• 
• 

Three Part Question

In [adult patients on warfarin with a minor head injury] does a [normal CT brain scan] allow [safe
discharge home]?

Clinical Scenario

An elderly woman attends your emergency department (ED) following a mechanical fall. She takes warfarin for atrial fibrillation and has a small occipital haematoma. Her Glasgow Coma Score (GCS) is 15; she has no amnesia and a normal neurological examination but did briefly lose consciousness. The International Normalised Ratio (INR) comes back within the therapeutic range at 2.9 and a CT scan is requested according to the National Institute of Health and Care Excellence (NICE) guidelines.

The scan is reported as normal, and her social circumstances are adequate in that she lives with her husband who can keep an eye on her. You wonder, though, whether it is safe to discharge her or if there is a possibility of delayed intracranial haemorrhage (DICH) due to her coagulopathy, and therefore she should be admitted for a period of neurological observation so that it can be identified and acted upon at the earliest opportunity.

Search Strategy

MEDLINE 1946 to August Week 4 2013 using the OVID interface.
[(exp Craniocerebral trauma/ OR head injur$.mp) AND (exp Warfarin/ OR warfarin.mp OR exp Coumarins/ OR exp Anticoagulants/ OR anticoagula$.mp OR phenoprocoumon.mp OR acenocoumarol.mp OR dicumarol.mp OR 4-hydroxycoumarins.mp OR sintrom.mp OR sinthrome.mp OR coumadins.mp)] LIMIT to humans AND english language.

