Characteristics, management and outcomes of adults with major trauma taking pre-injury warfarin in a Western Australian population from 2000 to 2005: a population-based cohort study

Trauma patients anticoagulated with warfarin are theoretically at risk of prolonged major bleeding. Studies of traumatic intracranial haemorrhage (ICH) suggest that patients taking anticoagulants have 2–6 times greater mortality.1-3 Those without ICH may not have increased mortality but they may have longer, more complex admissions.4-6 Patients taking anticoagulants seem to have no more traumatic ICH than age-matched controls, but bleeding, when it occurs, is more clinically apparent, with bleed size similar at presentation but with greater haematoma progression.2,7,8 Progressive bleeding is associated with high mortality in patients with traumatic coagulopathic ICH (35%–85%), severely disabled survivors and worse outcomes for initially awake patients with “minor” head injuries.2,7,9 Rapid, early warfarin reversal is recommended to improve outcomes.4,5,7,9,10

Authoritative guidelines strongly recommend rapid warfarin reversal for major bleeding, although there are few supportive prospective or randomised trial data.11-13 Australasian guidelines recommend urgent reversal of international normalised ratio (INR) to < 5.0, with further actions correlated with clinical bleeding.11 British and American guidelines suggest reversal of INR to < 2.0.12,13 Vitamin K (intravenous), prothrombin complex concentrates (Prothrombinex-VF [PTX], CSL Bioplasma, Melbourne, Vic), or fresh frozen plasma (FFP) are recommended for warfarin reversal in Australia.11

Evidence shows that practice varies significantly from published guidelines. In Italian emergency departments, < 30% of patients with anticoagulant-associated traumatic ICH underwent reversal, and attempts were delayed.2,14 One small prospective study of reversal effectiveness in anticoagulant-associated trauma (n < 20) suggested that rapid diagnosis and reversal reduced mortality from 48% to 10%.9,14,15

We were prompted to undertake this study by the observation, made by two of us (D M and V S) when reviewing trauma deaths, that patients taking warfarin had very poor outcomes and were over-represented among trauma deaths. It was also observed that warfarin reversal attempts were often delayed or ineffective. Our review of the literature, which found only a few small studies, suggested that our observations may not be unusual, but gave limited evidence for decision making and almost no Australasian data.2,14,16

We undertook an exploratory observational study of Australasian patients with major trauma who were taking warfarin. The aims of our study were to characterise the Western Australian major trauma population taking warfarin (2000–2005); describe warfarin reversal practice compared with guidelines; and study whether rapid reversal was associated with improved survival.

Our study is among the largest published cohort studies of major trauma in patients taking warfarin. It is also among the largest to use data that are prospective and of consecutive patients from an isolated population. The observed demographic characteristics of the study sample — advanced age; nearly equal sex distribution; taking warfarin most commonly for atrial fibrillation and embolic disease; high prevalence of traumatic ICH; and mainly therapeutic INRs — are similar to most previous studies.1,5,9,10,18 The estimated incidence of major trauma for patients taking warfarin in WA was low, at 0.7 per 100 000 population per year, but it is probably underestimated because the numbers of deaths on-scene, before transfer to hospital or where transfer was inappropriate, were not included in the data. Morbidity and mortality were high, even without ICH (unlike some reports).4-6 The 35% mortality rate (28/80) is higher than that found in a previous study (9.4%), which included trauma of any severity.4 The traumatic ICH mortality of 40% is comparable to that found in other studies (23%–85%).1-5,9,10,18 The severe morbidity after traumatic ICH for patients taking warfarin is demonstrated by the functional status being worse at discharge than before injury in 74% of patients (although some might improve with rehabilitation); this finding is comparable to the 68.1% of patients with decreased function found in another study.2 There were long delays from injury to initiating reversal (mean, 17.4 hours). The longest delays occurred at transferring hospitals, but there were still significant delays at trauma hospitals (5 hours). Delays were associated with late presentations, delayed transfers and slow initiation or lack of reversal agents at referring facilities. Although 5 hours between patient arrival and the start of reversal at a trauma hospital seems long, similar delays are reported for overseas tertiary centres.2,9

Our study did not show survival benefit for early initiation and completion of warfarin reversal on multivariate analysis, although it was present on univariate analysis. However, only 17 patients (13 with ICH) started reversal under 4 hours from injury and only eight (six with ICH) under 2 hours. Documented completion of reversal under 4 hours from injury was very rare and inadequate dosing regimens were common (Box 5). It has been suggested that initiating reversal under 2 hours may improve survival (from 10% to 48%) in traumatic ICH.9 Another study documented that progressive bleeding mainly occurs under 6 hours from injury.8 As only six patients with ICH started reversal under 2 hours in our study, a significant mortality reduction was unlikely to be seen even if present. Reversal testing and documentation were probably delayed and, therefore, reversal for some patients was probably achieved more quickly than recorded. However, the lack of early repeat testing of INRs also reflects a non-urgent approach to reversal. The literature and guidelines are unclear about what constitutes rapid reversal.11-13 We suggest early reversal should be defined as complete reversal by 4 hours from injury or 2 hours from hospital arrival. This needs to be investigated further.

