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

Data sources and search

The following databases were searched in July 2012 for primary studies: MEDLINE^®, EMBASE^®, SCOPUS, CINAHL, International Pharmaceutical Abstracts, clinicaltrial.gov, and the Cochrane Library. An analytic framework depicting our population of interest, interventions tested for prevention of VTE, intermediate and patient-oriented outcomes of treatment, as well as the harms of the interventions was developed³.

Study selection

Titles were reviewed followed by abstracts to identify randomized controlled trials (RCTs) or observational studies with comparison groups reporting on the effectiveness or safety of VTE prevention in TBI patients. Two investigators independently reviewed abstracts meeting our inclusion criteria; abstracts were excluded if both reviewers agreed that the article met one or more of the exclusion criteria (Table 1).

Data abstraction and quality assessment

Evidence Partners 2010 web-based database management program, DistillerSR, was used to manage the screening and review process. Standardized forms for data extraction from the articles were created. Paired investigators reviewed all extracted data.

The risk of bias was assessed independently and in duplicate, using the Downs and Black instrument⁶. Ten items that were most relevant to this review were prioritized in our assessment of risk of bias. Studies were assessed to have a low risk of bias if all of the following were true: the article completely described the hypothesis, the outcomes (in the introduction or methods section), the characteristics of the included subjects, the distribution of the potential confounders in each group, the interventions and comparisons (if relevant) the main findings, adverse events, and characteristics of the subjects lost to follow up. Additionally, we judged studies to be at low risk of bias if they randomized subjects to the intervention and concealed the assignment until randomization was complete, and if they attempted to blind the study participants and to blind those who measured the main outcomes. By this system, non-randomized studies could only be at moderate or high risk of bias. Studies were rated as having a moderate risk of bias if one of those items was not true, even if all of the others were true, or if the reporting on the distribution of potential confounders in each group was at least partially done. If two of the elements were not true, studies were rated to have a high risk of bias.

Data synthesis and analysis

A detailed set of evidence tables was created containing all information abstracted from eligible studies. Given the substantial statistical and clinical heterogeneity, we do not report pooled results but display the individual magnitude of effect and statistical significance for the individual studies.

Outcomes assessed

The effectiveness of pharmacological and mechanical strategies in preventing patient-oriented outcomes such as VTE, deep vein thrombosis (DVT) and pulmonary embolism (PE), mortality and progression of intracranial hemorrhage.

Grading the evidence and applicability

The quantity, quality, and consistency of the best available evidence was graded by adapting an evidence-grading scheme recommended in the Agency for Healthcare Research and Quality: Methods Guide for Conducting Comparative Effectiveness Reviews⁷.

Study characteristics

Seven studies that evaluated the effectiveness of pharmacological and mechanical strategies to prevent VTE in hospitalized patients with TBI were identified^8–14, four that evaluated the optimal timing of initiation of pharmacological prophylaxis^1,15–17 and one study that evaluated both¹⁸. Most of the studies were conducted in North America^{1,8,9,11–18}. Two RCTs were included in this review^10,14. The remaining were cohort studies; nine retrospective studies^{1,8,9,11,13,15–18} and one prospective¹². The majority of studies included patients admitted in level 1 trauma centers.

Participant characteristics

The number of participants in the included studies ranged from 32 to 812; the mean age of the participants ranged from 36 to 47 years. The Injury Severity Score (ISS) of TBI patients was reported in eight studies; the mean ranged from 15.7 to 33.8 indicating severe multi-system trauma^{8–12,14,15,18}. The ethnicity or race of the participants was not reported in any study (Table 2).

Intervention characteristics

Eight studies were included to assess the effectiveness and safety of pharmacological and mechanical interventions to prevent VTE in patients with traumatic brain injury^8–14,18. The interventions compared in these studies were highly heterogeneous; studies varied in drugs compared, the dosages and timing of initiation of therapy. Many studies had a control group in which active therapy was withheld from participants. The dose of pharmacological drugs used was reported in five studies; dalteparin was administered as 5000 U once daily. Unfractionated heparin (UFH) as 5000 U thrice daily, and enoxaparin as 30 mg twice daily or 40 mg daily.

Five studies independently assessed the optimal timing of the initiation of chemoprophylaxis in the same population^1,15–18. Although enoxaparin and UFH were the only pharmacological agents employed in these studies, two studies were unclear about the pharmacological agents used and were classified as “any heparin” intervention^16,17. Four out of five studies compared the effectiveness and safety of pharmacoprophylaxis in preventing VTE when initiated less than 72 hours (early prophylaxis) of hospital admission versus greater than 72 hours (late prophylaxis).

