Pharmacological thromboprophylaxis to prevent venous thromboembolism in patients with temporary lower limb immobilization after injury: systematic review and network meta‐analysis

Abstract Background Thromboprophylaxis has the potential to reduce venous thromboembolism (VTE) following lower limb immobilization resulting from injury. Objectives We aimed to estimate the effectiveness of thromboprophylaxis, compare different agents, and identify any factors associated with effectiveness. Methods We undertook a systematic review and network meta‐analysis (NMA) of randomized trials reporting VTE or bleeding outcomes that compared thromboprophylactic agents with each other or to no pharmacological prophylaxis, for this indication. An NMA was undertaken for each outcome or agent used, and a series of study‐level network meta‐regressions examined whether population characteristics, type of injury, treatment of injury, or duration of thromboprophylaxis were associated with treatment effect. Results Data from 6857 participants across 13 randomized trials showed that, compared with no treatment, low molecular weight heparin (LMWH) reduced the risk of any VTE (odds ratio [OR]: 0.52; 95% credible interval [CrI]: 0.37‐0.71), clinically detected deep vein thrombosis (DVT) (OR: 0.39; 95% CrI: 0.12‐0.94) and pulmonary embolism (PE) (OR: 0.16; 95% CrI: 0.01‐0.74), whereas fondaparinux reduced the risk of any VTE (OR: 0.13; 95% CrI: 0.05‐0.30) and clinically detected DVT (OR: 0.10; 95% CrI: 0.01‐0.86), with inconclusive results for PE (OR: 0.40; 95% CrI: 0.01‐7.53). Conclusions Thromboprophylaxis with either fondaparinux or LMWH appears to reduce the odds of both asymptomatic and clinically detected VTE in people with temporary lower limb immobilization following an injury. Treatment effects vary by outcome and are not always conclusive. We were unable to identify any treatment effect modifiers other than thromboprophylactic agent used.


| BACKG ROU N D
Venous thromboembolic (VTE) disease is a major global cause of morbidity and mortality. 1,2 An estimated 10 million episodes are diagnosed yearly; more than one-half of these episodes are provoked by hospital admission/procedures and result in significant loss of disability-adjusted life years. 3 Temporary lower limb immobilization after injury is a significant contributor to overall VTE burden. 4 This risk may be modifiable.
Existing evidence suggests that pharmacological prophylaxis could reduce overall VTE event rates in these patients, but the proportional reduction of symptomatic events remains unclear. 5 Recent randomized controlled trials (RCTs) have used different pharmacological agents (low molecular weight heparin [LMWH] and fondaparinux), dosing regimens and outcome measures. [6][7][8][9] In addition, some centers are reporting recent experience with use of the direct oral anticoagulants (DOACs) for this indication, despite the lack of appropriate licensing and trial data within this specific population. 10,11 Consequently, there is wide variation in thromboprophylaxis strategies, and international guidelines continue to offer conflicting advice for clinicians. [12][13][14][15] The overall clinical effectiveness of thromboprophylaxis for this indication and the optimal agent/dosing strategy are yet to be defined.
We undertook a systematic review and network meta-analysis (NMA) to assess the effectiveness of pharmacological thromboprophylaxis at preventing VTE in patients with temporary lower limb immobilization after injury. Our aim was to estimate the clinical effectiveness for each pharmacological thromboprophylaxis option and further compare regimens and agents to identify an optimal strategy.

| ME THODS
The systematic review was undertaken in accordance with the general principles recommended in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 16 This review was part of a larger project on thromboprophylaxis for lower limb immobilization that was registered on the PROSPERO international prospective register of systematic reviews (CRD42017058688). 17 The full protocol is available here (https :// www.journ alsli brary.nihr.ac.uk/progr ammes/ hta/15187 06/#/).

