NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
National Guideline Centre (UK). Venous thromboembolism in over 16s: Reducing the risk of hospital-acquired deep vein thrombosis or pulmonary embolism. London: National Institute for Health and Care Excellence (NICE); 2018 Mar. (NICE Guideline, No. 89.)
December 2019: In recommendation 1.3.5 the British Standards for anti-embolism hosiery were updated because BS 6612 and BS 7672 have been withdrawn. August 2019: Recommendation 1.12.11 (1.5.30 in this document) was amended to clarify when anti-embolism stockings can be used for VTE prophylaxis for people with spinal injury.
Venous thromboembolism in over 16s: Reducing the risk of hospital-acquired deep vein thrombosis or pulmonary embolism.
Show details27.1. Introduction
Elective knee replacement surgery involves a large number of patients per year, with an increasing application in younger age groups. The general risks of this surgery, including infection, are well documented.
An objection to using pharmacological VTE prophylaxis is the increased risk of bleeding as a result of anticoagulation. The benefit of VTE prophylaxis has to be weighed against the risks and consequences of a post-operative bleed.
27.2. Review question: What is the effectiveness of different pharmacological and mechanical prophylaxis strategies (alone or in combination) for people undergoing elective knee replacement surgery?
For full details see review protocol in appendix C.
Table 80
PICO characteristics of review question.
27.3. Clinical evidence
Twenty-eight studies were included in this evidence review, these are summarised in Table 81 below.
Fourteen studies were previously included in the previous guideline (CG92) 17 ,34 ,64 ,66 ,88 ,93 ,95 ,105 ,106 ,191 ,192 ,233 ,310 ,318 and fourteen studies were added in the update 4 ,52 ,53 ,180 ,188 ,189 ,216 104 ,252 ,300 ,330 31 186 151.
Two technology appraisals were previously included in the previous guideline; 228 229. These technology appraisals 229; evaluated evidence identified in the update 252 300 and evidence included in the CG92 88 180.
Six studies that were previously included in CG92, have been excluded from this evidence review due to incorrect interventions and incorrect comparisons 125 141 ,144 ,194 ,209 ,315.
Three Cochrane reviews 139 98 261 were identified which looked at continuous passive motion, heparin and vitamin K antagonists for the prevention of venous thromboembolism people undergoing elective knee replacement. The reviews included studies which were included in the previous guideline (CG92) and this current update.
Evidence from these studies is summarised in the clinical evidence summary tables below (Table 82, Table 83, Table 84, Table 85, Table 86, Table 87, Table 88, Table 89, Table 90, Table 91, Table 92, Table 93, Table 94, Table 95, Table 96, Table 97, Table 98, Table 99, Table 100, Table 101, Table 102, Table 103, Table 104, Table 105, Table 106, Table 107, Table 108, Table 109, Table 110, Table 111, Table 112, Table 113, Table 114, Table 115, Table 116, Table 117 and Table 118). See also the study selection flow chart in appendix E, forest plots in appendix L, study evidence tables in appendix H, GRADE tables in appendix K and excluded studies list in appendix N.
In order to input the clinical effectiveness data of multiple possible interventions into the economic model, it was proposed that a network meta-analysis be carried out on the outcome data for DVT, PE and major bleeding. For full details on the NMA methodology and results, please see appendix M.
Table 81
Summary of studies included in the review.
Table 82
Clinical evidence summary: LMWH (standard dose; standard duration) versus no prophylaxis.
Table 83
Clinical evidence summary: LMWH (standard dose; standard duration) versus apixaban.
Table 84
Clinical evidence summary: LMWH (standard dose; standard duration) versus dabigatran.
Table 85
Clinical evidence summary: LMWH (standard dose; standard duration) versus rivaroxaban.
Table 86
Clinical evidence summary: LMWH (standard dose; standard duration) versus aspirin.
Table 87
Clinical evidence summary: LMWH (standard dose; standard duration) versus AES.
Table 88
Clinical evidence summary: LMWH (standard dose; standard duration) versus IPCD.
Table 89
Clinical evidence summary: LMWH (standard dose; standard duration) versus foot pump + AES.
Table 90
Clinical evidence summary: LMWH (standard dose; standard duration) + AES versus foot pump + AES.
Table 91
Clinical evidence summary: LMWH (standard dose; standard duration) versus UFH.
Table 92
Clinical evidence summary: LMWH (standard dose; standard duration) + AES versus UFH + AES.
Table 93
Clinical evidence summary: LMWH (standard dose; extended duration) versus LMWH (standard dose; standard duration).
Table 94
Clinical evidence summary: LMWH (standard dose; standard duration) + AES versus LMWH (low dose; standard duration) + AES.
Table 95
Clinical evidence summary: LMWH (standard dose; standard duration) + AES versus AES.
Table 96
Clinical evidence summary: LMWH (standard dose; standard duration) versus LMWH (low dose; standard duration).
Table 97
Clinical evidence summary: LMWH (standard dose; standard duration) + CPM versus CPM.
Table 98
Clinical evidence summary: LMWH (low dose; standard duration) versus no pharmacological prophylaxis.
Table 99
Clinical evidence summary: LMWH (low dose; standard duration) + AES versus AES.
Table 100
Clinical evidence summary: LMWH (high dose; standard duration) versus no prophylaxis.
Table 101
Clinical evidence summary: LMWH (high dose; standard duration) versus UFH.
Table 102
Clinical evidence summary: LMWH (high dose; standard duration) versus VKA.
Table 103
Clinical evidence summary: LMWH (high dose; standard duration) versus fondaparinux.
Table 104
Clinical evidence summary: LMWH (high dose; standard duration) + AES versus fondaparinux + AES.
Table 105
Clinical evidence summary: LMWH (high dose; standard duration) versus apixaban.
Table 106
Clinical evidence summary: LMWH (high dose; standard duration) versus dabigatran.
Table 107
Clinical evidence summary: LMWH (high dose; standard duration) versus rivaroxaban.
