Diagnosing Acute Cardiac Ischemia in the Emergency Department:
A Cost-Effectiveness Analysis

Catherine Milch, MD

Ethan Balk, MD, MPH

Deeb Salem, MD

Joseph Lau, MD

 

Evidence-based Practice Center

Division of Clinical Care Research, and Division of Cardiology

Department of Medicine

New England Medical Center

Boston, Massachusetts

 

This study was conducted by the New England Medical Center Evidence-based Practice Center under contract to the Agency for Healthcare Research and Quality, contract No. 290-97-0019, Rockville, Maryland.

 

KEY WORDS: cost-effectiveness analysis, acute cardiac ischemia, myocardial infraction, emergency department, triage

Abstract

Objective: To assess the effectiveness of and costs associated with diagnostic tests for detecting acute cardiac ischemia (ACI) among patients presenting to the emergency department (ED).

Design: We developed a decision model with an institutional perspective which evaluates diagnostic test performance, patient outcomes, and associated costs of ED triage.

Data Sources: All studies of diagnostic test performance in the ED and national cost data for patient care.

Target Population: We applied the decision model to two different patient populations: all ED patients presenting with possible ischemia and a low-risk subgroup.

Time Horizon: 30 days from ED presentation.

Interventions: Biomarkers, imaging studies, stress tests, algorithms and computer-based models, and combinations of tests.

Outcome Measures: Appropriate triage (hospitalization) for patients with ACI.

Results: Biomarkers were least costly and least effective for diagnosing ACI; imaging and stress testing were more costly but more effective. Among all ED patients, the Acute Cardiac Ischemia Time-Insensitive Predictive Instrument (ACI-TIPI) was the most effective with an incremental cost-effectiveness ratio of $7,860 per appropriate triage for ACI compared with serial troponin T. The combination troponin T-echocardiography was the next most effective test, but it was associated with an incremental cost-effectiveness ratio of $27,000 compared with serial troponin T. It was substantially less cost-efficient than ACI-TIPI. Among low-risk patients, exercise electrocardiogram testing (ETT) and sestamibi imaging were the most effective diagnostic tests. ETT was nearly $700 per patient less costly than sestamibi imaging and had a low incremental cost-effectiveness ratio of $2,705 per appropriate triage for ACI compared with single troponin T. Increasing the prevalence (likelihood) of ACI reduced cost-effectiveness ratios.

Conclusions: ACI-TIPI and combination troponin T-echocardiography among all ED patients, and ETT and sestamibi imaging among low-risk ED patients, are effective diagnostic tests for detecting ACI in the ED. ACI-TIPI and ETT are effective and cost-efficient options at low to high rates of ACI prevalence.

Introduction

Over six million patients present to emergency departments (EDs) in this country with symptoms of possible acute cardiac ischemia (ACI), yet over half do not have ACI.1 An ED physician must decide which patients require admission and treatment for ACI, which includes both acute myocardial infarction (AMI) and unstable angina pectoris (UAP). Over the past decade, a variety of tests have gained popularity for the diagnosis of ACI, including the biochemical markers troponin and myoglobin, and cardiac perfusion imaging studies such as sestamibi scans.

Few studies have compared the costs and implications of diagnostic tests for ACI in the ED. The more accurate diagnostic tests may also be more costly. An explicit analysis of the trade-offs between cost and effectiveness of alternative tests may assist ED physicians in their choice of diagnostic tests.

To address these issues, we developed a decision analytic model that assesses triage outcomes and costs for 12 individual tests and four combinations of tests for the diagnosis of ACI in ED patients. Because some of these diagnostic strategies are not applicable for certain patients, we applied the decision model to two different patient populations: all ED patients presenting with possible ischemia, and a low-risk subgroup in whom the presenting ECG was normal or nondiagnostic. We focused our analyses on the diagnosis of and triage for ACI rather than long-term management and life expectancy. Thus, our effectiveness measure was appropriate triage for ACI. We used actual test diagnostic performance data obtained from high-quality studies performed in the ED.

We focused our evaluation on ACI, which includes both myocardial infarction and unstable angina, because unstable angina is a common condition, is often hard to differentiate from AMI in the ED, and necessitates appropriate evaluation and management. The decision model shows how the diagnostic performance of a test affects total costs and appropriate triage for patients with ACI when the test is applied to patients presenting to the ED with signs and symptoms of ACI.

