Background and Objectives Usage of oral anticoagulants (OACs) or adenosine diphosphate inhibitors (ADPi) is known to increase the risk of bleeding. We aimed to investigate the impact of OAC and ADPi therapies on short-term outcomes after traumatic brain injury (TBI).
Methods All adult patients hospitalized for TBI in Finland during 2005–2018 were retrospectively studied using a combination of national registries. Usage of pharmacy-purchased OACs and ADPi at the time of TBI was analyzed with the pill-counting method (Social Insurance Institution of Finland). The primary outcome was 30-day case-fatality (Finnish Cause of Death Registry). The secondary outcomes were acute neurosurgical operation (ANO) and admission duration (Finnish Care Register for Health Care). Baseline characteristics were adjusted with multivariable regression, including age, sex, comorbidities, skull or facial fracture, OAC/ADPi treatment, initial admission location, and the year of TBI admission.
Results The study population included 57,056 persons (mean age 66 years) of whom 0.9% used direct OACs (DOACs), 7.1% vitamin K antagonists (VKA), and 2.3% ADPi. Patients with VKAs had higher case-fatality than patients without OAC (15.4% vs 7.1%; adjusted hazard ratio [aHR] 1.35, CI 1.23–1.48; p < 0.0001). Case-fatality was lower with DOACs (8.4%) than with VKAs (aHR 0.62, CI 0.44–0.87; p = 0.005) and was not different from patients without OACs (aHR 0.93, CI 0.69–1.26; p = 0.634). VKA usage was associated with a higher neurosurgical operation rate compared with non-OAC patients (9.1% vs 8.3%; adjusted odds ratio 1.33, CI 1.17–1.52; p < 0.0001). There was no difference in operation rate between DOAC and VKA. ADPi was not associated with case-fatality or operation rate in the adjusted analyses. VKAs and DOACs were not associated with longer admission length compared with the non-OAC group, whereas the admissions were longer in the ADPi group compared with the non-ADPi group.
Discussion Preinjury use of VKA is associated with increases in short-term mortality and in need for ANOs after TBI. DOACs are associated with lower fatality than VKAs after TBI. ADPi were not independently associated with the outcomes studied. These results point to relative safety of DOACs or ADPi in patients at risk of head trauma and encourage to choose DOACs when oral anticoagulation is required.
Classification of Evidence This study provides Class II evidence that among adults with TBI, mortality was significantly increased in those using VKAs but not in those using DOACs or ADPi.
- adenosine diphosphate inhibitor;
- adjusted HR;
- adjusted OR;
- acute neurosurgical operation;
- Anatomical Therapeutic Chemical;
- direct oral anticoagulant;
- factor Xa inhibitor;
- hazard ratio;
- International Classification of Diseases, 10th Revision;
- interquartile range;
- oral anticoagulation;
- odds ratio;
- traumatic brain injury;
- vitamin K antagonist
Traumatic brain injury (TBI) in patients using oral anticoagulation (OAC) and adenosine diphosphate inhibitors (ADPi) has become increasingly common as the population ages.1,–,3 Recent studies have shown direct oral anticoagulants (DOACs) to be noninferior and in some cases superior to vitamin K antagonists (VKA) in reducing the risk of ischemic stroke in atrial fibrillation.4,–,6 DOACs are also associated with better long-term outcomes compared with VKAs in patients with atrial fibrillation and ischemic stroke.7 In patients without atrial fibrillation, ADPi seem superior to aspirin in secondary prevention of ischemic events, especially long-term.8,9
Patients with TBI who received OACs and ADPi before injury are at high risk for traumatic intracranial hemorrhage.10,11 The incidence for intracranial hemorrhage in patients with mild TBI and OACs seems to be higher than previously thought.12 Mild TBIs are common in older patients13,14 who are at higher risk for atrial fibrillation and cardioembolic events. However, many studies that have examined the risk of intracranial hemorrhage and outcome in patients with TBI have not distinguished between patients taking OACs and ADPi.15,–,17
In a recent systematic review and meta-analysis, preinjury OACs were associated with significantly higher overall and in-hospital mortality, but not with the need for acute neurosurgical operations (ANOs) compared with TBI patients without OACs.18 Mortality rates and the need for ANOs seem to be quite similar between VKAs and DOACs after TBI,19 although conflicting results in favor of DOACs have been presented.20,21 ADPi do not seem to be associated with increased mortality after TBI in single-center studies.22,23
The current literature remains nebulous due to the lack of large-scale studies in outcome of TBI in patients on OACs and ADPi. Furthermore, given the growing aging population at increased risk for TBI and the increasing prescription of agents that affect blood clotting, it is important to have clear indications for different OACs and ADPi for this subset of patients. Herein, we used a large nationwide cohort to examine whether OAC and ADPi therapies have differential impact on short-term outcomes, primarily mortality, after TBI.
