[Skip to Navigation]
Sign In
Figure 1.  Study Population
Study Population

LMWH indicates low-molecular-weight heparin; TBI, traumatic brain injury; UH, unfractionated heparin; VTE, venous thromboembolism.

Figure 2.  Unadjusted Probabilities of Venous Thromboembolism (VTE) and Repeated Neurosurgery as a Function of Prophylaxis Delay
Unadjusted Probabilities of Venous Thromboembolism (VTE) and Repeated Neurosurgery as a Function of Prophylaxis Delay

Shading represents the 95% CIs.

Figure 3.  Comparison of Unadjusted Probabilities for Venous Thromboembolism (VTE), Repeated Neurosurgery, and Death as a Function of Prophylaxis Delay in the Craniotomy/Craniectomy and Intracranial Monitor/Drain Insertion Subgroups
Comparison of Unadjusted Probabilities for Venous Thromboembolism (VTE), Repeated Neurosurgery, and Death as a Function of Prophylaxis Delay in the Craniotomy/Craniectomy and Intracranial Monitor/Drain Insertion Subgroups

Shading represents the 95% CIs.

aSignificant in the risk-adjusted analysis.

Figure 4.  Hospital-Level Variation in Use of Early Prophylaxis After Urgent Neurosurgical Intervention
Hospital-Level Variation in Use of Early Prophylaxis After Urgent Neurosurgical Intervention

Early prophylaxis was defined as initiation of prophylaxis within 3 days after neurosurgical intervention.

