Author + information
- Received February 20, 2011
- Revision received March 18, 2011
- Accepted March 18, 2011
- Published online May 1, 2011.
- Virginia L. Priest, BS⁎,
- Paul A. Scuffham, PhD⁎,
- Rory Hachamovitch, MS, MD‡ and
- Thomas H. Marwick, MD, PhD†,‡,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Thomas Marwick, Cardiovascular Medicine J1-5, Cleveland Clinic, Euclid Avenue, Cleveland, Ohio 44195
Emergency department presentations with chest pain are expensive and often unrelated to coronary artery disease (CAD). Coronary computed tomographic angiography (CTA) may allow earlier discharge of low-risk patients, resulting in cost savings. We modeled clinical and economic outcomes of diagnostic strategies in patients with chest pain and at low risk of CAD: exercise electrocardiography (ECG), stress single-photon emission computed tomography (SPECT), stress echocardiography, and a CTA strategy comprising an initial CTA scan with confirmatory SPECT for indeterminate results. Our results suggest that a 2-step diagnostic strategy of CTA with SPECT for intermediate scans is likely to be less costly and more effective for the diagnosis of a patient group at low risk of CAD and a prevalence of 2% to 30%. The CTA strategies were cost saving (lower costs, higher quality-adjusted life-years) compared with stress ECG, echocardiography, and SPECT. Confirming intermediate/indeterminate CTA scans with SPECT results in cost savings and quality-adjusted life-year gains due to reduced hospitalization of patients who returned false-positive initial CTA test. However, CTA may be associated with a higher event rate in negative patients than SPECT, and the diagnostic and prognostic information for the use of CTA in the emergency department is evolving. Large comparative, randomized, controlled trials of the different diagnostic strategies are needed to compare the long-term costs and consequences of each strategy in a population of defined low-risk patients in the emergency department.
- chest pain
- coronary computed tomographic angiography
- single-photon emission computed tomography
- stress electrocardiography
Annually, more than 6 million patients in the United States present to emergency departments (EDs) with chest pain (1). Conservative diagnostic strategies lead to high admission rates (>70%), with many lacking a discharge diagnosis of acute coronary syndrome (ACS) (2). Admission costs are augmented by increasing volumes of imaging tests, and the costs to adopt new technologies. Despite the conservative approach, a small proportion of patients (0.4% to 4%) with ACS are misdiagnosed and discharged without appropriate intervention with double the risk-adjusted mortality of admitted patients (3). Failure to diagnose ischemia as the cause of chest pain is the leading cause of malpractice suits against ED physicians in the United States (4). The adoption of dedicated chest pain centers and chest pain evaluation protocols in the ED may extend the group to include patients with lower probabilities of chest pain.
Patients presenting with chest pain suggestive of ischemia but with no history of coronary artery disease (CAD) and no enzyme or electrocardiographic abnormalities are generally at low risk but can include individuals with CAD. Of several technologies available for the diagnostic workup of these patients, provocative testing such as exercise electrocardiography (ECG), exercise or pharmacological echocardiography (echo), and single-photon emission computed tomography (SPECT) provide functional information about infarction and inducible ischemia, and the strong evidence base for these tests in the prediction of outcomes supports their use in risk stratification (5,6). However, all these strategies require a period of observation before stress testing is safe, and the healthcare costs associated with these protocols for acute chest pain evaluation have been estimated to be between US$10 and US$12 billion per year in the United States (1). Because coronary computed tomographic angiography (CTA) does not need cardiac stress testing, it has the advantage of avoiding this observation period. Although the negative predictive value of a normal scan is reported to be high, the predictive value of a positive test result is variable and low in the presence of a low prevalence of CAD. Therefore, patients with lesions of intermediate severity may require further investigations to identify the true positive patients (7). We sought to identify the relative cost-effectiveness of diagnostic imaging strategies in a low-risk patient population presenting to the ED with chest pain.