Search Outcome

796 papers were found, of which 789 were irrelevant or of insufficient quality. The remaining 7 were directly relevant to the three part question, and are summarised in the table below.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Menditto et al 2012 Italy	97 prospectively enrolled consecutive warfarinised (for at least 1/52) patients >/=14 years old in Level II trauma centre without ICH on 1st CT after minor head injury (any head trauma, other than superficial injury to face, presenting with GCS 14 or 15), regardless of presence of absence of LOC, within 48hrs of injury, with ISS <15 between Jan 2007-Mar 2010. Structured clinical pathway implemented, comprising 24-hour period of observation & 2nd CT prior to discharge	Case series	Immediate TICH	19/97 (16%) +ve initial CT scan	None had GCS 14 or received concomitant antiplatelet therapy. Only 5 developed ICH by 2nd CT - therefore lacking statistical power to analyse multivariate predictors of such haemorrhage. Not designed to investigate optimal period of observation before repeat CT
			Death	No deaths reported
			DICH	5/97 (6%) [95% CI 1-11%]. Only 1 showed signs of neurological deterioration during observation period, 2/5 were discharged anyway as ICH regarded as minimal. 2 discharged after completing study protocol with -ve CT admitted 2/7 & 8/7 later with symptomatic SDH; neither required surgery. 2/5 with DICH at 24hrs had initial INR >3 as did both beyond 24hrs (RR DICH with INR >3 was 14 [95% CI 4-49]). 10 refused 2nd CT & were well during 30/7 follow-up
			Hospitalized	3 patients
			Surgery	1 craniotomy
Nishijima et al 2012 USA	Prospective observational study at 2 trauma centres & 4 community hospitals’ EDs in California of patients >/=18 with blunt head trauma (most commonly ground level fall [83.3%]), regardless of LOC/amnesia & preinjury warfarin or clopidogrel use (but not both) within previous 7/7 between Aug 2009-Jan 2011. Followed for 2/52 by review of in-patient electronic medical record or by telephone survey if already discharged. 1064 patients enrolled (768 warfarin [72.2%] & 296 clopidogrel [27.8%]). 364 (34.2%) from Level I or II trauma centres & 700 (65.8%) community hospitals. 1000 received head CT in ED. Both warfarin & clopidogrel groups had similar demographics & clinical characteristics, although concomitant aspirin use more prevalent among clopidogrel group. Enrolled after screening by treating physician (16.7% missed by screening failure, but these had similar characteristics & outcomes). Repeat CT was at clinicians’ discretion. Excluded patients with known injury transferred in as their injuries would falsely inflate prevalence of TICH. 78.5% had INR checked, median 2.5; IQR 2.0-3.3	Cohort study	Immediate TICH (warfarin group)	37/724 (5.1%) [95% CI 3.6-7.0%]. Follow-up of 63/64 not undergoing initial CT showed no subsequent diagnosis of TICH. Majority of patients (62.2%) had GCS 15, & 4/37 (10.8%) had no LOC, GCS 15 & no evidence of trauma above clavicles	Observational - not everyone scanned initially for ethical reasons, or routinely before discharge. Clinical follow-up though to elicit clinically important outcomes. Warfarin users more aware of risks and so more likely to present with less severe mechanisms of injury
			In-hospital mortality after immediate TICH (warfarin group)	8/37 (21.6%) [95% CI 9.8-38.2]
			Neurosurgical intervention after immediate TICH (warfarin group)	5/37 (13.5%) [95% CI 4.5-28.8]
			DICH (TICH within 2/52 after initially normal CT in absence of further head trauma (warfarin group)	4/687 (0.6%) [95% CI 0.2-1.5]. 2 were inoperable & died from extensive TICH
Kaen et al 2010 Spain	137 consecutive adult anti-coagulated patients with minor head injuries (GCS 14-15) sustained in previous 48hrs over 15/12 period with normal initial CT scans. All admitted for 24hrs observation with a control CT scan performed before discharge. All warfarinised but 3 also on aspirin	Cohort study	Intracranial bleeding on 2nd CT scan	135 (98.6%) had no evidence of intracranial lesions on control CT and none developed subsequent neurological deterioration or needed neurosurgical intervention during observation period. Only 2 patients (1.4%) (INR 3.1 & 2.88) developed haemorrhagic lesions on control CT. Both patients on both warfarin & aspirin; difference in incidence of bleeding in these 2 cases compared with those only anticoagulated was statistically significant (p=0.01). Likewise, they were amongst 14 (10%) who had LOC; difference in frequency of bleeding compared with 90% without LOC also statistically significant (p=0.004)	Observational, nonrandomised without a control group. Needs larger number of patients to establish definite conclusions