Table.  Hospital-Level Analysis of Outcomes as a Function of Early Prophylaxis Usea
Hospital-Level Analysis of Outcomes as a Function of Early Prophylaxis Usea
1.
Rutland-Brown  W, Langlois  JA, Thomas  KE, Xi  YL.  Incidence of traumatic brain injury in the United States, 2003.   J Head Trauma Rehabil. 2006;21(6):544-548. doi:10.1097/00001199-200611000-00009PubMedGoogle ScholarCrossref
2.
Reiff  DA, Haricharan  RN, Bullington  NM, Griffin  RL, McGwin  G  Jr, Rue  LW  III.  Traumatic brain injury is associated with the development of deep vein thrombosis independent of pharmacological prophylaxis.   J Trauma. 2009;66(5):1436-1440. doi:10.1097/TA.0b013e31817fdf1cPubMedGoogle Scholar
3.
Knudson  MM, Ikossi  DG, Khaw  L, Morabito  D, Speetzen  LS.  Thromboembolism after trauma: an analysis of 1602 episodes from the American College of Surgeons National Trauma Data Bank.   Ann Surg. 2004;240(3):490-496. doi:10.1097/01.sla.0000137138.40116.6cPubMedGoogle ScholarCrossref
4.
Phelan  HA, Eastman  AL, Madden  CJ,  et al.  TBI risk stratification at presentation: a prospective study of the incidence and timing of radiographic worsening in the Parkland Protocol.   J Trauma Acute Care Surg. 2012;73(2)(suppl 1):S122-S127. doi:10.1097/TA.0b013e3182606327PubMedGoogle ScholarCrossref
5.
Spano  PJ  II, Shaikh  S, Boneva  D, Hai  S, McKenney  M, Elkbuli  A.  Anticoagulant chemoprophylaxis in patients with traumatic brain injuries: a systematic review.   J Trauma Acute Care Surg. 2020;88(3):454-460. doi:10.1097/TA.0000000000002580PubMedGoogle ScholarCrossref
6.
Lu  VM, Alvi  MA, Rovin  RA, Kasper  EM.  Clinical outcomes following early versus late pharmacologic thromboprophylaxis in patients with traumatic intracranial hemorrhage: a systematic review and meta-analysis.   Neurosurg Rev. 2020;43(3):861-872. doi:10.1007/s10143-018-1045-yPubMedGoogle ScholarCrossref
7.
Byrne  JP, Mason  SA, Gomez  D,  et al.  Timing of pharmacologic venous thromboembolism prophylaxis in severe traumatic brain injury: a propensity-matched cohort study.   J Am Coll Surg. 2016;223(4):621-631.e5. doi:10.1016/j.jamcollsurg.2016.06.382PubMedGoogle ScholarCrossref
8.
Scudday  T, Brasel  K, Webb  T,  et al.  Safety and efficacy of prophylactic anticoagulation in patients with traumatic brain injury.   J Am Coll Surg. 2011;213(1):148-153. doi:10.1016/j.jamcollsurg.2011.02.027PubMedGoogle ScholarCrossref
9.
Vella  MA, Crandall  ML, Patel  MB.  Acute Management of Traumatic Brain Injury.   Surg Clin North Am. 2017;97(5):1015-1030. doi:10.1016/j.suc.2017.06.003PubMedGoogle ScholarCrossref
10.
Phelan  HA.  Pharmacologic venous thromboembolism prophylaxis after traumatic brain injury: a critical literature review.   J Neurotrauma. 2012;29(10):1821-1828. doi:10.1089/neu.2012.2459PubMedGoogle ScholarCrossref
11.
American College of Surgeons. ACS TQP Best Practices Guidelines. Accessed February 1, 2020. https://www.facs.org/quality-programs/trauma/tqip/best-practice
12.
Pastorek  RA, Cripps  MW, Bernstein  IH,  et al.  The Parkland Protocol’s modified Berne-Norwood criteria predict two tiers of risk for traumatic brain injury progression.   J Neurotrauma. 2014;31(20):1737-1743. doi:10.1089/neu.2014.3366PubMedGoogle ScholarCrossref
13.
Shafi  S, Nathens  AB, Cryer  HG,  et al.  The Trauma Quality Improvement Program of the American College of Surgeons Committee on Trauma.   J Am Coll Surg. 2009;209(4):521-530.e1. doi:10.1016/j.jamcollsurg.2009.07.001PubMedGoogle ScholarCrossref
14.
Shafi  S, Nathens  AB, Parks  J, Cryer  HM, Fildes  JJ, Gentilello  LM.  Trauma quality improvement using risk-adjusted outcomes.   J Trauma. 2008;64(3):599-604.PubMedGoogle Scholar
15.
American College of Surgeons. Trauma Quality Improvement Program. Accessed February 1, 2020. https://www.facs.org/quality-programs/trauma/tqp/center-programs/tqip
16.
Yuan  Y.  Multiple imputation using SAS software.   Journal of Statistical Software. 2011;45(6). doi:10.18637/jss.v045.i06Google Scholar
17.
Sterne  JA, White  IR, Carlin  JB,  et al.  Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls.   BMJ. 2009;338:b2393. doi:10.1136/bmj.b2393PubMedGoogle ScholarCrossref
18.
Cohen  RA. An introduction to PROC LOESS for local regression: SAS Institute Inc. Accessed February 1, 2020. https://stats.idre.ucla.edu/wp-content/uploads/2016/02/loesssugi.pdf
19.
Merlo  J, Yang  M, Chaix  B, Lynch  J, Råstam  L.  A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people.   J Epidemiol Community Health. 2005;59(9):729-736. doi:10.1136/jech.2004.023929PubMedGoogle ScholarCrossref
20.
Mickey  RM, Greenland  S.  The impact of confounder selection criteria on effect estimation.   Am J Epidemiol. 1989;129(1):125-137. doi:10.1093/oxfordjournals.aje.a115101PubMedGoogle ScholarCrossref
21.
Merlo  J, Chaix  B, Ohlsson  H,  et al.  A brief conceptual tutorial of multilevel analysis in social epidemiology: using measures of clustering in multilevel logistic regression to investigate contextual phenomena.   J Epidemiol Community Health. 2006;60(4):290-297. doi:10.1136/jech.2004.029454PubMedGoogle ScholarCrossref
22.
Knudson  MM, Gomez  D, Haas  B, Cohen  MJ, Nathens  AB.  Three thousand seven hundred thirty-eight posttraumatic pulmonary emboli: a new look at an old disease.   Ann Surg. 2011;254(4):625-632. doi:10.1097/SLA.0b013e3182300209PubMedGoogle ScholarCrossref
23.
Denson  K, Morgan  D, Cunningham  R,  et al.  Incidence of venous thromboembolism in patients with traumatic brain injury.   Am J Surg. 2007;193(3):380-383. doi:10.1016/j.amjsurg.2006.12.004PubMedGoogle ScholarCrossref
24.
Ho  KM, Burrell  M, Rao  S, Baker  R.  Incidence and risk factors for fatal pulmonary embolism after major trauma: a nested cohort study.   Br J Anaesth. 2010;105(5):596-602. doi:10.1093/bja/aeq254PubMedGoogle ScholarCrossref
25.
Phelan  HA, Wolf  SE, Norwood  SH,  et al.  A randomized, double-blinded, placebo-controlled pilot trial of anticoagulation in low-risk traumatic brain injury: the Delayed Versus Early Enoxaparin Prophylaxis I (DEEP I) study.   J Trauma Acute Care Surg. 2012;73(6):1434-1441. doi:10.1097/TA.0b013e31825ac49ePubMedGoogle ScholarCrossref
Original Investigation
December 15, 2021

Association of Venous Thromboembolism Prophylaxis After Neurosurgical Intervention for Traumatic Brain Injury With Thromboembolic Complications, Repeated Neurosurgery, and Mortality

Author Affiliations
  • 1Department of Surgery, Johns Hopkins Hospital, Baltimore, Maryland
  • 2Division of Neurosurgery and Spinal Program, Department of Surgery, University of Toronto, Toronto, Ontario, Canada
  • 3Department of Neurosurgery, Penn Presbyterian Medical Center, Philadelphia, Pennsylvania
  • 4Division of Traumatology, Surgical Critical Care and Emergency Surgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia
  • 5Department of Surgery, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada
JAMA Surg. 2022;157(3):e215794. doi:10.1001/jamasurg.2021.5794
Key Points

Question  Is early initiation of pharmacologic venous thromboembolism (VTE) prophylaxis after urgent neurosurgical interventions for traumatic brain injury (TBI) associated with reduced risk of thromboembolic complications without increasing risk of intracranial complications?

Findings  In this cohort study of 4951 patients who underwent decompressive craniotomy/craniectomy or intracranial monitor/drain insertion for TBI, delayed prophylaxis after intervention was associated with increased odds of VTE, but earlier prophylaxis was also associated with increased risk of a repeated neurosurgical intervention.

