Address correspondence to: Linda Davies, Reader, Health Economics, School of Psychiatry and Behavioural Sciences, Manchester Royal Infirmary, Oxford Road, Manchester M13 9WL United Kingdom. E-mail: Linda.Davies@man.ac.uk
Objectives: Clinical trials indicate enoxaparin thromboprophylaxis (Clexane) can be effective and safe when used in an outpatient setting and that extending the length of thromboprophylaxis with enoxaparin to the postdischarge period may be more effective than inpatient thromboprophylaxis alone. This may increase the cost of thromboprophylaxis. The objective of the study was to estimate the expected cost-effectiveness of using enoxaparin for hospital admission only vs. enoxaparin for hospital admission and for 21 days postdischarge.
Methods: Decision analysis was used to combine probability, resource use and unit cost data, using the framework of cost-effectiveness analysis. The model used a societal perspective to estimate the expected costs of treatment and outcomes to patients undergoing orthopedic surgery for elective hip replacement. Incremental cost-effectiveness ratios were calculated to provide estimates of the cost per life gained, cost per year life year gained and cost per quality-adjusted life year gained with extended use of enoxaparin thromboprophylaxis.
Results: There was an expected cost per quality-adjusted life year gained of £5732 associated with extended enoxaparin thromboprophylaxis. The results were sensitive to the percentage of patients who could administer enoxaparin injections at home, the rate of DVT associated with standard enoxaparin thromboprophylaxis and the rate of PE associated with standard and extended enoxaparin thromboprophylaxis.
Conclusions: The analyses indicated that in most cases extended enoxaparin thromboprophylaxis resulted in increased costs for health care services. In all cases, extended thromboprophylaxis with enoxaparin was associated with improved survival and life-years gained.
Deep Vein Thrombosis (DVT) is a common complication of major surgery or serious illness. DVT is associated with a number of sequelae and events that are clinically important and affect patient survival and quality of life. These include fatal and nonfatal pulmonary embolism (PE), recurrence of thromboembolic events and long-term complications such as the post–thrombotic syndrome [1,2]. The postoperative risk of fatal PE is uncertain with estimates ranging from 0.1% to 5%[3,4]. The risk of DVT and its complications continues for at least several weeks postoperatively.
The resource use associated with diagnosis and treatment of DVT is significant, not least because the length of inpatient stay following surgery can be considerably increased. In addition, readmission may be required. Because of the cost, morbidity and mortality associated with the sequelae of DVT, thromboprophylaxis with mechanical methods, unfractionated heparin (UFH) or low molecular weight heparin (LMWH) is an important component of clinical guidelines in the UK [e.g. 5,6].
The method of thromboprophylaxis depends on the likely risk of DVT, which varies between patient groups, and surgical procedures. For patients, undergoing elective hip replacement, who are at high risk of DVT, chemical thromboprophylaxis with UFH or LMWH is recommended [1,4–6]. Furthermore, evidence suggests that LMWH is an effective and cost-effective prophylaxis for the prevention of DVT in this high risk group [7–15]. Recently, clinical trials have also indicated that LMWH is an effective and safe treatment for DVT when administered at home rather than hospital [16,17].
Late venous complications occur because of prolonged impaired hemostasis and hemodynamics, postsurgical venous injury and poor mobilization following discharge. Recent clinical evidence suggests that extending the length of thromboprophylaxis with LMWH to the postdischarge period may be more effective than inpatient thromboprophylaxis alone [18–21].
In the UK, thromboprophylaxis has become an important part of surgical patient management, pre- and postoperatively [22,23]. A recent UK survey indicated that 99% of surgeons used routine thromboprophylaxis for patients undergoing elective hip replacement and 79% used a combination of mechanical and chemical methods. Nearly two thirds of respondents routinely used LMWH for chemical thromboprophylaxis . For the UK, the extent to which patients receive thromboprophylaxis on discharge from hospital is unclear. A recent study in the US indicated that approximately 60% of patients who had elective hip replacement surgery were discharged with some form of thromboprophylaxis. Of these, LMWH was prescribed for about a third of patients .