Itshayek et al 2006 Israel	All anticoagulated head injuries are scanned in their Level I Trauma Centre & during 2½-year period they describe 4 patients (aged 65-86) presenting following minor head injury. All chronically anticoagulated (3 on warfarin 1 on Enoxaparin & aspirin). INRs of 3 warfarinised patients were 2.99, 3.03 & 3.2. All GCS 15, no LOC, no focal neurological deficit, no evidence of cranial fracture & normal CT on arrival in dept	Case series	Development of DASH, morbidity & mortality	3 warfarin patients developed DASH with rapid neurological deterioration within 24hrs, the patient on Enoxaparin & Aspirin deteriorated after 3/7. 3 out of 4 patients underwent craniotomy for evacuation of their haematomas. The 2 male patients died after complicated post-operative courses. 1 female patient underwent surgical evacuation & rehabilitation, eventually achieving a GOS of 3. In the 2nd female patient, the haematoma was treated conservatively & she achieved a GOS of 4	Only a small case series, but quantification of risk impossible. We are told that they treat a population of 800,000 but we are not told how many such patients with normal examinations and normal CT scans they saw out of which 4 developed DASH
Cohen et al, 2006, USA	77 patients from 2 trauma databases over 3-year period on warfarin with minor head injury (GCS 13–15). Average age 68. INR obtained in 57% with average 4.4 and values >3 in 47%, range 1.8–9.5. (There was another group of 49 patients who had GCS <8, average age 65. Average INR 6.5, 50% >5. Mortality 87.8%)	Cohort study	Mortality	20 evaluated and sent home from ED. Of these, 35% had CT and all were normal. 18 returned and subsequently diagnosed with significant ICH. 2 patients died at home, 1 with autopsy-confirmed acute SDH. Overall mortality in these 20 patients was 88.8%. 45 patients admitted for observation for head injury± treatment of other injuries. CT obtained before admission in 70%, with only 4 showing any ICH. Within 8–18 h of injury (mean 12 h), 80% deteriorated to GCS <10 with ICH. Mortality in this group 84%. 12 patients presented within h or days of injury with ICH. All underwent emergent craniotomy with a resultant mortality of 83.3%	No matched control group. Majority of patients supra-therapeutically AC, and of those undergoing CT on initial presentation, only slightly more than 30% had any evidence of ICH. Retrospective
Garra et al 1999 USA	65 anticoagulated (AC) patients suffering minor head injury without LOC or acute neurological abnormality identified from retrospective chart review of electronic records from 6 community hospital EDs, including 1 trauma centre over 2-year period. Only 38 patients had PT assessment (range 12–30.7 s)	Cohort study	Clinically significant intracranial injury	No intracranial injury found in any of the 39 patients who had a CT. Telephone follow-up of the remaining 26 patients revealed no evidence of complications related to their head injuries	Their computer system may not have identified all eligible patients leading to a selection bias. Retrospective. In the 38 patients in whom PT was checked, none was >30 s and almost 1/3 were <14 s, indicating that even though these patients were on warfarin, few were actually AC
Peck et al 2011 USA	Retrospective review of protocol in level 1 trauma centre between 01/01/06 and 31/08/09, whereby those with blunt head trauma and preinjury use of an anticoagulant or antiplatelet agent (ACAP) (defined as warfarin, clopidogrel, heparin, or dipyridamole and aspirin in combination) received an admission CT head (CT1) regardless of symptomatology, and for those without ICH on CT1, a period of observation followed by a routine 2nd CT (CT2) in 6 h. Excluded those solely on warfarin if INR <1.3. Blinded review of CTs by radiologist. Attempts at follow-up across San Diego area. 500 patients qualified for protocol, of which 424 (85%) had –ve CT1. Mean age 75 with almost equal sex distribution, and 84% were a fall from standing. Mean GCS was 14.8. Mean initial INR for 312 warfarinised patients was 2.5 with 22 >4: so largely a therapeutic cohort	Cohort study	Neurological deterioration during observation	15 patients (3.5%) had clear documentation of neurological deterioration; 397 (93.6%) showed no change; 12 (2.8%) had insufficient data. None of the 15 had a +ve CT2	Retrospective. Neurological examination not clear in all patients. CT2 not performed in 15% of CT1 –ve patients. Platelet inhibition unable to be measured. Clopidogrel and other agents confound the warfarin data of interest here