Meaning  The findings suggest that early pharmacologic VTE prophylaxis after neurosurgical intervention for TBI is associated with reduced risk of thromboembolic complications, but also with increased risk of repeated neurosurgical intervention.

Abstract

Importance  There is a lack of evidence regarding the effectiveness and safety of pharmacologic venous thromboembolism (VTE) prophylaxis among patients who undergo neurosurgical interventions for traumatic brain injury (TBI).

Objective  To measure the association between timing of VTE prophylaxis after urgent neurosurgical intervention for TBI and thromboembolic and intracranial complications.

Design, Setting, and Participants  This retrospective cohort study included adult patients (aged ≥16 years) who underwent urgent neurosurgical interventions (craniotomy/craniectomy or intracranial monitor/drain insertion within 24 hours after admission) for TBI at level 1 and 2 trauma centers participating in the American College of Surgeons Trauma Quality Improvement Program between January 1, 2012, and December 31, 2016. Data were analyzed from January to August 2020.

Exposures  Timing of pharmacologic VTE prophylaxis initiation after urgent neurosurgical intervention (prophylaxis delay) measured in days (24-hour periods).

Main Outcomes and Measures  The primary outcome was VTE (deep vein thrombosis or pulmonary embolism). Secondary outcomes were repeated neurosurgery (neurosurgical reintervention after initiation of VTE prophylaxis) and mortality. Hierarchical logistic regression models were used to evaluate the association between prophylaxis delay and each outcome at the patient level and were adjusted for patient baseline and injury characteristics.

Results  The study included 4951 patients (3676 [74%] male; median age, 50 years [IQR, 31-64 years]) who underwent urgent neurosurgical intervention for TBI at 304 trauma centers. The median prophylaxis delay was 3 days (IQR, 1-5 days). After adjustment for patient baseline and injury characteristics, prophylaxis delay was associated with increased odds of VTE (adjusted odds ratio [aOR], 1.08 per day; 95% CI, 1.04-1.12). Earlier initiation of prophylaxis was associated with increased risk of repeated neurosurgery. During the first 3 days, each additional day of prophylaxis delay was associated with a 28% decrease in odds of repeated neurosurgery (aOR, 0.72 per day; 95% CI, 0.59-0.88). After 3 days, each additional day of prophylaxis delay was associated with an additional 15% decrease in odds of repeated neurosurgery (aOR, 0.85 per day; 95% CI, 0.80-0.90). Earlier prophylaxis was associated with greater mortality among patients who initially underwent intracranial monitor/drain procedures, such that each additional day of prophylaxis delay was associated with decreased odds of death (aOR, 0.94 per day; 95% CI, 0.89-0.99).

Conclusions and Relevance  In this cohort study of patients who underwent urgent neurosurgical interventions for TBI, early pharmacologic VTE prophylaxis was associated with reduced risk of thromboembolism. However, earlier initiation of prophylaxis was associated with increased risk of repeated neurosurgery. These findings suggest that although timely initiation of prophylaxis should be prioritized, caution should be used particularly during the first 3 days after the index procedure, when this risk appears to be highest.

Introduction

Patients with traumatic brain injury (TBI) are at elevated risk for developing venous thromboembolism (VTE).1-3 However, patients with TBI are also at risk for progression of intracranial hemorrhage (ICH).4 There is evidence that early initiation of pharmacologic VTE prophylaxis within 24 to 72 hours after admission is effective at reducing VTE and can be safe without increased risk of ICH expansion.5-8 In practice, prophylaxis is often withheld during the initial phase of care until ICH is demonstrated to be stable on repeated computed tomography (CT).9 Nonetheless, practice variation persists with respect to the timing of initiation and pharmacologic agent used (unfractionated or low-molecular-weight heparin).

Patients who undergo urgent neurosurgical interventions such as decompressive craniotomy/craniectomy or intracranial monitor/drain insertion are at highest risk for ICH progression.10-12 These patients are routinely excluded from observational studies. Therefore, there is a lack of evidence to guide the optimal timing of pharmacologic VTE prophylaxis in this population. In the presence of such uncertainty, the American College of Surgeons Trauma Quality Improvement Program (TQIP) management guidelines for TBI, citing the modified Berne-Norwood Criteria, recommend forgoing pharmacologic prophylaxis after neurosurgical intervention.10,11

The objective of this study was to evaluate the association of timing of pharmacologic VTE prophylaxis after urgent neurosurgical intervention for TBI with risk of thromboembolism and intracranial complications. We hypothesized that early initiation of pharmacologic VTE prophylaxis would be associated with reduced risk of VTE without increasing the risk of ICH-related complications.

