- Tom Marshall, clinical lecturer ()1
- Correspondence to: T Marshall, c/o Partners Healthcare, 1620 Tremont Street, Boston, MA 02120, USA
- Accepted 10 October 2003
Objective To determine which treatments for preventing coronary heart disease should be offered to which patients by assessing their incremental cost effectiveness.
Design Modelling study
Data sources Cost estimates (for NHS) and estimates of effectiveness obtained for aspirin, antihypertensive drugs, statins and clopidogrel.
Data synthesis Treatment effects were assumed to be independent, and cost per coronary event prevented was calculated for treatments individually and in combination across patients at a range of coronary risks.
Results The most cost effective preventive treatments are aspirin, initial antihypertensive treatment (bendrofluazide and atenolol), and intensive antihypertensive treatment (bendrofluazide, atenolol and enalapril), whereas simvastatin and clopidogrel are the least cost effective (cost per coronary event prevented in a patient at 10% coronary risk over five years is £3500 for aspirin, £12 500 for initial antihypertensives, £18 300 for intensive antihypertensives, £60 000 for clopidogrel, and £61 400 for simvastatin). Aspirin in a patient at 5% five year coronary risk costs less than a fifth as much per event prevented (£7900) as simvastatin in a patient at 30% five year risk (£40 800).
Discussion A cost effective prevention strategy would offer aspirin and initial antihypertensive treatment to all patients at greater than 7.5% five year coronary risk before offering statins or clopidogrel to patients at greater than 15% five year coronary risk. Incremental cost effectiveness analysis of treatments produces robust, practical cost effectiveness rankings that can be used to inform treatment guidelines.
Coronary heart disease is a major cause of morbidity and mortality, and its prevention has assumed increasing importance in UK health policy.1 Several treatments reduce risk of coronary disease, the absolute benefits of treatment are proportional to pretreatment risk, and individual patients may be eligible for more than one treatment. Moreover, it is argued that treatments for lowering blood pressure and cholesterol are equally effective whether or not blood pressure or cholesterol levels exceed arbitrary thresholds.2 This means that virtually all patients might benefit from risk lowering treatments.
A previous analysis explored rational identification strategies for coronary heart disease prevention in primary care, ranking patients by their likelihood of benefiting from treatment.3 Given that health service resources are finite, a rational approach to treatment would offer patients treatments in order of their expected cost effectiveness. This requires knowledge of the incremental benefits of risk lowering treatments in relation to their incremental costs. Incremental cost effectiveness analysis provides a means of ranking treatments by calculating the incremental changes in both costs and benefits. Although widely advocated, it has seldom been used outside the evaluation of screening programmes. This paper presents an incremental cost effectiveness analysis of risk lowering treatments in patients at varying levels of risk. The treatments analysed are aspirin, initial antihypertensive treatment, intensive antihypertensive treatment, a statin, and clopidogrel.
Costs are considered from the perspective of the health service and are discounted at 6% per year.4 There are two main components to the costs of long term treatment–follow up costs and prescribing costs.
Follow up costs are based on two clinic appointments (of 15 minutes each) a year with a practice nurse. The total health service cost of a practice nurse clinic is £31 an hour.5 Patients taking thiazide diuretics require annual measurement of serum electrolytes and uric acid. Patients taking statins require annual measurement of serum lipid concentrations and liver function tests. I derived costs of blood tests from local standard costs of pathology services compiled by the London School of Hygiene and Tropical Medicine in 1996 (personal communication, Rheinold Gruen) and adjusted for annual price inflation of 6%.
Prescribing costs include drug and dispensing costs. I obtained drug costs from the British National Formulary6 and calculated dispensing costs at 87.4 pence per prescribed item on the assumption that four prescriptions are issued a year.7 Initial antihypertensive treatment is with bendrofluazide (bendroflumethiazide) 2.5 mg and atenolol 50 mg; further antihypertensive treatment adds enalapril 20 mg to these treatments. Cholesterol lowering is with simvastatin 40 mg. Clopidogrel is given at a dose of 75 mg daily.
