Venous thromboembolism prophylaxis in adults with acute traumatic brain injury: a systematic review

Abstract

Background

The optimal timing and type of venous thromboembolism chemoprophylaxis (VTEp) for adults with acute traumatic brain injury (TBI) remains unknown. This systematic review synthesized evidence on the timing and type of VTEp in adults with TBI and highlights evidence gaps.

Methods

We searched Ovid MEDLINE, Embase, and the Cochrane Library through November 4, 2024 for English language, randomized or non-randomized studies with a comparator that reported VTEp timing or agent in adults with acute TBI, and reported intracranial hemorrhage progression (ICHP), VTE, pulmonary embolism (PE), deep vein thrombosis (DVT), neurosurgical intervention (NSI), adverse events (AEs), length of stay (LOS), function, or mortality. Risk of bias (ROB) was assessed with ROBINS-I as low, moderate, serious, or critical risk in studies that used advanced analytic methods (AAMs) to control selection bias; all other studies were deemed critical risk. We qualitatively synthesized evidence and focused text reporting on studies with AAM.

Results

From 738 citations, 21 non-randomized studies met criteria (1 moderate, 20 critical ROB): 16 on VTEp timing, 7 on agents. Seven studies used AAM. For VTEp timing (five AAM studies), one study of adults with mostly mild TBI found no significant difference between Early and Late VTEp on ICHP (moderate ROB); no timing studies with AAM reported ICHP for adults with moderate or severe TBI. Findings were mixed for VTE timing on NSI and PE. Early VTEp was associated with fewer DVTs. There was no difference in mortality or serious AEs by VTEp timing. For VTEp agents, three studies with AAM found no significant difference between low molecular weight heparin (LMWH) and unfractionated heparin (UH) on ICHP, PE, DVT, and serious or overall AEs. Results were mixed for LMWH versus UH effects on NSI after VTEp, VTE, LOS and mortality. TBI severity labeling varied across studies and within measures, reducing comparability.

Conclusions

Clinical evidence on the timing and type of VTEp for adults with acute TBI is of insufficient quality for clinical decision-making. Prospective research designs, standardization of TBI severity labeling, and improved reporting of interventions and outcomes would advance the field.

PROSPERO registration number

CRD42023421534

Level of evidence

Systematic Review, Level IV

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Clinical uncertainty persists about the optimal timing and type of venous thromboembolism chemoprophylaxis (VTEp) for adults with acute traumatic brain injury (TBI). Prior systematic reviews have omitted discussions of study quality, reporting issues, and evidence gaps that limit the field.

WHAT THIS STUDY ADDS

• This systematic review without meta-analysis found that the clinical evidence on the timing and type of VTEp for adults with acute TBI is of insufficient quality for clinical decision-making, especially for adults with moderate to severe TBI.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Prospective designs, harmonization of TBI severity grading across scales, and detailed intervention reporting would advance the field and improve study quality and applicability to better inform clinical care.

Introduction

Venous thromboembolism (VTE) is a highly prevalent and potentially fatal complication among adults hospitalized with acute traumatic brain injury (TBI).1,3 VTE risks are strongly associated with TBI severity. Without pharmacologic chemoprophylaxis, up to one-third of adults with acute TBI will develop VTE during their initial hospitalization.4,7

Recent evidence suggests that some form of early pharmacologic VTE chemoprophylaxis (VTEp) is associated with reduced VTE complications among adults with acute TBI.8,14 There is considerable ongoing clinical interest in identifying the optimal time to initiate VTEp in adults with acute TBI that best balances the potential benefits of VTEp against the risks of facilitating intracranial hemorrhage progression (ICHP).

There is a need for living systematic reviews that periodically query the literature for new evidence and support rapid integration of these studies into previous reviews. How often the evidence is queried may vary. For example, in diseases such as COVID-19 where evidence was generated daily, a much quicker review period is needed (ie, weekly). In cases with slower generation of evidence such as TBI VTEp, quarterly or annual reviews may be more appropriate. This systematic review represents the first step in creating a living systematic review for TBI VTEp whereby the evidence review will be periodically updated. Future review iterations aim to encompass real-world data from the deployment of a clinical decision support tool for VTEp in TBI across multiple US trauma systems.

During the past 15 years, researchers have been investigating subsets of patients with TBI where early VTEp is not associated with increased expansion of ICH (ie, mild disease). Most notably, the Parkland Protocol began to classify and risk stratify adults with acute TBI based on the likelihood of ICHP with VTEp.¹⁵ The risk categories (low, moderate, or high) are based on initial head CT findings, the need for neurosurgical intervention (NSI) (intracranial pressure (ICP) monitor or craniotomy), and ICHP on subsequent CT imaging.^{15 16} The protocol was based on a modification of injury patterns identified by Berne and Norwood years earlier.16,18 The Parkland Protocol developers conducted one pilot randomized trial of early versus later VTEp in adults with low-risk TBI.¹⁹ Their subsequent validation study¹⁶ found that a modified, two-tiered protocol for low-risk versus moderate-to-high-risk TBI adequately classified risk of ICHP between initial and subsequent CT scans.¹⁶

Early investigations were focused on the clinical evidence surrounding pharmacologic VTEp versus none in adults with acute TBI.20,23 More recent studies have focused on the timing of VTEp initiation after acute TBI, the pharmacologic agents used (eg, low molecular-weight heparin (LMWH) or unfractionated heparin (UH)), dosing regimens, and special populations (eg, elderly and/or taking preinjury anticoagulation).11,1424 25 Several studies show benefits of LMWH in reducing VTE events or suggest a protective effect of LMWH over UH26,29 potentially by reducing cerebral edema.³⁰ Facility-level assessments of implementing VTEp after the modified Berne-Norwood criteria have also emerged.³¹

Nonetheless, there remains great clinical uncertainty about the optimal timing and type of pharmacologic VTEp in adults with acute TBI to simultaneously reduce VTE risk and minimize risks of ICHP.^{12 32} National consensus guidance supports the modified Berne-Norwood criteria,³³ with some recent movement toward earlier VTEp in adults with moderate-to-high risk TBI after a stable follow-up CT scan.³ However, in the absence of high-quality randomized trials or prospective studies,⁸ professional society consensus guidance varies on the timing of VTE chemoprophylaxis initiation, particularly for adults with TBI at moderate to high risk of ICHP.^{3 33}

Existing systematic reviews on VTEp in TBI have largely focused on broad, summary reporting across a range of studies and interventions, with less focus on identifying and reporting the highest quality studies with the lowest risk of bias (ROB) to best inform clinical decisions.9,14 Moreover, there are well-known inconsistencies in the reporting of the severity of TBI and concurrent injuries, VTEp timing categories, and longstanding outcomes surveillance issues that are infrequently reported but impact review conclusions.9,14

The goal of this systematic review was to summarize the evidence on outcomes associated with the timing and pharmacologic agent used for VTEp in adults hospitalized with acute TBI. We aimed to report the highest-quality evidence to best inform clinical decisions, and to highlight sample and intervention differences that impact outcomes in this population.