			CT1 −ve → CT2 +ve	4 patients (1%), all warfarinised with INRs of 2.2, 2.2, 3.9 and 1.7. CT3 was stable or –ve in these cases; 3 were discharged and 1 died of cardiac disease following an orthopaedic procedure, all without intervention
			Readmission	3 (0.6%) of original 500 patients admitted for medical problems. 1 (0.2%) for ‘progression of neurological insult,’ having trace of intraventricular blood documented on CT3 but resolved on CT4 without intervention or treatment
Karni et al, 2001, USA	Retrospective review of approx. 2000 patients admitted to trauma service of regional trauma centre between Sept 1998 and May 2000. 278 patients >65 years old with CT-documented TICH. 21 warfarinised but 5 excluded as thought more likely spontaneous ICH. Average age 78, average GCS 11, average INR 3.0	Case–control study	Mortality	Use of FFP and cryoprecipitate to reverse coagulopathy did not impact on mortality. Nearly half of patients studied underwent craniotomy with 67% 30-day mortality. Overall mortality rate in WHI patients was 50% (8/16) compared with 20% (51/256) in those without coagulopathy (p=0.011). In subgroup of patients with INR >3.5, the mortality rate approached 75%	Really no data displayed to appreciate. Inadequate sample size for those with INR >3 (n approx. 8) from which to draw meaningful conclusions. Retrospective
Li et al, 2001, USA	Retrospective chart review from 2 centres, May 1996–May 2000 from 1 and Jan–Dec 1998 from another. 144 WHI patients identified that had CTB. Excluded those with high-risk and moderate-risk findings. Median (IQR) age 83 (77–87)	Cohort study	Clinically important CT injury that results in change in disposition	10 patients found to have such injuries (7%, 95% CI 3% to 11%). No significant demographic or case-characteristic differences between groups with and without CT-identified injuries. Median (IQR) INR 2.1 (1.8–3.0) CT abnormal vs 2.1 (1.6–2.7) CT normal (p=0.6)	Retrospective design using different time periods from 2 centres for an unexplained reason. Selection bias from including only those who had CTB and no follow-up data to ensure no DICH
Mina et al, 2003, USA	Prospective evaluation of all WHI patients seen in ED of level 1 trauma centre between Jan 2001 and Feb 2002 via a ‘Coumadin protocol’ and compared with a group of age-matched patients over the same time period admitted with head injury but not on warfarin. 94 WHIs, mean age 77±11. Control group mean age 75±12 with normal INR values (mean 1.1±0.1)	Case–control study	Demographics	WHI group: no significant differences between those with and those without ICH in terms of age, gender, INR (3.2±1.9 with ICH vs 3.2±2.5 without, p=0.914), or MOI. ISS significantly higher (21.3±8.2 vs 3.4±7.1, p<0.001) and GCS significantly lower (12.0±4.2 vs 14.7±1.6, p<0.001) for patients with ICH. Control group: GCS not significantly different from WHI group but significantly higher ISS than WHI group	Well designed but no mention of impact of level of anticoagulation with regards to mortality. Apparently, most patients were therapeutic though
			Mortality	Significantly higher WHI group 12/25 (48%) vs 5/47 (10%) control group, p<0.001. WHI group: INR similar (3.3±1.6 dead vs 3.0±2.1 survivors, p=0.585). ISS significantly lower and GCS significantly higher in survivors. Control group: ISS not significantly different but GCS significantly higher in survivors
			ICH	25/94 (27%) WHI group. 47/70 (67%) control group. No significant differences in age, gender, ISS, GCS or MOI
Reynolds et al, 2003, USA	32 WHI patients over 7-year period identified from trauma registry database at level 2 trauma centre. Group 1–24 patients (mean age 82.5). All GCS 15. 8 had INR checked (mean 2.45, range 1.6–3.6). Group 2a—4 patients. All GCS 15. Mean INR 2 (range 1.5–2.6). Group 2b—4 patients. All GCS 15 but all became comatose within 6 h. Mean INR 2.5 (range 2.3–3.1)	Cohort study	ICH	Failed to observe a statistically significant difference in mean INR between groups 1 and 2 (p=0.59) although only 8/24 patients from group 1 tested. No statistical difference between subgroups 2a and 2b (p=0.12). Group 1: Only 3 had CTB (all normal). All discharged home from ED. 22 alive 6/12 post-injury without evidence of DICH. 2 patients lost to follow-up. Group 2a: All had evidence of ICH on CTB. 2 had FFP and vitamin K. All treated conservatively and survived to return to their location of origin. Group 2b: All had evidence of ICH on CTB. All had FFP±vitamin K. 3 had craniotomy with decompression (2 died; 1 discharged to nursing/rehab facility) and the 4th declined intervention and subsequently died	Small observational study really with only 8 patients with ICH in 7 years. Retrospective. Would have been more informative if all of group 1 had an INR and CTB. Note that delay to reversal occurred from failure to send appropriate blood samples from patients who appeared neurologically normal after arrival in ED
