clinical article J Neurosurg 124:750–759, 2016

Timing of the resumption of antithrombotic agents following surgical evacuation of chronic subdural hematomas: a retrospective cohort study

Daipayan Guha, MD,1,2 Shona Coyne, BSc,3 and R. Loch Macdonald, MD, PhD1,2

1Division of Neurosurgery, St. Michael’s Hospital; 2Keenan Research Centre for Biomedical Science and the Li Ka Shing Knowledge Institute of St. Michael’s Hospital, Department of Surgery, University of Toronto, Ontario, Canada; and 3Faculty of Medicine, The Royal College of Surgeons in Ireland, Dublin, Leinster, Ireland

Objective Antithrombosis (AT), defined here as either antiplatelets or anticoagulants, is a significant risk factor for the development of chronic subdural hematomas (cSDHs). Resuming AT following the evacuation of cSDH is a highly variable practice, with scant evidence in the literature for guidance. Here, a retrospective analysis of a cohort of patients from a single institution undergoing surgical drainage of cSDH was performed to evaluate postoperative complications and determine the optimal timing of the resumption of common antithrombotic agents. Methods This retrospective analysis was performed on 479 patients undergoing surgical evacuation of cSDH at St. Michael’s Hospital over a 5-year period (2007–2012). The collected variables included the type of AT agent, indications for AT, timing and type of postoperative complications, and the restart intervals for the AT agents, when available. Post- operative complications were classified as major hemorrhages, minor hemorrhages, or thromboembolic events. Results Among all 479 study patients, 71 experienced major hemorrhage (14.8%), 110 experienced minor hemor- rhage (23.0%), and 8 experienced thromboembolism (1.67%) postoperatively. Patients on any type of preoperative AT regimen were at a higher risk of major hemorrhage (19.0% vs 10.9%; OR 1.93; 95% CI 1.15–2.71; p = 0.014). The type of AT agent did not affect the frequency of any postoperative complications. Patients on any preoperative AT regimen expe- rienced earlier postoperative major hemorrhages (mean 16.2 vs 26.5 days; p = 0.052) and thromboembolic events (mean 2.7 vs 51.5 days; p = 0.036) than those patients without a history of AT; the type of AT agent did not affect timing of com- plications. Patients who were restarted on any AT therapy postoperatively were at decreased risk of major rebleeding following resumption than those patients who were not restarted (OR 0.06; 95% CI 0.02–0.2; p < 0.01). Conclusions Patients with a history of preoperative AT experienced thromboembolic complications significantly ear- lier than those patients without AT, which peaked at 3 days postoperatively with no increase in hemorrhage risk when AT was restarted. Cursory evidence is presented that shows resuming AT early following the surgical evacuation of cSDH at 3 days postoperatively may be safe. However, much larger prospective studies are required prior to providing any defini- tive recommendations regarding the optimal timing and method of resumption of individual agents. http://thejns.org/doi/abs/10.3171/2015.2.JNS141889 KEY WORDS anticoagulation; antiplatelet; antithrombosis; chronic subdural hematoma; vascular disorders

hronic subdural hematoma (cSDH) is a disease States population older than 65 years grows from 12% in of the elderly and demonstrates an incidence of 2003 to a projected 20% by 2030.17 approximately 3.4 patients per 100,000 persons Other identified risk factors for the development of Cyounger than 65 years of age and 8 to 58 patients per cSDH include male sex, history of falls, chronic alcohol 100,000 persons older than 65 years.1,4,9 The prevalence of use, and antiplatelet or anticoagulant therapy, with war- cSDH is expected to rise as the percentage of the United farin increasing the relative risk of developing cSDH by

ABBREVIATIONS ADP = adenosine diphosphate; ASA = acetylsalicylic acid; AT = antithrombosis; CAD = coronary artery disease; cSDH = chronic subdural hematoma; DVT = deep venous thrombosis; INR = international normalized ratio; LMWH = low-molecular-weight heparin; TIA = transient ischemic attack. submitted August 16, 2014. accepted February 27, 2015. include when citing Published online September 11, 2015; DOI: 10.3171/2015.2.JNS141889.

750 J Neurosurg Volume 124 • March 2016 ©AANS, 2016

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC Antithrombosis resumption in chronic subdural hemorrhage