Extended thromboprophylaxis with LMWH would increase the drug costs of elective hip replacement. However, recent economic analyses indicate that extended thromboprophylaxis with enoxaparin may be cost-effective in Sweden and France, by reducing the incidence of postdischarge DVT and hospital readmissions [25,26]. The aim of this paper is to assess the expected cost-effectiveness in the United Kingdom of enoxaparin used for extended prophylaxis postdischarge for DVT in hip surgery compared to enoxaparin administration prior to hospital discharge only.
The objective of the study was to estimate the incremental expected cost-effectiveness of using enoxaparin for the index hospital admission and 21 days postdischarge (extended enoxaparin) compared to using enoxaparin for the index hospital admission only (standard enoxaparin). The 21-day postdischarge period for extended enoxaparin was chosen to reflect the timing of the primary assessment in the clinical trial used for the analysis .
The study addresses a limited question in the management of patients undergoing elective hip replacement. That is, whether extended thromboprophylaxis with enoxaparin is cost-effective compared to no extended thromboprophylaxis in the UK setting. It is important to assess whether the findings of economic evaluations of extended enoxaparin also apply in the UK, particularly if patients do not routinely receive thromboprophylaxis on discharge from hospital following elective hip replacement surgery. In addition, there is insufficient clinical evidence to compare the relative effectiveness of enoxaparin with alternative methods of extended thromboprophylaxis after hospital discharge, for this patient group.
A societal perspective was used [27,28]. In the United Kingdom the key components for this perspective are the costs to the National Health Service (NHS) of providing treatment and care and the outcomes to patients. The NHS is funded by general taxation and is the principal provider of health care, which is free at the point of consumption.
The model only considered the expected cost-effectiveness for patients undergoing orthopaedic surgery for elective hip replacement. Decision analysis was used to combine probability, resource use and unit cost (price) data . This approach allowed a range of clinical alternatives and clinical outcomes to be described and expected costs and consequences to be estimated.
Figure 1 presents the decision tree for events and management choices. The decision tree starts at the point at which the decision has been taken to give enoxaparin prophylaxis against DVT. The choice is then between extended enoxaparin and standard enoxaparin. There is insufficient evidence to support differences in the sequence of events between the two choices, which are therefore assumed to be identical. On this basis, only the extended enoxaparin arm is illustrated in full. However, clinical trials evidence does suggest that the probability of one or more subsequent events may differ between the two choices [18,19]. The decision tree extends the model developed for the UK setting, which was used in a previous economic analysis of enoxaparin .
The expected costs and benefits of the standard and extended regimes of thromboprophylaxis were estimated for a cohort of 1000 patients undergoing elective hip replacement. The expected outcomes of thromboprophylaxis were estimated using three measures. These were lives gained, life-years gained and quality-adjusted life-years gained.
The costs of events were estimated as the quantity and type of health care resources used, multiplied by the unit costs or prices of those resources. All estimates of unit cost were for 1997–8 prices. Where necessary, unit costs were inflated to 1997–98 prices using the Hospital and Community Health Services Index .
The costs included in the analysis were for extended thromboprophylaxis and treatment of events within the 21-day postdischarge period. It was not necessary to discount this. The long-term patient outcomes of life-years gained and QALYs were discounted to present values. The discount rate for the base case analysis was 1.5% (the rate recommended by the UK Treasury). In the sensitivity analysis, these were varied between 0% and 5%[27,28].
The data for the model were based on point estimates from a variety of sources. However, there was insufficient information to generate plausible distributions for the data. This meant that it was not possible to use statistical techniques to assess the level of uncertainty associated with the results. This could arise from uncertainty in the data used, such as the probability of DVT from a limited number of trials, or imprecision in the measurement or estimation of data, such as the resources used to treat DVT.
Sensitivity analysis was used to test the impact on the results of changing those variables where the use of a minimum or maximum value would increase the cost and/or decrease the benefits associated with extended enoxaparin.