Methods

The systematic review protocol was registered with the International Prospective Register of Systematic Reviews (CRD42023421534, (online supplemental material SM-1). Our reporting follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guideline for a systematic review without meta-analysis (online supplemental SM-2). This review addressed three questions:

1. What evidence informs the timing of VTE chemoprophylaxis (VTEp) initiation after acute TBI in hospitalized adults to minimize the risk of ICHP?

   1a. How does the timing of VTEp initiation differ by TBI severity on initial CT scan?

2. What is the comparative effectiveness of pharmacologic agents used for VTEp in hospitalized adults with acute TBI?

   2a. How does comparative effectiveness differ by TBI severity on initial CT scan?

   2b. What evidence informs dosing for VTEp in acute TBI?

3. What measures were used to report TBI severity for hospitalized adults with acute TBI in included studies?

A medical librarian searched Ovid MEDLINE, Embase, and the Cochrane Library for clinical studies published from 1990 forward on April 28, 2023, and ran update searches on November 4, 2024 (online supplemental SM-3). We selected studies based on prespecified study and PICOT (Population, Intervention, Comparator, Outcomes and Timing) selection criteria (online supplemental SM-4). We included English-language, peer-reviewed randomized or non-randomized studies with a comparison group that reported VTEp timing, agent, or dosing, and reported VTE and related outcomes among adults hospitalized with acute TBI. Studies that examined VTE chemoprophylaxis versus no prophylaxis were retained if they reported VTEp timing or agent subgroup outcomes within the VTEp group. We excluded studies with regression results for VTEp timing or agent without event counts reported or determinable by group.

Two researchers independently screened titles and abstracts from the search results in a web-based tool, PICO portal.³⁴ PICO Portal uses machine learning to sort and first present citations most likely to be eligible, given the selection criteria. We screened citations until the machine learning system reported 100% certainty that none of the remaining citations would meet criteria. Studies that met criteria were reviewed full text by one researcher and checked by another. Disagreements were resolved through discussion. One researcher extracted study data into standardized forms, including study characteristics and reported results; a second researcher checked extraction accuracy.

The primary outcome was ICHP, which included expansion or new hemorrhage sites. Other outcomes included VTE, deep vein thrombosis (DVT), pulmonary embolism (PE), VTEp adverse effects, NSI after VTEp initiation, hospital and intensive care unit length of stay (LOS), inpatient mortality, and function (online supplemental SM-4). We excluded studies that did not report VTE and other outcome events by treatment group as number and percent (eg, reported only overall events²⁶ or ORs³⁵). We report event outcomes (ICHP, VTE, mortality) by group (Early vs Late) for matched or propensity-score matched or weighted cases, or if unmatched, as the unadjusted number and percentage of events by treatment group. We extracted regression outcomes for studies that used advanced analytic methods (AAMs) to control selection bias.

Individual study ROB was assessed with the ROBINS-I tool for non-randomized studies.³⁶ ROBINS-I assesses factors that affect the ROB (study validity or believability) preintervention, at intervention, or postintervention.³⁶ ROBINS-I ROB categories are Low, Moderate, Serious, or Critical risk.³⁶ Domain ratings are made first, followed by a summary judgment across seven domains. If one domain has “Serious risk” then the effect estimate is at serious ROB or worse, even if other domain estimates have less ROB.³⁶ Since retrospective observational studies with inadequate causal inference analytic approaches have critically high ROB, we limited ROBINS-I assessments to retrospective studies that had the potential to be rated as moderate ROB (eg, propensity-score or case matching to reduce selection bias).^{37 38} Retrospective observational studies that did not use these analytic techniques were summarily deemed Critical ROB. One experienced researcher rated the ROB; assessments were checked by a second researcher. Discrepancies were resolved through discussion.

We qualitatively synthesized data. We report basic study information and results for all included studies for VTEp timing (Question 1) and VTEp agent (Question 2) in tables and text. We report outcomes for matched cases only for studies that used propensity score or case matching and regression. For Question 3, we summarized measures used to assess TBI severity in a table and text. We did not conduct a meta-analysis for Question 1 or 2 because studies with high ROB are deemed too problematic for meta-analysis³⁶ (eg, may overestimate treatment effects compared with higher quality studies).³⁹

Results

Our searches identified 738 citations; 69 articles underwent full-text review (figure 1). 21 studies published between 2008 and 2024 met selection criteria (1 prospective, 1 secondary analysis of prospective data, and 19 retrospective cohort studies).1124 27,29 40 When reported, the vast majority of adults had blunt-force TBI.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) diagram detailing literature search and study selection results.

Risk of bias

Of the 21 included studies, 7 retrospective studies used AAMs (propensity-score matching or weighting, or case matching). Four of seven studies examined VTE timing,^{27 48 50 51} one of VTEp timing and agents,⁴¹ and two of VTEp agents^{24 42} (table 1). Only one retrospective observational study with advanced methods was rated as moderate ROB using ROBINS-I (online supplemental SM-5), meaning that the study provides sound evidence for a non-randomized study but cannot be considered comparable to a randomized trial.³⁶ The other six studies with AAMs were rated as having overall critical ROB, meaning that the studies are too problematic to provide useful evidence³⁶ (online supplemental SM-5).

Table 1. Studies of VTE chemoprophylaxis in adults with acute traumatic brain injury that used advanced analytic methods^*.

