Acute and Delayed Intracranial Hemorrhage in Head-Injured Patients on Warfarin versus Direct Oral Anticoagulant Therapy.

INTRODUCTION: Direct oral anticoagulant (DOAC) use for thrombosis treatment and prophylaxis is a popular alternative to warfarin. This study compares rates of traumatic intracranial hemorrhage (ICH) for patients on anticoagulant therapies and the effect of combined anticoagulant and antiplatelet therapies.
METHODS: A retrospective observational study of trauma patients was conducted at two level I trauma centers. Patients aged ≥18 years with preinjury use of an anticoagulant (warfarin, rivaroxaban, apixaban, or dabigatran) who sustained a blunt head injury within the past day were included. Patients were evaluated by head CT to evaluate for ICH.
RESULTS: Three hundred and eighty-eight patients were included (140 on warfarin, 149 on a DOAC, and 99 on combined anticoagulant and antiplatelet therapies). Seventy-nine patients (20.4%) had an acute ICH, while 16 patients (4.1%) had a delayed ICH found on routine repeat CT. Those on combination therapy were not at increased risk of acute ICH (relative risk [RR] 0.90, confidence interval [CI]: 0.56-1.44; P > 0.5) or delayed ICH (RR 2.19, CI: 0.84-5.69; P = 0.10) compared to anticoagulant use only. Those on warfarin were at increased risk of acute ICH (RR 1.75, CI: 1.10-2.78, P = 0.015), but not delayed ICH (RR 0.99, CI 0.27-3.59, P > 0.5), compared to those on DOACs. No delayed ICH patients died or required neurosurgical intervention.
CONCLUSION: Patients on warfarin had a higher rate of acute ICH, but not delayed ICH, compared to those on DOACs. Given the low rate of delayed ICH with no resultant morbidity or mortality, routine observation and repeat head CT on patients with no acute ICH may not be necessary. Copyright:

INTRODUCTION

Oral anticoagulant use has been increasing throughout the United States. The number of prescriptions for oral anticoagulants increased by 2.5 times from 2010 to 2017.[1] Direct oral anticoagulant (DOAC) use for thrombosis treatment and prophylaxis has become a popular alternative to warfarin due to their ease of use, safety profile, and efficacy.[12] DOACs comprised 46.5% of over 600,000 oral anticoagulant prescriptions in the United States during the fourth quarter of 2017.[1] This is of particular concern because patients prescribed these medications tend to be older and at higher risk of falls.[34]

Head injuries are a frequent cause of severe morbidity and mortality in geriatric patients, with ground level falls being the most common cause.[3] A third of geriatric patients experience a fall each year and have a reported mortality rate as high as 16%.[34] Both anticoagulant and antiplatelet medications are known to increase the risk of traumatic intracranial hemorrhage (ICH), though there is considerable variability in the literature regarding the rate of traumatic ICH in these patients. Previous studies of patients taking warfarin or DOACs who experienced mild traumatic brain injury have demonstrated rates of ICH between 4% and 22%.[56789101112]

Anticoagulated patients that experience head trauma are routinely evaluated for ICH in the emergency department (ED) with a computed tomography (CT) scan of the head. However, there is no clear consensus about how to appropriately disposition those whose initial CT scan of the head is normal. Some experts argue that normal initial imaging of the head is not sufficient to rule out ICH due to concern that impaired hemostasis may result in delayed ICH.[13] Numerous clinical guidelines and trauma center policies share this concern and recommend that all anticoagulated patients who experience head trauma be admitted to the hospital for observation and routine repeat CT scan of the head.[14151617] For this reason, several studies have attempted to quantify the risk of delayed ICH in patients taking anticoagulants, antiplatelets, or both. However, results have been disparate and sometimes even conflicting.[10111618192021]

Because of the variability in findings across the literature, our study attempts to better quantify the risk of both acute and delayed ICH in anticoagulated patients taking DOAC or warfarin presenting to a trauma center after suffering head trauma.

METHODS

This study was a multicenter retrospective observational study of head-injured trauma patients. Patients from two level-I trauma centers who presented between January 1, 2016 and December 31, 2017 were included. The first site located in Florida is a 36-bed ED with annual census of 70,000 patients and 463 inpatient beds. The second site located in Michigan is a 68-bed ED with annual census of 100,000 patients and 676 inpatient beds. Study approval was obtained by both hospitals' IRBs.