A decision analytic model using standard software (TreeAge Pro, Williamstown, Massachusetts) was developed to estimate and compare the cost and health outcomes of strategies used to identify patients with chest pain who, on presentation to the ED, are assessed as having a low risk of CAD (after exclusion of patients with proven ACS on the basis of electrocardiographic changes). Diagnostic strategies compared were exercise ECG, exercise and pharmacological stress echo, and SPECT, a CTA-only strategy and CTA with confirmatory exercise SPECT for intermediate or indeterminate scans (CTA with SPECT) (Fig. 1). Cost and health outcomes (including death and infarction) relating to CAD and its correct diagnosis were assessed at 12 months to capture the immediate consequences of diagnosis and because published evidence of patient outcomes is available at this time point to inform the model.
Accuracy of the Tests
Sensitivities and specificities of the various tests were obtained from key studies in the literature (Table 1). The diagnostic accuracy of exercise ECG was sourced from relevant guidelines, and those for stress echo and SPECT were sourced from a large published meta-analysis (5). The accuracy of CTA was sourced from a published trial of its use in the ED setting (8). The prevalence, or pre-test probability, of CAD was varied from 2% to 30%.
Positive Diagnostic Test Results
Patients testing positive proceeded to invasive coronary angiography. For the CTA with SPECT strategy, positive results for CTA were classified into high-grade stenoses (>75% luminal diameter; incidence, 8%) and intermediate/indeterminate stenoses. All patients with >75% stenoses then underwent invasive coronary angiography, with the remainder undergoing a period of observation and confirmatory SPECT (7). Positive findings on angiography were associated with costs and health outcomes for the treatment of CAD and an increased risk of cardiovascular events (4.6% per annum) compared with patients judged normal on diagnostic tests (9). Patients with positive diagnostic tests who subsequently tested negative on angiography were assumed to have a cardiovascular event rate of 0.6% per annum compared with patients with normal test results (10).
Negative Diagnostic Test Results
Patients with negative test results on the initial screen were expected to be discharged from the ED without further investigation. For the base case analysis, the accrued event rates in these patients at 12 months were 0.58% for SPECT and 1.03% for echo (11). Comparable results for CTA and ECG were calculated using the data from 2 studies that compared the outcomes in patients tested with stress echo and stress ECG (12) and 1 in which all patients were tested with SPECT and CTA (13).
For those who experienced an event during the follow-up period, the risk of mortality was assumed to be 7% (14).
All costs were converted to 2010 values using the U.S. Consumer Price Index for medical care (Table 2). Costs associated with diagnostic strategies were calculated from average national reimbursement in the United States using Current Procedural Terminology codes and Clinical Classifications Software (15,16). Because of the composite contribution of evaluation, monitoring, and therapies, the cost of observation was based on the average reimbursement at our facility. Patients positive on angiography were assumed to incur an average cost of $27,539 for the initial hospitalization and $4,818 in further treatment costs in the year after discharge (17). The cost of an event was approximated by the weighted average of costs associated with relevant diagnoses as classified by Diagnostic Related Group codes.
Health-related quality of life
Utility weights associated with the health states in the model were sourced from the published literature where available (Table 2) (18,19). A disutility of 2% of normal health was applied to patients with a false-positive diagnosis to account for the effects of stress for being falsely diagnosed with CAD and undergoing angiography.
Costs and outcomes
The average costs, quality-adjusted life-years (QALYs), and comparative cost-effectiveness results from testing a population with 2% to 30% prevalence of CAD with each diagnostic strategy at 12 months are reported in Table 3.
In all cases, the CTA with SPECT strategy accrued slightly higher QALYs than the CTA-only strategy because the additional diagnostic test ruled out some patients with false-positive CTA scans who were spared having to undergo angiography. Overall, exercise echo accrued more QALYs than the exercise SPECT strategy because echo has a lower rate of false positives. However, SPECT had a lower incidence of events than exercise echo in patients who tested negative. Exercise ECG, which had the lowest sensitivity and specificity of the diagnostic strategies modeled, correspondingly accrued fewer QALYs.
Costs increased with increasing prevalence of CAD. The strategies that included CTA as a first-line test were least costly. CTA scans with SPECT were less expensive than the CTA-only strategy due to reduced hospital costs in patients who return a false-positive CTA test.