42.5 times.2,3,25 The recurrence rates for cSDH in the re- subdural hematoma on CT scan), and those treated surgi- cent literature range from 9% to 33%.10,21,22 The predic- cally for the first time at this institution. Surgical drainage tors of recurrence are highly variable but include patient was performed via a single bur hole or craniotomy at the characteristics such as age and bilateral cSDH, radio- surgeon’s discretion, as dictated by the patient’s clinical graphic variables such as preoperative hematoma width and imaging findings. A minimum of 1 documented clinic and morphology, and surgical and perioperative factors follow-up visit was required. such as operative technique and choice of drainage sys- Patients with nonhead injuries that might otherwise tem.7,10,21–23,26,28,29 promote immobility and its associated thromboembolic The use of antithrombosis (AT), which is defined here complications, traumatic or aneurysmal subarachnoid as either antiplatelets or anticoagulants, is expanding with hemorrhage, iatrogenic SDH, underlying coagulopathies, a progressive rise in the prevalence of patients with ath- or platelet disorders (e.g., liver disease, myeloproliferative erosclerotic risk factors,12 as well as thrombogenic cardiac disorders) were excluded from the search. arrhythmias which typically affect the elderly.14 Acute The final search yielded 479 patients with spontaneous coronary syndromes affect more than 750,000 Ameri- or posttraumatic cSDH. A clear history of trauma or fall cans per year,32 with dual antiplatelet therapy being the could not be identified in all patients. current standard of care.24 Atrial fibrillation afflicts more than 2.4 million Americans, and warfarin or targeted an- Data Collection ticoagulants are used to treat those at high risk of arte- The collected variables included patient age, sex, num- rial embolism.8 The use of any AT is known to predispose ber and type of surgical procedures, type of preoperative the patient to developing cSDH, and it is well accepted AT, indications for AT, time of AT discontinuation, rever- that these agents should be stopped at presentation and sal of AT, time of postoperative resumption of AT, time of reversed in the case of warfarin.10,15 The impact of AT on postoperative complications, and follow-up interval. Pre- recurrence following surgical drainage, however, is con- operative patient comorbidities and functional status were troversial.7,11,15,25,27,29 Balancing the risk of rebleeding with not always readily identifiable and, hence, were not includ- potential thromboembolic complications in these patients ed in our data set. CT scans are not kept on the hospital is an ongoing dilemma with scant evidence available to imaging server for more than 2 years, and hematoma size guide if and when therapeutic AT should be resumed post- was not always specified in the radiology reports; hence, operatively. 6 data on preoperative hematoma size and morphology were Here, we retrospectively analyzed a cohort of 479 pa- not readily identifiable. tients who were treated surgically for cSDH at a single Antithrombotics were defined to include both antico- institution. We collected data on postoperative hemor- agulant and antiplatelet agents. The antiplatelets assessed rhagic and thromboembolic complications, when avail- in this study include acetylsalicylic acid (ASA; 81 mg or able, along with preoperative AT agents and indications. 325 mg daily), clopidogrel (75 mg daily), and dipyridam- By analyzing one of the largest cohorts of cSDH patients ole (200 mg twice daily); anticoagulants included warfa- in the literature, all of whom were from a single institu- rin (dosed to the therapeutic international normalized ra- tion, underwent similar surgical techniques performed by tio [INR] target), low-molecular-weight heparin (LMWH; 6 staff neurosurgeons, and have rates of recurrence and dalteparin or enoxaparin administered at a weight-based thromboembolism available, we hope to provide more ro- therapeutic dosage), and targeted coagulation inhibitors bust evidence to guide the resumption of therapeutic AT (dabigatran, fondaparinux, rivaroxaban, or apixaban). as well as direct further definitive studies. The indications for AT were categorized as primary prevention (the presence of hypertension, diabetes, hy- Methods percholesterolemia, and/or smoking history), atrial fibril- lation, prior transient ischemic attack (TIA), or stroke, Patient Selection documented coronary artery disease (CAD; history of The study subjects were collected from a database of acute coronary syndrome or angiographic coronary dis- patients admitted to St. Michael’s Hospital, Toronto, Can- ease, with or without surgical or endovascular revascular- ada, between 2000 and 2012. We searched the database ization), mechanical heart valve placement, arterial stent for the following codes included in the ICD-10: codes placement (coronary, carotid, or peripheral), active periph- I62 (other nontraumatic intracranial hemorrhage), S06.5 eral arterial/venous thrombosis or pulmonary embolism, (traumatic subdural hemorrhage), S06.6 (traumatic sub- venous thrombosis prophylaxis, or another indication. Re- arachnoid hemorrhage), S06.8 (other intracranial injuries), versal of AT was assessed only for patients on a regimen S06.9 (intracranial injury, unspecified).30 The search was of warfarin with any combination of vitamin K, fresh- restricted to patients with electronic records and limited to frozen plasma, or prothrombin complex concentrate. The those patients admitted after January 1, 2007. The records reversal of other agents was attempted inconsistently, for were then manually filtered to identify eligible patients. instance with pooled platelet transfusions for antiplatelets. Study approval was obtained from the St. Michael’s Hos- Given the lack of reliable efficacy, however, the reversal of pital Research Ethics Board. agents other than warfarin was not assessed here. Patients eligible for study inclusion included those old- Postoperative complications were classified as ma- er than 18 years of age with their first presentation to St. jor hemorrhage, minor hemorrhage, or thromboembolic Michael’s Hospital, those with an isolated cSDH (defined events. Major hemorrhage denotes a recurrent intracranial as an untreated hematoma for > 3 weeks or a hypodense hemorrhage requiring either repeat operative drainage or