There were no studies that assessed the effectiveness of inferior vena cava filters in preventing PE in TBI patients.

Ascertainment of VTE

Most studies did not routinely screen for VTE^{1,8–10,13,14,16,18}. Weekly surveillance using duplex ultrasound examination was carried out in four studies^11,12,15,17, although two of these studies performed it in high risk patients exclusively^11,17.

Outcomes

Venous thromboembolism. Five studies assessed the effectiveness of pharmacological agents in preventing VTE in patients with TBI^8,11–13,18. One study demonstrated lower rates of VTE in the UFH group compared to the control group (3% vs. 1%, respectively, p=0.019)¹¹; two studies showed increased rates of VTE in the enoxaparin and sequential compression devices group compared to the control group (enoxaparin vs. control, 3.9% vs. 2.2%, p=0.29; Sequential compression devices (SCD) vs. control, 28.6% vs. 22.2%, p=0.7)^12,18, while the last study demonstrated no difference in rates of VTE between dalteparin and control groups (0% vs. 0%)¹³. Head-to-head comparison available in a study demonstrated marginally increased rates of venous thromboses in patients treated with dalteparin compared to those treated with enoxaparin (7.5% vs. 7.0%, p value not significant)⁸.

A single study demonstrated increased rates of VTE with early enoxaparin prophylaxis when compared to late prophylaxis. (5.56% vs. 2.72% percent, Odds ratio (OR) 2.10, p=0.26)¹⁸ (Table 3).

Overall, the evidence was concluded to be insufficient to comment on the effectiveness and optimal timing of initiation of VTE prophylaxes in TBI patients (Table 4).

Deep vein thrombosis. Four studies were included to evaluate the efficacy of enoxaparin, UFH and sequential compression devices in preventing the development of DVT in patients with TBI^9,10,12,14. A single study demonstrated reduced rates of DVT in enoxaparin and UFH heparin groups compared to control (1% vs. 1% vs. 2% respectively, p value not reported)⁹. Two more studies demonstrated lower rates of DVT in patients treated with enoxaparin compared to those treated with placebo and sequential compression devices (0% vs. 3.6%, p=0.45 and 5% vs. 6.6%, p=0.07)^10,14. In contrast to this, a fourth study demonstrated that patients treated with sequential compression devices experienced fewer events when compared to a control group (0% vs. 11.1%)¹².

In two “any heparin” studies, the rates of DVT were consistently higher in the late prophylaxis group^16,17. The same was observed in patients treated with UFH; rates of DVT were higher when UFH was commenced later than 72 hours (4.3% vs. 5.9%, p value not significant)¹⁵ (Table 3).

Three individual studies demonstrated that rates of DVT were lower in patients treated with enoxaparin when compared to controls or patients treated with sequential compression devices only^9,10,14. Consistent, direct, yet imprecise results, which included one RCT with a low risk of bias, led to the conclusion that low-grade evidence supported the effectiveness of enoxaparin over control/sequential compression devices in reducing DVT in hospitalized patients with TBI. However, the evidence is insufficient to comment on the optimal timing of initiation of chemoprophylaxis in the same population (Table 4).

Pulmonary embolism. Five out of the eight included studies assessed the effectiveness of prophylaxis with enoxaparin, dalteparin, UFH and sequential compression devices in preventing development of PE in patients hospitalized with TBI. The results of these studies were equivocal. One study demonstrated that patients treated with enoxaparin failed to develop PE, whilst those in the control and UFH intervention groups did, the rate being lower in the control group⁹. In contrast, a RCT demonstrated that there was no difference in rates of PE in enoxaparin-treated patients and controls (0% vs. 0%)¹⁴. Two studies showed varying outcomes in patients treated with sequential compression devices only; a RCT demonstrated lower rates of PE, all of which were fatal, in this group compared to treatment with enoxaparin (3.3% vs. 6.6%, p=0.04)¹⁰. However, in another study, the patients in the sequential compression devices intervention group were reported to have experienced an increase in pulmonary embolic events in comparison to control patients (28.6% vs. 11.1%, p value not reported)¹². The last study reported the rate of development of PE in patients treated with dalteparin only, limiting an assessment of comparative effectiveness⁸.

Optimal timing of initiation of chemoprophylaxis in TBI populations to prevent development of PE was analyzed in three studies. Two studies demonstrated increased incidence of PE with early prophylaxis (3.5% vs. 0% and 4.3% vs. 0%), whereas in the third study, patients treated with enoxaparin within 72 hours of admission experienced fewer pulmonary embolic events (1.5% vs. 2.2%, respectively, p=0.49)^1,17,18 (Table 3).