Author, yearPMIDCountryData sourceStudy type	Matched sample nMean/median age% male% blunt TBITBI severity/GCSMean/median ISSICH pattern(s)	VTEp intervention	Comparison	Reported outcomesDVT screening	Risk of biasAnalytic method
Q1: VTEp timing
Jakob et al, 20224834,809,908USAACS-TQIP 2013–2017Retrospective	n=4,30456 years69% male100% blunt“Severe” but median GCS=14 (IQR 12–15)ISS: 16 (IQR 10–20)Isolated SDH+SAH	Early ≤48 hoursn=2,152UH or LMWHDose NRDuration NR	Late >48 hoursn=2,152UH or LMWHDose NRDuration NR	Delayed craniectomy, VTE, PE, DVT, LOSICU use, ICU LOS, mortalityDVT screening NR	CriticalMatched cohort
Jakob et al, 20222734,364,655USAACS-TQIP 2013–2016 Retrospective	n=3,32058 years63% male100% blunt“Severe” but median GCS=15 (IQR 14–15); GCS ≤8: 8.8%ISS: 17 (IQR 16–21)Isolated ASDH	Early ≤48 hoursn=1,660UH or LMWHDose NRDuration NR	Late >48 hourn=1,660UH or LMWHDose NRDuration NR	VTE (PE+DVT)ICU use, ICU LOShospital LOSmortalityDVT screening NR	CriticalMatched cohort
Frisoli et al, 20175028,973,510USAL1TC 2011–2013Retrospective	n=28254 years77% male% blunt: NR82% mild TBI; mean GCS 13.2, range 3–15Mean ISS: 13.9–15.324% preinjury AC/AP62.1% mixed ICH	Early <24 hoursn=94UH 5,000 u 2×/day orEnoxaparin 40 mg/dayDuration NR	Delayed >48 hoursn=188 UH 5,000 u 2 ×/day orEnoxaparin 40 mg/dayDuration NR	ICH expansionVTE, ICP monitor,neurosurgery after VTEp (new or return), mortalityDVT screening NR	ModeratePS-matched cohort
Byrne et al, 20165127,453,296CanadaACS-TQIP 2012–2014 Retrospective	n=2,46843 years77% male100% blunt100% severe TBI; all GCS ≤889% ICH (11% other head injury), 52% SAH, 56% SDH, 10% EH	Early <72 hoursn=1,234LMWH or UHDose NRDuration NR	Late ≥72 hoursn=1,234LMWH or UHDose NRDuration NR	VTE (PE, DVT), mortalityLate ICP monitor or craniotomy/ectomy13% of centers had >10% of adults who had US for DVT	CriticalPS-matched cohort
Q1 and 2: VTEp timing and agent
Johnson et al, 20244138,745,357USAMI-TQIP 2017–202235 L1-2TCsRetrospective	Matched n (%) NR; n=8,715 unmatched had VTEp; matched n varied by comparison (n=875–1,995)58–70 years (mean NR)58–67% male (NR)100% bluntGCS NR (only GCS motor); Head AIS 3–5.Mean ISS NR (all ≥5), multiple injuries incl.Prior AC/AP included	LMWH ≤48 hours LMWH >48 hours UH ≤48 hours UH >48 hours n differed by comparisonDose NRDuration NR	No VTEp	Mortality; NSI after VTEp, VTE, mortality after (VTEp+NSI), LOS, serious AENo routine DVT screening	CriticalSome outcomes are PS-matched
Q2: VTEp agent
Condon et al, 20242437,306,760USAACS-TQIP 2017–2019Retrospective	n=14,926 (unmatched; matched NR)57% maleAll ≥65 yearsAll head AIS ≥3% blunt: NROverall GCS: NRExcluded: serious non-head injuries, on LMWH or UH (other AC/AP NR).Included NSI ≤24 hoursICH patterns: NR	LMWHn=11,036 unmatched;Dose NRDuration NR	UHn=3,890 unmatched;Dose NRDuration NR	VTE, PE, DVT, mortality, NSIHas time to VTEp (table 1) but not event countsDVT screening NR	CriticalPS-matched case control (third analysis)
Ratnasekera et al, 20244238 614 835EAST multicenter retrospective2014–202024 L1-2TCs	n=984 (979 VTEp)Isolated TBI, 96% blunt63 years.63% male66% white raceMedian GCS 14 (IQR 11–15). AIS ≥3: 43%=3, 30%=4, 27%=5ISS NRPatterns: 47% SAH, 72% SDH, 6% EDH, 3% IVHPrior AC/AP excluded	LMWHn=502Dose and timing varied;Duration NR	UHn=482Dose & timing varied;Duration NR	ICHE, VTE, mortality, NSI, RTOR, LOS, AEs, MTP, discharge siteICHE (or new) per radiologist. VTE per venogram, DVUS, or CT angiogram.No routine DVT screening	CriticalPS-weighted Cox PH, clustered on facility

^*Advanced analytic methods: propensity score or case matching with regression to control for selection bias.

AC, anticoagulation; ACS, American College of Surgeons; ACS-TQIP, ACS Trauma Quality Improvement Program data; AE, adverse event; AIS, Abbreviated Injury Scale; AP, antiplatelet; ASDH, acute subdural hematoma; DVT, deep vein thrombosis; DVUS, duplex venous ultrasound; EH, epidural hematoma; GCS, Glasgow Coma Scale; ICH, intracranial hemorrhage; ICHE, intracranial hemorrhage expansion; ICP, intracranial pressure; ICU, intensive care unit; ISS, Injury Severity Score; IVH, interventricular hemorrhage; L, Level (trauma center); LMWH, low-molecular-weight heparin; LOS, length of stay; L1TC, Level 1 Trauma center; MI-TQIP, Michigan Trauma Quality Improvement Program; MTP, Massive Transfusion Protocol; NR, not reported; NSI, neurosurgical intervention; PE, pulmonary embolism; PH, Proportional Hazards; PS, propensity score; RTOR, Return to OR; SAH, subarachnoid hemorrhage; SDH, subdural hemorrhage/hematoma; TBI, traumatic brain injury; u, units; UH, unfractionated heparin; US, ultrasound; VTE, venous thromboembolism; VTEp, VTE chemoprophylaxis.

14 studies did not use AAMs to reduce selection bias and were deemed critical ROB without formal assessment (online supplemental SM-6 2).1128 29 40 43,47 49 52

Across all included studies, the main factors that contributed to high ROB were baseline confounding, outcomes measurement issues, and lack of intervention details.

