Different strategies for pharmacological thromboprophylaxis for lower-limb immobilisation after injury: systematic review and economic evaluation

Identification of studies

Studies were identified by searching the following electronic databases and research registers:

• Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily, MEDLINE and Versions(R) (OvidSP), 1946 to April 2017.

• EMBASE (OvidSP), 1974 to April 2017.

• Cochrane Database of Systematic Reviews (Wiley Online Library), 1996 to April 2017.

• Database of Abstracts of Review of Effects (Wiley Online Library), 1995 to March 2015.

• Cochrane Central Register of Controlled Trials (Wiley Online Library), 1898 to April 2017.

• Health Technology Assessment (HTA) database (Wiley Online Library), 1995 to April 2017.

• NHS Economic Evaluation Database NHS EED (Wiley Online Library), 1995 to March 2015.

• Science Citation Index Expanded (Web of Science), 1900 to April 2017.

• ClinicalTrials.gov (US National Institutes of Health), 2000 to April 2017.

• International Clinical Trials Registry Platform (World Health Organization), 1990 to April 2017.

The search strategy used free text and thesaurus terms and combined synonyms relating to the condition (i.e. VTE in people with lower-limb immobilisation) with synonyms relating to the interventions (e.g. LMWH, aspirin and oral anticoagulants). No language restrictions were used. However, as the search strategy of the current review updated the search strategy of an existing review on LMWH,15 searches were limited by date from 2013 (the last search date from the earlier review) to April 2017 for this intervention. For the other interventions, the search strategy was amended to include terms for aspirin and oral anticoagulants and searched from inception to April 2017. Further details of the search strategy can be found in Appendix 1. 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 internet using the Google search engine (Google Inc., Mountain View, CA, USA).

All identified citations from the electronic searches and other resources were imported into and managed using EndNote bibliographic software version X8 [Clarivate Analytics (formerly Thomson Reuters), Philadelphia, PA, USA].

Inclusion and exclusion criteria

The inclusion of potentially relevant articles was undertaken using a two-step process. First, all titles were examined for inclusion by one reviewer (AP) and any citations that clearly did not meet the inclusion criteria (e.g. non-human, unrelated to VTE) were excluded. Second, all abstracts and full-text articles were examined independently by two reviewers (AP and DH). When necessary, non-English-language studies were translated using Google Translate (Google Inc., Mountain View, CA, USA) to facilitate study selection and subsequent data extraction. Any disagreements in the selection process were resolved through discussion or, if necessary, arbitration by a third reviewer (SG), and articles were included by consensus.

Studies were considered eligible for inclusion if they met the following criteria: (1) study design – RCTs and controlled clinical trials; (2) population – adults (aged > 16 years) requiring temporary immobilisation (e.g. leg cast or brace in an ambulatory setting) for an isolated lower-limb injury; (3) interventions – chemical thromboprophylaxis with LMWH (e.g. dalteparin, enoxaparin, tinzaparin), fondaparinux or oral anticoagulants (e.g. apixaban, dabigatran etexilate, rivaroxaban); (4) comparators – these included placebo, no treatment, aspirin or alternative treatment (although the original protocol considered aspirin to be an option for VTE prophylaxis, NICE guidelines on venous thromboembolism (CG92)24 do not consider aspirin or other antiplatelet agents to be appropriate for VTE prophylaxis; in addition, aspirin is not indicated as a treatment for VTE prophylaxis in lower-limb immobilisation);21,25,26 and (5) outcomes – these included symptomatic or asymptomatic DVT, PE, major bleeding or mortality. Exclusion criteria for selection included studies that had not been designed as experimental studies (e.g. cohort studies and case–control studies), studies that had involved hospital inpatient care or any patient requiring hospital admission of > 5 days and studies in which patients received mechanical thromboprophylaxis or underwent ambulant orthopaedic surgery (e.g. arthroscopy, arthroscopic surgery).

Data abstraction and quality assessment strategy

Data relating to study design, methodological quality and outcomes were extracted by one reviewer (AP) into a standardised data extraction form and independently checked for accuracy by a second reviewer (DH). Any discrepancies were resolved through discussion or, if necessary, arbitration by a third reviewer (SG), and articles were included by consensus. If required, authors of primary studies were contacted to obtain additional data, clarify uncertainties and/or confirm data that had been extracted. When multiple publications of the same study were identified, data were extracted and reported as a single study.

The methodological quality of each included study was evaluated using a revised Cochrane Risk of Bias tool for randomised trials (RoB 2.0). 39 The original tool40 was updated because of questionable inter-rater agreement, subjectivity in assigning risk-of-bias judgements and bias judgements assigned at the trial level. 41–44 In general, RoB 2.0 redefined the potential for bias to five domains: (1) bias arising from the randomisation process, (2) bias as a result of deviations from intended interventions, (3) bias as a result of missing outcome data, (4) bias in the measurement of the outcome and (5) bias in the selection of the reported result. To limit subjectivity in assigning bias judgements, the RoB 2.0 tool provides detailed guidance and contains decision algorithms. An overall judgement of bias was assigned as ‘low risk’ if all domains were judged as being at a low risk of bias, a judgement of bias was assigned as ‘high risk’ if at least one domain was judged to be at a high risk of bias (or if the study had some concerns for multiple domains in a way that substantially lowers confidence in the result) and as ‘some concerns’ if some concerns of bias were noted in at least one domain. 39 The methodological quality of each included study was independently evaluated by two reviewers (AP and DH). Any discrepancies were resolved through discussion or, if necessary, the involvement of a third reviewer (SG). Blinding of the quality assessor to author, institution or journal was not considered necessary.

Methods of data synthesis and analysis

The extracted data and quality assessment variables were presented for each included study, both in structured tables and as a narrative description. For each outcome of interest, a network meta-analysis (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 randomised to each intervention, which were assumed to arise from an underlying binomial distribution. The probabilities of an event for each intervention were modelled using a logistic model to estimate odds ratios (ORs). The control intervention was defined as placebo, no treatment or aspirin, and the reference intervention defined in the NMA was the control intervention. Aspirin was grouped with placebo and no treatment on the basis that aspirin is not indicated as a treatment for VTE prophylaxis in lower-limb immobilisation21,25 and NICE guidelines on VTE (CG9224 and NG891) do not consider aspirin or other antiplatelet agents to be appropriate for VTE prophylaxis. It was planned to analyse different types of thromboprophylaxis drugs as separate interventions (i.e. LMWH, DOACs and fondaparinux) in the NMA on the basis of having different mechanisms of action and, therefore, potentially different effects.

The analysis was implemented using a Markov chain Monte Carlo simulation using WinBUGS software version 1.4.3 (MRC Biostatistics Unit, Cambridge, UK). 45 A fixed-effect model was used to estimate the effects of LMWH and fondaparinux relative to control in the available studies, namely 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 or not the interventions can have an effect in future studies. The random-effects model was the primary analysis. The baseline log odds in each study were given normally distributed prior distributions with mean 0 and variance 1000, namely N(0, 1000). The log-odds ratios for LMWH and fondaparinux versus control were given normally distributed prior distributions with a mean of 0 and variance of 1000, namely N(0, 1000). The between-study standard deviation (SD) was given a half-normal prior distribution with a mean of 0 and precision of 1.82, namely HN(0, 0.5495); this prior distribution was chosen to have, a priori, 95% of the study-specific odds ratios lie within a factor of 5 from the median odds ratio for each comparison. Convergence of the Markov chains to their stationary distributions was assessed using the Gelman–Rubin convergence statistic. 46 For all outcomes other than major bleeding, convergence occurred within 30,000 iterations of the Markov chain and within 100,000 samples for major bleeding; a burn-in of 100,000 iterations was used in all analyses. There was some evidence of high autocorrelation between successive iterations of the Markov chain; parameters were estimated after retaining every 10th sample of the Markov chain to limit the number of unnecessary runs of the decision model that are informed by the results of the NMA. Results were presented using ORs, 95% credible intervals (CrIs) and the 95% predictive intervals47 for the OR in a randomly chosen study relative to the control, and the probability of each intervention being the best. 48

It was planned to assess the following potential treatment effect modifiers in a series of meta-regressions: (1) population characteristics (e.g. proportion who were male, baseline risk of VTE), (2) type of injury (i.e. fractures, Achilles tendon rupture, other soft-tissue injury), (3) treatment of injury (surgical vs. conservative, above- vs. below-knee immobilisation), (4) thromboprophylactic agent used and (5) duration of thromboprophylaxis.

Quantity and quality of research available

The literature searches identified 1105 citations. Of these, 13 studies (all RCTs) met the inclusion criteria. 23,49–60 A flow chart describing the process of identifying relevant literature can be found in Figure 1. A total of 23 full-text articles were excluded as they did not meet all of the prespecified inclusion criteria. The majority of the articles were excluded primarily on the basis of inappropriate study design (i.e. non-randomised controlled trial or controlled clinical trial), wrong target population (i.e. not isolated lower-limb injury requiring temporary immobilisation) or unsuitable publication type (i.e. reviews, commentaries, editorials or multiple publications of the same study). A full list of excluded studies with reasons for exclusion is presented in Appendix 2.

FIGURE 1.

Study flow chart (adapted):37 review of pharmacological thromboprophylaxis for preventing VTE. © 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

© 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Description of included studies (design and participant characteristics)

The design and participant characteristics of the 13 included studies23,49–60 that evaluated the effectiveness of pharmacological thromboprophylaxis for preventing VTE in ambulatory trauma patients with temporary lower-limb immobilisation are summarised in Table 1.

All studies were published between 1993 and 2017. In total, 6857 patients were included and randomised across 10 countries (i.e. Canada,50,58 China,60 Denmark,51,56 France,57 Germany,23,52,53,57 Italy,57 the Netherlands,49,57,59 Russia,57 Spain57 and Sweden54,55) to receive either intervention or control treatment. LMWH injections were the primary intervention, using variable agents (i.e. certoparin,52 dalteparin,50,54,55,58 nadroparin,49,53,57,59 reviparin23,56 and tinzaparin51) and dosing regimes (e.g. administered once daily without dose adjustment for bodyweight), but two studies used fondaparinux. 49,57 One study used aspirin as a control group treatment,23 with others using placebo injections or nothing dependent on design. 50–56,58–60 In general, most studies excluded patients at highest risk of VTE, namely those with active cancer,49,54,55,58,60 previous VTE49,50,52–56,58–60 or first-degree family history of VTE. 59,60

Five identified studies used open-label methodology with subjective screening outcomes (duplex sonography or phlebography on cast removal). 51–53,57,59 Six studies used double blinding within the design. 50,54–56,58,60 The single largest study had symptomatic VTE only as an identified primary outcome, confirmed with imaging. 59 Although all studies included adult patients with an isolated lower-limb injury requiring temporary immobilisation, there was wide variation in terms of injury type. Five studies focused on patients with fractures,49,50,54,58,60 one focused on Achilles tendon ruptures55 and the remaining seven studies included patients with mixed pathology. 23,51–53,56,57,59 Depending on the type of injury, the management of lower-limb injury included conservative treatment,49,52,53,57 surgical management50,54,55,58,60 or both. 51,56,59 In eight studies,49,50,54–58,60 patients were recruited within 4 days of injury, whereas, in the remaining studies,23,51–53,59 the time to recruitment was not stated. The duration of immobilisation ranged from 14 days50 to 44 days. 54,56 In two studies, all54 or some (approximately one-third)56 patients first received prophylaxis prior to randomisation; these studies were included, as any final impact on outcome would be likely to take the form of a reduction in VTE outcome events. In addition, the results of these trials remain relevant to the study question in the light of current regimes, suggesting that prophylaxis continue for the duration of immobilisation (usually 4–6 weeks). The sample sizes of the included studies ranged from 10555 to 151959 patients, with the mean age of participants ranging from 34 years52,53 to 49 years. 58 The number of male participants ranged from 42%58 to 79%. 55

Quality characteristics

The overall methodological quality of the 13 included studies is summarised in Figure 2 and Table 2. Overall, risk of bias was present in all studies. Ten studies raised some concerns of bias. 49–51,54–60 The potential sources of bias most frequently identified included concerns about the randomisation process (allocation concealment was not reported in nine studies),23,50–56,60 blinding (open-label design)23,49,51–53,57,59 and analyses intentions (only one study provided sufficient information on the selection of the reported result). 59 A high risk of bias was noted in three studies. 23,52,53 High risk of bias was principally attributable to outcome assessment; in three open-label studies,23,52,53 outcome assessment was performed on all patients with routine screening compression ultrasonography and phlebography for confirmation of positive findings. Finally, all of the included studies were conducted outside the UK, making generalisability of the findings to the UK setting uncertain.

FIGURE 2.

Risk-of-bias assessment graph:39 review authors’ judgements about each methodological quality item across all included studies – review of pharmacological thromboprophylaxis for preventing VTE.

Quantitative data synthesis

Details of the results of the primary studies are provided in Appendix 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% with a median of 12.2%. 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 bleeding event across all the control groups. Clinically relevant (proximal or symptomatic) DVTs were reported in 10 out of 13 studies, with control event rates ranging from 0.0% to 6.4% (median 1.5%). Clinically detected (symptomatic) DVTs were reported in all 13 studies, with control event rates ranging from 0.0% to 5.5% (median 0.7%). Any proximal or distal asymptomatic DVTs were reported in 10 out of 13 studies, with control event rates ranging from 1.6% to 25.7% (median 6.9%). Asymptomatic proximal DVTs were reported in 8 out of 13 studies, with control event rates ranging from 0.0% to 6.4% (median 0.7%). Asymptomatic distal DVTs were reported in 8 out of 13 studies, with control event rates ranging from 0.8% to 16.0% (median 3.0%).

A NMA was undertaken to compare the effectiveness of two alternative forms of thromboprophylaxis (i.e. LMWH or fondaparinux) with no thromboprophylaxis (i.e. 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 analysed. Eleven of the studies compared LMWH thromboprophylaxis with no thromboprophylaxis, one three-arm study compared LMWH with fondaparinux with no thromboprophylaxis, and one study compared LMWH with fondaparinux. A summary of the key results of fixed-effect and random-effects NMA are provided in Table 3.

FIGURE 3.

Network diagram of different pharmacological thromboprophylaxis interventions vs. no thromboprophylaxis for preventing VTE. The nodes are the interventions. The numbers against each outcome represent the number of times that each pair of interventions has been compared. There was one multiarm study comparing LMWH vs. fondaparinux vs. control. Diagrams for specific outcomes depend on the number of studies that provide data and the number of non-zero event studies; not all outcomes involve feedback loops.

Summary of key findings

• Thromboprophylaxis with LMWH approximately halves the risk of any VTE. The effects on different types of VTE are variable and uncertain (in accordance with random error), but all are consistent with a halving of risk.

• Thromboprophylaxis with fondaparinux appears to have a greater effect on the risk of VTE and a greater probability than LMWH of being the more clinically effective, but estimates for fondaparinux are based on only two trials.

• Major bleeding is very uncommon; therefore, 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.

Identification of studies

Studies were identified by searching the following electronic databases and research registers:

• Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily, MEDLINE and Versions(R) (via OvidSP), 1946 to May 2017.

• EMBASE (via OvidSP), 1974 to May 2017.

• Cochrane Database of Systematic Reviews (via Wiley Online Library), 1996 to May 2017.

• Database of Abstracts of Review of Effects (via Wiley Online Library), 1995 to March 2015.

• Cochrane Central Register of Controlled Trials (via Wiley Online Library), 1898 to May 2017.

• HTA database (via Wiley Online Library), 1995 to May 2017.

• NHS EED (via Wiley Online Library), 1995 to March 2015.

• Science Citation Index Expanded (via Web of Science), 1900 to May 2017.

• ClinicalTrials.gov (via US National Institutes of Health), 2000 to May 2017.

• International Clinical Trials Registry Platform (via World Health Organization), 1990 to May 2017.

The search strategy used free-text and thesaurus terms and combined synonyms relating to the condition (i.e. VTE in people with lower-limb immobilisation) with risk factor assessment or risk prediction modelling terms (used in the searches of MEDLINE, The Cochrane Library and EMBASE only). No language or date restrictions were used on any database. Further details on the search strategy can be found in Appendix 5. 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 internet using the Google search engine.

All identified citations from the electronic searches and other resources were imported into and managed using EndNote bibliographic software.

Inclusion and exclusion criteria

The inclusion of potentially relevant articles was undertaken using a two-step process. First, all titles were examined for inclusion by one reviewer (AP) and any citations that clearly did not meet the inclusion criteria (e.g. non-human, unrelated to VTE) were excluded. Second, all abstracts and full-text articles were then examined independently by two reviewers (AP and DH). When necessary, non-English-language studies were translated using Google Translate to facilitate study selection and subsequent data extraction. Any disagreements in the selection process were resolved through discussion or, if necessary, arbitration by a third reviewer (SG), and articles were included by consensus.

Studies were considered eligible for inclusion if they met the following criteria: (1) any study design that included a measurement of VTE patient outcome, (2) a study population of adults (aged > 16 years) requiring temporary immobilisation (e.g. leg cast or brace in an ambulatory setting) for an isolated lower-limb injury, (3) any studies that reported and analysed data on individual risk factors associated with DVT or PE. Exclusion criteria for the selection included studies that involved hospital inpatient care or any patient requiring hospital admission for > 5 days, or studies that involved patients undergoing ambulant orthopaedic surgery (e.g. arthroscopy, arthroscopic surgery).

Data abstraction and quality assessment strategy

Data relating to study design, methodological quality and outcomes were extracted by one reviewer (AP) into a standardised data extraction form and independently checked for accuracy by a second reviewer (DH). Any discrepancies were resolved through discussion to achieve agreement. When differences were unresolved, the opinion of a third reviewer (SG) was sought. When multiple publications of the same study were identified, data were extracted and reported as a single study.