I calculated benefits of treatment as major coronary events (myocardial infarctions, new cases of angina, and cardiac deaths) prevented over five years, with the benefits discounted at 1.5% per annum in keeping with current guidelines from the National Institute for Clinical Excellence.4
In the base case analysis, I calculated cost effectiveness for a patient whose pretreatment five year coronary risk is 10%.8 This is the coronary risk of a non-diabetic, non-smoking man aged 62 with blood pressure of 160/98 mm Hg, total serum cholesterol concentration of 6.5 mmol/l, and high density lipoprotein cholesterol concentration of 1.3 mmol/l. Under current guidelines he is eligible for antihypertensive treatment, a statin, and aspirin.9
For each intervention, I derived relative risk of coronary heart disease from a recent meta-analysis. The relative risk of coronary events for patients taking aspirin is 0.72 (95% confidence interval 0.60 to 0.87).10 Aspirin also increases in incidence of major bleeding by 0.3% (0.2% to 0.4%) over five years of treatment.10 To take account of this, I offset the absolute reduction in coronary risk over five years by 0.3%, thus giving major bleeding events equal weight to coronary events. Compared with aspirin, the relative risk of a coronary event while taking clopidogrel is 0.88 (0.76 to 1.01).11 Empirical studies suggest that an indirect estimate of the effects of clopidogrel compared with placebo should be accurate provided that the population groups in studies are similar.12 Compared with placebo, the relative risk of a coronary event with clopidogrel is therefore 0.63 (0.45 to 0.82) (0.63 = 0.72x0.88).
Compared with placebo, the relative risk of a coronary event with antihypertensive treatment is 0.83 (0.72 to 0.91).13 Compared with initial antihypertensive treatment, the relative risk of a coronary event with intensive antihypertensive treatment is 0.81 (0.67 to 0.98).14 The relative risk of intensive antihypertensive treatment in comparison with placebo is therefore 0.67 (estimated 95% confidence interval 0.49 to 0.85). The relative risk of a coronary event with a statin is 0.69 (0.64 to 0.74).15
In clinical trials the relative risk of coronary events with preventive treatments is similar in patients taking additional treatments and in those who are not.16 This supports the view that treatment effects are independent.17 This view is also biologically plausible, since treatments act through different mechanisms. If treatment effects are independent the relative risk with two or more treatments is the product of the relative risk on each treatment. For example, the relative risk of a coronary event with aspirin is 0.72, with a statin is 0.69, and with aspirin and a statin is 0.50 (0.50 = 0.72x0.69).
Average cost effectiveness
I calculated the cost of each intervention over a five year time horizon, and calculated the reduction in absolute coronary risk by subtracting post-treatment risk from pretreatment risk. Post-treatment risk is the product of pretreatment risk and the relative risk with treatment. In the case of aspirin, 0.3% is subtracted from the reduction in absolute coronary risk to take account of major adverse effects. The cost effectiveness ratio (cost per event prevented) is the total cost divided by the reduction in absolute coronary risk. I estimated the average cost per coronary event prevented for each treatment used alone. In a sensitivity analysis I calculated maximum and minimum costs per event prevented for each of the interventions using the upper and lower 95% confidence limits for effectiveness. The average cost effectiveness rankings inform the order in which treatments would be offered in an incremental cost effectiveness analysis. To test the robustness of cost effectiveness rankings, I explored the effects of changes in the costs of interventions and the frequency of adverse effects alongside changed assumptions about effectiveness.
Incremental cost effectiveness
An efficient prevention strategy would offer the most cost effective treatment first, then the next most cost effective treatment, and so on. This enables the largest possible proportion of the benefits of treatment to be achieved at the lowest possible cost. The incremental cost effectiveness ratio is the additional cost associated with adding each treatment divided by the additional benefit of the treatment.
