Antithrombotic regimens and need for critical care interventions among patients with subdural hematomas
a b s t r a c t
Background: Antithrombotic-associated subdural hematomas (SDHs) are increasingly common, and the possibil- ity of clinical deterioration in otherwise stable antithrombotic-associated SDH patients may prompt unnecessary admissions to intensive care units. It is unknown whether all antithrombotic regimens are equally associated with the need for critical care interventions. We sought to compare the frequency of critical care interventions and poor functional outcomes among three cohorts of noncomatose SDH patients: patients on no antithrombotics, patients on anticoagulants, and patients on antiplatelets alone.
Methods: We performed a retrospective cohort study on all noncomatose SDH patients (Glasgow Coma Scale > 12) presenting to an academic health system in 2018. The three groups of patients were compared in terms of clinical course and functional outcome. Multivariable logistic regression was used to determine predictors of need for critical care interventions and poor functional outcome at hospital discharge.
Results: There were 281 eligible patients presenting with SDHs in 2018, with 126 (45%) patients on no antithrombotics, 106 (38%) patients on antiplatelet medications alone, and 49 (17%) patients on anticoagulants. Significant predictors of critical care interventions were coagulopathy (OR 5.1, P < 0.001), presence of contusions (OR 3, P = 0.007), Midline shift (OR 3.4, P = 0.002), and maximum SDH thickness (OR 2.4, P = 0.002). Significant predictors of poor functional outcome were age (OR 1.8, P < 0.001), admission Glasgow Coma Scale score (OR 0.3, P < 0.001), dementia history (OR 4.2, P = 0.001), and coagulopathy (OR 3.5, P = 0.02). Isolated antiplatelet use was not associated with either critical care interventions or functional outcome.
Conclusion: Isolated antiplatelet use is not a significant predictor of need for critical care interventions or poor functional outcome among SDH patients and should not be used as a criterion for triage to the intensive care unit.
(C) 2021
The incidence of subdural hematomas (SDHs) is increasing in the United States, [1] and SDH is projected to become the most common cranial neurosurgical condition by 2030. [2] SDHs most commonly re- sult from trauma, complicating 10-20% of all Traumatic brain injuries (TBIs). [3] The rising incidence of SDHs has been attributed to an aging population and increased use of antithrombotics, with approximately 50% of SDHs being associated with antithrombotic use. [4,5]
E-mail addresses: Robin2dv@ucmail.uc.edu (D. Robinson), pylelm@upmc.edu (L. Pyle)
, foremabo@ucmail.uc.edu (B. Foreman), ngwenyla@ucmail.uc.edu (L.B. Ngwenya), adeoyeo@ucmail.uc.edu (O. Adeoye), woodl@ucmail.uc.edu (D. Woo), kreitzne@ucmail.uc.edu (N. Kreitzer).
As more patients present with antithrombotic-associated SDHs to emergency departments (EDs), triage decisions will become increas- ingly important. When compared to other forms of traumatic intracra- nial hemorrhage (tICH), SDHs tend to occur in older patients with less trauma and require surgery at a high rate; it is thus difficult to deter- mine their optimal disposition, particularly in patients with preserved neurologic exams. [6] Clinical tools have recently been proposed to as- sist with decision-making, [7,8] but these tools either excluded all antithrombotic-associated SDHs or treated antiplatelet use and antico- agulant use as equivalent risk factors for more severe bleeds.
While anticoagulant use has been consistently associated with higher risk SDHs, [9] the data on antiplatelet use has been less consis- tent. A few studies have suggested that antiplatelet use may be associ- ated with worse outcomes in general tICH patients; however, many of these studies used mortality as the primary outcome (an outcome
https://doi.org/10.1016/j.ajem.2021.03.035
0735-6757/(C) 2021
heavily influenced by decisions regarding withdrawal of life-sustaining treatment), [10] did not control for possible confounders (e.g. age and premorbid conditions), and did not distinguish between the different types of tICH. [11-13] A more recent study found that acetylsalicylic acid (ASA) use may actually be associated with a reduced mortality in SDH patients, although this was only found in elderly patients with min- imal trauma histories. [14] No prior studies have examined the impact of antiplatelet use on outcomes relevant to decision-making in the ED in a complex cohort of SDH patients.
Despite limited evidence, antiplatelet use significantly increases the likelihood that noncomatose SDH patients (i.e. patients with a score on the Glasgow Coma Scale >12) will be triaged to the intensive care unit (ICU). [15] This is potentially problematic, as noncomatose SDH patients with minimal extracranial trauma may receive overly aggressive care, including unnecessary ICU admissions. [16-19] If antiplatelet use does not increase the risk of complications or poor outcomes, unnecessary ICU admissions could be avoided in this growing population. To help clarify triage decisions, we thus sought to determine whether isolated antiplatelet use is associated with the need for critical care interventions and overall functional outcome at hospital discharge among non- comatose SDH patients when compared with patients on anticoagulants and patients on no antithrombotics.
- Methods
This study was approved by our Institutional Review Board with a waiver of informed consent.
-
- Study design and setting
We performed a retrospective cohort study by identifying all SDH cases at a two-hospital academic health system comprised of a Level 1 Trauma Center and a Level 3 Trauma Center in 2018 using the ICD-10 codes S06.5 and I62.0. [4] This academic health system acts as a major referral center for a large geographic area and contains the only Level 1 Trauma center in the region. The same trauma surgery and neurosur- gery teams staff patients at both centers, although some patients seen at the Level 3 Trauma Center are staffed with the neurosurgeons using telemedicine when deemed safe by the treating trauma surgeons. Pa- tients with intracranial hemorrhage have protocolized coagulopathy management targets that include a platelet count >75,000 along with an International normalized ratio <1.5. During the time of this study, our institution managed many mild TBI patients with intracranial hemorrhage with an observational protocol (including some SDH pa- tients on antiplatelets), which has been previously described. [20] A portion of SDH patients in this study qualified for this protocol and were thus discharged home from the ED after the observational period ended if no clinical changes or complications occurred.
