Prevention and treatment of venous thromboembolism in hospital and the community: a research programme including the ExACT RCT

Hospital-acquired thrombosis

The largest risk for acquiring VTEs is hospital admission. 4 Current UK guidelines for preventing HAT recommend using the Department of Health and Social Care’s risk assessment tool to assess appropriate thromboprophylaxis for individuals. 7 Thromboprophylaxis has been shown to reduce the risk of VTE by 75% in surgical patients and 50% in medical patients.

In 2010, Commissioning for Quality and Innovation (CQUIN) agreements were introduced that required all UK acute trusts to assess risk for VTEs for at least 90% of patients, or risk financial penalties. 8 However, an All-Party Parliamentary Thrombosis Group (APPTG) survey found that implementation of risk assessment was poor,9 and only 58% of trusts carry out a regular clinical audit of thromboprophylaxis. Furthermore, HAT can occur up to 90 days post discharge, but there is little or no understanding of the role that primary care can play in the process; there is also no evidence of the use of care plans for community VTE prophylaxis.

Unprovoked deep-vein thrombosis

The risk of recurrence among patients with unprovoked thrombosis is higher than for those with identifiable risk factors, such as surgery or long-haul travel. The cumulative rate of recurrence for those with unprovoked VTE is about 25% at 5 years and 30% at 10 years.

The risk varies with time, with the highest risk evident during the first 6–12 months; the risk never reaches zero. Furthermore, these recurrent events are fatal in about 5–9% of people. Most recurrences can be prevented by appropriate antithrombotic therapy, but there is a concern in terms of the duration of treatment owing to the increased risk of bleeding from prolonged OAT.

Several studies have investigated the optimal duration of OAT, with the British Committee for Standards in Haematology recommending between 6 weeks and 6 months, depending on the aetiology. 6 However, the optimal duration of OAT for patients with a first unprovoked DVT remains uncertain, with randomised controlled trials (RCTs) suggesting that the duration has little effect on the rate of recurrence. 10 Research to address the most appropriate approach to preventing recurrences is still required.

Identification of patients at the highest risk of recurrence

Given the balance of risks and benefits of anticoagulation therapy (AT) described above, it may be beneficial to identify those at the highest risk of recurrence. It may be possible to stratify individual risk based on a variety of clinical factors, including sex or comorbidities, or to measure laboratory markers such as coagulation factors. Alternatively, risk prediction models can be used, such as the Vienna Prediction Model, although this is most accurate in people already at low risk of recurrence. 11 There is evidence that normal D-dimer levels, measured around 4 weeks after cessation of OAT, are associated with lower risk of recurrence. 12

D-dimer is a fibrin degradation product present in the blood after a blood clot is degraded by fibrinolysis; levels of D-dimer can be tested using laboratory and point-of-care testing devices. The primary clinical use of levels of D-dimer has been in conjunction with probability scores to determine whether or not further investigation is required for the diagnosis of VTE.

More recently, interest has been shown in studies that have investigated whether or not the levels of D-dimer can be utilised as a guide for determining who is at risk of recurrent VTE following treatment of the initial acute episode. Some small studies have suggested that the levels of D-dimer could act as a useful predictor of recurrent VTE in patients whose OAT is discontinued. 13 There are limited data on the utility of levels of D-dimer as a predictor in patients still on therapy, but Rodgers et al. 14 have shown that the levels of D-dimer can be used to predict recurrence in low-risk female patients. Two systematic reviews of D-dimer testing after cessation of treatment for VTE (seven studies with total of 1888 patients) concluded that ‘additional research is needed to establish the optimal interval between stopping anticoagulation and performing D-dimer testing, (and) to identify the optimum cut-off point that predicts recurrence’. 14

No studies to date have investigated the utility of D-dimer testing while a patient is still undergoing therapy, and there are no studies that have investigated the utility of D-dimer testing in conjunction with clinical algorithms to prevent recurrence of DVT.

Post-thrombotic syndrome

Post-thrombotic syndrome is a frequent and costly complication of DVT that can lead to chronic venous insufficiency and ulceration, and reduced quality of life for patients. Prevalence of PTS in one study was found to be 37% after 2.2 years, with 4% of people having severe PTS. 15 PTS has a cumulative incidence after 2 years of around 25%, and it has been postulated that prolonged treatment with OAT could prevent the development of PTS. 16 However, the only randomised data available suggested no association between long-term low-dose warfarin treatment and the development of PTS,15 and there are no data available for patients treated with long-term therapeutic warfarin.

Barriers to implementation of thromboprophylaxis

For the reasons discussed above, the prevention of VTE and appropriate management of the risk of recurrence is important. However, despite the introduction of guidelines8 and CQUIN agreements, it seems that the guidelines and recommendations are not universally implemented.

It is known that there is little knowledge about VTE in the public arena,4 which is perhaps unsurprising, as it is apparent that HCPs also underestimate the extent and impact of the problem. The APPTG report highlighted the role that primary care could play in improving the management of VTE. 17 General practitioners (GPs) are in a good position to deliver education to patients, and to improve the management of patients post hospital discharge. However, there is little evidence to date of the use of care plans for community VTE prophylaxis.

When considering patient barriers, diabetes studies have shown the desire to avoid injectable drugs, so low-molecular-weight heparin therapy may, therefore, introduce concordance issues. 18 Patients were educated about the risk of heparin-induced thrombocytopenia, a life-threatening, immune-mediated prothrombotic adverse drug effect. Patients appreciated the information about the potential adverse reactions to the drug; the information given did not lead to a treatment refusal, with all patients choosing treatment. 19

The lack of public awareness could be addressed through public education programmes, as improving understanding will empower patients with the knowledge to request a risk assessment on admission to hospital. 20 However, we do not know patients’ attitudes to education and information in this area. Furthermore, we do not know how best to deliver this information; no educational measures have been developed.

The HCP barriers to initiating VTE prophylaxis may be manifold. There is some evidence that non-adherence to guidelines is an issue for clinicians but barriers to implementation are unclear. Knowledge to practice translational issues are extremely important for the successful integration of thromboprophylaxis into the community and clarity around processes is necessary.

The guidelines stipulate a supporting role for GPs, based on notification that a patient has been discharged from hospital and the treatment prescribed. However, communication between care settings is known to be problematic, leading to the role performed by primary care being unclear. If primary care is to contribute more effectively to the prevention of HAT, then a better understanding of its role and of the factors that influence the role is required.

Summary

The VTE prevention and treatment programme was designed to provide an evidence base to improve the prevention and treatment of VTE in the NHS, through addressing some of the existing gaps in the evidence, as discussed above. The programme consists of a series of studies including a RCT; a survey of stakeholders; qualitative interviews with patients, doctors, nurses and other stakeholders; cost-effectiveness analyses; patient preference survey; and a number of economic modelling studies. Each work package is detailed below.

Patient characteristics

A total of 878 patients were recruited and sent the study pack and 564 of these returned completed questionnaires (64.2%). The mean age of respondents was 64.3 years [standard deviation (SD) 11.8 years], 38.8% (219/564) were male and the large majority were from a white background (97.5%, 550/564). A total of 50.5% (285/564) had either a university/college degree or a professional/commercial qualification, over half (53.4%, 301/564) were retired and 33.1% (187/564) were in either full- or part-time employment.

Most admissions were planned (91.7%, 517/564) as opposed to emergencies (7.3%, 41/564). Over two-thirds of admissions were for orthopaedic surgery (69.3%, 391/564) and all other admissions were general medical or surgical cases. The mean length of stay was 8 days (SD 14 days) and the mean time since discharge from hospital to the interview was 39 days (SD 21 days).

Primary care characteristics

Of the 817 GPs and 583 practice nurses who were sent study packs, 92 GPs (11.3%, 92/817) and 19 practice nurses (3.3%, 19/583) returned the surveys. The level of seniority and experience varied within the sample. The sizes of practices ranged from 472 to 52,000 patients (mean = 10,162 patients, SD = 6425 patients). Over one-third of the respondents indicated that they had an anticoagulation clinic at their practice (36.7%, 40/111).

Participants

Fourteen participants from two primary care trusts in the UK were interviewed for the study: 12 GPs and two advanced nurse practitioners who were commissioned by primary care trusts to provide a 24-hour rapid-response service for suspected DVTs.

Methods and analysis

A qualitative research design was used with data collected via face-to-face or telephone interviews. Interviews took place with personnel from acute trusts and other relevant organisations and examined the current practice of thromboprophylaxis, and the knowledge and experience of VTE prevention. The interviews explored the interface between primary and secondary care in terms of VTE prevention and the perceived role of primary care, and examined interdisciplinary communication, perceived barriers to VTE management, training provision and future requirements for VTE management. Key informants were identified, followed by snowball sampling. The recorded and transcribed semistructured interview data were analysed using constant comparative methods.

Interviews were conducted with a purposive sample of staff in four acute trusts and a sample of people from relevant organisations such as the Lifeblood Charity (www.thrombosis-charity.org.uk) and Anticoagulation Europe (www.anticoagulationeurope.org).

Participants

Seventeen participants were interviewed for the study. This group comprised a primary care trust commissioner, charity personnel, clinicians and VTE nurses and trainers.

Participants were provided with an information pack comprising a covering letter and a participant information sheet. Participants were asked to complete a consent form at the time of contact or provided verbal consent for telephone interviews. All interviews were digitally recorded with the permission of each participant. Content of the recordings were transcribed verbatim.

Methods

Participants were provided with an information pack comprising a covering letter and a participant information sheet and were asked to complete a consent form at the time of contact. All interviews were digitally recorded with the permission of each participant. Voice recordings were transcribed verbatim. Transcripts were identified by code number only. Participants were not identified in any written material resulting from the interviews. The recorded and transcribed semistructured interview data were analysed using constant comparative methods. Data were managed using NVivo 9 software. All transcripts were independently reviewed and codes developed in an iterative process to identify emerging patterns in the data and an initial coding framework. Using constant comparison,21 similarity and differences were identified within and across the transcripts. By comparing each part of the data, analytical categories were established and key concepts selected. Final themes were agreed between four of the authors to enhance reliability. Representative quotations that illustrate both typical responses and a range of views have been selected to illustrate the study findings.

Ethics

Ethics approval was provided by the National Research Ethics Committee (reference 11/H0605/5). The trial grant number was NIHR RP-PG-0608-10073.

Participants

A total of 31 surgical, medical, cancer and trauma patients from five separate hospitals in two NHS trusts were interviewed.

Summary of findings

Although the findings from the individual studies have been published23–25 for the purposes of this report an overview of the findings will be presented.

The following common themes were identified:

• communication

• knowledge

• role of primary care

• education and training

• barriers to patient adherence.

Communication

Data from all parts of this work package highlighted the issue of communication, be it between HCPs and patients or between the HCPs themselves.

From the patient survey, 138 out of 878 (16%) patients received information regarding VTE prevention from primary care, whereas 702 out of 878 (79%) reported receiving information on hospital admission, and 471 out of 713 (66%) of those discharged on extended prophylaxis reported receiving information about this.

A total of 843 out of 878 (96%) patients reported receiving treatment for VTE prevention and 713 out of 878 (81%) reported continued treatment after they were discharged. The majority of patients received pharmacological prophylaxis alone, antiembolism stockings alone, or both pharmacological prophylaxis and stockings during their hospitalisation.