The methodological quality of each included study was assessed using the Risk Of Bias In Non-randomized Studies – of Interventions tool (ROBINS-I) [formerly called A Cochrane Risk of Bias Assessment Tool for Non-Randomized Studies of Interventions (ACROBAT-NRSI)]. 61 The tool is based on the original Cochrane Risk of Bias tool for randomised studies40 and also builds on related tools, such as the quality assessment of diagnostic accuracy studies (QUADAS-2). 62

The ROBINS-I61 tool provides a detailed framework for assessment and judgement of risk-of-bias domains that may arise in three phases: (1) at pre-intervention, bias arising from confounding and selection of participants into the study; (2) at intervention, bias in measurement of interventions; and (3) at post-intervention, bias due to deviations from intended interventions, missing data, measurement of outcomes and selection of reported results. Each domain is rated as being at low, moderate, serious or critical risk of bias. A low risk of bias indicates that the study is comparable to a well-performed randomised trial in the domain being evaluated. A moderate risk of bias indicates that the study is sound for a non-randomised study but is not comparable to a well-performed randomised trial. A serious risk of bias indicates the presence of important problems in the domain and a critical risk of bias indicates that the study is too problematic to provide any useful evidence on the intervention effects. If insufficient information is provided to determine the risk of bias of a certain domain, the domain is marked as having no information. In general, the overall risk of bias of each study was determined to be equal to that of the most severe level of bias found in any domain.

All studies were analysed using ROBINS-I,61 regardless of whether or not the original study design included randomisation to other exposures, thus ensuring that risk of bias was assessed specifically for the comparisons of interest to this review. It is important to note that the quality assessment reflects how well a specific result evaluated the association of interest to this review, regardless of the objectives of the original study.

Methods of data synthesis and analysis

Venous thromboembolism was considered to comprise any subsequent recorded diagnosis of DVT or PE, or death attributable to either pathology. No attempt was made to distinguish between anatomical location, thrombus burden or clinical sequelae of VTE for this project, in accordance with the definitions of hospital-acquired thrombosis produced by NHS England. 63 Individual risk factors highlighted through regression, OR analysis or parametric testing as significantly associated with an increased or decreased likelihood of subsequent VTE were extracted. In particular, each paper was scrutinised for evidence of individual risk factors, especially those highlighted within current risk stratification tools,17,31–33 and their predictive performance was recorded, when available. Other risk factors demonstrating an association with VTE in the context of individual studies were also reported. A meta-analysis was not possible owing to significant levels of heterogeneity between studies, variable reporting items and the high risk of attributable bias. Descriptive statistics and thematic analysis were used to synthesise risk factors acting in a reproducible fashion across studies. Thematic analysis took an inductive/semantic form, using familiarisation and coding directed by data content. Consistent risk factor themes were then highlighted in ordinal fashion. All analyses were conducted using Microsoft Excel^® 2010 (Microsoft Corporation, Redmond, WA, USA).

Quantity and quality of research available

The literature searches identified 4771 citations. Of these, 15 studies9,11,23,50,52,53,60,64–71 met the inclusion criteria. Figure 4 presents a flow chart describing the process of identifying relevant literature. Sixty full-text articles were excluded as they did not meet all the prespecified inclusion criteria. The majority of the articles were excluded primarily on the basis of an inappropriate target population (not isolated lower-limb injury requiring temporary immobilisation), no data or analysis of risk factors associated with VTE, or an unsuitable publication type (i.e. reviews, commentaries, editorials or abstracts of excluded/included full-text papers). More specifically, two potentially relevant papers72,73 were excluded as they included a specific elective surgical population who were not considered to meet the inclusion criterion of lower-limb injury. A potentially relevant prospective observational cohort study74 was excluded, based on the authors’ conclusion of a low event rate precluding any subsequent analysis for predictors of VTE. Finally, a case–control study32 specifically seeking to derive a decision rule for the cohort of interest was excluded, based on the creation of this rule from a generic thrombosis cohort rather than a subgroup of patients with temporary lower-limb immobilisation. A full list of excluded studies with reasons for exclusion is presented in Appendix 6.

FIGURE 4.

Study flow chart (adapted):37 review of individual risk factors associated with VTE risk. © 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

© 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Description of included studies (design and patient characteristics)

The design and patient characteristics of the 15 included studies9,11,23,50,52,53,60,64–71 that provided data on individual patient identifiable risk factors associated with VTE risk in ambulatory trauma patients with temporary immobilisation following lower-limb injury is summarised in Table 4.

All studies were published between 1993 and 2017 (five were RCTs with conservative arms,23,50,52,53,60 three were prospective observational cohort or cross-sectional studies,65,67,69 one was a case–control study70 and six were retrospective cohort studies)9,11,64,66,68,71 and conducted in 10 countries (Australia,11,64,65 Canada,9,50 China,60 Denmark,71 France,69 Germany,23,52,53 Iran,67 the Netherlands,70 the UK66 and the USA). 68 Most of the studies (n = 11) were entirely outpatient based,11,23,50,52,53,60,64–67,69 whereas the remaining studies9,68,70,71 included patients with a short-duration inpatient stay to facilitate day-case surgery. In total, data were collated on 80,678 patients with a subsequent reported outcome of VTE positive or negative following temporary lower-limb immobilisation. The incidence of VTE across the studies with interpretable outcome data (79,202 patients) ranged from 0.22%66 to 23.5%9 (median 4.8%), mean age ranged from 33.8 years52 to 52.6 years71 and the proportion of male patients ranged from 45.8%65 to 86.1%,9 with a median across those studies with reported data of 56.3%.

The duration of follow-up varied between studies. Ten studies reported follow-up over a period of at least 3 months9,50,60,64–66,68–71 and one study followed up patients for up to 14 days. 67 Although four studies failed to record the duration of follow-up,11,23,52,53 two of these appeared to report follow-up only for the duration of the plaster cast, which averaged at 15.7 days53 and 17 days. 52 Eight studies collected data on risk factors prospectively via physician assessment or questionnaire,23,50,52,53,60,65,69,70 whereas six studies collected these data through clinical records, electronic patient notes or registries. 9,11,64,66,68,71 One study did not report the methodology for this aspect of data collection. 67 Analysis and methodology of VTE diagnosis subsequent to immobilisation varied markedly across studies and included prospective screening in all patients following plaster cast removal (seven studies),23,50,52,53,60,65,67 adjudicated diagnostic evaluation in those with symptoms (two studies)69,70 and retrospective identification of VTE through the interrogation of clinical records/health databases (six studies). 9,11,64,66,68,71 A single study66 looked only at the subsequent diagnosis of PE as an outcome, with reduced prevalence as expected. The association of individual risk factors with subsequent VTE was highlighted through regression analyses (nine studies),11,50,60,64,66,68–71 non-parametric tests (two studies)9,65 and descriptive statistics (four studies). 23,52,53,67

Quality characteristics

The overall methodological quality of the 15 included studies is summarised in Figure 5 and Table 5.

FIGURE 5.

ROBINS-I61 risk-of-bias assessment graph: review authors’ judgements about each methodological quality item across all included studies – review of individual risk factors associated with VTE risk.

All studies were deemed to be at overall moderate (seven studies)23,50,52,53,60,68,71 or serious (eight studies)9,11,64–67,69,70 risk of bias, using the ROBINS-I61 framework for assessment and judgement. Studies scoring a serious risk of bias did so predominantly on the selection of participants into the study, perhaps highlighting the issue with retrospective observational work into VTE outcomes; patients deemed to be at high risk in these cohorts are often treated with thromboprophylaxis (as highlighted in Table 4), or managed in a different manner from other patients. If this is not highlighted within a prospective analysis plan, a false low event rate is seen and risk is marginalised.

Narrative data synthesis

Age was the most consistent individual risk prediction factor, highlighted across 11 studies. 9,11,50,52,53,60,65,68–71 ORs reported for age varied from 1.0560 to 3.48,11 with limited estimates of precision. Meta-analysis was not undertaken owing to limitations in the data reported and concerns about the heterogeneous nature of the study populations. Injury type was the second most highlighted risk factor across six studies,11,50,52,53,69,70 all using multivariate logistic regression to suggest that severe traumatic injuries and fractures (when compared with soft-tissue injuries) were independently associated with VTE. Body mass index (BMI) was the third most consistent individual risk highlighted, noted as independently predictive of VTE across four studies,53,60,70,71 with ORs ranging from 1.20160 to 17.2. 70 Other risk factors were highlighted in fewer than four studies. Despite being present within several currently used risk stratification tools, pregnancy, recent hospital admission and preceding immobility prior to injury failed to demonstrate an independent association with VTE in any of the selected studies. Individual risk factors currently used within risk stratification tools and their association with VTE across all studies are shown in Table 6.

Other potential risk factors associated with subsequent development of VTE after lower-limb immobilisation included recent air travel (one study),64 coagulopathy and peripheral arterial disease (one study). 71 A single paper67 looked at the cumulative incidence of risk factors per patient and reported the presence of three or more factors to be significantly associated with the development of VTE. Methodology of reporting individual variables to have no association with subsequent VTE was inconsistent and heterogeneous. Six studies reported no association between sex and VTE in this cohort,11,50,52,65,66,70 five studies reported no association between exogenous oestrogen use and VTE9,50,52,53,65 and six studies reported no association between smoking and subsequent VTE. 9,50,52,65,69,71 Several papers produced conflicting results; six studies reported no association between raised BMI and subsequent risk of VTE9,50,52,65,68,69 and one study reported no association of VTE with increasing age. 66 These other identified risk factors and all negative associations are reported in Table 7.

Summary of key findings

• Increasing age and injury severity were the most consistent risk factors associated with the development of VTE in patients with lower-limb injury and temporary immobilisation.

• Many clinical features considered to be risk factors for VTE were not examined or associated with VTE in the studies.

• All studies included in the review were deemed to be at moderate or serious risk of bias.

• The evidence base for tailored risk prediction in people with lower-limb immobilisation due to injury is very weak.

Identification of studies

Studies were identified by searching the following electronic databases and research registers:

• Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily, MEDLINE and Versions(R) (via OvidSP), 1946 to May 2017.

• EMBASE (via OvidSP), 1974 to May 2017.

• Cochrane Database of Systematic Reviews (via Wiley Online Library), 1996 to May 2017.

• Database of Abstracts of Review of Effects (via Wiley Online Library), 1995 to March 2015.

• Cochrane Central Register of Controlled Trials (via Wiley Online Library), 1898 to May 2017.

• HTA database (via Wiley Online Library), 1995 to May 2017.

• NHS EED (via Wiley Online Library), 1995 to March 2015.

• Science Citation Index Expanded (via Web of Science), 1900 to May 2017.

• ClinicalTrials.gov (via US National Institutes of Health), 2000 to May 2017.

• International Clinical Trials Registry Platform (via Word Health Organization), 1990 to May 2017.

The search strategy used free text and thesaurus terms and combined synonyms relating to the condition (i.e. VTE in people with lower-limb immobilisation) with risk factor assessment or risk prediction modelling terms (used in the searches of MEDLINE, The Cochrane Library and EMBASE only). No language or date restrictions were used on any database. Further details on the search strategy can be found in Appendix 5. 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.

All identified citations from the electronic searches and other resources were imported into and managed using EndNote bibliographic software.

Inclusion and exclusion criteria

The inclusion of potentially relevant articles was undertaken using a two-step process. First, all titles were examined for inclusion by one reviewer (AP) and any citations that clearly did not meet the inclusion criteria (e.g. non-human, unrelated to VTE) were excluded. Second, all abstracts and full-text articles were then examined independently by two reviewers (AP and DH). When necessary, non-English-language studies were translated using Google Translate to facilitate study selection and subsequent data extraction. Any disagreements in the selection process were resolved through discussion or, if necessary, arbitration by a third reviewer (SG), and articles were included by consensus.

Studies were considered eligible for inclusion if they met the following criteria: (1) any study design that included a measurement of VTE patient outcome, (2) studies that recruited adults (aged > 16 years) who required temporary immobilisation (e.g. leg cast or brace in an ambulatory setting) for an isolated lower-limb injury and (3) any studies that reported any validation, or estimates of utility and performance of VTE risk assessment models for people with lower-limb cast immobilisation.

Data abstraction and quality assessment strategy

Data relating to study design, methodological quality and outcomes were extracted by one reviewer (AP) into a standardised data extraction form and independently checked for accuracy by a second reviewer (DH). Any discrepancies were resolved through discussion or, if necessary, arbitration by a third reviewer (SG) and included by consensus. When multiple publications of the same study were identified, data were extracted and reported as a single study.

There are no validated (or widely agreed on) tools for the assessment of prognosis research studies and there is little empirical evidence to support the importance of particular study features affecting the reliability of findings, including the avoidance of bias. 75 For this review, a generic list of important methodological features recommended by Altman76 and Moons et al. 77 for prediction modelling studies was deemed to be the most appropriate (i.e. useful) to assess the internal validity of the included studies. In general, five domains were considered important for assessing biases sufficiently large to distort the findings of prognosis research. These included (1) participant selection, (2) predictor assessment, (3) outcome assessment, (4) sample size and missing data and (5) statistical analysis. An overall judgement of bias was assigned as ‘low risk’ if all domains were judged as low risk, as ‘high risk’ if at least one domain was judged to be at high risk and as ‘unclear risk’ if an unclear risk of bias was noted in at least one domain and it was ‘low risk’ for all other domains. The methodological quality of each included study was independently evaluated by two reviewers (AP and DH). Any discrepancies were resolved through discussion or, if necessary, through the involvement of a third reviewer (SG). Blinding of the quality assessor to author, institution or journal was not considered necessary.

Methods of data synthesis and analysis

Venous thromboembolism was considered to comprise any subsequent recorded diagnosis of DVT, PE or death attributable to either pathology. No attempt was made to distinguish between anatomical location, thrombus burden or clinical sequelae of VTE for this project, in accordance with the definitions of hospital-acquired thrombosis produced by NHS England. 63 A narrative review of all identified scoring systems was performed, to compare design characteristics with thresholds for prophylaxis. Estimates of sensitivity, specificity, predictive values or likelihood ratios were directly extracted from validation studies or retrospectively calculated using available baseline data when applicable. All analyses were conducted using Microsoft Excel 2010.

Quantity and quality of research available

The literature searches identified 4771 citations. In total, only six studies32,78–82 met the inclusion criteria. Of these, one paper32 focused on prediction model development with external validation in independent data. The remaining papers focused on external model validation without model updating. 78–82 A flow chart describing the process of identifying relevant literature is given in Figure 6. A total of 69 full-text articles were excluded as they did not meet all of the prespecified inclusion criteria. The majority of the articles were excluded on the basis of inappropriate target population (i.e. not isolated lower-limb injury requiring temporary immobilisation), having no relevant RAMs or outcome evaluations and being an unsuitable publication type (i.e. reviews, commentaries, editorial or abstracts of included full-text papers). A full list of excluded studies with reasons for exclusion is presented in Appendix 7.

FIGURE 6.

Study flow chart (adapted):37 review of RAMs that predict the risk of VTE. © 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

© 2009 Moher et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Description of included studies (design and patient characteristics)

Study characteristics are described in Table 8. Two case–control studies32,82 and four observational studies78–81 that described the evaluation of seven RAMs were found. A single study32 presented data on derivation and validation; this study used a generic thrombosis database (including all acute thrombosis patients seen in an outpatient clinic matched with partner controls) to derive a new rule, rather than a relevant cohort of patients with temporary lower-limb immobilisation. All other papers looked to provide measures of external validation or implementation metrics regarding previously derived scores, with no description of the initial derivation methodology.

The seven identified RAMs varied in design, structure, output and threshold for prophylaxis. Several scores were dichotomous, with others providing ordinal measure of risk. Design characteristics and threshold levels for each RAM are presented for comparison in Table 9. The majority of RAMs focused solely on the estimate of thromboembolic risk; a single method featured characteristics designed to balance the risk of bleeding with thromboprophylaxis. 17 The individual predictors and their weighting varied markedly between RAMs. Variables (and their definitions) for each RAM are presented for comparison in Table 10.

Quality characteristics

The overall methodological quality of the six included studies is summarised in Figure 7 and Table 11. The majority of studies were deemed to be at high risk of bias,32,78,79,81,82 based primarily on participant selection; all reviewed studies attempted validation of proposed decision rules in heterogeneous cohorts, with pragmatic observational follow-up only. Outcomes were also non-standardised and varied by site in terms of description and inclusion.

FIGURE 7.

Risk-of-bias assessment graph:76,77 review authors’ judgements about each methodological quality item across all included studies in the review of RAMs that predict the risk of VTE.

Both case–control studies32,82 had specific and notable limitations in methodology. The larger study, by Nemeth et al. ,32 used three distinct generic thrombosis data sets to derive and validate the proposed RAM. As such, the tool is derived from a group of patients with unprovoked or hospital-acquired VTE and has limited potential generalisability to the cohort of interest in this study. The authors looked at a subgroup of patients with temporary lower-limb immobilisation within the study data, finding only a small proportion in whom to attempt validation (230 patients, 2% of the original derivation data set). The smaller study, by Watson et al. ,82 derived case–control data using an equal measure of appropriate patients with thrombosis following cast immobilisation, alongside a separate cohort of those without. As such, the prevalence of thrombosis within this study cohort was 50%. This is 20 times the estimated prevalence within the literature and, therefore, renders their estimates of predictive value at high risk of error.

