Publication

• Title: Effect of intravenous corticosteroids on death within 14 days in 10 008 adults with clinically significant head injury (MRC CRASH trial): randomised placebo-controlled trial
• Acronym: CRASH (Corticosteroid Randomisation After Significant Head injury)
• Year: 2004
• Journal published in: The Lancet
• Citation: CRASH trial collaborators. Effect of intravenous corticosteroids on death within 14 days in 10 008 adults with clinically significant head injury (MRC CRASH trial): randomised placebo-controlled trial. Lancet. 2004;364(9442):1321-8.

Context & Rationale

• Background

  • Small, heterogeneous head-injury corticosteroid trials and a widely cited synthesis had left uncertainty about benefit versus harm, with possible small mortality effects but imprecision and risk of bias. ¹
  • Despite uncertainty, corticosteroids were used in some settings for traumatic brain injury (TBI) on biological plausibility (anti-oedema, anti-inflammatory effects) and extrapolation from other neurotrauma contexts.
  • A large, pragmatic, globally generalisable trial was needed to resolve a clinically important question where even small absolute outcome changes would translate into substantial population impact.
• Research Question/Hypothesis

  • In adults with clinically significant head injury (GCS ≤14) treated within 8 hours, does a 48-hour intravenous methylprednisolone regimen reduce death and disability compared with placebo? ²
  • Pragmatic “uncertainty principle”: enrol when the responsible clinician was substantially uncertain whether corticosteroids would be beneficial or harmful. ²
• Why This Matters

  • Head injury is common globally; even a modest relative effect on mortality/disability would have major implications for emergency and critical care practice, especially in resource-limited settings.
  • The accompanying commentary stressed the importance of a definitive answer, while noting the limited mechanistic granularity typically available in large pragmatic trials. ³

Design & Methods

• Research Question: Among adults with acute clinically significant head injury (GCS ≤14) within 8 h, does a 48-hour methylprednisolone infusion improve outcomes compared with placebo?
• Study Type: International, multicentre, randomised, placebo-controlled, double-blind, pragmatic trial (239 hospitals in 49 countries); allocation by central telephone randomisation or sequential treatment packs; minimisation for key prognostic factors.
• Population:
  • Adults (≥16 years) with clinically significant head injury, GCS 14 or less, randomised within 8 hours of injury; enrolled when the treating clinician was substantially uncertain about corticosteroid use. ²
  • Key exclusions: clear indication or contraindication to corticosteroids (per clinician judgement), and presentation beyond the time window; protocol details specify eligibility/consent approaches suitable for emergency research. ²
• Intervention:
  • Intravenous methylprednisolone: loading dose 2 g over 1 hour, followed by infusion 0.4 g/hour for 48 hours (total planned duration 49 hours including loading). ²
• Comparison:
  • Matching placebo infusion using identical trial packs/administration schedule; all other care according to local practice (pragmatic co-interventions). ²
• Blinding: Double-blind (participants, clinicians, and outcome assessors); emergency unmasking permissible when clinically essential (unmasked in 24 patients overall).
• Statistics: Planned sample size 20,000 to detect a 2% absolute difference in mortality (assuming ~15% mortality in placebo) with >90% power at the 1% significance level; primary analyses by intention-to-treat.
• Follow-Up Period: Early outcome at 14 days; longer-term follow-up planned to 6 months (later report obtained 6-month data for 9673/10,008 patients). ⁴

Key Results

This trial was stopped early. Recruitment was stopped after interim unblinded review by the Data Monitoring Committee because mortality was higher in the corticosteroid group than in the placebo group.

• Across 10,008 randomised adults, early mortality was higher with methylprednisolone than placebo (21.1% vs 17.9%; RR 1.18; 95% CI 1.09 to 1.27; P=0.0001).
• At 6 months (later report), mortality remained higher with methylprednisolone (25.7% vs 22.3%; RR 1.15; 95% CI 1.07 to 1.24; P=0.0001). ⁴
• Pre-specified subgroup displays showed no convincing evidence of benefit in any injury-severity stratum; early harm appeared broadly consistent, with a suggestion (in 14-day data) of greater harm when treatment started >3 hours after injury (interaction P=0.05 in the 14-day report).