All ward admissions with TBI (International Classification of Diseases, 10th Revision [ICD-10] codes S06.* as the primary diagnosis) for patients aged 18 years or older in Finnish hospitals and healthcare ward units between January 1, 2005, and December 31, 2018, were retrospectively collected from the Care Register for Health Care. This mandatory by law database held by the National Institute for Health and Welfare, Helsinki, Finland, captures all healthcare ward discharges in Finland and includes information on performed surgical operations. Transfers between and within healthcare providers related to a particular admission episode were combined as 1 admission. The study included data from 338 hospitals or other healthcare units treating TBI, 5 of which provide neurosurgery services. The first admission of each patient during the study period was included. Patients with operated chronic subdural hematoma (ICD-10 codes S06.5 or I62.0 with operation for evacuation of chronic subdural hematoma [operational codes AAD10 or AAD12]) were not included in the study. Patients with missing follow-up data (n = 485) were excluded. The primary outcome of interest was death within 30 days after TBI admission. The secondary outcomes were an ANO and admission duration.
Validity of the ICD-10 TBI codes was studied by reviewing patient records of randomly selected patients corresponding to our inclusion criteria with S06.* as the primary discharge diagnosis admitted to the Turku University Hospital, Turku, Finland, and the North Karelia Central Hospital, Joensuu, Finland. Of 300 reviewed patients, 294 patients fulfilled the diagnostic criteria for S06.* resulting in a positive predictive value of 0.98.
Pharmacy purchases of anticoagulation and ADPi medications within 90 days before TBI admission were recognized using Anatomical Therapeutic Chemical (ATC)-codes (eTable 1, links.lww.com/WNL/C152). OACs and ADPi are only available from pharmacies by prescription in Finland, while aspirin is available as over the counter medication. The daily pill-counting method based on standardized defined daily dose (eTable 1) was used to estimate the usage of prescribed medication at the day of TBI. OAC therapy was classified as DOAC or VKA based on the latest OAC purchase, and ADPi therapy was classified as clopidogrel or prasugrel/ticagrelor based on the latest ADPi purchase. Patients with purchase of parenteral anticoagulants without OAC purchase were excluded (n = 421). Patients who purchased OAC within 90 days before TBI, but in whom the duration of purchased OACs did not cover the day of TBI were excluded (Figure 1). Skull or facial fracture was defined as ICD-10 diagnosis S02.* as codiagnosis. Comorbidities were detected from the combination of included registries.24 ANOs were detected as previously defined.25 Admission duration included ward and hospital transfers.
Standard Protocol Approvals, Registrations, and Patient Consents
The CRHF Registry and Finnish Cancer Registry data were obtained from the National Institute for Health and Welfare of Finland/Findata (Permission No.: THL/2245/5.05.00/2019). Fatality data were obtained from a nationwide cause of death registry held by Statistics Finland (Permission No.: TK-53-484-20). Prescription medication purchase data (including ATC codes, strength and amount, and purchase dates) and drug reimbursement permission data were obtained from the Social Insurance Institution of Finland (Permission No.: 91/522/2015). The collection and reporting of data within the included registries are mandated by law; therefore, the data from these registries provide a full picture of the Finnish population. Follow-up data were complete for all included patients. The requirements for permissions from individual hospital review boards and informed consent were waived by the law and participants were not contacted due to the nationwide registry study setting. The data underlying this article were provided by the Findata by permission. The legal basis for processing personal data is public interest and scientific research (EU General Data Protection Regulation 2016/679, Article 6(1)(e) and Article 9(2)(j); Data Protection Act, Sections 4 and 6).