Narrative data synthesis

The study by Nemeth et al. 32 derived and validated a clinical risk score for plaster cast patients: the Leiden thrombosis risk in plaster (cast) [L-TRiP(cast)] score. In this study,32 data from a large population-based case–control study of approximately 10,000 patients (4446 consecutive patients with a first episode of venous thrombosis and 6118 controls) were used in developing the model (included in patients with disease and many other confounding factors). After minimising the variables in an attempt to produce a clinical (14 environmental predictor variables) and pragmatic rule (11 predictor variables), the authors validated this rule in two subsequent VTE case–control data sets.

Watson et al. 82 assessed the diagnostic accuracy of five RAMs (only two of these, the GEMNet model17 and the Plymouth model,31 were specific to patients with lower-limb trauma and cast immobilisation) in a case–control study of 42 patients with lower-limb immobilisation following injury (21 cases and 21 controls). The reported sensitivity and specificity of the GEMNet model (85.7% and 47.6%, respectively) did not seem to be compatible with the numbers of cases and controls, so contact was made with the authors for clarification. They identified an error in the sensitivity, which should have been reported as 4.76%, and provided the raw numbers for both RAMs, which were used to calculate all diagnostic parameters and confidence intervals (CIs).

Prognostic accuracy measures for the three scores evaluated in these two studies are presented in Table 12. Sensitivity ranged from 57.1% to 92.6% across the RAMs and specificity ranged from 4.76% to 60.8%. The L-TRiP(cast) data are displayed in this table using thresholds denoting optimal performance and to allow direct comparison with other validated scores. The estimates of positive and negative predictive value for the L-TRiP(cast) score were modelled using an appropriately low prevalence of VTE, whereas the estimates for the GEMNet and Plymouth models used the artificial 50% prevalence from the case–control study. This explains the relatively high positive predictive values and the relatively low negative predictive values for these scores.

The area under the curve (AUC) for the L-TRiP(cast) score ranged from 0.77 (95% CI 0.66 to 0.87) in the derivation cohort to 0.77 (95% CI 0.58 to 0.96) and 0.95 (95% CI 0.91 to 0.99) in the two subsequent validation cohorts.

In addition to these rules, four additional models were identified (Saragas et al. ,81 Haque et al. ,80 Giannadakis et al. 79 and Eingartner et al. 78). No measures of external validation of these RAMs were found. All were found as single-centre small-scale implementation studies, revealing no further additional information on performance, utility or reliability.

Summary of key findings

• A number of RAMs have been developed using a variety of methods and based on a variety of predictor variables.

• External validation studies have weak designs and limited generalisability, so estimates of prognostic accuracy are very uncertain.

• The limited data available suggest that the L-TRiP(cast) score with a cut-off point of 8 can achieve reasonable sensitivity for predicting VTE without an excessive loss of specificity.

Delphi methodology

A three-round Delphi study was conducted between August 2017 and April 2018 using a panel of international topic experts, identified from the published literature and national clinical research network (injuries and emergency theme).

Expertise was ascribed using two criteria: (1) evidence of experience in relevant guideline or risk assessment tool design and (2) routine clinical experience with the relevant patient cohort and topic of interest. These criteria were selected to allow disclosure of unpublished methodology deriving existing rules, confer practicality of any rule and to ensure broad dissemination and uptake of consensus findings. Experts were identified through national bodies, relevant literature and local thrombosis committee groups. National bodies included The British Orthopaedic Association, The Royal College of Emergency Medicine and The Clinical Research Network Injuries and Emergencies Specialty group. Additional national or international societies were not approached for independent representation owing to time constraints and workload. All individuals approached agreed to participate and provided written informed consent. Experts were entirely independent of the core review team (n = 11), local colleagues independent of the review work (n = 3) and members of the core review team (n = 6).

The open first round of the classical Delphi approach was replaced with a systematic literature review to identify possible VTE predictors, as previously described. Round 1 of the Delphi study was then delivered via a web-based platform using SmartSurvey Ltd (Tewkesbury, UK)90 through a subscribed account. All potential individual predictor variables identified through the previous systematic review of existing decision rules and the wider literature on risk prediction in the relevant cohort were presented to participants as potential candidate predictors. Participants were then asked to rate the strength of the VTE prediction risk for each variable using a four-point Likert scale, with the option to record uncertainty. Their replies were collated into quantitative and qualitative output for each individual question. An opportunity to identify new relevant predictor variables was provided and participants were encouraged to identify missing themes.

New candidate variables were proposed, and those failing to achieve consensus in round 1 of the Delphi exercise were carried forward to a second round. Participants were presented with these variables together with a summary of the panel results from round 1, when applicable. Participants were asked to complete the same Likert scale as before, with the advantage of having additional insight into comments and quantitative results revealed by the rest of the group.

At the end of round 2 of the Delphi exercise, all variables were carried forward to a facilitated round-table discussion where consensus results and comments were provided to all participants. Data were collated and analysed to calculate frequencies, mean and range of scores.

Data synthesis and analysis

Criteria for inclusion in the decision rule were defined by a variable identified as a moderate or strong predictor by consensus between two, three or more respondents. Variables identified as uncertain or not or weakly predicting VTE risk by consensus between two, three or more respondents were likewise excluded from the exercise. Quantitative data from rounds 1 and 2 were presented to participants as bar charts with percentages. All analyses were conducted using Microsoft Excel 2010.

Expert engagement

Twenty participants were identified to participate in the study. All (100%) completed round 1; 19 participants (95.0%) completed round 2. Ten participants (50.0%) contributed to the final facilitated round-table discussion. A list of participants and the clinical scope of the Delphi panel is provided in Appendix 8.

Systematic review and identification of candidate predictors

Trial flow results from the relevant systematic reviews prior to the Delphi exercise have been previously described (see Review of individual risk factors associated with venous thromboembolic risk and Review of risk assessment models for predicting venous thromboembolic risk). Thirty-five individual candidate predictors were identified and included for dissemination in round 1. Predictors were subdivided into risks related to injury/immobilisation and generic thrombosis risks. Initial proposed candidate predictors are shown in Table 13, which also shows the results of subsequent Delphi rounds.

Round 1 results

Seven variables were identified as predictive of VTE risk by consensus criteria during round 1. Thirteen variables were identified as not predictive of VTE by consensus criteria and excluded from the exercise. No consensus was achieved on 15 variables; 14 predictors were carried forward to the second round, with moderated peer feedback and tabular display. A single variable was excluded from the exercise after collated comments, group discussion and feedback regarding the lack of primary suitability for inclusion. Four new variables were suggested during the round 1 exercise and these were also carried forward. All candidate predictors and their round 1 results are presented in Table 13.

Round 2 results

In the second round, consensus was achieved on six further variables as predictive of VTE risk. Five variables were identified on reflection as not predictive of VTE risk by consensus criteria and these were excluded from the exercise. No consensus remained for 7 of the 17 variables carried forward from round 1. No further risk predictors were suggested by participants during round 2. Candidate variables taken forward and their round 2 results are presented in Table 13.

Variables failing to achieve consensus

Of the seven variables failing to achieve consensus, two failed because of a dichotomous split with clear unresolvable disagreement by experts (intravenous drug use and comminuted fracture). The other five variables appeared to be categorised as weakly to moderately predictive by all, but fell short of agreed criteria for inclusion. Further rounds were deemed unlikely to generate further consensus at this stage and the trial team proposed a move to the facilitated round-table discussion to achieve consensus.

Facilitated round-table discussion

There was general round-table agreement about all variables for which consensus had been achieved by the Delphi exercise. A specific point was made by the group regarding the inclusion of several variables depicting the degree of immobilisation of the calf pump, and whether or not this should become a single ordinal variable. A single variable (active intravenous drug use) on which the group did not reach consensus was discussed in further detail, with the majority of the round table proposing that the variable be refined or this entire cohort of patients be excluded based on safety concerns. Several discussion points followed regarding the need for strict inclusion criteria when applying any decision rule, with particular regard to the type of immobilisation in this group. The final agreed variables considered to be predictors of VTE risk via the Delphi expert consensus were eight generic VTE risk predictors (i.e. thrombophilia, pregnancy/puerperium, active cancer, surgery in the preceding 3 months, prior VTE, exogenous oestrogen/hormone therapy, lower-limb paralysis and superficial thrombophlebitis), two patient demographics (i.e. age and BMI) and three variables specific to lower-limb immobilisation or injury (i.e. Achilles tendon rupture, rigid immobilisation and above-knee cast).

Table 14 compares the expert consensus variables to those included in the RAMs. Most of the expert consensus variables were included in one or more of the RAMs but the RAMs also included many variables that were not supported by expert consensus.

Originally, it was planned to use the expert consensus methods to refine existing RAMs or construct up to five new RAMs from the selected risk factors, and then produce a consensus estimate of sensitivity and specificity for each RAM that would allow the exploration of the trade-off between sensitivity and specificity in decision-analytic modelling. It was decided not to proceed with this for the following reasons:

• The difficulty of achieving consensus on individual risk factors (which was felt to be unsurprising given the limited available evidence) suggested that it would not be possible to achieve the necessary consensus on the content and structure required to refine an existing RAM or construct a new RAM.

• The receiver operating characteristic (ROC) analysis of the L-TRiP(cast) score published by Nemeth et al. 32 (identified in the systematic review of RAMs) provided an estimate of the trade-off between sensitivity and specificity that could be used in decision-analytic modelling, and would be more credible and usable than expert-derived estimates of sensitivity and specificity for expert-derived RAMs.

Summary of key findings

• Expert consensus on 13 variables most likely to predict VTE risk for outpatients with lower-limb injury and temporary immobilisation has been established: eight generic VTE risk predictors (i.e. thrombophilia, pregnancy/puerperium, active cancer, surgery in the preceding 3 months, prior VTE, exogenous oestrogen/hormone therapy, lower-limb paralysis and superficial thrombophlebitis), two patient demographics (i.e. age and BMI) and three variables specific to lower-limb immobilisation or injury (i.e. Achilles tendon rupture, rigid immobilisation and above-knee cast).

• It was not possible to achieve expert consensus on the following seven variables: intravenous drug use, significant injury in the preceding 3 months, significant medical comorbidity, preceding immobility, comminuted injury, non-weight-bearing status and family history of VTE.

• Injury- and plaster-associated risk was proposed as a single ordinal variable based primarily on the degree of calf pump immobilisation.

Identification of studies

Studies were identified by searching the following electronic databases and research registers:

• Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily, MEDLINE and Versions(R) (via OvidSP), 1946 to October 2017.

• EMBASE (via OvidSP), 1974 to October 2017.

• Cochrane Database of Systematic Reviews (via Wiley Online Library), 1996 to October 2017.

• Database of Abstracts of Review of Effects (via Wiley Online Library), 1995 to March 2015.

• HTA database (via Wiley Online), 1995 to October 2017.

• NHS EED (via Wiley Online Library), 1995 to March 2015.

For the MEDLINE and EMBASE searches, the aim was to maximise the specificity of the search in these databases that do not target economic evaluations. In these searches, the keyword strategies developed in the review of clinical effectiveness (see Chapter 3, Objective), which combined terms related to VTE with terms related to lower-limb immobilisation, were used with sensitive economic evaluation, quality of life and cost search filters91–93 aimed at restricting search results to economic and cost-related studies. For the searches of HTA and NHS EED, terms for lower-limb immobilisation were not included, in order to maximise sensitivity and find all economic evaluations relating to VTE. All resources were searched initially from inception to October 2017 (or, in the case of discontinued sources such as NHS EED, the date at which coverage ceased). Further details on the search strategy can be found in Appendix 9.

All identified citations from the electronic searches and other resources were imported into and managed using EndNote bibliographic software.

Inclusion and exclusion criteria

The inclusion of potentially relevant articles was undertaken using a two-step process by a single reviewer. During the screening stage, all titles were examined for inclusion by the health economist and any citations that clearly did not meet the inclusion criteria (e.g. non-human, unrelated to VTE) were excluded. During the screening stage, abstracts were also examined if the paper could not be excluded based on the title alone. During the eligibility stage, all full-text articles that could not be excluded based on title and abstract were examined.

Studies were selected for inclusion on the basis of pre-determined inclusion and exclusion criteria. The population, interventions and comparators were defined as per the clinical effectiveness review of pharmacological thromboprophylaxis for preventing VTE (see Chapter 3, Inclusion and exclusion criteria). In addition, the study population was restricted to patients treated in the UK NHS because estimates of resource use and costs may not be transferable between different health-care settings. Criteria for the study design and outcomes were specified to identify studies meeting the NICE reference case. 94 In terms of study design, the review was restricted to cost-effectiveness studies, thereby excluding cost-minimisation studies. In terms of outcomes, the review was restricted to studies that measured benefits using QALYs because the QALY has been defined by NICE as the reference case measure for benefit for UK cost-effectiveness studies94 and the NICE methods guide94 provides guidance on the range of cost-per-QALY values that can be considered to represent good value for money within the UK NHS. Studies reporting alternative outcomes, such as cost per VTE avoided, were therefore excluded. Cost-consequence studies and other studies that reported data on costs, resource use or utility values in relevant populations were excluded from the cost-effectiveness review but were examined for relevant data that might inform the modelled estimates of cost-effectiveness (described in Independent economic assessment methods).

Identification of data sources

A summary of the model parameters and data sources used to populate the model are provided in Table 15, with the exception of utility values, which are provided in Treatment-related utility decrement. The number of patients receiving thromboprophylaxis under the various treatment strategies is estimated by combining data on sensitivity and specificity, from the systematic review of RAMs (see Chapter 3, Review of risk assessment models for predicting venous thromboembolic risk), with data on the risk in untreated patients, from the review of RCTs comparing thromboprophylaxis with no thromboprophylaxis (see Chapter 3, Review of pharmacological thromboprophylaxis for preventing venous thromboembolism). Parameters relating to the relative risk (RR) of VTE events in patients receiving thromboprophylaxis compared with those not receiving thromboprophylaxis have been taken from the review of clinical effectiveness of thromboprophylaxis to prevent VTE in lower-limb immobilisation. For all other parameters, data were obtained by reviewing the data sources used in published cost-effectiveness analyses and other published sources. As no relevant models were identified that were specific to the population having lower-limb immobilisation, the search for parameter sources was broadened to models in related populations or indications. This search focused, first, on models used to inform NICE guidance on anticoagulants to prevent and treat VTE as the models used to inform any existing NICE guidance were considered highly likely to contain data relevant to the UK context and the NICE reference case, and have been critically scrutinised by independent academic groups. Second, the additional studies identified in the literature review for the related population of patients having elective knee surgery were used (as discussed in Results of cost-effectiveness review). Finally, ad hoc searches were conducted using the reference lists of recently published reviews to identify any relevant sources not identified during the review of published models and clinical experts were consulted to see if any key data sources known to them had been missed. This included any sources used in relevant models known to the authors. The published cost-effectiveness analyses identified as potential sources of model inputs are summarised in Appendix 11.

Characteristics of population having lower-limb immobilisation

The Prevention of Thrombosis after Lower Leg Plaster Cast (POT-CAST) trial by van Adrichem et al. 59 was selected as being representative of the population having lower-limb immobilisation due to injury who are at risk of thrombosis. This was selected as it is a recent, large RCT conducted exclusively in Europe (the Netherlands) and the inclusion and exclusion criteria were not too restrictive; patients with a prior history of VTE were excluded but those with cancer or family history of DVT were not excluded. The starting age in the model was set to be 46 years of age and 51.5% of the starting population were male.

Risk of venous thromboembolism and types of venous thromboembolism

The risk of VTE was based on data taken from the systematic review of clinical effectiveness for LMWH versus control (see Chapter 3, Review of pharmacological thromboprophylaxis for preventing venous thromboembolism). The rate of any form of VTE (i.e. PE, symptomatic DVT or asymptomatic DVT) was 8.39% across the 12 studies23,49–56,58–60 reporting data for control (i.e. placebo, aspirin, no thromboprophylaxis). The risk of PE, including those that experienced PE with DVT, was 0.43%. Therefore, the risk of DVT without PE was 7.96%. Only nine studies23,49–52,54,56,58,60 reported outcomes for both symptomatic and asymptomatic DVT. From these studies, it was estimated that 11.4% of DVT are symptomatic (16/140), giving an absolute incidence of 0.91% for symptomatic DVT and 7.05% for asymptomatic DVT in patients having lower-limb immobilisation without thromboprophylaxis. These are slightly higher than the raw figures (Table 16) as three studies53,55,59 did not report outcome data for both symptomatic and asymptomatic DVT but did report data for all VTE and PE. These estimates effectively assume the same split of symptomatic and asymptomatic DVT for the three studies with missing data to achieve the total VTE rate observed across all 12 studies. For asymptomatic DVT, only six studies23,50–52,58,60 reported outcomes for both distal and proximal asymptomatic DVT. From these six studies, it was estimated that 16% (9/56) of asymptomatic DVTs are proximal (and, therefore, 84% are distal). The data on the location of symptomatic DVTs (distal or proximal) were very sparse, with only 46 events reported across all trial arms and many studies not reporting the location. It was possible to extract data on the location of symptomatic DVTs for four studies54,57–59 included in the review, covering a total of 34 events across all trials arms. The proportion of symptomatic DVTs that were proximal varied from 14% to 100% across these four studies, with a median estimate of 36%. The POT-CAST trial59 was the only one of these four studies that focused exclusively on symptomatic VTE. The other three studies54,57,58 screened patients for asymptomatic DVTs as well as recording the number of DVTs that were symptomatic. There is a risk that the process of screening patients for asymptomatic DVTs may lead to patients attributing symptoms to screening-detected DVTs that may otherwise not have been sufficiently symptomatic to result in the patient presenting with symptoms in routine care. For this reason, it was decided to use the POT-CAST trial59 as the single source of estimates for the proportion of symptomatic DVTs that are proximal, as no routine screening was used in this trial to detect asymptomatic DVTs. As this single study provided a low number of events, it was decided to pool the data across trial arms. This resulted in 50% of the 16 symptomatic DVTs being proximal. Given that these data were based on a single study and a small number of events, two extreme scenarios were examined, assuming that either 0% or 100% of symptomatic DVTs are proximal, to see if the overall conclusions were sensitive to uncertainty around this parameter.