Inclusion criteria were initial presentation during 2018, age >= 18, ini- tial Glasgow Coma Scale Score (GCS) > 12, and isolated head injury (de- fined as abbreviated injury severity score < 3 for all non-head regions).
[21] The exclusion criteria were penetrating brain injury, presentationto an outpatient provider, or inability to visualize an SDH on preopera- tive imaging. Patients were followed through hospital discharge.
Using a standardized case report form, two board-certified neurolo- gists (DR and LP) reviewed images and abstracted data on demo- graphics, labs, vital signs, GCS scores, discharge destination, trauma history, and surgical management for each patient. No or minimal trauma was defined as no trauma or a fall <3 ft. [22] Imaging was reviewed for all patients, utilizing the first available Head CT. SDHs
were categorized as either acute or not acute (with the not acute category including heterogeneous, homogeneous hypodense, and ho- mogeneous isodense SDHs). [23] SDHs were assessed for their location, which was categorized as along the convexity or not (e.g. parafalcine or tentorial). Follow-up CT scans within 48 h of the initial scan were reviewed for SDH expansion, which was defined as >20% increase in volume or > 2 mm increase in thickness, and symptomatic expansion was defined as any worsening in the GCS or neurological exam at the time in which the SDH expanded. [24] Similar to prior work, coagulop- athy was defined as admission INR > 1.3 or platelet count <100,000.
[25] Data was collected for every patient who received any blood prod-ucts or other therapies used to reverse the effects of antithrombotics, along with the timing of these interventions.
The first 10 cases were dual-abstracted; between the two reviewers, there was a concordance of 95% for all categorical data, and an intraclass coefficient of 0.985 for continuous variables. All subsequent cases underwent single review, with 10% of charts randomly audited. Discrep- ancies were resolved by discussion.
-
- Antithrombotic exposure status
Patients were divided into three cohorts based on antithrombotic exposure, which was determined by the primary team on admission; those not on antithrombotics, those on antiplatelets alone (both single and dual antiplatelet regimens), and those on anticoagulation (includ- ing those on both antiplatelets and anticoagulants).
-
- Outcome
The primary outcome was need for critical care intervention(s). We utilized a previously defined composite variable, [26] which is based upon guidelines for ICU utilization laid out by the American College of Critical Care Medicine. [27] Specifically, critical care interventions were defined as the following: need for blood product transfusions, in- vasive hemodynamic monitoring, vasopressor therapy, neurosurgical intervention, osmotherapy, and mechanical ventilation. [26] Neurosur- gical intervention included Burr holes, craniotomies, hemicraniec- tomies, Skull fracture repairs, placement of External ventricular drains, and intracranial pressure monitors. While all antithrombotic Reversal strategies were recorded, blood product transfusions were only consid- ered a critical care intervention if they occurred after hospital admission or transfer to observational status, similar to prior studies. [26] The pri- mary outcome was available for all patients. The secondary outcome was dichotomized Modified Rankin scale (mRS) at hospital discharge, with scores >= 2 being considered a poor outcome. mRS scores were de- termined by the reviewers for all patients based on clinician notes at time of discharge.
-
- Statistical analyses
All statistical analyses were performed in R, version 4.0.2. [28] To de- termine sample size, we reviewed prior studies which found that ap- proximately 40% of noncomatose tICH patients eventually require critical care interventions. [21] A review of our institutional electronic medical record suggested that approximately 250-300 noncomatose SDH patients present to our health system each year, yielding an esti- mated ~100 patients requiring critical care interventions. Thus, based on the rule of 10 events per fitted parameter for logistic regression,
[29] a chart review of all patients admitted over 1 year would provideadequate sample size for a multivariable logistic regression.
The antiplatelet and anticoagulant cohorts were compared to the no antithrombotic cohort using Chi-Squared tests for all categorical vari- ables and Kruskal-Wallis tests for all continuous variables (none of the continuous variables were normally distributed). An omnibus test was first performed across all three cohorts and then pairwise comparisons were performed if the initial test was suggestive of statistically
Fig. 1. Flow diagram of case identification and antithrombotic exposure classification. Abbreviations: GCS = Glasgow Coma Scale, SDH=Subdural hematoma, ED = Emergency Department.
significant differences. P values obtained from pairwise comparisons were adjusted using the Bonferroni method.
For multivariable analysis of critical care interventions and func- tional outcome, a univariable analysis was first carried out for all puta- tive predictors using logistic regression. Variables that had more than 5% missing values were excluded from further consideration. The antithrombotic variables considered for the models were isolated anti- platelet use and anticoagulant use (with anticoagulant use including patients on only anticoagulants and patients on anticoagulants and
antiplatelets). After univariable analysis, all variables with a P value
<0.2 were placed in a multivariable model. [30] Modeling was per- formed with complete case analysis, as no tested variables had more than 5% missingness. Then, a manual backwards and forwards stepwise logistic regression based on P values was performed to find significant predictors of a poor functional outcome, with a P value of 0.05 consid- ered evidence of a significant association on multivariable analysis; age was kept in all models regardless of statistical significance consider- ing its known impact on SDHs. [31]
- Results
- Identification of SDH cohort
We identified 444 incident visits from 442 unique patients in 2018, of whom 281 cases met all inclusion criteria (Fig. 1). Of the 281 SDH cases in 2018, 126 (45%) patients were on no antithrombotics, 106 (38%) were on antiplatelet medications alone, and 49 (17%) were on anticoagulants.
-
- Comparison of the three cohorts
Univariable analysis of the three cohorts is shown in Table 1. Age, history of no/minimal trauma, SDH thickness, presence of skull fracture, likelihood of being discharged home from the ED, need for critical care interventions, and poor functional outcome were significantly different among the three cohorts. On pairwise comparison with the no anti- thrombotic cohort, the antiplatelet and anticoagulant cohorts were both significantly older (median age 56.2 in the no antithrombotic co- hort compared with median ages of 76.9 and 79.2 in the anticoagulant and antiplatelet cohorts, with P < 0.001 for both comparisons), more likely to have no/minimal trauma history (56% in the no antithrombotic
Comparison of the three cohorts.