The CTA strategies were cost saving (lower costs, higher QALYs) compared with exercise stress ECG, echo, and SPECT. Confirming intermediate/indeterminate CTA scans with SPECT results in cost savings and QALY gains due to reduced hospitalization of patients who returned a false-positive initial CTA test.
Sensitivity analyses were used to explore differences in the type of stress-based test used and variations in the incidence of event rates and costs attributed to each test.
Use of Pharmacological Stress
Use of pharmacologically induced stress, rather than exercise, did not change the order of the results, and the CTA strategies remained cost saving compared with stress-based tests. Dipyridamole was the least costly agent for echo, and adenosine performed best in the SPECT arm (Table 4).
Variations in Event Rates in Patients with Negative Test Results
Increasing the event rate in patients who test negative on CTA did not appreciably change the results. However, the model was limited to the medical costs incurred for events and did not incorporate those that may arise from litigation or malpractice.
Changes in the Relative Costs of the Diagnostic Tests
At a CAD prevalence of 5%, CTA with SPECT was the least expensive strategy up to a price per CTA scan of $4,600. The CTA-only strategy was less expensive than the stress-based tests up to a cost per scan of $3,500 or if the cost of observation was greater than $400. Changes in other cost variables did not affect the order of the results.
The diagnostic accuracies of the tests (sensitivity, specificity, and the corresponding positive and negative predictive values), rather than prognostic outcomes (event rates) at 12 months, were used to validate the results (Table 5). SPECT has low accuracy and a high cost differential at low levels of prevalence due to its low specificity rate and high number of false-positive results. However, the high sensitivity of SPECT confers a low false-negative rate. The high specificity of echo produces fewer false positives and greater overall accuracy at lower cost than SPECT, but with a higher proportion of false negatives due to its lower relative sensitivity. Adding a confirmatory SPECT test to the CTA strategy improves accuracy, particularly at low levels of prevalence when the majority of patients who test positive on CTA do not have CAD. CTA with SPECT is the least costly strategy at the point of diagnosis and associated with considerable savings due to reduced observation and hospitalization costs compared with the stress-based tests.
Our results suggest that a diagnostic strategy of CTA with confirmatory SPECT for indeterminate scans may be cost saving in the United States despite modeling a higher event rate in patients who test negative on CTA than SPECT. This work differs from previous work in its inclusion of exercise ECG and echo as diagnostic strategies and prognostic information for events at 12 months follow-up from testing.
Coronary CTA in the ED
The results of this study are consistent with those of other studies on CTA use in the ED. In the randomized, controlled trial by Goldstein et al. (7), CTA with SPECT for intermediate scans resulted in lower costs due to reduced time to diagnosis. A key limitation is the evidence base; our study relied on the diagnostic efficacy of CTA as reported by the only trial conducted in low-risk patients with acute coronary syndrome using 64-slice CTA (8), and prognostic events as reported by van Werkhoven et al. (13) in a prospective observational study. This highlights the need for further research to determine the true diagnostic and prognostic accuracy of CTA in the ED patient population.
Cost-effectiveness of CTA in the ED
There is a dearth of comparative effectiveness and cost-effectiveness information for CTA from clinical trials. To date, 2 studies have considered CTA as a diagnostic strategy in stable presentations with CAD (19,20). Ladapo et al. (20) reported that in a low-risk population presenting with stable chest pain, CTA was associated with an increased cost and an incremental cost-effectiveness ratio of $6,400 in men; however, it was cost saving in women because of the lower prevalence of CAD. Their analysis predicted that CTA would be associated with an extended life expectancy of 10 days in men and 6 days in women. Min et al. (21) presented a lifetime model of costs and consequences in a diagnostic population with intermediate (30%) prevalence of CAD. These authors also report that the CTA with confirmatory SPECT strategy is the least expensive strategy, which is in line with our findings (Table 3).