All-cause mortality

It was not considered necessary to model males and females separately as the risks of VTE during lower-limb immobilisation were not expected to vary according to sex. However, life-expectancy is dependent on sex and, consequently, any QALY gains from deaths prevented would be dependent on sex. For this reason, a weighted mortality risk was applied, based on the proportion of males (51.5%) in the POT-CAST trial,59 using data on the risk of death for males and females by age obtained from lifetables. The weighting across males and females was allowed to vary over the course of the model to allow for the fact that lower mortality risks in females leads to a slight increase in the proportion of people alive who are female over time.

Case fatality rate for pulmonary embolism

None of the studies included in the review of clinical effectiveness reported any fatal PEs. Given that 14 PEs were reported within the 13 studies23,49–60 included in the review, this suggests that the rate of fatality due to PE is < 7% in the population having lower-limb immobilisation due to injury. The research team looked for alternative data sources on the case fatality rate of PE. Data were identified from the Computerized Registry of Patients with Venous Thromboembolism (RIETE) (an extensive data registry of consecutive patients with VTE), which found that the rate of all-cause mortality at 30 days had fallen over time from 6.6% in 2001–05 to 4.9% in 2010–13 (reported by Jiménez et al. 118). However, some of this mortality may not be related to PE. When looking at only PE-related mortality, the rate at 30 days was 1.8%. Another analysis of RIETE, by Maestre et al. ,108 found that rates of fatal PE at 90 days were lower for outpatients than inpatients. They found that of 7591 outpatients with clinically overt PE, 219 (2.9%) had fatal PE at 90 days; this was lower than the 4.1% (119/2870) rate with clinically overt PE having a fatal PE in the inpatient population. Given that patients having lower-limb immobilisation due to injury would generally be managed as outpatients, this suggests that a lower risk may be seen in this population.

The Emergency Medicine Pulmonary Embolism in the Real World Registry (EMPEROR) cohort study (reported by Pollack et al. 119), which recruited patients with PE from EDs in the USA, reported a 30-day all-cause mortality rate of 5.4% for patients with confirmed PE, but the rate of mortality was only 1.1% when restricted to PE-related in-hospital mortality. These data seem reasonably consistent with the 30-day data from RIETE.

It was decided to apply the case fatality rate for outpatients from the RIETE study (2.9%) in the model, as patients having lower-limb immobilisation would be managed as outpatients and there was a slight increase from 30 days to 90 days, which is captured in the analysis by Maestre et al. 108 but not in the other two analyses, which were limited to 30 days.

Combining the incidence and case fatality rate for PE, it is estimated that one fatal PE per 8097 patients having lower-limb immobilisation without thromboprophylaxis would be expected, which is consistent with there being no fatal PEs observed in the studies included in the systematic review for treatment effectiveness.

Effectiveness of thromboprophylaxis on prevention of venous thromboembolism during lower-limb immobilisation

We applied the OR for LMWH versus control for the outcome of any VTE estimated from the NMA to all forms of VTE within the model. Although the NMA provides estimates of clinical effect that are specific to different types of VTE (i.e. symptomatic DVT, asymptomatic DVT and PE) the estimates for these specific VTE types are more uncertain as they are based on fewer events. The pooled effect for the OR (0.52, 95% CrI 0.37 to 0.71) was converted to a RR and applied to the baseline risk within the model.

For the sensitivity analyses examining thromboprophylaxis using alternative drug classes, it was assumed that DOACs would have the same effectiveness as LMWH, as no studies examining the effectiveness of DOACs in this population were included in the systematic review. For the analysis assuming that fondaparinux is used, the OR for fondaparinux versus control from the NMA was applied (0.13, 95% CrI 0.05 to 0.30). In both cases, the same adverse effects profile for LMWH, DOACs and fondaparinux were implicitly assumed, as the only adverse event included in the model was bleeding and it was assumed that all prophylaxis had the same impact on bleeding rates.

Risk of bleeding during lower-limb immobilisation

There is a risk of major bleeding within the general population regardless of whether or not they receive treatment with anticoagulants. A cohort study (reported by Hippisley-Cox and Coupland101), conducted in a large primary care population using data from 4.4 million patients with 16.4 million person-years of follow-up, estimated an age-standardised incidence rate of 1.34 (95% CI 1.32 to 1.36) per 1000 person-years for upper GI bleeding and 0.55 (95% CI 0.54 to 0.56) per 1000 person-years for intracranial bleeds for patients not taking anticoagulants. Given that the study cohort was patients aged 21–99 years registered with a general practice and that the only relevant exclusion criteria appears to be existing anticoagulated patients, this study was considered to be reflective of the risk of GI bleeds and ICH in the general population at risk of lower-limb injury who are not currently taking anticoagulants. The baseline characteristics of patients not taking anticoagulants in the study by Hippisley-Cox and Coupland101 were compared with the characteristics of those in the comparator arm of the POT-CAST trial59 (Table 17). Although Hippisley-Cox and Coupland101 did not present a mean age, a weighted mean age was calculated using the distribution across age bands and the mid-point ages for each band: this was found to be 40 years (with a range of 36–45 years when using the lower and upper limits of the age bands), which is lower than the mean age of 46 years from the POT-CAST trial. 59 The proportion of non-smokers was slightly lower in the POT-CAST trial and the proportion of patients with previous cancer was slightly higher.

To see if the rates of major bleeding in the RCTs included in the clinical effectiveness review were similar, data on the rates of major bleeding (see Appendix 3) were combined with data on the number of person-years people were at risk of bleeding within each study. It was assumed that the period at risk was the period of cast immobilisation, unless the study explicitly reported a longer duration of follow-up. The number of person-years at risk could be estimated for only 11 of the 13 studies. 23,49–56,59,60 One study used an active control group,57 whereas another did not explicitly report the duration of follow-up. 23 There was only one episode of major bleeding across 447 person-years, giving an incidence of major bleeding of two per 1000 person-years. This appears to be relatively consistent with the estimates for ICH and GI bleeds combined from the cohort reported by Hippisley-Cox and Coupland. 101 Therefore, it was decided to use the age-standardised estimates from the cohort reported by Hippisley-Cox and Coupland101 within the model.

Thromboprophylaxis would be expected to increase the risk of a major bleed but the very small number of major bleeding events in the systematic review meant that the estimate of the risk of major bleeding due to thromboprophylaxis in people with lower-limb immobilisation was very uncertain (OR 1.45, 95% CrI 0.08 to 32.17). Although the baseline risk of bleeding will vary between patient groups, there is no strong reason to believe that the relative effect of thromboprophylaxis will vary. Therefore, data from the NICE clinical guideline (CG) on reducing the risk of VTE (NICE CG92)24 was used, which pooled all indications into a single meta-analysis. 24 These data were reanalysed to use a random-effects approach. In this meta-analysis, the median OR was found to be 1.643 (95% CrI 0.90 to 2.53). The OR was converted to a RR and applied to the baseline risk within the model.

The significance of major bleeding depends primarily on whether or not it is fatal; for non-fatal bleeds, the significance depends on whether or not it is intracranial, as ICH has the potential to result in long-term disability. Fang et al. 102 examined death and disability following bleeds associated with warfarin (Coumadin^®, Bristol-Myers Squibb Company, New York City, NY, USA). These data were used to estimate the proportion of major bleeding episodes that were fatal. Fang et al. 102 found that 48.6% of patients having ICH died within 30 days, whereas only 5.1% of patients having non-intracranial bleeds died within 30 days. 102 Although it is acknowledged that bleeds associated with warfarin use may be more likely to result in fatalities than bleeds in patients not having anticoagulation,120 it seemed reasonable to apply the case fatality rates from bleeds associated with warfarin use in the model as these would be directly applicable to the additional cases of bleeding associated with thromboprophylaxis. Any overestimation of mortality in those having bleeds not related to anticoagulant use would be similar across arms and, therefore, would affect estimates of absolute QALYs for each thromboprophylaxis strategy but not the incremental QALY gains when comparing thromboprophylaxis strategies.

The case fatality rate for non-intracranial bleeds reported by Fang et al. 102 was lower than the case fatality rates observed for GI bleeds by Rockall et al. 121 and Button et al. ,103 but this may be because GI bleeds have a greater mortality risk than other non-ICHs. Although Button et al. 103 found that the case fatality rate had fallen over time, the case fatality rate of 10% estimated by Button et al. 103 appears to be consistent with a 2015 report122 from the National Confidential Enquiry into Patient Outcome and Death, based on data collected in 2013. As the data from QBleed101 related only to the incidence of ICH and GI bleeds, the 10% risk of mortality from Button et al. 103 was used instead of the 5% for extracranial bleeds from Fang et al. 102

In addition to the risk of ICH being fatal at the time of the bleed itself, there is an increased risk of death in patients having ICH compared with population norms. Fogelholm et al. 107 found that patients with primary ICH who survived to 28 days had a standardised mortality rate that was 4.5-fold higher in the first year and 2.2-fold higher in years 2–6 than that of age- and sex-matched controls. These figures were applied as RRs to the all-cause mortality rates to increase the risk of mortality in those patients who survived ICH. Although Fogelholm et al. 107 reported a 10% reduction in mortality compared with general population norms in years 7–16, no increased or decreased risk beyond 6 years was assumed as there was doubt over the clinical plausibility of a reduced risk of death following ICH.

Risk of bleeding during anticoagulation treatment of venous thromboembolism

Several sources of data on bleeding risks during treatment of VTE were identified. These included the RIETE105 and Prevention of thromboembolic events – European registry in venous thromboembolism (PREFER-VTE)123 studies and a systematic review of RCTs and cohort studies by Carrier et al. 124 In the review by Carrier et al. ,124 56 of the 69 studies included were RCTs, which the authors say are not generalisable to all patients with VTE because those with additional comorbid conditions are often excluded from clinical trials. Carrier et al. 124 reported a rate of major bleeding for any VTE of 1.6% (95% CI 1.3% to 2.0%) at 3 months and a rate of fatal bleeding of 0.2% (95% CI 0.1% to 0.3%). The case fatality rate for bleeding following any VTE was 11.3%. RIETE reports a rate of major bleeding of 2.24% (95% CI 2.05% to 2.42%) at 3 months and a rate of fatal major bleeding of 0.55% (95% CI 0.46% to 0.65%) giving a case fatality rate equivalent to 24.7%. 105 This higher case fatality rate may be because of a higher rate of comorbidities in an unselected cohort. In the PREFER-VTE cohort, the rate of major bleeds was 1.5% at 12 months but the rate of fatal bleeds was not reported. 123 In addition, the timing of the bleeds was not reported so it is unclear what the rate of bleeding would have been during the treatment period. The data from PREFER-VTE were considered to be more applicable to current UK practice than the data from RIETE as PREFER-VTE included only European sites, included UK patients and recruited patients after the introduction of DOACs. However, it was noted that only 56.7% of patients in the PREFER-VTE cohort had a low risk of bleeding at baseline [a HAS-BLED (Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalised ratio, Elderly, Drugs/alcohol concomitantly) score of 0 or 1]. 123 As thromboprophylaxis for lower-limb injury is likely to be limited to those with a low risk of bleeding, the research team looked for alternative data sources. A study by Kooiman et al. 104 estimated the incidence of major bleeding at 6 months for patients having treatment for VTE stratified by HAS-BLED score. The combined incidence for patients with a HAS-BLED score of 0 or 1 was 0.9% (3/335). Kooiman et al. 104 did not report risks separately for PE and DVT but Cohen et al. 123 reported that, in PREFER-VTE, the rates of major bleeds were not significantly different after PE and DVT, so the same rate of 0.9% has been applied to both DVT and PE. This seems reasonable given that the same treatment is assumed in both groups.

Owing to the greater detail reported by RIETE, the breakdown of major bleeding events into fatal events, non-fatal ICH and other non-fatal major bleeds was based on the event rates from RIETE. 105 The analysis of RIETE also found that symptomatic PE was not a significant predictor of whether a fatal bleed occurred and the site of DVT was not a significant predictor unless the DVT was distal when the OR was 0.39 (95% CI 0.16 to 0.95). This lower rate may be as a result of less aggressive anticoagulation therapy or another confounding factor not accounted for in the analysis. It was decided to apply the same rates of fatal bleeding in all patients treated for VTE regardless of the VTE event type.

Clinically relevant non-major bleeding during prophylaxis

Clinically relevant non-major bleeding was not included in the base-case analysis as it has a low impact on quality of life and would be likely to incur a relatively small cost compared with major bleeding. However, to test whether or not the model was sensitive to this structural assumption, the cost of one ED attendance was applied to those experiencing CRNMB. To estimate the rate of CRNMB, the ratio of CRNMB to major bleeding (2.05) from Cohen et al. 123 (PREFER-VTE) was applied to the rate of major bleeding for those having prophylaxis to estimate the proportion of patients having CRNMB during prophylaxis. Owing to the low rate of bleeding during prophylaxis, the cost of CRNMB was only £0.19 for each patient receiving prophylaxis. Therefore, the exclusion of this cost from the base-case analysis is not expected to have significantly biased the base-case results.

Risk of chronic thromboembolic pulmonary hypertension

The incidence of CTEPH following PE was based on a systematic review by Ende-Verhaar et al. ,111 which reported an overall incidence of CTEPH of 2.3% (95% CI 1.5% to 3.1%). The review111 also reported the incidence separately for all patients with PE (0.56%, 95% CI 0.13% to 0.98%) and for those who survived the initial 3- to 6-month treatment period for PE (3.2%, 95% CI 2.0% to 4.4%). The follow-up period in the studies included in the review varied from 3 months to 8 years, with most reporting follow-up of ≈2 years. One study, by Pengo et al. ,112 which followed patients prospectively for 10 years, found that no cases occurred beyond 2 years, despite a median follow-up of 94 months. It is therefore assumed in the model that all cases are diagnosed between 6 months and 2 years after PE. The rate of CTEPH for those surviving the initial 3- to 6-month treatment period (3.2%) was considered to be the most relevant estimate for the model as no diagnoses of CTEPH are assumed until patients complete their anticoagulant treatment, and deaths during VTE and during anticoagulation are already accounted for in the decision tree part of the model.

Mortality risks in patients with CTEPH were based on data from an international prospective registry reported by Delcroix et al. 113 A recent HTA report by Goodacre et al. 114 used data from this registry to estimate survival curves for patients having medical or surgical management of CTEPH. The same survival curves were used in this model to estimate the hazard of death over time for patients diagnosed with CTEPH. The proportion having surgical management in the registry was 60% and this was applied in the model. Given that deaths related to PE occurring within 6 months of PE are already accounted for in the model, all-cause mortality has been applied to those with PE in the first year regardless of whether or not CTEPH occurs and the rate of mortality for those with CTEPH is applied from year 2 onwards. To ensure that the risk of death in the CTEPH group was not artificially low compared with the risk of death in the general population, general population mortality risks were applied whenever these were higher than the risk in the CTEPH population, based on the survival curves.

Risk of post-thrombotic syndromes

A review of epidemiological studies by Galanaud et al. 125 describes how the incidence of PTS is dependent on the scoring system used to diagnose PTS, the location and extent of the DVT (e.g. proximal vs. distal), and the duration of time since DVT. Galanaud et al. 125 also discuss how pre-existing chronic venous insufficiency and the symptomatic versus asymptomatic nature of the DVT may also affect the risk of PTS.