Incremental cost per event prevented is calculated in much the same way as the average cost per event prevented. The incremental cost of treatment includes only additional costs of treatment. The incremental reduction in absolute coronary risk is calculated by subtracting post-treatment risk from pretreatment risk. Post-treatment risk is the product of pretreatment risk and the relative risk on the additional treatment. However, pretreatment risk is the post-treatment risk after any previous treatments. In the case of aspirin, 0.3% is subtracted from the reduction in absolute coronary risk to take account of adverse effects. The cost effectiveness ratio (cost per event prevented) is the total cost divided by the reduction in absolute coronary risk.
A sensitivity analysis tested the robustness of cost effectiveness ratios by changing assumptions about effectiveness and identifying the threshold costs at which cost effectiveness rankings would change. Since the cost per coronary event prevented decreases as patients' coronary risk increases, I also investigated the cost effectiveness of coronary disease prevention in patients at a range of five year coronary risks.
Average cost effectiveness
In a patient at 10% coronary risk over five years, aspirin is the most cost effective risk lowering treatment, at £3500 per coronary event prevented. Initial antihypertensive treatment costs £12 500, intensive antihypertensive treatment costs £18 300, clopidogrel costs £60 000, and simvastatin costs £61 400 per coronary event prevented (table 1).
In a sensitivity analysis I calculated the cost per event prevented for each intervention if its effectiveness was given by the upper and lower 95% confidence limits of the estimates (see figure). The cost effectiveness of aspirin, initial antihypertensive treatment, and intensive antihypertensive treatment are sensitive to changes in assumptions about effectiveness. However, for simvastatin to be of similar cost effectiveness to intensive antihypertensive treatment, the relative risk with treatment must be at the lower 95% confidence interval and the cost of the drug 65% lower. This is unlikely, as drug prices typically fall by less than 50% when they come off patent. There is a wide degree of uncertainty about the cost effectiveness of clopidogrel, reflecting uncertainty about the relative risk on treatment.Image/caption references in XML:
Varying the discount rates for either costs or benefits from 0% to 10% has no effect on rankings. Using a general practitioner for follow up has no effect on cost effectiveness rankings. Cost effectiveness of initial and intensive antihypertensive treatment is sensitive to increases in the price of drugs. If sufficiently high cost drugs are used (such as for brand name calcium channel blockers) the cost per event prevented with initial antihypertensive drugs is as high as with a statin.
Incremental cost effectiveness of additional treatments
Costs of follow up clinic visits do not increase with extra treatments. The incremental costs of additional treatments therefore include only additional drug costs and additional laboratory investigations. The incremental effectiveness of additional drugs is also smaller than their effectiveness as initial treatments because incremental effects act on progressively smaller pretreatment risks.
If a patient at 10% five year coronary risk is given combination treatments in order of their cost effectiveness, the incremental cost per event prevented rises with each additional treatment. Compared with placebo, clopidogrel is more cost effective than simvastatin. However, clopidogrel as a replacement for aspirin provides little additional benefit at substantial extra cost. It is therefore the least cost effective in an incremental analysis. Incremental costs per event prevented are £3500 for aspirin, £12 000 for initial antihypertensive treatment, £33 900 for enalapril, £122 400 for simvastatin, and £527 200 for clopidogrel (table 2).
I carried out a sensitivity analysis to investigate the effects of varying the costs and effectiveness of treatments. The most favourable assumption for simvastatin is that relative risks for all other treatments are at the upper 95% confidence limit and for simvastatin is at the lower 95% confidence limit. If this is the case, the incremental costs per event prevented are £8700 for aspirin, £18 800 for initial antihypertensive treatment, £243 000 for intensive antihypertensive treatment, £65 800 for simvastatin, and £177 300 for clopidogrel. Even under these assumptions, the price of simvastatin would have to fall by 70%, and the price of clopidogrel by more than 90%, to be of similar cost effectiveness to initial antihypertensive treatment.