No antithrombotic (N = 126) |
Anticoagulant users (N = 49) |
Antiplatelet users (N = 106) |
Omnibus P value |
|
Age (years) |
56.2 (38.6, 68.5) |
76.9 (66.3, 83.8)+ |
79.2 (70.6, 86.8) + |
<0.001 |
Male |
83 (66) |
32 (65) |
68 (64) |
0.96 |
Coagulopathy? |
16 (14) |
40 (82)+ |
6 (6)? |
<0.001 |
No or minimal trauma?? |
77 (56) |
46 (94)+ |
91 (86)+ |
<0.001 |
Maximum thickness (mm) |
5 (3, 8.8) |
7 (5, 14)+ |
5.5 (3.8, 10) |
0.01 |
Presence of midline shift |
38 (30) |
14 (29) |
21 (20) |
0.2 |
Location along convexity |
105 (83) |
39 (80) |
78 (74) |
0.2 |
Radiological characteristics |
||||
Acute/Hyperdense |
96 (76) |
35 (71) |
85 (80) |
0.5 |
Heterogeneous |
23 (19) |
8 (16) |
18 (17) |
0.9 |
Chronic/Hypodense |
4 (3) |
3 (6) |
7 (6) |
0.45 |
Subacute/Isodense |
3 (2) |
3 (6) |
2 (2) |
0.3 |
Associated contusion |
22 (17) |
8 (16) |
14 (13) |
0.67 |
Associated subarachnoid hemorrhage |
35 (28) |
16 (33) |
34 (32) |
0.7 |
Associated Skull fracture |
37 (29) |
4 (8)+ |
16 (15)+ |
0.002 |
SDH expansion |
12 (10) |
6 (12) |
16 (15) |
0.43 |
Symptomatic SDH expansion |
4 (3) |
1 (17) |
3 (3) |
0.9 |
Seizures |
1 (1) |
6 (12)+ |
3 (3) |
<0.001 |
Critical Care interventions |
36 (29) |
26 (53)+ |
19 (18) ? |
<0.001 |
Blood product transfusion |
20 (16) |
14 (29) |
12 (11) |
|
Arterial line or central line |
6 (5) |
5 (10) |
4 (4) |
|
Osmotherapy |
13 (10) |
9 (18) |
3 (3) |
|
Pressors |
3 (2) |
3 (6) |
0 (0) |
|
Ventilated |
6 (5) |
8 (16) |
8 (8) |
|
Neurosurgery |
19 (15) |
9 (18) |
7 (7) |
|
Transferred to Level 1 and discharged within 48 h |
18 (14) |
8 (16) |
24 (23) |
0.2 |
Discharged home from ED |
27 (21) |
5 (10) |
36 (34)? |
0.004 |
Poor functional outcome |
42 (33) |
41 (84)+ |
61 (58)+ ? |
<0.001 |
Data are n (%) or median (IQR). P values shown are omnibus values for overall Kruskal Wallis or Chi Square tests, with bolded variables indicating statistical significance at P < 0.05. Abbreviations: SDH = subdural hematoma, mm = millimeter, ED = Emergency Department.
+ Statistically different at P<0.05 from the no antithrombotic cohort with pairwise testing after Bonferroni adjustment.
? Statistically different at P<0.05 from the anticoagulant cohort with pairwise testing after Bonferroni adjustment.
* Defined as INR>1.3 or platelet count <100,000.
?? Defined as no trauma or fall <3 feet.
cohort compared with 94% and 86% in the anticoagulant and antiplatelet cohorts, P < 0.001 for both comparisons), less likely to have an associ- ated Skull fractures (29% in the no antithrombotic cohort compared with 8% and 15% in the anticoagulant and antiplatelet cohorts, P = 0.02 and P = 0.05 respectively), and more likely to have a poor dis- charge functional outcome (33% of patients in the no antithrombotic co- hort compared with 84% and 58% in the anticoagulant and antiplatelet cohorts, P < 0.001 for both comparisons). There was no difference in the rate of seizures between patients on antiplatelets and patients on no antithrombotics (3% versus 1%, p = 0.15), but seizures were signi- ficantly more common in patients on anticoagulants (12% versus 1%, p = 0.006). Antiplatelet users were commonly discharged home from the ED without admission (34%), a rate that was significantly higher than in the anticoagulant cohort (10%, P = 0.01). Critical care interven- tions were most common in the anticoagulant cohort, occurring in 53% of Anticoagulated patients compared with 29% in the no antithrombotic cohort (P = 0.01) and 18% in the antiplatelet cohort (P < 0.001). The an- ticoagulant cohort also had the thickest SDHs, although this only reached statistical significance when compared with the no antithrom- botic cohort (median of 7 in the anticoagulant cohort versus a median of 5 in the no antithrombotic cohort, P = 0.01).
-
- Antithrombotic regimens and reversal
Antithrombotic regimens used in the antiplatelet and anticoagulant cohorts along with reversal strategies are shown in Table 2. Reversal strategies listed in this table include all treatments, including those given in the ED. In the antiplatelet cohort, 72% of patients were on ASA monotherapy and most of the remaining patients were on dual an- tiplatelet therapy with ASA and clopidogrel (19%). A subset of patients had their antiplatelet efficacy tested via the VerifyNow Platelet function
Antithrombotic regimens and reversal strategies for the antiplatelet and anticoagulant cohorts.
assay (Accriva diagnostics, San Diego CA), which showed that ~75% of the patients were in the therapeutic range. Platelet transfusions and desmopressin were used for reversal in 14% and 21% of antiplatelet users, respectively; when combined, only 28% of the antiplatelet cohort received a reversal agent. In the anticoagulant cohort, 65% of patients were on warfarin, 33% were on a direct oral anticoagulant, and one pa- tient was on Low Molecular Weight Heparin. Most patients on warfarin were therapeutic based on International normalized ratio at ad- mission (median INR 2.3, IQR 1.8-2.9). 53% of the anticoagulant cohort also used Antiplatelet agents, typically ASA alone. Several different re- versal agents were used in the anticoagulant cohort, with plasma and prothrombin complex concentrate being the most common; overall, 74% of anticoagulant users were reversed.