In terms of communication between HCPs, it was clear that there was dissatisfaction from both primary and secondary care.

From the primary care perspective, there was concern about both the quality and timeliness of information received on hospital discharge with delays of up to 6 weeks mentioned:

No not really, we don’t really have that much contact really. A lot of the time the correspondence that we get from hospital is quite delayed. So when a consultant’s decided that there’s going to be a planned surgery, it can sometimes be delayed by up to 4/5/6 weeks before we get told about it. And also, after the procedure, and I guess this probably makes it more difficult, after they’ve had the procedure done, again it can take up to 5 or 6 weeks before we even find out that they’ve had a surgical procedure done . . . So I guess that would be one hindrance to us actually, getting people started on things when we don’t find out sometimes 5 or 6 weeks later. And that’s probably the most critical period, when they’re going to get a DVT or a PE.

GP1

I think it’s one of these areas, it’s like a lot of areas in medicine, where there is no connection between secondary care and primary care. The secondary care somehow thinks that we’ll pick it up. Well we may not because we never get the discharge summary for 2 weeks. We don’t necessarily receive the discharge summary for a couple of weeks after the patient’s been discharged . . . Well now, because the number of appointments has risen and the number of patients on their books has risen an extra bit, it rises by the month, some patients say, well we’ve got this letter and the receptionist just takes it off the patient and so we don’t actually get to see them. It’s all to do with the amount of appointments we have is just limited.

GP6

The content and quality of patient discharge notes received by the GP vary widely and are a concern for many participants. They suggest that discharge notes are often brief and lack basic information:

Completely pot luck it just depends on the quality of the discharge summary. Some discharge summaries are very good, they tell you what the dose of Clexane that they want you to give and for how many weeks and what they’re treating it for. And then on other hand, you just don’t really get any feedback at all. And it’s only sometimes when patients ring you up and say, I’m on a Clexane injection and I’ve been told to get some more from my GP. So it varies largely.

GP1

One participant acknowledged that there is a gap in the care pathway where no one has responsibility for a discharged patient on extended prophylaxis. One GP questioned if the responsibility to make sure that a patient has been discharged with prophylaxis lies with the consultant having done the operation or with primary care:

A lot of the time . . . we get a phone call saying, oh can you come and do a home visit, leg swelling or pain in the chest. And a lot of the time, when you diagnose it, you then look back and think, actually somebody should have been responsible for looking after this patient. I think once they’ve had their operation done, I think it’s a grey area, in terms of where the responsibility lies. Does it lie with the consultants who’ve done the operation to make sure that they’ve sent patients home with prophylaxis, or whether it’s our job then to just make sure that they are on prophylaxis when they’ve come out?

GP1

It was suggested that a solution to this grey area in patient care would be an official handover where acute trusts request that primary care takes over:

Some consultants will probably say responsibility lies with them, they’ve done the operation, they should decide on prophylaxis and how long for. Whereas some consultants would say it should fall upon the GP because it is an extended course and it’s the GP’s responsibility to follow it up. But yes, I think up until we probably get good communication or some kind of pro forma where there’s an official handover occurring saying, right, I’ve done this operation, this is what I’ve started them on, please would you carry on either prescribing this or looking after the patient. So I think that probably needs to be improved.

GP1

Secondary care should be responsible until they have effectively communicated the patient’s needs to primary care and their GP has agreed to take over responsibility.

GP9

However, participants confirmed that they do not make contact with a discharged patient on the assumption that the hospital had made the necessary arrangements. Shortfalls in the procedure are put down to hospitals not doing their job properly:

We wouldn’t make active contact, no, if somebody was discharged, I would assume that the hospital had sorted it all out really. We don’t really have time to phone people up just in case the hospital hasn’t done their job properly.

GP5

Generally, if I think they’re a fit young person or they haven’t got any physical or mental issues, then I would tend to leave the patient alone, unless they contact me with any issues, but otherwise, no.

GP1

The GPs in rural practices are more likely to contact a patient discharged from hospital on low-molecular-weight heparin:

In our practice, when we see they’re on low-molecular-weight-heparin, sometimes I’ll give them a ring and say, are you alright giving yourself a jab or do you want to come up to the surgery to have it done? But that’s because we’re a rural practice. But if you’re in a big city or a big town practice, no you don’t have the time to even breathe let alone ring them.

GP6

Delays in receiving discharge notification could make prescribing difficult for primary care and one participant felt that commissioning should not be split between providers:

Sometimes you don’t hear until the patient has been discharged for a couple of weeks and if you can provide a week’s anticoagulation I really can’t see why you can’t provide a month’s anticoagulation because it would be a much neater package commissioning it this way rather than splitting it between various providers.

GP7

Similar issues were raised by secondary care clinicians with regard to communication both to and from primary care. It was pointed out that VTE occurs and is usually diagnosed in the community and, thus, the GPs have a role to play:

I think we’ve missed a trick here – there’s this belief that thrombosis is an issue that is seen and dealt with within the hospital. Most people who will have a hospital-acquired thrombosis will have it in the community because you don’t tend to get a DVT or a PE until you are discharged. So before you even look at prevention of VTE – the majority of thrombosis will be diagnosed in the community. Prevention even more so . . . So primary care has a large role to do here. Now maybe it’s unfair on them that it’s all being dumped into the community but actually that is the price you pay for having less beds and it’s the price you pay for having to free up beds to admit the acutely ill patients.

Consultant and VTE lead

Hospital staff expressed concerns regarding whether or not patients complete the course of extended prophylaxis when they have been discharged from the hospital as ‘a bit of a grey area’:

We send patients home on Clexane, they’ve been shown how to do it, but they [midwives] feel they are not actually administering it at home. Some will be visited by community midwives more regularly and they can check up on that but you might have a lady that isn’t seen and no one seems to check up that they are actually taking their thromboprophylaxis or not. It could be the same in other patients.

VTE prevention lead

From our root cause analysis we’ve done on patients most of them have gone home with the necessary prophylaxis and certainly now we’ve developed a standard pro forma root cause analysis form. One of the questions on it is now, ‘did you complete the course?’ or ‘did the patient complete the course?’ and we’ve started asking them that if we can, if we are able to contact them. So yes that is a bit of a grey area. You can’t know for certain if they are self-administering. I suppose that is something we could ask when they come back to clinic but then it’s after the horse has bolted then, isn’t it?

Clinical nurse tutor

There also remains a problem of used sharps disposal for the patient:

Used sharps disposal is also a challenge and often the patients are asked to bring them back to hospital.

Consultant nurse for anticoagulation

It was suggested that ‘patients need to know what to expect’. When this participant was asked who could do this, they responded with: ‘The GP’ (consultant nurse for anticoagulation).

There were few positive comments regarding communication between secondary and primary care, with the one notable example being in maternity services:

It seems to work very well for maternity-related thromboprophylaxis with very few problems between secondary and primary care.

Community pharmacist

The need for effective communication between every level of the NHS for the safe management of extended thromboprophylaxis has been recognised at the commissioning level:

Extended thromboprophylaxis can be needed for quite a long time. The whole system needs to manage that safely and that’s about communication . . . Whilst the individual provider or the individual surgery thinks that everything is fine – it’s only when you look at the overall picture you realise, actually patients came out of this clinic and information wasn’t sent to the GP and, therefore, the GP doesn’t know what’s going on, and so we try and monitor that system from an overall point of view . . . the incentives are much more focused on the business of a hospital to deliver a business bottom line in terms of funding than working across a whole system to actually ensure that all the safety and all the protocols are properly joined up with other providers outside their business. There is something there about actually the market system around secondary and primary care it’s not joined up, and that’s a problem.

Commissioner

In areas of high risk such as anticoagulation and that type of management that the commissioner needs to be really understanding, looking at what’s going on and looking at where the risks are and really trawling, quite actively, trawling to try and see where they can identify risks . . . given that things like anticoagulation and the top 10 risks and the top 10 examples of avoidable admission we should be focusing, as commissioners, much more upon those clinical areas we’re being encouraged to do than to focus on other things that commissioners get involved in, which might be less outcome based.

Commissioner

Knowledge

Patients

Patient understanding and awareness of hospital-acquired VTE appeared vague and incomplete. Many patients had an elusive idea about their VTE preventative treatment but they generally deferred to the health professional ‘knowing what they were doing’:

All I had was morphine and my usual tablets. That’s all I had.

Did you say you had stockings, surgical stockings?

Yep, yeah, those nice white ones.

Did you understand why you needed to use those?

Well, it’s obviously to stop the blood clots forming. I don’t understand why, it’s just there and not anywhere else on the body. But there we go, they must know. Yeah.

When I went to see the surgeon, in the follow-up interview, I asked him why I’d been given these anticoagulants, and he said that that was something that everybody was given who had an operation.

P1 Orthopaedic patient

Recollection of information was unspecific. P11 suggested the required information would have been given among all the other information received but the amusing anecdote evokes the best recall:

When you saw the surgeon, he mentioned all of the risks, he went from, you’re going to have this, you’re going to have this but this is what could happen, you could get an infection, you could get this, and he listed so much stuff and by the time he’d finished I went ‘Oh my God I’m gonna die.’. And in there somewhere would’ve been blood clots. But I don’t remember specifically talking to the nurse about, at the hospital.

P11 Orthopaedic patient NOC53

Many patients said that they knew about blood clots but did not link them to their own situation. Recall of the information provided in hospital was vague, even suggesting that they were told about blood clots when flying:

I’m quite ignorant about blood clots. I mean you always think when you’re flying, people say you can get blood clots.

P19 Orthopaedic patient ORH31

Can you tell me about the stockings? Did they explain to you why you were wearing them? Not really, it was to do with the aircraft.

P22 Medical patient ORH60

A patient was aware of the connection between immobility and blood clots:

Nowadays you’re not kept in bed for any longer than absolutely necessary to avoid blood clots. ’Cause years ago they used to keep people in bed and then they had thrombosis and things as a result which wasn’t very good but they know more now.

P24 Orthopaedic patient NOC120

A medical patient felt that the hospital was very informative regarding VTE information but was too poorly on the first day to consider the information:

I was aware that they risk assess and also they actually, even though I was having aspirin, I had the injections . . . because I’d been in hospital probably 6 months earlier as well, I’d gone through all that anyway. So I actually, actually I felt too poorly to be bothered the first day, but they were very informative.

P27 Medical patient UHB47

One orthopaedic patient felt that there was too much emphasis placed on blood clot prevention. This patient experienced haemorrhaging after hospital discharge:

If anything I would say probably, over, people are really, really quite frightened of it, the staff are and things like that, there’s no doubt about it. They concentrate very heavily on it . . . What the mistake was, it was the medics in the hospital that were warfarinising me too heavily too quickly, which comes full circle to my analysis of, they’re too paranoid about the blood clots.

P4 Orthopaedic patient NOC99

From the patient survey, it was found that, prior to admission, 15.7% received information from GPs or practice nurses in primary care. Secondary care health professionals provided advice to 79.3% of patients on admission to hospital and to 66.4% of those patients discharged from hospital on extended prophylaxis. Post discharge, 12.8% received information regarding blood clot prevention from GPs or practice nurses in primary care.