Table 108
Clinical evidence summary: Fondaparinux versus no pharmacological prophylaxis.
Table 109
Clinical evidence summary: Fondaparinux + AES versus AES.
Table 110
Clinical evidence summary: Fondaparinux + IPCD + AES versus VKA + IPCD + AES.
Table 111
Clinical evidence summary: Apixaban versus VKA.
Table 112
Clinical evidence summary: Dabigatran versus no prophylaxis.
Table 113
Clinical evidence summary: Rivaroxaban versus aspirin.
Table 114
Clinical evidence summary: Foot pump versus no prophylaxis.
Table 115
Clinical evidence summary: AES versus no prophylaxis.
Table 116
Clinical evidence summary: IPCD versus no prophylaxis.
Table 117
Clinical evidence summary: IPCD versus AES.
Table 118
Clinical evidence summary: CPM versus no prophylaxis.
27.4. Economic evidence
Published literature
Thirty economic studies, in 32 publications, relating to this review question were identified but were excluded due limited applicability, methodological limitations, a combination of limited applicability and methodological limitations or the availability of more applicable evidence.10 ,32,33,39,47,75,79,80,117,119,126,128,197,207,208,214,224,226,228–230 ,246 ,253–255 ,259 ,281 ,282 ,305 ,320 ,321 ,329 Of these, 10 publications were previously included in CG46.10 ,32 ,33 ,68 ,70 ,79 ,119 ,126 ,197 ,253 They also included 3 NICE TAs, 2 evidence review group [ERG] reports and the CG92 model for standard duration and post discharge prophylaxis. All excluded studies are listed in appendix O, with reasons for exclusion given.
See also the health economic study selection flow chart in appendix F.
New cost-effectiveness analysis
The committee considered the available evidence of cost effectiveness of prophylaxis strategies for people admitted for elective knee replacement (eTKR). The original guideline (CG92) model was considered but it was considered that it required updating given the availability of more recent trial data and the exclusion of the some of the older studies that were included in the CG92 NMAs from the current updated NMAs. The original model also included some interventions that are not routinely used in current practice including high doses of aspirin, VKA and UFH. The committee also discussed that since the publication of CG92, three TAs covering the use of DOACs in this population have also been published the latest in 2012.228–230 It was agreed that it would be more convenient for clinicians to be able to consult a single source for recommendation regarding the most cost-effective prophylaxis strategy for this population. This would also help in standardising current practice. Moreover, as the size of the population covered by this review question is very large; which means that changes to more costly prophylaxis options would lead to substantial resource implications, the committee agreed that this question should be prioritised for economic modelling. Hence, the original economic model presented here sought to address the question about the cost-effectiveness of different VTE prophylaxis strategies (alone or in combination) in people admitted for eTKR. A summary of the analysis is presented below and a full description can be found in appendix P in the full guideline.
Model overview
A cost-utility analysis was undertaken in Microsoft Excel® where costs and quality-adjusted life years (QALYs) were considered from a UK NHS and personal social services (PSS) perspective. A Markov model was constructed in order to estimate the costs and QALYs associated with different VTE prophylaxis strategies. Both costs and QALYs were discounted at a rate of 3.5% per annum in line with NICE methodological guidance231 Uncertainty was explored through probabilistic and sensitivity analyses. The time horizon considered was lifetime.
Population
The population entering the model are adults who are admitted to hospital for an eTKR. The cohort characteristics were based on the data reported in the National Joint Registry 13th annual report;36 which represented data collected up to December 2015 in England, Wales, Northern Ireland and the Isle of Man. The mean age of this population was 69.3 years and 44% were male.
Comparators
Thirteen prophylaxis strategies were selected for inclusion based on the availability of evidence from the clinical review, direct and network meta-analyses (N)MAs and discussion with the committee around which regimens are considered to be relevant to current clinical practice in the UK. These were:
- LMWH (std,std) + AEs
- Fondaparinux+ AES
- Foot pump + AES
- Foot pump
- AES
- LMWH (std,std)
- LMWH (std,extd)
- Aspirin
- Dabigatran
- Apixaban
- Rivaroxaban
- No prophylaxis
Model structure
The model consists of a simple decision tree covering the acute phase from admission up to 90 days post-operatively, to cover the period included in the definition of hospital-acquired VTE, followed by a Markov chain for the remaining model time horizon. The structure is repeated for each prophylaxis strategy.
The acute phase of the model is represented by a decision tree consisting of the primary clinical events: DVT (symptomatic proximal, symptomatic distal, asymptomatic proximal and asymptomatic distal), non-fata PE, fatal PE, Surgical site bleeding, non-surgical site bleeding (gastrointestinal (GI) bleeding, intracranial haemorrhage (ICH)/haemorrhagic stroke, other major bleeding), fatal major bleeding (MB), clinically-relevant non-major bleeding (CRNMB) and heparin-induced thrombocytopaenia (HIT). The structure of the decision tree is presented in Figure 4.
The long-term part is represented by a Markov cohort model. Individuals enter the model in one of the following states; based on where they end up at the end of the 90 days post-operatively: Well, post-symptomatic proximal DVT, post-symptomatic distal DVT, post-asymptomatic proximal DVT, post-asymptomatic distal DVT, post-PE, amputated post-HIT, disabled post-stroke, post-revision for infection. In the first two years, individuals in a post-VTE state can develop post-thrombotic syndrome (PTS). Those in the post-PE state can also develop chronic thromboembolic pulmonary hypertension (CTEPH). Transitioning to death is allowed from any state in the model, to represent allcause mortality. The structure of the Markov cohort model is illustrated in Figure 5.
Model inputs
The relative effects of treatments on the baseline transition probabilities were derived from clinical evidence identified in the systematic review undertaken for the guideline, the results of the NMA and supplemented by additional data sources as required. Health utility data were obtained from the literature. Cost inputs were obtained from recognized national sources such as the drug tariff, NHS reference costs and Personal Social Services Research Unit (PSSRU) publications. All inputs and assumptions made were validated by the committee.
