Outcome data were available for 12,639 randomised patients (6359 patients allocated to the TXA group and 6280 patients to the placebo group). For patients randomised within 3 hours of injury, outcome data were available for 9127 patients (4613 patients allocated to the TXA group and 4514 patients to the placebo group). A total of 12,561 (98.6%) patients were known to have completed the loading dose, and 11,866 (93.2%) patients completed the 8-hour maintenance dose.
Primary outcome
Figure 2 shows the number of deaths and cause of death by days since injury in all patients randomised. There were 2560 deaths in total and the median time to death was 59 hours after injury (interquartile range 20–151 hours). Among patients treated within 3 hours of injury, there were 1878 deaths overall. Appendix 4, Figure 12, shows the cumulative incidence of head injury death in patients randomised within 3 hours of injury.
FIGURE 2 Mortality by days since injury among all participants randomised. (Figure not reproduced here; see the original article.)
Table 3 shows the effect of TXA on head injury death in the 9127 patients randomised within 3 hours of injury with outcome data. Among patients treated within 3 hours of injury, the risk of head injury death was 18.5% in the TXA group versus 19.8% in the placebo group (855 vs. 892 events; RR = 0.94, 95% CI 0.86 to 1.02). In the prespecified sensitivity analysis that excluded patients with a GCS score of 3 or bilateral unreactive pupils at baseline, the results were 12.5% in the TXA group versus 14.0% in the placebo group (485 vs. 525 events; RR = 0.89, 95% CI 0.80 to 1.00).
| Head injury death | TXA | Placebo | RR (95% CI) | ||||
|---|---|---|---|---|---|---|---|
| n | N | % | n | N | % | ||
| All | 855 | 4613 | 18.5 | 892 | 4514 | 19.8 | 0.94 (0.86 to 1.02) |
| Excluding GCS score of 3, both unreactivea | 485 | 3880 | 12.5 | 525 | 3757 | 14.0 | 0.89 (0.80 to 1.00) |
Subgroup analysis
The effect of TXA on head injury death stratified by baseline GCS and pupillary reactions was examined (Figure 3). There was a reduction in the risk of head injury death with TXA in mild to moderate head injury (RR 0.78, 95% CI 0.64 to 0.95), but in severe head injury (RR 0.99, 95% CI 0.91 to 1.07) there was no clear evidence of a reduction (p-value for heterogeneity = 0.030). When we examined the impact of baseline GCS score in a regression analysis, there was evidence (p = 0.007) that TXA is more effective in less severely injured patients. Among patients with reactive pupils, head injury deaths were reduced with TXA (RR 0.87, 95% CI 0.77 to 0.98).
FIGURE 3 Effect of TXA on head injury death stratified by baseline severity in participants randomised within 3 hours of injury. (Figure not reproduced here; see the original article.)
We examined the effect of TXA on head injury death stratified by time to treatment and recorded no evidence of heterogeneity (p = 0.96). The RR of head injury death with TXA was 0.96 (95% CI 0.79 to 1.17) in patients randomised ≤ 1 hour after injury, 0.93 (95% CI 0.85 to 1.02) in those randomised > 1 to ≤ 3 hours after injury and 0.94 (95% CI 0.81 to 1.09) in those randomised > 3 hours after injury. However, as anticipated in the statistical analysis plan, patients treated soon after injury often have more severe head injury and so the impact of time to treatment could be confounded by severity.
Figure 4 shows the impact of time to treatment on the effect of TXA in patients with a mild or moderate head injury and in those with severe head injury after adjusting for GCS score, SBP and age in a multivariable model including all participants. Early treatment was more effective in patients with mild or moderate head injury (p = 0.005), but there was no obvious impact of time to treatment in severe head injury (p = 0.73). The effectiveness of TXA by time to treatment stratified by severity is further demonstrated in Figure 5. We recorded no evidence of heterogeneity in the effect of TXA by patient age (p = 0.45).
FIGURE 4 Effect of TXA on head injury death by severity and time to treatment in all participants with (a) mild and moderate head injury; and (b) severe head injury. Dotted lines represent 95% confidence limits. Mild/moderate, n = 8107; severe, n = 2703. (Figure not reproduced here; see the original article.)
FIGURE 5 Effectiveness of TXA on head injury death vs. time to treatment stratified by severity in all patients. Severe, n = 2703; both pupils react, n = 2204; moderate, n = 3897; mild, n = 4275. (Figure not reproduced here; see the original article.)
