Heart failure (HF) is a major health problem in the U.S., with 4,790,000 patients affected and 550,000 new cases each year (1). The incidence of HF rises sharply with age, implying a large increase in the burden of disease over the next 20 years as the U.S. population ages. Survival is dismal following the development of HF, with only 20% of men age 65 years and over surviving to 5 years following development of symptoms (1). Not surprisingly, there is strong interest in preventing HF by identifying patients at risk and treating them before symptoms develop (2).

Screening for depressed left ventricular (LV) systolic function already meets many of the criteria for a successful screening program. The disease (LV dysfunction) is common and associated with significant morbidity or mortality. There is a recognizable latent phase: Stage B HF as defined by the American College of Cardiology and the American Heart Association (3). The disease is treatable during the latent phase; angiotensin-converting enzyme (ACE) inhibitors improved outcome in the Studies Of Left Ventricular Dysfunction (SOLVD) prevention trial (4). However, an important criterion that has yet to be established before widespread application can be recommended is that screening be cost-effective.

There are several candidate screening tests to detect patients with depressed LV function. The gold standard is direct imaging of the LV using contrast left ventriculography, nuclear angiography, or echocardiography. A less costly but less accurate test is the recently developed blood test for B-type natriuretic peptide (BNP). The BNP is released primarily by the cardiac ventricles in response to abnormal loading conditions (5). The test (currently priced near $30) is available as a rapid bedside assay and has been shown to accurately determine if HF is the cause of dyspnea in patients presenting to the emergency room (6). The BNP has also been found to be elevated in asymptomatic patients with LV dysfunction (7). The purpose of this study was to determine if population screening for LV systolic dysfunction would be cost-effective.

We examined four screening strategies. The first strategy was BNP testing and, if abnormal, echocardiography. Patients with an ejection fraction (EF) <40% are treated (ACE inhibitors) to prevent the development of HF. The second strategy was BNP only, with treatment based on the results. The third strategy was echocardiography for all patients (treatment based on the results). The fourth strategy was not to screen for depressed LV function. Each test leads to one of four results (true positive, false positive, true negative, false negative). Patients with true and false positive test results are treated. Patients with false negative results are treated only when HF develops. Patients with true negative results have a normal age-specific life expectancy. False-positive patients receive a decrement in quality-adjusted survival to account for side effects of treatment.

We developed a decision model using DATA software (Version 4.0, TreeAge Software, Boston, Massachusetts). We determined the lifetime health and economic outcomes for hypothetical cohorts of 60-year-old men and women with: 1) depressed EF (40% or less) but no history of HF and treated with ACE inhibitors, 2) depressed EF but no history of HF and no treatment until HF developed, and 3) without depressed EF.

Each month, patients with a low EF and without a history of HF can remain asymptomatic, develop HF, or die (Figure 1). Of those patients developing HF, we assumed 33% would be hospitalized during their initial episode of HF (4). Once patients develop HF they can remain stable, be hospitalized, or die during each time period. The model follows patients until all have died (age 120). Patients without depressed EF are assumed to have an average age-specific mortality based on U.S. life table data (8). Development of HF for patients without depressed EF is accounted for using age-specific medical costs and survival for the general population.

The gender-specific sensitivity and specificity of BNP testing for detecting depressed LV EF (based on echocardiography) were obtained from a recent report from the Framingham Heart Study (9) (Table le1). This study excluded patients with HF and used the Shionogi BNP assay with a threshold of 24 mg/dl for men and 34 mg/dl for women. The gender-specific prevalence of disease was taken as the average of estimates from the Framingham Heart Study and a population study from Rochester, Minnesota (10). Prevalence was defined as moderate-severe dysfunction (EF ≤40%). In sensitivity analysis we examined Biosite BNP test-characteristics from a population survey in Rochester, Minnesota (threshold at the 95th percentile for gender and age: 52 mg/dl for men age 55 to 64 years and 93 mg/dl for women age 55 to 64 years) (11). In the base case we assumed nuclear angiography was the gold standard for measurement of EF because it was the predominant measure used to determine EF in the SOLVD prevention study (4) (from which we based benefit of ACE inhibitors). We assumed that echocardiography would be slightly less accurate (sensitivity of 92% and a specificity of 96%) when compared to this standard (12), but examined a scenario in sensitivity analysis where echocardiography had 100% accuracy.