Clinical Trials Data
Two randomized, double-blind, placebo-controlled trials have been conducted to evaluate the outcome of extended enoxaparin thromboprophylaxis in patients undergoing elective hip replacement [18,19]. In both trials, patients were given enoxaparin 40 mg once daily by subcutaneous injection for the index hospital admission. Prior to discharge, patients were randomized to receive either once daily 40 mg enoxaparin subcutaneous injection or a placebo (saline) injection on an outpatient basis for 21 days. All patients were evaluated at follow up using objective tests (bilateral venography). Both trials found statistically significant differences in the rates of DVT between active and placebo groups in favour of extended enoxaparin. No statistically significant differences in adverse events were found in either trial. However, the sample sizes were calculated to detect differences in the rate of DVT, not adverse events. There were a number of key differences in the design of the trials that meant that the trial by Bergqvist et al.  was a more appropriate source of clinical data for an economic analysis in the UK setting. The trial by Planes et al.  used bilateral ascending venography on all patients prior to discharge following surgery and subsequent enrollment in the trial. Patients who did not have a normal venogram were excluded from the study. The use of venography was specific to the trial and not common practice [18,30]. In addition, it has been argued that venography may be a confounding factor in the analysis of differences between the intervention and control groups [18,30]. In contrast, Bergqvist et al. did not use an objective test to assess patients' thromboembolic status prior to discharge from the index admission.
Secondly, the trial by Planes et al.  only included patients who could walk without assistance other than crutches. It has been suggested that the incidence of DVT in the control group was high for this subset of patients undergoing elective hip replacement, given the strict eligibility criteria .
Probability of Events
For the base case analysis, data on the probability of confirmed DVT, PE, and hospital readmission were estimated from the Bergqvist et al. trial  comparing standard and extended enoxaparin (Table 1). For the sensitivity analysis, the minimum and maximum rates of DVT for both standard and extended enoxaparin thromboprophylaxis were assumed to be between 18% and 39%. These are the rates of DVT found in the clinical trial by Bergqvist et al. . In addition, the sensitivity analysis used the rates of DVT and PE from the Planes et al. trial , to assess the impact of a lower incidence of DVT on the results.
The probability of clinically and economically important DVT may have been over-estimated in the clinical trial, which used objective diagnosis to detect all DVTs. To adjust the objectively diagnosed rates of DVT, the decision tree included the probability that the patient had signs and symptoms of DVT allowing a clinical diagnosis to be made. The probabilities of a clinical diagnosis of PE, accuracy of clinical diagnosis and survival were taken from the previous economic analysis of enoxaparin in the UK , which were also used in the economic evaluation of extended enoxaparin in the French setting .
The number of lives gained per 1000 cohort of patients undergoing elective hip replacement with extended enoxaparin was calculated as 1000 multiplied by the net expected probability of survival. The number of life-years gained with extended enoxaparin was calculated as the number of lives gained following surgery multiplied by the average life expectancy for the cohort. The average (median) age at surgery of patients enrolled in the clinical trial used for this analysis was 70 . The life expectancy of people aged 70 was estimated to be 81 years using national statistics for the UK .
National surveys conducted for the government Office of National Statistics  indicate that the self-reported general health of the population declines with age. In the 1998 survey, 78%–86% of people aged 16–54 reported their general health to be good or very good, compared to 57%–68% of people aged 55 years or over. In addition, 52%–66% of people over the age of 55 reported a long-standing illness compared to 22%–37% of people aged 16–54.
These data, and the older age of the population for this analysis, indicate that gains in survival and life-years from the use of extended enoxaparin will over-estimate the potential health gains which could be realized. Therefore, the number of life-years gained were weighted by an age-specific health-related utility value to estimate quality-adjusted life-years (QALYs) gained. The utility values used for the analysis were those reported by a national survey of self-reported health status and health-related quality of life . The instrument used in the survey was the EuroQol, a validated measure of health status and utility. The utility weights reported were generated from a population survey using the time trade off technique. It was assumed that the utility value would be the same for DVT/PE and non-DVT/PE patients. This reduces the potential QALY gain from any reductions in nonfatal DVT or PE that may be associated with the use of extended enoxaparin.
A search of the literature was conducted to estimate the use of inpatient and community-based health care services that were specific to the UK setting (Table 2). For the base case analysis, it was estimated that all patients with symptoms of DVT or PE would have objective diagnostic evaluations to confirm the clinical diagnosis .