Methods

Overview

The decision analysis model compared the costs and patient outcomes associated with using different diagnostic tests for patients presenting to the ED with symptoms suggestive of ACI. The model represents the triage decisions and 30-day patient outcomes as a consequence of a negative or positive test result. All patient dispositions and outcomes occurring within 30 days of ED presentation, including initial and subsequent admission to the hospital, sequelae from “missed” AMI or UAP, outpatient follow-up, and death, were included in the model (Figure 1). Patients with either UAP or AMI were considered to have ACI; patients with stable angina were considered to have nonacute cardiac ischemia. Patients in cardiac arrest were not considered in the analyses.

The tests and combinations of tests evaluated in the decision analysis included those that are commonly available and have been evaluated in ED patients. Tests were chosen based on the National Heart Attack Alert Program recent update1 and an extensive systematic review of diagnostic tests for ACI.2–4 Tests and combinations of tests that had not been evaluated in a population of ED patients were excluded (such as stress tests with imaging). We evaluated 16 tests:

Serum biochemical markers: single and serial CK-MB, troponin T, and myoglobin;

Electrocardiogram-based tests, algorithms, and instruments: continuous and/or serial electrocardiograms (ECG), Acute Cardiac Ischemia Time-Insensitive Predictive Instrument (ACI-TIPI), Goldman chest pain protocol, exercise stress ECG testing;

Imaging studies: rest echocardiography and rest sestamibi perfusion scans;

Combinations of two tests: single and serial CK-MB and myoglobin; single CK-MB and serial ECG; single troponin T and echocardiography.

A single biochemical test was defined as one that occurred within the initial 4-hour period after presentation to the ED; serial testing was defined as repeated testing occurring within a period of up to 6 hours after presentation to the ED.

Decision Model

Because certain diagnostic tests (such as stress testing) can not or should not be used in some patients with possible ACI, we applied the decision model to two different patient populations with possible ACI: all ED patients and a low-risk subgroup with normal or non-diagnostic ECG on presentation to the ED. Thus, tests that may be applicable to or have been evaluated in only low-risk patients, such as stress testing, serial ECGs, and sestamibi imaging, were evaluated in the low-risk subgroup only. Because some ED patients may be at high risk for AMI, we did not evaluate these diagnostic tests in the analyses of all ED patients. Additionally, we did not include two tests, ACI-TIPI and the Goldman chest pain protocol, in the low-risk subgroup analyses because their diagnostic performance has not been evaluated in ED subgroups.

To reflect the differences in pretest likelihood of ACI among patients in the two populations, we used different prevalence rates for both AMI and ACI. Prevalence rates for ACI and AMI in the general ED patient population were obtained from a large clinical trial of patients presenting to EDs with any sign or symptom suggestive of ACI.5 The prevalence rate for AMI among “low-risk” patients was estimated to be approximately half that among all ED patients.6 The prevalence of UAP was the same in both models. Additionally, because ACI prevalence rates vary among EDs, we performed sensitivity analyses altering the prevalence rates of ACI in both models.

Test Performance Data

Test diagnostic performance data were obtained from published results of all studies performed in ED patients between 1966 and 1999 and that were included in a recent extensive systematic review.1,2 Because patient inclusion criteria, ACI prevalence rates, and reported diagnostic test performance varied among individual studies, we relied on meta-analyses or pooled results. For biochemical tests, values for AMI sensitivity and ACI specificity were based on results obtained from meta-analyses of all studies of ED patients.3 For the low-risk subgroup, we used test sesitivity based on the lower 95% confidence interval from the meta-analyses to reflect the lower test sensitivity in low prevalence populations. Because data on UAP sensitivity were sparse, we used test sensitivity for detection of coronary artery disease requiring revascularization in those studies that reported results.

Test diagnostic data for nonbiochemical tests were obtained from individual studies, or pooled results, in the appropriate ED population.2,4 Because ACI-TIPI reports a predicted probability of ACI for a patient instead of a dichotomous test result, its diagnostic performance values were based on a probability cut-point that approximates its clinical effectiveness as observed in a large clinical trial.5 Table 1 shows the sensitivity and specificity of each test and data sources.