Timing of VTEp initiation (Q1)

16 studies examined the timing of Early compared with Late VTEp: one prospective cohort study,¹¹ a secondary analysis of prospective data,⁴⁶ and 14 retrospective observational studies2740 41 43,45 47 (table 1, online supplemental SM-6). Mean patient age varied from 29⁴⁶ to 62 years.⁴⁴ Patients were predominantly male (range 53%⁴⁵ to 95%).⁴⁷ TBI severity was inconsistently reported between text (eg, severe TBI) and severity scale metrics, and was sometimes indeterminable (table 1, online supplemental SM-6; see Question 3 below). For example, two studies with sample median Glasgow Coma Scale (GCS) scores of 14 and 15 were author-labeled as severe TBI studies.^{27 48}

VTEp intervention reporting was varied and often incomplete. The definition of Early VTEp initiation varied across studies and within a study,⁵⁴ and ranged from less than 24 hours up through 72 hours (online supplemental SM-7). Five studies reported Early VTEp as less than or equal to 24 hours,^{40 44 45 49 50} five as up to and including 48 hours,^{27 41 46 48 52} and five studies as less than^{11 51} or equal to 72 hours.^{47 53 54} One study reported four VTEp initiation time windows in hours, without text labels.⁴³ Individual study details on agents (LMWH, UH, or both) within VTEp timing groups, dosing, and duration are in table 1 and online supplemental SM-6. 11 VTEp timing studies (69%) did not report drug dose or duration.

Eligible sample sizes ranged from 53¹¹ to 4,304⁴⁸ (table 1, online supplemental SM-6). Data sources included single or multiple Level 1 trauma centers, or the American College of Surgeons Trauma Quality Improvement Program (ACS-TQIP) trauma databases. Six studies used the ACS-TQIP^{27 40 41 43 48 51} with data years 2008⁴³ through 2022.⁴¹ Data exclusions in the ACS-TQIP studies were commonly due to facility-level missing data, missing VTEp or outcomes within records, and clinical reasons (eg, early mortality). Missing data sometimes accounted for more than one-third of case exclusions⁵¹ or was indeterminable, given the manner of excluded records reporting.^{27 48}

Outcomes ascertainment varied (table 1, online supplemental SM-6). Criteria for initiating diagnostic imaging for ICHP or PE were rarely reported. Only two studies (13%) conducted routine DVT surveillance with lower extremity duplex ultrasound.^{11 54} Most studies did not report about DVT screening or noted that surveillance was up to treating clinicians or local facility protocols. One study without routine DVT screening reported that only 42% of the LMWH group had undergone screening during their hospital stay.⁵²

Summary of findings on VTEp timing

The findings of five studies of VTEp timing used AAMs to control for treatment selection bias are summarized in table 2.^{27 41 48 50 51} Although only one was rated as moderate ROB⁵⁰ and four as having critical risk,^{27 41 48 51} all five studies are summarized in table 2 as the best available evidence on VTEp timing for adults with acute TBI. Individual study results can be found in online supplemental SM-8a.

Table 2. Summary of findings: Early versus Late VTE prophylaxis in adults with acute TBI: studies with advanced analytic methods^*.

OutcomeComparison	Study # and typeData sourceN analyzed†	Sample TBI severity	Summary of findings
ICH progression	1 PS-matched cohort50L1TCn=282	Mostly mild (per authors):82% GCS ≥13; mean GCS 13.2, range 3–15	1 found no difference between <24 hours and >48 hours VTEp
Delayed craniectomy or craniotomy; or neurosurgical intervention (NSI) after VTEp	4 cohort studies: 2 PS-matched,41 51 2 case matched27 483 ACS-TQIP: 1 MI-TQIPn ≤13,508	Mixed:1: Severe, all GCS≤81: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15, 9% GCS ≤8)1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	3 found no difference using 48 hours or 72 hours cut-offs between Early and Late VTEp1‡ found significantly higher incidence and risk of NSI after Early LMWH ≤48 hours vs Late >48 hour LMWH for both Isolated TBI and All TBI for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)
Delayed ICP monitor	2 cohort studies: 1 PS-matched,51 1 case matched27ACS-TQIPn=5,788	Mixed:1: Severe, all GCS ≤81: Isolated SDH, median GCS=15 (IQR 14–15), 9% GCS ≤8	2 found no difference using a ≤48 hours or 72 hours cut-off between Early and Late VTEp
ICP monitor and/or craniectomy	1 PS-matched cohort50L1TCn=282	Mostly Mild:82% mild; mean GCS 13.2, range 3–15	1 found no difference between <24 hours and >48 hours VTEp
VTE	4 cohort studies: 2 PS-matched,41 50 2 case matched27 482 ACS-TQIP, 1 MI-TQIP, 1 L1TCn≤11,322	Mixed:1: 82% mild; mean GCS 13.2, range 3–15. Mean ISS 14–151: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15, 9% GCS ≤8)1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	2 found benefit of Early VTEp ≤48 hours vs VTEp >48 hours for isolated SDH or isolated (SDH+SAH) with median GCS 14 (IQR 12–15) or 15 (IQR 14–15)1‡ study:All TBI: found benefit of Early LMWH ≤48 hours vs LMWH >48 hours for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)Isolated TBI: found no difference between LMWH ≤48 hours and LMWH >48 hours for unknown GCS, Head AIS 3–5, and mean ISS NR (all≥5)1 found no difference between <24 hours and >48 hours VTEp for mostly mild TBI
Pulmonary embolism	3 cohort studies: 1 PS-matched,51 2 case matched27 483 ACS-TQIPn=10,092	Mixed:1: Severe, all GCS≤81: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15), 9% GCS ≤8	2 found benefit of Early VTEp using a 48 hours or 72 hours cut-off between Early and Late VTEp1 found no difference between ≤48 hours and >48 hours VTEp
Deep vein thrombosis	3 cohort studies: 1 PS-matched,51 2 case matched27 483 ACS-TQIPn=10,092	Mixed:1: Severe, all GCS ≤81: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15), 9% GCS ≤8	3 found benefit of Early VTEp using 48 hours or 72 hours cut-offs between Early and Late VTEp
Hospital length of stay	3 cohort studies: 2 case-matched27 48; 1 PS-matched412 ACS-TQIP, 1 MI-TQIPn ≤11,040	Mixed:1: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15), 9% GCS ≤81: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	2 found benefit of Early VTEp ≤48 hours vs >48 hours VTEp1‡ found benefit of Early LMWH ≤48 hours vs LMWH >48 hours for All TBI and Isolated TBI with unknown GCS, Head AIS 3–5, and mean ISS NR (all≥5)
MortalitySerious adverse events	5 cohort studies: 3 PS-matched,41 50 51 2 case matched27 481 L1TC, 3 ACS-TQIP, 1 MI-TQIPn ≤13,7901 PS-matched41MI-TQIPn=2,334 isolated,3,416 all TBI	Mixed:1: Severe, all GCS≤81: 82% mild; mean GCS 13.2, range 3–151: Isolated (SDH+SAH), median GCS=14 (IQR 12–15)1: Isolated SDH, median GCS=15 (IQR 14–15), 9% GCS ≤81: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS≥5Unable to analyze‡:1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS≥5	4 found no difference between Early and Late VTEp using <24 hours, ≤48 hours or <72 hours cut-offs between Early and Late VTEp1‡ found no difference between Early LMWH ≤48 hours and Late LMWH >48 hours for All TBI and Isolated TBI with unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)1‡ found no difference in risk between Early LMWH ≤48 hours and Late LMWH >48 hours for All TBI and Isolated TBI