Baseline features were analyzed with the χ2 test or analysis of variance as appropriate. Case fatality was analyzed with the Kaplan-Meier method, log-rank test, and adjusted Cox-regression. Proportional hazard assumptions were examined with Schoenfeld residuals. ANOs were analyzed with logistic regression. Admission duration was analyzed with linear regression (log-transformed and standardized dependent variable). Regression analyses were adjusted with age, sex, comorbidities listed in Table 1 (except for atrial fibrillation), skull or facial fracture, OAC/ADPi treatment, and initial admission location and stratified by the year of TBI admission. The potential modulative influence of combined OAC and ADPi therapies was studied with interaction analysis in adjusted regression models. Potential interaction of age (≤65 or >65 years) with association between OAC and ADPi and case-fatality was also studied. The results are given as the mean, median, percentage, hazard ratio (HR), or odds ratio (OR) with a 95% CI, interquartile range (IQR), or ±SD. Statistical significance was inferred at p value <0.05. Analyses were performed with SAS version 9.4 (SAS Institute, Inc., Cary, NC).
The data are available from Findata (findata.fi) by permission.
The total study population included 57,056 patients with TBI, of whom 0.9% were treated with DOAC and 7.1% with VKA. Warfarin was the only oral VKA used by the study patients. The mean age of all study patients was 66.0 years (SD 21.3, range 18–105), and 55.6% were male. Patients with OAC were older, were more often female, and had higher comorbidity burden compared with TBI patients without OAC (Table 1). An initial admission location was more frequently local health center than hospital with surgical capability in patients with OAC treatment. Patients receiving DOAC were more often female and more likely to have a history of alcohol abuse than patients receiving VKA (Table 1). The most common recorded primary TBI diagnosis was concussion/mild TBI (eTable 2, links.lww.com/WNL/C153). There was no difference in the proportion of skull or facial fractures between the OAC regimens. Of 527 patients with DOAC, 125 (23.7%) used dabigatran and 402 (76.3%) used factor Xa inhibitors (FXais; apixaban, edoxaban, and rivaroxaban). An ADPi was used by 2.3% of the study patients. Clopidogrel was used by 95.8% and ticagrelor or prasugrel by 4.2% of the ADPi users. All patients receiving both OAC and ADPi were taking clopidogrel. Patients taking ADPi were older and had a higher rate of comorbidities than patients not taking ADPi (Table 2). Patients taking either OAC or ADPi had a lower frequency of skull or facial fractures than patients not taking OAC (Table 1) or ADPi (Table 2).
Of all patients with TBI, 4,369 died within 30 days. The 30-day case-fatality rate was 8.4% in the DOAC group, 15.4% in the VKA group, and 7.1% in the non-OAC group (Figure 2). Case-fatality in patients treated with DOAC did not differ from that in patients not treated with OAC in either the nonadjusted (p = 0.263) or the multivariable adjusted analysis (adjusted HR [aHR] 0.93; CI 0.69–1.26; p = 0.634). Patients treated with VKA had a higher risk of dying within 30 days after TBI compared with non-OAC patients in both the nonadjusted (p < 0.0001) and multivariable adjusted analyses (aHR 1.35; CI 1.23–1.48; p < 0.0001). Patients treated with DOAC had lower case-fatality than patients treated with VKA in both the unadjusted (p < 0.0001) and adjusted analyses (aHR 0.62; CI 0.44–0.87; p = 0.005). The case-fatality was 6.4% in the dabigatran group and 9.7% in the FXai group (nonadjusted p = 0.048) with an aHR of 0.33 (CI 0.11–0.94; p = 0.039). The 30-day case-fatality rate was 10.2% in the ADPi group and 7.6% in the non-ADPi group (nonadjusted p = 0.0004) (Figure 3). ADPi were not associated with case-fatality in the multivariable model (aHR 0.89; CI 0.74–1.06; p = 0.194). There were no interactions between OAC and ADPi therapies (interaction p = 0.524) or between age and OAC (interaction p = 0.905) or ADPi (interaction p = 0.270) in analysis of case-fatality. Case-fatality was 9.4% in patients taking clopidogrel and 20.0% in patients taking prasugrel/ticagrelor (nonadjusted p = 0.008; aHR 1.74; CI 0.84–3.60; p = 0.134). The underlying cause of death was determined to be external in 81.1% and disease in 18.9% of the deceased patients, with falls being the most common mechanism of trauma (eTable 3, links.lww.com/WNL/C154). Underlying causes of death did not differ between study groups. Association of baseline features with case-fatality is presented in eTable 4 (links.lww.com/WNL/C155). Of comorbidities, coagulopathy, liver disease, and heart failure were associated with the highest risk of case-fatality. Patients treated in the university hospitals had over 5-fold risk of death compared with patients treated in the health centers.