-
- Factors associated with need for critical care interventions
Univariable and multivariable analysis of factors associated with need for critical care interventions is shown in Table 3. While isolated antiplatelet use had a modest protective effect against the need for crit- ical care interventions on univariable analysis, it had no association with critical care interventions in the multivariable analysis. The significant factors on multivariable analysis were coagulopathy (defined as an INR > 1.3 or platelet count <100,000, aOR 5.1 95% CI 2.6-10.5; P < 0.001), presence of any contusion on initial CT (aOR 3, 95% CI 1.3-6.7; P = 0.007), presence of any midline shift (aOR 3.4, 95% CI 1.6-7.4; P = 0.002), and maximum SDH thickness (aOR 2.4 for every 10 mm increase in thickness, 95% CI 1.4-4.3; P = 0.002). Of note, a sim- ilar model could be built with “anticoagulant use” being used in place of coagulopathy (Supplemental Table 1), but when both variables were placed in the same model, only coagulopathy was significant. A sensitiv- ity analysis was conducted where transfusions were excluded from the critical care intervention outcome, which showed similar results (Sup- plemental Table 2).
To understand how often patients would be undertriaged based on the above model, we examined the 28 patients (26%) in the antiplatelet cohort that would be lowest risk (absence of coagulopathy, lack of con- tusion, no associated midline shift, and a maximum SDH thickness less
Antithrombotic regimen
Antiplatelet cohort (N = 106)
Anticoagulant cohort (N = 49)
than or equal to 4 mm). Two of these patients needed critical care inter- ventions, but neither were for SDH related reasons (summarized in Table 4).
Apixaban N/A 11 (23)
Rivaroxaban N/A 5 (10)
Low molecular weight heparin N/A 1 (2)
ASA? 77 (72) 23 (47)
Clopidogrel? 8 (8) 1 (2)
Dual antiplatelet regimen 21 (20) 2 (4)
ASA/clopidogrel regimen 20 (19) 2 (4)
Laboratory evaluation of antithrombotic activity
INR in patients on warfarin N/A 2.3 (1.8-2.9)
-
- Factors associated with functional outcome
Factors associated with poor functional outcome at discharge are shown in Table 5. Only age (aOR 1.8 per 10 years, 95% CI 1.5-2.3;
p < 0.001), admission GCS (aOR 0.3 per 1 point, 95% CI 0.5-0.7;
p < 0.001), dementia history (aOR 4.2, 95%CI 1.5-15.1; p = 0.001),
and coagulopathy (aOR 3.5, 95% CI 1.7-7.5; p = 0.02) were significant
ASA activity (PRU)? N = 77; 480
N = 17; 479 (446,
predictors of functional outcome at discharge. Antiplatelet use was not
Clopidogrel Activity (PRU)??
(431-534)
N = 22; 183 (126,
214)
537)
N = 2; 174 (118, 231)
a significant predictor of functional outcome in our analysis.
- Discussion
vitamin K N/A 22 (45)
Fresh Frozen Plasma N/A 17 (35)
In this retrospective cohort study of SDHs in an academic health sys-
Prothrombin complex concentrate
N/A 17 (35)
tem, antiplatelet use was not associated with either the need for critical care interventions or poor functional outcome among noncomatose pa-
Platelet Transfusion 15 (14) 5 (10)
Desmopressin 22 (21) 2 (4)
Other reversal agent+ N/A 4 (8)
Reversal strategies include those given at any time, not just those given after hospital ad- mission. Data are n (%) or median (IQR). In cases where incomplete data is available, the number of patients for which data was available is listed as N = Abbreviations: PRU = platelet reactivity units, INR = International normalized ratio.
* Therapeutic is defined as <550 PRU.
?? Therapeutic is defined as <210 PRU.
+ Coagulation factor Xa (recombinant) in 3 cases and Factor IX complex in 1.
? This does not include patients on dual antiplatelet regimens, which are summarized separately.
tients with SDHs. Instead, a significant portion of antiplatelet-associated SDHs had a benign hospital course and outcome, suggesting that iso- lated antiplatelet use should not be used as a criterion for admission to an ICU. Among the lowest risk antiplatelet users (patients without co- agulopathy, contusions, or midline shift, and a maximum SDH thickness
<= 4 mm), the only critical care interventions that occurred were unre- lated to the SDH. As antiplatelet-associated SDHs are expected to con- tinue rising in incidence, [2] future prospective studies should build upon these data to determine which antiplatelet-associated SDHs can be safely monitored with floor-level care or observational protocols.
Multivariable analysis of need for critical care interventions.