Role of primary care

The large majority of the GPs and practice nurses reported that they never or only occasionally conducted VTE risk assessments (94.6%, 105/111), yet 34.2% (38/111) believed that this role should fall within the remit of primary care. Even more respondents (50.5%, 56/111) believed that they should be providing advice to patients prior to elective hospital admissions, but, in practice, 79.3% (88/111) never or only occasionally do this. Involvement in VTE risk management post hospitalisation was higher in those surveyed, with 39.6% (44/111) reporting that they never or only occasionally managed extended thromboprophylaxis and 64.2% reporting that they never or only occasionally provided advice about VTE risk to their patients following discharge from hospital. Three-quarters of respondents (74.8%, 83/111) believed that primary care should manage patients requiring extended prophylaxis and 58.6% (65/111) believed that primary care should provide advice to patients requiring extended prophylaxis following hospital admission.

Despite a substantial proportion of respondents indicating that primary care should take on a greater role, the majority of GPs and practice nurses perceived barriers to conducting VTE risk assessments (84.7%, 94/111), managing extended prophylaxis (73.0%, 81/111) and providing VTE management advice (69.4%, 77/11) in primary care. The main barriers identified were lack of time, resources and expertise; lack of continuity of care and poor communication between primary and secondary care; lack of awareness of planned hospitalisations; lack of knowledge of exact regimes and risk/benefit ratio of prophylaxis; lack of understanding of risk associated with different procedures and the potential complications; lack of consistency over VTE risk management protocols at different hospitals, the cost of prophylaxis and confusion over whether primary or secondary care is responsible; the mobility/wellness of patients to attend surgery for prophylaxis; and the pressure from many other areas in primary care.

One participant who was aware of the NICE guidelines7 for VTE felt that they were hospital focused:

NICE guidance exists but I think current guidance is predominantly hospital focussed.

GP15

Some participants felt that they, as GPs, should have a role in reducing the risk of VTE, but were unaware of any specific guidelines and were vague regarding what that role would entail or how it could be implemented. They suggested that their role should only encompass preventative medicine and that primary care should not be considered as the safety net for VTE prevention. Participants quoted a series of reasons why their involvement would be difficult, in particular the logistical problem of having no contact with patients before hospitalisation and the burden of adding to their already exhaustive workload:

All GPs should have a role in reducing risk of VTE. However cannot recollect any specific guidelines for general practice.

GP8

A good question, we should do [have a role in reducing hospital-acquired VTE]; I don’t think we know what our role is at the moment.

GP4

I don’t think it’s a safe system to rely on primary care alone because there are times when we don’t have anything to do with the patients before they go in. So I think that it’s something that we could be doing, sort of preventative medicine, but I don’t think we could necessarily be the safety net.

GP4

Additional funding and resources to enable primary care to take on a role in VTE prevention was a major factor for several participants. It was felt that primary care was well placed to improve outcomes, particularly as they had access to patient medical details and histories, but their involvement would require clear planning to encompass training, resources and regular audit:

Limited role for primary care due to capacity in general practice to take on additional work which is not funded.

GP8

A small role – but not one that I am prepared to take on without extra time and remuneration.

GP10

Primary care is well placed to improve outcomes in this area but clear planning, training provision and resource, with regular audit, would be needed to ensure this was done effectively and safely.

GP9

A lot of the time we know the patients and their background history. I can’t see why we can’t do a risk assessment. I guess the cynics amongst us would probably say, well where are we going to get the time to sit down with every patient who’s going in for a planned elective admission, to go through the risk assessment? I guess that’s just a matter of time and resources really.

GP1

However, one participant would not want to take on risk assessment in general practice because it would complicate the process and impose additional appointments on their workload. In addition, there could then be a problem in linking the patient record between general practice and the hospital:

In terms of assessing people who go up to hospital, I would not want to take that on in general practice. I think that that would be complicated; it would be an unnecessary appointment for the patient. So it would be duplicating patient’s, you know, it would make things more complicated for the patient. What should happen is, that should happen either when somebody is listed for surgery, that they should go and see a nurse at that appointment, or they should, if they’re having a pre op clinic, they should be discussed at the pre op clinic. For them to come and see us about it, that means a third appointment, which is inappropriate. So it’s not good for the patient, it’s not good for us and it doesn’t make sense from the hospital’s point of view because they haven’t got a record of whether we’ve done it or not. So on no count does that make sense.

GP5

Some participants felt that primary care could step in if there was a breakdown in the care pathway but essentially thought that acute trusts should be the main provider:

In terms of context – side effects experience or if there is a breakdown and the district nurse does not turn up then primary care has a role to play there but in routing provision of that, that should be commissioned as one package delivered by the same team and in terms of delivery I don’t think primary care should take over that.

GP7

Not our role, should be in their routine provision of information from hospital.

GP14

Similarly, participants felt that there was a role for primary care in maintaining patient compliance:

There is some role for them [GPs], from my point of view, this is just on a personal level as a nurse, if we have supplied the stocking for a prophylaxis, we should be following that through, it’s a prescription. If I prescribe an antibiotic, because when I go out to these patients, some of them don’t have a DVT, they’ve got cellulitis, I prescribe an antibiotic for them. If they’re not then taking their course of antibiotic, then I have to do something about that. But just supplying a prescription for a tens [TED stockings (thromboembolic deterrent)] stocking, and then the patient doesn’t do anything about it, then, you know, there’s a failure in the system.

Advanced nurse practitioner 11

I think there’s a role in trying to maintain compliance, yes totally. But whether we can do much more than that, I don’t know.

GP4

Raising awareness of hospital-acquired VTE risk was a more acceptable role for primary care to take on, but, as the task is already set up in acute trusts, participants were reluctant to see why they should take it on as well:

I think we should make them more aware of the risks definitely, so that, I guess that’s another way to tackle it, whether we somehow collate data on patients who are going in to hospital and send them this information beforehand. And get the patients to prompt their consultants and question them whether they should be coming home with any prophylaxis. That could be another strategy, yes.

GP1

Patient awareness, that would be fine but only if there is a system for doing this systematically, otherwise it will be haphazard.

GP13

One participant felt happy to provide prescriptions to patients and send out district nurses to help elderly patients with injections:

Yes, we’re more than happy to give that out to our patients. And patients who are confident enough to self-administer it will be quite happy to ask us for some more. Those patients who are elderly and are unable to administer it, we also get our district nurses to go out and give them their Clexane injections. So yes, we’re quite happy prescribing it.

GP1

A specific concern was when GPs accepted hospital procedures without question:

I think a lot of the time the ones that probably do get missed are the ones who’ve been released by hospital and we’ve probably not questioned why they haven’t been sent home, especially if they’re at high risk of DVTs, on Clexane or stockings. So I think we, as GPs, should question discharges probably a bit more and especially after big operations. And I think probably, the tendency at the moment is that we do leave it in the hands of the consultants. And if they are on prophylaxis afterwards we accept that and if they’re not, again we just accept that. So I think we should probably question it a bit more.

GP1

Time pressures and financial resources were cited as the main barriers to primary care having a role in VTE management:

The key barrier, I think for me, would be the time and the resources because I’ve done risk assessments before and I’ve not had any difficulty in doing that. And I think especially if you use something that’s been validated, like the Wells’ scoring system, it’s mostly a tick-box exercise and it gives you a score at the end. So I don’t see why we shouldn’t be able to but I think, like I said it’s just the time and the resources.

GP1

One participant suggested that any funding provided to primary care for their involvement in VTE management should be set against a system of target outcomes that shows that DVT rates have gone down. Furthermore, it is felt that acknowledging the cost of VTE in terms of health burden would offset the financial resources required:

I think it’s just highlighting the costs of post-operative DVTs, PEs, not just the financial cost but also the health burden as well. And getting the health professionals to acknowledge the costs and kind of balancing it up with the time and the resources that they’ll need to put in to prevent those DVTs and PEs from occurring.

GP1

Education and training

The issues of education and training, across the range of health professionals and patients, was repeatedly noted.

There was little evidence of any training in VTE across the primary care participants:

No, no official training. I think that would be the same across the whole number of staff. We’ve not really had any official training.

GP1

It’s non-existent actually.

GP7

Not specifically, no.

GP3

Participants had varied suggestions as to how training could be implemented. This included outside speakers attending the practice at lunch time, training using a CD (compact disc) or YouTube presentation (www.youtube.com; YouTube, LLC, San Bruno, CA, USA), a purpose-written online module, training in the form of continuing professional development, and training form information published in GP magazines:

Training I think would be good generally across all staff members, nurses and doctors. And how could it be implemented? I mean yes, I think either outside speakers could come in and give the practice a talk during the lunch time session or probably inviting GPs to go to a seminar that’s being held. That would probably be a way of approaching it.

GP1

The easiest thing is for a, somebody to come and deliver in-house training to the whole team.

GP2

Theory and practical aspects provided as a day course or over the internet.

GP8

Junior doctor training in terms of risk assessment was repeatedly highlighted as an area of concern. Ongoing training may be necessary to continue to promote awareness and develop understanding to prevent the risk assessment becoming merely a tick-box exercise. It is necessary for clinical staff to understand why the exercise is so important. Engaging with the problem may subsequently ensure that there is follow-on action from the risk assessment:

‘There is very little awareness; there is little awareness amongst primary care staff . . . Also amongst secondary care staff because many see this as a chore, as a tick-box exercise.

Clinician

It can just end up being another piece of paper another tick-box exercise. I think that’s where the importance of the training comes in because you need people to understand why it’s so important.

Nurse tutor, VTE committee member

There was concern regarding current risk assessment guidelines7 in that CQUIN payments are based on the number of patients risk assessed with disregard as to whether or not there has been any action based on that assessment:

One of the weaknesses of the current strategy is that the outcome that is being measured for the moment is – the number of risk assessment forms completed. The focus needs to be on whether they have been completed correctly and clinicians have acted on that assessment. People seem to think that it is all about identifying whether a patient is at risk of thrombosis but a risk assessment tool is also there to identify whether a patient is at risk of complications of thromboprophylaxis and, therefore, it is essential that the information is used to guide practice.

Consultant and VTE lead

Individual trusts have developed their own literature for the education of both junior doctors and patients. In some cases, staff information has been adapted for patient use. Previously expressed concerns regarding low reading ability and poor levels of understanding among patients suggest that a nationally developed leaflet, targeted at an appropriate reading level, may be a simple but important requirement. Equally, standardised training material for clinical staff would help to regulate the VTE assessment and care:

We’ve adapted it [leaflet] to be appropriate to our patients.

Nurse

Bigger trusts have specific thrombosis teams or VTE nurses . . . we don’t have that but we have just put together a document that is going through the approval process so hopefully that will help. I’m not saying patients just don’t get the appropriate treatment but I think maybe the actual process gets a bit blurred sometimes.

Nurse tutor, VTE committee member

We have followed the NICE guidance and written our own local trust guidance and that’s available on the intranet and available in a little booklet form that we give to the junior doctors when they start working here.