Sensitivity analysis
A probabilistic analysis was carried out whereby distributions were assigned to model inputs in order to account for the uncertainty around the point estimates of these inputs and capture the effect of this uncertainty on model outputs. Additionally, a number of one-way sensitivity analyses were conducted whereby, for each analysis one key model input was changed in order to explore the sensitivity of model results to changes in that parameter (Table 119).
Table 119
One-way sensitivity analyses.
Results
Base case
The results of the base case analysis are presented in Table 120 and Figure 6. These show that the most effective intervention in terms of QALYs-gained is foot pump, with mean discounted QALYs per patient of 9.814 (95% CI: 7.86 to 11.58) over life-time time horizon followed by aspirin, with mean discounted QALYs per patient of 9.809 (95% CI: 7.86 to 11.58) and LMWH (standard dose, standard duration)+AES with a mean of 9.807 (95% CI: 7.86 to 11.58) over life-time time horizon. The least effective was dabigatran; with 9.71 QALYs (95% CI: 7.53 to 11.56). Aspirin had the lowest mean total cost of £187 (95% CI: £118 to £304) followed by foot pump with a mean total cost of £219 (95% CI: £119 to £473) and rivaroxaban with a mean total cost of £256 (95% CI: £82 to £1,205). The highest mean total cost was seen for fondaparinux + AES; with mean total cost of £904 (95% CI: £358 to £3,016).
The incremental net monetary benefit (INMB) vs the comparator (LMWH [standard, dose, standard duration]+ AES) was calculated at a cost-effectiveness threshold of £20,000 per QALY gained. Based on the INMB, the most cost-effective strategy (the one with the highest INMB) was found to be foot pump; with mean INMB of £353 (95% CI: −£101 to £665); with 18% probability of being the most cost-effective. It was followed by aspirin, with mean INMB of £281 (95% CI: −£195 to £703), then foot pump + AES (mean INMB £72 [95% CI: −£379 to £343]).
The full ranking based on the mean INMB of each strategy; together with the 95% confidence intervals that were calculated probabilistically, are presented in Table 120. This shows that there is considerable uncertainty in relation to the ranking of these interventions; with wide and overlapping 95% CIs. Based on the rank of the INMB; all interventions except dabigatran were more cost-effective than no prophylaxis. Foot pump and IPCD were more cost-effective compared to AES in this population.
Of the DOACs included in the model; rivaroxaban dominated both apixaban and dabigatran. However, the model comparator (LMWH [standard dose, standard duration]+AES) was more cost-effective compared to rivaroxaban (ICER: £7,686).
Sensitivity analysis
In all the SAs undertaken, the most cost effective option (foot pump) and the ranking of all interventions remained largely the same.
Table 120
Probabilistic base case analysis results for elective total knee replacement (eTKR) population.
Discussion
Interpretation and limitations
The results of this analysis reflect the very large uncertainty seen in the eTKR NMAs and in particular the uncertainty in the PE NMA which appeared to be driving the results of the economic model. This has been reflected in the very small differences in QALYs gained, the very wide 95% CIs around the ranks and the fact that the optimal intervention (foot pump) only had 18% probability of being the most cost-effective option. On average, however, the results seem to support the conclusion that VTE prophylaxis is cost-effective compared to no prophylaxis. However, the choice of a prophylaxis strategy is not clear cut. This is likely to be the result of the uncertainty around the relative effectiveness estimates for the different strategies.
Nevertheless, based on the results of this economic model; low intensity and lower cost strategies appeared to be more cost-effective for individuals undergoing eTKR, which might be the result of the lower risk of symptomatic DVT and PE in this population compared to the eTHR population. This has been reflected in the most cost-effective options being foot pump, aspirin and a combination of foot pump and AEs. Of the DOACs included in the model; rivaroxaban dominated both apixaban and dabigatran. This was in line with the results of the economic models assessed as part of TA170 and TA245 and a more recent analysis funded by the NIHR.229 ,230 ,281 Of the mechanical prophylaxis options considered in the analysis, foot pumps and IPCD were more cost-effective compared to AES. This supported the clinical experience that AES are not a practical option in this population.
Similar to the eTHR population, the model was an update of the CG92 model; so we attempted to address the limitations of that model which were highlighted by the orthopaedic surgeons’ community in a number of publications. One limitation was the use of relative effectiveness from the DVT NMA for the PE outcomes; where we used the PE NMA for all the interventions for which PE data were available to avoid making this assumption unless absolutely necessary; where the strategy was not included in the PE network. However, we have verified this assumption with the committee and externally validated it using the observational data analysis that used NJR data;152 where the ratio of the relative effectiveness of LMWH vs aspirin for the DVT outcome was found to be approximately the same as for the PE outcome.
Another issue was the lack of differentiation between proximal and distal DVT. We have addressed this issue by differentiating between the proximal and distal DVT for both symptomatic and asymptomatic events. We also allowed for different probabilities of progressing from each of these DVT events to PTS; to acknowledge the fact that progression from treated and untreated DVT to PTS would be different. We emphasised the fact that asymptomatic DVT also does not have an impact on costs and outcomes in the short term as it is not diagnosed in practice and its only consequence in the model is its future progression to PTS.
There was also a concern regarding the baseline risk used in the model which was based on data from the “no prophylaxis” arm in the RCTs. This was not considered to be reflective of current incidence of VTE with some trials dating back to the 70s, especially as practice has changed in terms of encouraging early mobilisation as well as the difference in surgical techniques. Based on this, we have used a strategy consisting of LMWH (standard dose, standard duration)+AES as our model comparator and obtained its baseline risk data from observational cohort studies that used the UK NJR data (see model write-up appendix P).152
However, this updated model may have some limitations. Due to lack of data on either DVT or PE outcomes for some strategies, an assumption still had to be made about the equivalence of relative effectiveness on the DVT and PE outcomes for these strategies. However, we have limited this only to instances where data was available for one of these outcomes but not for the other. This assumption may have affected the results. The relative effectiveness of foot pump, aspirin and foot pump + AES in relation to the PE outcome was assumed to be the same as their relative effectiveness obtained from the DVT NMA. This has resulted in a much lower PE rate for these interventions compared to all the others.