We based rates for the development of HF and death for asymptomatic patients with and without ACE inhibitor treatment using published data from the SOLVD prevention trial (4). We used actual event rates during the four years of reported follow-up. To model outcome after four years, we used an average of the yearly event rates weighted by the number of subjects still enrolled at each year of follow-up. Using this method we estimated that the yearly rate of progression to symptomatic HF would be 6.5% for patients treated with ACE inhibitors and 9.8% for those not treated. We used a similar method to determine the yearly relative risk of death (compared to the general population) for patients with asymptomatic LV dysfunction who are treated (2.9) and not treated (3.3) with ACE inhibitors. We assumed that 32% of patients would not be adherent to therapy on the basis of past trials of ACE inhibitors for patients without symptomatic HF (13).

We used SOLVD treatment trial data to estimate hospitalization and death rates for patients with HF treated with ACE inhibitors (14). We used actual event rates during the four years of reported follow-up for the SOLVD treatment trial. To model outcome following four years of living with HF, we used an average of the yearly event rates weighted by the number of subjects participating during each year of the trial. We assumed that survival with HF is further improved (relative risk 0.65) by the use of beta-blockers (15). This method estimated that the yearly relative risk of death (compared to the general population) for patients with symptomatic LV dysfunction was 4.9 when treated with ACE inhibitors and beta-blockers.

To determine quality-adjusted survival we assigned a utility value of 0.71 to each year of life for patients living with HF on the basis of prior studies using the time-tradeoff utility of patient preferences in HF (16). Asymptomatic patients were assumed to have a utility value of 0.87, also based on the time-tradeoff utility (16). We also examined a scenario where the utility for being asymptomatic was 1 (perfect health).

As recommended by the Panel on Cost-Effectiveness in Health and Medicine, we analyzed a reference case assuming a societal perspective by including all costs of medical care (Table le1), including medical costs incurred due to increased survival (17). Because HF survivors will incur additional costs for non-HF treatments, we assigned all patients a yearly age-specific (decile) cost of medical care based on medical expenditures for residents of the U.S. (18). To this baseline cost we added the costs of hospitalization of HF, ACE inhibitor treatment, and other outpatient HF care. We adjusted all costs to 2001 dollars using the medical component of the Consumer Price Index (Bureau of Labor Statistics).

We determined costs for hospitalization by applying the average U.S. cost (from cost-to-charge ratios for Medicare admissions) for diagnosis-related group 127, and added to this the Medicare- allowed charge for physicians visits of intermediate intensity (mean length of stay 5 days in 2000), including one consultation. For costs of ACE inhibitor treatment (lisinopril 20 mg QD) we used an average price from three national U.S. pharmacies. We assumed carvedilol would be the beta-blocker used for HF treatment, given the results of the Carvedilol or Metoprolol European Trial in Patients With Chronic Heart Failure (COMET) indicating a survival benefit of carvedilol over metoprolol tartrate (19). We determined the cost of outpatient HF care by using published estimates of resource use (20- 21) with adjustments for treatment with additional medications (such as spironolactone). Costs and benefits were discounted at 3% per year (17).

The costs of two-dimensional echocardiography including technical and professional components were obtained from Medicare-allowed charges for 2001. In sensitivity analysis we assumed that the Doppler and color Doppler components of the study would not be performed as part of a screening echocardiogram. The cost of BNP ($32) was based on the average cost of testing for the Biosite and Bayer BNP test. We estimated the costs of BNP testing assuming 5,000 tests were done per year per laboratory: phlebotomy and technician time ($9); reagent, calibration, controls (if applicable), and maintenance ($22); and equipment rental ($1). Lower volume laboratories will have a higher cost per test; we therefore examined costs up to $100 per test in sensitivity analysis.

A diagnosis of depressed EF may lead to additional tests (stress testing, angiography). Therefore, we added $2,200 in test costs for each patient with a positive test result assuming 100% would undergo stress echocardiography and 50% would receive coronary angiography. Because the benefit of this additional testing is difficult to quantify, we made the conservative assumption (biased against screening) that additional testing provided no additional benefit.

For each analysis we first rank the strategies by increasing cost. We then compare the cost-effectiveness of the first strategy with the strategy that has the next highest cost. Strategies that provide less effectiveness at a higher cost are eliminated (dominated).

We varied all parameters through the ranges listed in (Table le1). Although there is no universally accepted threshold for cost-effectiveness (17), $50,000 per QALY gained is commonly used (22). We also varied multiple parameters simultaneously using a Monte Carlo analysis with 5,000 simulations. Each parameter was assigned a distribution (beta-distribution for probabilities, log-normal for other parameters including costs). Because little data are available for assigning a variation to each distribution, we chose a standard deviation equal to the mean for cost variables. For parameters that are probabilities, we chose a standard deviation that approximated the range given in (Table le1) (2.5% lower bound to 97.5% upper bound).