VTE trainer

There were examples of low levels of knowledge of VTE risk and prevention among staff in some acute trusts, including hospitals where the majority of patients would be assessed to be at high risk:

The major deficiencies are actually among health professionals and that we need to address those first before we start educating patients anymore.

Consultant

When I’m doing training . . . it’s only an awareness not high fluting – signs, symptoms, prevention, risk assessment, all the key things. It is improving now because we’ve done a lot of work but even the knowledge amongst people who work in hospitals, in an orthopaedic hospital where it’s always been higher risk, is low. If it’s low for that group then the patients themselves are unlikely to have a huge amount of knowledge and then things like on your medical training, nurse training there should be a whole module on VTE and the risks associated with it.

Nurse tutor, VTE committee member

A considerable variation in VTE teaching for a range of medical staff has been identified, highlighting a call for improvements:

I’ve been looking at education and its huge variability in the amount of teaching that medical students get in haematology where most of the VTE teaching is concerned so it varies from virtually nothing to 8 weeks haematology teaching between the different medical schools and if one looks at the nursing syllabus – the midwives have nothing, there’s no module at all on VTE and the nursing modules vary so there is a huge need for improvement in education.

Charity executive

Critical care staff who see one or two VTEs per month felt that they did not know enough about thromboprophylaxis:

I see one or two cases of VTE a month. I don’t think I have enough knowledge or information about VTE and thromboprophylaxis.

Critical care charge nurse

Similarly, training to cover the management of VTEs may be inadequate:

Possibly the thing we don’t cover so well at the moment is the management of suspected or actual VTEs . . .

Nurse tutor, VTE committee member

Even when there is a clear training programme in place, a nurse tutor suggested that attention can slip and compliance rates drop off:

You almost have to police it. You do things and you think, ‘right they’ve got that now, they know that every patient needs to be risk assessed’ but then something else will come along that takes their attention for a while and before you know it, it’s starting to drop off again.

Nurse tutor, VTE committee member

There were examples of excellence in staff commitment, responsibility and training. In several acute trusts training was mandatory. The staff were well trained and knowledgeable, keeping up with new policies and guidelines to maintain an up-to-date knowledge base. The most successful implementation exists where members of the team are allocated to the task and feel dedicated to the role:

VTE training is mandatory in our organisation and this is a very useful driver.

Consultant nurse for anticoagulation

In this trust we ask, not just nurses, but everyone that has front line direct patient contact to complete the e-learning VTE learning module, which is mandatory and we also provide, for the nurses specifically, some VTE awareness sessions.

Clinical nurse tutor

We provide slots on all induction programmes for new doctors and nurses, regular lunchtime teaching for pharmacists and an established link nurse/midwife network with study days and monthly lunchtime meetings incorporating teaching. Teaching of new FY1’s who are involved with VTE trust wide audit.

Consultant nurse for anticoagulation

In addition to specific medical training, there were examples where the importance of effective communication with the patient was emphasised:

We have e-learning or face-to-face training. When we do face to face we put more emphasis on the importance of telling the patient the importance of giving them the written information, when to tell them. Getting them to think about, when some ones refusing their TED [thromboembolism-deterrent] stockings. Getting them to think ‘have you actually explained to this person what they are for’. If you haven’t, ‘why are they going to wear them?’. Why would you wear them if they put them on you and you didn’t know what they were for?

VTE prevention lead

It was recognised that documentation outlining the appropriate treatment was required in some hospitals because they do not have specialist teams to manage VTE. This was most evident in specialist orthopaedic hospitals, where staff skills are appropriate to their specialist nature with little access to specialist skills in other medical conditions. Explicitly, orthopaedic surgeons are knowledgeable with regard to risk factors related to surgery and anaesthetics but do not see cases of VTE because they are referred to a general hospital and they may be unfamiliar with the risk factors associated with cancers and other comorbidities. This includes the management of suspected or actual VTEs. This situation also calls for the provision of standardised, nationally available information:

For some patients, there are other risk factors. It would be hard to have a form that covers every eventuality. Even though we are orthopaedic speciality only, within orthopaedics there are actually spinal, oncology patients, your hips and knees, etc. So even within that small group there are lots of different risk factors and if you think about big trusts you might have medical neuro’ [neurology] as well. It would be difficult to have something that covered absolutely every risk factor.

Clinical nurse tutor

The thing we don’t cover so well at the moment is the management of suspected or actual VTEs. Because, we are a specialist orthopaedic trust so we don’t have the input of, I mean a lot of bigger trusts have specific thrombosis teams or VTE nurses or . . . Whereas in this trust we don’t have that but we have just put together a document that is going through the approval process so hopefully that will help. I’m not saying patients just don’t get the appropriate treatment but I think maybe the actual process gets a bit blurred sometimes.

Clinical nurse tutor

Specific training requirements in acute care

Barriers to patient adherence

Participants gave several reasons for thinking that they did not need to wear their surgical stockings. Being mobile very quickly constituted a substitution theory. Furthermore, patients did not wear stockings because they were irritating, they were uncomfortable, they were too tight, the patient’s legs were too swollen, the patient did not think that they were necessary, or the patient was having anticoagulant injections:

I was walking again within a week, so I was sort of up and about more, so it wasn’t as if I was just lying still for, yeah, so I didn’t wear them when I came out no, no. Well they did give me four or five pairs. Do you think perhaps they’d intended you to wear them? Yeah probably. So why did you not? I just think ‘cause I was out, up and about and, feeling y’know, that I’m moving so everything is moving sort of thing so, yeah.

P7 Orthopaedic patient ORH46

No I didn’t. I was carrying on for a little while but they made my legs worse, I felt, they were too tight and that. My daughter in law is a physiotherapist, because I was so active, she thought I would be alright without so, ‘cause they were so uncomfortable they fit in too tight around the top leg here. I was getting up two or three, about four times a night, I was always, on the move, and all the time I didn’t — wasn’t lying in bed for days on end without getting out of bed.

P3 Orthopaedic patient NOC82

To be honest, I found them incredibly irritating. So I wore it for the first week and then I just scrapped it, and I thought, well I’ve got the injections anyway. So what I did, my view was, I won’t wear the stockings but what I’ll do is I’ll move a lot more. Even if I was in bed I was doing leg exercises and keeping movement going there, I kind of thought I’d rather do more of that, than wear the stockings, now whether or not that’s advisable who knows but, I can only apply my own logic.

P5 Orthopaedic patient ORH35

No no I didn’t wear them. ’Cause I was up and about.

P15 Medical/cancer patient

I came home with the ones from the hospital and I was told to keep then on for, I think it’s something like a couple of weeks. But I didn’t, not with these. No.

P16 Orthopaedic patient ORH38

I was told I wouldn’t need the stockings anymore just use the fragmin injections.

P9 Orthopaedic patient NOC138

Several participants suggested that, although they had been given stockings to take home, they did not wear them because they had not been specifically told to. These participants also suggested that they did not wear the stockings because they were unpleasant. Without clear instructions, participants are unsure whether they have to wear the stockings and for how long they should wear them. Several participants thought that it was not necessary to wear stockings once they were home:

I don’t recall being told, you’ve got to wear these for X amount of time. They’re not easy to get on at all, I had to put that on before and I don’t think I’d put it on right, it was all ruckled and they said it’s got to be smooth. And I had that on for a few days but I can’t remember, I don’t think it was on when I came out and I was only in for 4 days. But nobody said, ‘we’re giving you these because we want you to put them on’.

P26 Orthopaedic patient ROH68

They were, my legs were going in and out like this, and I think I took them off myself later on that evening. Nobody actually said to me, take them off or, do you want them taken off? I think I knew what they were for – they may’ve said, but I think I can’t really remember.

P17 Orthopaedic patient ORH61

I was given the tight socks to wear. I’ve got two at home, I was wearing them when I came home. Yeah. I haven’t worn any since I got home at all.

P18 Medical patient ORH17

For 6 weeks supposedly after. Not completely the 6 weeks no, 4 weeks of that I should imagine. I stopped wearing them because of irritation and, you know, just general tightness of the leg sort of thing with a stocking on, not being used to a full stocking.

P33 Orthopaedic patient ROH65

In hospital. I didn’t when I came out. I think I probably should have done for a bit longer. They were so uncomfortable, they are so uncomfortable, they really are. And yes I got them at home but I did stop using them as soon as I ever thought I could . . . I knew they were important but I didn’t think they were totally necessary when I came out.

P34 Orthopaedic patient ROH104

Both medical and orthopaedic patients felt that there was too much information to take in at a time when they were concerned about their hospitalisation. Knowing what to prioritise would help:

I think the difficulty is when you’re going in for an operation like this, there’s so much information that you need to take in anyway, and you’re worried about y’know, going into hospital, who’s going to look after your animals/children/husband or whatever, so, I think it would have to be carefully sort of planned in to, to have an impact really.

P1 Orthopaedic patient ORH4

You don’t know what is important in that information, so if you’re given a load, you can’t prioritise, and nobody else prioritises it for you.

P15 Medical patient ORH8

Several orthopaedic patients suggest that conflicting advice was given with regard to wearing stockings, which usually led to the stockings not being worn once the patient had left hospital:

I suppose that’s probably the most unclear part of the whole procedure, the injections were fine, I was quite happy with, y’know, doing and administering it and that whole process. Obviously the stockings were worn in the hospital sort of continuously and then I think the information you get, the information sheets is: wear of up to 4 weeks post op[eration], but there was definitely conflicting advice from the nursing staff. Some of the nurses were, ‘well it’s not that serious if you don’t wear them’ others were ‘absolutely must wear them’. So, y’know there was definitely conflicting advice.

P5 Orthopaedic patient ORH35

I did ask should I wear them, when I came home and nobody seemed to know whether I should or shouldn’t but I presumed it was OK, y’know you’ve had them on for a fortnight.

P19 Orthopaedic patient ORH31

The nurse said wear them for a fortnight which is what I did and then reading the leaflet afterwards it said keep them, wearing the stockings for after 6 weeks but, I only wore them for a fortnight.

P24 Orthopaedic patient NOC120

Many participants found the stockings uncomfortable to wear and difficult to put on. The nature of hip and knee surgery meant that patients did not have the flexibility to reach their feet:

Very uncomfortable and, very hot but I knew I had to keep using them. I wore them for the full time. Oh I hated them, hated them. I found that the design was awful because they had this little hole in the bottom, which, at night got trapped around your toes, so you’d wake up and your toes were freezing because they’d been stopping the circulation, and then when you were wearing shoes to go out, they trap around your toes as well, they’d work up, no matter what we did, and of course my husband had a terrible job getting them on, I couldn’t.

P6 Orthopaedic patient NOC100

Never managed to do it, now I’ve got short arms unfortunately and of course if you’ve got hip problems it’s a bit difficult to do that, so I had to have help with those.

P8 Orthopaedic patient NOC110

Help with putting the stockings on came from family members:

My wife done it. Even like, with socks, I can’t bend. So anything below the knee normally, my wife does it.

P2 Orthopaedic patient NOC58

I have a wife for that. It was difficult, and that was an area which I think somehow that can be improved, how anybody on their own does it, I don’t know. You cannot do it on your own, and however much you’ve got this, the gadget.

P4 Orthopaedic patient NOC99

I’m very fortunate because my husband’s very good, he does the cooking and cleaning.