Gittleman et al, 2005, USA	89 patients being treated with heparin or coumadin who had a head injury and underwent a CTB at a level 1 trauma centre identified over a 4-year period (April 1997–Jan 2002) using hospital information database and neuroradiology case log. 77 taking coumadin, 8 taking heparin and 4 taking both	Cohort study	ICH	7 patients with ICH and all had GCS <15. Included 3 cerebellar haemorrhages that were more suggestive of hypertensive rather than traumatic aetiology. No significant difference found between those with ICH and those without with respect to coagulation profile (INR 2.2±1.1 with ICH vs 2.5±1.2 without ICH)	Relatively small numbers and failed to meet sample size required by pretest power calculation. Selection bias from only including those who had CTB. Retrospective. No breakdown of ICH patients to say who was on coumadin or heparin and presumably this could skew the mean INR values of the groups. No mortality data or follow-up data either regarding the possibility of DICH
Ivascu et al, 2005, USA	82 WHI patients identified prospectively between Feb 2003 and April 2007, of which 19 (23%) had evidence of ICH on CTB. Compared with 2 control groups: a group identified during this protocol and a group of historic controls treated before implementation of this protocol to fast-track anticoagulation reversal	Cohort study	Age, sex, MOI, presenting GCS	Not statistically significant between groups with ICH and those without	The validity of comparing the median INR is questionable, as opposed to the comparison of mean INR. There are minimal details given of the level of coagulopathies, such as range, from which more informed conclusions could be drawn
			INR	All patients with ICH had therapeutic INR, and there was not a statistically significant difference in degree of anticoagulation between groups. Median INR 2.7 (with ICH) vs 2.5 (without), p=0.350
			ICH	63 patients without ICH on initial CTB were admitted for 23 h of observation. None subsequently developed ICH, including 12 patients with an INR >3.5
			Protocol implementation	Improved time from hospital presentation to physician evaluation, 50% less time in triage, significant reduction in time to obtain CTB, full anticoagulation reversal with FFP, significant reduction in rates of ICH progression and mortality
Franko et al, 2006, USA	Retrospective analysis of consecutive series of 1493 adult blunt head injury patients between Jan 2001 and May 2005. 159 warfarinised patients identified and were significantly older, with average age 78±10 and average INR 2.4±1.06	Case–control study	ISS	Significantly greater 14.5±8.4 WHI vs 12.4±9.4 control, p<0.01	They cite a selection bias through education of AC patients, encouraging the seeking of early medical attention, even after seemingly minimal trauma, and so more of these patients present for evaluation. It is suggested that non-therapeutic users and non-users (NU) had similar results, but they were not compared directly. Retrospective
			LOS	Significantly longer 6.7±11.1 WHI vs 4.1±6.3 control, p<0.001
			ICH	Significantly more likely 96/159 (60.4%) WHI vs 536/1334 (40.2%) control, p<0.001, OR=2.2, 95% CI 1.6 to 3.1
			Mortality	Significantly higher 38/159 (23.9%) WHI vs 66/1334 (4.9%) control, p<0.001, OR=6.0, 95% CI 3.8 to 9.3
			Mortality in those with ICH (n=632)	Significantly higher 36/96 (37.5%) WHI vs 61/536 (11.4%) control, p<0.001, OR=4.6, 95% CI 2.8 to 7.6
			Effect of preinjury anticoagulant level	Mortality and occurrence of ICH both significantly increased with increasing INR (Cochran's linear trend p<0.001)
			Age-dependent mortality	Mortality of patients >70 significantly higher than that of younger patients (p<0.001). In control group mortality significantly higher with age >70 (38/465, 8.2% vs 28/869, 3.2%) p<0.001. In WHI group mortality significantly higher with age >70 (34/133, 25.6% vs 4/26, 16.4%) p<0.001