Data were obtained through a query of the trauma registry at each hospital. Inclusion criteria consisted of patients seen by the trauma team for any head trauma, isolated or as part of polytrauma, with any Glasgow Coma Scale (GCS) and preinjury use of an anticoagulant (defined as warfarin, rivaroxaban, apixaban, or dabigatran). Patients with age <18 years, no use of anticoagulant therapy, and those suffering head trauma >24 h before ED presentation were excluded. The study sample consisted of all patients who met these inclusion and exclusion criteria.

When evaluating head injury patients, routine trauma team workup at both study hospitals consisted of CT head imaging and blood tests, including platelet count and international normalized ratio (INR). A board-certified radiologist interpreted the head CT images at both institutions. Patients with negative workups were admitted based on the attending trauma surgeons' discretion for neurological observation and repeat head CT. Management of these patients was not standardized, though anticoagulant reversal agents were not routinely administered, anticoagulant medications were often continued, and repeat head CT was typically performed within 24–48 h.

A single investigator at each institution performed the chart reviews, utilizing a standardized data abstraction form containing: patient demographics, mechanism of injury, preinjury anticoagulant and antiplatelet use, labs (INR and platelet count), GCS, Injury Severity Score, head CT findings, neurosurgical intervention (treatment with mannitol or hypertonic saline, need for an intracranial pressure monitor, a subdural drain or intraventricular catheter, or a craniotomy/craniectomy), and mortality.

CT head findings were used to determine the study's primary outcome – presence of ICH (acute or delayed). Acute ICH was defined by the presence of an acute ICH on initial ED head CT. A delayed ICH was defined by the presence of an acute ICH on repeat head CT after an initially negative head CT. Mortality and need for neurosurgical intervention during the hospitalization were the study's secondary outcomes.

Risk ratios were calculated for acute and delayed ICH. Patients were analyzed in subgroups by anticoagulant type, comparing those on DOACs to warfarin. In a separate analysis, patients taking combined anticoagulant and antiplatelet (aspirin, clopidogrel, or ticagrelor) therapies were compared to patients taking only anticoagulants. Background characteristics of patients were compared between the various anticoagulant categories. If any data points were missing for a patient, they were excluded only from that specific analysis. Statistics were performed using Stata Release 16 (StataCorp, College Station, TX, USA).

RESULTS

Three hundred and eighty-eight patients met the study inclusion criteria: 140 were on warfarin only, 149 were on a DOAC only (62 on rivaroxaban, 73 on apixaban, and 14 on dabigatran), and 99 were on combined anticoagulant and antiplatelet therapies. Patients had similar baseline characteristics [Table 1]. There were no missing data for any of the independent or dependent variables.

Table 1

Baseline characteristics of patients on anticoagulants by anticoagulant category and antiplatelet use

	VKA (n=180)	DOAC (n=208)	No antiplatelet (n=289)	Yes antiplatelet (n=99)
Hospital, n (%)
Site 1	54 (30)	50 (24)	80 (28)	24 (24)
Site 2	126 (70)	158 (76)	209 (72)	75 (76)
Age (years), mean (SD)	78 (14)	77 (13)	78 (14)	77 (10)
Gender, n (%)
Male	101 (56)	108 (52)	150 (52)	59 (60)
Female	79 (44)	100 (48)	139 (48)	40 (40)
Ethnicity, n (%)
Caucasian	170 (94)	185 (89)	266 (92)	89 (90)
African American	7 (4)	9 (4)	14 (5)	2 (2)
Hispanic/Latino	2 (1)	6 (3)	5 (2)	3 (3)
Other	1 (1)	8 (4)	4 (1)	5 (5)
Domicile, n (%)
Private home	145 (81)	161 (77)	229 (79)	77 (78)
Nursing home	20 (11)	27 (13)	33 (11)	14 (14)
ALF/Rehab	15 (8)	18 (9)	25 (9)	8 (8)
Homeless	0 (0)	2 (1)	2 (1)	0 (0)
Mechanism of injury, n (%)
Fall	156 (87)	183 (88)	249 (86)	90 (91)
MVC	17 (9)	16 (8)	27 (9)	6 (6)
Bicycle accident	1 (1)	0 (0)	1 (0)	0 (0)
Motorcycle accident	1 (1)	0 (0)	0 (0)	1 (1)
Pedestrian versus MVC	1 (1)	3 (1)	4 (1)	0 (0)
Blunt injury to head	4 (2)	6 (3)	8 (3)	2 (2)
INR result, mean (SD)	2.9 (1.7)	1.1 (0.2)	2.0 (1.5)	1.9 (1.3)
Platelet count in thousands, mean (SD)	191 (62)	206 (68)	198 (67)	202 (64)
Injury severity score, mean (SD)	10 (20)	9 (12)	10 (17)	9 (10)