Another limitation of this analysis is that the relative safety of aspirin compared to LMWH was based on an observational cohort analysis based on NJR data. 152 This was due to the lack of any randomised controlled trials that report major bleeding outcomes for aspirin in these populations. However, as the data for MB from trials are likely to be imprecisely estimated due to the rarity of these events, it was considered that the use of observational data would be appropriate.
Generalisability to other populations/settings
This analysis has been undertaken from a UK NHS and PSS perspective; hence its results might not be generalisable beyond these settings. The population modelled also represents a cohort whose characteristics might be different from eTKR cohorts in other settings.
Conclusions
27.5. Evidence statements
Clinical
Pairwise meta-analysis statements
Pharmacological interventions versus pharmacological interventions
LMWH (standard dose; standard duration)
LMWH at a standard dose for a standard duration was compared with no prophylaxis, the outcomes DVT (symptomatic and asymptomatic), PE, major bleeding, wound haematoma, technical complications of mechanical interventions (examples given were skin rash, swelling above the appliance, pressure necrosis of the skin and peroneal nerve palsy) and wound infection were reported in one study. Moderate quality evidence showed clinical benefit of LMWH in terms of DVT (symptomatic and asymptomatic). Very low quality evidence suggested possible clinical benefit of LMWH in terms of PE and wound infection; however the uncertainty around this result was also consistent with both no difference or clinical harm. There was possible clinical harm of LMWH in terms of wound haematoma and no clinical difference in terms of major bleeding and technical complications of mechanical interventions, however there was also considerable uncertainty around these results. The quality of the evidence ranged from very low to moderate due to risk of bias, imprecision, indirectness and inconsistency.
LMWH at a standard dose for a standard duration was compared with apixaban, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, fatal PE, clinically relevant non-major bleeding and wound haematoma were reported in one study. There was possible clinical benefit of LMWH in terms of all-cause mortality, PE, fatal PE and wound haematoma. However the uncertainty around these results also related to no difference and clinical harm. Moderate quality evidence showed clinical harm of LMWH in terms of DVT (symptomatic and asymptomatic). There was possible clinical harm of LMWH in terms of major bleeding and clinically relevant non-major bleeding, although these outcomes also had serious uncertainty. The quality of the evidence ranged from very low to moderate due to risk of bias and imprecision.
LMWH at a standard dose for a standard duration was compared with dabigatran, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, fatal PE and clinically relevant non-major bleeding were reported across two studies. High quality, precise evidence showed no clinical difference between LMWH and dabigatran for DVT. There was a suggestion of clinical harm of LMWH in terms of fatal PE and no clinical difference in terms of all-cause mortality, PE, major bleeding and clinically relevant non-major bleeding, although all of these results were associated with considerable uncertainty. The quality of the evidence ranged from low to high due to imprecision. The outcome with evidence of high quality of was DVT (symptomatic and asymptomatic).
LMWH at a standard dose for a standard duration was compared with rivaroxaban, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, clinically relevant non-major bleeding and wound infection were reported across two studies. There was clinical harm of LMWH in terms of DVT (symptomatic and asymptomatic). There was, possible clinical harm of LMWH in terms of all-cause mortality, PE and wound infection, although these findings could also have been consistent with no difference. There was no clinical difference in terms of major bleeding and clinically relevant non-major bleeding, however the uncertainty around the bleeding results were also consistent with both benefit and harm. The quality of the evidence ranged from very low to moderate due to risk of bias and imprecision.
LMWH at a standard dose for a standard duration was compared with aspirin, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. There was no clinical difference between the two interventions for both of the outcomes reported, although there was very serious imprecision around both results. The quality of the evidence was very low due to risk of bias, imprecision and indirectness.
LMWH at a standard dose for a standard duration was compared with UFH, the outcome wound haematoma was reported in one study. There was possible clinical benefit of LMWH in terms of this outcome of wound haematoma, however the uncertainty around this result was also consistent with no difference and clinical harm. The quality of the evidence was very low due to risk of bias and imprecision.
LMWH at a standard dose for a standard duration was compared with LMWH at a low dose for a standard duration, the outcome major bleeding was reported in one study. There was possible clinical harm of LMWH at a standard dose in regards to this outcome, however there was very serious uncertainty around the result. The quality of the evidence of the evidence was very low due to risk of bias and imprecision.
LMWH (standard dose; extended duration)
LMWH at a standard dose for an extended duration was compared with LMWH at a standard dose for a standard duration, the outcomes DVT (symptomatic and asymptomatic), PE, major bleeding and heparin-induced thrombocytopaenia were reported in one study. There was possible clinical benefit of LMWH for an extended duration in terms of PE and major bleeding. There was no clinical difference in terms of DVT (symptomatic and asymptomatic) and heparin-induced thrombocytopaenia. However for all four outcomes the results were considerably uncertain and could be associated with harm, no difference and benefit. The quality of the evidence was low due to imprecision.
LMWH (low dose; standard duration)
LMWH at a low dose for a standard duration was compared with no prophylaxis, the outcome was major bleeding was reported in one study. There was possible clinical harm of LMWH in terms of major bleeding, however this result was uncertain and could also be consistent with no difference. The quality of evidence was very low due to risk of bias and imprecision.
LMWH (high dose; standard duration)
LMWH at a high dose for a standard duration was compared with no prophylaxis, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic) and major bleeding were reported in one study. High quality evidence showed clinical benefit of LMWH in terms of DVT (symptomatic and asymptomatic). Low quality evidence suggested possible clinical benefit of LMWH in terms of major bleeding and no clinical difference in terms of all-cause mortality, however there was uncertainty around both of these results. The quality of evidence ranged from low to high due to imprecision.