^*Advanced analytic methods: propensity score or case matching with regression to control for selection bias.

^†Number analyzed per outcome varied when Johnson 2024 was included because N differed for All TBI and Isolated TBI.

^‡Johnson et al 202441: applicability indeterminable due to lack of sample details on TBI and injury severity for PS-matched and unmatched samples.

AC, anticoagulation; ACS, American College of Surgeons; ACS-TQIP, ACS Trauma Quality Improvement Program; AIS, Abbreviated Injury Scale; Found benefit, statistically significant difference between groups favoring Early VTEp; GCS, Glasgow Coma Scale; ICH, intracranial hemorrhage; ICP, intracranial pressure; ISS, Injury Severity Score; LMWH, low molecular weight heparin; L1TC, Level 1 Trauma center; MI-TQIP, Michigan Trauma Quality Improvement Program; n, number analyzed; NR, not reported; PS, propensity score; SAH, subarachnoid hemorrhage; SDH, subdural hemorrhage/hematoma; TBI, traumatic brain injury; VTE, venous thromboembolism; VTEp, VTE prophylaxis.

There were no significant differences in ICHP (one study, 82% GCS ≥13⁵⁰), delayed ICP monitor,^{27 51} or ICP monitor and/or craniectomy⁵⁰ between Early (<24 hours or ≤48 hours) and Late VTEp (table 2). Results were mixed for NSIs, including delayed craniotomy, delayed craniectomy, or any NSI after VTEp initiation, where three studies with heterogeneous samples^{27 48 51} found no significant difference between Early and Late VTEp using 48 or 72 hours cut-offs, and one study found significantly higher risk of NSI with Early compared with Late LMWH in both isolated and across all adults with TBI.⁴¹

Four studies reported VTE outcomes with mixed results in adults with varied or indeterminable TBI severity (table 2). Two reported benefits of Early VTEp within 48 hours for isolated subdural hematoma (SDH), or isolated SDH with subarachnoid hemorrhage,^{27 48} one study found no difference between Early (<24 hours) and Late (>48 hours)⁵⁰ VTEp, and one study found a benefit of Early VTEp for adults with all TBI but not for adults with isolated TBI.⁴¹

Three studies reported PE among adults with varied TBI severity.^{27 48 51} One study found no significant difference in PE between Early and Late VTEp,⁴⁸ one reported a benefit of Early VTEp,²⁷ and one reported an adjusted OR that favored Early VTEp.⁵¹

For DVT, three ACS-TQIP studies reported significantly lower DVT with Early VTEp within a 48-hour or 72-hour VTEp initiation window compared with Late.^{27 48 51}

Three studies reported hospital LOS by VTEp timing. All three reported an Early VTEp benefit (≤48 hours)^{27 41 48} among adults with mixed TBI severity and with or without extracranial injuries (table 2, online supplemental SM-8a).

There were no significant differences in inpatient mortality between Early and Late VTEp groups in any of the five studies that used AAMs (table 2, online supplemental SM-8a). One study found no difference in serious adverse events (AEs) between Early and Late LMWH (table 2, online supplemental SM-8a).⁴¹

In summary, among VTEp timing studies with AAMs (table 2), only one moderate ROB study of adults with mostly mild TBI addressed the primary ICHP outcome and found no significant difference between Early and Late VTEp.⁵⁰ No studies of VTEp timing reported on ICHP for adults with moderate or severe TBI. Findings on VTEp timing were mixed for NSI and PE. Early VTEp was associated with fewer DVTs. All five studies reported no difference in mortality by VTEp timing, and one found no difference in serious AEs.

For Question 1a, we are unable to report potential outcomes differences for Early versus Late VTEp by TBI severity because severity measures varied (Head Abbreviated Injury Scale (AIS) and GCS, total or motor only) and text labels (eg, severe TBI) did not match standard definitions in the field.56,58

Other timing studies

11 critically high ROB studies that did not use AAMs (online supplemental SM-6) reported mixed results for VTEp timing effects.1140 43,47 49 52 Detailed results for these studies are in the online supplemental SM-8b.

VTEp agents (Q2)

Seven retrospective studies published between 2010 and 2024 compared VTEp agents among adults hospitalized with acute TBI. Three studies that used AAMs^{24 41 42} (table 1) are summarized here. Details on the four studies without AAMs are in online supplemental SM-6; their study results are in online supplemental SM-8b.^{28 29 40 55}

Across the three studies with advanced methods, mean patient age was over 63 years²⁴,⁴² or not reported for the matched samples.⁴¹ One study focused exclusively on adults age 65 or older, and that sample also included adults who underwent early NSI (≤24 hours).²⁴ Males dominated the samples (≥57%). TBI severity varied (table 1). One study focused on adults with isolated TBI,⁴² one included older adults with other non-serious injuries,²⁴ and one lacked details about any matched sample characteristics but reported outcomes for all TBI and isolated TBI.⁴¹ Sample sizes ranged from just under 1,000⁴² to almost 15,000.²⁴ Study data came from the ACS-TQIP^{24 41} or 24 Level 1 or 2 trauma centers.⁴² All three studies compared LMWH with UH.^{24 41 42} None of the three studies reported drug dose or duration, and only one reported that dose and timing varied.⁴² None of the three studies reported routine DVT screening.