Table 2. Resource use: thromboprophylaxis, diagnosis and treatment
Furthermore, it was estimated that asymptomatic patients would not have objective tests to detect DVT or PE . The rate of readmission for the diagnosis and treatment of DVT occurring in the community and the average length of stay per readmission was estimated from observed rates reported in the clinical trial by Bergqvist et al. . In the absence of data to the contrary, the following assumptions were made. First, the length of treatment for DVT for patients treated at home was assumed to be the same as that for patients admitted to hospital. Second, for the base case analysis it was conservatively assumed that none of the patients or their carers would administer enoxaparin thromboprophylaxis themselves and would require one district nurse visit per day of thromboprophylaxis. This increases the costs of extended enoxaparin thromboprophylaxis, and may over-estimate the total costs of this option. This assumption was tested in the sensitivity analysis by varying the proportion of patients or carers who could administer enoxaparin thromboprophylaxis from 0% to 100%[16,34,35]. Thirdly, it was assumed that the cost of monitoring associated with warfarin therapy would be the same for both groups and was excluded from the analysis.
However, the costs of weekly platelet counts to monitor for drug-related adverse events were included in the extended enoxaparin arm. The clinical trials of extended enoxaparin did not identify any statistically significant differences in hematological adverse events between the intervention and control arms. On this basis these events were excluded from the analysis. However, there may be concern that extended enoxaparin could potentially increase the risk of these events, leading to increased use of laboratory based monitoring.
Finally, it was also assumed that all patients undergoing elective hip replacement would have three visits from a community-based health care professional to monitor progress. This may over-estimate the costs of the standard enoxaparin group, and the cost-effectiveness of extended enoxaparin. Again this was tested in the sensitivity analysis, by excluding the costs of these visits from the standard enoxaparin group.
Unit Cost Data
The unit costs of enoxaparin (40 mg, 0.4 ml syringe) and warfarin, compression stockings, syringes and needles were estimated from the published retail prices (Table 3). No allowance was made for hospital discounts, which can vary considerably between pharmacy departments.
Table 3. The costs of resource use per event (£'s, 1997-98)
The costs of blood tests, objective diagnosis and hospital inpatient stay were estimated from the tariffs of NHS provider units [37–39]. The costs per hour of hospital-based health care professionals were estimated from the reports of national review bodies [40,41]. The costs of district nurse home visits and general practitioner surgery consultations were estimated from published costs .
Table 4 presents the expected costs and outcomes associated with the use of extended enoxaparin and standard enoxaparin as prophylaxis for DVT. For the base case analysis, the expected cost per 1000 individuals was £428,139 for extended enoxaparin and £184,244 for standard enoxaparin, giving a net cost of £243,895 per 1000 people for extended enoxaparin. In contrast, extended enoxaparin was associated with a reduction in mortality and expected benefit of 6 lives gained per 1000 patients. For the extended enoxaparin group, this translated to a cost per life gained of £42,898, cost per life-year gained of £4257 and cost per quality-adjusted life-year (QALY) gained of £5732.
Table 4. Expected costs and outcomes of extended and standard enoxaparin, per 1000 patients
Expected cost & survival
Incremental cost-effectiveness ratio
Cost/life yr gained
QALY: quality adjusted life year.
Discount rate 1.5% (base case)
Discount rate 0%
Discount rate 5%
The sensitivity analyses indicated a large range in the net expected costs and benefits of extended enoxaparin. The majority (88%) of the sensitivity analyses estimated the cost per quality-adjusted life-year gained to be below £7000. The analysis was sensitive to the values assigned to the percentage of patients or carers who could administer enoxaparin injections at home, the rate of PE for standard and extended enoxaparin and the rate of DVT with standard enoxaparin. Figures 2–4 present the results of the sensitivity analyses for these variables.
If more than 10% of patients or informal carers could administer enoxaparin injections at home, the expected cost/QALY gained would range between £144 and £5100 (Fig. 2). In contrast, it could be assumed that only the rate of DVT will vary between standard and extended enoxaparin, and that the rate of PE, for patients with DVT is the same for both thromboprophylaxis regimes . Figure 3 presents the results of the sensitivity analysis for alternative rates of PE following DVT, when it is assumed that the rate if PE is equivalent between the two interventions. In this analysis, the expected cost per QALY gained with extended enoxaparin thromboprophylaxis could range from £4500 to £27,000.
