|Home | About | Journals | Submit | Contact Us | Français|
Non-invasive stress testing might guide the use of aspirin and statins for primary prevention of coronary heart disease (CHD), but it is unclear if such a strategy would be cost-effective.
We compared the status quo, in which the current national use of aspirin and statins was simulated, with three other strategies; 1) full implementation of Adult Treatment Panel (ATP III) guidelines, 2) a “treat-all” strategy in which all intermediate-risk persons received statins (men and women) and aspirin (men only) and 3) a “test and treat strategy,” in which all persons with an intermediate-risk of CHD underwent stress testing and those with a positive test were treated with high-intensity statins (men and women) and aspirin (men only). Healthcare costs, CHD events, and quality-adjusted life years from 2011 to 2040 were projected. Under a variety of assumptions, the “treat-all” strategy was the most effective and least expensive strategy. Stress electrocardiography was more effective and less expensive than other “test-and-treat” strategies, but it was less expensive than “treat all” only if statin cost exceeded $3.16/pill or if testing increased adherence from below 22% to above 75%. However, stress electrocardiography could be cost-effective in persons initially non-adherent to the “treat all” strategy if it raised their adherence to 5% and cost-saving if it raised their adherence to 13%.
When generic high-potency statins are available, non-invasive cardiac stress testing to target preventive medications is not cost-effective unless it substantially improves adherence.
Although coronary heart disease (CHD) mortality has declined in the United States over recent decades, a substantial proportion of CHD remains attributed to hypercholesterolemia.1 National guidelines for lipid-lowering and aspirin prophylaxis for the primary prevention of CHD have proven effectiveness in high-risk patients, whereas costs and potential adverse side effects have tempered enthusiasm for treating low-risk individuals.2–5 Less clear is the approach for intermediate-risk patients (10–20% ten year risk of CHD), even though this group constitutes the largest population recommended for primary prevention with lipid-lowering agents and aspirin under current guidelines.3 Based on the declining cost of statins, pharmacologic lipid lowering may now be cost-effective for primary prevention for the large intermediate-risk population.6,7
Non-invasive cardiac stress testing can improve risk stratification of intermediate-risk patients.8,9,10 A positive test may identify intermediate-risk patients who would benefit from more intensive risk factor modification, whereas negative test results may identify persons appropriate for lower intensity interventions.11 Additionally, stress testing might improve medication adherence in persons with positive test results.12,13 We used the CHD Policy Model, a computer model of CHD in the United States (US) adult population, to test the hypothesis that further risk stratification of intermediate-risk persons with stress testing might be more effective and less costly than other primary prevention approaches in this group.
The study population initially consisted of U.S. men age 45 years in 2011 and women age 55 years in 2011 who had an intermediate-risk (10–20%) of developing a CHD event in the next 10 years based on the ATP III guideline point system.6 In each subsequent year, new intermediate-risk 45 year old men and 55 year old women were added to the simulation, and any persons who died or became high-risk before their next 10 year screening without first developing CHD left the simulation. Intermediate-risk persons who developed CHD were followed until their deaths or 2040, whichever came first.
In this population we compared 1) a “treat-all” strategy in which all persons were treated with high-intensity statins (men and women) and aspirin (men only) without undergoing stress testing, 2) a “test-and-treat” strategy, in which the initiation or intensification of statin therapy (men and women) and aspirin prophylaxis (men only) was guided by stress testing; and 3) an ATP III guideline-based approach in which all persons with an LDL of at least 130mg/dl (3.36 mmol) were placed on low-intensity statin therapy. In the “treat-all” and “test-and-treat” strategies, aspirin therapy was limited to men, as data suggest benefit for preventing CHD in men but not in women.13 Three “test-and-treat” strategies were analyzed: stress electrocardiography (ECG) alone, stress ECG combined with a nuclear perfusion scan, and stress echocardiography alone. We also projected results of a status quo strategy that simulated the current use of statins and aspirin prophylaxis in intermediate-risk adults in the U.S. (Table 1).