Summary of findings on VTEp agents

The findings of three studies of VTEp agents (LMWH compared with UH) that used AAMs to control treatment selection bias are summarized in table 3. Although all three studies of VTEp agents were rated as having critical ROB,^{24 41 42} they are summarized below as the best available evidence on VTEp agents for adults with acute TBI. Their individual study results are in online supplemental SM-8a.

Table 3. Summary of findings: low molecular weight heparin versus unfractionated heparin for VTE prophylaxis in adults with acute traumatic brain injury from three retrospective studies with advanced analytic methods^*.

OutcomeComparison	Study # and typeData sourceN analyzed†	Sample TBI severity(matched or weighted sample)	Summary of findings
ICH expansion (ICHE) or new ICH	1 PS-weighted cohort4224 L1–2TCsn=979	Mixed, isolated TBI:Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; 96% blunt TBI	1 found no difference between LMWH and UH
Craniectomy, craniotomy, or neurosurgical intervention (NSI) after VTEp	2: 1 PS-weighted,42 1 PS-matched cohort4124 L1–2TCs, 1 MI-TQIPn=2,729 isolated;2,314 all TBI	Mixed:1: Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; isolated TBI, 96% blunt1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	1 found no difference in craniotomy or craniectomy between LMWH and UH1‡ found higher risk of NSI after early LMWH ≤48 hour vs early UH ≤48 hour for both Isolated TBI and All TBI for unknown GCS, Head AIS 3–5, and ISS all ≥5
VTE	2 PS-matched cohort studies24 411 ACS-TQIP, 1 MI-TQIP,n>1,750 but indeterminable	Mixed and indeterminable:1: GCS and Head AIS distribution NR for matched cohort; 78% of unmatched cohort had GCS 14–15; <9% unmatched had GCS 3–8. All age ≥65 year.1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	1 found no difference in the AOR between LMWH and UH in a non-descript, elderly, matched sample1‡ study: All TBI: found benefit of LMWH ≤48 hour vs UH ≤48 hour for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)Isolated TBI: found no difference between LMWH ≤48 hour and UH ≤48 hour for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5). Borderline p value (0.052)
Pulmonary embolism	1 PS-weighted cohort4224 L1–2TCsn=979	Mixed, isolated TBI:Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; 96% blunt TBI	1 found no difference between LMWH and UH
Deep vein thrombosis	1 PS-weighted cohort4224 L1–2TCsn=979	Mixed, isolated TBI:Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; 96% blunt TBI	1 found no difference between LMWH and UH
Hospital length of stay	two cohort studies: 1 PS-matched,41 1 PS-weighted421 MI-TQIP, 1 multicenter L1–2TCs n=2,729 isolated TBI; 2,314 not isolated TBI	Mixed:1: Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; isolated TBI, 96% blunt1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5	1 found no difference between LMWH and UH1‡ found:All TBI: benefit of LMWH ≤48 hour vs UH ≤48 hour for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)Isolated TBI: no difference between LMWH and UH for unknown GCS, Head AIS 3–5, and mean ISS NR (all ≥5)
MortalitySerious adverse eventsOverall complications	2 PS-matched cohorts24 41; 1 PS-weighted cohort1 ACS-TQIP, 1 MI-TQIP, 1 multicenter L1–2TCs n=2,729 isolated TBI; 3,293 all TBI1 PS-matched cohort41MI-TQIPn=1,750 isolated TBI,2,314 not isolated TBI1 PS-weighted cohort4224 L1–2TCsn=979	Mixed or indeterminable:1: GCS and Head AIS distribution NR for matched cohort; 78% of unmatched cohort had GCS 14–15; <9% unmatched had GCS 3–8. All age ≥65 year.1: GCS NR; all Head AIS 3–5; % NR for head AIS 3, 4 or 5. Mean ISS NR (all ≥5)1: Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%, isolated TBI, 96% bluntMixed:1: GCS NR (motor only); Head AIS 3–5 but n (%) NR for Head AIS 3, 4 or 5. All ISS ≥5Mixed, isolated TBI:Median GCS 14 (IQR 11–15), Head AIS: 3=43%, 4=30%, 5=27%; 96% blunt TBI	1 found benefit in the AOR favoring LMWH compared with UH in a non-descript, elderly matched sample.1‡ found no difference between LMWH ≤48 hour and UH ≤48 hour for unknown GCS, Head AIS 3–5, and ISS all ≥5for both All TBI and Isolated TBI1 found no difference between LMWH and UH for mostly blunt, isolated TBI of mixed severity1‡ found no difference in risk between LMWH ≤48 hour and UH ≤48 hour for unknown GCS, Head AIS 3–5, and ISS all ≥5 for both All TBI and Isolated TBI1 found no difference between LMWH and UH

^*Advanced analytic methods: propensity score matching, weighting, or case matching with regression to control for treatment selection bias.

^†Number analyzed per outcome varied when Johnson 2024 was included because N differed for All TBI and Isolated TBI.

^‡Johnson et al 202441: applicability indeterminable; lack of sample details on TBI and injury severity for PS-matched and unmatched samples.

ACS, American College of Surgeons; ACS-TQIP, ACS Trauma Quality Improvement Program; AIS, Abbreviated Injury Scale; AOR, adjusted OR; Found benefit, statistically significant difference between groups; GCS, Glasgow Coma Scale; ICH, intracranial hemorrhage; ISS, Injury Severity Score; LMWH, low molecular weight heparin; L1–2TCs, Level 1–2 Trauma centers; MI-TQIP, Michigan Trauma Quality Improvement Program; n, number analyzed; NR, not reported; PS, propensity score; TBI, traumatic brain injury; UH, unfractionated heparin; VTE, venous thromboembolism; VTEp, VTE prophylaxis.