LMWH at a high dose for a standard duration was compared with UFH, the outcomes DVT (symptomatic and asymptomatic), PE and major bleeding were reported in one study. There was possible clinical benefit of LMWH in terms of DVT (symptomatic and asymptomatic) and PE. There was no clinical difference in terms of major bleeding. However all three of these outcomes were associated with a high level of uncertainty. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
LMWH at a high dose for a standard duration was compared with VKA, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, fatal PE, wound haematoma and wound infection were reported across three studies. There was possible clinical benefit of LMWH in terms of all-cause mortality, DVT (symptomatic and asymptomatic), major bleeding and wound infection, although these results were uncertain. There was no clinical difference in terms of PE, fatal PE and wound haematoma, however these results were also uncertain. The quality of evidence ranged from very low to moderate due to risk of bias and imprecision.
LMWH at a high dose for a standard duration was compared with fondaparinux, the outcome major bleeding was reported in one study. Low quality, precise evidence showed clinical benefit of LMWH in terms of this outcome. The quality of the evidence was low due to risk of bias and indirectness.
LMWH at a high dose for a standard duration was compared with apixaban, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, fatal PE, clinically relevant non-major bleeding and wound infection were reported across two studies. There was possible clinical benefit of LMWH in terms of fatal PE and wound infection. There was possible clinical harm of LMWH in terms of all-cause mortality, major bleeding and clinically relevant non-major bleeding. There was no clinical difference in terms of DVT (symptomatic and asymptomatic) and PE. There was considerable uncertainty around all of the outcomes for this comparison. The quality of the evidence ranged from low to moderate due to imprecision and inconsistency.
LMWH at a high dose for a standard duration was compared with dabigatran, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding and clinically relevant non-major bleeding were reported in one study. There was possible clinical benefit of LMWH in terms of all-cause mortality, possible clinical harm in terms of major bleeding and no clinical difference in terms of major bleeding and PE. There was considerable uncertainty around all of the outcomes for this comparison. The quality of evidence ranged from very low to moderate due to risk of bias and imprecision.
LMWH at a high dose for a standard duration was compared with rivaroxaban, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, clinically relevant non-major bleeding and wound infection were reported in one study. There was possible clinical benefit of LMWH in terms of all-cause mortality and major bleeding. There was possible clinical harm of LMWH in terms of DVT (symptomatic and asymptomatic) and PE. There was no clinical difference in terms of clinically relevant non-major bleeding and wound infection. There was considerable uncertainty around all of the outcomes for this comparison. The quality of evidence ranged from very low to moderate due to risk of bias and imprecision.
Fondaparinux
Fondaparinux was compared with no prophylaxis, the outcome major bleeding was reported in one study. There was no clinical difference for this outcome; however the quality of the evidence was very low due to risk of bias and very serious imprecision around the effect estimate, meaning the result could also be consistent with clinical benefit or harm.
Apixaban
Apixaban was compared with VKA, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding, fatal PE and wound infection were reported one study. Moderate quality evidence showed clinical benefit of apixaban in terms of DVT (symptomatic and asymptomatic). There was possible clinical harm of apixaban in terms of all-cause mortality, major bleeding and fatal PE, however these results may also be consistent with no difference and clinical benefit as they were so uncertain. There was no clinical difference in terms of PE and wound infection. These results were similarly uncertain. The quality of the evidence ranged from very low to moderate due to risk of bias and imprecision.
Dabigatran
Dabigatran was compared with no prophylaxis, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE, major bleeding and clinically relevant non-major bleeding were reported in one study. Moderate quality, precise evidence showed clinical benefit of dabigatran in terms of DVT (symptomatic and asymptomatic). Low quality evidence suggested possible clinical benefit of dabigatran in terms of clinically relevant non-major bleeding, possible clinical harm of dabigatran in terms of major bleeding, and no clinical difference in terms of all-cause mortality and PE. There was considerable uncertainty around these results. The quality of evidence ranged from low to moderate due to risk of bias and imprecision.
Rivaroxaban
Rivaroxaban was compared with aspirin, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. High quality evidence showed clinical benefit of rivaroxaban in terms of DVT (symptomatic and asymptomatic). Very low quality evidence suggested no clinical difference in terms of PE, however this was uncertain. The quality of the evidence ranged from very low to high due to risk of bias, indirectness and imprecision. The outcome with evidence of high quality was DVT (symptomatic and asymptomatic).
Pharmacological interventions versus mechanical interventions
LMWH at a standard dose for a standard duration was compared with AES, the outcomes DVT (symptomatic and asymptomatic), PE, technical complications of mechanical interventions (examples given were skin rash, swelling above the appliance, pressure necrosis of the skin and peroneal nerve palsy) and wound infection in one study. There was possible clinical benefit of LMWH in terms of DVT (symptomatic and asymptomatic), PE and wound infection. There was no clinical difference in terms of technical complications of the mechanical intervention. The evidence for all four of these outcomes exhibited considerable uncertainty. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
LMWH at a standard dose for a standard duration was compared with IPCD, the outcomes DVT (symptomatic and asymptomatic), PE, technical complications of mechanical interventions (examples given were skin rash, swelling above the appliance, pressure necrosis of the skin and peroneal nerve palsy) and wound infection in one study. There was possible clinical benefit of LMWH in terms of DVT (symptomatic and asymptomatic) and wound infection. There was no clinical difference in terms of PE and technical complications of the mechanical intervention. The evidence for all four of these outcomes exhibited considerable uncertainty. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
Combination interventions versus single interventions
LMWH at a standard dose for a standard duration was compared with foot pump in combination with AES, the outcomes DVT (symptomatic and asymptomatic) and fatal PE were reported in one study. There was possible clinical benefit of LMWH for both outcomes, however the DVT outcome was also consistent with no difference, and the fatal PE outcome with both no difference and clinical harm. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
LMWH at a standard dose for a standard duration in combination with AES was compared with AES, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. There was possible clinical benefit of LMWH in combination with AES in terms of DVT (symptomatic and asymptomatic) and no clinical difference in terms of PE, however there was uncertainty associated with both of these results. The quality of the evidence ranged from very low to low due to risk of bias and imprecision.
