The author replies

Abstract

O n behalf of my coauthors, I thank Gusmao-Flores et al (1) for their interest in our meta-regression analysis (2) that sought to explore whether ICU interventions, both pharmacologic and nonpharmacologic, that are more effective at reducing delirium duration are associated with a reduction in short-term mortality. In their letter, they suggest that for the trials we identified where a pharmacologic intervention was evaluated, we should have included only those studies comparing a medication with a placebo because "the positive results in this meta-analysis can only express the impact in inducing delirium caused by some of these drugs (i.e., morphine and midazolam)." Based on the results of a new delirium duration meta-analysis conducted by the authors, which includes only medication versus placebo studies along with the early mobilization study by Schweickert et al (3), and which represents only seven of the 15 studies included in our delirium duration meta-analysis, they conclude that average delirium duration is not affected by the interventions hypothesized to reduce delirium burden and then go on to suggest that the title and conclusions of our meta-regression analysis are not valid.A number of important concerns exist surrounding the statements that Gusmao-Flores et al (1) make. In a study that is randomized, the ability to measure differences between groups for a particularly outcome is not dependent on whether the intervention is compared to a placebo or to an existing gold standard therapy. Instead, factors such as sample size, use of blinding, the method and intensity of delirium assessment, delirium prevalence at study entry, and the exposure of the control group to a nonstudy intervention that could affect delirium occurrence (4) are more likely to influence the duration of delirium reported. The sensitivity analyses we reported in our article accounted for many of these factors. Although Gusmao-Flores et al (1) emphasize that only studies evaluating a pharmacologic intervention (vs a placebo) should have been included in our duration of delirium meta-analysis, in their duration of delirium meta-analysis, they include the early mobilization study by Schweickert et al (3) but exclude the Awakening and Breathing Controlled (ABC) trial by Girard et al (5). The ABC trial, which reported reduced sedative exposure and coma when daily awakening was added to a spontaneous breathing trial, should be included in any duration of delirium meta-analysis given the well-established relationship between coma and delirium (5).It is important to note that the duration of delirium metaanalysis we present in our article should not be considered a "stand-alone" result given the substantial heterogeneity that was found to exist between studies and the fact that the sole objective of our meta-regression analysis was to explore whether interventions that reduce delirium duration are associated with a reduction in short-term mortality. The methods used in our article are state-of-the-art, and the conclusions we make are data-...