Acute Neurosurgical Operations
ANO was performed to 4,770 patients with TBI. The ANO rate was 4.7% in the DOAC group, 9.1% in the VKA group, and 8.3% in the non-OAC group (nonadjusted p = 0.003). Patients with VKA had higher odds for ANO than patients without OAC (adjusted OR [aOR] 1.33; CI 1.17–1.52; p < 0.0001). Adjusted odds for ANOs did not differ between the DOAC vs VKA groups (aOR 0.90; CI 0.55–1.47; p = 0.126) or between the DOAC vs non-OAC groups (aOR 1.08; CI 0.70–1.66; p = 0.110). Of patients using dabigatran, 4.8% and, of those using an FXai, 4.7% underwent ANO (aOR 1.15; CI 0.36–3.73; p = 0.300). ANOs were performed in 6.5% of the patients treated with ADPi and in 8.4% of the patients not treated with ADPi (aOR 0.93; CI 0.73–1.18; p = 0.542). There was no interaction between OAC and ADPi therapies (interaction p = 0.739). The rate of surgery was 6.2% in patients treated with clopidogrel and 12.7% in patients treated with prasugrel/ticagrelor (aOR 2.31; CI 0.79–6.77; p = 0.128).
The median length of all TBI admissions was 4 days (IQR 2–11 days). The median duration was 4 days (IQR 2–11) in the DOAC group, 5 days (IQR 2–15) in the VKA group, and 3 days (IQR 2–10) in the non-OAC group. Admission duration did not differ between patients treated with DOAC and patients without OAC in nonadjusted (p = 0.328) or multivariable adjusted (p = 0.354) analyses. Treatment with VKA was associated with longer admission in the nonadjusted analysis (p < 0.0001), but not after multivariable adjustments (p = 0.738). The median duration was 5 days (IQR 2–12) in the ADPi group and 4 days (IQR 2–10) in the non-ADPi group with p < 0.0001 in the nonadjusted and p = 0.0003 in the adjusted analyses. There was no interaction between OAC and ADPi therapies (interaction p = 0.182).
Classification of Evidence
This study provides Class II evidence that among adults with TBI, mortality was significantly increased in those using VKAs but not in those using DOACs or ADPi.
This was a nationwide population-based study examining the impact of OAC and ADPi therapies on short-term outcome after TBI. The main findings from the adjusted analyses are (1) case-fatality was higher with VKA than without OACs, (2) case-fatality was lower with DOACs than with VKA, (3) case-fatality was similar with DOACs and without OACs, (4) case-fatality was similar with ADPi and without ADPi, (5) OR for ANOs was higher with VKA than without OACs, (6) OR for ANOs was similar with DOACs and without OACs, and (7) OR for ANOs was similar with ADPi and without ADPi. Other findings were (1) in DOACs, case-fatality was higher with FXais than dabigatran and (2) case-fatality was similar between ADPi types.