Methods
Study Design and Data Source

This was a retrospective cohort study of patients treated at level 1 and 2 trauma centers participating in TQIP between January 1, 2012, and December 31, 2016. This study was approved by the University of Pennsylvania institutional review board. Informed consent was not required because data were retrospective and deidentified. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

TQIP is a performance improvement initiative of the American College of Surgeons that provides a platform for risk-adjusted benchmarking of trauma center outcomes.13,14 Patient and institutional variables are recorded by trained abstractors, and the reliability of data is ensured through abstractor training and interrater reliability audits. At the time of data collection for this study, there were more than 450 participating American College of Surgeons–verified and state-designated level 1 and 2 trauma centers in the US and Canada.15

Study Participants

Figure 1 shows derivation of the study cohort. Adult patients (aged ≥16 years) with blunt, isolated TBI (head abbreviated injury score [AIS], ≥3; AIS <3 in other body regions) who underwent urgent neurosurgical interventions were included. Urgent neurosurgical interventions were defined as craniotomy/craniectomy, intracranial monitor/drain insertion, or both within 24 hours of hospital arrival. Neurosurgical interventions were identified using International Classification of Diseases, Ninth Revision, Clinical Modification procedure codes (eTable 1 in the Supplement). The cohort was limited to patients who received pharmacologic VTE prophylaxis. TQIP began collecting data on pharmacologic VTE prophylaxis (type and timing of initiation) from participating trauma centers in 2012. To ensure reliability of the data analyzed, trauma centers not yet collecting these data in a given year were excluded for that year. Remaining trauma centers missing more than 25% of VTE prophylaxis data were also excluded. To mitigate the competing risks of early discharge or death, the cohort was further limited to patients who remained alive and in the hospital after 5 days. Patients with bleeding disorders or chronic anticoagulation were excluded.

Exposure

The exposure was defined as timing of VTE prophylaxis initiation after urgent neurosurgical intervention (prophylaxis delay). Prophylaxis delay was measured as the number of days (24-hour periods) elapsed between urgent neurosurgical intervention and first administration of pharmacologic VTE prophylaxis. For example, prophylaxis given within 24 hours of urgent neurosurgical intervention was a prophylaxis delay of 0 days and, between 24 and 48 hours, a prophylaxis delay of 1 day.

Outcomes

The primary outcome was VTE, defined as deep vein thrombosis or pulmonary embolism. Secondary outcomes sought to capture ICH-related complications. Because CT findings or acute neurologic changes are not captured in TQIP, repeated neurosurgery and in-hospital death were identified as outcomes that might result from clinically significant expansion of ICH. Repeated neurosurgery was defined as craniotomy/craniectomy or intracranial monitor/drain insertion after initiation of pharmacologic VTE prophylaxis.

Potential Confounders

We considered patient baseline and injury characteristics that might confound the association between prophylaxis delay and the outcomes. Patient baseline characteristics included age, sex, race (in TQIP, race is captured in keeping with the National Trauma Data Standard, which prioritizes self- or family-reported race identification), insurance status, comorbidities, and interfacility transfer. Injury characteristics included mechanism, injury severity (head AIS), patterns of intracranial injury, shock in the emergency department (systolic blood pressure ≤90 mm Hg), and initial Glasgow Coma Scale motor component. Patterns of intracranial injury were characterized by identifying specific lesions using AIS predot (1998 version) or International Classification of Diseases, Ninth Revision diagnosis codes (eTable 2 in the Supplement). The type of urgent neurosurgical intervention and the pharmacologic agent used for prophylaxis (unfractionated or low-molecular-weight heparin) were also considered. Presenting systolic blood pressure and Glasgow Coma Scale scores were missing for 16 309 patients with blunt, isolated TBI (5%); missing values were imputed using a multiple imputation technique.16,17

Statistical Analysis

Data were analyzed from January to August 2020. Estimated probabilities of VTE, repeated neurosurgery, and mortality were plotted as a function of prophylaxis delay using locally estimated scatterplot smoothing.18 Risk-adjusted patient- and hospital-level analyses were then performed.

First, hierarchical logistic regression models were used to evaluate the association between prophylaxis delay and each outcome at the patient level and were adjusted for patient baseline and injury characteristics. All models were multilevel mixed models that accounted for clustering of patients within trauma centers.19 Piecewise linear splines were used to improve model fit when breakpoints were identified on locally estimated scatterplot smoothing analysis. Potential confounders were evaluated for inclusion using the 10% change-in-estimate approach described by Mickey and Greenland.20 Final models included variables that changed the exposure point estimate by more than 10% in addition to variables deemed to be clinically important. Multicollinearity was ruled out using the variance inflation factor and tolerance statistic. Model calibration was ensured using the Hosmer-Lemeshow goodness-of-fit test, and model discrimination was reported using the C statistic.

Next, we performed a hospital-level analysis to evaluate the association between a hospital’s tendency to administer early prophylaxis and center-level outcomes. Early prophylaxis was defined by the median prophylaxis delay as prophylaxis initiation within 3 days after urgent neurosurgical intervention. This analysis was limited to hospitals that treated 10 or more patients who met the inclusion criteria. Trauma centers were grouped into quartiles based on the hospital-specific proportion of patients who received early prophylaxis. Hierarchical logistic regression was then used to measure the association between early prophylaxis quartile and the outcomes, adjusting for patient-case mix. This analysis assessed whether hospital practices for prophylaxis timing after urgent neurosurgical interventions are associated with center-level outcomes and minimized confounding at the individual patient level.

Finally, the proportion of inter-hospital variation in outcomes that could be explained by early prophylaxis use was calculated using the proportional change in variance (PCV). The PCV uses the ability of a multilevel model to estimate cluster-level (in this case, between-hospital) variance, and was calculated using the following equation19,21: PCV = [(V1 – V2) / V1] × 100%, where V1 is the inter-hospital variance in the multilevel model containing only patient characteristics and V2 is the inter-hospital variance in the same model including the hospital-specific early prophylaxis rate. This analysis quantified the degree to which practices for prophylaxis timing after urgent neurosurgical interventions contributed to observed differences in the outcomes between hospitals.