ALF: Assisted living facility, MVC: Motor vehicle crash, DOAC: Direct oral anticoagulant, SD: Standard deviation, VKA: Vitamin K antagonist

In total, 79 patients (20.4%; 95% confidence interval [CI]: 16–25) had an acute ICH, while 16 patients (4.1%; 95% CI: 2–7) had a delayed ICH. Those on warfarin had an increased rate of acute ICH compared to those on a DOAC [26% vs. 16%; relative risk (RR) 1.75, 95% CI: 1.10–2.78, P = 0.015; Table 2], but not delayed bleed [4% vs. 4%; RR 0.99, 95% CI 0.27–3.59, P > 0.5; Table 2]. Of those with delayed ICH, 9 (56%) were discovered within the first 24 h after injury, and another 7 (44%) in the next 24 h period – no delayed ICH was found after 48 h from the time of injury. Only one patient was found to have a delayed ICH (intraparenchymal hemorrhage) on a repeat head CT performed due to a clinical change, and this patient did not require any neurointervention. All other patients with delayed ICH were found on routine repeat CT (six subdural hemorrhages, five subarachnoid hemorrhages, three intraparenchymal hemorrhages, and one combined subdural and intraparenchymal hemorrhage). Of patients with an initial negative head CT, there was no difference in INR (2.07 vs. 1.92; difference = 0.15, 95% CI − 0.60–0.89, P = 0.699) or platelet count (199 vs. 201; difference = −2.8, 95% CI − 36.5–30.9, P = 0.870) between those who developed a delayed ICH and those who did not.

Table 2

Computed tomography results by anticoagulant type

Anticoagulant	No ICH	Acute ICH	Delayed ICH	Total
VKA	127 (71%; CI: 63- 77)	46 (26%; CI: 19- 33)	7 (4%; CI: 2- 8)	180
DOAC	166 (80%; CI: 74- 85)	33 (16%; CI: 11- 22)	9 (4%; CI: 2- 8)	208
Xarelto	68 (81%; CI: 71- 89)	15 (18%; CI: 10- 28)	1 (2%; CI: 0- 6)	84
Eliquis	85 (80%; CI: 71- 87)	15 (14%; CI: 8- 22)	6 (6%; CI: 2- 12)	106
Pradaxa	13 (72%; CI: 47- 90)	3 (17%; CI: 4- 41)	2 (11%; CI: 1- 35)	18

All CIs are 95%. CIs: Confidence intervals, ICH: Intracranial hemorrhage, DOAC: Direct oral anticoagulant, VKA: Vitamin K antagonist

Patients on combination anticoagulant and antiplatelet therapy were not at increased risk of acute ICH (RR 0.90, CI: 0.56–1.44; P > 0.5) or delayed ICH (RR 2.19, CI 0.84–5.69; P = 0.10) compared to anticoagulant use only [Table 3].

Table 3

Computed tomography results by anticoagulant versus anticoagulant with antiplatelet

Anticoagulant type	No ICH	Acute ICH	Delayed ICH	Total
Anticoagulant only	219 (76%; CI: 70- 81)	61 (21%; CI: 17- 26)	9 (3%; CI: 1- 6)	289
Anticoagulant with antiplatelet	74 (75%; CI: 65- 83)	18 (18%; CI: 11- 27)	7 (7%; CI: 3- 14)	99

All CIs are 95%. CIs: Confidence intervals, ICH: Intracranial hemorrhage

There was one death during hospitalization in the acute ICH group and none in the delayed ICH group. No patients with a delayed ICH required neurosurgical intervention.

DISCUSSION

Patients taking anticoagulants are known to be at higher risk of ICH after head trauma.[22] However, the comparative risk between the various types of anticoagulants had not been fully elucidated. Our study found the rate of acute ICH to be higher in patients taking warfarin when compared to those taking a DOAC (26% vs. 16%), without a significant difference in the rates of delayed ICH between the specific anticoagulant medications (4% each). Furthermore, we found similar rates of ICH between each of the DOACs, without significant difference. This is consistent with previous reports in the literature, although the number of published studies that have looked at the specific type of anticoagulant is limited.[781218] One such retrospective study found that those taking warfarin had significantly higher rates of operative intervention and mortality compared to the patients taking a DOAC.[23] Our research provides further support for the notion that DOACs may confer lower risk of acute traumatic ICH than warfarin.