P9 Orthopaedic patient NOC138

Oh, it was awkward getting them on. Oh, my son came in. I can’t even get my tights on now.

P22 Medical patient ORH60

All participants given a course of injections to self-administer after hospitalisation claimed to have completed the course. Most participants claimed to have received one training session, were uncertain about giving themselves an injection and would have liked reassurance that they were injecting themselves correctly.

A medical patient did not know why she had been given the injections and complained that the alternative was a daily trip to the doctors. It was not until the research interview that she was made aware that the injections were to prevent blood clots and not for pain relief:

You’ve done it for about a week, and you go, ‘I can’t do this anymore’, don’t be silly of course you can, you’ve done it for a week, of course you can do it and then you go, just do it but there does seem to be this, oh I really don’t want to put this thing in me again. But, realising effectively that, there isn’t a choice, just get on and do it.

P9 Orthopaedic patient NOC138

Not at all happy. But they said that alternative was going to the doctors every day. Well you need a lift for that, you know, when you don’t feel good. So I just had to do it for a month. Yes, bruises everywhere. They just gave me the thing. Injections in the stomach are to prevent blood clots. Oh, are they? See that’s what it was for. Thought it was for relief.

P22 Medical patient ORH60

The nurse did the injections when I was in hospital and then explained that I’d have to do it for a month afterwards. And, my wife wouldn’t do so I had to do it myself.

P24 Orthopaedic patient NOC120

Some participants experienced no problems with the course of injections and complied with the treatment advice:

I injected my belly every day and when I came out I had to do it for 2 weeks I think after. No trouble at all. They give you a quick lesson.

P33 Orthopaedic patient ROH65

I was quite aware of all this blood clotting, so I didn’t need to ask. I was just aware that it was there and you just do what you’re told. It’s doing what you’re told really isn’t it?

P34 Orthopaedic patient ROH 104

I did the whole course. Yeah, I was gonna complete the course, never miss.

P3 Orthopaedic patient NOC82

It was clear from the patient interviews that there were issues relating to retention of information, whereby patients cannot remember, they forget or the information does not sink in. Patients did not remember if anyone talked to them about VTE prevention:

I take the wife with me, she tends to ask the questions and remembers more than I do, because I have troubles, I can remember things but I’m not that great in remembering things.

P2 Orthopaedic patient NOC58

I honestly don’t remember but I’m getting quite forgetful in my old age.

P18 Medical patient ORH17

I don’t remember any particular discuss[ion] but at the same time, I mean what I’ve been told, it didn’t sink in. I’m sorry to be so vague about it.

P18 Medical patient ORH17

I’m pretty certain we were given injections for, to prevent blood clotting, but I mean I was, there was so much, happening to me, that I, can’t a hundred per cent remember.

P20 Medical patient ORH51

Just an operation can cause DVTs, so when you come out you’ll be, I suppose, cared for enough. I can’t remember exactly what was said.

P4 Orthopaedic patient NOC99

They did say blood clots, you know, stops your blood clotting, but they didn’t tell you what symptoms you react from it. Well to be honest, I mean whether they’ve told me I don’t know because I was under a lot of anaesthetic in the hospital for a few days. Whether it was mentioned earlier on and, you know, obviously my minds all over the place, they could have mentioned something.

P28 Medical emergency patient UHB23

One patient suggested that most people are aware because airlines have taken up the issue and if they have done so it must be important:

I think probably more people know about it now because the airlines have taken it up and I think that’s really where you learn most, that if they think it’s worth telling you to wear their stockings or whatever, then, yes it is important.

P8 Orthopaedic patient NOC110

When told the symptoms of a DVT during the interview, a participant remembered reading the information but did not take it in because he did not recognise the importance or significance. This suggests that information needs emphasis and prioritisation for the patient:

That’s certainly was in that information I read, which just shows you that, but, there was no real emphasis put on that. Y’know, it’s most probably quite important to you.

P21 Orthopaedic patient NOC125

There were concerns raised regarding the volume of information given, with one patient overwhelmed by the volume of information provided by the hospital. It was suggested that information was given for the sake of ticking a box with little concern given to the patient’s capacity to understand or take in the information:

You can say you’ve given the information, you can’t say that people have understood it or you’ve actually explained it. I’ve had so much information not just about blood clots but about, cancer and chemotherapy and radiotherapy and I’m a nurse and honestly, I cannot make head and tail of it. And there’s things that, you just think, can’t they just give me a priority like, number one do this do this, do that, you’re just given so much information, I’ve got piles of it upstairs and now and again I might say, ‘Oh I think I’ll, let’s have a look at that now and see if it makes sense now’. I just think it’s, complete waste of time to be honest. I think it’s people saying, they feel they can tick the boxes that, we’ve told you but, actually, what goes in is very different. And what’s sits in your brain’s very different so, I don’t think it’s very impressive myself.

P15 Medical/cancer patient ORH8

Participants suggested that there may be an optimal time frame to receive VTE information to avoid times when it is difficult to take in details when recovering from an operation. The first day was considered to be the worst time to receive information:

I would say that I think the most useful thing would’ve been, maybe a sort of, about 48 hours after having had the operation, to have had a nurse come in who specialised in that sort of thing and explain it to me at that stage.

P1 Orthopaedic patient ORH4

I don’t think I would’ve done [read leaflet]. Probably not the first day, ‘cause you’re still a bit drowsy, probably on the second, third day or something like that.

P2 Orthopaedic patient NOC58

Furthermore, participants suggested that, rather than just being handed the leaflets, it would be helpful if there was someone to go through and point out the key points:

Although you get the booklet of information, to be honest you’re not always that, mentally, alert when you’re still on heavy meds, you, y’know you might read a bit but you don’t really process it, so, maybe just sort of having someone there saying, this is the information what’s in this pack is, there’s this, this and this. Let’s go through these key important things again. It makes a massive difference having someone physically go through it with you rather than just giving you a batch of paper to read. You read it and on the medication at the time you probably don’t process it all. Then you forget where you’ve put it when you need it. It’s much better I think if you have one on one, and something that may be even better would be 2 or 3 days once you’re home. Or you could even have someone do it over the phone.

P5 Orthopaedic patient ORH35

Participants suggested that before the operation would be an appropriate time to do this:

Before the operation yeah.

P7 Orthopaedic patient ORH46

Sometimes when they give it you, you’re not actually at your best to take it in. Some of the information they give you obviously at your pre-op[eration]. I don’t recall the blood clot bit being given to me at the pre-op[eration]. Now that might’ve been a good place to mention it because you’re a bit more aware at that point of things, because they’ve already gone through all the other risks and things.

P9 Orthopaedic patient NOC138

However, the worry about an operation could blot out the ability to take in information:

Too much of everything else is going on isn’t it? And you’re worried about your operation as it is anyway. All they basically say is you’ve gotta have your injections for blood clots. They come round and give you that but there’s not a lot else said after that about it really.

P7 Orthopaedic patient ORH46

All you’re thinking about is the operation so you’re not really yeah, not really switched on to it are you.

P7 Orthopaedic patient ORH46

When you’re not dozy from the operation. To be given some of the information at that point’s probably not the best time to get to read it.

P9 Orthopaedic patient NOC138

Just prior to discharge was also suggested as the best time to receive the information. The information is then relevant to the requirement for patients to self-inject when at home:

They didn’t tell me the detail of how to inject myself until the day I was going home. So by that time I was feeling reasonably sort of fit and looking forward to going home, and I was in a mood for sort of taking those sorts of things in. The first days after the operation, when I was on morphine and things like that, then I possibly wouldn’t have wanted too much detail.

P25 Orthopaedic patient ROH55

Model population

The patient population comprised adult individuals having already completed at least 3 months of AT in response to their first unprovoked VTE. Individual patients were generated from the ExACT patient data set with characteristics created by randomly sampling the patient-level data by means of a uniform distribution. Patient characteristics comprised age, sex and type of index VTE event (DVT or PE).

Model pathways and clinical events

The economic model extrapolated the results of the ExACT trial and compared the strategy of discontinuing anticoagulation with lifetime extended anticoagulation. The model contained the same potential pathways in both strategies (Figure 4), and their characteristics in part determined the probability of clinical events, costs and utilities. The model had a 1-month time cycle and lifetime time horizon.

FIGURE 4.

Model patient pathways.

In 1 month, an individual had a probability of experiencing a clinical event: death from other causes, recurrent VTE (non-fatal DVT, fatal or non-fatal PE), fatal or non-fatal major bleeds (intracranial bleed, gastrointestinal bleed, and other major bleeds). Recurrent VTE carried a risk of PTS. A recurrent VTE could be a PE or DVT and the recurrent VTE type was assumed to be influenced by an individual’s initial VTE site. Once a patient suffered a recurrent VTE, they were put on AT for life with cessation of therapy only if a later major bleeding event occurred. VTE events were assumed to incur a one-off quality-of-life reduction, with a proportion of surviving patients assumed to suffer from severe PTS for life.

All major bleeding events had short-term costs and quality-of-life decrements. In addition, an intracranial bleed was assumed to be associated with ongoing costs and a permanent quality-of-life decrement along with a sustained increased lifetime risk of other-cause mortality. For the ‘other major bleeds’ category, it was assumed that this heterogeneous category of bleeds should have the same costs and quality-of-life decrement as a gastrointestinal bleed.

Any major bleeding event led to discontinuation of AT. A recurrent VTE in a later cycle was assumed to restart therapy. It was assumed that there was no effect of AT on VTE recurrence risk more than 1 month post cessation of therapy.

Model type

A Markov patient-level simulation was developed in TreeAge 2014 (TreeAge software, Williamstown, MA, USA) and built on previous work on extended anticoagulation. 46 The patient-level simulation allowed the use of individual patient characteristics (age, sex, index VTE type) from the ExACT trial. The model was run with a large number of simulated patients (50,000) to account for interpatient variability.

Clinical parameters

The risk of a patient’s first recurrent VTE in the first 2 years was calculated by converting the trial primary outcomes for both arms from the rates per 100 person-years to an annual probability. 14 The annual risk of a further VTE event after a VTE recurrence on and off therapy was obtained from the PREVENT trial. 11 Only major bleeds were included in the model, and the risk of major bleeding on and off therapy was obtained from the ExACT trial data, converting rates per 100 person-years to annual probabilities. All annual probabilities were subsequently converted to monthly probabilities.

Resource use and costs

Costs of therapy and clinical events were included in the model. Costs of VTE and bleeding events were obtained from NHS Reference Costs 2012/13. 47 In line with the within-trial analysis, the cost hospital international normalised ratio (INR) monitoring was assumed and the cost per visit was obtained from previous research and updated (see Appendix 2). All costs were updated to 2012/13 prices using the Hospital and Community Health Services (HCHS) Index. 18

Quality of life

Quality-of-life (utility) values were assumed to be age related as they entered the model using EQ-5D-3L UK normative values. 12 Utility values for clinical events were multiplied by the age-specific utility to derive quality-of-life reductions for patients experiencing a clinical event. The model assumed no disutility with warfarin in the base case.