Patient Disposition

The triage of an ED patient was determined by the test result. A positive test result would lead to hospitalization and a negative test result would lead to discharge from the ED. Appropriate triage was defined as hospitalization for any patient with ACI; inappropriate triage was defined as discharge from the ED of a patient with ACI. Thirty-day outcomes were determined by patient’s true diagnosis (ACI or non-ACI), patient triage (hospitalization or discharge from the ED based on diagnostic test result), follow-up evaluation, and the risk of survival or death from appropriate or inappropriate triage. Inpatient and outpatient mortality rates for patients with ACI appropriately hospitalized or inappropriately discharged from the ED and subsequent hospitalization rates for patients with ACI inappropriately discharged were based on national data from large clinical trials of ED patients.45–47 The transition probabilities affecting patient disposition are shown in Table 2.

Patient disposition assumptions included: 1) all patients with ACI survive or die, and a certain percentage of patients will die from ACI, regardless of appropriate or inappropriate triage; 2) all patients with ACI who survive are either hospitalized or undergo outpatient evaluation; 3) all patients with ACI who are hospitalized receive definitive treatment for ACI during the hospitalization; 4) a small percentage of patients with UAP who are inappropriately discharged from the ED will subsequently develop AMI; 5) all patients without ACI survive; and 6) all patients with serious non-ACI disease (but in whom the diagnosis of ACI was entertained, such as those with valvular heart disease, pulmonary embolus, biliary tract disease, etc) receive definitive treatment for their condition but their outcomes and associated costs of care are not included in the decision analysis. Table 3 lists the 14 possible patient dispositions.

We did not include complications that may arise from some of the tests, such as stress testing, because of the extremely low rate of death or clinically significant complications reported in studies. (None of the studies evaluating the use of sestamibi imaging or exercise ECG testing in ED patients with possible ACI reported deaths or significant morbidity such as complications altering the care given to a patient). Also, these tests were only evaluated in low-risk patients.28–30,48 We also excluded complications from in-hospital treatment for ACI (such as restenosis after angioplasty) because the focus of the analysis was on triage for, rather than management of, ACI.

Costs

The total costs used in the analysis represent the total reimbursement to the hospital and outpatient clinic for patient services for 30 days from the initial ED visit. The total cost of using a test included not only the cost of the test itself, but also the cost of subsequent patient management (such as, hospitalization and outpatient follow-up) and outcome (Table 4). Costs associated with treatment for conditions other than suspected ACI were not considered. The diagnostic test used in the ED and the level of suspicion for ACI determined the extensiveness and costs of further diagnostic testing after discharge from the ED. Four different outpatient follow-up scenarios, based on likelihood of ACI, were evaluated in the low-risk subgroup model because more intensive and costly studies, such as echocardiography and sestamibi imaging, may lead to less costly outpatient evaluations after ED discharge. For example, an outpatient evaluation of a patient in whom the likelihood of ACI is considered very low may not have a follow-up stress test after a negative sestamibi scan in the ED.

Test costs and reimbursements for hospital admission were based on 1999 national median fees49 and average national payments for specific DRG codes.50 These are shown in Table 5. Because the standard of care is to obtain an ECG on all ED patients evaluated for ACI, we did not use an additional cost for an initial ECG. Outpatient visit reimbursements were calculated from median fees for tests performed as part of the outpatient work-up and for the professional component of the outpatient visit.50 Discounting was unnecessary because of the short time horizon of the analysis.

Cost Effectiveness

The cost-effectiveness (CE) of a specific diagnostic test was the sum of all the costs incurred during the 30-day period from ED presentation, divided by its effectiveness. The effectiveness of a test, determined by its ability to detect UAP or AMI in the ED, was defined as the proportion of appropriately triaged (hospitalized) patients with ACI. The decision model projects the costs and number of patients with ACI appropriately triaged from a cohort of 1000 ED patients for each test.

The cost-effectiveness analyses involved comparing incremental cost-effectiveness ratios of tests. First, diagnostic tests were ranked by increasing cost. More expensive tests that were less effective were eliminated by “simple dominance.” For remaining strategies, the incremental cost-effectiveness ratio was calculated as the additional cost to diagnose and appropriately triage one additional patient with ACI compared with the next less costly and less effective alternative. Tests that had a higher incremental CE ratio than more effective alternatives were eliminated by “weak dominance” because they were not as “cost-efficient” as the other tests. Sensitivity analyses on relevant variables were performed to assess the stability of the results. Decision models and cost-effectiveness analyses were created in and performed with Data TreeAge 3.5 software (Williamstown, MA).