The Agency for Health Care Research and Quality provided funds but had no role in model construction, data analysis, or report preparation.

For a population of men age 60 years with no history of HF (prevalence of low EF 3.5%) we found that a strategy of initial screening with BNP followed by echocardiography improved outcome at a cost of $22,300 per QALY gained compared to no screening (Table le2). If quality of life is ignored, BNP screening still costs only $23,500 per life-year gained compared to no screening. The number of men needed to screen with BNP was 44 to identify one with depressed EF, 133 to gain one year of life, and 127 to gain one QALY.

Screening with echocardiography was less attractive, with an incremental cost-effectiveness compared to initial BNP screening of more than $100,000 per QALY gained. Screening with BNP alone was more expensive and led to worse outcome (dominated) compared to BNP testing followed by echocardiography.

For a population of women age 60 years with no history of HF (prevalence of low EF 0.45%), initial screening with BNP followed by echocardiography improved outcome at a cost of $77,700 per QALY ($91,800 per life-year) gained compared to no screening (Table le3). The number of women needed to screen was 278 to identify one with depressed EF, 909 to gain one year of life, and 769 to gain one QALY. Screening women with echocardiography or BNP alone was dominated by BNP screening followed by echocardiography. The BNP-alone strategy is not discussed further.

We tested the robustness of our base-case findings by varying each of the assumptions in (Table le1) over the ranges listed. The decision to screen was sensitive to the prevalence of low EF and the accuracy of the screening tests. The model was insensitive to the benefit of ACE inhibitor treatment, costs of care, and cost of screening including echocardiography and BNP testing. We performed a probabilistic sensitivity analysis where the parameters were replaced with distributions and random sampling of each distribution was performed for 5,000 simulations. Screening men with BNP followed by echocardiography compared to no screening cost <$50,000 per QALY gained in 88% and <$100,000 per QALY gained in 98% of simulations. For women, screening with BNP followed by echocardiography compared to no screening cost <$50,000 per QALY gained in 27% and <$100,000 per QALY gained in 72% of simulations. Details of selected sensitivity analyses are described in the following text.

For the base-case analysis we assumed an asymptomatic population of men and women (mean age 60 years) would be screened. If the prevalence of low EF is at least 1% then the incremental cost-effectiveness of BNP screening is $50,000 or less per QALY gained for both men and women (Figure 2). For the cost-effectiveness ratio with BNP screening to be <$20,000 per QALY gained, the prevalence must be >3% for women and 4% for men. Screening all patients with echocardiography costs <$50,000 per QALY gained if the prevalence of depressed EF is at least 9% for men and 14% for women.

For our base case we chose the BNP threshold from the Framingham Heart Study that maximized the sums of sensitivity and specificity (21 ng/dl for men and 34 ng/dl for women, Shionogi assay). If we use BNP test characteristics from Rochester, Minnesota (sensitivity 80% for men, 100% for women; specificity 86% for men, 92% for women) (11) that used a threshold at the 95th percentile for the Biosite assay (52 ng/dl for men, 93 ng/dl for women), the cost-effectiveness of screening is more attractive for men ($19,500 per QALY gained) and women ($63,600 per QALY gained).

If the accuracy of echocardiography is equal to radionuclide or contrast echocardiography for determination of EF (sensitivity and specificity 100%), then screening men with BNP becomes more attractive ($18,100 per QALY gained compared to not screening). However, screening all men with echocardiography becomes more attractive as outcome is improved over initial BNP screening at a cost of $75,700 per QALY gained. For women, BNP screening would cost $57,500 per QALY gained compared to no screening if echocardiography had 100% accuracy.

Screening men with BNP remained economically attractive over a wide range of test costs (Figure 3). We assumed that the cost of echocardiography (including two-dimensional and Doppler) is $420. However, if only a limited study (focused two-dimensional, no Doppler: $98) were required, screening first with BNP would cost $15,700 per QALY compared to no screening in men and $52,600 per QALY gained in women. At this reduced cost, echocardiography compared to initial screening with BNP would cost $55,600 per QALY gained for men but still be dominated by initial BNP screening for women.

There may be unforeseen economic consequences of labeling a patient with cardiac disease, such as difficulty obtaining insurance or employment. Screening would be unattractive (>$100,000 per QALY gained) if a diagnosis led to a loss of lifetime income of $30,000 for males or $7,000 for females. However, if treatment delays the onset of symptomatic HF, then a diagnosis may lead to an increase in income.