The analyses were performed using the CHD Policy Model, which is a computer-simulation, state-transition (Markov cohort) model of CHD incidence, prevalence, mortality, and costs among persons aged 35–84 years in the United States.14–16 The demographic-epidemiologic submodel predicts CHD incidence and non-CHD mortality among persons without CHD, stratified by age, sex, blood pressure, smoking status, high-density lipoprotein (HDL) cholesterol level, low density lipoprotein (LDL) cholesterol level, diabetes mellitus, and use of aspirin or statin therapy. Population-weighted joint distributions of risk factors and use of statins for lipid-lowering were estimated from the U.S. National Health and Nutrition Examination Surveys (NHANES) for the years 1999–2004.17 Baseline aspirin use was estimated from the Behavioral Risk Factor Surveillance Survey 2003 (BRFSS, personal communication, U.A. Ajani, U.S. Centers for Disease Control). After CHD develops, a bridge submodel characterizes the initial CHD event and its sequelae over 30 days. The disease history submodel then predicts subsequent CHD events, revascularization procedures, CHD mortality, and non-CHD mortality among patients with CHD, stratified by age, sex, and CHD event histories. Each health state and event has an annual cost and quality-of-life adjustment. All population distributions, risk factor levels, coefficients, event rates, case-fatality rates, costs, and quality-of-life adjustments can be modified to evaluate different scenarios.
CHD and non-CHD death rates were estimated from the United States Vital Statistics.18 Case-fatality, revascularization rates, and the number of people who had myocardial infarctions or out-of-hospital cardiac arrests and survived to hospital discharge were obtained from the National Hospital Discharge Survey (NHDS).19 Risk factor beta coefficients, conditioned on age and sex, were estimated by using examinations nine to 13 of the original Framingham cohort and examinations one to six of the Framingham offspring cohort.20 Total direct health care costs were estimated from the Centers for Medicare & Medicaid Services.21 Health-related quality-of life weights were based on observational data,22 and costs and quality-adjusted life-years (QALYs) were discounted at 3% per year. All costs for all simulations were inflated to year 2009 costs.
For our primary analysis, testing and treatment began at age 45 years for men and 55 years for women. Persons at intermediate-risk were tested every ten years through age 65. We also examined different age thresholds for initiation of therapy, including 45, 55, and 65 years, and also considered different thresholds based on sex. Simulations, which started in 2011 and ended in 2040, evaluated cumulative direct medical costs, QALYs, and CHD outcomes from 2011–2040. Incremental cost-effectiveness ratios (ICER) were calculated. An ICER of $50,000 to $75,000 per QALY gained was used as a cost-effectiveness reference point.16, 23
We derived estimates for the probability of a positive stress test, for each type of test stratified by age and gender, and the resulting change in risk of future CHD events from studies of non-invasive stress testing in asymptomatic intermediate-risk adults.8,9,11 Test costs were estimated from Medicare reimbursement rates.24 Nuclear stress tests were assigned an additional cost and QALY decrement due to the estimated increased lifetime cancer risk as a result of test associated radiation exposure.25–27 No QALY decrements were assumed for stress ECG or stress echocardiography. The base-case simulations utilized wholesale generic prices for statins and generic prices for aspirin. Costs for wholesale generic statins were taken from the lowest average wholesale price reported in the 2006 National Drug Data File for low-intensity and high-intensity statins as well as informal telephone and Web site surveys conducted in 2008.6 The costs for brand name statins and generic aspirin were estimated from the Redbook database.28 All simulations included a $95.50 cost for a yearly physician visit and lipid profile.29
Because reliable evidence for effective primary prevention of acute myocardial infarction with aspirin exists only for men (relative risk of 0.77),30 only eligible male patients were prescribed aspirin. The rates of adverse events from aspirin and statin therapy, as well as resultant costs and QALY decrements, were assumed from the literature.31–33 Side effects were assumed to occur twice as often for high-intensity statins.6,34 We assumed that 5.7% of men would be aspirin-allergic and unable to tolerate it.35 All intermediate-risk patients who were taking statins at baseline (“status quo”) were assumed to be taking low-dose statins (simvastatin 10 mg, or pravastatin 20 mg) daily. For the primary analysis, effectiveness and treatment adherence rates were assumed to be equivalent to those observed in clinical trials.36 Non-adherent patients were assumed to incur the costs of stress testing, in stress-test based simulations, and only one month of medical treatment.