Pieracci et al, 2007, USA	Retrospective study (2004–2006) of all trauma patients aged ≥65 (n=275) evaluated by a trauma service at a level 1 trauma centre who had a CTB following a head injury, including 40 WHI. 3 cohorts compared: (1) WHI with INR ≥2 (therapeutic group (TG)), n=22, 11 of whom had INR >3. Mean INR 3.33, range 2.12–7.28. (2) WHI with INR <2 (non-therapeutic group (NT)), n=18. Mean INR 1.51, range 1.00–1.96. (3) Warfarin NU, n=235. Mean INR 1.11, range 0.87–4.01	Case–control study	Admission GCS	TG 9 patients (40.9%), p=0.001; NT 2 patients (11.1%); NU 22 patients (11.9%). OR=5.13, 95% CI 1.97 to 13.39, p=0.001 comparing TG to NU group	Relatively small sample size, therefore, unable to fully compare warfarin users with INR 2–3 with those with INR >3. Results suggest threshold rather than a linear relationship between level of anticoagulation and risk of ICH, which is consistent with exponential scale of INR. However, both a small number of therapeutic users and a relatively narrow range of INRs among them preclude more detailed analysis. Retrospective
			ICH	TG 17 patients (77.2%), p=0.10; NT 9 patients (50.0%); NU 105 (56.8%). OR=2.59, 95% CI 0.92 to 7.32, p=0.07 comparing TG to NU group. Subgroup analysis revealed no difference in likelihood of ICH between those with INR 2–3 (9/11, 81.8%) and INR >3 (8/11, 72.7%)
			Overall mortality	TG 7 patients (31.8%), p=0.009; NT 2 patients (11.8%); NU 17 patients (9.4%). OR=4.48, 95% CI 1.60 to 12.50, p=0.004 comparing TG to NU group
			Mortality after ICH	TG 6 patients (35.3%), p=0.01; NT 1 patient (12.5%); NU 14 patients (13.7%). OR=3.42, 95% CI 1.09 to 10.76, p=0.03 comparing TG to NU group
Grandhi et al, 2008, USA	Retrospective review of all patients evaluated at level 1 trauma centre between Jan 2000 and Dec 2006, to include patients ≥65 coded with a closed head injury. 52/491 (11%) were documented as taking warfarin (AC) and subsequently compared with those not AC (NAC) by 1:3 propensity matching. Mean admission INR in AC group 2.4±1.2	Case control study	Ventilator LOS (days)	2.8±7.9 AC vs 1.5±5.8 NAC, p=0.08	Numbers too small to determine if there was a certain level of anticoagulation for which outcomes significantly worsened ?type 2 error. Analysis of association between degree of anticoagulation and various measures of morbidity and mortality with larger sample populations may be able to determine a ‘cut-off’ INR value for which the benefits of anticoagulation are outweighed by its potential complications. Retrospective
			ICU LOS (days)	6.4±11.8 AC vs 4.4±7.3 NAC, p=0.19
			Hospital LOS (days)	10.5±13.6 AC vs 9.1±12.1 NAC, p=0.97
			Mortality	19/49 (38.8%) AC vs 34/147 (23.1%) NAC, p=0.04
Major et al, 2009, USA	399 patients presenting to University Hospital ED with head injury and coexistent anticoagulant (warfarin) or antiplatelet (aspirin, clopidogrel or dipyridamole) therapy who were admitted to the hospital over a 3-year period (Jan 2005–Dec 2007), identified through search of electronic patient records	Cohort study	ICH	110 patients (28% had CTB, of which 24 showed ICH. 4 died and 2 had neurosurgical intervention, but none of these were warfarinised. Of 89 patients on warfarin (including 5 also on aspirin), 27 (30%) underwent CTB, with 4 of these (15%) having ICH. There were 63 patients on warfarin who had an INR <3 (2/17 +ve scans) and 26 who had an INR >3 (2/10 +ve scans). The RR of a patient having a +ve CTB with an INR >3 compared with an INR <3 was 1.7 (95% CI 0.3 to 10.3)	Patients discharged from ED excluded leading to a selection bias. Retrospective. Only 44/110 patients scanned had the indication recorded. No information as to whether those with +ve scans were the ones with indications for scanning. May suggest that a significant proportion of this cohort was low risk
Brewer et al, 2011, USA	Retrospective review of trauma registry at level 2 trauma centre. All trauma registry patients with MHI registered between 2004 and 2006 who were taking clopidogrel or warfarin, GCS 15 and had CTB included. Trauma registry includes all patients admitted to or consulted by the trauma service. 141 patients (mean age 79, range 36–101)	Case-control study	ICH	41 patients (29%) diagnosed with ICH. 23/84 (27%) on warfarin. Mean presenting INR with ICH 1.97±0.92 compared with 2.3±1.2 without ICH (p=0.0987). 15/36 (41%) on clopidogrel, 3/21 (14%) on combination therapy. 39 (95%) of patients with ICH underwent reversal± discontinuation of clopidogrel± warfarin. 5 patients required surgical evacuation of ICH. 4 patients died. LOC (Wald=7.468, β=1.179, p=0.008) predicted a +ve CT. Type of medication (warfarin, clopidogrel or aspirin) did not reach statistical significance as a predictor of a +ve CT	Patient population only includes those from trauma registry and may explain a selection bias. Relatively small numbers and retrospective design