We tested for interaction between prophylaxis delay and type of urgent neurosurgical intervention in each multivariable model. This was important because the indications for each procedure likely reflect differing patterns of injury, risk for complications, and prognostication. When interaction terms were significant, we reported the risk-adjusted associations between prophylaxis delay and the outcomes of interest in each subgroup. A 2-sided P < .05 was considered statistically significant. Analyses were performed using SAS, version 9.4 (SAS Institute Inc).

Results
Study Population

The study included 4951 patients who underwent urgent neurosurgical interventions at 304 trauma centers. The median age was 50 years (IQR, 31-64 years), and 3676 patients (74%) were male. Falls were the most common mechanism (2596 [52%]). The median head AIS was 5 (IQR, 4-5), and ICH was present in 4516 patients (91%). The median Glasgow Coma Scale score at hospital arrival was 7 (IQR, 3-14). Craniotomy/craniectomy was the urgent procedure most frequently performed (2244 [45%]), followed by intracranial monitor/drain insertion (1688 [34%]) and both procedures (1019 [21%]). The median prophylaxis delay after urgent neurosurgical intervention was 3 days (IQR, 1-5 days). Venous thromboembolism occurred in 372 patients (8%; deep vein thrombosis in 336 [7%] and pulmonary embolism in 69 [1%]). Repeated neurosurgery was performed for 272 patients (6%), and 581 (12%) died.

Patient-Level Analysis

Plots of estimated probabilities for VTE and repeated neurosurgery as a function of prophylaxis delay are shown in Figure 2. The probability of VTE increased in association with longer prophylaxis delay (Figure 2A). Conversely, earlier prophylaxis was associated with greater probability of repeated neurosurgical intervention (Figure 2B).

Results of the hierarchical logistic regression models for VTE, repeated neurosurgery, and mortality are included in eTables 3 and 4 in the Supplement. After adjusting for patient baseline and injury characteristics, longer prophylaxis delay was associated with increased risk of VTE (eTable 3 in the Supplement). Specifically, each additional day of delay was associated with an 8% increase in odds of VTE (adjusted odds ratio [aOR], 1.08 per day; 95% CI, 1.04-1.12). However, earlier initiation of VTE prophylaxis was associated with increased risk of repeated neurosurgery (eTable 4 in the Supplement). During the first 3 days, each additional day of prophylaxis delay was associated with a 28% decrease in odds of repeated neurosurgery (aOR, 0.72 per day; 95% CI, 0.59-0.88). After 3 days, each additional day was associated with an additional 15% decreased odds of repeated neurosurgery (aOR, 0.85 per day; 95% CI, 0.80-0.90). Low-molecular-weight heparin (vs unfractionated) was associated with lower odds of VTE (aOR, 0.64; 95% CI, 0.49-0.84). Pharmacologic agent was not associated with repeated neurosurgery. Interaction terms between prophylaxis delay and the type of urgent neurosurgical intervention were not significant in the models for VTE or repeated neurosurgery.

The interaction term between prophylaxis delay and type of urgent neurosurgical intervention was significant in the model for mortality (P = .049). Therefore, we explored the association between prophylaxis delay and the outcomes of interest in the patient subgroups that initially underwent craniotomy/craniectomy and those that initially underwent intracranial monitor/drain insertion. Figure 3 compares plots for the probabilities of VTE, repeated neurosurgery, and mortality as a function of prophylaxis delay between subgroups. As expected from the tests for interaction, the risk-adjusted associations between prophylaxis delay, VTE (eTables 5 and 6 in the Supplement), and repeated neurosurgery (eTables 7 and 8 in the Supplement) were comparable between subgroups. However, among patients who underwent intracranial monitor/drain insertion as the index procedure, earlier prophylaxis was associated with higher mortality. Specifically, each additional day of prophylaxis delay was associated with a 6% decrease in odds of death (OR, 0.94 per day; 95% CI, 0.89-0.99) (eTable 10 in the Supplement). There was no association between prophylaxis delay and mortality among patients who underwent craniotomy/craniectomy (eTable 9 in the Supplement).

Hospital-Level Analysis

During the study period, 147 trauma centers treated 10 or more patients who underwent urgent neurosurgical interventions for TBI. There was notable variation in early prophylaxis use between hospitals (Figure 4). Specifically, hospitals in the highest quartile (quartile 1) began early prophylaxis in 82% (1217 of 1482 patients) with a median prophylaxis delay of 1 day (IQR, 1-2 days) compared with hospitals in the lowest quartile (quartile 4), where 10% received early prophylaxis (81 of 826 patients) and the median prophylaxis delay was 6 days (IQR, 4-8 days).

After adjusting for patient-case mix, there was no association between early prophylaxis use and VTE at the hospital level. However, hospitals in the highest quartile of early prophylaxis use were associated with higher odds of repeated neurosurgery compared with hospitals in the lowest quartile (120 of 1482 patients [8%] vs 29 of 826 patients [4%]; OR, 2.36; 95% CI, 1.46-3.82) (Table).