J Neurosurg Volume 124 • March 2016 751

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC D. Guha, S. Coyne, and R. L. Macdonald readmission to the hospital for observation. Minor hem- TABLE 1. Study patient clinical characteristics orrhage refers to residual subdural hematomas seen on Variable Value* outpatient postoperative imaging that necessitated further radiographic follow-up, but not readmission to hospital or Mean age ± SD in yrs reoperation. Thromboembolic events were categorized All patients 72.3 ± 13.6 as peripheral venous thrombosis (including pulmonary Preop AT 76.4 ± 12.4 embolism), peripheral arterial thrombosis, myocardial in- Restarted 76.0 ± 11.2 farction, or TIA/stroke. Not restarted 77.1 ± 14.1 Our institution does not have guidelines for the routine administration of postoperative venous thromboprophy- No preop AT 68.4 ± 14.3 laxis to our patient cohort. Most patients were started on Sex prophylactic subcutaneous heparin or LMWH at appro- Male 341 (71.2) priate weight-based dosing on postoperative Day 2 or 3; Female 138 (28.8) however, there was significant variability across patients Total 479 (100) and staff neurosurgeons and, hence, this was not captured Side of cSDH in our data set. Left 175 (36.5) Data Analysis Right 184 (38.4) The incidence of postoperative complications and Bilateral 120 (25.1) timing were compared among patients on any AT regi- Initial surgical management men, patients off an AT regimen, and patients receiving Bur hole 409 (84.9) ASA (81 mg or 325 mg daily), clopidogrel, or warfarin. Craniotomy The number of patients receiving other AT agents was too All patients 72 (14.9) small to allow a stratified analysis. We found no differ- Preop AT 38 (7.9) ences in the complication incidence or timing between patients on preoperative ASA versus clopidogrel versus No preop AT 34 (7.1) warfarin, and, hence, subsequent analyses following the Craniectomy 1 (0.2) resumption of AT were performed by pooling the patients * Values are presented as the number of patients (%) unless stated otherwise. on any of these 3 agents preoperatively in order to improve the statistical power. Differences in complication timing between multiple (i.e., > 2) AT agents were assessed using Kruskal-Wallis single bur hole drainage of their cSDH; a small subset un- 1-way ANOVA with Dunn’s multiple-comparisons tests. derwent mini-craniotomy as their initial operation due to Comparisons between 2 groups of patients were made an acute-on-chronic or highly membranous appearance on using the Mann-Whitney U-tests. Differences in compli- CT imaging. The proportion of patients requiring initial cation frequencies among multiple (> 2) antithrombot- mini-craniotomy was not affected by the use of preopera- ics were analyzed using the chi-square tests; differences tive AT. One patient underwent mini-craniotomy without between 2 groups of patients were compared using the replacement of the bone flap and hence was classified as a Fisher exact tests. Odds ratios and confidence intervals craniectomy. A closed subdural drainage system to thumb were computed for each test as appropriate. In order to suction, using a Jackson-Pratt device, was placed in all 479 obtain the cumulative hazard rates, differences in time- patients. The mean follow-up time was 3.1 months. dependent outcomes (time to postoperative hemorrhage or thromboembolism) based on the timing of AT resumption AT Profiles were assessed by Cox proportional hazards regression us- Preoperative and postoperative AT use in our cohort is ing maximum likelihood methods. The significance level shown in Fig. 1. In total, 231 of 479 study patients (48%) in all patients was set at p ≤ 0.05. were receiving AT therapy on admission: 127 patients All statistical analyses were performed using Graph- were receiving 81 mg ASA (27%), 9 were receiving 325 Pad Prism (GraphPad Software, Inc.). mg ASA (2%), 28 were receiving clopidogrel (6%), 86 were receiving warfarin (18%), 8 were receiving LMWH (2%), Results 1 was receiving dabigatran (0.2%), and 6 were receiving dipyridamole (1%). Thirty-two patients were receiving Patient Characteristics multiple agents, and the majority were receiving either The study patient characteristics are shown in Table 1. ASA+clopidogrel or ASA+warfarin; the former were typi- Patients had a mean age of 72.3 years, which was high- cally on an AT regimen for recently implanted vascular er among those patients on preoperative AT (mean 76.4 stents, and the latter for atrial fibrillation and either pri- years) than patients not receiving preoperative AT (mean mary prevention or known CAD. All patients on an AT 68.4 years), with significant male predominance. There regimen at presentation had their medication discontinued were similar numbers of left- and right-sided cSDH. One- at or before surgery. The median time from discontinua- quarter of patients presented with bilateral hematomas; tion to initial operation was 1 day, with the exception of all but 5 of these patients required bilateral evacuation in clopidogrel which was stopped a median of 3 days preop- their initial operation. The majority of patients underwent eratively. In total, 59 of 86 patients on warfarin (69%) un-

752 J Neurosurg Volume 124 • March 2016

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC Antithrombosis resumption in chronic subdural hemorrhage

Fig. 1. Antithrombotic agent use among the study patients. The use of antiplatelet and anticoagulant medications pre- and postop- eratively are shown. 81 and 325 represent the dosage of ASA in milligrams. derwent some form of reversal, which generally included mary prevention due to atherosclerotic risk factors, 72 for a combination of intravenous vitamin K and either fresh- atrial fibrillation, 34 for prior TIA or stroke, 45 for CAD frozen plasma or a prothrombin complex concentrate. with or without revascularization, 8 for mechanical car- The indications for preoperative AT are shown in Fig. diac valves, 13 for indwelling arterial stents (coronary, ca- 2. Eighty-four patients were on an AT regiment for pri- rotid, or peripheral), 14 for the treatment of active periph- eral arterial/venous thrombosis or pulmonary embolism, 2 for deep venous thrombosis (DVT) prophylaxis, and 6 for other indications.