LMWH at a standard dose for a standard duration in combination with CPM was compared with CPM, the outcomes DVT (symptomatic and asymptomatic), PE and major bleeding were reported in one study. There was possible clinical benefit in terms of DVT (symptomatic and asymptomatic). There was no clinical difference in terms of PE and major bleeding. All three outcomes has considerable uncertainty associated with them. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
LMWH at a low dose for a standard duration in combination with AES was compared with AES, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. There was possible clinical benefit of LMWH in combination with AES in terms of DVT (symptomatic and asymptomatic), although this finding was also consistent with no difference. And no clinical difference was suggested in terms of PE, although this finding was very uncertain and could also be consistent with benefit and harm. The quality of the evidence ranged from very low to low due to risk of bias and imprecision.
Fondaparinux in combination with AES was compared with AES, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE and major bleeding were reported in one study. High quality evidence showed clinical benefit of fondaparinux in combination with AES in terms of DVT (symptomatic and asymptomatic). There was no clinical difference in terms of all-cause mortality, PE and major bleeding. However the findings for these three outcomes were also consistent with benefit and harm. The quality of the evidence ranged from very low to high due to risk of bias and imprecision. The outcome with evidence of high quality was DVT (symptomatic and asymptomatic).
Combination interventions versus combination interventions
LMWH (standard dose; standard duration) + AES
LMWH at a standard dose for a standard duration in combination with AES was compared with UFH in combination with AES, the outcomes DVT (symptomatic and asymptomatic), PE and wound infection were reported in one study. There was possible clinical benefit of LMWH in combination with AES in terms of wound infection and no clinical difference in terms of DVT (symptomatic and asymptomatic) and PE. However all three of these outcomes were associated with considerable uncertainty. The quality of the evidence was very low due to risk of bias and imprecision.
LMWH at a standard dose for a standard duration in combination with AES was compared with foot pump in combination with AES, the outcomes DVT (symptomatic and asymptomatic) and fatal PE were reported in one study. There was possible clinical benefit of LMWH in combination with AES in terms of fatal PE, although this finding was very uncertain. There was no clinical difference suggested for DVT (symptomatic and asymptomatic), however the uncertainty around this result was also consistent with clinical benefit. The quality of evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
LMWH at a standard dose for a standard duration in combination with AES was compared with LMWH at a low dose for a standard duration in combination with AES, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. There was no clinical difference for both of these outcomes, although there was considerable uncertainty associated with both. The quality of evidence ranged from very low to low due to risk of bias and imprecision.
LMWH (high dose; standard duration) + AES
LMWH at a standard dose for a standard duration in combination with AES was compared with fondaparinux in combination with AES, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic), PE and fatal PE were reported in one study. There was possible clinical harm of LMWH in combination with AES in terms of all-cause mortality, DVT (symptomatic and asymptomatic) and PE. However there was uncertainty around these results. There was no clinical difference in terms of fatal PE. The quality of the evidence ranged from very low to low due to risk of bias and imprecision and indirectness.
Fondaparinux + IPCD + AES
Fondaparinux in combination with IPCD and AES was compared with VKA in combination with IPCD and AES, the outcomes all-cause mortality, DVT (symptomatic and asymptomatic) and PE were reported in one study. There was no clinical difference for all the outcomes; although all outcomes were very uncertain. The quality of the evidence was very low due to risk of bias and imprecision.
Mechanical interventions versus mechanical interventions
Foot pump
Foot pump was compared with no prophylaxis, the outcomes DVT (symptomatic and asymptomatic) and PE were reported in one study. Moderate quality evidence showed clinical benefit of foot pump in terms of DVT (symptomatic and asymptomatic) and very low quality evidence suggested no clinical difference in terms of PE. There was uncertainty around the PE result. The quality of the evidence ranged from very low to moderate due to risk of bias, indirectness and imprecision.
AES
AES was compared with no prophylaxis, the outcomes DVT (symptomatic and asymptomatic), PE, major bleeding, technical complications of mechanical interventions and wound infections were reported in one study. There was possible clinical benefit of AES in terms of DVT (symptomatic and asymptomatic). There was no clinical difference in terms of PE, major bleeding, technical complications of mechanical interventions and wound infection. There was considerable uncertainty around the effect estimates for all five outcomes. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
IPCD
IPCD was compared with no prophylaxis, the outcomes DVT (symptomatic and asymptomatic), PE, major bleeding, technical complications of mechanical interventions and wound infections were reported in one study. There was possible clinical benefit of IPCD in terms of DVT (symptomatic and asymptomatic), PE and wound infection, and no clinical difference in terms major bleeding and technical complications of mechanical interventions. However these results were all very uncertain. The quality of the evidence ranged from very low to low due to risk of bias, indirectness and imprecision.
IPCD was compared with AES, the outcomes DVT (symptomatic and asymptomatic), PE, major bleeding, technical complications of mechanical interventions and wound infections were reported in one study. There was possible clinical benefit of AES in terms of DVT (symptomatic and asymptomatic), PE and wound infection, and no clinical difference in terms major bleeding and technical complications of mechanical interventions. However there was considerable uncertainty around all these results. The quality of the evidence was very low due to risk of bias, indirectness and imprecision.
Continuous passive motion
Continuous passive motion compared with no prophylaxis, the outcome DVT was reported in one study. There was no clinical difference for this outcome, however it was associated with considerable uncertainty. The quality of the evidence was very low due to risk of bias, indirectness and imprecision.
Network meta-analysis statements
DVT (symptomatic and asymptomatic)
23 studies were included in the network meta-analysis (NMA) for the outcome of DVT (symptomatic and asymptomatic), involving 19 treatments. Treatments included no VTE prophylaxis, pharmacological and mechanical interventions as single agents as well as combination interventions of both pharmacological and mechanical interventions. Results from the network meta-analysis presented rivaroxaban, apixaban and LMWH at a high dose for a standard duration as the most clinically effective interventions in terms of DVT (symptomatic and asymptomatic). The least clinically effective interventions were no prophylaxis, AES (length unspecified) and LMWH at a high dose for a standard duration in combination with AES (length unspecified). Three inconsistencies were identified when relative risk values from pairwise meta-analyses were compared with relative risk values from the NMA. There was also a fair amount of uncertainty around the rank-point estimates with very wide credible intervals.
