Citations 0
December 23/30, 1998

Methylprednisolone for Unresolving ARDS

Author Affiliations

Margaret A.WinkerMDIndividualAuthorPhil B.FontanarosaMD, Senior EditorsIndividualAuthor


Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998

JAMA. 1998;280(24):2074. doi:10-1001/pubs.JAMA-ISSN-0098-7484-280-24-jbk1223

To the Editor.—Dr Meduri and colleagues1 present data from 24 patients with acute respiratory distress syndrome (ARDS) persisting for a week or longer who were initially randomized to either methylprednisolone treatment or placebo. We disagree with Meduri et al that the data are sufficient to conclude methylprednisolone offers clinical benefit. As noted in the accompanying Editorial,2 the trial's small size and its crossover design limit conclusions that can be drawn from this study. Of the 8 patients initially randomized to receive placebo, 4 (50%) crossed over to methylprednisolone, and 3 (75%) of these 4 patients subsequently died. We are concerned that use of an intent-to-treat analysis in a small crossover study may be uninterpretable. If the outcomes of the patients randomized to placebo but treated with methylprednisolone are included in the methylprednisolone arm, 17 (85%) of 20 treated patients survived the intensive care unit (ICU), whereas 3 (75%) of 4 patients randomized to receive and treated only with placebo (75%) survived (P = .54). Similarly, 15 (75%) of 20 methylprednisolone-treated patients survived to hospital discharge as did 2 (50%) of 4 placebo recipients (P = .55). With the crossover design, one cannot determine whether patients randomized to placebo but treated with methylprednisolone died because they received methylprednisolone too late or because they received methylprednisolone at all, or if methylprednisolone had no effect on their outcome. Infection complication data should be presented not only as "intention to treat" but also as "actual exposure" to corticosteroids. As shown, the data could overestimate the risk of infection in patients randomized to receive placebo but treated with methylprednisolone. Also, we are concerned that statistically insignificant but potentially clinically important imbalances in the rate of sepsis and severity of lung dysfunction and organ failure at study entry favored the methylprednisolone group.

First Page Preview View Large
First page PDF preview
First page PDF preview