No critical care intervention |
Critical care intervention |
Univariable P |
Univariable OR |
Multivariable P |
Adjusted odds ratio |
|
(N = 200) |
(N = 81) |
value |
(95% CI) |
value |
(95% CI) |
|
Gender (Male) |
133 (66) |
50 (62) |
0.35 |
0.8 (0.5-1.3) |
||
Age (per 10 years) |
70.4 (55, 81.9) |
72.3 (56.7, 81.9) |
0.34 |
1.1 (0.93-1.2) |
0.77 |
1.02 (0.9-1.2) |
Initial GCS |
15 (14, 15) |
15 (14, 15) |
0.002 |
0.5 (0.4-0.8) |
||
Cirrhosis |
3 (2) |
7 (9) |
0.01 |
5.9 (1.6-27.7) |
||
Coagulopathy |
26 (14) |
36 (45) |
<0.001 |
5 (2.8-9.3) |
<0.001 |
5.1 (2.6-10.5) |
Chronic kidney disease |
15 (8) |
8 (10) |
0.85 |
1.1 (0.4-2.7) |
||
3 (2) |
2 (2) |
0.6 |
1.6 (0.2-9.6) |
|||
Alcohol abuse |
20 (10) |
12 (15) |
0.33 |
1.5 (0.7-3.1) |
||
Associated contusion |
26 (13) |
18 (22) |
0.07 |
1.9 (0.9-3.7) |
0.008 |
3 (1.3-6.7) |
Associated SAH |
63 (32) |
22 (27) |
0.44 |
0.8 (0.4-1.4) |
||
Skull fracture |
48 (24) |
9 (11) |
0.03 |
0.4 (0.2-0.9) |
||
Acute SDH on imaging |
161 (80) |
55 (68) |
0.04 |
0.5 (0.3-0.97) |
||
Presence of midline shift |
32 (16) |
41 (51) |
<0.001 |
5 (2.9-9.1) |
0.002 |
3.4 (1.6-7.4) |
No or minimal trauma |
140 (70) |
68 (84) |
0.02 |
2.3 (1.2-4.6) |
||
Anticoagulant use |
23 (12) |
26 (32) |
<0.001 |
3.4 (1.8-6.5) |
||
Location along convexity |
151 (76) |
71 (88) |
0.001 |
2.7 (1.3-6.3) |
||
Isolated antiplatelet use |
87 (44) |
19 (23) |
0.003 |
0.4 (0.2-0.7) |
||
Thickness (per 10 mm) |
4.8 (3.1, 8) |
9 (5.5, 17) |
<0.001 |
3.6 (2.3-5.8) |
0.002 |
2.4 (1.4-4.3) |
Initial creatinine |
0.9 (0.8-1.1) |
1 (0.8-1.2) |
0.94 |
0.99(0.8-1.2) |
||
Initial sodium |
139 (137, 141) |
139 (136, 141) |
0.99 |
1 (0.94-1.06) |
Data are n (%) or median (IQR). Bolded variables are statistically significant at P < 0.05 on multivariable analysis. Abbreviations: SDH = subdural hematoma, GCS = Glasgow Coma Scale, SAH=Subarachnoid hemorrhage, OR = Odds ratio, CI=Confidence Interval), mm = millimeter.
Studies could also explore whether this can reduce costs [32] and/or im- prove patient satisfaction. [33]
Prior studies have attempted to identify noncomatose SDH patients at low risk for clinical deterioration, but restricted their analysis to pa- tients with isolated SDHs. [8,34] Thus, in addition to our focus on anti- thrombotic users, our results expand upon prior work by including patients with both isolated and nonisolated SDHs. Similar to prior stud- ies, we found that SDH thickness and midline shift are strongly associ- ated with higher risk patients even in this more heterogeneous population. Among nonisolated SDHs, we found that patients with the combination of subarachnoid hemorrhage and SDH were not more likely to need critical care interventions in our study, while pa- tients with a contusion and an SDH were at a higher risk. This suggests that some patients with SDHs and SAHs can be safely triaged to a lower level of care (assuming they are otherwise low risk), while patients with contusions in addition to an SDH may benefit from more aggressive triaging. Since SDHs commonly occur with additional intracranial le- sions (~40% of patients in this study had lesions in addition to their SDH), including both isolated and nonisolated SDHs may improve the generalizability of future decision support tools.
There has also been prior work examining the impact of antiplatelet use on SDHs, although these studies have arrived at conflicting conclu- sions in the setting of heterogeneous designs and outcome measures. Isolated ASA use has been associated with no effect or even a protective effect on SDH mortality in some cohorts, [14,35] concordant with our data showing no impact of isolated antiplatelet use on functional out- come. Other studies have found that some antiplatelets (particularly
Characteristics of antiplatelet users who needed critical care interventions despite being low risk.
Demographic Critical Care Comments Intervention
clopidogrel) are associated with an increased risk of SDH expansion, but not necessarily with worse outcomes. [34,36] We did not observe any association between antiplatelet use and SDH expansion, although this study was not powered to assess SDH expansion as an outcome and this finding should be interpreted with caution. We further found that symptomatic SDH expansion was exceptionally rare in our cohort. Taken together, our data would suggest that any possible association between antiplatelet use and SDH expansion is not large enough to sig- nificantly increase the need for critical care interventions or the risk of a poor outcome, consistent with recent work. [36] Thus, while it may be reasonable to include antiplatelet use in protocols for repeat head imag- ing in SDH patients, [7] these repeat images can likely be performed without admission to an ICU in many cases. The safety of this approach will need to be evaluated prospectively.
There are other key differences between our work and prior studies exploring antiplatelet use in tICH. First, unlike prior studies, we made the pragmatic choice to focus on isolated antiplatelet use rather than general antiplatelet use as it can facilitate more straightforward triage decisions. This means that patients on both antiplatelets and anticoagu- lants were assigned to the anticoagulant cohort. Second, there is the po- tential for variation in the exact antiplatelet regimens used across different studies. This could lead to significant heterogeneity and ex- plain the conflicting results in the literature, as regimens containing clopidogrel may be associated with more severe SDHs than ASA mono- therapy. [34,35] We did not observe this trend (in fact, dual antiplatelet regimens with ASA and clopidogrel were quite common and were not associated with poor outcome), but our sample size is not large enough to conduct meaningful subgroup analyses of the antiplatelet cohort. More work will be needed to determine if some isolated antiplatelet regimens are associated with more aggressive Clinical courses among noncomatose SDH patients; while our results provide evidence that iso- lated antiplatelet use alone should not dictate triage decisions for SDH patients, there may be particular antiplatelet regimens that require spe- cial attention.