Assessment of cost-effectiveness

The incremental analysis calculated the cost per QALY gained for extended anticoagulation versus discontinued anticoagulation. Cost-effectiveness was assessed in relation to the NICE lower threshold of £20,000 per QALY, where a value of £20,000 per QALY is judged to be cost-effective. 48 All costs and outcomes were discounted at the recommended 3.5%. 13

Deterministic sensitivity analysis

To test the robustness of base-case results, deterministic sensitivity analyses were run to determine the impact of changing key parameters on results.

Probabilistic sensitivity analysis

Where available, data were input into the model as distributions to assess parameter uncertainty in the form of a probabilistic sensitivity analysis (PSA). The model was rerun with 10,000 simulations for each trial of 1000 simulated patients and the results expressed as cost-effectiveness planes and cost-effectiveness acceptability curves (CEACs).

Base-case results

Under base-case assumptions, lifetime extended anticoagulation was more costly but more effective than extended anticoagulation, with an incremental cost-effectiveness ratio (ICER) of £9530 per QALY gained.

Deterministic sensitivity analysis results

Shortening the time horizon to 5 years increased the ICER to £15,975 per QALY, which was still below the £20,000 per QALY threshold. Any further increase in time horizon beyond 5 years led to a reduction in the ICER. Increasing the risk of death from PE improved the cost-effectiveness of extended anticoagulation to £5836 per QALY. Attaching a disutility of 0.997 for being on anticoagulation increased the ICER, but it was still within the acceptable threshold. Changing the type and, therefore, cost of anticoagulation monitoring had the greatest impact on the ICER, with the scenario of all practice-led monitoring giving an ICER of £20,970 per QALY gained, slightly above the threshold, and an equal split of hospital and practice monitoring was £15,219 per QALY.

Probabilistic sensitivity analysis results

At a willingness-to-pay threshold of £20,000 per additional QALY, there is a 56% probability of extended anticoagulation being the cost-effective option, with a 60% probability at £30,000 per QALY.

Risk of serious bleeding with extended warfarin to avoid deep-vein thrombosis/pulmonary embolism

A total of 72 patients (37%) reported that they were willing to take a 100% risk of a bleed to stay on warfarin for a further 5 years to avoid a further DVT/PE. A total of 155 patients (80%) were willing to take greater than the stated 5% risk of a bleed, and only 15% (n = 28) were not willing to take any risk of serious bleed.

Risk of deep-vein thrombosis without extended warfarin to avoid serious bleeding

A total of 24 patients (12%) were willing to take a 100% risk of a DVT/PE to avoid a serious bleed. A total of 44 patients (23%) were willing to take greater than the stated 30% risk of a DVT/PE, and 84 patients (44%) were not willing to risk a DVT/PE to avoid a serious bleed.

Comparison of risk of bleeding and risk of deep-vein thrombosis

A total of 63 patients (33%) showed a very strong preference for extended warfarin by being willing to take a 100% risk of a bleed, but were not willing to take any risk of a DVT/PE. A total of 21 patients (11%) chose the opposite (i.e. had a very strong preference for not extending warfarin). Comparing the risk of DVT/PE with the risk of bleeding, the majority (135 patients, 70%) had a preference for avoiding DVT/PE (and extending warfarin); 40 patients (21%) preferred to avoid a serious bleed and the remaining 18 patients (9%) gave very similar values for both DVT/PE and bleeding.

The majority of participants were willing to take a higher risk of a bleed (compared with DVT/PE) to receive warfarin for 5 years. Twenty out of the 24 patients (83%) who suffered a DVT/PE and responded to the questionnaire preferred extended warfarin. Those who had suffered a major bleed reported only a slightly lower preference for extended warfarin (6/9, 67%).

Model pathways and clinical events

The economic model compared a strategy of no therapy (usual care) with a number of decision rule strategies, where therapy was restarted (effectively continued) if the predicted annual risk of VTE recurrence was equal to or greater than the given threshold risk. In line with the extended warfarin analysis, arbitrary but clinically relevant thresholds were explored in the analyses (1%, 3%, 5%, 7.5% 10%, 12.5%, 15%, 17.5%, 20%, 22.5% and 25%) and a treat-all strategy was also included as a comparator. The specified VTE risks were then used as different decision rule comparators.

Any major bleeding event led to discontinuation of AT and a subsequent recurrent VTE in a later cycle was assumed to restart therapy.

Clinical parameters

The model was run for each DOAC separately (i.e. apixaban, dabigatran, rivaroxaban) as well as warfarin. Owing to weak calibration statistics of the prognostic model after 3 years, beyond this point an annual constant risk for the first recurrent VTE event off therapy was used. 41

Resource use and costs

Costs of therapy and clinical events were included in the model, and details are provided in Appendix 2.

Quality of life

Quality-of-life (utility) values were assumed to be age related because they enter the model using EQ-5D-3L UK normative values. Utility values for clinical events and disutility on warfarin were multiplied by the age-specific utility to derive quality-of-life reductions for patients on warfarin or experiencing a clinical event. It was assumed that all three DOACs did not result in any disutility. Warfarin was assigned a disutility of 0.997.

Assessment of cost-effectiveness

A sequential incremental analysis was designed to calculate the cost per QALY gained for applying a decision rule versus the next most effective option, applying the rules of dominance and extended dominance. Cost-effectiveness was assessed in relation to NICE lower threshold of £20,000 per QALY, where a value of £20,000 per QALY is judged to be cost-effective.

Deterministic sensitivity analysis

To test the robustness of base-case results, deterministic sensitivity analyses were run to determine the impact of changing key parameters on results:

• The probability of death from a PE was increased to 30% because of uncertainty among clinical experts.

• Subgroup analysis was undertaken for index PE patients, as the subgroup of PE patients were at a higher risk of recurrence and mortality.

Base-case results

Under base-case assumptions, restarting warfarin therapy for patients with a predicted annual VTE recurrence risk of 25% gave the lowest cost per QALY of £2814. This recurrence risk was also the most cost-effective for apixaban, dabigatran and rivaroxaban, with ICERs ranging between £5520 and £6312 per QALY.

When considering a threshold of £20,000 per QALY gained, resuming AT with warfarin for patients with a predicted annual VTE recurrence risk of 17.5% yielded the highest number of QALYs while also being considered cost-effective with an ICER of £15,572 per QALY gained. Both dabigatran and rivaroxaban were cost-effective at a higher VTE recurrence risk of 20%, with a risk of 17.5% resulting in ICERs above the £20,000 threshold but below £30,000. The lowest risk threshold at which apixaban was cost-effective was even higher, at 22.5%.

Deterministic sensitivity analysis results

Increasing the risk of death from PE led to lower risk decision rule strategies also being cost-effective. This was ≥ 12.5% in the case of warfarin, and ≥ 17.5% for apixaban, dabigatran and rivaroxaban.

All decision rule strategies of ≥ 10% were cost-effective with extended warfarin when the patients’ index event was a PE, reflecting the high-risk nature of such index events. All strategies of ≥ 12.5% were cost-effective for dabigatran and rivaroxaban, and all strategies ≥ 15% were cost-effective for apixaban.

Methods

The economic evaluation took the form of a cost–utility analysis using QALYs based on the EQ-5D-3L instrument. Relevant patient costs used in the economic evaluation alongside the clinical trial were based on individual-level data collected during the 2-year follow-up period of the trial. Costs were estimated from an NHS perspective, and were based on 2012/13 values, consistent with the time period of data collection. In line with the main trial analysis, 273 patients were included in the economic evaluation, with 139 randomised to the extended anticoagulation group and 134 to the group discontinuing anticoagulation.

Cost and resource use

The cost analysis adopted a UK health-care perspective. Resource use data were collected alongside the trial. The total costs for each participant consisted of treatment costs (anticoagulation and monitoring), PTS costs, and adjudicated adverse event costs (thrombotic and haemorrhagic). Resource use concentrated on primary care, medicines, hospital consultant appointments, A&E attendances and hospital admissions.

As part of the treatment cost, the cost of anticoagulation has two components:

1. The cost of the medication was assumed to be the unit cost per tablet (per day) multiplied by the duration of treatment. An assumption on the average dose of warfarin as being 5 mg per day was made for costing purposes. Patients who were on a different anticoagulant (rivaroxaban) were costed in the same way using unit costs of that medication.

2. Visit for an INR test (if on warfarin). The total number of INR test visits per patient, based on individual resource use data, was multiplied by the unit cost per visit, assuming a hospital clinic visit.

For participants on anticoagulation, where data on anticoagulation visits were not available because of loss to follow-up, it was assumed that treatment continued for the full 24 months. The mean overall frequency of monitoring was assumed for the time frame where INR visit data were missing. Tables 10–12 present the unit costs related to treatment. PTS costs were based on visits to (or by) HCPs, and specified dressings and medications.

For adverse events, only events that were relevant to this study were costed, based on the resource use requirements of each event. Therefore, only adjudicated thrombotic events, haemorrhagic events and death (if resource use was utilised) were included in the costing, and unrelated adverse events were not included. For this reason, no data on chronic pulmonary hypertension were included in the analysis as no data for this outcome were collected within the ExACT trial.

For each event, data were collected on any relevant health-care contacts in primary and secondary care, and, although data on the ExACT trial length of hospitalisation were not reported, information was given if the stay in hospital was prolonged. Therefore, in the absence of data on actual length of stay for hospitalisation, the unit costs of thrombotic events were based on whether or not the event required hospitalisation and, if so, whether the hospitalisation was prolonged. If the event had not led to hospitalisation, the ‘event only’ cost was calculated as the difference between the long-stay cost and the product of weighted average length of stay in long stay and unit cost of an excess bed-day. If the adverse event occurred as a complication to an elective procedure, then only the cost of an estimated additional amount of time in hospital was included. Costs were based on NHS Reference Costs 2012/1347 (Table 13) and Unit Costs of Health and Social Care 201357 for visits to HCPs, as shown in Table 14. As the follow-up period was 2 years, all costs for year 2 were discounted at a rate of 3.5%.

Quality of life

The outcomes for the health economic analyses of the ExACT trial were based on QALYs estimated using the EQ-5D-3L instrument. A health utility is a continuum where values of zero and one are assigned to health conditions judged equivalent to death and optimal health, respectively. The EQ-5D-3L instrument comprises five questions dealing with aspects of physical and mental health (i.e. mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), for which the response is one of three possible degrees of impairment. The questionnaire was administered to study participants at baseline, 6, 12, 18 and 24 months. Data from the EQ-5D-3L were converted into utility values using UK preference-based utility weights. 58 Mean values for completed questionnaires were presented for each follow-up time point. Multiple imputation was used in cases of missing EQ-5D-3L data at one or more of the time points. 59 For participants who had died, a utility score of 0 was applied from the date of death and these participants were included in the complete-case analysis. The total QALY score for each study participant over their 24 months’ follow-up in the trial was estimated by calculating the area under each patient’s health utility curve using linear interpolation. To avoid bias, adjustment for differences in baseline EQ-5D-3L scores was undertaken using a regression-based adjustment. 60 All QALY scores reported in the base-case analysis reflect imputation and adjustment. As the follow-up period was 2 years, all QALYs for year 2 were discounted at a rate of 3.5%.