Earlier and More Frequent Major Hemorrhages With Preoperative AT A summary of all postoperative complications encoun- tered in our patient cohort, as stratified by pre- and post- operative AT use, is shown in Table 2. Among all study patients, postoperative major hemor- rhage occurred in 71 of 479 patients (14.8%); of these, 64 patients required reoperation (16 of whom required a third procedure and 3 of whom required a fourth) and 7 patients were readmitted to the hospital for observation without further intervention. Patients on any AT preoperatively were more likely to have a major hemorrhage than those off AT (19.0% vs 10.9%; OR 1.93; 95% CI 1.15–2.71; p = 0.014). Subgroup analysis revealed that patients on preop- erative ASA (OR 1.75; 95% CI 0.97–2.53; p = 0.083) or warfarin (OR 2.17; 95% CI 1.12–3.20; p = 0.023) were also more likely to rebleed than those patients off AT (Fig. 3). Patients on clopidogrel preoperatively did not rebleed at a Fig. 2. Indications for AT. The bars represent the numbers of patients significantly higher rate than patients off AT, but there was with each indication. Some patients had multiple indications. no statistical difference in the incidence of major hemor-

J Neurosurg Volume 124 • March 2016 753

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC D. Guha, S. Coyne, and R. L. Macdonald

TABLE 2. Summary of the postoperative complications* Major Hemorrhage Minor Hemorrhage Thromboembolism AT Prerestart Postrestart Prerestart Postrestart Prerestart Postrestart No AT preop (n = 248) 27 59 2 AT preop (n = 231) ASA preop (n = 136) Restarted (n = 78) 7 1 11 9 2 0 Downgraded to ASA (n = 14)† 3 2 0 1 0 0 Not restarted (n = 58) 16 14 2 Clopidogrel preop (n = 28) Restarted (n = 7) 0 0 0 0 0 0 Not restarted (n = 21) 3 3 1 Warfarin preop (n = 86) Restarted (n = 35) 2 0 5 2 1 0 Not restarted (n = 51) 16 9 0 * The number of postoperative complications stratified by pre- and postoperative AT status are shown. The complications are categorized by their occurrence at pre- or postresumption of AT, where applicable. † Patients in this category were on either clopidogrel or warfarin preoperatively and resumed ASA (81 mg) or ASA (325 mg) postoperatively. rhage between patients on ASA versus clopidogrel versus stents, and 2 radial artery thrombi in patients on AT for warfarin (chi-square = 1.52; p = 0.47). Among patients on primary prevention or prior stroke). One in-hospital stroke warfarin, preoperative reversal did not affect the rate of and 1 DVT were seen in patients not on preoperative AT. major hemorrhage. No cases of stent thrombosis were seen. There was no sta- Postoperative major hemorrhages occurred earlier in tistical difference in the frequency of thromboembolism patients receiving any preoperative AT at a median inter- between patients with or without a history of preoperative val of 16.2 days versus 26.5 days (Fig. 4). This just misses AT (2.60% vs 0.81%; OR 3.28; 95% CI 0.66–5.90; p = the level of statistical significance (p = 0.052). There was 0.16). There was also no statistical difference in throm- no difference in the timing of major rebleeds between pa- boembolism frequency among patients on ASA versus tients receiving ASA versus clopidogrel versus warfarin clopidogrel versus warfarin preoperatively (chi-square = (Fig. 4, lower), demonstrating median intervals of 14.2, 7.08; p = 0.069); however, patients on clopidogrel dem- 8.7, and 12.8 days, respectively. onstrated the highest postoperative thromboembolic rate (7.7%), which barely misses the level of significance (p = Minor Hemorrhage Incidence and Timing Is Unaffected by 0.052) (Fig. 3). Preoperative AT Among all study patients, postoperative minor hemor- rhage occurred in 110 of 479 patients (23.0%). There was no difference in the minor hemorrhage frequency be- tween patients with or without a history of preoperative AT (23.8% vs 22.1%), or between patients on ASA versus clopidogrel versus warfarin preoperatively (chi-square = 3.06; p = 0.22) (Fig. 3). There was no statistical difference in the timing of minor hemorrhage between patients with or without a history of any AT (median 53.2 vs 46.4 days), nor between patients on ASA versus clopidogrel versus warfarin preoperatively (median 46.7 vs 40.0 vs 50.2 days, respectively) (Fig. 5).

Earlier Postoperative Thromboembolism With Preoperative AT Thromboembolic complications occurred in 8 of 479 study patients (1.67%). Six of the 8 thromboembolic events occurred in patients on preoperative AT (Table 2): 2 myo- cardial infarctions in patients on ASA for known CAD, Fig. 3. Incidence of postoperative complications. Incidence of post- 1 DVT in a patient on warfarin for previous DVT, and 3 operative complications stratified by preoperative AT status as the events unrelated to the indication for initial AT (1 in-hos- percentage of all patients within a given AT category. *p < 0.05. TE = pital stroke in a patient on ASA+clopidogrel for coronary thromboembolism.

