For the primary analysis, the relative risk reduction from the CRASH-2 trial was applied to estimate the number of premature deaths that could be averted (1) if all Imatinib cell line patients received TXA within one hour of injury, and

(2) if all patients received TXA within three hours of injury. The numbers of deaths averted in each Inhibitors,research,lifescience,medical country were combined to give an overall global estimate. To identify the countries with the greatest potential for benefit from TXA, countries were ranked in order of the estimated number of premature deaths averted. To investigate the impact on the results of uncertainty in the parameter estimates used in the modelling, a number of sensitivity analyses were conducted. First, the analysis was repeated using the lower and upper bounds Inhibitors,research,lifescience,medical of the 95% confidence intervals for the parameter estimates to explore the effect of parameter uncertainty. Second, we repeated the analysis using the relative risk estimate for all-cause mortality rather than death due to bleeding. This analysis was conducted to take account of the possibility that some patients who Inhibitors,research,lifescience,medical do not die from bleeding because of TXA administration would nevertheless die of

other causes such multi-organ failure or brain injury. Third, we repeated the analyses using the relative risk estimate Inhibitors,research,lifescience,medical for all-cause mortality with TXA when given at any time within eight hours of injury. Although, previously published subgroup analyses show that early treatment is more effective it is possible that treatment within three hours is not possible in some settings. For each estimate, to reflect statistical uncertainty around the relative risks of TXA, an uncertainty range was estimated by calculating the numbers of deaths averted based on the

95% confidence intervals for the relative risks. The analyses were conducted using Microsoft Excel and STATA 11 (TX: StataCorp LP) software. Results Estimation of the effect of TXA on death due to bleeding by geographical Inhibitors,research,lifescience,medical region Figure ​Figure11 shows the found effect of TXA given within three hours of injury on death due to bleeding by geographical region. There was no evidence for heterogeneity in the effect of TXA by region (χ2 = 1.445; p = 0.70). The overall RRs for the effect of TXA on death due to bleeding when given within one hour (RR = 0.68; 95% CI 0.57 to 0.82) and within three hours (RR = 0.72; 95% CI 0.63 to 0.83) of injury were therefore taken as the most reliable guide as to the approximate RRs in all regions, and were used to estimate the number of deaths that could be averted with TXA. Figure 1 Risk ratio (95% CI) for death due to bleeding with TXA given within three hours of injury, overall and by geographical region.