Results

Analysis for All Emergency
Department Patients

The prevalence of ACI and AMI used for the decision model to represent all ED patients was 18% and 8%, respectively. Thus, a test with perfect sensitivity would lead to appropriate triage of all 180 patients with ACI per 1000 ED patients evaluated for ACI.

Triage Accuracy: The proportion of ED patients with ACI who would be correctly detected and hospitalized by each test is shown in Figure 2. As expected, tests with higher diagnostic accuracy for both AMI and UAP had higher values for appropriate triage for patients with ACI. The biochemical tests and the Goldman protocol did not perform as well as echocardiography because they are generally not designed or used to detect UAP. Serial testing or combinations of biochemical tests improved ACI detection, with serial troponin T having the best triage accuracy for ACI among the biomarkers. ACI-TIPI and the combination of troponin T-echocardiography had the best detection rates for ACI because they detect both AMI and UAP.

Base Case Cost-effectiveness Analysis: Figure 3 shows the proportion of patients with ACI appropriately triaged (effectiveness, on the y-axis) by and the associated costs (on the x-axis) of each diagnostic test. Test costs generally increase with increasing accuracy because the cost of appropriate hospitalization is more costly than that of inappropriate discharge in the base case. The single biochemical tests, clustered near the lower left corner of the graph, are the least costly and have the lowest values for appropriate triage. Serial testing improves effectiveness and raises costs. Echocardiography, troponin T-echocardiography, and ACI-TIPI are more effective and more costly.

The tests connected by the line or that lie very close to the line are the nondominated tests and are thus considered the most cost-efficient. That is, for a given cost they are more effective than tests that lie far from the line. Thus, single myoglobin and troponin T, serial myoglobin and troponin T, and ACI-TIPI are cost-efficient tests compared with tests that lie far below the line such as the combination CK-MB and myoglobin or echocardiography. The slope of the line reflects the inverse of the cost-effectiveness ratio. Thus, the flatter the slope, the higher the incremental cost-effectiveness, indicating less additional effectiveness for a given additional cost. The slope of the line connecting single myoglobin and serial troponin T is fairly steep, indicating that the additional cost associated with serial testing leads to substantially more appropriate triage, making serial testing cost-efficient. The slope of the line connecting serial troponin T and ACI-TIPI is less steep, indicating a higher incremental cost-effectiveness ratio between these two tests than between single myoglobin and serial troponin T.

The costs, effectiveness values, and, for nondominated tests, the incremental CE ratios, are shown in Table 6. Eight tests are more effective and less costly than the alternatives: single myoglobin and troponin T, serial CK-MB, myoglobin, troponin T, rest echocardiography, troponin T-echocardiography, and ACI-TIPI. Five of these tests are dominated by “weak” dominance because their incremental CE ratios are higher than that for the next more effective and more costly test. For example, troponin T-echocardiography costs approximately $27 more per patient than echocardiography alone but leads to appropriate triage for 10 additional patients. Thus, its incremental CE ratio is $2,700 (as calculated among 1000 ED patients), which is higher than the incremental CE ratio of the next more effective test, ACI-TIPI ($1,477, calculated by comparing costs and number of patients with ACI appropriately triaged between troponin T-echocardiography and ACI-TIPI). Thus, troponin T-echocardiography is not as cost-efficient as other tests and is eliminated by weak dominance.

Three tests are not dominated by other tests: single myoglobin, serial troponin T, and ACI-TIPI. Although the CE of ACI-TIPI is relatively high compared with serial troponin T, it leads to appropriate triage for 60 additional patients with ACI at an additional cost of $473 per patient.