Internal Validity

• Randomisation and allocation concealment: Allocation through central randomisation or sequentially numbered treatment packs; minimisation used for key prognostic variables (age, GCS, pupillary response); low risk of selection bias.
• Blinding and contamination: Double-blind placebo-controlled design; unmasking occurred in 24/10,008 (0.2%), limiting performance/detection bias.
• Follow-up completeness: Vital status at 14 days available for 9964/10,008 (99.6%), limiting attrition bias for the early primary endpoint.
• Protocol adherence: Full loading dose given in 9748/10,008 (99.0%); infusion continued for at least 24 hours in 8286/10,008 (82.8%).
• Post-randomisation exclusions: Protocol deviations included 62 patients <16 years, 21 randomised >8 hours post-injury, and 3 infusions stopped by relatives; these participants were retained in the intention-to-treat analysis (minimising bias from exclusions).
• Baseline comparability: Groups were well balanced for injury severity and clinical prognosticators (including GCS strata and pupillary response) and for CT-use rates (CT performed in 7781/10,008; no CT in 2227/10,008), supporting exchangeability.
• Outcome assessment: Primary endpoint (all-cause mortality) is objective, robust to measurement bias; complication outcomes were collected but were not the principal driver of inference.
• Timing and dose: Intervention delivered in the hyperacute phase (within 8 hours) with a high-dose regimen; early stopping was triggered by harm signal rather than futility.
• Statistical rigour: Intention-to-treat analysis; large sample size and narrow confidence intervals around mortality RR increase; early stopping may influence effect-size precision but direction of effect was consistent in later 6-month follow-up. ⁴

Conclusion on Internal Validity: Strong — randomisation and blinding were robust, follow-up for mortality was near-complete, protocol adherence was high, and the mortality harm signal was reproduced at 6 months, although early stopping may modestly affect precision of the estimated effect size.

External Validity

• Population representativeness: Broad international enrolment (239 hospitals, 49 countries) and pragmatic “uncertainty principle” inclusion support applicability to real-world adult TBI care pathways, including varied resource settings.
• Clinical spectrum: Included mild-to-severe clinically significant head injury (GCS ≤14), reflecting a common emergency/ICU TBI cohort; findings may not apply to paediatric patients, penetrating TBI, or patients with a specific separate indication for corticosteroids (e.g., adrenal crisis).
• Intervention generalisability: Tested a specific high-dose methylprednisolone regimen; conclusions most directly apply to routine administration of this regimen after acute head injury (not to physiological steroid replacement for proven adrenal insufficiency).
• Systems applicability: Because co-interventions were not protocolised, the harm signal is likely robust across diverse practice environments, but mechanistic inference (why harm occurred) is less transportable than the treatment-effect estimate itself.

Conclusion on External Validity: High for adult acute blunt head injury managed within 8 hours in emergency/ICU settings — the trial’s pragmatic, multinational design supports generalisability of the finding that routine high-dose corticosteroids worsen outcomes.

Strengths & Limitations

• Strengths: Very large sample; international multicentre pragmatic design; double-blind placebo control; near-complete early mortality follow-up; clinically decisive effect estimate; subsequent 6-month outcomes available for most participants. ⁴
• Limitations: Early stopping (potential influence on magnitude estimation); limited mechanistic data collection (cause-specific mortality, hyperglycaemia/insulin protocols, detailed infection adjudication); pragmatic co-interventions not standardised; disability endpoint reporting requires the later 6-month publication for full interpretation. ³

Interpretation & Why It Matters

• Clinical practice

  Routine high-dose methylprednisolone for acute traumatic brain injury should be avoided, as it increases mortality both early (14 days) and at 6 months. ⁴
• Trial methodology lesson

  CRASH illustrates the danger of adopting biologically plausible therapies based on small trials and meta-analyses; definitive, adequately powered pragmatic trials can overturn entrenched practice and prevent large-scale iatrogenic harm. ⁵
• Guideline impact

  Modern severe TBI guidelines explicitly recommend against corticosteroids to improve outcome or reduce intracranial pressure because of increased mortality demonstrated by CRASH. ⁶