These studies also report that their analyses are sensitive to changes in the accuracy and relative costs of the diagnostic tests, as well as the prevalence of CAD. Despite differences in diagnostic accuracy among CTA and exercise ECG, SPECT, and echo, the differences in QALYs across diagnostic strategies at 12 months were small in this low-risk population. Interestingly, both Ladapo et al. (20) and Min et al. (21) also found small differences in QALYs across the strategies in lifetime models.
Our results also demonstrate that it is important to include prognostic information to predict longer term health outcomes of the test rather than relying solely on diagnostic accuracy information. Given the paucity of information to inform the CTA arm of our model, we believe that further research should be undertaken to identify its true accuracy (sensitivity and specificity), and trials with longer term clinical follow-up to capture prognostic information in the form health outcomes and demonstrating the comparative cost-effectiveness are needed before widespread adoption of this technology. Clinical trials currently under way may provide evidence of the long-term prognostic value of CTA compared with SPECT, although the results of our analysis also warrant a direct comparison of a CTA strategy with echo, perhaps including a CTA with confirmatory echo strategy.
Our analysis relied on assumptions derived from published evidence, and decision models inevitably simplify a complex situation. Patients re-attending to the ED with recurrent chest pain are not evaluated separately and are anticipated to represent a component of the patients characterized as having no CAD. The model does not incorporate the benefits of revascularization on potentially reducing ED presentations in survivors. Finally, although previous studies categorized different levels of risk in patients with mild stenoses and no stenoses (10), the diagnostic accuracies used in this study relate to a binary cutoff of disease severity (6,8).
The time frame for the analysis is short (12 months). Because data are available to inform the analysis, less extrapolation and reliance on assumptions are required, which reduces uncertainty. However, it does not capture the longer term consequences or costs from clinical decisions made at diagnosis. The accuracy and outcome data for functional testing were derived from the literature, but few of these studies incorporate newer developments, such as the use of echocardiographic contrast agents and gated ventriculography and attenuation correction for SPECT, that may change the diagnostic performance of the tests. Indeed, the use of exercise capacity and heart rate recovery may improve the performance of ordinary stress ECG. On the other hand, none of these are used universally in practice. Finally, the analysis does not incorporate issues of radiation exposure, the safety of which with nuclear and CTA imaging has recently received intense scrutiny. CTA can be performed with low radiation exposure, and avoidance of radiation is prudent in younger patients, especially women.
The salient points of this analysis are summarized in Table 6. The results indicate that a 2-step diagnostic strategy of CTA with SPECT for intermediate scans is likely to be less costly and more effective than stress-based regimens such as ECG, SPECT, and echo. However, the diagnostic and prognostic information for the use of CTA in the ED is evolving. Large comparative, randomized, controlled trials of the different diagnostic strategies are needed to compare the long-term costs and consequences of each testing strategy in a population of defined low-risk patients.
Supported in part by a Program grant (519823) from the National Health and Medical Research Council, Canberra, Australia. The authors have reported that they have no relationships to disclose. Dr. H. William Strauss, MD, served as Guest Editor for this article.
- Received February 20, 2011.
- Revision received March 18, 2011.
- Accepted March 18, 2011.
- American College of Cardiology Foundation
- Heijenbrok-Kal M.H.,
- Fleischmann K.E.,
- Hunink M.G.
- Gibbons R.J.,
- Balady G.J.,
- Bricker J.T.,
- et al.
- Goldstein J.A.,
- Gallagher M.J.,
- O'Neill W.W.,
- et al.
- Meijboom W.B.,
- Mollet N.R.,
- Van Mieghem C.A.,
- et al.
- Antman E.M.,
- Anbe D.T.,
- Armstrong P.W.,
- et al.
- Metz L.D.,
- Beattie M.,
- Hom R.,
- et al.
- Gaibazzi N.,
- Reverberi C.,
- Badano L.
- van Werkhoven J.M.,
- Schuijf J.D.,
- Gaemperli O.,
- et al.
- Heller G.V.,
- Stowers S.A.,
- Hendel R.C.,
- et al.
- Current Procedural Terminology Codes of the American Medical Association
- Healthcare cost and utilization project (HCUPnet)
- Davies A.,
- Hutton J.,
- O'Donnell J.,
- et al.