We varied adherence rates to medications for each strategy from zero to 100% and compared outcomes at varying rates of adherence for each strategy. We based this variable adherence sensitivity analysis on studies that observed adherence rates to statin therapy that ranged from as low as 19% to 52% in persons not undergoing non-invasive cardiac imaging,13, 37,38 to as high as 90% in persons who had positive non-invasive cardiac imaging.12 Simulations were also performed in which a negative test result decreased adherence from the assumed baseline adherence rate of 40% to 21%, using the inverse of the likelihood ratio that a positive stress test raised adherence from 40% to 90%. An additional adherence-based sensitivity analysis assessed the value of a positive stress test to improve adherence in persons who were not adherent to an initial “treat-all” strategy.
Another sensitivity analysis modeled discontinuing current statin therapy in persons with a negative stress test and an LDL < 130 mg/dl (3.36 mmol/l). Other sensitivity analyses assessed the impact of assuming that aspirin therapy reduces the risk of CHD by 27% in women just as in men, assuming that aspirin in men reduces the risk of ischemic stroke by 23% and increases the risk of hemorrhagic stroke by 32%, and assuming that statins reduce the risk of stroke in men and women by 17%.14, 39,40 Sensitivity analyses assigned statins a non-wholesale generic Redbook annual price of $367 and a brand name price of $1,116, or $3.06 a pill,28 and evaluated the costs of twice-yearly lipid and liver function laboratory tests for persons on high-dose statins. Additional analyses evaluated assigning nuclear stress testing a lower dose of radiation, performing interventions for only men or only women, performing only one-time testing when a person reached each age threshold without subsequent re-testing, prescribing only statins but not aspirin for all interventions, and varying the age thresholds of intervention from a low of 45 years to a high of 65 years (Supplemental Table 6).
We also estimated possible downstream effects in which persons with a positive stress test would then have either a stress echocardiogram or stress nuclear test.11,41 If this second test were also positive, the patient would then undergo diagnostic coronary angiography.42 Asymptomatic persons with positive angiography would then be prescribed atenolol (100 mg daily) as well as a statin (men and women) and aspirin (men only).43–44 This analysis was labeled “test-angio-treat-atenolol”.
About 22.5 million individual, asymptomatic, intermediate-risk adults met the pre-specified inclusion criteria during the 30 year simulation period and were included in the analysis. Over the 30-year period evaluated, the “treat-all” simulation would lead to approximately 368 million person-years of high-intensity statin therapy and 164 million person-years of aspirin therapy, whereas the “test-and-treat” strategy would lead to only 62 million person-years of high-intensity statin therapy and 32 million person-years of aspirin therapy. With the “test-and-treat” strategy, 40.4 million stress tests would be performed by the year 2040, with an estimated 21% being positive and leading to high-intensity statin (and in men, aspirin) therapy.
The “treat-all” intervention was less expensive and more effective than either full ATP III implementation or any of the “test-and-treat” simulations (Table 3). All “test-and-treat” simulations led to relatively similar results, and stress ECG alone, which was the least expensive and would be easiest to implement (Supplemental Table 7), was the “test-and-treat” strategy of choice for subsequent simulations.
Even when assuming perfect medication adherence for the stress ECG strategy, “treat all” remained more cost-effective until the adherence rate decreased to below 30%. If, however, a positive stress ECG increased medication adherence to above 75%, “test and treat” would be the preferred strategy once the adherence rate for “treat all” fell below 22% (Table 4 and Figure 1).