The risk of PTS in patients with symptomatic DVT in the model was estimated from a prospective registry [Thrombosis and Pulmonary Embolism in Out-Patients – plus (TULIPA-PLUS)], reported by Hach-Wunderle et al. 109 This source was chosen as the study design excluded patients with previous DVT or signs of PTS prior to the index DVT. The incidence is reported separately for proximal and distal DVT and is stratified into mild, moderate and severe PTS using a validated scoring system (Villalta scale), which is the scale preferred by the International Society on Thrombosis and Haemostasis. 126 The prevalence of symptomatic PTS at 3 years was 24.4% for all DVTs, 32.4% for proximal DVT and 15.6% for distal DVT, with 70% being mild, 24% being moderate and 6% being severe (average across both proximal and distal). 109

One applicability issue related to this study was that patients received anticoagulation for a whole year, which is longer than the period of anticoagulation assumed in the model. Therefore, the rates of PTS may be underestimated if longer anticoagulation is protective of PTS. However, as it was not known if longer anticoagulation was protective, and the risk of PTS in this study may reasonably be lower than other estimates as a result of careful exclusion of PTS symptoms at the time of the index DVT, it was decided that it was reasonable to apply the incidence from Hach-Wunderle et al. 109 in the model given that the incidence was in the range generally reported (20–50%). 125

There are only limited data on the incidence of PTS in patients with asymptomatic DVT because these patients are less likely to be followed up for a PTS diagnosis as their DVT is less likely to be diagnosed outside a research setting. However, a study by Persson et al. 127 examining PTS incidence in 83 patients operated on for Achilles tendon rupture, which prospectively identified DVT using colour duplex scanning 3 and 6 weeks after surgery and assessed PTS (assessors were blinded to DVT diagnosis) 5 years post operatively, reported a low rate of a PTS diagnosis (Villalta score of ≥ 5) of 8% (3/38) in the DVT group and 4% (2/45) in the no-DVT group. All of the diagnosed cases had a Villalta score of < 15 and were therefore of mild to moderate severity. However, it should be noted that most of those patients with asymptomatic DVT diagnosed post operatively received warfarin for 3 months and were advised to wear compression stockings for 1 year. As a result, the rate of PTS in asymptomatic, undiagnosed and, therefore, untreated DVT is unknown but is likely to be at least 8%. Furthermore, only three of the 38 DVTs were proximal, making it difficult to estimate the rate of PTS in patients with asymptomatic proximal DVTs from this study. 127

A study by Schindler and Dalziel128 in a related population (those having total hip or knee arthroplasty) found that four of 34 (11%) with asymptomatic distal DVT and three of eight (32.5%) with asymptomatic proximal DVT developed PTS 15 to 24 months after surgery. In the knee arthroplasty subgroup (which is considered to be more clinically relevant to those having lower-limb immobilisation), 28 patients had distal DVT with one of these having clot propagation leading to proximal DVT. Four of these 28 patients (14.3%) with DVT developed PTS, with one of these being the patient who had clot propagation. However, PTS was also diagnosed in 5.3% of those patients without DVT. In this study,128 all those with proximal DVT were treated with anticoagulants (i.e. 3 months of warfarin), whereas those with distal DVT received 3 months of aspirin. Therefore, the rate of PTS in untreated asymptomatic DVT following hip and knee surgery remains unknown.

van Dongen et al. 110 studied the relationship between inadequate anticoagulation with vitamin K antagonist (VKA) and the risk of PTS with symptomatic proximal DVT and found that patients who spend > 50% of their time beneath the international normalised ratio (INR) level of 2.0 are at a higher risk for PTS (OR 2.71, 95% CI 1.44 to 5.10). Based on this, one would expect a higher risk of PTS in patients with asymptomatic proximal DVT that is untreated than in those with treated asymptomatic proximal DVT. Although van Dongen et al. 110 did not include untreated patients in their study, they reported a fourfold increase (OR 3.69, 95% CI 1.29 to 10.53) for those patients spending > 90% of their time outside the target INR range, which suggests that the risks may be higher still in those patients not offered any anticoagulation.

The rate of PTS in patients with asymptomatic proximal DVTs reported by Schindler et al. 128 (32.5%) was similar to that reported in symptomatic proximal DVT by Hach-Wunderle et al. 109 (32.4%). This suggests that the risk of PTS is similar regardless of whether or not the proximal DVT is symptomatic, provided that it is treated with anticoagulants. However, given that patients who are asymptomatic in the model will not receive any anticoagulant treatment as they are not being proactively screened for asymptomatic DVT, a higher rate for asymptomatic proximal DVT was applied using the OR reported by van Dongen et al. 110 for good versus poor anticoagulation (OR 2.71 for those patients spending > 50% of their time outside the target INR). This gave a rate of PTS of 56.5% for untreated asymptomatic proximal DVT. It is recognised that the rate of PTS may be higher still for those receiving no anticoagulation at all, as discussed in the previous paragraph, but this may be partially offset by asymptomatic DVTs being less extensive and, therefore, less likely to cause PTS.

The rate of PTS in asymptomatic untreated distal DVT was set as equal to that for treated symptomatic distal DVT, namely 15.6% as reported by Hach-Wunderle et al. 109 Inflating this risk further by applying the OR from van Dongen et al. 110 did not seem appropriate, given that this would give a risk of 33.4% and the clinical experts believed that the rate for untreated distal DVT should be lower than for treated proximal DVT. Furthermore, it is noted that the rate from Hach-Wunderle et al. 109 of 15.6% is nearly twice the rate reported for treated asymptomatic distal DVT by Persson et al. ,127 which would support not inflating the rate further.

The timing of PTS within the model has been based on data reported by van Dongen et al. ,110 who reported the incidence of PTS at 0.5, 1, 2 and 4 years (Figure 11). The shape of the curve for incidence over time was compared with that reported by Prandoni et al. 129 and was found to be similar, although the magnitude better matched the upper 95% CI of the estimates of incidence reported by Prandoni et al. 129 The hazards based on the incidence reported by van Dongen et al. 110 were scaled to match the rate reported at 3 years by Hach-Wunderle et al. 109 for symptomatic proximal and distal DVT. As van Dongen et al. 110 reported no further incidence beyond 4 years, no new incidence of PTS from 4 years was assumed. The prevalence of PTS over time incorporated within the model is shown in Figure 12.

FIGURE 11.

Prevalence of PTS in years following proximal DVT from studies.

FIGURE 12.

Prevalence of PTS in the model for treated symptomatic and untreated asymptomatic DVTs.

Costs of prophylaxis

For thromboprophylaxis during lower-limb immobilisation, it was assumed that patients receive 6 weeks of treatment using LMWH at the dose recommended for long-term treatment after hip or knee replacement. In the base-case analysis, it was assumed that the lowest-cost formulation of LMWH would be used (dalteparin), but the impact of treatment with either fondaparinux or dabigatran was explored in sensitivity analyses as these represented the highest- and lowest-cost alternatives to LMWH. The dosages and drug costs are summarised in Table 18.

Menakaya et al. 115 estimated the costs of outpatient VTE prophylaxis for patients having lower-limb injuries based on 388 patients receiving prophylaxis who were recruited from 7048 patients attending a fracture clinic in England. The total cost for an average of 46 days of anticoagulant treatment was £107.54 for the self-administration of LMWH, with £46 of this being the drug cost (the authors assumed £1 per day). Therefore, the total cost of administration was £61.54. This covered clinical time for counselling and taking blood samples (including a 5-day platelet count), pharmacy dispensing and initial administration. For patients requiring district nurse administration (£23 per home visit), the average cost of administration across the treatment course was £1096.54 but this was required in only 4% of cases, giving an average cost of £102.94 across the cohort. Adjusting the number of doses from 46 to 42, which affects only the district nurse administration costs, and updating this from 2011/12 to 2016/17 prices (Curtis and Burns100), gives a total administration cost of £105.13. The total cost for the 6-week course of dalteparin, including administration costs, was £223.70.

For the sensitivity analysis on fondaparinux, the same administration costs as those required for LMWH were assumed. Patients initiating treatment with a DOAC were assumed to require one nurse-led telephone consultation at 10 days. The range tested in sensitivity analysis for the whole course of treatment was £84.23 for dabigatran to £355.66 for fondaparinux.

When assessing the cost-effectiveness of using RAMs to determine thromboprophylaxis, it is necessary to include the cost of the time spent by the clinician to determine the individual’s risk score. It was assumed that this would require 5 minutes of time by a hospital consultant, which was estimated to be £8.83 based on an hourly cost of £106. 100 This was applied to all patients in the RAM-based thromboprophylaxis strategies but to none of the patients in the treat-all and treat-none strategies.

Cost of treatment

Patients who have lower-limb immobilisation may later present with symptomatic DVT or PE, requiring diagnosis and treatment. The costs of VTE have been estimated for three broad categories of costs: health-care contacts, diagnostic costs and drug treatment costs. It was assumed that all patients presenting with symptomatic VTE would require an ED assessment, but that a proportion would present first to a general practitioner (GP) before attending an ED, and that only a proportion of those attending an ED would require admission. It was assumed that those patients requiring admission would also be likely to require ambulance transfer to the ED. The proportion of patients attending GPs prior to ED attendance and the proportion requiring admission (Table 19) were based on clinical expert opinion. Discussions with clinical experts were informed by the assumptions used in the model described in the 2017 draft update to the NICE guideline on managing VTE130 and the assumptions used in relevant NICE single technology appraisals (STAs). 131–139

Diagnostic tests that occur during the ED episode of care, such as electrocardiography monitoring and chest radiography, are not costed separately. However, scans occurring later as outpatient activity were costed separately as these diagnostic tests are ‘unbundled’ within the NHS reference cost data. 116 Assumptions regarding the use of diagnostic tests to assess patients with suspected DVT and PE were based on clinical expert opinion. The main costs not covered within the ED episode of care are assumed to be leg vein ultrasonography for patients with a suspected DVT and computerised tomography pulmonary angiography (CTPA) or ventilation/perfusion scans for patients with a suspected PE. A minority of patients (20%) with suspected PE are also assumed to require echocardiography. The diagnostic tests are summarised in Table 19. It was assumed that all patients presenting with PE or symptomatic DVT (proximal or distal) would be treated with anticoagulants.

Treatment for VTE was assumed to consist of 3 months of either a VKA (warfarin) or a DOAC. Patients being treated with a VKA would also require initial anticoagulation with LMWH until adequate oral anticoagulation is established. The proportion receiving DOACs versus VKA was informed by data from the PREFER-VTE study,123 which found that 40% of patients in countries where DOACs had been launched were receiving a DOAC as their initial treatment following DVT. The split between drugs in the LMWH class was informed by assumptions used in a previous appraisal. 140 The costs of administering treatment with LMWH are based on the costs estimated by Menakaya et al. 115 for initiating prophylaxis using LMWH but with a reduction in district nurse administration based on the shorter duration of treatment (7 vs. 42 days). Patients receiving VKA are assumed to require nine visits to an anticoagulation service in 3 months to monitor their INR. Patients initiating treatment with a DOAC are assumed to require one nurse-led telephone consultation at 10 days. All patients are assumed to require a consultant-led face-to-face visit at 3 months to assess the need for ongoing treatment. The overall cost, including monitoring, when averaged across the treatment strategies was £290.24 (Table 20).

The total cost for non-fatal PE was estimated to be £1788, which includes health-care contacts, diagnostic costs and 3 months of drug treatment for VTE. Fatal PE was assumed to incur the same costs for health-care contacts and diagnostics, but without the cost of VTE treatment, giving a cost of £1498. Symptomatic proximal and distal DVTs had a total cost of £688 and £560, respectively, with the higher cost of proximal DVTs driven by a greater likelihood of admission.

Cost of bleeding complications

A paper by Luengo-Fernandez et al. 117 on the acute and long-term costs of care after stroke in patients with atrial fibrillation (AF) was identified as a source of costs in those having ICH from the NICE appraisal of edoxaban tosylate (Lixiana^®; Daiichi Sankyo UK Ltd, Gerrards Cross, UK) for the treatment and secondary prevention of DVT and/or PE. 140 The paper117 provides estimates of acute (up to 90-day) and long-term costs (average annual costs up to 5 years post stroke) using data from the Oxford Vascular Study (OXVASC), which was a prospective study of 91,000 UK patients registered with general practices in Oxfordshire. The paper by Luengo-Fernandez et al. 117 reports costs for the subset of patients who experienced a stroke during the study and who had a history of AF prior to that index stroke (n = 153). The mean age of patients in this subset was 80 years, with 26% of patients having previous stroke and 20% of patients having pre-morbid warfarin use.

The costs were reported separately on the basis of disability level and whether the type of stroke was ischaemic, haemorrhagic or not known. Only 17 of the 191 strokes experienced by the 153 patients were haemorrhagic. The mean acute cost of fatal haemorrhagic strokes was £1592 (SD £1886; n = 8). This mean cost was substantially lower than the average for all haemorrhagic strokes, which was £10,683 (SD £12,885; n = 17). Given that these estimates were based on a small number of patients, the estimates for all stroke types were also considered. The cost of fatal strokes across all types was also lower, at £2680 (SD £2661; n = 40), than the average for all strokes (mean £10,413, SD £15,105; n = 191), suggesting that fatal strokes do incur a lower acute cost than non-fatal strokes. The cost of fatal haemorrhagic strokes was used in the base-case analysis for fatal intracranial bleeds. The cost was updated from 2008/9 prices to 2016/17 prices (using Curtis and Burns100), giving an acute cost of £1802 over 3 months.

The paper by Luengo-Fernandez et al. 117 also provided relevant data on the cost of non-fatal ICHs. The cost of a non-fatal haemorrhagic stroke was estimated to range from £3401 to £24,234, depending on the level of disability following the stroke. A weighted average cost for non-fatal haemorrhagic strokes of £18,764 was calculated across the non-disabling, moderately disabling and totally disabling haemorrhagic strokes. However, it was noted that these estimates were based on only nine non-fatal haemorrhagic strokes. The equivalent average across all non-fatal strokes was £12,461. The higher cost for haemorrhagic non-fatal strokes appeared reasonable given that haemorrhagic strokes appeared to result in higher than average costs across all categories except fatal strokes. This is supported by data presented by Fernando et al. ,141 which show that patients with haemorrhagic stroke who are admitted to an intensive care unit (ICU) incur higher costs than matched patients (matched for age, sex and comorbidity) admitted to an ICU for other (non-ICH) reasons. Furthermore, ICHs associated with oral anticoagulant use were also found to be significantly more costly than those not associated with oral anticoagulant use. 141 The weighted average cost for non-fatal haemorrhagic stroke from Luengo-Fernandez et al. 117 was used in the base-case analysis for the acute cost of non-fatal ICH. This cost was updated from 2008/9 prices to 2016/17 prices (using Curtis and Burns100), giving non-fatal ICH a cost of £21,245 over 3 months.

It was noted that this cost for non-fatal ICH is significantly higher than the weighted average cost of stroke based on NHS reference costs for non-elective long- and short-stay inpatients (including excess bed-days), which was £3290 [average of reference costs116 for Healthcare Resource Group (HRG) codes AA35A to AA35F]. However, the average NHS reference cost is similar to the cost reported by Luengo-Fernandez et al. 117 for non-disabling stroke costs. This may reflect the fact that rehabilitation costs are reported separately in NHS reference costs and these costs are between £332 and £402 per day for admitted rehabilitation. These rehabilitation costs would have been included in the hospitalisation costs reported by Luengo-Fernandez et al. 117 and would be much higher for moderately and totally disabling strokes, which may explain the higher costs in these groups than the NHS reference cost for admitted care in stroke patients, which excludes rehabilitation.

The model also required post-acute costs (post 90 days) for patients having non-fatal ICH. Luengo-Fernandez et al. 117 estimated the post-acute costs per annum relative to the year before stroke (baseline). The total post-acute costs were found to be non-significantly higher versus baseline costs at £804 (95% CI –£832 to £2440). However, the GP and emergency care components were significantly higher at £98 (95% CI £27 to £169) and £99 (95% CI £56 to £141), respectively. In addition, the paper separately reported long-term residential care costs for patients not previously admitted to residential care. They reported that 18% of strokes resulted in a new admission to long-term care (i.e. warden housing, residential care or nursing home care), with an average annual cost of £6880 across all patients surviving the acute period (90 days). Post-acute and long-term costs were not reported separately by stroke type (haemorrhagic vs. other) by Luengo-Fernandez et al. 117 The total costs of GP, emergency care and long-term care were averaged across all stroke types and uplifted to current prices (using inflation indices from Curtis and Burns100) to give a total cost per year for non-fatal ICH patients of £8013 in the post-acute (beyond 90 days) period. This was applied in the Markov phase of the model to patients in the non-fatal ICH health state. It was also applied pro rata to patients having non-fatal ICH more than 90 days before the end of the decision tree model.

The cost of non-fatal extracranial bleeds was estimated using average inpatient costs for GI bleeds, based on the approach taken in the NICE Technology Appraisal (TA) 354. 140 A weighted average was taken across bleeds requiring single, multiple or no intervention and across short- and long-stay non-elective inpatients (including excess bed-days). This gave an average cost of £1198 based on 2015/16 reference costs (HRG codes FZ38G to FZ38P). 116

Cost of post-thrombotic syndrome and chronic thromboembolic pulmonary hypertension

Several reviews of studies relating to PTS were examined to identify any cost papers cited. 125,142–145 None of the cost papers identified were specific to the UK NHS. Three examined costs from a US perspective (Caprini et al. ,146 MacDougall et al. 147 and Olin et al. 148), one examined costs from a Brazilian perspective (Ramacciotti et al. 149), one examined costs from a Swedish perspective (Bergqvist et al. 150) and one examined costs from a Canadian perspective (Guanella et al. 151). Although the paper by Bergqvist et al. 150 was from a European country, the data related to patients diagnosed between 1970 and 1985.

Several of the NICE STAs identified in Appendix 11 cited cost assumptions from Goodacre et al. 152 One cited the US study by Caprini et al. 146 and another cited costs from a 2001 abstract by Cohen et al. 153

It was assumed in the model that management of PTS would involve one first and one follow-up vascular surgery outpatient appointment in the first year after diagnosis and two follow-up GP appointments every year thereafter. This was based on clinical expert advice regarding current management of PTS in the UK and is consistent with the assumptions made by Goodacre et al. ,152 which have been applied in several previous NICE STAs. 131,132,140,154

A review by Grosse et al. 143 on the costs attributable to VTE in the USA identified one paper155 on the cost burden of CTEPH. Kirson et al. 155 compared costs for patients with CTEPH with those for controls (matched for age, sex, employee status and geographical region) in a cohort of privately insured US patients. They found that costs per month were US$2368 (p < 0.0001) higher in patients with CTEPH than in matched controls, but they also found higher rates of comorbidities, suggesting that some of the differences may not be directly attributable to CTEPH. Grosse et al. 143 commented that CTEPH patients are far more likely to have chronic heart failure or chronic pulmonary disease, and no risk-adjusted treatment cost estimates were identified in their review.

The cost of CTEPH was based on the approach taken in technology appraisal (TA) number 354 (i.e. TA354)140 in which pulmonary endarterectomy costs were applied based on reference costs for complex thoracic procedures. Updating this estimate by using current reference costs116 gave a cost of £7898. The Evidence Review Group’s preferred assumption for TA354140 was to use the cost of drugs from CG92,24 which estimated a cost of £1219 per 4 weeks, giving a cost of £17,942 per annum when uplifted to 2016/17 prices (using inflation indices from Curtis and Burns100). Data from Delcroix et al. 113 were used to estimate the proportion having surgical management and the proportion receiving medical management. The fact that some surgical patients receive bridging therapy, as reported by Delcroix et al. ,113 was taken into account by allowing them 4.6 months of medical therapy based on the average time to surgery for these patients. This increased the costs in the first year for patients having surgical management to £9890. It was assumed that no further costs are incurred in year 2 and beyond for those having surgical management.