We examined the effect of TXA on head injury death stratified by World Bank country income level (high income vs. low and middle income). This analysis was not prespecified. Although the reduction in the risk of head injury death with TXA was higher in high-income countries (RR 0.76, 95% CI 0.55 to 1.04) than in LMIC (RR 0.92, 95% CI 0.81 to 1.04), there was no statistical evidence of heterogeneity by country income level (p = 0.258). As early head injury deaths are more likely than late head injury deaths to result from intracranial haemorrhage, we examined the effect of TXA on head injury deaths within 24 and 48 hours of injury. The RRs of head injury death were 0.81 (95% CI 0.69 to 0.95) and 0.89 (95% CI 0.79 to 1.02) within 24 and 48 hours, respectively. When patients with a GCS score of 3 and those with bilateral unreactive pupils at baseline were excluded, the corresponding values were 0.72 (95% CI 0.56 to 0.92) and 0.84 (95% CI 0.69 to 1.01).
The models are adjusted for GCS score, age and SBP. In patients with a mild and moderate GCS score (9–15) there were 537 head injury deaths. In patients with a severe GCS score (3–8) there were 918 head injury deaths, excluding those with a GCS score of 3 and those with unreactive pupils.
Secondary outcomes
In patients randomised within 3 hours of injury, the RRs for non-head injury deaths and for all-cause mortality were 1.31 (95% CI 0.93 to 1.85; 75 vs. 56 events) and 0.96 (95% CI 0.89 to 1.04; 930 vs. 948 events), respectively. The results for non-head injury deaths broken down by cause and all-cause mortality in all patients randomised are presented in Table 4.
| Cause of death | TXA group, N = 6359 | Placebo group, N = 6280 | RR (95% CI) | ||
|---|---|---|---|---|---|
| n | % | n | % | ||
| Bleeding | 9 | 0.1 | 7 | 0.1 | 1.27 (0.47 to 3.41) |
| PE | 9 | 0.1 | 7 | 0.1 | 1.27 (0.47 to 3.41) |
| Stroke | 10 | 0.2 | 4 | 0.1 | 2.47 (0.77 to 7.87) |
| MI | 9 | 0.1 | 3 | 0.0 | 2.96 (0.80 to 10.94) |
| Multiorgan failure | 27 | 0.4 | 24 | 0.4 | 1.11 (0.64 to 1.92) |
| Aspiration/pneumonia | 30 | 0.5 | 34 | 0.5 | 0.87 (0.53 to 1.42) |
| Sepsis | 9 | 0.1 | 6 | 0.1 | 1.48 (0.53 to 4.16) |
| Cervical spine injury | 3 | 0.0 | 4 | 0.1 | 0.74 (0.17 to 3.31) |
| Other | 16 | 0.3 | 11 | 0.2 | 1.44 (0.67 to 3.09) |
| Any cause | 1262 | 0.2 | 1298 | 0.2 | 0.96 (0.90 to 1.03) |
We assessed the effect of TXA on disability in survivors by comparing the mean DRS score (lower score means less disabled) between the TXA and placebo groups. The scores were similar between groups for patients treated within 3 hours of injury (mean = 4.99, SD = 7.6, for TXA group, vs. mean = 5.03, SD = 7.6, for placebo group) and for those treated after 3 hours of injury (mean = 4.52, SD = 7.0 for TXA group, vs. mean = 5.00, SD = 7.4 for placebo group). We also examined the effect of TXA on disability (Table 5) using an outcome measure designed by patient representatives by estimating the RR of being in the most extreme category for six areas of functioning: (1) walking, (2) washing, (3) pain and discomfort, (4) anxiety or depression, (5) agitation or aggression and (6) fatigue. The prevalence of disability among survivors was similar in the TXA and placebo groups.