Cost-effectiveness analysis

An incremental cost-effectiveness analysis was conducted according to the intention-to-treat principle, to determine the difference in costs and outcomes between extended and discontinued anticoagulation. The unit of outcome was the incremental cost per QALY gained (incremental costs divided by incremental QALYs). Although the cost data were skewed, the arithmetic mean was calculated along with its non-parametric 95% CI. Mean differences in costs and outcomes were also presented with their non-parametric 95% CIs. To account for uncertainty because of sampling variation in cost-effectiveness, non-parametric bootstrapping was applied to the patient-level data to derive 5000 paired estimates of mean differences in costs and QALY scores. These were presented graphically on a cost-effectiveness plane and were used to construct a CEAC.

Costs

The costing analysis based on the trial data showed that the mean cost per participant during the 24-month trial period was £326.30 for the extended anticoagulation arm and £350.38 for the discontinued anticoagulation arm (Table 15). This demonstrates an overall mean cost saving of £24.08. As expected, mean costs in the intervention arm were greater for anticoagulation and haemorrhagic events, but much less for thrombotic events. Costs for treatment of PTS were very similar, in line with clinical results demonstrating no differences in PTS severity.

Outcomes

The baseline EQ-5D-3L score was higher in the discontinued anticoagulation arm (0.814) than in the extended anticoagulation arm (0.807) and slightly higher for the majority of time points (Table 16). Complete EQ-5D-3L data at all time points were available for 218 participants (80%), with a slightly higher rate in the extended anticoagulation arm (81%) compared with the discontinued anticoagulation arm (78%).

The differences in the mean imputed QALY score (adjusted for baseline differences) over 24 months’ follow-up were very similar in the two trial arms, although slightly favouring extended anticoagulation, with 0.009 additional QALYs (see Table 16). The complete-case analysis without any imputation and adjustment slightly favoured the discontinued anticoagulation arm. In all cases the 95% CIs around the mean differences did not suggest significance.

Base-case analysis

Table 17 provides the base-case results of the economic evaluation alongside the ExACT trial and reports the within-trial cost-effectiveness analysis undertaken using the adjusted and imputed data. Although differences in both cost and QALYs were small, the results favour the extended anticoagulation arm, with slightly lower costs and more QALYs.

The cost-effectiveness plane in Figure 6 highlights the uncertainty around the result, with cost–QALY difference points in all four quadrants. The CEAC in Figure 7 reports the probability of cost-effectiveness for extended anticoagulation over a range of willingness-to-pay thresholds up to £50,000 per QALY. The CEAC shows that at a willingness-to-pay threshold of £20,000 per additional QALY, there is a 62% probability of extended anticoagulation being the cost-effective option, with a 61% probability at £30,000 per QALY.

FIGURE 6.

Cost-effectiveness plane of extended AT vs. discontinued AT.

FIGURE 7.

The CEAC of extended AT vs. discontinued AT.

Model population

The patient population comprised adult individuals having already completed at least 3 months of AT in response to their first unprovoked VTE. Individual patients were generated from the ExACT patient data set with characteristics created by randomly sampling the patient-level data by means of a uniform distribution. Patient characteristics comprised age, sex and type of index VTE event (DVT or PE).

Model pathways and clinical events

The economic model extrapolated the results of the ExACT trial and compared the strategy of discontinuing anticoagulation with lifetime extended anticoagulation. The model contained the same potential pathways in both strategies (Figure 8), and their characteristics in part determining the probability of clinical events, costs and utilities. The model had a 1-month time cycle and lifetime time horizon.

FIGURE 8.

Model patient pathways.

In 1 month, an individual had the probability of experiencing a clinical event: death from other causes, recurrent VTE (non-fatal DVT, fatal or non-fatal PE), fatal or non-fatal major bleeds (intracranial bleed, gastrointestinal bleed, and other major bleeds). Recurrent VTE carried a risk of PTS. Other-cause mortality was dependent on the current age and sex of the patient and was taken from UK life tables. 61 Recurrent VTE risk depended on time spent in the model, previous history of a recurrent VTE event taking place in the model, and treatment status. A recurrent VTE could be a PE or DVT and the recurrent VTE type was assumed to be influenced by an individual’s initial VTE site. Once a patient suffered a recurrent VTE, they were put on AT for life with cessation of therapy only if a later major bleeding event occurred. VTE events were assumed to incur a one-off quality-of-life reduction, with a proportion of surviving patients assumed to suffer from severe PTS for life.

The risk factors for a major bleed in the model were presence of anticoagulation treatment and age if on treatment, with major bleeds categories as gastrointestinal bleeds, intracranial bleeds and ‘other major bleeds’. All major bleeding events had short-term costs and quality-of-life decrements. In addition, an intracranial bleed was assumed to be associated with ongoing costs and a permanent quality-of-life decrement along with a sustained increased lifetime risk of other-cause mortality. For the ‘other major bleeds’ category, it was assumed that this heterogeneous category of bleeds should have the same costs and quality-of-life decrement as a gastrointestinal bleed.

Any major bleeding event led to discontinuation of AT. A recurrent VTE in a later cycle was assumed to restart therapy. It was assumed that there was no effect of AT on VTE recurrence risk more than 1 month post cessation of therapy.

Model type

A Markov patient-level simulation was developed in TreeAge 2014 and built on previous work on extended anticoagulation. 46 The model was adapted was to estimate the cost-effectiveness of extending AT versus discontinuing AT (usual care) in patients with a first unprovoked VTE event. A Markov model was appropriate here as it can represent a clinical situation where patients move between health states over a long period of time. The patient-level simulation allowed the use of individual patient characteristics (age, sex, index VTE type) from the ExACT trial, and these characteristics and clinical events that affect subsequent risks were remembered in the model with tracker variables. The model was run with a large number of simulated patients (50,000) to account for interpatient variability.

A time cycle of 1 month was selected to represent an assumption that this reflects a period in which a single clinical event might occur. Costs, utilities and probabilities were transformed into monthly equivalents as per the time cycle length. A half-cycle correction was applied to costs and effects. The base-case cost–utility analysis was undertaken from a UK NHS/Personal Social Services (PSS) perspective and considered a lifetime horizon.

Clinical parameters

Parameter estimates and their sources are listed in Table 18. The risk of a patient’s first recurrent VTE in the first 2 years was calculated by converting the trial primary outcomes for both arms from the rates per 100 person-years to an annual probability. This probability of VTE recurrence on therapy was assumed to be the same beyond 2 years, whereas the risk of VTE recurrence off therapy beyond 2 years was assumed to fall to 5%. 14 The annual risk of a further VTE event after a VTE recurrence on and off therapy was obtained from the PREVENT trial. 11 The probability of the type of any recurrent VTE being a PE was dependent on the index event, using primary data from the ExACT trial. Only major bleeds were included in the model, and the risk of major bleeding on and off therapy was obtained from ExACT trial data, converting rates per 100 person-years to annual probabilities. All annual probabilities were subsequently converted to monthly probabilities.

Resource use and costs

Costs of therapy and clinical events were included in the model (Figure 9 and Table 19). Costs of VTE and bleeding events were obtained from NHS Reference Costs 2012/1347. In line with the within-trial analysis, the cost hospital INR monitoring was assumed and the cost per visit was obtained from previous research and updated. 45 All costs were updated to 2012/13 prices using the HCHS Index. 18

FIGURE 9.

Model patient pathways.

Quality of life

Quality-of-life (utility) values were assumed to be age related as they entered the model using EQ-5D-3L UK normative values. 64 As patients aged in the model, their utility score changed to reflect their updated quality of life for their age. Utility values for clinical events (Table 20) were multiplied by the age-specific utility to derive quality-of-life reductions for patients experiencing a clinical event. The model assumed no disutility with warfarin in the base case.

Assessment of cost-effectiveness

The incremental analysis calculated the cost per QALY gained for extended anticoagulation versus discontinued anticoagulation. Cost-effectiveness was assessed in relation to the NICE lower threshold of £20,000 per QALY, where a value of £20,000 per QALY is judged to be cost-effective. 48 All costs and outcomes were discounted at the recommended 3.5%. 13

Deterministic sensitivity analysis

To test the robustness of base-case results, deterministic sensitivity analyses were run to determine the impact of changing key parameters on results.

• The model time horizon was restricted to 5, 10 and 15 years.

• The utility of warfarin therapy was reduced to 0.997 from 1.

• The probability of death from a PE was increased to 30% because of uncertainty among clinical experts on the true value.

• The cost of anticoagulation was increased to assume all patients were monitored in primary care and the site of monitoring was equally split between hospital- and general practice-led monitoring.

Probabilistic sensitivity analysis

When available, data were input into the model as distributions to assess parameter uncertainty in the form of a PSA. The model was rerun with 10,000 simulations for each trial of 1000 simulated patients and the results expressed as cost-effectiveness planes and CEACs.

Base-case results

Under base-case assumptions, lifetime extended anticoagulation was more costly but more effective than extended anticoagulation, with an ICER of £9530 per QALY gained (Table 21).

Deterministic sensitivity analysis results

Deterministic sensitivity scenario results are shown in Table 22. Shortening the time horizon to 5 years increased the ICER to £15,975 per QALY, but this was still below the £20,000 per QALY threshold. Any further increase in time horizon beyond 5 years led to a reduction in the ICER. Increasing the risk of death from PE improved the cost-effectiveness of extended anticoagulation to £5836 per QALY. Attaching a disutility of 0.997 for being on anticoagulation increased the ICER, but it was still within the acceptable threshold. Changing the type and, therefore, cost of anticoagulation monitoring had the greatest impact on the ICER, with the scenario of all practice-led monitoring giving an ICER of £20,970 per QALY gained, which was slightly above the threshold, and an equal split of hospital and practice monitoring was £15,219 per QALY.

Probabilistic sensitivity analysis results

The cost-effectiveness plane in Figure 10 highlights the uncertainty around the result, with cost–QALY difference pairs all both north-west and north-east quadrants. The CEAC in Figure 11 reports the probability of cost-effectiveness for extended anticoagulation over a range of willingness-to-pay thresholds up to £100,000 per QALY. The CEAC shows that at a willingness-to-pay threshold of £20,000 per additional QALY, there is a 56% probability of extended anticoagulation being the cost-effective option, with a 60% probability at £30,000 per QALY.

FIGURE 10.

Cost-effectiveness plane of extended AT vs. discontinued AT.

FIGURE 11.

The CEAC of extended AT vs. discontinued AT.

Research aims

The aim of this research was to determine patient preferences for extended anticoagulation treatment, taking into account both the benefits (reduction in thrombotic events) and risks (bleeding events) associated with anticoagulation.

Methods for data collection and analysis

A questionnaire was developed to elicit patient preferences for the extension of anticoagulation (specified as warfarin) (see Appendix 1). This used a technique regularly used to represent risks and benefits on anticoagulation as part of a decision tool, where 100 faces were represented. 49

The questionnaire was in two parts, looking at the risk of clots and risk of bleeding separately. Participants were first shown a representation of current evidence where the risk of DVT was 30% over 5 years without warfarin, and 0% with warfarin. They were then asked for the risk of serious bleeding they were willing to take over the 5 years by staying on warfarin (and, therefore, avoiding DVT), by crossing out faces on the picture. The second half of the questionnaire showed a representation of current evidence on serious bleeding, with 0% risk with no warfarin and 5% risk with warfarin. They were then asked what risk of DVT they were willing to take to avoid being on warfarin (and, therefore, avoiding serious bleeding), again by crossing out faces on the picture.