Current Australasian guidelines recommend early reversal with combined therapy (including PTX) and continuous assessment until an INR < 5.0 is achieved and bleeding stops.11 Use of the recommended regimen should achieve complete reversal in many patients. Unfortunately, the guidelines do not clearly deal with the more common situation of life-threatening bleeding with therapeutic INRs (the situation for approximately three-quarters of the patients in this study). The confusion around reversal decisions in this situation seems borne out by the major delays and inadequate interventions we found in our study. It would seem sensible to clarify and emphasise the need for urgent complete reversal in future updates of warfarin reversal guidelines.

Initial vitamin K doses were generally adequate (mean, 7.4 mg), although two patients had oral dosing.11 Mean FFP dosing was 2.9 units (approximately 725 mL), but median dose was 2 units (eg, fewer than half of the patients had the recommended 3 units). Initial dosing of PTX (mean, 1.5 vials; approximately 10 IU/kg) was well below recommended doses (25–50 IU/kg)11 and was used in only 13 patients. We do not have data on why PTX was used so infrequently. However, current guidelines were only formulated in 2004; anecdotally, advice from haematologists was variable and concerns about full reversal were commonly expressed because of perceived thromboembolic risks. Only eight patients had all three recommended reversal agents, but 47 had either FFP plus vitamin K or PTX plus vitamin K.11 Although often slow and inadequate,13-15 attempted reversal was more prevalent than that found in Italian emergency departments.4 We did not assess clinicians’ knowledge of reversal guidelines (or access to them), but it seems clear that implementation is patchy.

In our study, only two patients had thromboembolic events after reversal, substantially fewer than the number reported in another study (4/35).18 However, anticoagulation was recommenced for many patients after they were deemed stable and, for a third, reversal was either incomplete or not initiated. Even so, the low rate of thromboembolic events in our study suggests if reversal is indicated it should not be avoided because of thromboembolic risk, and anticoagulation should be reinstated only when bleeding stops and if still indicated.13-15

Previous studies have suggested outcome predictors for patients with anticoagulant-associated traumatic ICH. Increasing age strongly predicts death and we found a 22% decrease in survival chance per year.1,3,4 Similarly, GCS was strongly associated with survival, with each point increase about doubling survival odds.4,12 Another study showed that 91.5% of patients with a presentation GCS < 8 died,12 while we found that all patients with GCS < 8 died, suggesting low GCS may predict futility of treatment. Warfarin is strongly associated with traumatic ICH expansion, which probably mediates mortality.10,11 Progressive bleeding decreased patients’ survival chances profoundly. Twelve of the 36 patients with progressive bleeding had an initial GCS of 15, seven of whom died. Two of the seven patients who died did not undergo anticoagulation reversal, having been deemed neurologically stable until suddenly deteriorating. These data suggest that all patients with traumatic ICH and taking warfarin should be viewed as immediately life-threatened even when “stable”. It has been suggested that high INRs increased mortality but we did not find this association, with 64 patients having therapeutic INRs.3

A limitation of our study was our need to use some retrospective data, although patient selection and most data were from prospective high-quality registries. A lack of agreed uniform criteria in this area of research meant we defined data differently from some studies. Despite being a relatively large series compared with previous studies, the study’s overall numbers were small, so only very large clinical effects can be observed. We defined morbidity outcomes by functional status at hospital discharge because direct follow-up was made difficult by high expected mortality rates. Perth is an isolated city with distinctive regional problems, but the population is highly centralised, mainly metropolitan and has a mature system for trauma treatment. We used an INR of < 1.5 (rather than < 2.0) to define reversal because this is the level at which our neurosurgeons are willing to operate electively.

Further studies need larger populations, achievable either by expanding the study period (infeasible for prospective studies) or including more centres. Following up survivors at prespecified times to determine outcomes would allow better definition of medium-term dysfunction. Studies examining levels of knowledge among clinicians and availability of, and attitudes to, guidelines for anticoagulation reversal are required. Australasian guidelines should be more definitive about reversal and more widely promulgated. Larger studies may confirm whether the initial GCS should inform decisions to withdraw therapy. Prospective interventional before-and-after studies would allow assessment and confirmation of the utility and feasibility of rapid reversal protocols. The findings of our study, combined with those of other recent studies, could guide the design of such trials.

The incidence of major trauma in those taking warfarin in WA is low, but morbidity and mortality among those taking warfarin who suffer major trauma are high. Increasing age, low GCS and progressive ICH strongly predict mortality. In our study, clinicians did not follow guidelines when faced with the possible need to reverse the effects of warfarin. Use of reversal agents was often delayed or not attempted (particularly at peripheral hospitals), and reversal was often completed slowly, with inadequate dosing and inappropriate combinations of agents. In particular, initial doses of PTX were well below recommendations.13 Rapid initiation of warfarin reversal did not independently predict survival, but it was so rarely performed that potential survival benefits were unlikely to be seen.15 Reversal was associated with a low rate of thromboembolic events. Further prospective investigation into compliance with guidelines and their clinical effectiveness is needed.