| Eligibility criteria
Studies were considered eligible for inclusion if they met the following criteria: (a) RCTs or controlled clinical trials; (b) adults (age >16 years) requiring temporary immobilization (e.g., leg cast or brace in an ambulatory setting) for an isolated lower limb injury; (c) chemical thromboprophylaxis with any LMWH agent, fondaparinux, or oral anticoagulant (e.g., apixaban, dabigatran etexilate, rivaroxaban, edoxaban); (d) comparators included placebo, no treatment, aspirin, or alternative treatment; and (e) outcomes included symptomatic or asymptomatic deep vein thrombosis, pulmonary embolism (PE), major bleeding (as defined within each study), or mortality. Exclusion criteria for selection included studies that had not been designed as experimental studies (e.g., cohort studies, case control studies); involved hospital inpatient care or any patient requiring hospital admission longer than 3 days; patients receiving mechanical thromboprophylaxis or undergoing ambulant orthopedic surgery (e.g., arthroscopy, arthroscopic surgery).

| Outcome definitions
Given the challenges of outcome reporting in this population, we chose to prospectively define VTE events according to anatomical location and symptomatology. Our aim was to provide full transparency of all potentially relevant outcomes and to highlight the specific data informing assessment of intervention.
We defined proximal deep vein thrombosis (DVT) as thrombosis occurring at or above the level of the popliteal trifurcation. 18 Symptomatic disease was defined as reported within individual trials; any diagnosis of PE was considered to be symptomatic, as were presentations outside routine study follow-up with acute DVT symptoms and subsequent confirmation of disease. However, in several studies patients were questioned on the symptoms of DVT (e.g., pain, swelling) when the cast was removed, at routine follow-up. If the patients reported any positive symptoms and routine sonography had detected DVT, the event was classified as symptomatic. The limitations with this approach are highlighted later in the discussion section. "Any VTE"was defined as the composite of any PE and/or any distal or proximal DVT, with or without symptoms. • This network meta-analysis analyzed 6857 patients randomized to thromboprophylaxis or control.
• Overall, pharmacological prophylaxis significantly reduced the odds of any venous thromboembolism (VTE).
• Individualized treatment may be an optimal strategy and requires further study.
We could not retrospectively apply consensus definitions of symptomatology or major bleeding to individual study results. 19 These issues and their potential impact on study results are explored further in the discussion section.

| Information sources and searches
Ten electronic databases (including MEDLINE, EMBASE, and the Cochrane Library) were searched. The search strategy used free text and thesaurus terms and combined synonyms relating to the condition (e.g., venous thromboembolism in people with lower limb immobilization) with synonyms relating to the interventions (e.g., LMWH, aspirin, oral anticoagulants). No language restrictions were used.
Searches were supplemented by hand-searching the reference lists of all relevant studies (including existing systematic reviews), performing a citation search of relevant articles, contacting key experts in the field, and undertaking systematic keyword searches of the World Wide Web using the Google search engine. Further details on the search strategy can be found in Table S1 (supporting information).

| Study selection
All titles were examined for inclusion by one reviewer (A.P.); any citations that clearly did not meet the inclusion criteria were excluded.
All abstracts and full-text articles were then examined independently by two reviewers (A.P. and D.H.). Any disagreements in the selection process were resolved through discussion or if necessary, arbitration by a third reviewer (S.G.) and included by consensus.

| Data extraction and quality assessment
Data relating to study design, methodological quality, and outcomes were extracted by one reviewer into a standardized data extraction form and independently checked for accuracy by a second.
The methodological quality of each included study was evaluated using a revised Cochrane Risk of Bias tool for randomized trials (RoB 2.0). 20 The original tool 21 was updated because of questionable inter-rater agreement, subjectivity in assigning risk of bias judgments, and bias judgments assigned at the trial level. [22][23][24][25] An overall judgement of bias was assigned as low risk if all domains were judged as low risk of bias; high risk if at least one domain was judged to be at high risk of bias (or if the study has some concerns for multiple domains in a way that substantially lowers confidence in the result), and some concerns if any bias (other than high risk) was noted in at least one domain. 20

