Albumin: does not clearly increase mortality

Clinical bottom line (level 1a)

  1. Patients who received albumin compared with control are not clearly more likely to die.
Wilkes and Navickis: Ann Intern Med 2001; 135 : 149-164
Expires May 2004

The study

Systematic review of all randomised controlled trials of
  • Patients: unwell
  • Intervention: albumin therapy compared with crystalloid therapy, no therapy or a lower dose of albumin
  • Outcome: death

Articles found in all languages using Medline, Embase, Cochrane Controlled trials register, Cochrane Medical Editors Trial Amnesty, to November 2000 (search terms: not given ) and searching relevant Internet sources, suchs as conference reports, abstracts, compilations of references and full-text journal articles. JAMA, N Engl J Med, Lancet and BMJ were handsearched over the previous 10 years. Albumin suppliers and authors of published RCTs were contacted, and bibliographies of selected articles were also checked.

Selection criteria: by 2 independent reviewers with differences resolved by consensus.
Appraisal criteria: by 2 independent reviewers: based on randomisation concealment, intention-to-treat analysis, blinding
Articles excluded if:
  • control regimen included synthetic colloids, blood products or plasma protein fraction
  • duplicate reports
55 RCTs found involving 3504 patients
  • 27 involving surgery or trauma
  • 4 involving burns
  • 5 involving hypoalbuminaemia
  • 5 involving high-risk neonates
  • 5 involving ascites
  • 3 involving other conditions

Subgroup results were not found to be significantly heterogeneous.

The evidence

Outcome Time to outcome CER OR
(95% CI)		 NNH
(95% CI)
death 11 days 252/1502
(16.8%)		 1.11
(0.95 to 1.28) 		66
(NNT = 140 to infinity;
NNH = 27 to infinity)

  • Albumin was not found to increased mortality compared with control for any subgroup analysed.

Comments

  1. There was wide variation in the regimens used between the trials.
  2. Median number of patients enrolled per trial was 52 (range 10 to 300). Median duration of follow-up was 11 days (range 0.04 to 1096 days). The small size of many studies and the short follow-up means that important differences between albumin and control may have been missed.
  3. Note that the confidence intervals for mortality were not able to exclude the possibility of significant harm from using albumin. Other clinical effects of albumin were not reported in this systematic review.

Citation

  1. Wilkes MM, and Navickis RJ: patient survival after human albumin administration: a meta-analysis of randomized, controlled trials. Ann Intern Med 2001; 135 : 149-164
Search Terms: from ACP Journal Club
Contributor: Chris Ball, May 2002
Reviewer:

Clinical Question.
Patient
Intervention or Exposure
Outcome