Summary of methods used to identify studies reporting utility values relevant to the model

The searches for the systematic review of published cost-effectiveness studies included a filter to identify health-related quality of life (HRQoL) data. While sifting these studies, one study156 reporting HRQoL data in patients with lower-limb immobilisation was identified; this study by Arverud et al. ,156 reported EuroQol-5 Dimensions (EQ-5D)157 scores in patients who had an Achilles tendon rupture. The study reported mean EQ-5D scores at 1 year for patients who did and patients who did not experience DVT following surgical repair and immobilisation. The average EQ-5D score was lower in patients with DVT but the difference was not statistically significant (p = 0.6). A second study158 identified in the search reported the validation of a disease-specific HRQoL measure, Deep Venous Thrombosis Quality of Life (DVTQoL), in patients with DVT but this study reported only the correlation between the EQ-5D and the domains of the DVTQoL; utility values based on EQ-5D scores could not be extracted. No other studies reporting preference-based measures of utility in patients with lower-limb immobilisation were identified from the searches conducted for the economic review.

One published study by Wolowacz et al. 99 was identified from the literature review that examined the cost-effectiveness of thromboprophylaxis for a similar, but not directly relevant, population (elective knee replacement) as described in Results of cost-effectiveness review. This was supplemented by searches of the NICE website to identify models developed to inform NICE TAs and CGs, as these models would be expected to have a relevant setting and methodological approach and would not always be identified through database searches. The cost-effectiveness analyses identified are summarised in Appendix 11.

In addition, ad hoc searches for systematic reviews of utilities for relevant model states (e.g. PTS) were conducted using the Google Scholar search engine (Google Inc., Mountain View, CA, USA). This identified a published systematic review of HRQoL studies for VTE and related complications by Lubberts et al. 159 This review focused on studies with a minimum follow-up of 1 year in order to capture the long-term impact of complications such as PTS and CTEPH. Only two of the studies included by Lubberts et al. 159 reported a generic preference-based utility measure (EQ-5D). One of the two studies reporting EQ-5D was conducted in patients with subclavian DVT and is therefore less applicable to this population. The other study, by Haig et al. ,160 reported EQ-5D scores for patients with and without PTS following DVT.

In addition, the systematic review by Lubberts et al. 159 reports outcomes for 15 studies that used either the Short Form questionnaire-36 items (SF-36)161 or the Short Form questionnaire-12 items (SF-12). Data from these studies were meta-analysed to assess the difference in HRQoL between patients and members of the general population matched by country, age and sex. Lubberts et al. 159 report the meta-analysed decrement in SF-36/SF-12 mental component summary (MCS) and physical component summary (PCS) scores compared with general population norms (matched by country, age and sex) using the standardised mean difference (SMD) as a common currency across studies. These meta-analyses by Lubberts et al. 159 do not provide a preference-based utility measure that can be used directly in the model, but their outcomes are summarised below as they provide evidence as to whether or not each VTE-related outcome has an important impact on long-term HRQoL based on more than a single study. They are therefore useful in interpreting the utility estimates from individual studies identified from existing cost-effectiveness models.

Although Lubberts et al. 159 included only studies reporting outcomes more than 1 year after VTE, they also listed nine studies158,162–169 that were excluded for reporting outcomes at shorter follow-up points. Of these, one study by Hogg et al. 163 reported useful data (see Appendix 12); the remaining eight studies did not report utility values for relevant health states. 158,162,164–169

One additional HRQoL study170 was identified from ad hoc searches; Roberts et al. 170 examined the predictors of HRQoL for patients following DVT using the SF-36. This is also discussed below as it provides useful additional information on PTS as a determinant of HRQoL.

The sources of utility data identified from both the published models listed in Appendix 11 and the systematic review by Lubberts et al. 159 are summarised in Appendix 12.

Utility values in pulmonary embolism

Lubberts et al. 159 report that PE is associated with a significant reduction in SF-36 PCS score (SMD –0.30, 95% CI –0.45 to –0.14) but not SF-36 MCS score (SMD 0.14, 95% CI –0.01 to 0.30), relative to general population norms at time periods of > 1 year based on an analysis of two studies. 171,172 This suggests that PE has an important long-term impact on patients. Three studies173–175 reported EQ-5D outcomes for patients with PE, although two of these studies173,174 were published only in abstract form and the other study175 used Danish rather than UK population norms. One study used an alternative generic preference-based measure, the Short Form questionnaire-6 Dimensions (SF-6D). 163 In addition, two studies176,177 reported utility values for PE measured directly using either standard gamble or time trade-off (TTO). Data from the study abstract by Cohen et al. 173 were reported in additional detail in the company submission for TA354140,178 (Figure 13).

FIGURE 13.

Data from Cohen et al. ,173 as reported in the company submission for TA354. 140,178

The data from Cohen et al. 153 were selected for use in the model because these were measured using the EQ-5D, were sourced from a large registry database (PREFER-VTE) including patients recruited in the UK and the paper included estimates of the change in utility over time (see Figure 13). The data reported by Cohen et al. 173 demonstrated that utility values varied depending on the time since PE, with utility values appearing to stabilise at between 3 and 6 months. The average utility in patients with PE during the 6 months following PE was calculated using the data from Cohen et al. 173 This was compared with the utility value at 6 months in patients with DVT (who do not have a long-term utility decrement according to the review Lubberts et al. 159) to calculate the utility multiplier for PE, which was then applied to the age appropriate utility for patients without PE. A linear change in utility between the points measured was assumed, giving an average utility of 0.775 over the 6 months after PE, which is a 9% decrement relative to utility at 6 months post DVT. Given that there was some evidence of a reduction in HRQoL for patients with PE found by Lubberts et al. 159 at 1 year, it was decided to use the 6-month utility decrement for PE relative to DVT, when no long-term reduction was found, to estimate the long-term utility reduction for patients following PE. This gave a 5% reduction in utility beyond 6 months for a patient surviving PE beyond 6 months. Tavoly et al. 175 reported mean EQ-5D values of 0.8 in patients post PE versus 0.86 in age- and sex-matched general population controls, suggesting a 7% reduction in utility despite a median time since diagnosis of 3.8 years. This supports the assumption that there are some ongoing post-PE symptoms beyond 6 months.

Utility values in deep-vein thrombosis

The data sources for utility values in DVT were similar to those identified for PE, except that the study by Tavoly et al. 175 was not relevant because it examined only patients with PE. However, a similar study by Utne et al. 179 examining DVT outcomes from the same hospital registry was identified instead. One additional study by Arverud et al. 156 examining DVT versus no DVT after lower-limb immobilisation due to injury was also identified. This study found that DVT identified by screening (colour duplex sonography) 2 weeks post operatively was a non-statistically significant predictor of worse EQ-5D scores (0.918 vs. 0.906; p = 0.604). This may be expected given that the DVTs were detected by screening and, therefore, may have been asymptomatic. Utne et al. 179 found that EQ-5D score was statistically significantly lower for patients with DVT compared with age- and sex-matched buddy controls. However, these patients were recruited any time after DVT, and a significant proportion had PTS, so the reduction compared with controls may have been partly due to PTS symptoms, which are likely to be persistent beyond 6 months. In the first 6 months of the model, the data from Cohen et al. 173 were used to estimate the average utility, which was compared with the utility at 6 months. This gave a utility decrement of 4%, on average, during the first 6 months after DVT. Beyond 6 months, no utility decrement for patients without PTS was assumed, as the systematic review by Lubberts et al. 159 reported that there was no statistically significant difference in SF-36 PCS and MCS scores for DVT at time periods of > 1 year based on an analysis of nine studies. 172,180–187 Utne et al. 179 did not find a statistically significant difference in the odds of having impaired HRQoL, measured by EQ-5D, depending on whether the DVT was proximal or distal. In the absence of any other data on the difference in HRQoL impact of proximal versus distal DVT, the same utility loss was assumed during the first 6 months for symptomatic proximal DVT and symptomatic distal DVT.

Utility values in post-thrombotic syndrome

Lubberts et al. 159 reported that patients with PTS had significantly lower MCS and PCS scores than matched population norms (SMD for MCS score –0.27, 95% CI –0.43 to –0.11; SMD for PCS score –0.89, 95% CI –1.21 to –0.57), based on an analysis of seven studies. 180–182,184–186,188 Five studies reported utility values for patients with PTS,160,177,189–191 with two studies160,189 reporting EQ-5D scores and three studies177,190,191 using direct measures of utility such as standard gamble or TTO. The data reported by Haig et al. 160 and Enden et al. 189 were the 2-year and 5-year follow-up points from the catheter-directed thrombolysis versus standard treatment for acute iliofemoral deep vein thrombosis (CaVenT) study that compared catheter-directed thrombolysis (with 24 months of compression stockings) against standard treatment (6 months of anticoagulant and 24 months of compression stockings) for patients with high proximal DVT. The CaVenT study189 found no difference in EQ-5D scores between treatment arms at 6 months, 2 years or 5 years, but did find a statistically significant difference in HRQoL between those with PTS (n = 92) and those without PTS (n = 97) at 24 months and 5 years. At 6 months, the mean EQ-5D scores were lower for those with PTS but the difference was not statistically significant (p = 0.062).

The decrement measured at 2 years in the CaVenT study189 was applied to all patients with PTS in the Markov phase of the model. It was decided not to model a change in HRQoL for patients with PTS over time as some HRQoL decrement is likely to be present from the time of diagnosis of PTS (i.e. 6 months after DVT) and the HRQoL impact of PTS appears to remain apparent at 5 years. Therefore, a 10% decrement was applied based on the 24-month data from Enden et al. 189

Supportive evidence is provided in the study by Roberts et al. ,170 which reported that patients without PTS recovered their SF-36 PCS scores from below population norms by 6 months, whereas those with mild or moderate/severe PTS did not experience a similar recovery. This study170 also reported that the location of DVT (i.e. proximal or distal) was not an independent predictor of HRQoL, with only sex, comorbidities and PTS being significant predictors in the multivariate analysis.

As the CaVenT study189 reported the average utility decrement across all PTS, without stratifying by PTS severity, this value was applied in the model for all patients with PTS. This may overestimate the utility decrement if patients in the CaVenT study had more severe PTS than would be seen in the modelled population and severity was a predictor of utility decrement. However, in the study by Roberts et al. ,170 the coefficients for mild versus no PTS and for moderate/severe versus no PTS were similar in the regressions for SF-36 PCS and MCS scores, suggesting that both mild and moderate/severe PTS had similar implications for HRQoL. Furthermore, the review by Lubberts et al. 159 shows that the 5-year estimates from the CaVenT study reported by Haig et al. 160 are fairly consistent with the average across the 10 studies160,180–182,184–186,188,192,193 that reported outcomes for the disease-specific measure of HRQoL Venous Insufficiency Epidemiological and Economic Study – Quality of life (VEINES-QoL). A sensitivity analysis was conducted to see if halving the utility decrement for PTS had a significant impact on the model results. A sensitivity analysis was also conducted using the data from Lenert and Soetikno190 and estimating a weighted average utility across severe and mild/moderate PTS using the data from Hach-Wunderle et al. 109 to estimate the proportion of PTS that is severe (6%). This gave an average utility decrement of 2% rather than the 10% assumed in the base-case analysis.

Utility values in chronic thromboembolic pulmonary hypertension

Only one study included in the review by Lubberts et al. 159 reported outcomes for CTEPH, and this reported statistically significantly lower MCS (SMD –0.31, 95% CI –0.54 to –0.08) and PCS scores (SMD –1.72, 95% CI –1.95 to –1.48) than matched population norms. Six studies were identified as providing information on preference-based measures of HRQoL in patients with CTEPH. Three of these studies reported data based on the EQ-5D,194–196 a fourth reported data from an alternative generic preference-based measure, the SF-6D197 and a fifth reported data from a condition-specific preference-based measure, the Cambridge Pulmonary Hypertension Outcome Review (CAMPHOR). 198 Finally, one study reported values based on a standard gamble. 199 Only one of the papers reported values specifically for patients with CTEPH195 but this paper was restricted to those with resistant/recurrent or inoperable CTEPH (i.e. those in whom pulmonary endarterectomy had failed or was not possible). The paper by Ara and Brazier194 reported data for ‘other heart problems’ (i.e. not heart attack/angina). The study by Keogh et al. 197 only included patients with idiopathic pulmonary arterial hypertension (PAH) or PAH related to connective tissue disease. Two studies, by Olschewski et al. 196 and Meads et al. ,198 included patients with CTEPH and other forms of PAH but Meads et al. 198 reported utility outcomes separately for those with CTEPH. Meads et al. 198 also reported a statistically significant difference in utility between CTEPH and those with PAH associated with connective tissue disorder. There was some inconsistency in the estimates provided by Ghofrani et al. 195 and Meads et al. ,198 with higher baseline scores in the population with inoperable/recurrent resistant CTEPH (0.66 for placebo arm) in Ghofrani et al. 195 than in the CTEPH subgroup reported by Meads et al. 198 (0.56). This may be because patients accepted into trials have slightly better general health than those receiving treatment in general practice. It was decided to use the data from Meads et al. 198 in the model; this was calculated based on the difference between the average utility for patients with CTEPH (0.56) and the average utility for those with New York Heart Association (NYHA) class 1 (0.89), who would be expected to have minimal HRQoL impact. This gave a multiplier of 0.63 or, equivalently, a 37% decrement relative to patients who are well. The decrement was applied for a patient’s lifetime for those who were medically managed but only applied for the first year for those being surgically managed, who had the decrement for PE applied thereafter.

Utility values following intracranial haemorrhage

Of the 10 studies177,190,194,200–206 reporting utility values for stroke patients, six studies 194,200–204 measured utility using the EQ-5D. Another four studies177,190,205,206 reported utility values using a direct standard gamble or TTO measure, but these were disregarded as generic measures of HRQoL are preferred by NICE. 94 The values based on the EQ-5D ranged from 0.31 (as reported by Pickard et al. )203 to 0.70 (as reported by Lunde et al. ). 202 Luengo-Fernandez et al. 201 reported that the decrement ranged from 0.22 at 1 month to 0.09 at 5 years. Studies that reported multiple time points found that the absolute utility value increased and the decrement versus controls decreased over time. The study with the longest follow-up201 found that utility values were relatively stable from 6 months to 5 years post stroke. The data from the 5-year follow-up of the OXVASC study201 were chosen for application in the model as this was the largest sample of stroke patients, the population was selected from the UK, the duration of follow-up allowed time since stroke to be accounted for, and a comparison was made against general population norms. A decrement of 0.22 was assumed in the decision tree part of the model where time since stroke was < 6 months and a decrement of 0.09 was assumed in the long-term part of the model.

Utility values for major non-intracranial haemorrhage bleeds

For major non-ICH bleeds, no studies were identified that measured utility using the EQ-5D or another generic preference-based measure. Two studies176,206 reported values using the standard gamble and a third used TTO. 177 Two of the studies that recruited at least some patients with VTE reported a utility value of 0.65,176,177 whereas the study in patients with AF reported a much higher value of 0.841. 206 The study by Locadia et al. 177 (median 0.65, interquartile range 0.49 to 0.86) was considered to be the most relevant as it used TTO, which is the same method used to generate utility vales for the EQ-5D, and it explicitly included patients with experience of major bleeding episodes during VKA treatment in addition to some patients with VTE. However, using the data from Locadia et al. 177 directly would have resulted in a lower utility value for non-ICH bleeds than for ICH, which would not have had face validity. As the study by Locadia et al. 177 showed similar valuations for the PE and GI bleed health states, and these states were ranked as equivalent, it was decided to use the data for PE from Cohen et al. 173 in the model as the longer-term data provided by Cohen et al. 173 allows the utility multiplier to be calculated relative to a patient who has recovered from VTE (i.e. relative to a patient 6 months after a DVT). The utility multiplier was, therefore, estimated to be 0.80 (0.68/0.84). As major non-ICH bleeds are not expected to lead to long-term disability, the utility multiplier for major bleeding was applied for 28 days.

Treatment-related utility decrement

Given that thromboprophylaxis involves giving patients treatment even though they may not go on to experience a VTE event without treatment, it is important to include in the model the possibility that treatment itself is associated with harm.

Several sources of data were identified for the decrement associated with prophylaxis and treatment of VTE. Six studies reported the utility decrement for warfarin or other VKAs. Four of these studies177,205–207 used a direct measure of utility (e.g. TTO or standard gamble) and two of these studies173,208 measured utility using the EQ-5D. The study by Monz et al. 208 examined HRQoL in a subset of the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial, which was a randomised comparison of dabigatran and warfarin in patients with AF. Therefore, it was unable to measure the utility impact of warfarin alone. However, it did report that there was no statistically significant difference in HRQoL based on treatment type (DOAC vs. warfarin) and the utility scores observed were similar to age-comparable population norms. Cohen et al. 173 reported EQ-5D data by treatment type (heparin only, heparin/VKA, DOAC) from the PREFER-VTE registry. However, it was not clear whether or not patients receiving heparin alone in the PREFER-VTE registry would have different characteristics from those receiving other forms of treatment for VTE, in which case the difference in utility between heparin and other VTE treatments may have been confounded by patient characteristics and their influence on clinicians’ choice of anticoagulant.