Comment(s)

There is much debate in the literature as to how best to manage this group of patients. Their risk of immediate traumatic intracranial haemorrhage is increased to about one in six compared with those not warfarinised. The NICE guidelines tell us only to perform a CT scan in the presence of loss of consciousness or amnesia, but tell us nothing about the impact of coagulopathy or about a period of observation, especially if no CT is to be performed. From my research, it is clear that these patients are managed quite differently even within the same centre, in the absence of robust guidelines. Italian guidelines published in 1999, derived largely from a published case series of two patients, advise a CT scan in all patients with coagulopathy, admission for observation for 24 h and then a second CT prior to discharge. Kaen et al here suggest that this second CT may not be necessary, with only 2/137 (1.4%) found to have DICH, both neurosurgically irrelevant. Peck et al show that a routine repeat CT after 6 h is also unnecessary. DICH in this setting, with a normal admission CT and an International Normalised Ratio (INR) not supra-therapeutic, would appear to be a fairly rare occurrence, quoted as <1% in large studies by Nishijima et al and Peck et al. Indeed, the relative risk of DICH with an INR >3 was found to be 14 (95% CI 4 to 49) by Menditto et al. However, more often than not in the literature it is described following a previous discovery of a cranial fracture or following a clinical deterioration in a previously asymptomatic patient that had therefore not been initially scanned. An important question to ask is whether such findings of DICH really matter? By admitting more patients, and ordering more scans, is outcome improved? Or do we just reveal positive findings of little clinical significance? Furthermore, very few of the DICH found in these studies occurred within 24 h; instead often after several days. So do we admit these patients for our own false reassurance? Our practice seems largely to be based on anecdotal evidence, fuelled probably by such case series as presented here, but the larger studies described should reassure us as to the very low risk of DICH, particularly with an initial INR <3 and a normal CT scan. Therefore, it would seem reasonable to discharge these patients, but with robust and clear instructions. Some have even shown the effectiveness of telephone follow-up the following day.