754 J Neurosurg Volume 124 • March 2016

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC Antithrombosis resumption in chronic subdural hemorrhage

Fig. 4. Timing of postoperative major hemorrhage. Upper: Survival Fig. 5. Timing of postoperative minor hemorrhage. Upper: Survival curve for patients with major hemorrhage stratified by preoperative AT curve for patients with minor hemorrhage stratified by preoperative AT status. Lower: Time of major hemorrhage, measured in days from the status. Lower: Time of minor hemorrhage measured in days from the initial operation and stratified by preoperative AT status. The results initial operation and stratified by preoperative AT status. The results shown are the median with the 95% confidence interval. *p < 0.05 vs no shown are the median with the 95% confidence interval. *p < 0.05 vs no antithrombosis. antithrombosis.

The restart frequencies and intervals of all AT agents Thromboembolic complications occurred significantly assessed in this study are shown in Table 3. AT was re- earlier in patients with a history of preoperative AT (me- sumed at a median of 52 days postoperatively, and at > dian 2.7 vs 51.5 days; p = 0.036) (Fig. 6). There was no sta- 2 weeks in 75% of cases. Of all agents, clopidogrel was tistical difference in thromboembolism timing for patients resumed least often and with the largest delay. No agent on ASA versus clopidogrel versus warfarin preoperatively was resumed earlier than 3 days postoperatively. There (median 3.5 vs 3.5 vs 2.0 days, respectively) (Fig. 6 lower). was no difference in the rate of resumption of any agent between patients undergoing initial craniotomy versus Postoperative AT Resumption Associated With Reduced bur hole evacuation (p = 0.52). Recurrence Fifty of 231 patients on preoperative AT suffered a ma- Of the 231 study patients on preoperative AT, 120 were jor hemorrhage, of whom 15 resumed AT treatment post- restarted on a postoperative AT agent (51.9%), 106 on their operatively and in only 3 did the resumption of AT precede home regimen. The remaining 14 patients resumed ASA rebleeding (Table 2). Patients with a history of preopera- or converted from either ASA+clopidogrel (typically for tive AT who were restarted on any AT had a reduced risk indwelling stents or CAD) or warfarin (typically for atrial of rebleeding following resumption in comparison with fibrillation with questionable indications for therapeutic those patients who did not resume AT (26.9% vs 2.2%; OR anticoagulation). No cases of inferior vena cava filter in- 0.06; 95% CI 0.02–0.2; p < 0.01) (Fig. 7 upper). sertion in lieu of AT resumption were found. Of our cohort’s 6 total thromboembolic complications

J Neurosurg Volume 124 • March 2016 755

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC D. Guha, S. Coyne, and R. L. Macdonald

Fig. 7. Complication incidence with the resumption of AT. Incidence of postoperative major hemorrhage (upper) and thromboembolism (lower) stratified by the resumption of any postoperative AT. Bars represent the number of patients in a given category. *p < 0.05. Fig. 6. Timing of postoperative thromboembolism. Upper: Survival curve for patients with thromboembolism stratified by preoperative AT status. Lower: Time of thromboembolism measured in days from the initial operation and stratified by preoperative AT status. The results of thromboembolism against time to resumption demon- shown are the median with the 95% confidence interval. *p < 0.05 vs no strated a hazard ratio of 1.00 (95% CI 0.98–1.02; p = 0.86), antithrombosis. indicating no influence of the resumption interval on the subsequent thromboembolism risk. in patients receiving preoperative AT, 4 occurred in pa- tients who resumed postoperative AT and all occurred Discussion prior to AT resumption. Patients who resumed AT did not Antiplatelet and anticoagulant therapy are known risk show any difference in thromboembolism frequency prior factors for the development of cSDH, which are expected to AT resumption relative to the patients who did not re- to rise in prevalence with the aging population that in- sume AT (3.3% vs 1.8%; OR 1.34; 95% CI 0.29–6.13; p = creasingly requires AT to treat the diseases that afflict it. 0.89) (Fig. 7 lower). Cox proportional hazards regression The reported AT use in prior studies ranges from 18% to

TABLE 3. Postoperative resumption of AT* Preop Agent % Restarted Early Restart Late Restart Median Restart Interval (range) in Days All ATs 51.9% (120/231) 24.4% 75.6% 52 (3–801) ASA 57.4% (78/136) 30.8% (24/78) 69.2% (54/78) 52 (3–187) Clopidogrel 25.0% (7/28) 0.0% (0/7) 100.0% (7/7) 67 (28–104) Warfarin 40.7% (35/86) 14.3% (5/35) 85.7% (30/35) 47 (4–801) LMWH 37.5% (3/8) 33.3% (1/3) 66.7% (2/3) 17.5 (7–57) Targeted anticoagulants 100.0% (1/1) 0.0% (0/1) 100.0% (1/1) 53 (53–53) Dipyridamole 66.7% (4/6) 0.0% (0/4) 100.0% (4/4) 54 (32–66) * The number and percentage of patients on a given preoperative agent restarted on the same agent postoperatively. Early/late restart refers to the percentage of restarts at ≤ 2 weeks (early) or > 2 weeks (late) from the time of the initial operation.