If stress ECG testing were performed only in persons who were non-adherent to aspirin and statin therapy, such targeted testing would be cost-savings compared with “treat all” if 13% or more of patients with a positive stress ECG then became adherent and received the benefits of therapy and would be cost-effective at $46,000 per QALY if adherence rose to 5% in those with a positive stress ECG.
At an ICER of $50,000/QALY, “treat all” remained more cost-effective than the stress ECG “test-and-treat” strategy until annual statin therapy costs exceeded $1,152 (Table 4).
“Test-and-treat” strategies that discontinued statins in persons with a negative stress test and an LDL <130mg/dl (3.36mmol/dl) were always more expensive and less effective than maintaining the prior statin dose (Table 4). Although “treat all” was cost-saving compared with “test and treat” for all men-only simulations, it was not cost-saving but remained cost-effective for women-only simulations (Table 4). For “test-and-treat” simulations, one-time testing was not more cost-effective compared with repeat testing every 10 years (Table 4).
The “test-angio-treat-atenolol” strategy had an ICER of approximately $200,000/QALY gained as compared with the initial “test-and-treat” strategy using stress ECG alone. In addition, this “test-angio-treat-atenolol” strategy was less effective and more costly than “treat all” (Table 4).
Even when prescribing only statins without aspirin, “treat all” was both more effective and less expensive than “test and treat”, and “treat all” continued to be the strategy of choice when assigning the same CHD reduction of 27% to women taking aspirin as assigned to men in the primary analysis (Table 4). The “treat-all” strategy remained more effective than “test and treat” even assuming twice yearly lipid and liver function testing. Assuming that aspirin and statins reduce the risk of ischemic stroke and that aspirin increases the risk of hemorrhagic stroke, the “treat-all” strategy would be even more attractive compared with all other interventions (Supplemental Table 8). Lastly, the ideal age threshold to initiate each intervention was determined to be age 45 for men and 55 for women, as in each simulation that age threshold resulted in more added QALYs and lower costs as compared with initiating interventions at older age thresholds (Supplemental Table 6).
At the current low price for generic statins and aspirin, extending treatment to all intermediate-risk men and women is more cost-effective than targeting a selected subgroup of intermediate-risk individuals either with positive non-invasive stress tests or higher LDL levels. This finding is consistent with multiple studies showing a limited improvement in CHD risk stratification based on non-invasive stress testing, and analyses that demonstrated the cost-effectiveness of routine treatment of moderate risk persons with low-cost statins.7,45–46 However, since current adherence rates to primary prevention pharmacotherapy strategies are low, our simulations indicate that a “test-and-treat” based strategy would be cost-effective only if it markedly improves adherence to about 75% in all intermediate-risk persons. However, testing persons who are initially non-adherent to a “treat all” strategy would be cost-effective if it improved adherence to aspirin and statin therapy to greater than 5% and would be cost-saving if it raised adherence to more than 13% in initially non-adherent patients after a subsequent positive stress ECG.
Studies have shown that multiple factors, including one’s perception of future risk of disease and knowledge of carrying a diagnosis of CHD, are important determinants of one’s likelihood to adhere to recommended statin therapy.47,48 Most importantly, in a study of asymptomatic persons undergoing coronary artery calcium testing to asses for CHD risk, those who had a calcium score of zero only exhibited a 44% one year adherence rate to recommended statins after testing, whereas those with a calcium score of 100–399 exhibited a 75% adherence rate, and those with a calcium score of over 400 exhibited an adherence rate of 90% to recommended statin therapy.12 Therefore our sensitivity analysis, which assumed an increased adherence rate to statin therapy for patients with a positive non-invasive stress test as compared with those with a negative stress test, is reasonable based on previously published studies.
