Colloid Use for Fluid Resuscitation: Evidence and Spin

• Fluid therapy
• Colloids
• Crystalloids
• Albumin
• Critical care

In this issue, Wilkes and Navickis (1) present a meta-analysis of albumin versus crystalloids in seriously ill patients. Their study has many strengths. The comprehensive search strategy minimized the chance of publication bias and English-language bias. Explicit selection criteria were used, including studies with randomized allocation; comparison of albumin administration with crystalloid, no albumin, or lower-dose albumin; and an end point of mortality. The investigators selected trials and abstracted data in duplicate. Other criteria for quality assessment of the trials included treatment allocation, crossovers, and blinding. Blinding of whom (clinicians, researchers, outcome assessors, analysts, or patients) and for what purpose (minimization of co-interventions and outcome ascertainment) is a useful technique for reporting meta-analyses that acknowledges the inherent problems associated with conventional blinding labels (2). The analysis is clear and replicable, and the authors used the QUORUM (Quality of Reporting of Meta-Analyses) format to aid in transparent reporting (3).

Wilkes and Navickis conclude that their findings “should serve to allay concerns regarding the safety of albumin” (1). However, the pooled relative risk for death was 1.11 (95% CI, 0.95 to 1.28) in all patients receiving albumin, 1.12 (CI, 0.85 to 1.46) in surgery or trauma patients, 1.76 (CI, 0.97 to 3.17) in patients with burns, and 1.59 (CI, 0.91 to 2.78) in patients with hypoalbuminemia. These results are reassuring only insofar as they fail to show a statistically significant increase in mortality. In each case, the point estimate—the best estimate of the true effect of treatment—shows an increase in the relative risk for death of more than 10% overall and up to 76% in subgroups. Confidence intervals estimate the range within which the true effect plausibly lies. These confidence intervals indicate that the results are consistent with a relative overall increase in mortality of 28% and a threefold …