In the base case, we calculated an increase in 0.56 QALYs for men and 0.59 QALYs for women with low EF taking ACE inhibitors while asymptomatic, compared with those that wait to start treatment when they develop HF. This may be an overestimate because the SOLVD prevention trial, from which we based our estimate, may have included some patients with symptoms of HF (and more likely to benefit from ACE inhibitors). However, we may have underestimated the benefit of screening if beta-blockers provide a benefit when used in conjunction with ACE inhibitors to prevent HF (23). We examined a scenario where the addition of carvedilol increased survival by 0.66 QALYs for women and 0.78 QALYs for men (assumes a relative mortality risk of 0.7 with beta-blockade) (23). Screening with BNP remained cost-effective for men ($21,300 per QALY gained) and became more attractive for women ($63,800 per QALY gained) compared to no screening.

Although the prevalence of low EF increases with age (10), the potential benefit from treatment may decrease because of shorter life expectancy. The overall effect is that screening becomes slightly more attractive for women and remains attractive for men at older ages. Screening 80-year-old men with a prevalence of reduced EF of 4% would cost $27,306 per QALY gained, whereas screening 80-year-old women with a prevalence of reduced EF of 1.1% would cost $63,450 per life year gained.

We assumed a small decrement in quality-adjusted survival (0.001 years or 0.37 days per patient) to account for potential side effects of ACE inhibitor treatment and for possible increased stress of being labeled as someone with cardiac disease. We know of no studies that have evaluated healthy persons' preferences for quality of life versus length of life with ACE inhibitor use, so the negative health impact of taking unneeded medication is unclear. However, our findings were largely unchanged over a wide range of quality-of-life decrements for ACE inhibitor treatment. The cost-effectiveness of BNP screening compared to no screening in men ranged from $22,000 per QALY gained if there was no decrease in quality-adjusted survival to $41,400 per QALY gained if there was a thirty-day reduction in quality adjusted survival per person.

If we assume that new-onset HF (incidence 0.4% for women age 60 to 69 years and 1% for men age 60 to 69 years (24) is due to low EF in 50% (25), then an estimate of the incidence of asymptomatic low EF is 0.2% per year for women and 0.5% per year for men age 60 to 69 years. If these incidence rates are correct, then yearly screening would be expensive for both men ($113,200 per QALY gained) and women ($174,500 per QALY gained). Repeat screening of men in their 60s would be <$50,000 per QALY gained if screening occurred every three years.

Identifying patients with depressed LV function is important because there is treatment available (such as ACE inhibitors) that is likely to prevent the progression to HF (4). Although screening for depressed ventricular function has been considered (2,26), it has generated little enthusiasm, in part because a cost-effective screening test has not been identified. Our study suggests that BNP testing followed by echocardiography is a cost-effective screening strategy for men and possibly women at age 60 years. Although BNP has only moderate sensitivity and specificity compared to echocardiography for identifying LV dysfunction (9), we found that for every 125 men screened, one year of life would be gained at a cost of $23,500. This is comparable to the estimated $16,000 cost per life year gained with annual mammography for women age 50 to 79 years (27).

The cost-effectiveness of screening was highly sensitive to the prevalence of depressed EF. This explains why screening men (prevalence 3.5%) was more economically attractive than screening women (prevalence 0.45%). Our analysis indicates that screening with BNP will cost <$50,000 per QALY gained if the prevalence is at least 1%. Although we used the gender specific prevalence of disease from populations within the U.S. (Framingham, Massachusetts; and Rochester, Minnesota), studies from Europe are consistent with these estimates. Davies reported a prevalence of asymptomatic low EF (<40%) of 1% in primary care patients age 45 years and greater (28). Data from community cohorts in Scotland have found a 5% prevalence of depressed EF in the general population over age 55 years (29). One-half of these patients were asymptomatic, suggesting that as a group they would benefit from screening with BNP (prevalence of asymptomatic low EF of 2% to 3%).

Other populations will have a higher prevalence of depressed EF, and screening these groups should be highly cost-effective. Patients with prior atherosclerosis as manifested by cerebrovascular accidents, transient ischemic attacks, or peripheral vascular disease had a 28% prevalence of depressed EF with 44% being asymptomatic (30). Patients with a prior history of ischemic heart disease had a 12.1% prevalence of EF of 30% or less. Although some of these patients will already be treated with ACE inhibitors (where additional treatment may not improve outcome), screening should be considered for those untreated.