PE
12 studies were included in the NMA for the outcome of PE, involving 13 treatments. Treatments included no VTE prophylaxis, pharmacological and mechanical interventions as single agents as well as combination interventions of both pharmacological and mechanical interventions. Results from the network meta-analysis presented LMWH at a standard dose for an extended duration, rivaroxaban and IPCD (length unspecified) as the most clinically effective interventions in terms of the outcome of PE. The least clinically effective interventions were UFH, LMWH at a standard dose for standard duration in combination with AES and no prophylaxis. No inconsistencies were identified when relative risk values from pairwise meta-analyses were compared with relative risk values from the NMA. There was also a high amount of uncertainty around the rank-point estimates with very wide credible intervals.
Major bleeding
19 studies were included in the NMA for the outcome of major bleeding, involving 11 treatments. Treatments included no VTE prophylaxis and pharmacological interventions (mechanical interventions were combined with no prophylaxis as the assumption was made that these interventions do not contribute to bleeding risk). Results from the network meta-analysis presented LMWH at a low dose for a standard duration, LMWH at a standard dose for an extended duration, and VKA as the most clinically effective interventions in terms of the outcome of major bleeding. The least clinically effective interventions were fondaparinux, rivaroxaban and LMWH at a standard dose for a standard duration. No inconsistencies were identified when relative risk values from pairwise meta-analyses were compared with relative risk values from the NMA. There was also a high amount of uncertainty around the rank-point estimates with very wide credible intervals.
Economic
- One original cost-utility analysis found that, in people admitted for elective knee replacement surgery, the following interventions were cost-effective (having positive incremental net monetary benefit [INMB]) compared to LMWH (standard dose, standard duration) +AEs: Foot pump (INMB £353), aspirin (INMB £281), foot pump+ AES (INMB £72). This analysis was assessed as directly applicable with potentially serious limitations.
27.6. Recommendations and link to evidence
| Recommendations |
|
| Research recommendation | None |
| Relative values of different outcomes |
The committee considered all-cause mortality (up to 90 days from hospital discharge), deep vein thrombosis (symptomatic and asymptomatic) (7–90 days from hospital discharge), pulmonary embolism (7–90 days from hospital discharge), fatal PE (7–90 days from hospital discharge), major bleeding (up to 45 days from hospital discharge) and surgical site haematoma (up to 45 days from hospital discharge) as critical outcomes. The committee considered clinically relevant non-major bleeding (up to 45 days from hospital discharge), health-related quality of life (up to 90 days from hospital discharge), heparin-induced thrombocytopaenia (duration of study), and technical complications of mechanical interventions (duration of study) and infection (duration of study) as important outcomes. Three network meta-analyses were conducted for this population, evaluating the outcomes DVT (symptomatic and asymptomatic), PE and major bleeding across numerous interventions. Please see section 4.4.3 in the methods chapter for further detail on prioritisation of the critical outcomes. |
| Quality of the clinical evidence |
Evidence from direct pairwise comparisons was included in the network meta-analyses for the elective knee replacement population. The quality of the pairwise comparisons ranged from very low to high due to risk of bias, imprecision, indirectness and inconsistency. The DVT (symptomatic and asymptomatic) network evaluated 19 interventions, the PE network evaluated 13 interventions and major bleeding network evaluated 11 interventions. Inconsistencies were identified in the DVT (symptomatic and asymptomatic) and PE networks between the direct pairwise evidence and the NMA evidence but there was good calibration for all the outcomes with small differences between the residual deviance and DIC values for the network meta-analysis models that were ran. Very wide credible intervals around the median network meta-analyses values present some uncertainty around the NMA results, particularly for the PE and major bleeding networks. |
| Trade-off between clinical benefits and harms |
The clinical evidence presented to the committee and orthopaedic subgroup informed the economic model that was developed. The committee’s discussions on the clinical evidence guided the recommendations alongside discussions on the results of the economic model. The model evaluated cost effectiveness using clinical data from the network-meta analyses undertaken on the committee-specified critical outcomes of DVT (symptomatic and asymptomatic), PE, and major bleeding. The model also captured data from the included trials on additional outcomes such as symptomatic DVT and asymptomatic DVT, and more detailed bleeding outcomes such as surgical site bleeding, gastrointestinal bleeding, and wound haematoma. When assessing the results of the analysis of the clinical data, the committee noted the wide credible intervals presented in the network meta-analyses, particularly in the PE and major bleeding networks, and that the uncertainty in the clinical data would have a knock on effect for the certainty in the results of the economic modelling. The licenced DOACs (rivaroxaban, apixaban and dabigatran) ranked highly when considering the clinical data for DVT, with rivaroxaban and apixaban ranked as the top two interventions having relatively narrow credible intervals. Based on the point estimates in the ranking, rivaroxaban (for 14 days) also outperformed dabigatran and apixaban in the analysis of the clinical data for PE. However there was considerable overlap of the confidence intervals for all of the DOACs due to the large uncertainty around the ranking results. None of the DOACs performed as well with respect to major bleeding. The committee and orthopaedic subgroup noted that the network meta-analyses suggest that combination prophylaxis may not be highly beneficial, but acknowledged that there is a lot of uncertainty as indicated by the wide credible intervals. The orthopaedic subgroup discussed that the use of AES is common within clinical practice in the eTKR population, without any presence of clinical benefit, with AES showing low rankings in preventing VTE outcomes (DVT and PE). It was discussed whether mechanical prophylaxis may be used due to pharmacological contraindications, and if clinicians might consider IPCD as the intervention of choice as there is a suggested clinical benefit of these interventions in terms of DVT (symptomatic and asymptomatic) and PE, with some uncertainty. The ranking for foot pump based on the clinical data was relatively high in the DVT NMA but it was discussed that the study which influenced the rank of this intervention was conducted during a time period when clinical practice was very different. Foot pumps are not commonly used by people undergoing elective knee replacement surgery for very long in the post-operative period due to the fact that this device can limit early mobilisation. The inclusion of aspirin and LMWH combined with anti-embolism stockings (until discharge) in the recommendation was primarily based on the results from the economic model (see ‘Trade-off between net clinical effects and costs’ section for further discussion). The duration of the interventions were based on the durations presented in the relevant clinical trials. |
