I read with interest the arguments of my opponents. The questions posed are undoubtedly crucial. As I have originally foreseen, in my rebuttal, I will have to discuss the available data in more details.
First, does administration of albumin improve outcomes in severe sepsis or septic shock? I agree with my opponents that we do not yet have a definitive answer. Nonetheless, the actual data available firmly suggest a beneficial effect of albumin-containing solutions, particularly during shock. The post hoc analysis of the ALBIOS trial revealed a significant 6.3% absolute reduction in 90-day mortality associated with albumin replacement in patients with shock, which remained significant even after adjustments for baseline unbalances (P = .02). Such improvement in survival remained significant even after adjustments for clinically relevant variables, when considering patients with shock as defined with a broader definition. In addition, all the meta-analyses performed reported data in favor of a beneficial effect of albumin in patients with shock. A this point, two considerations merit our attention. First, the “signal” supporting a survival benefit of albumin administration appears to refer to patients with shock and not to those without shock. Second, we should not contrast a single P value against a biological plausibility of a hypothesis. Even the analysis mentioned by my opponents as not significant and reported in the ALBIOS trial (relative risk in the albumin group during shock of 0.88 [95% confidence interval, 0.77-1.01] after adjustments for clinically relevant variables, P = .07) does not necessarily confirm the null hypothesis of the trial. In contrast, it may indicate either that the outcome chosen to test the hypothesis (mortality) may be inappropriate and other clinically relevant outcomes should be tested, or that the beneficial effect of the treatment may have a greater extent (and less variability) in specific patient subcategories.