Similar to prior work, anticoagulant users had more severe SDHs and
93-year-old female
65-year-old male
Isolated Blood Transfusion
Mechanical ventilation
History of chronic anemia, developed
acute-on-chronic asymptomatic anemia from a superficial thigh hematoma 3 days after admission
Complex mandibular fracture in addition to SDH, required a tracheostomy to promote healing and facilitate staged operation
worse functional outcomes in our study; [9,37] however, our final mul- tivariable models for both outcome and critical care interventions used a previously defined “coagulopathy” variable and did not include anti- coagulant use. This was the result of the coagulopathy variable being a more robust predictor than anticoagulant use, likely due to the high pro- portion of patients on warfarin in our study (65%) and the fact that it
Multivariable analysis of functional outcome. Poor functional outcome is defined as a modified Rankin Scale of 3 or higher at discharge. Bolded variables are statistically significant at P < 0.05 on multivariable analysis.
Good functional outcome |
Poor functional outcome |
P Value |
Univariable OR (95% |
Multivariable P |
aOR (95% CI) |
|
(N = 147) |
(N = 144) |
CI) |
value |
|||
Gender (Male) |
98 (72) |
85 (59) |
0.04 |
0.6 (0.34-0.98) |
||
Age (per 10 years) |
58.8 (44, 71.6) |
78.5 (69.3, 87) |
<0.001 |
1.8 (1.5-2.2) |
<0.001 |
1.8 (1.5-2.3) |
Initial GCS |
15(15, 15) |
15 (14, 15) |
<0.001 |
0.3 (0.2-0.5) |
<0.001 |
0.3 (0.1-0.5) |
Cirrhosis |
6 (4) |
4 (3) |
0.4 |
0.6 (0.1-2.1) |
||
Coagulopathy |
15 (12) |
47 (33) |
<0.001 |
3.7 (1.9-7.2) |
0.001 |
3.5 (1.7-7.5) |
Chronic kidney disease |
4 (3) |
19 (13) |
0.008 |
4.4 (1.6-15.7) |
||
End stage renal disease |
1 (1) |
4 (3) |
0.3 |
3.6 (0.5-71.2) |
||
History of Dementia |
4 (3) |
41 (28) |
<0.001 |
11.6 (4.5-39.7) |
0.01 |
4.2 (1.5-15.1) |
Alcohol abuse |
21 (15) |
11 (8) |
0.03 |
0.4 (0.2-0.9) |
||
Associated Contusion |
18 (13) |
26 (18) |
0.3 |
1.4 (0.7-2.8) |
||
Associated SAH |
35 (26) |
50 (35) |
0.2 |
1.4 (0.9-2.5) |
||
Associated skull fracture |
37 (27) |
20 (14) |
0.02 |
0.5 (0.3-0.9) |
||
Acute SDH on imaging |
107 (78) |
109 (76) |
0.9 |
0.9 (0.5-1.7) |
||
Convexity location |
101 (74) |
121 (84) |
0.02 |
2 (1.1-3.8) |
||
Thickness (per 10 mm) |
4.5 (3, 8.3) |
6.7 (4.2, 13) |
0.002 |
1.9 (1.3-3) |
||
Presence of midline shift |
29 (21) |
44 (31) |
0.2 |
1.5 (0.8-2.5) |
||
No or minimal trauma |
85 (62) |
123 (85) |
<0.001 |
3.5 (2-6.4) |
||
Anticoagulant use |
8 (6) |
41 (28) |
<0.001 |
6 (2.8-14.4) |
||
Isolated antiplatelet use |
45 (33) |
61 (42) |
0.07 |
1.6 (0.9-2.7) |
||
Initial sodium |
139 (137, 141) |
139 (136, 141) |
0.4 |
1 (0.9-1) |
||
Initial creatinine |
0.9 (0.8, 1.1) |
0.97 (0.8, 1.3) |
0.7 |
1 (0.9-1.2) |
Data are n (%) or median (IQR). Abbreviations: SDH = subdural hematoma, GCS = Glasgow Coma Scale, CKD=Chronic Kidney Disease, SAH=Subarachnoid hemorrhage, OR = Odds ratio, CI = Confidence Interval, mm = millimeter.
captures other coagulopathic patients in addition to warfarin users. [25] There are advantages and disadvantages to this variable. A major advan- tage is that it can facilitate triage decisions by using widely available lab tests that can capture multiple types of coagulopathy. The disadvantage is that it will not capture patients on direct oral anticoagulants, which comprise an increasingly large proportion of anticoagulant users in the United States. [38] At the time of this study, our institution did not routinely perform Anti-Xa assays in patients suspected of using factor Xa inhibitors, but future work should seek to integrate these tests into decision-making. An alternative would be to define coagulopathy using thromboelastography, although this modality remains available only at a small number of centers and has yet to be validated in trau- matic brain injury. [39]
While imaging characteristics like SDH thickness and midline shift were strong predictors of critical care interventions and have been ob- served to influence outcome in other cohorts, [40] they were not strong predictors of functional outcome on in our cohort. This is likely the re- sult of multiple factors. First, we enrolled a heterogeneous, real-world cohort of SDH patients with varying degrees of trauma presenting to an urban ED; in such a setting, premorbid functional status influences outcome more than imaging features. [41] Second, our cohort was re- stricted to patients who presented with relatively preserved exams, as these are the patients most likely to be overtriaged; [17] this will likely exclude the SDH patients with the most severe imaging features who are likely to have the worst outcomes. Finally, we only had data on func- tional outcome at discharge available. It is well known that even mildly injured TBI patients recover their functional status slowly over time,
[42] and more subtle effects of SDH thickness may become more appar- ent with time.