The interaction term between early prophylaxis quartile and type of urgent neurosurgical intervention was significant in the center-level model for mortality (P = .047). Therefore, the Table reports results for mortality by subgroup. Hospitals in the highest quartile of early prophylaxis use were associated with higher odds of mortality among patients who underwent intracranial monitor/drain insertion as the index procedure compared with hospitals in the lowest quartile (99 of 554 patients [18%] vs 26 of 312 patients [8%]; OR, 2.08; 95% CI, 1.17-3.69). There was no association between hospital-specific early prophylaxis use and mortality among patients who underwent craniotomy/craniectomy.

The Table shows the proportional change in variance for each outcome attributable to hospital-specific early prophylaxis use. With use of the model for repeated neurosurgery, the inter-hospital variance in the model containing only patient characteristics was 0.1811. After including the hospital-specific rate of early prophylaxis use, the observed variance was 0.0779, corresponding to a PCV of −57%, indicating that 57% of observed differences in repeated neurosurgery between hospitals could have been attributable to hospital-based practices for prophylaxis timing after urgent neurosurgical interventions. By comparison, the PCV for VTE attributable to hospital-specific early prophylaxis use was 1.5%, indicating that prophylaxis timing after urgent neurosurgical interventions had relatively little influence over inter-hospital variation in thromboembolism in this population.

Discussion

In this retrospective cohort study of patients who underwent neurosurgical interventions for TBI, delay in initiating pharmacologic VTE prophylaxis was associated with increased risk of thromboembolic complications. However, earlier initiation of prophylaxis was associated with increased risk of repeated neurosurgery. This adverse association was reflected at the hospital level. Specifically, trauma centers with the greatest tendency for initiating early prophylaxis use were associated with significantly greater risk of repeated neurosurgery.

These data fill a knowledge gap regarding the effectiveness and safety of pharmacologic VTE prophylaxis among patients with TBI. Patients with brain injury are at elevated risk of thromboembolic complications owing to prolonged immobilization and systemic hypercoagulability.2,22,23 Pulmonary embolism is associated with an estimated attributable mortality of up to 50%.24 These findings suggest that pharmacologic VTE prophylaxis should be initiated as early as safely feasible when risk of ICH-related complications is low. The modified Berne-Norwood Criteria,10,25 cited in the TQIP Best Practices Guidelines,11 suggest that prophylaxis can be safely initiated when ICH is stable on repeated head CT at 24 hours for “low risk” patients or at 72 hours for “moderate risk” patients. This study is the first, to our knowledge, to evaluate pharmacologic VTE prophylaxis among patients considered to be at “high risk”—those who have undergone urgent neurosurgical procedures during the acute phase of care. The finding that prophylaxis delay was associated with increased odds of VTE suggests that early prophylaxis is effective in reducing risk of thromboembolic complications in this population.

However, the finding that earlier prophylaxis was associated with increased risk of repeated neurosurgery raises concern that the benefit of reduced thromboembolic risk might also come at the cost of increased risk for hemorrhagic complication. This risk appeared to be greatest during the first 3 days after the index procedure. These data contrast with previous studies5-8 that found no association between early prophylaxis, initiated within 24 to 72 hours of admission, and intracranial complications. However, these studies did not specifically evaluate patients who underwent urgent neurosurgical procedures, a population in which ICH progression manifests on more than 60% of repeated imaging studies.4

We also found that earlier prophylaxis was associated with increased mortality among patients who underwent intracranial monitor/drain insertion as the index procedure. This association was not observed in patients who underwent craniotomy/craniectomy. These data may reflect a greater potential harm of hemorrhage progression in patients selected for intracranial monitor/drain procedures owing to injuries that are typically more diffuse, intraparenchymal, or not amenable to cranial decompression. The finding also highlights the need for prospective evaluation to elucidate whether this observation is causal rather than due to unmeasured confounding, which is a major limitation of observational studies such as this.

The results of our hospital-level analysis suggest that center-specific thromboprophylaxis practices are associated with patient outcomes. Patients treated at trauma centers most favoring early prophylaxis use had significantly increased odds of undergoing repeated neurosurgery. Patients who underwent intracranial monitor/drain insertion as the index procedure were significantly more likely to die at these hospitals. Our calculation of the PCV suggests that a large proportion (57%) of inter-hospital variation in repeated neurosurgical interventions may be attributable to hospital-specific differences in early prophylaxis use. These findings are notable in the context of the marked practice variation in timing of prophylaxis initiation among trauma centers. Taken together, these data suggest that although VTE prophylaxis should be prioritized, trauma centers should evaluate whether institutional protocols that emphasize early prophylaxis could contribute to negative outcomes in this patient population that is at highest risk for hemorrhagic complications.

Limitations

This study has several important limitations. The study used repeated neurosurgical procedures as a surrogate marker for ICH-related changes requiring surgical intervention. This approach was necessary because TQIP does not capture data related to findings on head CT or changes in neurologic examination. Therefore, it remains uncertain which repeated neurosurgical interventions were actually attributable to treatment of hemorrhage.