756 J Neurosurg Volume 124 • March 2016

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC Antithrombosis resumption in chronic subdural hemorrhage

43% of patients with cSDH.3,7,11,27,29 Our cohort had slightly Multiple series have echoed the safety of resuming war- elevated numbers, with 48% of patients on any AT, 33% on farin at 3 to 5 days for patients with mechanical valves with antiplatelets, and 18% on warfarin. no increased risk of recurrence.6,15,18,31 Controversy remains The use of AT has not been as clearly associated with regarding warfarin. Majeed et al. reported 10 to 30 weeks postoperative recurrence in the literature, although retro- as the optimal restart interval, although this encompassed spective analyses reveal a recurrence rate as high as 33% any instance of warfarin-related intracranial hemorrhage for patients on AT at presentation versus as high as 15% for and included both operative and nonoperative cases.20 In a patients off AT.10 Forster et al. and Rust et al. showed in- series of 343 cSDH, Torihashi et al. reported no increased creased recurrence with antiplatelets,11,25 Chon et al. found risk of recurrence following the resumption of antiplate- increased recurrence with anticoagulants,7 and several oth- lets within 1 week postoperatively, though all patients were ers found no increase in recurrence with either anticoagu- resumed and hence no comparison was made to patients lants or antiplatelets.15,23,27,29 The overall recurrence rate of who did not resume preoperative antiplatelets.29 Okano et 14.8% in our cohort is consistent with prior studies. Our al., in a large cohort of 448 cSDH patients, also reported findings show increased major rehemorrhage in patients no increase in recurrence for patients on preoperative anti- with a history of any preoperative AT (19% vs 10.9%), and platelet agents who resumed postoperatively; however, the specifically ASA and warfarin (17.7% and 20.9%, respec- timing of resumption was not assessed.23 Neither study dis- tively). Interestingly, patients on clopidogrel did not show a tinguished between types of antiplatelet agents. In smaller proclivity to rebleed; we found no prior studies in the liter- studies, Tahsim-Oglou et al. reported increased recurrence ature on the recurrence rates in the context of clopidogrel. and reoperation rates for patients started on prophylactic In our cohort, clopidogrel was discontinued a median of 3 dose LMWH (enoxaparin 40 mg daily) on postoperative days prior to initial operation versus 1 day for both ASA Day 1, and Forster et al. reported increased recurrence in and warfarin, which may partially explain this finding. high-risk patients bridged postoperatively with therapeu- The incidence of thromboembolic complications in tically dosed nadroparin (also a LMWH).11,28 No studies our cohort was 1.67%, which is consistent with previous assessing the recurrence risk with newer targeted antico- reports. Few studies on cSDH have adequately evaluated agulants were found. thromboembolic complications in order to assess the rela- Our finding of decreased recurrence risk following tive rates in patients with and without a history of preoper- AT resumption is curious but was echoed previously by ative AT, 6 though the discontinuation/reversal of warfarin several other studies summarized in a systematic review in 1 study did not increase the thromboembolism risk.15 by Chari et al.6 This finding may be due to differences in Data on patients with any intracranial bleeding, without baseline hemorrhage risk across patients, with a tendency distinction of intra- and extraaxial hemorrhages or AT sta- to restart AT on those patients perceived to be at lower tus, has shown equivalent hemorrhage and thromboembo- risk for recurrence. Rehemorrhage risk stratification re- lism rates of roughly 6% to 7%.16 In our study there was quires detailed data on patient comorbidities, neurological a trend toward increased thromboembolism in patients on and functional status, including the risk of falls along with any preoperative AT (2.60% vs 0.81%), though this was the preoperative radiological predictors of hematoma re- not statistically significant. Patients on clopidogrel had the currence such as hematoma width and morphology. None most thromboembolic complications; this may have been of these data were available to us due to the retrospective due partly to the cessation of clopidogrel earlier than ASA nature of this study. or warfarin. The indications for clopidogrel typically in- Other similar limitations exist in this study. The met- cluded arterial stents, which are thought to have a higher rics of thromboembolic risk in atrial fibrillation, such as 13,19 risk of thrombosis; however, none of our complications in- CHADS2 and CHA2DS2-VASc, which are particularly cluded stent thrombosis. important for warfarin and newer anticoagulants, were un- For cSDH in the context of AT, it is standard practice to available. Preoperative and postoperative INR for patients discontinue AT at presentation and reverse warfarin prior on warfarin were not always available. Our study was to surgery where possible.15,18 The question then becomes if conducted at a tertiary referral center, and many patients and when to resume AT postoperatively. While patients on were already reversed at the referring institution with an a preoperative AT regimen in our cohort rebled earlier and undocumented INR on admission. Postoperatively, many more frequently than those patients off an AT regimen, at patients were restarted on warfarin in the hospital, but a median of 16.2 days versus 26.5 days, respectively, this were subsequently discharged with their INR levels to be by and large occurred irrespective of AT resumption, and, followed by their primary care physician. While the re- in fact, postoperative resumption was associated with a de- versal of warfarin has been shown to reduce recurrence crease in rehemorrhage. Patients on preoperative AT also and improve outcome,10,15 the reversal of antiplatelet agents experienced thromboembolic complications significantly with desmopressin or platelet transfusions is controversial earlier than those patients not receiving preoperative AT, and was not captured here.5 Our sample sizes for newer an- at an average of 2.7 days versus 51 days, respectively, albeit ticoagulants, such as dabigatran or fondaparinux, were too the complications were unrelated to the original indication limited to allow subgroup analyses of any antithrombotics for AT in half of the cases. However, given the earlier onset other than ASA, warfarin, or clopidogrel. Our cohort size of thromboembolism in patients on AT, with the apparently of 479 patients, while one of the largest series in the litera- minimal risk of rebleeding following resumption, our data ture, remains too limited to allow a proper comparison of provide cursory evidence that resuming therapeutic AT thromboembolic risk between patients with and without a early, starting at 3 days postoperatively, may be safe. history of AT. Based on our cohort, as well as the findings