Primary prevention strategies that tailor statin therapy based on a person’s overall risk of CHD may be cost effective compared with strategies that base the decision on a particular LDL level, but a more cost-effective approach often may include both a target LDL level in higher risk persons and a risk-based treatment approach in intermediate-risk persons.6 Others have debated the relative merits of basing statin treatment guidelines on 10-year CHD risk alone or basing treatment on LDL level. Our analysis suggests that the advent of low-cost statins will make treating all intermediate-risk persons, regardless of LDL level, a cost-effective primary CHD prevention strategy.
Our simulations demonstrated that even at near 100% medication adherence, neither a stress test-based approach nor more complete implementation of ATP III risk factor modification guidelines would optimally prevent CHD or reduce overall health care costs in intermediate-risk persons compared with a “treat-all” strategy. Although “treat all” places more people on statin and aspirin therapy, it is less expensive than any other strategy because it reduces incident CHD by over 1 million cases over the 30-year period evaluated compared with ATP III, which was the next most effective strategy.
Our study has several limitations. The sensitivities and specificities of non-invasive stress testing as well as the post-test prognosis given a positive or negative test were based on published studies and may not extend to all testing centers or to testing in all asymptomatic intermediate-risk persons. Our simulations also assumed that all intermediate-risk persons from age 45 to 65 years would be physically capable of performing an exercise stress test and that all tests could be classifiable as ‘normal’ or ‘abnormal’. Since the American College of Cardiology guidelines recommend exercise stress testing over pharmacologic testing until the age of 65,49 the broad implementation of stress ECG to improve CHD risk stratification for patients age 45 to 65 years might be feasible. Other than the increased risk of cancer from nuclear stress testing radiation exposure, we assumed no other disutility from stress testing. The sensitivity analyses that assumed a substantially higher adherence rate in persons with a positive stress test were based on limited adherence data. Even if a positive test result improved medication adherence, little is known about how long this effect is sustained, and there may be less costly ways to increase adherence to aspirin and statins. Additionally, our primary simulations may underestimate the health benefits and costs savings associated with statins and aspirin, because the main benefits did not include any impact of aspirin and statins on the reduction of cerebrovascular disease in the intermediate-risk population. Coronary artery calcium screening can improve risk stratification of intermediate-risk persons. When compared with routine statins, it has been reported to be cost-effective for men if a positive test leads to universal treatment with statins, aspirin, and antihypertensive medications, regardless of the person’s blood pressure level.10 Our analysis differs in assuming that men in the “treat all” strategy would already be prescribed statins and aspirin and in not assuming that antihypertensive medications would be used in normotensive persons. Biomarkers such as high sensitivity C-reactive protein may also help to stratify intermediate-risk patients50 but were not examined in our study.
In conclusion, using a national-scale computer simulation model of CHD in U.S. adults, we project that universal treatment of intermediate-risk women with statins and intermediate-risk men with statins plus aspirin, regardless of their LDL levels, would be a cost-effective CHD primary prevention policy. Universal non-invasive cardiac stress testing to guide the use of statins or aspirin is not cost-effective, unless testing markedly increases medication adherence to about 75%. By comparison, testing of non-adherent patients is potentially attractive since only a modest adherence (about 13%) among persons with a positive stress test would be cost-saving and even a very low adherence (5%) among persons with a positive stress test would be cost-effective in this previously non-adherent group. However, the worthiness of such an approach would have to be compared against other alternative and possibly more cost-effective ways, such as frequent reminders, to increase adherence.48
This manuscript was prepared using a limited access dataset obtained by the National Heart Lung and Blood Institute (NHLBI) and does not necessarily reflect the views or opinions of the Framingham Heart Study, Framingham Offspring Study, or the NHLBI. The authors thank Uman A. Ajani for providing the rate of aspirin use from the BRFSS.
Funding Sources: Dr. Goldman receives funding from the American Heart Association (10CRP4140089) to study the analysis of US blood pressure control guidelines. Drs. Goldman and Moran receive funding from the (NHLBI) (1K08HL089675-01A1) to support studies utilizing computer modeling of cardiovascular disease.
Conflict of Interest Disclosures: None