Controversies & Subsequent Evidence

• Pre-trial evidence base and trial justification: Prior systematic review evidence was criticised for heterogeneity and small-study limitations; correspondence debated whether the evidence base justifying a mega-trial was methodologically secure, and highlighted the need for transparency about assumptions and trial governance. ⁷⁸
• Mechanism of harm: The accompanying editorial emphasised that the marked mortality excess was unexpected from prevailing mechanistic narratives, and noted that large pragmatic trials may not capture sufficient granular pathophysiology (e.g., infection adjudication, hyperglycaemia, cause-of-death, and care-limitation decisions) to explain harm. ³
• Consistency over time: The later 6-month outcomes confirmed higher mortality with corticosteroids and found no evidence that the effect differed by injury severity or time to treatment (contrasting with the early suggestion of time-dependent heterogeneity in the 14-day report). ⁴
• Evidence synthesis after CRASH: Updated systematic review evidence incorporating CRASH concluded corticosteroids should not be used routinely in acute traumatic brain injury because of increased mortality. ⁵
• Neurosurgical commentary and broader neurotrauma framing: Commentary in neurosurgical literature highlighted the clinical implications of abandoning high-dose steroids for head injury and reinforced the primacy of adequately powered randomised evidence over extrapolation from other neurotrauma syndromes. ⁹
• Later critique and “reappraisal” discussions: More recent commentary has revisited interpretability (dose selection, early stopping, and limited mechanistic/withdrawal-of-care data), while not disputing that the tested high-dose regimen increased mortality in the enrolled population. ¹⁰
• Guidelines: Subsequent guideline updates incorporated CRASH and recommend against corticosteroids for outcome improvement/ICP control in severe TBI. ⁶

Summary

• CRASH was a large, pragmatic, international, double-blind RCT testing high-dose intravenous methylprednisolone versus placebo in adult acute head injury (GCS ≤14) within 8 hours.
• The trial was stopped early because mortality was higher in the corticosteroid arm.
• At 14 days, mortality was 21.1% with methylprednisolone versus 17.9% with placebo (RR 1.18; 95% CI 1.09 to 1.27; P=0.0001).
• At 6 months (later report), mortality remained higher (25.7% vs 22.3%; RR 1.15; 95% CI 1.07 to 1.24; P=0.0001), with no evidence of benefit in disability outcomes. ⁴
• CRASH decisively shifted practice and guideline recommendations away from routine corticosteroid use in traumatic brain injury. ⁶

Further Reading

Other Trials

• 2011Cooper DJ, Rosenfeld JV, Murray L, et al. Decompressive craniectomy in diffuse traumatic brain injury. N Engl J Med. 2011;364:1493-1502.
• 2016Hutchinson PJ, Kolias AG, Timofeev IS, et al. Trial of decompressive craniectomy for traumatic intracranial hypertension. N Engl J Med. 2016;375:1119-1130.
• 2015Andrews PJD, Sinclair HL, Rodriguez A, et al. Hypothermia for intracranial hypertension after traumatic brain injury. N Engl J Med. 2015;373:2403-2412.
• 2019CRASH-3 trial collaborators. Effects of tranexamic acid on death, disability, vascular occlusive events and other morbidities in patients with acute traumatic brain injury: randomised, placebo-controlled trial. Lancet. 2019;394:1713-1723.

Systematic Review & Meta Analysis

• 1997Alderson P, Roberts I. Corticosteroids in acute traumatic brain injury: systematic review of randomised controlled trials. BMJ. 1997;314:1855-1859.
• 2005Alderson P, Roberts I. Corticosteroids for acute traumatic brain injury. Cochrane Database Syst Rev. 2005;(1):CD000196.
• 2017Fatima N, Al Rumaihi G, Shuaib A, Saqqur M. The role of decompressive craniectomy in traumatic brain injury: a systematic review and meta-analysis. Sci Rep. 2017;7:8800.
• 2021Al Lawati K, Sharif S, Maqsood H, et al. Tranexamic acid in traumatic brain injury: a systematic review and meta-analysis. Intensive Care Med. 2021;47:14-27.
• 2022Chen Y, Li G, Wu K, et al. Efficacy and safety of different drug therapies for traumatic brain injury: a systematic review and network meta-analysis. Front Pharmacol. 2022;13:1021653.