| Data synthesis and analysis
For each outcome of interest, an NMA was performed to allow a simultaneous comparison between interventions using all available studies. The data were the number of events out of the number of patients randomized to each class of intervention, which were assumed to arise from an underlying binomial distribution. LMWH agents were collated and considered as a single intervention. The probabilities of an event for each intervention were modelled using a logistic model to estimate odds ratios (ORs). The reference intervention was defined as placebo, no treatment, or aspirin in the NMA.
The different thromboprophylaxis drugs were treated as separate interventions (i.e., LMWH, DOACs, and fondaparinux) in the NMA on the basis of having different mechanisms of action and different adverse event profiles.
The analysis was implemented using Markov chain Monte Carlo simulation using WinBUGS software Version 1.4.3 (MRC Biostatistics Unit). 26 A fixed effect model was used to estimate the effects of LMWH and fondaparinux relative to control in the available studies (i.e., a conditional inference). In addition, a random effects model was used to allow for heterogeneity in the effects of interventions between studies and to estimate whether the interventions can have an effect in future studies. Results were presented using ORs, 95% credible intervals (CrI), and 95% predictive intervals for the OR in a randomly chosen study relative to the control, with the probability of each intervention being the best.
We also evaluated the following potential treatment effect mod-
Studies excluded after full text review are listed in Table S2, along with the rationale for exclusion.
Five studies were open label with subjective screening outcomes (duplex sonography or phlebography on cast removal). [29][30][31]35,36 Six studies used double blinding within the design. 6,8,28,[32][33][34] Although all studies included adult patients with an isolated lower limb injury requiring temporary immobilization, there was wide variation in terms of injury type. Five studies included only patients with fractures, 6,8,9,28,32 one of patients with Achilles tendon ruptures, 33 and the remaining seven studies included patients with mixed pathology. 27,[29][30][31][34][35][36] Depending on the type of injury, the management of lower limb injury included conservative treatment, 9,30,31,35 surgical management, 6,8,28,32,33 or both. 29,34,36 In eight studies, 6,8,9,28,[32][33][34][35] patients were recruited within 4 days of injury, whereas, in the remaining studies, 27,[29][30][31]36 the time to recruitment was not stated. The duration of immobilization ranged from 14 days 28 to 44 days. 32,34 In two studies, all 32 or some (approximately onethird) 34 patients first received prophylaxis before randomization; these studies were included because any final impact on outcome would likely take the form of reduction in VTE outcome events. In addition, the results of these trials remain relevant to the study question in light of current regimes suggesting prophylaxis should continue for the duration of immobilization (usually 4-6 weeks).

| Ri™sk of bias within and across studies
The overall methodological quality of the 13 included studies is summarized in Figure 2 and Table 2. Overall, risk of bias was present in all studies. Ten studies raised some concerns of bias. 6,8,9,28,29,[32][33][34][35][36] The potential sources of bias most frequently identified included concerns about the randomization process

| Effects of interventions
Details of the total participant numbers in each analysis, event rates, and further key outcome results of the individual primary studies are provided in Table 3. All 13 studies reported outcomes for any VTE, PE, and major bleeding. The rate of any VTE in the control group ranged from 1.8% to 40.4%. The rate of PE in the control group was zero in eight studies and ranged from 0.7% to 2.1% in the other four.
There was only one major bleeding event across all control groups.
NMA was undertaken to compare the effectiveness of two alternative forms of thromboprophylaxis (LMWH or fondaparinux) to no thromboprophylaxis (aspirin, placebo, or no treatment). Figure 3 presents the network of evidence. All 13 studies were included in the analysis and provided information on at least one of the outcomes being analyzed. A summary of the results of fixed effect and random effects NMA are provided in Table 4.

| Clinically detected DVT (symptomatic)
Data were available from all 13 studies. 6

| Asymptomatic DVT: proximal segment
Data were available from eight studies. 6

| Asymptomatic DVT: distal
Data were available from eight studies. 6 Overall risk of bias judgment (equal to the most severe level of bias found in any domain) was judged as: 1) Low risk of bias: the study is judged to be at low risk of bias for all domains for this result; 2) Some concerns: the study is judged to have some concerns of bias in at least one domain for this result; 3) High risk of bias: the study is judged to be at high risk of bias in at least one domain for this result or have some concerns for multiple domains in a way that substantially lowers confidence in the result.

