Millions of patients sustain a serious head injury every year. Many die or are permanently disabled. Corticosteroids have been the standard treatment for serious head injuries for years because inflammation after trauma is believed to contribute to neuronal degeneration. A survey reported that 64 percent of U.S. trauma centers used corticosteroids to manage serious head injuries. Previous trials have found an insignificant decrease in the absolute risk of death (about 1 to 2 percent) and only small gains in neurologic function with corticosteroid treatment. The international corticosteroid randomization after significant head injury (CRASH) collaborators designed a trial to conclusively demonstrate any benefit, no matter how small, of corticosteroids in head injury patients.

The randomized, placebo-controlled trial included adults older than 16 years with a Glasgow Coma Scale (GCS) score of 14 or less within eight hours of head injury. The trial only included cases where the physician was substantially uncertain whether or not to treat the patient with corticosteroids. Therefore, patients with clear indications or contraindications to corticosteroids were excluded. Eligible patients were randomized to receive intravenous methylprednisolone (Depo-Medrol) therapy or placebo. The groups were matched for sex, age, time since injury, GCS score, and pupil reactivity. The treatment group received a loading methylprednisolone dosage of 2 g over one hour followed by 0.4 g per hour for 48 hours. The primary outcomes were death within two weeks of injury, and death or disability within six months of injury.

More than 10,000 patients were enrolled from 239 hospitals in 49 countries. The 5,007 patients in the treatment group were comparable with the 5,001 patients in the placebo group in all important characteristics. The mean age was 37 years, and the average time from injury to randomization was three hours. Forty percent of participants had severe head injuries (GCS score of three to eight), and 23 percent had major extracranial injuries. A computed tomography (CT) scan was performed on 78 percent of patients, but only 23 percent were normal. Approximately 99 percent of treatment patients received a full loading dosage, and 83 percent completed at least 24 hours of therapy.

During week 2 follow-up, mortality data were collected for 9,964 patients. Twenty-one percent of the treated group died within two weeks of treatment compared with 18 percent of the placebo group (equivalent to 159 excess deaths in the treated group). The treated patients had a 1.18 percent higher relative risk of death compared with placebo patients. This number was not affected by severity of injury, time since injury, CT scan results, or extracranial injury. When the CRASH data were added to a meta-analysis of the effect of corticosteroids on death after head injury, the relative risk of death in steroid-treated patients was 1.12 percent higher in treated patients compared with placebo patients. The authors stopped the trial to publish their week 2 mortality data early (deferring the publication of month 6 disability data to a later article) because they believed that the CRASH results could substantially alter the suggested treatment of patients with head injury.

The authors concluded that, despite previous beliefs, corticosteroid therapy was associated with a significant rise in mortality within two weeks of a serious head injury. The study did not include cause of death data and, therefore, found only an association between steroid treatment and higher mortality rates, not a causal relationship. The authors suggest that corticosteroids should not be used routinely to treat patients with head injury, regardless of the severity of the trauma.

editor’s note: In an associated editorial,¹ Sauerland and Maegele make the calculation that the treatment of patients with head injuries with corticosteroids could have been responsible for more than 10,000 avoidable deaths. The article and editorial call for more well-designed studies with adequate statistical power to provide reliable evidence on which to base the management of trauma patients.—a.d.w.