| Trade-off between net clinical effects and costs |
An original economic model was developed to assess the cost effectiveness of the prophylaxis options included in the clinical review NMAs. It models the outcomes from the NMAs and also differentiates between asymptomatic and symptomatic DVT. This takes into account that asymptomatic DVT does not have an impact on costs and outcomes in the short term as it is not diagnosed in practice and its only consequence in the model is its future progression to PTS. Thirteen options were included in this model:
The model results showed that the most cost-effective option for this population is foot pump. This intervention had the highest mean incremental net monetary benefit (INMB) per patient compared to LMWH (standard dose, standard duration) + anti-embolism stockings (£353) at a cost-effectiveness threshold of £20,000 per QALY gained. It was followed by aspirin with INMB of £281. Compared to no prophylaxis, all options ranked higher, except dabigatran. A number of sensitivity analyses were presented to the committee including changing the cost-effectiveness threshold to £30,000 per QALY gained; changing the discount rate for costs and QALYs to 1.5%; and using the licensed duration where applicable rather than the average RCT duration. The committee and the orthopaedic subgroup considered the results of the model and noted that there was considerable uncertainty in this analysis which is likely to be the result of the uncertainty in the NMAs that informed the model; particularly the PE NMA, where the results were very imprecise. However, the results overall suggested that low-intensity, single-component and low-cost interventions are the most likely to be cost-effective in this population, with foot pump and aspirin ranking first and second. This was thought to be a result of the lower PE and symptomatic DVT incidence in the modelled cohort for the eTKR population compared to the eTHR population. The committee and the orthopaedic subgroup noted that despite being the most cost-effective option, foot pump had a low probability of being the most cost-effective (18%). This further emphasised the fact that considerable uncertainty exists in the analysis, which was also reflected in the very wide 95% CIs around the mean ranks. Hence, the committee opted to give a choice of prophylaxis options, noting that some people may have contraindications. The committee and the orthopaedic subgroup noted that out of the three DOACs included in the model (rivaroxaban, apixaban and dabigatran), rivaroxaban was dominant (more effective and less costly) compared to both apixaban and dabigatran. The committee noted that this was in line with previously published economic evaluations, the economic models assessed as part of TA170 and TA245 and a more recent analysis funded by the NIHR.229 ,230 ,281 Dabigatran was also, on average, worse than no prophylaxis. The orthopaedic subgroup also noted recent reports of increased risk of wound complications and subsequent increased length of hospital stay when using dabigatran.35 The committee noted that despite being dominated and having low INMB, apixaban had high probability of being the most cost-effective (43%). However, there was higher uncertainty around its cost-effectiveness; with around 5% probability of being the worst (compared to 0% for rivaroxaban) and 95% CI around its mean rank of 1 to 13 (compared to 1 to 11 for rivaroxaban). Hence, the committee recommended rivaroxaban as the most cost-effective DOAC with the aim of standardising practice to minimise costs and reduce errors. Apixaban and dabigatran already have current technology appraisal guidance associated with them and are, therefore, also recommended. However, as both were not cost effective compared to rivaroxaban, the committee decided that these options could only be considered if all the three recommended options are not suitable for the person (for example due to contraindications or issues related to patient preference). For those with contraindications for pharmacological prophylaxis, the committee and the orthopaedic subgroup considered that foot pump/IPCD appeared to be more cost-effective in this population compared to AES. This was in line with the evidence from other populations where AES tended to be less effective than previously thought. The committee also noted the difficulty in using AES in this specific population where application is only possible on the opposite leg. Given the very large cost impact of using AES in this population, the considerable time required for nurses to fit them and the considerable uncertainty about their effectiveness; the committee and the subgroup considered that the use of AES as a sole prophylaxis option in this population should be discouraged. However, AES still ranked higher than no prophylaxis and the committee therefore determined there was not enough evidence to recommend against their use as a sole means of prophylaxis. The committee also noted that in general it was not possible to include any side effects for the mechanical prophylaxis options in the model, and hence their cost effectiveness might be over-estimated. Additionally, the trials of all mechanical prophylaxis options have used them for longer durations than how they are currently used in practice, where early mobilisation is encouraged, so the efficacy levels seen in the trials may not be possible to replicate in practice. Hence, a weaker “consider” recommendation would be more appropriate. |
| Other considerations | The committee noted the dose used for aspirin in the evidence represented a non-standard dose for the UK at 100mg per day. Clinicians can decide whether to use 75mg or 150mg. |
Footnotes
- n
At the time of publication (March 2018), aspirin did not have a UK marketing authorisation for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing unlicensed medicines for further information.
- o
At the time of publication (March 2018), LMWH did not have a UK marketing authorisation for use in young people under 18 for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing unlicensed medicines for further information.
- p
At the time of publication (March 2018), LMWH did not have a UK marketing authorisation for use in young people under 18 for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing unlicensed medicines for further information.
- q
At the time of publication (March 2018), LMWH did not have a UK marketing authorisation for use in young people under 18 for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing unlicensed medicines for further information.
- r
At the time of publication (March 2018), LMWH did not have a UK marketing authorisation for use in young people under 18 for this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed consent should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing unlicensed medicines for further information.
- Elective knee replacement surgery - Venous thromboembolism in over 16sElective knee replacement surgery - Venous thromboembolism in over 16s
Your browsing activity is empty.
Activity recording is turned off.
See more...