The evidence was concluded to be insufficient to comment on the effectiveness and optimal timing of initiation of prophylaxes in preventing PE in TBI patients (Table 4).

Total mortality. Three studies included in this review evaluated the efficacy of prophylaxis with UFH or enoxaparin versus no prophylaxis or treatment with sequential compression devices only. Two studies uniformly demonstrated increased mortality in control groups when compared to patients treated with enoxaparin and UFH^9,10. However, the third study demonstrated that rates of mortality were increased in patients treated with enoxaparin when compared to those prescribed sequential compression devices only (13.3% vs. 11.6%, p=0.08)¹⁰.

A single cohort study reported increased deaths with early UFH prophylaxis when compared to late prophylaxis (8.5% vs. 5.9%, p=1.0)¹⁵ (Table 3).

Low grade evidence supported the effectiveness of UFH over no pharmacoprophylaxis in reducing total mortality in patients hospitalized with traumatic brain injury (Table 4).

Progression of intracranial hemorrhage. The rates of progression of intracranial hemorrhage resulting from prophylaxis with dalteparin, enoxaparin, or UFH were reported in six studies^8–11,13,14. Two studies reported that there was no difference in rates of progression of intracranial hemorrhage between the control or sequential compression devices only group and the pharmacoprophylaxis (enoxaparin and dalteparin) group^10,13. Another set of two studies that compared prophylaxis with UFH and enoxaparin to control or placebo demonstrated equivocal results^11,14; patients treated with UFH had lower rates of progression of intracranial hemorrhage, while those treated with enoxaparin had higher rates. Two other studies demonstrated head-to-head comparisons of two pharmacological agents. According to one study, patients treated with enoxaparin and dalteparin had comparable rates of intracranial bleeding (0.001% vs. 0%)⁸, while the other demonstrated a statistically significant increase in intracranial bleed in patients treated with UFH compared to those treated with enoxaparin (12% vs. 5%, p<0.05)⁹.

Three studies evaluating the optimal timing of initiation of pharmacoprophylaxis reported on rates of progression of intracranial hemorrhage in TBI populations^1,15,18. Even though all three studies reported increased rates of intracranial hemorrhage when prophylaxis was initiated with enoxaparin or any other heparin after 72 hours of admission, the increase was only minimal in two studies (3.5% vs. 3.8%; 1.46% vs. 1.54%) (Table 3).

Overall, the evidence was insufficient to comment on the effect of pharmacological and mechanical prophylaxis and timing of initiation of pharmacoprophylaxis on progression of intracranial bleeding in TBI patients (Table 4).

Risk of bias

Of the twelve studies included in this review, only one RCT was at a low risk of bias¹⁴. With the exception of a single cohort study that was at a moderate risk of bias⁸, ten were estimated to be high risk of bias studies. Most cohort studies had incomplete descriptions of the important confounders and a lack of adjustment for differences between groups. They also had incomplete accounts of losses to follow-up. All of these are important confounders and threaten the internal validity of these studies.

Applicability

The participants that these studies recruited were typical of participants admitted to other trauma centers and hence findings are generalizable. The studies were generally representative of patients with TBI in the USA. Gender was inconsistently reported, thus we could not assess the applicability of these findings to females. We did not have details to assess the applicability of this evidence to other racial groups since the studies inconsistently reported on ethnicity or race. Some studies excluded patients with previous VTE^1,10 as well as those at higher risk of bleeding, such as those with low platelet counts^1,10,14,15, limiting generalizability to these high-risk subgroups.

Limitations

Our systematic review identified important weaknesses in the literature. We did not identify high quality RCTs for this review. The majority of observational studies included in this review were at a high risk of bias and did not report on several quality items of interest. The studies were heterogeneous in the definition of VTE and bleeding outcomes precluding any meaningful pooling in a meta-analysis. We also did not find data on several pharmacologic comparisons of interest or details about optimal timing of initiation of prophylaxis in this population. We were unable to assess the possibility of publication bias or selective outcomes reporting and its impact on our findings.

Future research

Studies among patients with TBI are needed to determine whether pharmacologic DVT prophylaxis should be employed in these patients and the timing of administration. Studies should also determine the role of appropriate classification and severity of TBI when deciding to administer pharmacologic prophylaxis. Our report shows that confounding by indication was a major problem in these studies. Patients at high risk for thrombotic outcomes were more likely to receive prophylaxis and more likely to have events-the treated and untreated patients were not comparable. Future studies should consider the use of appropriate analytic strategies such as instrumental variables that control for unobserved variables if an appropriate instrument can be identified for analysis. High-quality observational studies that control for confounding by indication, such as provider and practice patterns, and confounding by disease severity may be needed as RCTs typically exclude or do not report on these populations.