Only two studies173,207 provided data on the disutility associated with heparin. Given the potential bias described above for Cohen et al. ,173 it was decided to use the data provided by Marchetti et al. ;207 in this study, a TTO exercise was used in 48 patients attending an anticoagulation clinic who had experience of LMWH. The trade-off was based on clinical vignettes for hypothetical patients receiving LMWH. The average number of days they were willing to trade was 2.7 days over a 1-year time frame. Therefore, a mean decrement of 0.007 was applied during thromboprophylaxis.

Four studies provided directly elicited measures of utility for warfarin therapy. Two were in patients with AF. 205,206 Marchetti et al. 207 was the only study conducted exclusively in patients with a history of VTE. Half of the patients recruited by Marchetti et al. 207 had previous experience of receiving warfarin. All of the patients in the study reported by Locadia et al. 177 had been treated previously with a VKA, but not all had previous VTE. The utility decrement for VKA versus no VKA was 4% for Locadia et al. 177 (0.92 vs. 0.96) based on median estimates of utility for both states, whereas it was 1% for Marchetti et al. 207 based on a willingness to trade 4 of 365 days of full health to avoid warfarin. The estimate from Marchetti et al. 207 is similar to the estimates from Gage et al. 205 in patients with AF, whereas the estimate from Locadia et al. 177 is close to the estimate from Robinson et al. 206 It was decided to use the data from Marchetti et al. 207 to provide an estimate consistent with the methodology used to estimate the utility decrement for LMWH.

In the sensitivity values exploring the use of DOACs as prophylaxis, no utility decrement has been assumed, as DOACs are taken orally and do not require frequent blood tests to monitor INR levels. In the sensitivity analysis exploring the use of fondaparinux as prophylaxis, the same utility decrement as for LMWH was assumed as both treatments are given by subcutaneous injection.

The utility values applied in the model are summarised in Tables 21 and 22.

Timing and duration of utility decrements applied in the decision tree

To calculate the QALYs gained by patients having different paths through the decision tree, it is necessary to make some assumptions regarding the timing of events, as these are not explicitly modelled in a decision tree. The following assumptions were made when estimating QALYs in the decision tree:

• Baseline utilities using general population utility values for the starting age are applied to those not having treatment and not having any clinical events (e.g. VTE, bleeds).

• Bleeds during lower-limb immobilisation are assumed to occur halfway through the immobilisation period, which is assumed to last 8 weeks (56 days), namely bleeds occur at 28 days.

• VTEs occurring during lower-limb immobilisation are assumed to be diagnosed at the end of immobilisation (i.e. at 56 days) and utility decrements for PE and DVT are applied from diagnosis until the end of the decision tree period.

• Bleeds occurring during treatment for VTE are assumed to occur at 13, 32 and 12 days post diagnosis of VTE for fatal, ICH and other major bleeds, respectively (based on data from RIETE, reported by Nieto et al. 105).

• Disutilities for ICH are applied lifelong, but separate disutilties are applied in decision tree and Markov phases of the model (i.e. before and after 6 months post lower-limb immobilisation).

• Disutility of GI bleeding is assumed to last a maximum of 28 days.

• Disutility of prophylaxis applies for the duration of prophylaxis (6 weeks) and is therefore less in those stopping early because of major bleeding.

• Disutility of treatment for VTE applies for the duration of treatment (3 months) and is therefore less in those stopping early because of major bleeding.

The following assumptions were made when estimating QALYs in the Markov model:

• Utility values for patents without any long-term sequelae (e.g. ICH, CTEPH, PTS) are taken from general population values and decrease as patients age in the model.

• All other utility values are applied as multipliers, such that the absolute utility value decreases for all patients as a result of ageing.

• Utility decrements continue in the Markov model for the remainder of a patient’s lifetime for PE but not for DVT, for which patients are assumed to return to general population utility values at 6 months.

• Patients with CTEPH who are treated medically have a lifelong utility decrement, whereas those treated surgically return after 1 year to the same utility as those surviving PE without CTEPH.

• Patients with PTS have the same utility decrement from diagnosis to death.

• Patients with ICH have the same utility decrement from 6 months (i.e. the start of Markov model) to death.

Clinical outcomes predicted by the model

The symptomatic adverse clinical outcomes predicted by the model at 6 months (based on mid-point parameter values) are summarised in Figures 14 and 15. As patients may have more than one adverse outcome during the 6-month time period of the decision tree model, patients are summarised on the basis of the worst outcome experienced, such that those recorded as having non-fatal PE or symptomatic DVT are those who did not also have fatal bleed, fatal PE or ICH. In Figure 14 it can be seen that prophylaxis for all is the strategy that minimises the total number of symptomatic adverse outcomes at 6 months. The numbers of the most serious adverse clinical outcomes (e.g. fatal bleeds, fatal PEs or ICHs) are similar across the prophylaxis strategies and the differences are too small to see based on Figure 14. Figure 15 focuses solely on these serious adverse clinical outcomes and shows that giving prophylaxis based on a L-TRiP(cast) score of ≥ 9 minimises the total number of these outcomes, although the rate is very low (≈1 in 4000) and shows little variation between strategies.

FIGURE 14.

Symptomatic clinical outcomes predicted at 6 months for 10,000 patients having lower-limb immobilisation due to injury.

FIGURE 15.

Serious adverse clinical outcomes predicted at 6 months for 10,000 patients having lower-limb immobilisation due to injury.

It should be noted that major bleeds that are non-fatal and not ICHs are assumed to resolve within 6 months and are, therefore, not included in Figure 14. The number of major bleeding outcomes is summarised in Figure 16, which shows that the total number of major bleeds, including fatal and non-fatal ICHs, increased from 4.03 per 10,000 having lower-limb immobilisation when no prophylaxis is given to 5.38 per 10,000 when prophylaxis is given to all. Giving prophylaxis to all patients with lower-limb immobilisation, therefore, results in fewer than two additional cases of major bleeding per 10,000 patients receiving prophylaxis. The increase in major bleeding is smaller than would be expected based solely on the RR of 1.64 for bleeding with prophylaxis versus bleeding without prophylaxis, because prophylaxis prevents VTE, thereby lowering the rate of bleeding experienced during VTE treatment.

FIGURE 16.

Major bleeding predicted at 6 months for 10,000 patients having lower-limb immobilisation due to injury.

The adverse clinical outcomes predicted by the model at 5 years (based on mid-point parameter values) are summarised in Figure 17. Only those outcomes resulting in long-term morbidity are included; therefore, patients who experience DVTs without any subsequent PTS are not included in the adverse clinical outcomes shown. It can be seen that the number of people surviving to 5 years is fairly similar across all the thromboprophylaxis strategies. The strategy of thromboprophylaxis for all results in the smallest number of people surviving with CTEPH, PTS or ongoing morbidity due to previous ICH or PE. The numbers surviving with either ongoing morbidity due to ICH or CTEPH following PE are relatively small, at ≈1 per 10,000 patients for each outcome.

FIGURE 17.

Adverse clinical outcomes predicted at 5 years for 10,000 patients having lower-limb immobilisation due to injury.

Cost-effectiveness of thromboprophylaxis for all compared with thromboprophylaxis for none

Figure 18 shows the estimates of incremental costs and QALYs for the comparison of thromboprophylaxis for all with thromboprophylaxis for none for 10,000 PSA samples. The mean incremental costs and QALYs from the 10,000 PSA samples are shown as an × on the cost-effectiveness plane. The mean QALY gain is 0.015, but it can be seen that there is a wide spread of incremental QALY estimates, with 95% of the incremental QALYs falling in the range of 0.004 to 0.029 but with some PSA samples (0.53%) having a negative QALY gain. The mean incremental cost is £203 with 95% of the PSA estimates falling in the range £172–245. The incremental cost-effectiveness ratio (ICER) for thromboprophylaxis for all compared with thromboprophylaxis for none based on the mean costs and QALYs is £13,524. Therefore, if a decision-maker is willing to pay £20,000 per QALY, it would be cost-effective to give thromboprophylaxis to all patients with lower-limb immobilisation due to injury if this was the only alternative to giving thromboprophylaxis to none. However, it can be seen that a significant proportion of the PSA estimates fall above the £20,000 per QALY threshold; therefore, there is a 24% probability that thromboprophylaxis for all is not cost-effective.

FIGURE 18.

Cost-effectiveness plane for the comparison of thromboprophylaxis for all vs. thromboprophylaxis for none.

Cost-effectiveness of using a risk assessment model to determine thromboprophylaxis

The results of all of the thromboprophylaxis strategies considered in the model (based on the mean outputs of the PSA for 10,000 samples) are summarised in Table 23 and Figure 19. It can be seen in Figure 19 that, generally, the QALYs increase as the proportion receiving prophylaxis increases across the decision tools, with the QALY gains becoming more marginal as the proportion receiving prophylaxis increases. Furthermore, it can be seen in Figure 19 that L-TRiP(cast)-7 is extendedly dominated, as the ICER to move from L-TRiP(cast)-8 to L-TRiP(cast)-6 would be lower than the ICER to move from L-TRiP(cast)-8 to L-TRiP(cast)-7, but more QALYs are gained by L-TRiP(cast)-6. However, it should be noted that the point for L-TRiP(cast)-7 is close to falling on the cost-effectiveness frontier [i.e. on the line between L-TRiP(cast)-8 and L-TRiP(cast)-6]. The incremental analysis for the remaining non-dominated strategies is shown in Table 24. The optimal strategy would be L-TRiP(cast)-9 if the decision-maker values QALYs at £20,000 and L-TRiP(cast)-8 if the decision-maker values QALYs at £30,000.

FIGURE 19.

Base-case results on CE plane (mean from 10,000 PSA samples).

Optimal balance of sensitivity and specificity for a risk assessment model

From the base-case results, it can be seen that as we move from higher to lower thresholds of the L-TRiP(cast) score, the QALYs gained increase as the sensitivity of the risk assessment tool increases but the cost increases as the specificity decreases. The ROC curve for the L-TRiP(cast) score provided by Nemeth et al. 32 allows us to explore where the optimal balance of sensitivity and specificity lies when assuming that it lies somewhere on the ROC curve provided by L-TRiP(cast). To do this, a linear regression on the logit scale was fitted to the points on the ROC plane from L-TRiP(cast)-6 to L-TRiP(cast)-10 (intercept = –1.5256, gradient = 0.8098, R² = 0.9926). The sensitivity and specificity values were then varied between the L-TRiP(cast)-9 and L-TRiP(cast)-8 points of the curve to maximise the INMB using the deterministic model that uses mid-point parameter estimates. It was identified that, when valuing a QALY at £20,000, the INMB is maximised for a sensitivity of 84% and a specificity of 55%, resulting in 48% of patients being treated. When valuing a QALY at £30,000, the INMB is maximised for a sensitivity of 89% and a specificity of 46%, resulting in 57% of patients being treated. Therefore, even when using an optimised balance between sensitivity and specificity on the ROC curve predicted by the data from Nemeth et al. ,32 between 43% and 52% of patients would not receive thromboprophylaxis. The incremental costs and QALYs for these optimal points are shown in Figure 20. It can be seen from this that the optimal points on the ROC curve are actually extendedly dominated by the points for LTRiP(cast)-9 and LTRiP(cast)-8. This is because the ROC curve predicted by the linear regression on the logit scale passes below the data points for LTRiP(cast) scores of 9 and 8.

FIGURE 20.

Cost-effectiveness plane showing optimal points on the ROC curve when valuing a QALY at either £20,000 or £30,000 (generated using mid-point parameter values).

Uncertainty surrounding base-case results

Figure 21 shows the cost-effectiveness plane for L-TRiP(cast)-9 (the optimal strategy when valuing QALYs at £20,000) versus no prophylaxis. It can be seen that there is considerable uncertainty in the size of the QALY gain but there is a low chance of the optimal strategy resulting in QALY losses, compared with no prophylaxis, with 95% of the incremental QALYs falling in the range of 0.005 to 0.024. In terms of costs savings, 95% of the incremental costs fell in the range of £69 to £107, suggesting that the optimal strategy is unlikely to result in cost savings compared with no prophylaxis. This is expected, as prophylaxis results in definite costs to those receiving it but the benefits of preventing VTE are small and uncertain when averaged over the whole population receiving prophylaxis. Similar distributions of incremental costs and QALYs were seen for the other L-TRiP(cast) cut-off scores, with all predicting additional costs and additional QALYs and a mean ICER of < £20,000 relative to a strategy of no prophylaxis.

FIGURE 21.

Cost-effectiveness plane for L-TRiP(cast)-9 showing incremental costs and QALYs relative to no prophylaxis.

Figure 22 is the CEAC, which shows the probability that each strategy is optimal when varying the amount that the decision-maker is willing to pay for a QALY. It can be seen that L-TRiP(cast)-9 has the highest probability (66.8%) of being optimal when valuing a QALY at £20,000, which is consistent with the results in Table 24. L-TRiP(cast)-8 has a slightly higher probability of being optimal when valuing a QALY at £30,000, but there is also a high probability of L-TRiP(cast)-9 remaining optimal. This makes sense given that the ICER for L-TRiP(cast)-8 versus L-TRiP(cast)-9 is only just under £30,000 per QALY. It can also be seen that L-TRiP(cast)-7 always has a low probability of being optimal regardless of what the decision-maker is willing to pay and that the decision-maker would need to be willing to pay more than £100,000 per QALY before prophylaxis for all is the optimal strategy. This is interesting because if the risk-stratifying decision tools had not been included in the analysis, then prophylaxis for all would have been cost-effective compared with prophylaxis for none. This demonstrates the potential value of using decision tools to target prophylaxis in a cost-effective manner instead of focusing on evaluating all or nothing prophylaxis options.

FIGURE 22.

Cost-effectiveness acceptability curve.

To demonstrate the impact of different strategies on a health service, L-TRiP(cast)-9 can be compared with a strategy of using prophylaxis for all across the English NHS. In Chapter 1, Description of the health problem, it was estimated that 70,000 people per year across the English NHS experience temporary lower-limb immobilisation due to injury as an outpatient. Using the mean outputs from the PSA, it is found that prophylaxis for all would result in an additional 0.0023 QALYs per person, but for an additional cost of £117.40 per person when compared with L-TRiP(cast)-9. Applied to the 70,000 annual population for the English NHS, the cost saving that could be achieved by giving prophylaxis to only those with a L-TRiP(cast) score of ≥ 9 would be £8.2M pounds and the QALYs lost would be 160. Conversely, if the impact of giving prophylaxis to patients at a L-TRiP(cast) score of 9 compared with treatment for none is considered, it is found that the additional cost to the NHS is £6.0M, with QALY gains of 891 for a population of 70,000 patients having lower-limb immobilisation. One-tenth of this additional cost to the NHS is the upfront cost of performing risk assessment in all patients with lower-limb immobilisation following injury to identify those requiring prophylaxis.

Value to the decision maker of reducing uncertainty in the estimates of cost-effectiveness

The global EVPI was estimated, which tells a decision-maker the value of reducing all current decision uncertainty associated with the PSA parameters. It was found that the overall EVPI was £4.12 per patient treated, suggesting that although treating patients at the L-TRiP(cast)-9 cut-off score would be expected to achieve an INMB of £168.96 per patient compared with no prophylaxis, based on current information, only 2.4% more INMB could be achieved with perfect information. This suggests that the value of reducing the decision uncertainty further is low on a per-patient basis. However, if the EVPI is estimated over 5 years, assuming that 70,000 patients have lower-limb immobilisation due to injury every year, then the overall discounted population EVPI is £1.3M.

Parameter EVPI analysis was also used to assess the importance of each individual parameter in generating uncertainty around the optimal strategy. Based on this, the most important parameters for decision uncertainty were the utility value for PTS, the efficacy of thromboprophylaxis in reducing VTE, the probability of PTS for patients with distal DVT and the utility decrement associated with taking prophylaxis. This suggests that the main factors driving uncertainty in the model are the efficacy and disutility of receiving prophylaxis, and the long-term complications of PTS. The incidence of PTS in patients with distal DVT is particularly important probably because asymptomatic DVT is more common than symptomatic DVT and the majority of asymptomatic DVTs are expected to be distal in this population. However, the magnitude of the overall EVPI suggests that a large-scale RCT comparing thromboprophylaxis with no thromboprophylaxis would be unlikely to be cost-effective.

It should be noted that EVPI analysis can capture only uncertainty related to parameters included in the PSA and will not capture any uncertainty associated with structural assumptions or the choice of data inputs used to estimate parameter uncertainty. These uncertainties are quantified instead using deterministic sensitivity analyses.

Important factors driving the cost-effectiveness estimates identified during model validation were:

• incidence, long-term costs and utility decrement for PTS

• baseline risks of VTE and the proportion of DVTs that are distal

• costs and efficacy of prophylaxis and prophylaxis-related disutility.

These have been explored in deterministic sensitivity analyses by examining changes in the outputs of the deterministic model (the values produced when assuming mid-point parameter inputs) when changing one or more selected parameter inputs.

Scenario analyses and threshold analyses

The results of the individual scenario analyses are summarised in Table 25 by reporting the optimal thromboprophylaxis strategy when valuing a QALY at £20,000 and £30,000 for each scenario.

Whether or not the age of the cohort being offered prophylaxis would alter the conclusions was examined, as it is likely that those receiving prophylaxis under the various decision tools would be older than the mean age of the whole group who have lower-limb immobilisation due to injury. It was found that increasing the mean starting age of the whole cohort to 70 years increased the ICERs but L-TRiP(cast)-9 remained optimal at £20,000 and L-TRiP(cast)-8 remained optimal at £30,000 with an ICER of £29,555 versus L-TRiP(cast)-9. This suggests that the decision to use a cohort approach based on average age is unlikely to have over- or underestimated prophylaxis benefits sufficiently to alter the key conclusions regarding the optimal strategy.