| < 3 hours | ≥ 3 hours | All | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TXA (N = 4613) | Placebo (N = 4514) | RR (95% CI) | TXA (N = 1746) | Placebo (N = 1766) | RR (95% CI) | TXA (N = 6359) | Placebo (N = 6280) | RR (95% CI) | |||||||
| n | % | n | % | n | % | n | % | n | % | n | % | ||||
| Patient-derived disability measuresa | |||||||||||||||
| Confined to bed | 579 | 12.6 | 549 | 12.2 | 1.03 (0.93 to 1.15) | 190 | 10.9 | 222 | 12.6 | 0.87 (0.72 to 1.04) | 769 | 12.1 | 771 | 12.3 | 0.99 (0.90 to 1.08) |
| Unable to wash or dress | 580 | 12.6 | 583 | 12.9 | 0.97 (0.87 to 1.08) | 195 | 11.2 | 228 | 12.9 | 0.87 (0.72 to 1.04) | 775 | 12.2 | 811 | 12.9 | 0.94 (0.86 to 1.03) |
| Extreme pain or discomfort | 38 | 0.8 | 29 | 0.6 | 1.28 (0.79 to 2.08) | 10 | 0.6 | 10 | 0.6 | 1.01 (0.42 to 2.42) | 48 | 0.8 | 39 | 0.6 | 1.22 (0.80 to 1.85) |
| Extreme anxiety or depression | 43 | 0.9 | 41 | 0.9 | 1.03 (0.67 to 1.57) | 19 | 1.1 | 20 | 1.1 | 0.96 (0.51 to 1.79) | 62 | 1.0 | 61 | 1.0 | 1.00 (0.71 to 1.43) |
| Extreme agitation or aggression | 53 | 1.1 | 53 | 1.2 | 0.98 (0.67 to 1.43) | 14 | 0.8 | 27 | 1.5 | 0.52 (0.28 to 1.00) | 67 | 1.1 | 80 | 1.3 | 0.83 (0.60 to 1.14) |
| Extreme fatigue | 100 | 2.2 | 101 | 2.2 | 0.97 (0.74 to 1.27) | 40 | 2.3 | 43 | 2.4 | 0.94 (0.61 to 1.44) | 140 | 2.2 | 144 | 2.3 | 0.96 (0.76 to 1.21) |
| Complicationsb | |||||||||||||||
| All vascular occlusive events | 69 | 1.5 | 60 | 1.3 | 1.13 (0.80 to 1.59) | 32 | 1.8 | 42 | 2.4 | 0.77 (0.49 to 1.21) | 101 | 1.6 | 102 | 1.6 | 0.98 (0.74 to 1.28) |
| PE | 18 | 0.4 | 18 | 0.4 | 0.98 (0.51 to 1.88) | 6 | 0.3 | 14 | 0.8 | 0.43 (0.17 to 1.13) | 24 | 0.4 | 32 | 0.5 | 0.74 (0.44 to 1.26) |
| DVT | 15 | 0.3 | 12 | 0.3 | 1.22 (0.57 to 2.61) | 4 | 0.2 | 4 | 0.2 | 1.01 (0.25 to 4.04) | 19 | 0.3 | 16 | 0.3 | 1.17 (0.60 to 2.28) |
| Stroke | 29 | 0.6 | 23 | 0.5 | 1.23 (0.71 to 2.13) | 17 | 1.0 | 19 | 1.1 | 0.90 (0.47 to 1.74) | 46 | 0.7 | 42 | 0.7 | 1.08 (0.71 to 1.64) |
| MI | 9 | 0.2 | 12 | 0.3 | 0.73 (0.31 to 1.74) | 9 | 0.5 | 8 | 0.5 | 1.14 (0.44 to 2.94) | 18 | 0.3 | 20 | 0.3 | 0.89 (0.47 to 1.68) |
| Renal failure | 73 | 1.6 | 56 | 1.2 | 1.28 (0.90 to 1.80) | 27 | 1.5 | 28 | 1.6 | 0.98 (0.58 to 1.65) | 100 | 1.6 | 84 | 1.3 | 1.18 (0.88 to 1.57) |
| Sepsis | 297 | 6.4 | 279 | 6.2 | 1.04 (0.89 to 1.22) | 114 | 6.5 | 133 | 7.5 | 0.87 (0.68 to 1.10) | 411 | 6.5 | 412 | 6.6 | 0.99 (0.86 to 1.12) |
| Seizure | 130 | 2.8 | 105 | 2.3 | 1.21 (0.94 to 1.56) | 76 | 4.4 | 81 | 4.6 | 0.95 (0.70 to 1.29) | 206 | 3.2 | 186 | 3.0 | 1.09 (0.90 to 1.33) |
| Gastrointestinal bleeding | 16 | 0.3 | 22 | 0.5 | 0.71 (0.37 to 1.35) | 8 | 0.5 | 13 | 0.7 | 0.62 (0.26 to 1.50) | 24 | 0.4 | 35 | 0.6 | 0.68 (0.40 to 1.14) |
Adverse events
The risk of vascular occlusive events and other complications was similar in the TXA and placebo groups (see Table 5). There was no evidence that TXA increased fatal or non-fatal stroke (RR = 1.08, 95% CI 0.71 to 1.64). The risk of seizures was similar between groups (RR = 1.09, 95% CI 0.90 to 1.33). The numbers of other adverse events were similar between groups (see Appendix 5, Table 18).
Unblinding
Clinicians requested unblinding of the treatment allocation for 13 patients after randomisation for the following reasons: TXA became indicated after randomisation, n = 7; clinical management depended on knowing the treatment allocation, n = 3; patient requested unblinding, n = 1; required for suspected unexpected serious adverse reaction reporting, n = 1; unblinded in error, n = 1.
Forty patients received TXA in addition to the trial treatment after randomisation. In 36 cases, this was because the clinician believed that it was clinically indicated, and in four cases it was given in error instead of the trial drug.