BEST EVIDENCE IN ANESTHETIC PRACTICE
621
Commentary by G. Bryson
to 1.17) and stated that “there is no apparent differ-
ence in pulmonary edema, mortality, or length of stay
between isotonic crystalloid and colloid”.4
Wilkes and Navickis present a systematic review evalu-
ating the influence of albumin administration on mor-
tality. This is a well-conducted review by the standards
cited in the Users’ Guides to the Medical Literature.1
The review presents a focused clinical question, a rig-
orous search of the literature, appropriately selected
articles, duplicate evaluation, and a variety of sensitiv-
ity analyses based upon study quality. No statistically
significant increase in mortality was found when albu-
min was used in surgery/trauma (relative risk (RR)
1.12, 95% confidence interval (CI) 0.85 to 1.46),
burns (RR 1.76, 95% CI 0.97 to 3.17), or when all
indications were pooled (RR 1.11, 95% CI 0.95 to
1.28). Should the results of this systematic review
“allay concerns regarding the safety of albumin” as the
authors suggest?
One should always remain cautious when interpret-
ing the results of a systematic review, even a good one.
“The combining of heterogeneous material is a
commonly accepted threat to the validity of meta-
analysis”.2 More simply stated, you can’t combine
apples and oranges. Statistical tests for heterogeneity
in this review were negative but substantial clinical
heterogeneity remains. The surgery/trauma subgroup
included studies of albumin for acute normovolemic
hemodilution, priming of cardiopulmonary bypass cir-
cuits, preservation of orthotopic liver transplants, and
direct management of colloid osmotic pressure. A
clinician would be unlikely to consult these sources
when trying to decide what fluid to give the hypoten-
sive patient in the operating room.
It is also easy to be drawn in by the simplicity of a
single number to summarize the results of all this
research. The 95% CI surrounding the estimates of
albumin’s effect are not statistically significant, leading
one to conclude that albumin is harmless. One should
be more critical. Most subgroups in this study demon-
strated an excess of mortality in the group treated with
albumin. The 95% CI of all analyses include the possi-
bility of clinically relevant harm and in no case did the
use of albumin create statistically or clinically signifi-
cant reductions in mortality. The statistical absence of
harm does not prove safety.
Lastly, it is important to place the results of this sys-
tematic review in the context of previous research.
The Cochrane Injuries Group Albumin Reviewers
found that albumin increased the risk of mortality (RR
1.52, 95% CI 1.17 to 1.99) and concluded that “there
is no evidence that albumin administration reduces the
risk of death in critically ill patients…”.3 Similarly
Choi and colleagues found a trend to lower mortality
with crystalloid resuscitation (RR 0.86, 95% CI 0.63
So should clinicians feel safe in choosing albumin?
Despite differences in methods and interpretation, all
systematic reviews are unanimous in showing no clin-
ical benefit to albumin therapy. Without clear benefit
and with unresolved concerns regarding the transmis-
sion of new variant Creutzfeld Jacob disease in human
blood products,5 a compelling argument for the use of
albumin is difficult to make.
Gregory L. Bryson MD FRCPC MSc
Ottawa, Ontario
References
1 Oxman AD, Cook DJ, Guyatt GH, for the Evidence-
Based Medicine Working Group. Users’ guides to the
medical literature. VI. How to use an overview. JAMA
1994; 272: 1367–71.
2 Chalmers TC. Problems induced by meta-analyses. Stat
Med 1991; 10: 971–9; discussion 979–80.
3 The Albumin Reviewers. Human albumin solution for
resuscitation and volume expansion in critically ill
patients (Cochrane review). In: The Cochrane Library,
Issue 1, 2002. Oxford: Update Software. Document
CD001208.
4 Choi PTL, Yip G, Quinonez LG, Cook DJ. Crystalloids
vs. colloids in fluid resuscitation: a systematic review.
Crit Care Med 1999; 27: 200–10.
5 Department of Health. Further precautionary measures
on blood products announced. Available from:
Commentary by P. Choi
The long-standing debate about the use of crystalloids or
albumin in seriously ill patients is one that has been
brought to the public’s attention with the increased
awareness of risks of human blood products and the
1998 publication of a systematic review on this topic.1
Wilkes and Navickis present another synthesis of the evi-
dence on this subject. Their review has many strengths:
comprehensive search strategy, explicit selection criteria,
duplication selection and abstraction of eligible random-
ized controlled trials (RCTs), evaluation of methodolog-
ical quality, clear data analysis, and transparent reporting
of their study based on the QUOROM (Quality of
Reporting of Meta-Analyses) format.2
This systematic review raises several important issues.
First, research findings must be interpreted with cau-
tion. Wilkes and Navickis report non-statistically signif-
icant differences in mortality for all patients and for
various subgroups. They conclude that “[t]his finding