The global population ages: In 2015, people aged 60 years or older made up 12% of the world’s population, and this number is expected to reach 22% by 2040. In Europe, 24% of the population is already aged 60 years or older and that proportion is projected to reach 34% in 2050.26 Older adults are at higher risk for atrial fibrillation and cardioembolic events, but they are also at higher risk for TBIs.13,14,27 The major concern in patients taking medications that affect blood clotting who suffer a TBI is the increased risk of intracranial hematoma expansion and mortality. Studies examining outcome after mild TBI in patients with DOACs and VKAs reported lower mortality associated with DOAC use. In these 2 studies, patients did not have major intracranial hemorrhage and patients receiving DOACs did not have reversal of anticoagulation, whereas most patients receiving VKAs were treated with vitamin K.28,29
Regarding TBIs of all severities, the evidence is not so consistent. Two recent prospective studies and 1 retrospective study reported that VKA and DOAC users had comparable mortality rates after a TBI.19,22,30 In another recent study that included patients with all severities of TBI, there was no statistically significant difference in the expansion rates of intracranial hematomas between VKAs and DOACs. Nonetheless, DOAC-treated patients had better outcomes compared with patients receiving VKAs despite the low use of reversal strategies.31 Conflicting results have also been reported. A single-center study reported a higher rate of hematoma expansion in patients with traumatic intracerebral hematomas, higher rate in need for ANOs, and higher mortality on DOACs compared with VKAs.32 However, a recent systematic review and meta-analysis including 2,622 patients of whom 239 were on DOACs and 524 on VKA found that in-hospital mortality and the need for ANOs did not differ between patients with TBI on DOACs and VKAs. The authors noted that within each study, surgery rates, reversal agents used, hematoma progression, and in-hospital mortality differed significantly between DOAC and VKA cohorts.21 A recent study published after the above-mentioned review article reported that OAC use was associated with enlargement of intracranial hematomas, whereas the use of antiplatelet agents was not when compared with patients without medications affecting blood clotting.33
Our large-scale study of 57,056 patients with TBI, of whom 527 patients received DOACs and 4,053 patients received VKAs, brings clarity to the issue: Patients who received VKAs before the injury have twice the mortality rate of patients who received DOACs, who in turn have a similar mortality rate to patients who did not take OACs. Dabigatran was found to be associated with lower mortality than FXais. It is noteworthy that in a Norwegian study examining risk of intracranial hemorrhage associated with antithrombotic therapy, it was observed that all other antithrombotic therapies excluding dabigatran were associated with increased risk of hemorrhage.34 In this study, patients receiving VKA had higher odds for ANO than patients without OAC. However, we found no OR differences for ANOs between the DOAC vs VKA groups or between the DOAC vs non-OAC groups. The current results are largely consistent with previous findings on the impact of ADPi on mortality after TBI.17,22,23 We also found that the OR for ANOs was similar in patients who received ADPi before injury and patients without ADPi. In addition, we did not find any interaction between the impact of OAC and ADPi therapies on mortality and ANOs. Intriguingly, however, VKAs and DOACs were not associated with longer admission length compared with the non-OAC group, whereas the admissions were longer in the ADPi group compared with the non-ADPi group.
We used the need for ANOs and admission lengths as surrogate markers of TBI severity. When examining factors associated with case-fatality, we identified many comorbidities but also treatment at the university hospitals compared with health centers. This reflects the fact that most severe TBI cases are treated in centers where neurosurgical services are available.
We observed that patients who received DOAC were more often women. In another recent study, it was reported that women outnumbered men in patients who received DOAC and suffered a TBI.35 The observation in this study may be related to the results of our previous study, in which we found that the number of TBIs among elderly women in Finland increased during partially the same years as in this study. In addition, DOACs were introduced into general practice around the same time as the CHA2DS2VASc score,36 which gives an additional risk point to women. This may have influenced the gender distribution of patients with DOAC in Finland.