There are several limitations to this study. First, this work was per- formed at an academic, two-hospital system comprised of a level 1 and a level 3 trauma center. This system has standardized approaches to antithrombotic reversal and other aspects of TBI care and acts as a major referral center for complex cases in the region. Thus, our popula- tion is likely biased towards more clinically complex, traumatic SDHs; further study in non-trauma centers and smaller community hospitals will be critical to understand whether these results hold true in a broader SDH population. Second, we used a previously defined compos- ite outcome of critical care interventions for our primary outcome,
which is derived from expert opinion [27] and focuses on the interven- tions required for clinical changes, rather than the clinical changes themselves. Thus, we cannot entirely account for other potential bene- fits of ICU care (for instance neuromonitoring), which may only be available in the ICU in some settings and may actually prevent the need for downstream interventions. This limitation would not invali- date the central findings of our study, however, as it would impact all risk factors for critical care interventions equally, with no differential ef- fect on antithrombotic use in particular. Finally, a small portion of our antiplatelet cohort did receive antiplatelet-directed reversal agents, in- cluding platelet transfusions and desmopressin, which may bias our re- sults towards the null, and decrease the generalizability of these results.
- Conclusion
Many SDH patients on isolated antiplatelet regimens will have a be- nign clinical course and outcome. Isolated antiplatelet use should not be used as a criterion to determine the need for ICU admission among noncomatose SDH patients.
Previous presentations
Neurocritical care Society Annual Meeting, Neurocritical Care Society
Vancouver, BC, Canada October 17, 2019 Poster session
Funding
DR reports funding assistance from the National Institute of Neuro- logical Disorders and Stroke, T32NS047996-13. The funders had no role in conducting, analyzing, or reporting these results.
Credit authorship statement David Robinson: Conceptualization; Data curation; Formal analysis;
Methodology; Writing-original draft; Writing – review and editing. Lo-
gan Pyle: Conceptualization; Data curation; Validation; Investigation;
Writing- review and editing. Brandon Foreman: Conceptualization; Formal Analysis; Methodology; Writing- review and editing. Laura B. Ngwenya: Conceptualization; Methodology; Writing- review and editing. Opeolu Adeoye: Conceptualization; Formal Analysis; Method- ology; Writing- review and editing. Daniel Woo: Conceptualization; Formal Analysis; Methodology; Writing- review and editing. Natalie Kreitzer: Conceptualization; Data curation; Formal analysis; Methodol- ogy; Writing- review and editing.
Declaration of competing interest
DR reports no conflict of interest. LP reports no conflict of interest. BF reports no conflict of interest. LBN reports no conflict of interest. OA re- ports no conflict of interest. DW reports no conflict of interest. NK re- ports no conflict of interest.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2021.03.035.
References
- Frontera JA, Egorova N, Moskowitz AJ. National trend in prevalence, cost, and dis- charge disposition after subdural hematoma from 1998-2007. Crit Care Med. 2011;39:1619-25.
- Balser D, Farooq S, Mehmood T, Reyes M, Samadani U. Actual and projected inci- dence rates for chronic subdural hematomas in United States veterans administra- tion and civilian populations. J Neurosurg. 2015;123:1209-15.
- Massaro F, Lanotte M, Faccani G, Triolo C. One hundred and twenty-seven cases of acute subdural haematoma operated on. Correlation between CT scan findings and outcome. Acta Neurochir (Wien). 1996;138:185-91.
- Gaist D, Garcia Rodriguez LA, Hellfritzsch M, Poulsen FR, Halle B, Hallas J, et al. Asso- ciation of antithrombotic drug use with subdural hematoma risk. JAMA. 2017;317: 836-46.
- Connolly BJ, Pearce LA, Hart RG. vitamin K antagonists and risk of subdural hema- toma: meta-analysis of randomized clinical trials. Stroke. 2014;45:1672-8.
- Orlando A, Levy AS, Carrick MM, Tanner A, Mains CW, Bar-Or D. Epidemiology of mild traumatic brain injury with intracranial hemorrhage: focusing predictive models for neurosurgical intervention. World Neurosurg. 2017;107:94-102.
- Joseph B, Friese RS, Sadoun M, Aziz H, Kulvatunyou N, Pandit V, et al. The big (brain injury guidelines) project: defining the management of traumatic brain injury by acute care surgeons. J Trauma Acute Care Surg. 2014;76:965-9.
- Orlando A, Levy AS, Rubin BA, Tanner A, Carrick MM, Lieser M, et al. Isolated sub- dural hematomas in mild traumatic brain injury. Part 2: a preliminary clinical deci- sion support tool for neurosurgical intervention. J Neurosurg. 2018:1-8.
- Won SY, Dubinski D, Bruder M, Cattani A, Seifert V, Konczalla J. Acute subdural he- matoma in patients on Oral anticoagulant therapy: management and outcome. Neurosurg Focus. 2017;43:E12.
- Turgeon AF, Lauzier F, Simard JF, Scales DC, Burns KE, Moore L, et al. Mortality asso- ciated with withdrawal of life-sustaining therapy for patients with severe traumatic brain injury: a Canadian multicentre cohort study. CMAJ. 2011;183:1581-8.
- Grandhi R, Harrison G, Voronovich Z, Bauer J, Chen SH, Nicholas D, et al. Preinjury warfarin, but not antiplatelet medications, increases mortality in elderly traumatic brain injury patients. J Trauma Acute Care Surg. 2015;78:614-21.
- Ohm C, Mina A, Howells G, Bair H, Bendick P. Effects of antiplatelet agents on out- comes for elderly patients with traumatic intracranial hemorrhage. J Trauma. 2005;58:518-22.
- McMillian WD, Rogers FB. Management of prehospital antiplatelet and anticoagu- lant therapy in Traumatic head injury: a review. J Trauma. 2009;66:942-50.
- Alford EN, Rotman LE, Erwood MS, Oster RA, Davis MC, Pittman HBC, et al. Develop- ment of the subdural hematoma in the elderly (she) score to predict mortality. J Neurosurg. 2019:1-7.