Observational Studies

• 2005Cohan P, Wang C, McArthur DL, et al. Acute secondary adrenal insufficiency after traumatic brain injury: a prospective study. Crit Care Med. 2005;33:2358-2366.
• 2000Dickinson K, Bunn F, Wentz R, Edwards P, Roberts I. Size and quality of randomised controlled trials in head injury: review of published studies. BMJ. 2000;320:1308-1311.
• 2008Perel P, Arango M, Clayton T, et al. Predicting outcome after traumatic brain injury: practical prognostic models based on large cohort of international patients. BMJ. 2008;336:425-429.
• 2008Steyerberg EW, Mushkudiani N, Perel P, et al. Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics. PLoS Med. 2008;5:e165.

Guidelines

• 2017Carney N, Totten AM, O’Reilly C, et al. Guidelines for the Management of Severe Traumatic Brain Injury, Fourth Edition. Neurosurgery. 2017;80:6-15.
• 2020Carney N, Totten AM, Ullman JS, et al. Guidelines for the Management of Severe Traumatic Brain Injury: 2020 Update of the Decompressive Craniectomy Recommendations. Neurosurgery. 2020;87:427-434.
• 2019Hawryluk GWJ, Aguilera S, Buki A, et al. A management algorithm for adult patients with both brain oxygen and intracranial pressure monitoring: the SIBICC consensus. Intensive Care Med. 2019;45:1783-1794.
• 2019Kochanek PM, Tasker RC, Carney N, et al. Guidelines for the Management of Pediatric Severe Traumatic Brain Injury, Third Edition: Update of the Recommendations. Pediatr Crit Care Med. 2019;20:S1-S82.
• 2023Picetti E, Iaccarino C, Servadei F, et al. Consensus for the management of isolated severe traumatic brain injury in centres without neurosurgical capability. World J Emerg Surg. 2023;18:10.

Notes

• The methylprednisolone regimen mirrored high-dose neurotrauma steroid strategies used historically in other syndromes; CRASH shows that extrapolation across neurotrauma entities can be misleading.
• The early report includes subgroup displays by injury severity and time-to-treatment; the later 6-month report found no evidence of effect modification by these strata. ⁴

Overall Takeaway

CRASH is a landmark negative (and harmful) pragmatic RCT demonstrating that routine high-dose intravenous methylprednisolone after acute clinically significant head injury increases mortality, a finding sustained at 6 months. It permanently shifted global neurocritical care practice and guideline recommendations away from corticosteroids for traumatic brain injury, and remains a canonical example of why definitive large-scale trials are required before widespread adoption of biologically plausible therapies.

Bibliography

• 1.Alderson P, Roberts I. Corticosteroids in acute traumatic brain injury: systematic review of randomised controlled trials. BMJ. 1997;314:1855-1859.
• 2.Edwards P, Arango M, Balica L, et al; CRASH Trial Collaborative Group. The CRASH trial protocol (Corticosteroid Randomisation After Significant Head injury). BMC Emerg Med. 2001;1:1.
• 3.Sauerland S, Maegele M. A CRASH landing in severe head injury. Lancet. 2004;364:1291.
• 4.Edwards P, Arango M, Balica L, et al; CRASH trial collaborators. Final results of MRC CRASH, a randomised placebo-controlled trial of intravenous corticosteroid in adults with head injury—outcomes at 6 months. Lancet. 2005;365:1957-1959.
• 5.Alderson P, Roberts I. Corticosteroids for acute traumatic brain injury. Cochrane Database Syst Rev. 2005;(1):CD000196.
• 6.Carney N, Totten AM, O’Reilly C, et al. Guidelines for the Management of Severe Traumatic Brain Injury, Fourth Edition. Neurosurgery. 2017;80:6-15.
• 7.Gregson BA, Mendelow AD, Murray GD, et al. CRASH trial is based on problematic meta-analysis. BMJ. 1999;319:578.
• 8.Alderson P. Design of CRASH trial. BMJ. 1999;319:1068.
• 9.Bracken MB. Steroids for acute spinal cord injury and head injury. Neurosurgery. 2005;57:858-860.
• 10.Reconsideration of the MRC CRASH trial at 20: time to reappraise corticosteroids for traumatic brain injury? J Intensive Care Soc. 2025;26:1-3.