There are several strengths in this study. We have included all Finnish hospitals that provide facilities for the TBI patient follow-up and all tertiary care hospitals providing neurosurgical services. We used multiple national registries that provide a full picture of the Finnish population. The collection and reporting of data within the included registries are mandated by law. An important limitation is the possibility of selection bias because we included only admitted patients with TBI. Patients with OAC or ADPi may be more likely to be admitted to ward for monitoring after minor head trauma than patients without OAC, possibly resulting in less severe injuries in the treated groups. This could dilute the observed results, notwithstanding the fact that adjustments were made for skull or facial fractures. However, it is unlikely that this potential bias would affect the DOAC and VKA groups differently. Actual medication usage was not confirmed separately; we used a standard method of prescription medication purchase to study usage.37 Aspirin is mostly used as over the counter drug, and we could not assess its usage in the study population. We did not assess the use of OAC reversal strategies. However, the favorable impact of DOACs on mortality after TBI remains strong because (1) the Finnish national TBI care guideline advocates the use of vitamin K for VKA reversal,38 (2) the antidote idarusitsumab for dabigatran was not authorized in Finland until late 2015, and (3) the antidote andexanet alfa for FXA was authorized after the end of the current data collection in 2019. Because of the limitations of the ICD-10 coding system, TBIs with hemorrhagic intracranial lesions cannot be reliably distinguished from diffuse and nonhemorrhagic TBIs—especially in cases with multiple lesion types on head CT.
As a final limitation of the study, we must acknowledge the lack of data examining stroke rates in different groups of patients with TBI. The use of OACs or ADPi is a compromise between thrombosis and bleeding. The majority of the study patients on OACs had atrial fibrillation and a number of other known risk factors for stroke. In patients with risk factors for thrombosis and thromboembolic events, the use of OACs and ADPi is advocated, but the current results highlight the importance of preventive measures to reduce the risk of head trauma, particularly in patients taking OACs.
We found that VKA use is associated with increased short-term mortality and need for ANOs after TBI. DOACs are associated with lower fatality than VKAs after TBI, and case-fatality is similar with DOACs and without OACs. ADPi are not associated with mortality or need for ANOs after TBI. The results reinforce the safety of DOACs and ADPi in patients at risk of head trauma and encourage the use of DOACs when oral anticoagulation is required.
J.P. Posti has received funding from Academy of Finland (Grant No. 17379) and the Maire Taponen Foundation. J.P. Posti, V. Kytö, and P. Rautava received funding from Competitive State Research Financing of the Expert Responsibility area of Turku University Hospital, Finland. T.M. Luoto has received funding from the Competitive State Research Financing of the Expert Responsibility area of Tampere University Hospital, Finland, Finnish Brain Foundation sr, the Emil Aaltonen Foundation sr, the Maire Taponen Foundation, the Science Fund of the City of Tampere, and the Finnish Medical Society Duodecim. V. Kytö has also received a grant from the Paulo Foundation sr and the Finnish Foundation for Cardiovascular Research sr. No targeted funding reported for J.O. Ruuskanen and J.O.T. Sipilä.
J.O. Ruuskanen has received scientific consultancy fees (Merck and Sandoz), speaker fees (Merck, Biogen Idec, UCB Pharma, and Bayer), and travel grants and congress sponsorship (BMS, Sanofi-Genzyme, TEVA, and Bayer). P. Rautava has received speaker fee and travel grant (Roche Oy). J.O.T Sipilä has received honoraria (Merck, Pfizer, and Sanofi), a consultancy fee (Rinnekoti Foundation), travel grants and congress sponsorship (Abbvie, Orion Pharma, Merck Serono, Sanquin, Lundbeck, and Novartis), and holds shares (Orion Corporation). V. Kytö has received a scientific consultancy fee (AstraZeneca), speaker fee (Bayer, Astra-Zeneca, and Boehringer Ingelheim), and travel grants and congress sponsorship (AstraZeneca, Boehringer Ingelheim, Bayer, and Pfizer). J.P. Posti and T.M. Luoto report no relevant disclosures. Go to Neurology.org/N for full disclosures.
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
Submitted and externally peer reviewed. The handling editor was Rebecca Burch, MD.
Class of Evidence: NPub.org/coe
- Received November 22, 2021.
- Accepted in final form April 22, 2022.
- © 2022 American Academy of Neurology