There is notable variability in published data regarding the rate of acute and delayed ICH in anticoagulated patients after head trauma. The majority of studies in the literature report a rate of acute traumatic ICH between 5% and 15% for patients taking an anticoagulant.[678910111224] Overall, our study found that patients taking any anticoagulant medications who suffered blunt head trauma had a 20% rate of acute ICH. While our reported rate of acute ICH is higher than that found in many other studies, this could be attributed to differences in study methodology, as we included patients with any GCS and not just those with minor head trauma. In our study, patients were identified through a retrospective review of the trauma registry. Several other studies that were performed exclusively at urban tertiary care or level 1 trauma centers reported similarly high rates of acute ICH in these patients.[22252627] One such study that also utilized a retrospective review of the trauma registry found a rate of acute ICH in anticoagulated patients of 29%.[6] Many EMS protocols follow the CDC Guidelines for Field Triage of Injured Patients, recommending that patients on anticoagulants with head injury be transported to a trauma center.[28] Because these patients are more likely to be transported to a level I trauma center, it is reasonable to expect these centers to have higher rates of acute ICH when compared to all hospitals in general.

When comparing patients on combination anticoagulant and antiplatelet therapy to anticoagulants only, our study did not find a significant difference in acute or delayed ICH rates. Previous studies on the topic have produced variable results, with some showing no association between combination anticoagulant and antiplatelet use and higher rates of traumatic ICH,[78] while others have found increased rates of acute ICH, progression of ICH, and mortality.[29]

Although our study found a significant rate of acute ICH in anticoagulated patients after head trauma, the rate of delayed ICH was low and none of the patients with delayed ICH required neurosurgical intervention. Of these patients, only one patient had a delayed ICH found on a repeat CT due to clinical change; the rest were all found on routine repeat imaging. Neither INR nor platelet count had an association with delayed ICH. There is ongoing debate about the frequency and clinical significance of delayed ICH in anticoagulated patients, as a number of previous studies found similarly low rates of delayed ICH in these patients.[11171920213031] Given the low rate of delayed ICH and the lack of mortality or need for neurosurgical intervention found in our study, routine repeat head imaging on anticoagulated patients after head trauma may not be necessary. This conclusion supports the recent Austrian consensus statement regarding the therapeutic approach in adult patients with traumatic brain injury receiving oral anticoagulant therapy.[32] However, the current American College of Emergency Physicians guidelines regarding mild traumatic brain injury, last updated in 2008, indicate that there are insufficient data to recommend if admission and repeat head CT is indicated.[33] Clearly, this topic requires further investigation, ideally with a large prospective study.

This study should be interpreted in the framework of several limitations. Similar to previous research using retrospective trauma registry data, there is a selection bias toward more significant injuries or mechanisms of injury. Local EMS patterns preferentially take head-injured patients on anticoagulants to the trauma center in line with national guidelines.[28] This may lead to a higher overall incidence of ICH when compared to the general ED population but should be comparable between anticoagulant groups. Our results do not include a control cohort of patients taking neither anticoagulants nor antiplatelet medications. As such, our study focuses on the comparisons between different cohorts of patients grouped by type of anticoagulant medication and the results cannot be extrapolated to the population of all patients suffering traumatic head injury. This limits the applicability of our study to the general ED population. Although there may be regional population differences, this should be somewhat mitigated by the multicenter nature of our study. Of note, baseline characteristics between these two geographically distinct cohorts were similar. At both centers, the threshold for obtaining an initial head CT imaging for patients on anticoagulants is very low, but there still exists the possibility of selection bias. There is also the possibility of delayed ICH even after the second CT scan which would lead to underreporting, although the retrospective nature of our study limited our ability to follow-up all of the outcomes.

CONCLUSION

In comparison to patients on the DOACs, patients on warfarin had a higher rate of acute ICH, but not delayed ICH. The rates of acute and delayed ICH in patients on anticoagulant medications did not change with the addition of antiplatelet medications. Overall, 4% of patients taking anticoagulant medications who suffered blunt head trauma were found to have a delayed ICH, and none of these required any neurointervention. Given the low rate of delayed ICH without resultant morbidity or mortality, routine observation and repeat head CT on patients without acute ICH may not be necessary.

Research quality and ethics statement

This study was approved by the Institutional Review Board/Ethics Commmittee and the corresponding approval numbers are Metro West IRB#00002845 and Michigan State University human subjects research committee study #1296. The authors followed applicable EQUATOR Network () guidelines during the conduct of this research project.

Financial support and sponsorship

None.

Conflicts of interest

There are no conflicts of interest.