Finally, if the rate of DVT associated with standard thromboprophylaxis with enoxaparin is lower than that found in the Bergqvist trial, then the expected cost/QALY gained by extended enoxaparin will be between £6000 and £19,700 (Fig. 4). If the rates of DVT from the trial by Planes et al.  are used, the cost/QALY gained would be £13,365.
The base case and sensitivity analyses indicated that in most cases extended enoxaparin was associated with increased costs for health care services, when compared to standard enoxaparin. However, in all cases, extended enoxaparin was associated with net benefits in terms of improved survival and life-years gained.
These results need to be treated with caution. The data used for the model were not collected with the objective of estimating the cost-effectiveness of extended enoxaparin. The majority of the data for the model were point estimates from a limited number of sources or derived from assumptions. Given the nature of the data it was not feasible to construct a valid stochastic simulation model, which would have allowed statistical analysis of the results and corresponding measures of uncertainty. This means that the results of the model can only be indicative of the potential value for money of using enoxaparin thromboprophylaxis for an extended period following discharge for elective hip replacement.
The analysis did not include the impact of side effects, primarily bleeding episodes, which may be associated with the use of LMWH. There were no statistically significant differences in these rates of these adverse events in the clinical trials of extended prophylaxis with LMWH [18–21]. If there were clinically or economically important differences in these events, then the cost-effectiveness of extended enoxaparin may have been over-estimated. However, the base case analysis did include the costs of additional monitoring for hematological side-effects with extended enoxaparin.
Wherever possible, conservative assumptions were used, to ensure that any biases in the analysis worked against extended enoxaparin thromboprophylaxis. These were further tested in the sensitivity analysis. For those cases where extended enoxaparin was associated with a net expected cost, the cost per quality-adjusted life-year gained ranged between £144 and £27,000. These compare well with the cost/QALY for other health care interventions commonly used in the NHS (Table 5).
Table 5. Cost per quality-adjusted life year of alternative interventions
In addition, a number of factors may mean that enoxaparin is more cost-effective than the base case analysis would indicate. First, it is plausible that some patients on extended enoxaparin would require fewer visits than assumed. There is some evidence to suggest that patients or their carers can administer enoxaparin thromboprophylaxis without the assistance of a health care professional. The proportions of patients found eligible for home treatment ranges from 20 to 85%[16,34,35]. If 20% of patients could administer their own injections, the cost per quality-adjusted life-year gained was reduced to £4500. If 85% or more of patients or carers could administer enoxaparin, extending thromboprophylaxis to the postdischarge phase would result in an expected cost/QALY of less than £1000. Secondly, the rate of clinical diagnosis of DVT used for the analysis may be low, compared to current practice. If this were increased to 35% the analysis would show lower costs and higher survival in the extended enoxaparin group than the standard enoxaparin group.
Thirdly, the analysis only considered the short-term implications of DVT and PE. However, DVT can cause problems over a much longer period. Levin et al.  estimated the extra cost of a DVT over 15 years to be roughly £12,000 per person (converted to UK £'s and inflated to 1998 prices). Incorporating these additional costs into the model would give a net expected saving associated with extended enoxaparin of approximately £2300 per person.
Fourthly, the analysis weighted the life-years gained from the prevention of DVT, by using a utility measure to reflect the generally lower health-related quality of life of older people. However, the base case analysis assumed that people who had DVT and survived would have the same life expectancy and health-related quality of life as those who did not have DVT. This may under estimate the long-term impact of DVT on health status and morbidity. If extended enoxaparin thromboprophylaxis reduces the incidence of clinically important DVT, the QALY gains will also have been under-estimated.
Despite the limitations of analyses using modelling techniques, this conservative analysis indicates that extended thromboprophylaxis with enoxaparin may be cost-effective, compared to many other routine interventions. This will depend in part, on the monetary valuations health care policy-makers place upon gains in survival and quality of life.