ROSE trial - Early Neuromuscular Blockade in ARDS

CRITICAL APPRAISAL · LANDMARK TRIAL

Randomized Controlled Trial

In moderate-to-severe ARDS managed with a modern, higher-PEEP lung-protective strategy, routine early paralysis with cisatracurium and deep sedation did not reduce 90-day mortality — and came with more serious ICU-acquired weakness.

National Heart, Lung, and Blood Institute PETAL Clinical Trials Network (Moss M, Huang DT, Brower RG, et al.). Early Neuromuscular Blockade in the Acute Respiratory Distress Syndrome. N Engl J Med. 2019;380(21):1997–2008. · PMID 31112383 · NCT02509078

External validity

Conducted across 48 U.S. academic and community hospitals already running a high-PEEP, ARDSnet lung-protective ventilation protocol — a "usual care" comparator that is lighter and more protocolized than the deeply-sedated comparator ACURASYS used a decade earlier, and closer to how many ICUs now manage ARDS. Prone positioning was used in only about 16% of patients, far less than the roughly 50% seen in ACURASYS and PROSEVA-era practice, so ROSE mainly answers whether routine paralysis adds benefit on top of a modern, higher-PEEP, lighter-sedation strategy — not whether it helps when combined with aggressive proning. Applies most directly to moderate-to-severe ARDS (P/F <150) managed with contemporary lung-protective ventilation; patients already on neuromuscular blockade, pregnant patients, and those on ECMO were excluded.

Key limitations

●      Open-label design — paralysis and sedation depth can't be blinded, exposing softer endpoints (co-interventions, the weakness exam) to bias even though the primary endpoint (death) is not easily biased.

●        Stopped early for futility at 1,006 of a planned 1,408 patients — although they had not predefined a futility rule. It is probably not as problematic as stopping early for harm or benefit. Precision around secondary and safety outcomes is reduced, and the trial was not powered to rule out smaller true effects.

●        Meaningful crossover: 17.0% of usual-care patients received a neuromuscular blocker within 48 hours, which would tend to narrow rather than widen any true difference between arms.

●        Prone positioning was used in only ~16% of patients, far less than in ACURASYS or PROSEVA-era practice, so the trial speaks to paralysis added to a lighter-sedation ARDSnet strategy — not to paralysis combined with aggressive proning. The mortality was also higher than expected at 42% in both groups.

●        The weakness comparison (46.8% vs 27.5%) relied on a bedside exam that could only be completed in patients well enough to cooperate, so a sizeable and probably uneven slice of each arm went unassessed — the true gap between groups could be larger or smaller than reported, even though prolonged paralysis is a plausible mechanistic cause of the weakness itself.