Editor Comment

aOR, adjusted OR; AUC, area under the curve; CTB, CT brain; DICH, delayed intracranial haemorrhage; ED, emergency department; FFP, fresh frozen plasma; GCS, Glasgow Coma Score; ICU, intensive care unit; INR, International Normalised Ratio; ISS, injury severity score; LOC, loss of consciousness; LOS, length of stay; MHI, minor head injury; MOI, mechanism of injury; NICE, National Institute for Health and Care Excellence; NPV, negative predictive value; PPV, positive predictive value; PT, prothrombin time; ROC, receiver operator curve; RR, relative risk; SDH, subdural haematoma; TICH, traumatic intracranial haemorrhage; WHI, warfarinised head injury.

Clinical Bottom Line

The risk of delayed intracranial haemorrhage has been quantified as reassuringly small, rather than anecdotally inflated. Therefore, it would seem reasonable to discharge those with an International Normalised Ratio <3 and a normal initial CT scan, ensuring good instructions and follow-up.

References

1. Menditto VG, Lucci M, Polonara S, et al. Management of minor head injury in patients receiving oral anticoagulant therapy: a prospective study of a 24-hour observation protocol. Ann Emerg Med 2012; 59: 451-455.
2. Nishijima DK, Offerman SR, Ballard DW, et al. Immediate and delayed traumatic intracranial hemorrhage in patients with head trauma and preinjury warfarin or clopidogrel use. Ann Emerg Med 2012; 59: 460-468.
3. Kaen A, Jimenez-Roldan L, Arrese I, et al. The value of sequential computed tomography scanning in anticoaguated patients suffering from minor head injury. J Trauma 2010; 68: 895-898.
4. Itshayek E, Rosenthal G, Fraifeld S, et al. Delayed posttraumatic acute subdural haematoma in elderly patients on anticoagulation. Neurosurgery 2006; 58: 851-856.
5. Cohen DB, Rinker C, Wilberger JE. Traumatic brain injury in anticoagulated patients. J Trauma 2006; 60: 553-557.
6. Garra G, Nashed AH, Capobianco L. Minor head trauma in anticoagulated patients. Acad Emerg Med 1999; 6: 121-124.
7. Peck KA, Sise CB, Shackford SR, et al. Delayed intracranial hemorrhage after blunt trauma: are patients on preinjury anticoagulants and prescription antiplatelet agents at risk? J Trauma 2011;71:1600–4.
8. Karni A, Holtzman R, Bass T, et al. Traumatic head injury in the anticoagulated elderly patient: a lethal combination. Am Surg 2001;67:1098–100.
9. Li J, Brown J, Levine M. Mild head injury, anticoagulants, and risk of intracranial injury. Lancet 2001;357:771–2.
10. Mina AA, Bair HA, Howells GA, et al. Complications of preinjury warfarin use in the trauma patient. J Trauma 2003;54:842–7.
11. Reynolds FD, Dietz PA, Higgins D, et al. Time to deterioration of the elderly, anticoagulated, minor head injury patient who presents without evidence of neurologic abnormality. J Trauma 2003;54:492–6.
12. Gittleman AM, Ortiz AO, Keating DP, et al. Indications for CT in patients receiving anticoagulation after head trauma. Am J Neuroradiol 2005;26:603–6.
13. Ivascu FA, Howells GA, Junn FS, et al. Rapid warfarin reversal in anticoagulated patients with traumatic intracranial hemorrhage reduces hemorrhage progression and mortality. J Trauma 2005;59:1131–9.
14. Franko J, Kish KJ, O'Connell BG, et al. Advanced age and preinjury warfarin anticoagulation increase the risk of mortality after head trauma. J Trauma 2006;61:107–10.
15. Pieracci FM, Eachempati SR, Shou J, et al. Degree of anticoagulation, but not warfarin use itself, predicts adverse outcomes after traumatic brain injury in elderly trauma patients. J Trauma 2007;63:525–30.
16. Grandhi R, Duane TM, Dechert T, et al. Anticoagulation and the elderly head trauma patient. Am Surg 2008;74:802–5.
17. Major J, Reed MJ. A retrospective review of patients with head injury with coexistent anticoagulant and antiplatelet use admitted from a UK emergency department. Emerg Med J 2009;26:871–6.
18. Brewer ES, Reznikov B, Liberman RF, et al. Incidence and predictors of intracranial hemorrhage after minor head trauma in patients taking anticoagulant and antiplatelet medication. J Trauma 2011;70:E1–5.