J Neurosurg Volume 124 • March 2016 757

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC D. Guha, S. Coyne, and R. L. Macdonald in the literature and assuming a 2% thromboembolic risk DeRosa PL, Anderson K, et al: The surgical management of in patients with a history of AT and 0.5% risk in those chronic subdural hematoma. Neurosurg Rev 35:155–169, without AT, a study would require at least 860 patients to 2012 11. Forster MT, Mathé AK, Senft C, Scharrer I, Seifert V, Ger- achieve 80% statistical power with an alpha level of 0.05. lach R: The influence of preoperative anticoagulation on out- Lastly, no agent in our study was restarted prior to 3 days come and quality of life after surgical treatment of chronic postoperatively, precluding the comparison of outcomes subdural hematoma. J Clin Neurosci 17:975–979, 2010 between resumption earlier or later than 3 days. 12. Fuster V, Mearns BM: The CVD paradox: mortality vs prev- alence. Nat Rev Cardiol 6:669, 2009 13. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Conclusions Radford MJ: Validation of clinical classification schemes for Patients receiving preoperative AT suffered hemor- predicting stroke: results from the National Registry of Atrial rhagic complications earlier and more frequently, but also Fibrillation. JAMA 285:2864–2870, 2001 experienced thromboembolic complications significantly 14. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby earlier. Patients who were resumed on AT did not have an JV, et al: Prevalence of diagnosed atrial fibrillation in adults: increased risk of recurrent hemorrhage. We provide curso- national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial ry evidence that the resumption of preoperative antithrom- Fibrillation (ATRIA) Study. JAMA 285:2370–2375, 2001 botics at 3 days postoperatively may be safe; however, 15. Gonugunta V, Buxton N: Warfarin and chronic subdural hae- larger prospective studies are needed before any definitive matomas. Br J Neurosurg 15:514–517, 2001 recommendations can be made. The ultimate decision of 16. Hawryluk GW, Austin JW, Furlan JC, Lee JB, O’Kelly C, whether, when, and how to resume AT should be tailored Fehlings MG: Management of anticoagulation following cen- to individual patients and their unique risk profiles. tral nervous system hemorrhage in patients with high throm- boembolic risk. J Thromb Haemost 8:1500–1508, 2010 17. He W, Sengupta M, Velkoff VA, DeBarros KA: 65+ in the Acknowledgments United States: 2005. Washington DC: US Census Bureau, Dr. Macdonald receives grant support from the Physicians 2005 Services Incorporated Foundation, Brain Aneurysm Foundation, 18. Kawamata T, Takeshita M, Kubo O, Izawa M, Kagawa M, Canadian Institutes for Health Research, and the Heart and Stroke Takakura K: Management of intracranial hemorrhage associ- Foundation of Canada. ated with anticoagulant therapy. Surg Neurol 44:438–443, 1995 19. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ: Refin- References ing clinical risk stratification for predicting stroke and throm- 1. Asghar M, Adhiyaman V, Greenway MW, Bhowmick BK, boembolism in atrial fibrillation using a novel risk factor- Bates A: Chronic subdural haematoma in the elderly—a based approach: the euro heart survey on atrial fibrillation. North Wales experience. J R Soc Med 95:290–292, 2002 Chest 137:263–272, 2010 2. Aspegren OP, Åstrand R, Lundgren MI, Romner B: Antico- 20. Majeed A, Kim YK, Roberts RS, Holmström M, Schulman agulation therapy a risk factor for the development of chronic S: Optimal timing of resumption of warfarin after intracra- subdural hematoma. Clin Neurol Neurosurg 115:981–984, nial hemorrhage. Stroke 41:2860–2866, 2010 2013 21. Mori K, Maeda M: Surgical treatment of chronic subdural 3. Baechli H, Nordmann A, Bucher HC, Gratzl O: Demograph- hematoma in 500 consecutive cases: clinical characteristics, ics and prevalent risk factors of chronic subdural haematoma: surgical outcome, complications, and recurrence rate. Neurol results of a large single-center cohort study. Neurosurg Rev Med Chir (Tokyo) 41:371–381, 2001 27:263–266, 2004 22. Nakaguchi H, Tanishima T, Yoshimasu N: Factors in the 4. Bourgeois P, Sleiman M, Louis E, Haddad E, Touzet G, Fich- natural history of chronic subdural hematomas that influence ten A, et al: [Chronic subdural hematoma in patients over 80 their postoperative recurrence. J Neurosurg 95:256–262, years of age.] Neurochirurgie 45:124–128, 1999 (Fr) 2001 5. Campbell PG, Sen A, Yadla S, Jabbour P, Jallo J: Emergency 23. Okano A, Oya S, Fujisawa N, Tsuchiya T, Indo M, Nakamura reversal of antiplatelet agents in patients presenting with an T, et al: Analysis of risk factors for chronic subdural haema- intracranial hemorrhage: a clinical review. World Neurosurg toma recurrence after burr hole surgery: optimal manage- 74:279–285, 2010 ment of patients on antiplatelet therapy. Br J Neurosurg 6. Chari A, Clemente Morgado T, Rigamonti D: Recommence- 28:204–208, 2014 ment of anticoagulation in chronic subdural haematoma: a 24. Rogacka R, Chieffo A, Michev I, Airoldi F, Latib A, Cos- systematic review and meta-analysis. Br J Neurosurg 28:2– grave J, et al: Dual antiplatelet therapy after percutaneous 7, 2014 coronary intervention with stent implantation in patients tak- 7. Chon KH, Lee JM, Koh EJ, Choi HY: Independent predictors ing chronic oral anticoagulation. JACC Cardiovasc Interv for recurrence of chronic subdural hematoma. Acta Neuro- 1:56–61, 2008 chir (Wien) 154:1541–1548, 2012 25. Rust T, Kiemer N, Erasmus A: Chronic subdural haematomas 8. Connolly S, Pogue J, Hart R, Pfeffer M, Hohnloser S, and anticoagulation or anti-thrombotic therapy. J Clin Neu- Chrolavicius S, et al: Clopidogrel plus aspirin versus oral rosci 13:823–827, 2006 anticoagulation for atrial fibrillation in the Atrial fibrillation 26. Santarius T, Kirkpatrick PJ, Ganesan D, Chia HL, Jalloh Clopidogrel Trial with Irbesartan for prevention of Vascular I, Smielewski P, et al: Use of drains versus no drains after Events (ACTIVE W): a randomised controlled trial. Lancet burr-hole evacuation of chronic subdural haematoma: a ran- 367:1903–1912, 2006 domised controlled trial. Lancet 374:1067–1073, 2009 9. Cousseau DH, Echevarría Martín G, Gaspari M, Gonorazky 27. Stanisic M, Lund-Johansen M, Mahesparan R: Treatment SE: [Chronic and subacute subdural haematoma. An epi- of chronic subdural hematoma by burr-hole craniostomy in demiological study in a captive population.] Rev Neurol adults: influence of some factors on postoperative recurrence. 32:821–824, 2001 (Span) Acta Neurochir (Wien) 147:1249–1257, 2005 10. Ducruet AF, Grobelny BT, Zacharia BE, Hickman ZL, 28. Tahsim-Oglou Y, Beseoglu K, Hänggi D, Stummer W, Stei-

