Five RCTs were included (n=3,819 patients; 3794 included in the analysis). All RCTs were randomised, double blinded and placebo controlled; randomisation was in blocks of four and stratified by treatment centre. Four RCTs reported completion rates. 80% of placebo patients and 81% tirilazad patients completed treatment; it was reported that there was no significant baseline differences between groups.
Tirilazad was associated with a significant decrease in symptomatic vasospasm (OR 0.80, 95% CI 0.69 to 0.93; five RCTs). Sensitivity analysis showed that exclusion of any one of three trials significantly influenced the meta-analysis result. A tirilazad dose greater than 6mg/kg/day was associated with significantly decreased odds of symptomatic vasospasm (OR 0.75, 95% CI 0.61 to 0.93; two RCTs) but this was not significant for doses of less than 6mg/kg/day. The odds of symptomatic vasospasm also decreased progressively with increasing dose of tirilazad (number of trials unclear). Logistic regression analyses showed that tirilazad was independently associated with decreased vasospasm, but had no effect on Glasgow Outcome Scale or cerebral infarction. There were no significant interactions between tirilazad and gender, use of anticonvulsants and WFNS grade when considering Glasgow Outcome Scale and symptomatic vasospasm. Results were similar with adjustment for anticonvulsant use and whether sedation affected reliability of symptomatic vasospasm diagnosis.
Tirilazad was not significantly associated with unfavourable outcome (as measured by Glasgow Outcome Scale) (five RCTs) or cerebral infarction (four RCTs) compared with placebo.
Publication bias was not indicated. Significant statistical heterogeneity was only present for the outcomes of cerebral infarction (p=0.02) and the subgroup analysis for this outcome (less than 6mg/kg/day tirilazad dose) (p=0.01).
There was no significant difference between tirilazad and placebo groups in the likelihood of phlebitis (four RCTs), pulmonary complications (two RCTs), sepsis (four RCTs), hypotension (three RCTs), intracranial haemorrhage/re-bleeding (five RCTs), brain oedema (two RCTs), increased intracranial pressure (four RCTs) or arrhythmias (one RCT).