There were no significant differences in ICH expansion or new ICH between LMWH and UH in one study of adults with isolated TBI of mixed severity (table 3).⁴² Results from two studies were mixed for NSI after VTEp initiation.^{41 42} One study found no difference in delayed craniectomy or craniotomy between LMWH and UH,⁴² and one study found a higher risk of NSI after early LMWH compared with early UH for both isolated and all TBI.⁴¹

Two studies reported VTE outcomes with mixed results. One study of adults aged 65 or older reported no difference between LMWH and UH on VTE. The other study reported a benefit for LMWH over UH for all TBI, but no difference between agents when they limited their sample to adults with isolated TBI⁴¹ (table 3). One multicenter study separately reported PE and DVT and found no significant difference between LMWH and UH.⁴² None of the 24 sites in the study had routine DVT screening.⁴²

Hospital LOS had mixed results for VTEp agent. One multicenter study found no difference in LOS among adults with TBI who received LMWH or UH,⁴² and one study found shorter LOS with LMWH for all TBI, but no difference in LOS in isolated TBI.⁴¹

There were no significant differences in serious AEs of overall complications with LMWH compared with UH.^{41 42}

In summary, three studies with advanced methods to control selection bias found no significant difference between LMWH and UH on ICH expansion/new ICH, PE, DVT, serious AEs, and overall complications (table 3). However, results were mixed for LMWH compared with UH on NSI/craniotomy/craniectomy after VTEp, VTE, hospital LOS and mortality. Evidence is insufficient to address potential variation in drug effectiveness by TBI severity (2a) and dosing comparisons (2b).

TBI severity measures (Q3)

Measures used to analyze admitting TBI severity varied (table 4). Six of 21 studies reported TBI severity by the GCS, 2 reported only the GCS motor score,^{41 43} 3 reported only the Head AIS, and 8 studies reported both the GCS and Head AIS. One study reported TBI severity only by Marshall’s CT Classification.⁵⁴ One study used GCS (all ≤8) and Head AIS (all >2) for sample selection but did not report either metric with sample characteristics.⁴⁰

Table 4. Measures used to report TBI severity in the included literature.

Author, year	Reported TBI severity (text)	GCS	Head AIS	Other	Reported TBI severity/ranges
VTEp timing
Shulkosky et al, 202343	NR	X motor			GCS motor only (1–6, 65%=6)All Head AIS ≥3 but Head AIS NR
Wu et al, 202346	Severe	X	X		Only percent with GCS <9: 55%All Head AIS 3–5 (30%=3; 37%=5)
Stormann et al, 202245	Severe	X	X		Head AIS ≥3
Stormann et al, 201944	Severe		X		Mean Head AIS: 3.6±0.8
Al-Dorzi et al, 202247	Severe (80%) to moderate	X			GCS <9: severe (80%*)GCS 9–12: moderateMean GCS∼6.4 ± 2.9
Jakob et al, 202248	“Severe” per authors	X	X		Median GCS 14 (IQR 14–15);Head AIS 3–5 (56%=4)
Jakob et al, 202227	“Severe” per authors	X	X		Median GCS 15 (IQR 14–15)
Rivas et al, 202249	NR (average appears mild)	X			Mean GCS 13±4 and 14±3 by group
Hachem et al, 201811	Severe	X			All GCS ≤8: Mean GCS 4.7±2.0 Early; Mean GCS 5.1±1.9 Late
Frisoli et al, 201750	Mild (82%)	X			Mean GCS 13.2, range 3–15
Byrne et al, 201651	Severe	X	X		Head AIS ≥3; GCS ≤8
Kwiatt et al, 201252	Severe	X			Mean GCS=8
Koehler et al, 201153	Moderate (84%) to severe		X		39% Head/neck AIS=3;45% Head/neck AIS=4
Depew et al, 200854	NR (average appears mild)			Marshall’s Head CT	86% Marshall’s Head CT classification=2
VTEp timing and agents
Elkbuli et al, 202440	Severe	–	–		All GCS ≤8 but GCS NRAll Head AIS >2 but Head AIS NR
Johnson et al, 202441	Severe	X motor			All Head AIS >2 but Head AIS NRGCS motor 1–6 (58–75%=6)
VTEp agents
Condon et al, 202424	Severe	X	X		Head AIS ≥3: matched sample % NRGCS 3–15 (most 14–15); matched sample % NR
Ratnasekera et al, 202442	Severe	X	X		Median GCS 14 (IQR 11–15)All Head AIS ≥3 (43%=3; 27%=5)
Filiberto et al, 202328	“Severe” per authors	X	X		Median GCS 14 (IQR 8–15)All Head AIS 3–5 (67%=3; 12%=5)
Minshall et al, 201129	Severe		X		100% Head AIS >2
Dudley et al, 201055	Moderate to severe	X			Mean GCS 7.4 (SD 3);range 3–12

^*Al-Dorzi: GCS <9 is 80.1% of the 276 who received VTEp (excluding the no VTEp group).

AIS, Abbreviated Injury Scale; GCS, Glasgow Coma Scale; NR, not reported; TBI, traumatic brain injury; VTEp, venous thromboembolism prophylaxis.

There were considerable differences across studies in the labeling of severe TBI.²⁸ Jakob et al described samples from two studies as severe but the median GCS for each study was 14 (IQR 14–15)⁴⁸ and 15 (IQR 14–15).²⁷ Other authors defined severe TBI as GCS less than or equal to 8.^{11 40 46 47 51 52 55} The manner of Head AIS reporting also varied and only five studies that included adults with Head AIS 3–5 provided a distribution (number and percent) for each Head AIS value in their samples.^{28 42 46 48 53} There were inconsistencies in text severity labels when both the GCS and Head AIS scales were reported, whereby authors reported that their sample had severe TBI by Head AIS 3–5, but median GCS scores were 14.^{28 42} Cut points for TBI severity scales were not referenced.

Discussion

The peer-reviewed clinical evidence to date on the timing and type of VTEp for hospitalized adults with acute TBI is of insufficient quality for clinical decision-making. We did not find any low ROB (higher quality) studies that compared Early to Late VTEp in adults with acute TBI, or head-to-head drug comparisons for VTEp. Only one moderate ROB study with AAMs compared Early with Late VTEp in adults with mostly mild TBI (82% GCS ≥13) and found no significant difference in ICHP, VTE, NSI or mortality.⁵⁰ The remaining literature in this review, including all studies that reported on adults with moderate or severe TBI, was deemed critically high ROB, meaning that the studies are too problematic to provide useful evidence, and that higher quality evidence may change conclusions.³⁶ We are unable to report potential outcome differences for TBI severity subgroups because TBI severity labeling was heterogeneous and inconsistent within and across measures. Collectively, these findings represent a considerable knowledge gap surrounding VTEp for adults with acute TBI and highlight the need for higher-quality randomized and prospective studies with standardized severity labeling to address knowledge gaps in future research.