We based our gender-specific sensitivity and specificity of BNP screening on the Framingham population (9), which are less favorable than data from other community studies. A report from Rochester, Minnesota, compared the Biosite and Shionogi BNP assays and found that using the 95th percentile threshold for the Biosite assay (52 ng/dl for 60-year-old men, 93 ng/dl for 60-year-old women) led to a sensitivity 80% for men and 100% for women (specificity 86% for men, 92% for women) (11). The Shionogi assay (also used by the Framingham study) had lower sensitivities (50% for men, 50% for women) but higher specificities (92% for men, 94% for women). Because the Rochester random sample of the community included some symptomatic patients, it may not be representative of an asymptomatic population. However, even when the better test characteristics from Rochester were used in the analysis, the cost-effectiveness of screening improved only slightly for both men and women. Thus, the cost-effectiveness of screening was only mildly sensitive to the variation in test characteristics.

In contrast to screening middle-age to elderly patients, screening younger patients is economically unattractive. When two population studies from Great Britain are combined (28- 29), only two of 840 (prevalence 0.24%) asymptomatic women age 45 to 54 years had depressed LV function. Among 257 asymptomatic men <45 years of age only one (prevalence 0.4%) had depressed LV function (28- 29). The cost-effectiveness of screening with BNP compared to no screening for these younger populations would be more than $150,000 per QALY. Asymptomatic patients with a normal electrocardiogram (ECG) are another group that may not be attractive to screen with BNP. Among 676 patients with a normal ECG and without symptoms of HF, 5 (0.7%) had reduced EF (31).

Our conclusion, that screening men and perhaps women appears to be cost-effective when compared with other medical interventions, is different from the conclusion of the Framingham Heart Study on which we based many of our assumptions (9). The Framingham Study noted that BNP was not an “optimal” diagnostic test for depressed ventricular function, with which we agree. We confirmed that relying on BNP alone to decide on treatment is not a cost-effective strategy. However, when echocardiography is used to confirm positive BNP tests, screening appears to be economically attractive for certain patient groups.

The study is limited by the absence of data on the effect of ACE inhibitors in patients with no known cardiac disease. We relied on estimates from the SOLVD prevention trial to estimate the benefit of ACE inhibitors in asymptomatic patients. Although these patients were considered asymptomatic, this population came through unknown routes to clinical attention for identification of cardiac disease. Thus, their event rate is likely to be higher, and the effect of ACE inhibitors greater than for patients with unsuspected LV dysfunction. However, if beta-blockers are shown to prevent HF, then we may have underestimated the potential value of screening.

Although we accounted for a quality-of-life decrement for patients receiving a positive test, we may have underestimated the repercussions of a diagnosis of LV dysfunction. In addition, there are financial consequences if the ability to obtain insurance and employment is limited. These issues will be most significant for young patients, where many positive test results will be false positives because of the low prevalence of disease.

We did not include the potential screening benefits of identifying diastolic dysfunction or significant valvular disease that may be found with BNP screening. These patients may benefit from more aggressive treatment of hypertension or fluid overload. To the extent that these interventions improve quality of life or survival, including these benefits would make screening more economically attractive. A recent meta-analysis suggests that ACE inhibitors may be more effective for asymptomatic men than women with reduced LV function post myocardial infarction (32). If true for all patients with depressed EF, this would further support screening for men, but in women only at high-risk for heart disease.

We did not evaluate other tests that have been used to estimate LV function, such as chest radiography or electrocardiography. We did not consider chest radiography in our analysis because it has been found to have poor sensitivity in detecting depressed EF and is more expensive than BNP testing (33). A prior report from Scotland found that an abnormal ECG (pathological Q waves, left bundle branch block, ST-segment depression, T-wave abnormalities, LV hypertrophy, or atrial fibrillation or flutter) had a sensitivity of 60% and a specificity of 82% for depressed EF (30% or less) (29). Given the comparable cost of BNP and electrocardiography, BNP screening would be preferred given its slightly better test characteristics. We did not evaluate other blood tests such as pro-BNP (34) because data regarding prevalence and outcome were not available. Despite our positive findings with BNP, additional research should continue to determine the most cost-effective test for identifying depressed LV EF.

In summary, we found that screening with BNP followed by echocardiography in those with an abnormal test was economically attractive for 60-year-old men and possibly for women. Screening all patients with echocardiography was expensive, and relying on BNP alone to decide treatment led to higher cost and worse outcome compared to the sequential BNP-echocardiography strategy. In general, screening with BNP followed by echocardiography is likely to be economically attractive for patient groups with at least a 1% prevalence of moderate or greater LV systolic dysfunction. Additional studies of BNP or other diagnostic tests for low EF in asymptomatic patients are needed to determine the impact of population screening.