Confounding by indication is a concern in retrospective database studies such as this one. However, it is unlikely that early prophylaxis was systematically favored among patients at higher risk for repeated neurosurgery or death such that the results are attributable to such bias. To the contrary, it is common practice to defer prophylaxis among patients at high risk for ICH progression, and therefore bias should be toward the null.

Furthermore, repeated neurosurgery is unlikely to represent planned interventions such as closure or replacement of bone flap, which are identified by different International Classification of Diseases, Ninth Revision, Clinical Modification procedure codes. Therefore, although the true indications for these procedures are unknown, their occurrences likely represented unplanned adverse events.

The study is further limited by its observational nature such that we were unable to make concrete recommendations for safe timing of initiation of pharmacologic VTE prophylaxis after urgent neurosurgical interventions. Although difficult to demonstrate in a heterogeneous cohort using retrospective data, all patients are likely to achieve a health state in which it is clinically safe to begin pharmacologic VTE prophylaxis. Therefore, our observations should not be interpreted to indicate that prophylaxis is unsafe at all time points in this patient population. To the contrary, our findings support that timely initiation of prophylaxis should remain a priority to minimize risk of potentially fatal thromboembolism. However, our results also serve to caution that prophylaxis initiated too soon after neurosurgical intervention might contribute to adverse outcomes.

Conclusions

In this retrospective cohort study of patients who underwent urgent neurosurgical interventions for TBI, early pharmacologic VTE prophylaxis was associated with reduced risk of thromboembolic complications. However, earlier initiation of prophylaxis was also associated with increased risk of repeated neurosurgery. These findings suggest that although timely initiation of pharmacologic VTE prophylaxis should be prioritized, caution should be exercised particularly during the first 3 days after the index procedure, when this risk appears to be highest. Prospective evaluation led by multidisciplinary trauma, neurosurgical, and critical care teams is needed to define optimal timing and quantify risks of prophylaxis in this population.

Back to top
Article Information

Accepted for Publication: August 30, 2021.

Published Online: December 15, 2021. doi:10.1001/jamasurg.2021.5794

Corresponding Author: James P. Byrne, MD, PhD, Department of Surgery, Johns Hopkins Hospital, 1800 Orleans St, Sheikh Zayed Tower, Ste 6107B, Baltimore, MD 21287 (jpbyrne@gmail.com).

Author Contributions: Dr Byrne had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Byrne, Reilly, Seamon.

Acquisition, analysis, or interpretation of data: Byrne, Witiw, Schuster, Pascual, Cannon, Martin, Nathens.

Drafting of the manuscript: Byrne, Nathens.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Byrne.

Administrative, technical, or material support: Nathens.

Supervision: Schuster, Cannon, Martin, Reilly, Seamon.

Conflict of Interest Disclosures: Dr Pascual reported receiving grants from Grifols Inc outside the submitted work and having a patent pending for antithrombin III use for traumatic brain injury. Dr Cannon reported receiving personal fees from UpToDate and royalties paid to the University of Pennsylvania outside the submitted work. Dr Nathens reported being a salaried employee of the American College of Surgeons. No other disclosures were reported.

