. [abstract; full-text for subscribers]

Reviewed by Al Torres, MD, MS, University of Illinois College of Medicine at Peoria

Review posted August 21, 2002

---

I. Are the results of the study valid?

A. Primary questions:

1. Did the overview address a focused clinical question?

Yes. The authors state in the abstract that the purpose of the meta-analysis was to "test the hypothesis that albumin administration is not associated with excess mortality." However, the authors partially tested the hypothesis that more albumin administration is not associated with excess mortality since the authors included seven trials in which the control group received a lower dose of exogenous albumin (see below).

2. Were the criteria used to select articles for inclusion appropriate?

Yes. The authors included randomized trials that compared albumin therapy in the treatment groups to crystalloid therapy, no therapy with exogenous purified albumin, or a lower dose of exogenous purified albumin in the control groups. The 55 trials included in the meta-analysis were not limited by therapeutic endpoint or patient population studied. The authors clearly state they used such broad criteria to avoid exclusion bias and to increase statistical power and generalizability.

B. Secondary questions:

3. Is it unlikely that important, relevant studies were missed?

Yes. The authors not only performed computer searches of MEDLINE and EMBASE databases, the Cochrane Controlled Trials Register, and the Cochrane Medical Editors Trial Amnesty for published and unpublished trials, but they also performed manual searches of JAMA, New England Journal of Medicine, The Lancet, and BMJ for the period January 1990 to November 2000. They used alternate computer search engines to investigate conference reports, abstract, and reference lists for potential trials. The authors did not exclude foreign language articles and did consult medical directors of albumin suppliers and authors of published randomized trials.

4. Was the validity of the included studies appraised?

Yes. The authors only included trials in which mortality data were available, either published or available from the investigators. The authors classified the trials according to method used to conceal randomized group allocation (adequate, inadequate, or unclear), blinding vs. no blinding, crossover vs. no crossover, and trial size (< 100 patients vs. 100 patients). The authors did assess if the treatment and control groups received concomitant fluid therapy with asanguineous colloids and blood products. The authors did not mention if they assessed if the treatment and control groups were similar in severity or other patient characteristics. The authors made a significant effort to avoid repeated inclusion of data of patients who may have appeared in multiple reports.

5. Were assessments of studies reproducible?

Unclear. The authors reviewed the trials, extracted the data independently, and resolved differences in interpretation by discussion. Agreement between observers was not quantified since the kappa statistic or other measure of inter-observer variability was not performed.

6. Were the results similar from study to study?

Yes. The authors used a test of heterogeneity to assess if the differences in the size of effect between trials were due to chance (p > 0.05) or more likely due to differences in patients, exposures, outcomes, or study design (p < 0.05). The p value was 0.1 for all the trials combined or when the trials were stratified according to patient subgroup (e.g., burns, high-risk neonate, surgery), or study validity criteria (e.g., blinding, adequacy of concealment, crossover, mortality end point, and combination of two of these criteria). Since no significant heterogeneity was found between trials, a fixed-effects model was used to calculate pooled relative risk and 95% confidence intervals for the 42 trials with at least one death.

II. What are the results?

1. What are the overall results of the review?

The pooled relative risk for death for the 42 trails was 1.11 (95% CI, 0.95 to 1.28), indicating that albumin administration had no significant effect on mortality. Relative risk was lower for pooled trials with valid study criteria present (a RR > 1 implies a higher risk of death in the albumin groups):

Trial Categories	Relative Risk (95% CI)
Blinding vs. no blinding	0.73 (0.48 to 1.12) vs. 1.17 (1.00 to 1.37)
Mortality endpoint vs. no mortality endpoint	1.00 (0.84 to 1.18) vs. 1.49 (1.11 to 2.00)
Crossover vs. no crossover	1.04 (0.89 to 1.22) vs. 1.43 (1.00 to 2.05)

When the trials were analyzed according to trial size, smaller trials (< 100 patients) favored the control group and the larger trials (> 100 patients) favored the treatment groups.

Trial Categories	Small Trials	Large Trials
All (n of 32 vs. 10)	1.37 (1.10 to 1.70)	0.94 (0.77 to 1.14)
Blinding	1.22 (0.58 to 2.57)	0.54 (0.31 to 0.92)
Adequate allocation concealment	1.44 (1.00 to 2.07)	0.97 (0.73 to 1.29)
Mortality endpoint	1.20 (0.87 to 1.64)	0.93 (0.76 to 1.14)

The patient subgroups tended to favor the control groups.

Trial categories (n)	Relative Risk (95% CI)
Surgery or trauma (20)	1.12 (0.85 to 1.46)
Burns (4)	1.76 (0.97 to 3.17)
Hypoalbuminemia (4)	1.59 (0.91 to 2.78)
High-risk neonates (6)	1.19 (0.78 to 1.81)
Ascites (4)	0.93 (0.67 to 1.28)
Other (4)	0.91 (0.67 to 1.22)

2. How precise were the results?

The relative risk 95% CI of the larger trials are reasonably narrow. Since the majority of 95% CI include one (with the exception of the pooled blinded, large trials and pooled blinded trials with mortality endpoint), the risk of death between the treatment and control groups are likely to be equivalent in 95 out of 100 future trials.

III. Will the results help me in caring for my patients?

1. Can the results be applied to my patient care?

Although the trials included a more diverse group of patients and a larger number of trials than two previously published meta-analyses, (1,2) this overview did not assess the effect of albumin on critically ill children. The meta-analysis also didn't assess the effect of albumin on the mortality of critically ill populations with sepsis and/or acute respiratory distress syndrome. The concomitant use of blood products complicated the quantitative assessment of albumin on mortality.

It may be justified not to use albumin in burn patients and hypoalbuminemic patients since there is a trend towards albumin being harmful in these two subgroups.

2. Were all clinically important outcomes considered?

No. Length of stay was not assessed.

3. Are the benefits worth the harms and costs?

This meta-analysis certainly did not demonstrate an increase in survival for patients receiving albumin and since albumin is over 30 times more expensive than crystalloid solutions, the benefits do not appear to outweigh the high cost. The Cochrane Group's (1) systematic review of albumin in critically ill patients reported a significantly greater risk of dying in patients receiving albumin, especially in burn patients and hypoalbuminemic patients. In contrast, the Schierhout and Roberts (2) systematic review in critically ill patients reported an insignificant increase in mortality in the albumin groups included. In the face of the current evidence, pediatric intensivists should refrain from administering albumin to critically ill children, especially to those with burns or who are hypoalbuminemic, until a large, prospective randomized trial has demonstrated albumin administration to be beneficial in critically ill children.

References:

1. Cochrane Injuries Group Albumin Reviewers. Human albumin administration in critically ill patients: systematic review of randomized controlled trials. BMJ 1998; 317:235-240. [full-text]; PedsCCM EB Journal Club Review, by A Torres
2. Schierhout G, Roberts I. Fluid resuscitation with colloid or crystalloid solutions in critically ill patients: a systematic review of randomised trials. BMJ 1998;316:961-965. [full-text]; PedsCCM EB Journal Club Review, by M. Verive

Also see:

Choi PT, Yip G, Quinonez LG, Cook DJ. Crystalloids vs. colloids in fluid resuscitation: a systematic review. Crit Care Med 1999; 27:200-10. [abstract]; PedsCCM EB Journal Club Review, by M. Sanchez-Mendiola

Comments

Back to the EB Journal Club Index