Low-dose steroids in ARDS.

The recent study by Dr. Meduri and colleagues (April 2007)1 regarding treatment of severe ARDS with steroids has critical flaws in the presentation and analysis of the data. The methylprednisolone and control groups differed significantly in the proportion of patients who had catecholamine-dependent shock (CDS) at baseline in the treatment group (15 of 63 patients, 23.8%) vs those in the control group (13 of 28 patients, 46.4%) [p 0.03]. The authors state that mortality rates by day 7 for patients with catecholamine-dependent shock were similar in the treatment and control groups: 80% and 76.9%, respectively. This translates into 12 fatalities and 10 fatalities, respectively. This contradicts data in Table 2 that describe 56 survivors at 7 days in the treatment group and 22 survivors in the control group. ICU mortality data are equally confusing. The text states that “ICU mortality for patients with catecholamine-dependent shock was 73% vs 46% (p 0.24), and for patients without shock was 81% vs 67% (p 0.29).” This contradicts data shown in Table 3 showing there were 50 survivors of ICU admission (79.4%) in the methylprednisolone group vs 16 survivors (57.4%) in the placebo group. These discrepancies need to be resolved. Beyond the technical presentation of the data, the difference between the groups in the proportion of patients with CDS at baseline, and the extremely high mortality rate in this subgroup— apparently nearly 80% in both groups—render this study of questionable significance in its entirety. It is quite possible that the apparent mortality benefit from the treatment can be largely ascribed to the greater number of patients with CDS in the control group. Other apparent benefits may also be biased since the high mortality in the placebo group drastically reduces the number of patients who may improve despite the lack of treatment. The authors should reanalyze their data after removing the patients with CDS at baseline. Although this would be post hoc subgroup analysis, it is apparently the only way to eliminate the biases caused by the inclusion of the patients with CDS.