A threshold analysis was conducted to see how much the mid-point estimates of efficacy for LMWH in preventing VTE would need to change to alter the optimal prophylaxis strategies. It was found that a decrease in the hazard ratio (HR) (improvement in efficacy) from 0.54 to 0.44 would make L-TRiP(cast)-8 the optimal strategy at both £20,000 and £30,000 per QALY, whereas an increase in the HR (decrease in efficacy) to 0.60 would make L-TRiP(cast)-9 optimal at both £20,000 and £30,000 per QALY. An increase in the HR to 0.76 would be required for L-TRiP(cast)-9 to no longer be optimal at £20,000 per QALY. An increase in the HR to 0.84 would be required for treatment for none to be the optimal strategy at £20,000 per QALY.

It was examined whether or not the conclusions would differ if all patients were assumed to receive either fondaparinux or DOACs instead of LMWH for prophylaxis. For this, no difference in adverse event rates was assumed across the different prophylaxis options. Costs were assumed to vary in each case as per the costing analysis in Table 18. For DOACs, equivalent efficacy to LMWH was assumed and no utility decrement for DOACs was assumed because they are taken orally. For fondaparinux, the efficacy estimates from the NMA for fondaparinux versus control (OR 0.1, 95% predictive interval 0.05 to 0.3, from the random effects analysis) were used and the same utility decrement as applied for LMWH was assumed because both drugs are given by subcutaneous injection. Using DOACs shifted the optimal strategy when valuing a QALY at £20,000 from L-TRiP(cast)-9 to L-TRiP(cast)-8 and shifted the optimal strategy when valuing a QALY at £30,000 to treat all. Using fondaparinux resulted in the ICER for L-TRiP(cast)-8 versus L-TRiP(cast)-9 falling under £20,000 per QALY but L-TRiP(cast)-8 remained optimal when valuing a QALY at £30,000. It should be noted that these two scenarios are fairly crude and do not fully explore differences between the different drug classes in terms of adverse events. Furthermore, in the case of DOACs the analysis is based on an assumption of equivalent efficacy. The intention here was to explore whether or not the optimal cut-off point for providing thromboprophylaxis would differ when using a different drug class, rather than determining the optimal drug class to use for thromboprophylaxis.

In addition to a lack of precision in the estimates of utility decrement associated with PTS, there were differences in the estimates provided by different papers. The data from Lenert and Soetikno190 were used, which had lower decrements for PTS and showed that the optimal strategy when willing to pay £20,000 per QALY became L-TRiP(cast)-10, with L-TRiP(cast)-9 having an ICER of just under £30,000 in this scenario. This suggests that there is considerable decision uncertainty associated with the choice of utility data for PTS.

An assumption was made regarding the resource use associated with management of PTS based on clinical expert advice. A costing study from the USA146 reported higher costs, possibly related to more aggressive management of PTS. Whether or not using these higher cost estimates changed the conclusions was tested; it was found that the optimal strategies did not change, suggesting that the costs of managing PTS are not an important source of decision uncertainty.

It should also be noted that the PSA does not incorporate any uncertainty related to the estimates of sensitivity and specificity. Therefore, a scenario analysis was conducted in which, for each strategy, the specificity was fixed and the sensitivity was reduced so that it fell half-way between the observed value and the line on the ROC curve, which represents a tool that is no better than random selection. The alternative points are shown in Figure 23. The approximate AUC for the alternative points is 0.62, which is in the lower range of the CI for the AUC for the L-TRiP(cast) score is the plaster cast subgroup of the derivation cohort (AUC mean 0.76, 95% CI 0.66 to 0.86). 32 A similar AUC was achieved when L-TRiP(cast) score was validated in a subgroup of patients having plaster cast in the THE-VTE study (0.77, 95% CI 0.58 to 0.96) but a much better AUC was achieved when it was validated in the Milan study (0.95, 95% CI 0.91 to 0.99). 32 It can be seen in Figure 24 that, in this scenario assuming a lower AUC, the ICER for moving from each L-TRiP(cast) cut-off score to the next is fairly constant, such that the gains in QALYs when increasing the numbers treated are fairly proportional to the gains in cost across all L-TRiP(cast) cut-off scores. In this scenario, treatment for all is the optimal strategy when valuing a QALY at either £20,000 or £30,000. This tells us that the shape of the ROC curve is important in determining whether or not risk assessment tools can be used to target treatments more cost-effectively. If the tool used is less efficient at identifying the patients at risk, as indicated by a lower AUC, then a strategy of treating all patients would be most cost-effective. It was examined how much the sensitivity and specificity values could change before the RAM based thromboprophylaxis became extendedly dominated by treatment for all (i.e. when the ICER vs. no treatment was the same for both). It was found that for a specificity of 55%, the sensitivity would need to fall to 49% before thromboprophylaxis using a RAM became extendedly dominated by thromboprophylaxis for all. However, the absolute QALY gains would be 42% lower than the optimal point on the ROC curve and the costs would be only 16% lower.

FIGURE 23.

The ROC curve for the observed sensitivity and specificity values for L-TRiP(cast)-6 to L-TRiP(cast)-10 and the alternative values used in the sensitivity analysis.

FIGURE 24.

Cost-effectiveness plane when assuming that L-TRiP(cast) performs less well than observed by Nemeth et al. 32 (AUC of 0.62 instead).

Given that the use of a decision tool would result in clinical time being spent for every patient who has lower-limb immobilisation, it was important to evaluate whether or not the assumption regarding the time taken for risk assessment was associated with significant decision uncertainty. The time taken was changed from 5 minutes to 20 minutes and it was found that this had an impact on the ICER only for L-TRiP(cast)-10 versus no prophylaxis; as this remained < £20,000, L-TRiP(cast)-9 remained optimal.

Whether or not the exclusion of costs related to CRNMB had an important impact on the model conclusions was examined and it was found that the optimal strategies did not change when assuming that all patients with CRNMB attended ED for assessment.

It was initially planned to have a zero rate of PTS in patients with asymptomatic distal DVT based on clinical expectations that the rate of PTS would be low in this group. Given that we had identified that PTS following distal DVT was an important driver of uncertainty in the model, a scenario analysis was conducted with a zero rate of PTS for asymptomatic distal DVT. Under this assumption, L-TRiP(cast)-9 was still the optimal strategy when willing to pay £20,000 per QALY but L-TRiP(cast)-8 had an ICER of > £50,000 compared with L-TRiP(cast)-9. In addition, under this scenario, prophylaxis for all had an ICER of £26,962 versus prophylaxis for none. This is because the marginal gains of preventing VTE are smaller when PTS risk is limited to the minority with proximal DVTs or symptomatic distal DVTs.

There was some uncertainty regarding the rate of PTS in asymptomatic proximal DVT that remains undetected and untreated, as most of the studies that identified asymptomatic proximal DVT treated patients. It was examined whether or not the conclusions varied if assuming the same rate of PTS as observed in symptomatic proximal DVT but it was found that the optimal strategies were unchanged.

The estimates of the proportion of symptomatic DVTs that are proximal were based on 16 DVTs reported in a single study. 59 To determine if this factor was an important driver of cost-effectiveness, the proportion was varied from 0% to 100%; it was found that this had no impact on the optimal strategy when valuing a QALY at either £20,000 or £30,000.

Similarly, as the estimates of the proportion of asymptomatic DVTs that are distal was based on a small number of events, an extreme analysis was conducted to see if varying the proportion from 0% to 100% had a large impact on the conclusions. It was found that, when assuming that all asymptomatic DVTs are distal, the optimal strategy was L-TRiP(cast)-9 when valuing a QALY at either £20,000 or £30,000. Conversely, when assuming that all asymptomatic DVTs are proximal, the optimal strategy was L-TRiP(cast)-8 when using a willing-to-pay threshold of either £20,000 or £30,000 per QALY.

Given that the major risk of adverse effects to patients is the risk of a bleed following prophylaxis, a threshold analysis was conducted to see how much higher the bleed risk in the general population would need to be to change the optimal strategy. It was found that L-TRiP(cast)-9 was optimal at £20,000 per QALY until the bleed risk reached 30 times that assumed in the base-case analysis. However, a sixfold increase in bleed risk in the population was required to make L-TRiP(cast)-8 have an ICER of > £30,000 per QALY. Furthermore, when the bleed risk is 15-fold higher than assumed in the base-case analysis, the treat-all strategy is no longer cost-effective relative to the treat-none strategy at a threshold of £20,000 per QALY.

The risk of VTE without prophylaxis was increased and decreased threefold; it was found that this had a significant impact on the optimal strategy, with a threefold increase resulting in treating all being the optimal strategy when valuing a QALY at £30,000, and a threefold decrease resulting in treat none being optimal when valuing a QALY at £20,000. For a strategy of treat all to result in fewer QALYs gained than lost, namely for there to be a net harm to patients, the VTE risk would need to be < 9% of its base-case value, suggesting that baseline risks would need to be substantially lower than thought before prophylaxis would be at risk of doing more harm than good to a patient’s health. However, at this level of risk, none of the prophylaxis strategies would be cost-effective.

Pharmacological thromboprophylaxis for preventing venous thromboembolism

In the meta-analysis, it was found that thromboprophylaxis with LMWH reduced any VTE (OR 0.52, 95% CrI 0.37 to 0.71), clinically detected DVT (OR 0.40, 95% CrI 0.12 to 0.99) and PE (OR 0.17, 95% CrI 0.01 to 0.88) compared with no thromboprophylaxis. The effect of LMWH thromboprophylaxis varied between different VTE outcomes but all estimates were consistent with an approximate halving of risk. The number of deaths and bleeding events were too small to draw reliable conclusions. Thromboprophylaxis with fondaparinux also reduced any VTE (OR 0.13, 95% CrI 0.05 to 0.30) and clinically detected DVT (OR 0.10, 95% CrI 0.01 to 0.94) compared with no thromboprophylaxis, but the effect on PE was inconclusive (OR 0.47, 95% CrI 0.01 to 9.54). Event rates for symptomatic DVT and PE were generally low across the studies, so an approximate halving of risk may result in a small absolute risk reduction. Network meta-regression did not identify evidence of effect modification associated with population characteristics, type of injury, type of immobilisation or duration of treatment. There was evidence of difference in the effect of different types of LMWH used, with certoparin having the highest probability of the greatest effect on any VTE, but this was based on findings from one study. 52

Individual risk factors associated with venous thromboembolism risk

In the systematic review, it was found that only older age was consistently associated with increased risk of VTE across the included studies. Four studies53,60,70,71 showed that higher BMI was associated with increased risk and six studies11,50,52,53,69,70 show associations between injury type and risk of VTE. All studies were deemed to be at moderate23,50,52,53,60,68,71 or serious9,11,64–67,69,70 risk of bias overall following structured quality assessment. The use of individual risk factors to predict VTE risk in people with lower-limb immobilisation due to injury is, therefore, based on limited and weak evidence.

Risk assessment models for prediction of venous thromboembolism risk

The systematic review identified a number of RAMs but only two studies32,82 (both case–control design) provided estimates of sensitivity and specificity for three of the RAMs. The L-TRiP(cast) score32 using a threshold score of 8 (sensitivity 92.6%, specificity 39.7%) or 9 (sensitivity 80.8%, specificity 60.8%) may have appropriate accuracy for use in practice but these estimates are subject to considerable uncertainty.

Pharmacological thromboprophylaxis for preventing venous thromboembolism

The meta-analysis synthesised data from 6857 participants in 13 randomised trials that were judged to have a low risk of bias or some concerns only for most quality criteria. This represents a large, methodological robust data set across a variety of settings that allows us to draw generalisable conclusions. Reasonably precise estimates of the effect of thromboprophylaxis compared with control could be produced for most outcomes. Statistical power was more limited for uncommon outcomes, such as PE, but even here it was possible to generate an estimate that suggested a statistically significant effect. It was also possible to undertake a NMA to compare across three alternatives: (1) LMWH, (2) fondaparinux and (3) no thromboprophylaxis.

A recent Cochrane meta-analysis by Zee et al. 18 reported data from eight trials,49,51–56,59 including 3680 participants, that compared thromboprophylaxis with no treatment or placebo. They found that LMWH was associated with a significantly reduced risk of any DVT (OR 0.45, 95% CI 0.33 to 0.61) and symptomatic VTE (OR 0.40, 95% CI 0.21 to 0.76) but not PE (OR 0.50, 95% CI 0.17 to 1.47). Zee et al. 18 excluded four trials that were included in this analysis (Goel et al. ,50 Selby et al. ,58 Gehling et al. 23 and Samama et al. 57), and one additional trial was published after their updated meta-analysis (Zheng et al. 60). Two of the trials were excluded because they focused on operatively treated fractures rather than immobilisation (Goel et al. 50 and Selby et al. 58), one because the comparator was aspirin (Gehling et al. 23) and one because the intervention was fondaparinux rather than LMWH (Samama et al. 57). The inclusion of these trials has ensured that this analysis is more comprehensive and provides more precise estimates of effectiveness, but at the expense of greater heterogeneity.

This analysis was inevitably limited by the primary data. The variety of settings and patient groups may be a strength in terms of increasing generalisability; there was evidence of heterogeneity of effect for many of the outcomes, suggesting that effects may differ between populations. Three trials23,52,53 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. One trial that administered aspirin to the control group was included on the basis that NICE guidelines on VTE1,24 do not consider aspirin or other antiplatelet agents to be appropriate for VTE prophylaxis and aspirin is not indicated as a treatment for VTE prophylaxis in lower-limb immobilisation. 21,25,26 If aspirin has a significant prophylactic effect, then this trial may underestimate the effect of thromboprophylaxis.

The primary studies had a number of selection criteria that limit the ability to apply the findings to certain populations. Patients at a high risk of VTE (such as those with active cancer, thrombophilia, previous VTE or pregnancy/puerperium) and those with an increased risk of bleeding were often excluded. The studies generally included patients with rigid immobilisation rather than those with a degree of movement or a removable cast or splint, so the findings may be applicable only to those with full immobilisation.

The analysis included a substantial number of participants but the event rates for some outcomes were very low. As a consequence, it was not possible to produce reliable estimates of the effect of thromboprophylaxis on major bleeding or 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 at a higher risk of bleeding.

Individual risk factors associated with venous thromboembolism risk

This is the first systematic review conducted to look directly at individual risk factors increasing the risk of VTE in patients with lower-limb injury immobilised in plaster, discharged to an outpatient setting. This is an important distinction, as the population of interest differs markedly from generic thrombosis data sets; patients with lower-limb injury tend to be younger, more active and devoid of comorbidity than those presenting with unprovoked or cancer-associated thrombosis. Previous work has attempted to address a similar issue in patients undergoing elective foot and ankle surgery. 72,73,81 We consider this to be a different population because of the absence of acute blunt trauma causing local vessel/endothelial injury, expert postoperative immobilisation and often bespoke immobilisation regimes.

The review was undertaken in accordance with guidelines published by the Centre for Reviews and Dissemination75 and the protocol was registered with the PROSPERO register. 38 Clinical experts were involved throughout as checkers and to assess the validity and applicability of research during the project. Descriptive statistics were reported to provide plain insight into the limited evidence base applicable to the subject matter and the scientific concerns regarding validity of the data.

The review was limited by a lack of studies specifically examining risk prediction in lower-limb immobilisation due to injury. Therefore, studies, such as RCTs, that were designed for another purpose but reported risk predictors in their analyses were included. The studies of risk prediction were a combination of prospective cohorts and retrospective health database registries. Both have significant limitations. Retrospective studies of health database registries may have large numbers but may be limited by poor data quality and failure to accurately ascertain outcomes. Prospective cohorts may have better-quality data but lack statistical power because of smaller numbers. The included studies demonstrated high levels of heterogeneity, so a meta-analysis could not be undertaken.

Several studies included patients receiving operative intervention and short inpatient stays. Following the introduction of NICE guidance on thromboprophylaxis to reduce the risk of hospital-acquired thrombosis, it is reasonable to assume that, in a modern health-care environment, all these patients would receive some form of routine thromboprophylaxis. 24 For this reason, inclusion of these patients could lead to false reassurance regarding low incidence of VTE. However, patients with a short inpatient stay (< 5 days) were considered to fit within the scope of interest; these patients often remain temporarily immobilised for a period of 4–8 weeks in total, and debate persists about the role of thromboprophylaxis that is similar to debate around patients discharged directly from the ED.

All the studies within the review were classed as being at moderate or severe risk of bias. Therefore, any conclusions regarding the influence of risk factors on the subsequent development of VTE drawn are based on weak evidence and have the potential to be inaccurate.

Risk assessment models for prediction of venous thromboembolism risk

This review had strengths of following recognised guidance, prospective registration, a clear protocol and drawing on appropriate clinical and reviewing expertise. It was anticipated that the data would not allow meta-analysis; therefore, descriptive analyses only were planned.

A previous review by Watson et al. 82 identified two of the RAMS included in this review (Keenan et al. 31 and Roberts et al. 17), along with three generic RAMs that could be applied to people with lower-limb immobilisation due to injury. 210–212 Watson et al. ’s82 review then undertook primary analysis of these two RAMs to estimate sensitivity and specificity for predicting VTE. This review included these data and a number of other studies32,78–81 but only the study of Nemeth et al. 32 provided estimates of sensitivity and specificity.

The review is, therefore, limited by the paucity of data. Furthermore, case–control designs are known to be associated with a risk of bias,213 so sensitivity and specificity may have been overestimated. The study by Watson et al. 82 was relatively small (21 cases and 21 controls), so estimates of sensitivity and specificity are imprecise. The study by Nemeth et al. 32 involved validation using two data sets but each only contained a small number of patients with lower-limb immobilisation (32 cases, 7 controls, and 143 cases, 8 controls, respectively).