Questionnaires were completed by patients during their final follow-up study appointment, with assistance from the study nurses, with completed questionnaires entered onto a database.

An initial descriptive analysis was undertaken to determine the distribution of responses for both questions separately. Subsequently, both questions were considered together to determine whether a participant showed (1) a preference for warfarin (i.e. willing to risk a higher risk of bleeding than DVT), (2) a preference for no warfarin (i.e. willing to take a higher risk of DVT than bleeding) or (3) no preference (i.e. chose equivalent risks). Answers where it was clear that the participant had not understood the questionnaire, by offering no answer or responding 0% or 100% to both options, were reported separately. A descriptive analysis of counts and percentages was then presented. Finally, these data were analysed taking into account the following patient characteristics:

• age

• sex

• trial arm

• thrombotic event during the trial

• bleeding event during the trial.

This was to determine whether or not any pattern was emerging in relation to demographics, and whether or not allocation of treatment or an event during the trial had influenced their choice.

Risk of serious bleeding with extended warfarin to avoid deep-vein thrombosis/pulmonary embolism

Here responses ranged from 0% to 100%, with 72 participants (37%) reporting they were willing to take a 100% risk of a bleed to stay on warfarin for a further 5 years and to avoid a further DVT/PE. A total of 80% (n = 155) were willing to take greater than the stated 5% risk of a bleed, and only 15% (n = 28) were not willing to take any risk of serious bleed.

Risk of deep-vein thrombosis without extended warfarin to avoid serious bleeding

Again, responses ranged from 0% to 100% risk, with 24 participants (12%) willing to take a 100% risk of a DVT/PE to avoid a serious bleed. A total of 23% (n = 44) were willing to take greater than the stated 30% risk of a DVT/PE, and 44% (n = 84) were not willing to risk a DVT/PE to avoid a serious bleed.

Comparison of risk of bleeding and risk of deep-vein thrombosis

The responses to both questions were then considered together. A total of 63 participants (33%) showed a very strong preference for extended warfarin by being willing take a 100% risk of a bleed, but not willing to take any risk of a DVT/PE. Only 21 (11%) chose the opposite (i.e. a very strong preference for not extending warfarin). Comparing the risk of DVT/PE with bleeding, the majority (70%, n = 135) showed a preference for avoiding DVT/PE (and extending warfarin), 21% (n = 40) preferred to avoid a serious bleed, and the remaining 18 (9%) gave very similar values for both DVT/PE and bleed.

The majority of participants were willing to take a higher risk of a bleed (compared with DVT/PE) to receive warfarin for 5 years (Table 24). This did not differ greatly in terms of sex, age group or trial arm, although men and those aged ≥ 65 years had a slightly greater preference for extended warfarin. Incidence of a DVT/PE in the trial period had an impact, with 20 out of the 24 (83%) who suffered a DVT/PE and responded to the questionnaire preferring extended warfarin. Those who had suffered a major bleed only reported a slightly lower preference for extended warfarin (68%, 6/9).

Model type

The Markov patient-level simulation was developed in TreeAge 2014 (TreeAge software, Williamstown, MA, USA) to estimate the cost-effectiveness of using a decision rule for restarting AT with a DOAC versus no AT (usual care) in patients with a first unprovoked VTE event. The patient-level simulation allowed the use of patient-level data to create individual patients, each with a set of varying characteristics, and to be assigned a risk of VTE recurrence. Patient characteristics and clinical events that affect subsequent risks were remembered in the model with tracker variables. The model was run with a large number of simulated patients (50,000) to account for variability between patients.

A time cycle of 1 month was selected to represent a period in which a single clinical event might occur. Costs, utilities and probabilities were transformed into monthly equivalents as per the time cycle length. A half-cycle correction was applied to costs and effects. The base-case cost–utility analysis was undertaken from a UK NHS/PSS perspective and considered a lifetime horizon.

Model population

The modelled patient population was adult individuals having already completed at least 3 months of AT after their first unprovoked VTE. Patients entered the model having already had their D-dimer level measured 30 days after stopping at least 3 months of AT. Individual patients were generated from patient data (recurrent VTE collaborative database), which was used to develop the prognostic model. 67 Each patient had characteristics created by randomly sampling the patient-level data by means of a uniform distribution. Patient characteristics comprised age (mean 61.7 years), sex (61.8% male), type of index VTE event (distal DVT 9.2%, proximal DVT 58.5%) and PE (32.3%), and post-anticoagulation D-dimer level (mean 667.3 µg/l). The individual’s risk of a recurrent VTE within 12 months was then determined by inputting their newly created characteristics into the prognostic model risk equation.

Model pathways and clinical events

The economic model compared a strategy of no therapy (usual care) with a number of decision rule strategies, where therapy was restarted if the predicted annual risk of VTE recurrence was equal to or greater than the given threshold risk. In line with the extended warfarin analysis, arbitrary but clinically relevant thresholds were explored in the analyses: 1%, 3%, 5%, 7.5% 10%, 12.5%, 15%, 17.5%, 20%, 22.5%, 25% and a treat-all strategy was also included as a comparator. The specified VTE risks were then used as different decision rule comparators. Once the decision rule was applied, all the patients encountered the same potential pathways in all strategies (see Figure 9), with their characteristics determining the probabilities of clinical events, costs and utilities.

In each 1-month time cycle, an individual had the probability of experiencing one clinical event: death from other causes, recurrent VTE (non-fatal distal or proximal DVT, fatal or non-fatal PE), or fatal or non-fatal major bleeds (intracranial bleed, gastrointestinal bleed and other major bleeds). A recurrent VTE carried a risk of PTS. Other-cause mortality was dependent on the current age and sex of the patient and was taken from UK life tables. 61 Recurrent VTE risk depended on a patient’s characteristics, time spent in the model, previous history of a recurrent VTE event taking place in the model, and treatment status. A recurrent VTE could be a PE, distal DVT or proximal DVT. The recurrent VTE type was assumed to be affected by the site of an individual’s initial VTE. Once a patient suffered a recurrent VTE, they were put on AT for life, with therapy cessation only occurring with a later major bleeding event. VTE events were assumed to incur a one-off quality-of-life reduction, with a proportion of surviving patients assumed to suffer from severe PTS for life.

Treatment status and age (if on treatment) were considered to be risk factors for a major bleed, and bleeds were split into gastrointestinal bleeds, intracranial bleeds and other major bleeds. All major bleeding events had short-term costs and quality-of-life decrements. In addition, an intracranial bleed was assumed to be associated with ongoing costs and a permanent quality-of-life decrement along with a sustained increased lifetime risk of other-cause mortality. For model simplification purposes, it was agreed that the other major bleeds category should have the same costs and quality-of-life decrement as a gastrointestinal bleed.

Any major bleeding event led to discontinuation of AT, and a subsequent recurrent VTE in a later cycle was assumed to restart therapy. It was assumed that there was no effect of AT on VTE recurrence risk after 30 days post cessation of therapy.

Clinical parameters

The model was run for each DOAC separately (i.e. apixaban, dabigatran, rivaroxaban) as well as warfarin. The risk of a patient’s first recurrent VTE off therapy was calculated using the prognostic model for up to 3 years post D-dimer measurement (30 days after initial therapy cessation). Owing to weak calibration statistics of the prognostic model after 3 years, beyond this point an annual constant risk for the first recurrent VTE event off therapy was used. 46 Annual risk of a further VTE event after a VTE recurrence was an average of values for patients with normal and elevated D-dimer levels, on and off therapy respectively in the PREVENT trial. 14 The original model uses estimates on warfarin, and in this model it is assumed that these same risks of VTE and bleed on treatment are applied to each DOAC.

Resource use and costs

Costs of therapy and clinical events were included in the model (Table 25). The cost of a D-dimer test was incurred by the decision rule strategies as the D-dimer information was needed to enact the decision rules (Table 26). The 2012/13 costs in the original model were inflated to 2016/2017 prices using the HCHS Index, so that the most recent drug costs could be included. 68

Quality of life

Quality-of-life (utility) values were assumed to be age related as they enter the model using EQ-5D-3L UK normative values. 73 As patients aged in the model, their utility score changed to reflect the updated quality of life for their age. Utility values for clinical events (Table 27) and disutility on warfarin were multiplied by the age-specific utility to derive quality-of-life reductions for patients on warfarin or experiencing a clinical event. It was assumed that all three DOACs did not result in any disutility. Warfarin was assigned a disutility of 0.997.

Assessment of cost-effectiveness

A sequential incremental analysis was designed to calculate the cost per QALY gained for applying a decision rule versus the next most effective option, applying the rules of dominance and extended dominance. Cost-effectiveness was assessed in relation to the NICE lower threshold of £20,000 per QALY, where a value of £20,000 per QALY is judged to be cost-effective. 26 Strategies were compared by increasing effectiveness, and ICERs were calculated from the difference in costs and effects between a decision rule strategy and the next best alternative. A strategy is said to be dominated if it is more expensive and less effective than a comparator. All costs and outcomes were discounted at the recommended 3.5%. 26

Deterministic sensitivity analysis

To test the robustness of base-case results, deterministic sensitivity analyses were run to determine the impact of changing key parameters on results. Results from the original model suggested that the model was particularly sensitive to parameters relating to PE; therefore, additional analysis concentrated on these parameters.

• The probability of death from a PE was increased to 30% because of uncertainty among clinical experts.

• Subgroup analysis was undertaken for index PE patients, as the subgroup of PE patients were at a higher risk of recurrence and mortality.

Base-case results

Tables 28–31 present the base-case results for each anticoagulant, where a sequential incremental analysis was undertaken. Under base-case assumptions, restarting warfarin therapy for patients with a predicted annual VTE recurrence risk of 25% gave the lowest cost per QALY of £2814. This recurrence risk was also the most cost-effective for apixaban, dabigatran and rivaroxaban, with ICERs ranging between £5520 and £6312 per QALY.

When considering a threshold of £20,000 per QALY gained, resuming AT with warfarin for patients with a predicted annual VTE recurrence risk of 17.5% yielded the highest number of QALYs while also being considered cost-effective with an ICER of £15,572 per QALY gained. Both dabigatran and rivaroxaban were cost-effective at a higher VTE recurrence risk of 20%, with a risk of 17.5% resulting in ICERs above the £20,000 threshold but below £30,000. The lowest risk threshold where apixaban was cost-effective was even higher, at 22.5%.

Deterministic sensitivity analysis results

Deterministic sensitivity scenario results are shown in Tables 32 and 33. These results illustrate that increasing the risk of death from PE led to lower risk decision rule strategies also being cost-effective. This was ≥ 12.5% in the case of warfarin, and ≥ 17.5% for apixaban, dabigatran and rivaroxaban.

All decision rule strategies of ≥ 10% were cost-effective with extended warfarin when the patient’s index event was a PE reflecting the high-risk nature of such index events. All strategies of ≥ 12.5% were cost-effective for dabigatran and rivaroxaban, with all strategies ≥ 15% cost-effective for apixaban.