- Ratcliff JJ, Adeoye O, Lindsell CJ, Hart KW, Pancioli A, McMullan JT, et al. Ed disposi- tion of the Glasgow coma scale 13 to 15 traumatic brain injury patient: analysis of the transforming research and clinical knowledge in tbi study. Am J Emerg Med. 2014;32:844-50.
- Nishijima DK, Sena MJ, Holmes JF. Identification of low-risk patients with traumatic brain injury and intracranial hemorrhage who do not need intensive care unit ad- mission. J Trauma. 2011;70:E101-7.
- Bonow RH, Quistberg A, Rivara FP, Vavilala MS. Intensive care unit admission pat- terns for mild traumatic brain injury in the USA. Neurocrit Care. 2019;30:157-70.
- Borczuk P, Van Ornam J, Yun BJ, Penn J, Pruitt P. Rapid discharge after interfacility transfer for mild traumatic intracranial hemorrhage: frequency and associated fac- tors. West J Emerg Med. 2019;20:307-15.
- Sorensen MJ, von Recklinghausen FM, Fulton G, Burchard KW. Secondary overtriage: the burden of unnecessary interfacility transfers in a rural Trauma system. JAMA Surg. 2013;148:763-8.
- Rhame K, Le D, Ventura A, Horner A, Andaluz N, Miller C, et al. Management of the mild traumatic brain injured patient using a multidisciplinary observation unit pro- tocol. Am J Emerg Med. 2020. https://doi.org/10.1016/j.ajem.2020.06.088 Published online July 14, 2020.
- Nishijima DK, Shahlaie K, Echeverri A, Holmes JF. A clinical decision rule to predict adult patients with traumatic intracranial haemorrhage who do not require inten- sive care unit admission. Injury. 2012;43:1827-32.
- Kuhn EN, Erwood MS, Oster RA, Davis MC, Zeiger HE, Pittman BC, et al. Outcomes of subdural hematoma in the elderly with a history of minor or no previous trauma. World Neurosurg. 2018;119:e374-82.
- Alves JL, Santiago JG, Costa G, Mota Pinto A. A standardized classification for sub- dural hematomas- I. Am J Forensic Med Pathol. 2016;37:174-8.
- Connolly SJ, Crowther M, Eikelboom JW, Gibson CM, Curnutte JT, Lawrence JH, et al. Full study report of andexanet alfa for bleeding associated with factor xa inhibitors. N Engl J Med. 2019;380:1326-35.
- Wafaisade A, Lefering R, Tjardes T, Wutzler S, Simanski C, Paffrath T, et al. Acute co- agulopathy in isolated blunt traumatic brain injury. Neurocrit Care. 2010;12:211-9.
- Nishijima DK, Haukoos JS, Newgard CD, Staudenmayer K, White N, Slattery D, et al. Variability of icu use in adult patients with minor traumatic intracranial hemor- rhage. Ann Emerg Med. 2013;61 509-517.e4.
- Guidelines for intensive care unit admission, discharge, and triage. Task force of the American College of Critical Care medicine, Society of Critical Care Medicine. Crit Care Med. 1999;27:633-8.
- R Core Team. A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2018 URL https://www.R-project.org/ ver- sion 3.5.2.
- Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and cox regression. Am J Epidemiol. 2007;165:710-8.
- Mickey RM, Greenland S. The impact of confounder selection criteria on effect esti- mation. Am J Epidemiol. 1989;129:125-37.
- Howard MA, Gross AS, Dacey RGJ, Winn HR. Acute subdural hematomas: an age- dependent clinical entity. J Neurosurg. 1989;71:858-63.
- Tang A, Hashmi A, Pandit V, Joseph B, Kulvatunyou N, Vercruysse G, et al. A critical analysis of secondary overtriage to a level I trauma center. J Trauma Acute Care Surg. 2014;77:969-73.
- Wu C, Melnikow J, Dinh T, Holmes JF, Gaona SD, Bottyan T, et al. Patient admission preferences and perceptions. West J Emerg Med. 2015;16:707-14.
- Pruitt P, Ornam JV, Borczuk P. A decision instrument to identify isolated traumatic subdural hematomas at low risk of Neurologic deterioration, surgical intervention, or radiographic worsening. Acad Emerg Med. 2017;24:1377-86.
- Gaist D, Garcia Rodriguez L, Hellfritzsch M, et al. Association of antithrombotic drug use with subdural hematoma risk. JAMA. 2017;317:836-46.
- Mathieu F, Guting H, Gravesteijn B, Monteiro M, Glocker B, Kornaropoulos EN, et al. Impact of antithrombotic agents on radiological lesion progression in acute trau- matic brain injury: a center-tbi propensity-matched cohort analysis. J Neurotrauma. 2020;37:2069-80.
- Bershad EM, Farhadi S, Suri MF, Feen ES, Hernandez OH, Selman WR, et al. Coagulop- athy and inhospital deaths in patients with acute subdural hematoma. J Neurosurg. 2008;109:664-9.
- Barnes GD, Lucas E, Alexander GC, Goldberger ZD. National trends in ambulatory oral anticoagulant use. Am J Med. 2015;128 1300-5.e2.
- Kreitzer NP, Bonomo J, Kanter D, Zammit C. Review of thromboelastography in neurocritical care. Neurocrit Care. 2015;23:427-33.
- Leitgeb J, Mauritz W, Brazinova A, Janciak I, Majdan M, Wilbacher I, et al. Outcome after severe brain trauma due to acute subdural hematoma. J Neurosurg. 2012; 117:324-33.
- Weimer JM, Gordon E, Frontera JA. Predictors of functional outcome after subdural hematoma: a prospective study. Neurocrit Care. 2017;26:70-9.
- Nelson LD, Temkin NR, Dikmen S, Barber J, Giacino JT, Yuh E, et al. Recovery after mild traumatic brain injury in patients presenting to us level I trauma centers: a transforming research and clinical knowledge in traumatic brain injury (track-tbi) study. JAMA Neurol. 2019;76:1049-59.