Under the base case analysis, the cost effectiveness rankings of all five treatments are the same for any patient with a five year coronary risk greater than 1.5%. The incremental cost per event prevented in a patient at 5% five year coronary risk is £7900 with aspirin and £24 000 with initial antihypertensive treatment. This is less than the incremental cost per event prevented with simvastatin (£40 800) in a patient at 30% five year coronary risk (see table 3)
The most extreme assumptions we can make are to assume that relative risk on all treatments is at the upper 95% confidence limit (least effective) and assume that the relative risk with simvastatin is at the lower 95% confidence limit (most effective). Under these assumptions, the cost per event prevented with aspirin in a patient at 7.5% five year risk would be £12 900 and the cost per event prevented with simvastatin in a patient at 15% five year risk would be £13 200.
This analysis confirms the poor cost effectiveness of statins and clopidogrel compared with aspirin and antihypertensive treatment.18 19 By quantifying the treatments' cost effectiveness, the analysis suggests it is likely to be more cost effective to treat patients at 5% five year coronary risk with aspirin than to prescribe further antihypertensive treatment or statins to patients at 30% five year risk.
Limitations of study
A weakness of this analysis is that some of the findings are sensitive to the choice of drug. This is particularly true of the cost effectiveness of initial antihypertensive treatment, where drug prices range from £10 to £290 a year. However, consideration of every possible antihypertensive regimen is beyond the scope of this paper.
The analysis may overestimate the benefits of some interventions. Estimates of the effects of intensive antihypertensive treatment are derived from studies comparing less intensive with more intensive treatment.14 But the estimate of the benefits of initial antihypertensive treatment include all blood pressure lowering, not just less intensive treatment.13 The additional benefits of intensive antihypertensive treatment may therefore be exaggerated.
Apart from major bleeding due to aspirin, the analysis takes no account of adverse effects of treatment. Adverse effects may be considered as minor and reversible on stopping treatment or as major and irreversible. However, major adverse event rates would have to be 0% with a statin and to exceed 2.5% per five years with aspirin and 1.3% with initial antihypertensive treatment for the statin to be more cost effective than the latter two treatments. Minor, reversible adverse effects have little impact on the analysis because patients who stop treatment incur neither further costs nor further benefits. Even if all patients who experience adverse effects discontinue treatment but continue to be followed up, these differences make no differences to the cost effectiveness rankings. Evidence suggests that patients taking a statin report fewer adverse effects than those taking a placebo.20 Among patients taking two low dose antihypertensive drugs, 7.5% reported adverse effects, but few (< 1%) were sufficiently severe to stop treatment and almost all were reversible.21 With aspirin, 3.9% of patients report adverse effects,22 but excess risk of major bleeding is under 0.5%.10
Incremental cost effectiveness analysis of treatments produces robust, practical cost effectiveness rankings. Authors of guidelines should take account of this when making treatment recommendations. If the aim of treatment is to maximise prevention of coronary disease, these results have clear implications for current treatment recommendations. They cast doubt on the wisdom of present policy, which emphasises achievement of target blood pressures and the use of statins for people at 15% five year risk of a coronary event.23 A more efficient prevention strategy would be to offer aspirin and initial antihypertensive treatment to all people at over 7.5% five year coronary risk before offering statins to patients at 30% five year risk. According to national survey data, 87% of men and 56% of women aged over 65 are at over 7.5% five year risk.24
What is already known on this topic
Aspirin, antihypertensive treatment, statins, and clopidogrel are all effective in preventing coronary heart disease
These drugs vary in their cost effectiveness, and a rational prevention strategy would offer the most cost effective treatments first
What this study adds
Most of the benefits of prevention can be achieved with aspirin and antihypertensive treatment at a fraction of the cost of simvastatin or clopidogrel
Treating a patient with a five year coronary risk of 7.5% with aspirin and low cost antihypertensives is more cost effective than treating a patient with 30% coronary risk with a statin
Clinical guidelines should be informed by analysis of the incremental costs and incremental benefits resulting from each additional treatment
Editorial by Warburton
Competing interests None declared.
Ethical approval Not applicable.