References
1.
Rutland-Brown  W, Langlois  JA, Thomas  KE, Xi  YL.  Incidence of traumatic brain injury in the United States, 2003.   J Head Trauma Rehabil. 2006;21(6):544-548. doi:10.1097/00001199-200611000-00009PubMedGoogle ScholarCrossref
2.
Reiff  DA, Haricharan  RN, Bullington  NM, Griffin  RL, McGwin  G  Jr, Rue  LW  III.  Traumatic brain injury is associated with the development of deep vein thrombosis independent of pharmacological prophylaxis.   J Trauma. 2009;66(5):1436-1440. doi:10.1097/TA.0b013e31817fdf1cPubMedGoogle Scholar
3.
Knudson  MM, Ikossi  DG, Khaw  L, Morabito  D, Speetzen  LS.  Thromboembolism after trauma: an analysis of 1602 episodes from the American College of Surgeons National Trauma Data Bank.   Ann Surg. 2004;240(3):490-496. doi:10.1097/01.sla.0000137138.40116.6cPubMedGoogle ScholarCrossref
4.
Phelan  HA, Eastman  AL, Madden  CJ,  et al.  TBI risk stratification at presentation: a prospective study of the incidence and timing of radiographic worsening in the Parkland Protocol.   J Trauma Acute Care Surg. 2012;73(2)(suppl 1):S122-S127. doi:10.1097/TA.0b013e3182606327PubMedGoogle ScholarCrossref
5.
Spano  PJ  II, Shaikh  S, Boneva  D, Hai  S, McKenney  M, Elkbuli  A.  Anticoagulant chemoprophylaxis in patients with traumatic brain injuries: a systematic review.   J Trauma Acute Care Surg. 2020;88(3):454-460. doi:10.1097/TA.0000000000002580PubMedGoogle ScholarCrossref
6.
Lu  VM, Alvi  MA, Rovin  RA, Kasper  EM.  Clinical outcomes following early versus late pharmacologic thromboprophylaxis in patients with traumatic intracranial hemorrhage: a systematic review and meta-analysis.   Neurosurg Rev. 2020;43(3):861-872. doi:10.1007/s10143-018-1045-yPubMedGoogle ScholarCrossref
7.
Byrne  JP, Mason  SA, Gomez  D,  et al.  Timing of pharmacologic venous thromboembolism prophylaxis in severe traumatic brain injury: a propensity-matched cohort study.   J Am Coll Surg. 2016;223(4):621-631.e5. doi:10.1016/j.jamcollsurg.2016.06.382PubMedGoogle ScholarCrossref
8.
Scudday  T, Brasel  K, Webb  T,  et al.  Safety and efficacy of prophylactic anticoagulation in patients with traumatic brain injury.   J Am Coll Surg. 2011;213(1):148-153. doi:10.1016/j.jamcollsurg.2011.02.027PubMedGoogle ScholarCrossref
9.
Vella  MA, Crandall  ML, Patel  MB.  Acute Management of Traumatic Brain Injury.   Surg Clin North Am. 2017;97(5):1015-1030. doi:10.1016/j.suc.2017.06.003PubMedGoogle ScholarCrossref
10.
Phelan  HA.  Pharmacologic venous thromboembolism prophylaxis after traumatic brain injury: a critical literature review.   J Neurotrauma. 2012;29(10):1821-1828. doi:10.1089/neu.2012.2459PubMedGoogle ScholarCrossref
11.
American College of Surgeons. ACS TQP Best Practices Guidelines. Accessed February 1, 2020. https://www.facs.org/quality-programs/trauma/tqip/best-practice
12.
Pastorek  RA, Cripps  MW, Bernstein  IH,  et al.  The Parkland Protocol’s modified Berne-Norwood criteria predict two tiers of risk for traumatic brain injury progression.   J Neurotrauma. 2014;31(20):1737-1743. doi:10.1089/neu.2014.3366PubMedGoogle ScholarCrossref
13.
Shafi  S, Nathens  AB, Cryer  HG,  et al.  The Trauma Quality Improvement Program of the American College of Surgeons Committee on Trauma.   J Am Coll Surg. 2009;209(4):521-530.e1. doi:10.1016/j.jamcollsurg.2009.07.001PubMedGoogle ScholarCrossref
14.
Shafi  S, Nathens  AB, Parks  J, Cryer  HM, Fildes  JJ, Gentilello  LM.  Trauma quality improvement using risk-adjusted outcomes.   J Trauma. 2008;64(3):599-604.PubMedGoogle Scholar
15.
American College of Surgeons. Trauma Quality Improvement Program. Accessed February 1, 2020. https://www.facs.org/quality-programs/trauma/tqp/center-programs/tqip
16.
Yuan  Y.  Multiple imputation using SAS software.   Journal of Statistical Software. 2011;45(6). doi:10.18637/jss.v045.i06Google Scholar
17.
Sterne  JA, White  IR, Carlin  JB,  et al.  Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls.   BMJ. 2009;338:b2393. doi:10.1136/bmj.b2393PubMedGoogle ScholarCrossref
18.
Cohen  RA. An introduction to PROC LOESS for local regression: SAS Institute Inc. Accessed February 1, 2020. https://stats.idre.ucla.edu/wp-content/uploads/2016/02/loesssugi.pdf
19.
Merlo  J, Yang  M, Chaix  B, Lynch  J, Råstam  L.  A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people.   J Epidemiol Community Health. 2005;59(9):729-736. doi:10.1136/jech.2004.023929PubMedGoogle ScholarCrossref
20.
Mickey  RM, Greenland  S.  The impact of confounder selection criteria on effect estimation.   Am J Epidemiol. 1989;129(1):125-137. doi:10.1093/oxfordjournals.aje.a115101PubMedGoogle ScholarCrossref
21.
Merlo  J, Chaix  B, Ohlsson  H,  et al.  A brief conceptual tutorial of multilevel analysis in social epidemiology: using measures of clustering in multilevel logistic regression to investigate contextual phenomena.   J Epidemiol Community Health. 2006;60(4):290-297. doi:10.1136/jech.2004.029454PubMedGoogle ScholarCrossref
22.
Knudson  MM, Gomez  D, Haas  B, Cohen  MJ, Nathens  AB.  Three thousand seven hundred thirty-eight posttraumatic pulmonary emboli: a new look at an old disease.   Ann Surg. 2011;254(4):625-632. doi:10.1097/SLA.0b013e3182300209PubMedGoogle ScholarCrossref
23.
Denson  K, Morgan  D, Cunningham  R,  et al.  Incidence of venous thromboembolism in patients with traumatic brain injury.   Am J Surg. 2007;193(3):380-383. doi:10.1016/j.amjsurg.2006.12.004PubMedGoogle ScholarCrossref
24.
Ho  KM, Burrell  M, Rao  S, Baker  R.  Incidence and risk factors for fatal pulmonary embolism after major trauma: a nested cohort study.   Br J Anaesth. 2010;105(5):596-602. doi:10.1093/bja/aeq254PubMedGoogle ScholarCrossref
25.
Phelan  HA, Wolf  SE, Norwood  SH,  et al.  A randomized, double-blinded, placebo-controlled pilot trial of anticoagulation in low-risk traumatic brain injury: the Delayed Versus Early Enoxaparin Prophylaxis I (DEEP I) study.   J Trauma Acute Care Surg. 2012;73(6):1434-1441. doi:10.1097/TA.0b013e31825ac49ePubMedGoogle ScholarCrossref
×