Besides VTEp, we identified several confounding factors that may account for the observed outcomes differences between groups in non-randomized studies. Observational studies can only account for measured factors and the extent to which they were recorded. Early and Late VTEp groups were quite different at baseline, and these differences may account for provider decisions to use Late VTEp in more critically injured adults. Baseline injury and medical severity metrics that were not reported in a dataset, were categorized inaccurately or insufficiently (eg, binary comorbidity coding), or excluded from matching or analytic models, can overwhelm observational study conclusions about the effects of any intervention on outcomes. Uncommon or rare outcome events magnify these concerns and large datasets do not remedy them. Propensity score matching, even with many predictor variables, only accounts for measured factors. Some study authors noted that the Late VTEp groups had more severe injuries than adults in the Early VTEp groups on several critical factors. These differences drive outcomes estimates, especially for uncommon outcomes of ICHP and VTE. Likewise, decisions to use UH versus LMWH are impacted by TBI and related injury severity. Several studies noted that more severely injured adults received UH.^{28 29 35} Therefore, the lack of comparability of baseline samples in observational studies, combined with no standard screening or assessment criteria for the main outcomes of this review (ICHP, DVT, and PE) makes it extremely likely that numerous outcome events were missed in adults with TBI, many of whom were on ventilators.

Limitations of the literature

The TBI VTEp literature is observational, retrospective, and heterogeneous. The following limitations, if corrected, offer a roadmap to improve the quality of future TBI VTEp research:

1. Inconsistent tools for TBI severity grading: study authors used three measures for TBI severity grading (GCS total or motor, Head AIS, Marshall’s Head CT) and no study reported all three metrics.

2. Inconsistent labeling of TBI severity: when reported, author labels for TBI severity (ie, mild, moderate, severe) were sometimes inconsistent with numerical TBI severity ratings from various tools (eg, GCS).^{27 28 48} Although TBI is a heterogeneous and complex condition, inconsistency in TBI severity labeling complicates comparability across studies. Standardization of text severity labeling across TBI severity scales by harmonizing the GCS and AIS thresholds, establishing publishing standards based on professional society recommendations (eg, multiple TBI severity metrics with referenced cut-points and subcategory distribution reporting), plus journal editor buy-in, would advance the field.^{56 57}

3. Limited or inconsistent reporting of concurrent injuries: although some studies reported GCS and Injury Severity Score (ISS), some did not report concurrent fractures, injuries, or operations that may increase VTE risk. A recommended minimum set of co-injury identifiers for TBI-related studies would advance the field.

4. Lack of patient-centered outcomes: studies focused on inpatient event outcomes. Functional recovery and quality of life are important to patients and families and should be included in future, higher quality prospective studies.

5. Under the identification of DVT: only 2 of 21 studies routinely screened for DVT.^{11 54} Undercounting DVT also undercounts VTE. The lack of routine DVT screening is a significant issue in this neurologically impaired population, and biases outcomes estimates that are necessary to analyze best practices. Studies that conducted routine screening found more DVTs.^{11 54}

6. Limitations of the ACS-TQIP and reporting: the ACS-TQIP does not contain follow-up CT scan information and therefore cannot assess ICHP in TBI, the main outcome of interest in VTEp studies. Proxy markers using procedures sometimes associated with ICHP (eg, NSI) underestimate the frequency of ICHP. Additionally, ACS-TQIP does not include detailed information necessary to assess the risk of TBI progression (ie, SDH size). Aside from expected clinical exclusions (eg, early mortality), ACS-TQIP studies omitted 20% or more of eligible adults due to missing data on VTEp at the facility level, when reported.^{27 48} Combined with missing data for key variables at the individual level, these exclusions represent substantial missing data^{28 40} that is not random and undoubtedly impacts outcomes estimates. Other important factors that were missing or not reported in ACS-TQIP were DVT screening, and VTEp dose and duration. Several authors limited studies to adults with isolated TBI⁵¹ and/or specific CT bleed patterns,^{27 48} which reduces heterogeneity but may also reduce clinical applicability when certain ICH patterns do not commonly occur as isolated TBI injuries. The inclusion of additional variables into the national trauma data standard with routine collection in ACS-TQIP may improve future research to some degree, such as CT characteristics (eg, hematoma size), presence of TBI progression, and details on VTEp timing and initial dose. Until then, the ACS-TQIP cannot provide valid estimates of ICHP after VTEp.

7. Lack of randomized trials on VTEp timing and dose: we found no randomized trials for this review and 90% of included studies were retrospective cohort studies. Higher quality evidence, including randomized trials and large prospective cohort studies, with standardized VTEp protocols, follow-up CT imaging, and comprehensive baseline severity and comorbidity reporting, is essential to move beyond the current limitations of the TBI clinical evidence.⁸

Limitations of this review

We did not review the literature to date on VTEp versus no VTEp since the state of clinical practice suggests that this is a resolved question for TBI. Other interventions remain controversial but were beyond the scope of this review (eg, inferior vena cava filters). We conducted ROB assessments on a subset of studies but have presented table results for all studies. Given the lack of higher quality research, we did not assess the strength of evidence. Although some studies included small proportions of adults who were on anticoagulant or antiplatelet agents prior to their TBI, we did not focus on this growing subgroup of older adults whose outcomes are simultaneously impacted by anticoagulation effects at admission for acute TBI, and the impact of reversal agents.

Conclusion

The peer-reviewed clinical evidence to date on the timing and type of VTEp for hospitalized adults with acute TBI is of insufficient quality for clinical decision-making. This systematic review makes a valuable contribution to the field by consolidating the existing evidence on VTEp in TBI and identifying critical research gaps. Higher quality study designs, harmonization of TBI severity scale thresholds, and standardization of baseline reporting requirements would reduce bias and improve reporting issues that currently limit the field. Such improvements are necessary to advance clinical outcomes evidence that can improve TBI patient care.

Data availability statement

Data are available upon request.