Mirror, mirror on the wall, which is the fairest meta-analysis of all?*

    loading  Checking for direct PDF access through Ovid


Although the results of initial studies with activated protein C, hydrocortisone, and intensified insulin therapy in severe sepsis and septic shock were encouraging and seemingly convincing, we have had to accept that recent consecutive trials have failed to reproduce these same positive results. After >20 yrs of randomized, controlled sepsis trials, there is not one single drug or treatment concept that has proven its mortality-reducing effect by the highest evidence-based standard: two controlled trials with a high statistical power. What shall we do? Shall we tell the families of our sepsis patients they have to wait until we have achieved the highest level of evidence for all our therapies? Or, would it be better to gather all the information we have (high- and low-quality trials, retrospective analyses, case control studies, expert opinions), judge it as best we can, and then recommend to our colleagues what they could do … even with all the uncertainty we are used to in intensive care medicine?In cases of adjunctive sepsis therapy with intravenous immunoglobulins (ivIg), the story goes back to the 1980s when the first placebo-controlled trials had been performed. Small sepsis trials with polyvalent immunoglobulins competed at that time with much larger monovalent approaches with monoclonal anti-endotoxin and anti-cytokine antibodies. However, the latter all failed, whereas the small ivIg trials displayed at least some positive aspects. The complexity was and is further advanced by the existence of several polyvalent ivIgG preparations, but only a single ivIgGMA preparation might act differently in those patients.The Cochrane Institution, which started the ivIg meta-analysis business (1) in 2002, analyzed adjunctive sepsis therapy with monoclonal and polyclonal ivIg in adults and neonates. In a subgroup analysis of 11 trials (n = 492) using polyclonal ivIg, a significant reduction in all-cause mortality was demonstrated for the treatment group (relative risk [RR], 0.64; 95% confidence interval [CI], 0.51, 0.80). It was especially interesting that post hoc subanalysis, according to type of polyclonal ivIg, demonstrated a greater reduction in mortality among patients given immunoglobulin M (IgM)-enriched ivIg (n = 194; RR, 0.48; 95% CI, 0.30, 0.76) compared with standard polyclonal ivIg (n = 219; RR, 0.68; 95% CI, 0.51, 0.89). Subgroup analysis according to age group showed a significant decrease in mortality among adults (n = 222; RR, 0.62; 95% CI, 0.49, 0.79) and a similar but not statistically significant result among neonates with sepsis (n = 241; RR, 0.70; 95% CI, 0.42, 1.18). Often, in meta-analyses dealing with small trials of low quality, the authors come to the conclusion that “polyclonal ivIg has a very promising role as an adjuvant therapy in sepsis,” but they also state that “large, multicenter studies are needed to confirm the effectiveness of polyclonal ivIgs in reducing mortality in patients with sepsis” (1).So far, correct, but how helpful?Shall we rely on grade C recommendations (the third of five grades) based on small, randomized trials with uncertain results and a moderate to high risk of false-positive (alpha) or false-negative (beta) error (the second of five grades of evidence [2]) and treat our patients with ivIg, or shall we not?Is it more helpful for the intensivists to get more and more meta-analyses of the same trials? We have ivIg meta-analyses by Pildal and Gotzsche (3), Norrby-Teglund et al. (4), and in this issue of Critical Care Medicine, Laupland et al. (5) and Kreymann et al (6).

    loading  Loading Related Articles