758 J Neurosurg Volume 124 • March 2016

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC Antithrombosis resumption in chronic subdural hemorrhage

ger HJ: Factors predicting recurrence of chronic subdural haematoma: the influence of intraoperative irrigation and Disclosure low-molecular-weight heparin thromboprophylaxis. Acta Dr. Macdonald states that he has direct stock ownership in and is Neurochir (Wien) 154:1063–1068, 2012 Chief Scientific Officer of Edge Therapeutics, Inc. 29. Torihashi K, Sadamasa N, Yoshida K, Narumi O, Chin M, Yamagata S: Independent predictors for recurrence of chron- Author Contributions ic subdural hematoma: a review of 343 consecutive surgical cases. Neurosurgery 63:1125–1129, 2008 Conception and design: Macdonald, Guha. Acquisition of data: 30. World Health Organization: International Statistical Clas- Guha, Coyne. Analysis and interpretation of data: Guha. Draft- sification of Diseases and Related Health Problems, 10th ing the article: Guha. Critically revising the article: all authors. Revision. Geneva: World Health Organization, 2011 Reviewed submitted version of manuscript: all authors. Statistical 31. Yeon JY, Kong DS, Hong SC: Safety of early warfarin re- analysis: Guha. Administrative/technical/material support: Mac- sumption following burr hole drainage for warfarin-associat- donald. Study supervision: Macdonald. ed subacute or chronic subdural hemorrhage. J Neurotrau- ma 29:1334–1341, 2012 Correspondence 32. Yorkgitis BK, Ruggia-Check C, Dujon JE: Antiplatelet and R. Loch Macdonald, Division of Neurosurgery, St. Michael’s anticoagulation medications and the surgical patient. Am J Hospital, 30 Bond St., Toronto, ON M5B 1W8, Canada. email: Surg 207:95–101, 2014 [email protected].

J Neurosurg Volume 124 • March 2016 759

Unauthenticated | Downloaded 10/06/21 04:03 PM UTC