TA B L E 3 Summary of outcomes: PE and major bleeding
Author

| Compliance and adverse events
Compliance with study medication appeared generally good within trial participants; eight studies reported >90% compliance, two studies between 80% and 90%, and was unclear in three studies.
A single open-label study 9 recorded reports of pain on injection in 1.4% of participants within the intervention group.
There were few reported adverse events in the treated patients.
Subjective and composite overall adverse event rates ranged from 0% to 4.0% across individual studies with intervention, and 0% to 2.0% in control patients. Minor bleeding event rates varied from 0% to 10.5% in the LMWH intervention groups, 0% to 1.5% in the fondaparinux intervention groups, and 0% to 6.8% in the control groups. In the largest RCT to date, 36  and control groups (1.6% vs 2.0%, respectively). In four studies actively reporting the incidence of heparin-induced thrombocytopenia, no cases were found. 8,29,37,38 No deaths in any study were deemed attributable to either VTE or the use of intervention.

| Additional analyses
The results of the network meta-regressions are detailed in Table S3. The analysis showed that no covariate improved model fits and therefore explained variation in treatment effects.
A sensitivity analysis excluding the three studies at high risk of bias is detailed in Table S4. This analysis generally had negligible impact on the estimates of treatment effect, but as expected, tended to increase uncertainty.  caution, based on the heterogeneity between studies and the low event rates.

| Summary of evidence
Our NMA shows that thromboprophylaxis with LMWH for patients with lower limb immobilization after injury approximately halves the odds of any VTE in these studies. Thromboprophylaxis with fondaparinux appears to have a greater effect on reducing the risk of DVT and is likely to be more effective than LMWH. Event rates for symptomatic DVT and PE in untreated patients were generally low across the studies, so an approximate halving of odds may result in a small absolute risk reduction.
Major bleeding is very uncommon, so the effect of thromboprophylaxis on major bleeding in this group is uncertain. Meta-regression did not identify any reliable evidence of effect modification by key covariates.

| Strengths and limitations
Our NMA synthesized data from 6857 participants in 13 randomized trials. This represents a large, methodologically robust data set across multiple settings used to simultaneously estimate of relative treatment effects.
Our analysis was inevitably limited by the primary data. The variety of settings and patient groups may be a strength, but generated evidence of heterogeneity of treatment effect across studies for many of the outcomes. Previous work has shown evidence of publication bias such that studies with nonsignificant or unfavorable results on this topic are perhaps less likely to be published, have a delay to publication, or involve selectively reporting outcomes. 39 These issues have the potential to exaggerate any benefit to the intervention seen at NMA.
The studies were judged mainly to have low risk of bias or some concerns only for most quality criteria. However, three trials were judged as having a high risk of bias on the basis of outcome ascertainment being potentially subject to bias in an open-label trial. 27,30,31 This is particularly relevant to the issue of symptomatic VTE as an outcome. Several of these open-label trials performed routine sonographic screening on removal of plaster cast, followed by clinical assessment. This methodology introduces a high risk of bias with symptomatic VTE outcomes; patients may have been influenced by the sonographer, or party to the ultrasound results before disclosing information on symptomatology. A sensitivity analysis excluding these studies generally had negligible impact on the estimates of treatment effect but, as expected, tended to increase uncertainty.
This analysis does not take into account that several of the clinical events were likely not representative for events that would lead a patient to actively seek medical assistance (i.e., truly symptomatic events). This is reflected in the highly varying risks between studies found in Table 3. A further breakdown of symptomatic VTE outcomes is provided in Table S5 for transparency.
We included one trial 27 47 We present our outcomes in this study stratified by symptomatic disease and anatomical location of VTE to address these issues.
The analysis included a substantial number of participants but the number of events for some outcomes were very low, or zero, including zero events in two arms of a study. As a consequence, not all studies provide estimates of relative treatment effect for all outcomes. We were unable to produce precise estimates of the effect of thromboprophylaxis upon major bleeding or estimate the effect of treatment on death. The low rate of bleeding provides some reassurance that thromboprophylaxis is not causing a clinically important rate of serious adverse outcome in this population but this may not be applicable to patients with a higher risk of bleeding. Other surrogate datasets, such as patients receiving thromboprophylaxis for knee arthroscopy can provide further relevant information on bleeding risk. 48 This information could be used to support shared decision-making. However, the population undergoing elective arthroscopy has key differences to our population of interest, including acute exposure to surgical bleeding risk, hospitalization, and the absence of blunt forced injury. As such, extrapolation of bleeding risk to conservatively managed ambulatory patients has significant caveats.  27 ) and one because the intervention was fondaparinux rather than LMWH (Samama et al.). 35 The inclusion of these trials has ensured that our analysis is more comprehensive, but possibly at the expense of greater heterogeneity.

| Comparison to previous research
In addition, our study is also the first to perform network me- Within the meta-regression analyses, we were unable to identify any evidence to directly support stratified thromboprophylaxis. We found no association between treatment effect and patient characteristics, type of injury, treatment method, or duration of prophylaxis.
Several authors have recently suggested that selection of patients for thromboprophylaxis may be appropriate on the basis of an increased baseline risk 39 ; it does not appear from our analysis adjusting for baseline risk that prophylaxis should be offered based on an expectation of greater effectiveness in any specific group. We did not assess risk factors at a patient level within this work and so cannot draw any conclusions on the merits of risk adjusted thromboprophylaxis.

| The direction of future research
Although our findings suggest that thromboprophylaxis could reduce the rate of symptomatic events, further study of cost effectiveness is needed given the low absolute risk. In addition, stratified thromboprophylaxis may be able to select out patients at highest risk and maximize potential clinical and cost effectiveness. Several risk assessment models (RAM) have already been derived for use in this patient population, aiming to tailor thromboprophylaxis strategies at presumed high risk and limit financial costs, opportunity costs, and side effects. 12,52,53 The current evidence base for RAMs is very limited, and estimates of sensitivity and specificity are subject to substantial uncertainty. 17 Improving the evidence base for RAMs is a key research priority and external validation studies are urgently needed.
In addition to dichotomous RAMs, individualized treatment could potentially be optimized by further adaptation in very high-risk groups deemed to warrant thromboprophylaxis (e.g., higher dosing, longer duration). This latter question is yet to be addressed in the literature.
Oral medications could provide the benefits of thromboprophylaxis without the costs, inconvenience, and discomfort of injections.
However, evidence of effectiveness in our review was related only to LMWH or fondaparinux. If further research identifies a high-risk population that unequivocally benefit from thromboprophylaxis, the use of direct oral anticoagulants could potentially be compared with LMWH to assess differences in cost, clinical outcome, and patient satisfaction.
It is currently unclear whether people with limited lower limb immobilization (such as splints that allow some movement or removable splints or casts) carry similar risks of VTE to those with full immobilization. A study of this population could determine the risk of VTE and potentially identify patient-level risk predictors.

| CON CLUS IONS
Thromboprophylaxis for patients with lower limb immobilization after injury appears to be clinically effective, reducing the odds of symptomatic VTE. Given the low absolute risk of VTE in a broad population, individualized risk assessment and shared decision making may be optimal. This strategy requires further supporting research.

ACK N OWLED G M ENTS
The authors thank all additional members of the core project group for HTA 15/187/06 for input and commentary throughout the work.
The authors also thank Mark Clowes for his assistance with the initial literature searches and article sourcing and to Joanne Hinde for assistance with logistics and administration.

CO N FLI C T O F I NTE R E S T S
On behalf of all authors, I declare no known real or potential conflicts of interest to exist regarding this research article.