Sensitivity Analyses for All ED Patients

Variation of ACI Prevalence Rates: The ED physician’s a priori impression of the likelihood of ACI in a patient presenting to the ED with signs and symptoms suggestive of ACI often influences the interpretation of test results as well as the final triage decision. We attempted to model pretest likelihood of coronary artery disease by varying the prevalence of ACI in the model. For example, a 5% prevalence of ACI may represent a low pretest likelihood, such as for a 40-year-old woman with nonspecific chest pain. Moderate pretest likelihood may be modeled by an ACI prevalence rate of 30% (represented by a 45-year-old man with chest pain atypical for angina). A high pretest likelihood may be represented by an ACI prevalence rate of 80%, represented by a 65-year-old man with typical angina.51,52

Varying the ACI prevalence affects the costs and incremental cost-effectiveness ratios of tests. As ACI prevalence increases, the costs and effectiveness values of a test increase linearly. Costs increase because there are more patients with ACI and more hospitalizations. However, the number of appropriate triages for ACI increases more steeply (because test effectiveness is affected by both prevalence of ACI and test sensitivity for ACI), leading to an exponential decrease in incremental CE ratios. Thus, tests become more cost-efficient, because application of diagnostic tests to a cohort of patients with a high ACI prevalence makes each appropriate triage less costly than at low ACI prevalence. The relative effectiveness rankings of tests however are not altered.

Comparing the incremental CE ratios as prevalence changes provides information regarding the most cost-effective option for patients with different likelihood rates of ACI. At very low pretest likelihood rates of ACI (representing a population of ED patients at very low risk for ACI), the incremental CE ratios of all tests are substantially higher than in the base case. ACI-TIPI has a very high incremental CE ratio compared with serial troponin T, nearly $150,000, which is over 15 times its base case ratio, and greater than typical “thresholds” used in cost-effectiveness analyses. As ACI prevalence increases, the differences among incremental CE ratios decrease. At ACI prevalence rates of 30% and 50%, ACI-TIPI has CE ratios of only $4000 and $2600, respectively, compared with serial troponin T. At higher ACI prevalence rates, the Goldman protocol becomes a cost-efficient test along with serial troponin T and ACI-TIPI.

Variation of Test Performance Characteristics: Because ACI-TIPI does not have a cut-point or threshold probability for a “positive” result for detection of ACI, we used various thresholds to evaluate how the effectiveness and cost-effectiveness of ACI-TIPI changes relative to the other tests. As the cut-point for a positive test result increases from 10% to 25% (thereby decreasing the sensitivity but increasing the specificity of the predictive instrument), the costs associated with using ACI-TIPI decrease relative to other tests, so that it is no longer the most costly test to use. At a cut-point as high as 25% (ie, patient has a “positive” test result only if the predictive instrument gives a likelihood of ACI of 25% or more), ACI-TIPI remains the most effective test, with an incremental CE ratio of $6300 per additional patient with ACI appropriately triaged compared with serial troponin T, making it as cost-efficient as the base case. The sensitivity of ACI-TIPI for both AMI and UAP would have to fall to below 70% (a cut-point near 35%) for it to be no longer a cost-efficient alternative.

We also performed sensitivity analyses for the Goldman chest pain protocol because of the difficulty in estimating its diagnostic performance for patients with UAP. Because the protocol is not designed to aid in the detection of patients with UAP, we theoretically altered its sensitivity for UAP. The protocol is dominated by other tests until its theoretical sensitivity for UAP increases to over 40%. At a sensitivity near 50% for UAP, it is the second most effective test along with combination troponin T and echocardiography, with a cost-efficient incremental CE of $6400 compared with troponin T serial.

Variation of Cost of ACI-TIPI: Because ACI-TIPI can be incorporated into ECG machines, the base case analysis assumes no additional cost, over that for an ECG in the ED, for the actual test. However, not every ED has an ECG machine in which the predictive instrument has been incorporated. Thus, hospitals may have to purchase new ECG machines that have the predictive instrument. We performed sensitivity analyses to determine how much ACI-TIPI would have to cost per test to lose its dominant cost-effectiveness status. ACI-TIPI is no longer cost-efficient at a cost of approximately $4000 per patient use. However, the addition of the predictive instrument adds only $800 to $1000 to the total retail cost of each ECG machine.53

Variation of Costs of Patient Dispositions: One assumption of the decision analysis model is that the reimbursement for hospital admission for patients without ACI is less than that for patients with ACI, reflecting the 23-hour “observation” status of most non-ACI patients, as well as the exclusion of non-ACI related treatment costs in the analysis. This assumption favors tests that may be very sensitive but not very specific for ACI. However, the cost of hospital admission for some patients without ACI may exceed the reimbursement for a “rule-out” AMI admission assumed in the model. We therefore performed sensitivity analyses on the cost of hospital admission for patients without ACI to assess the effect on cost-effectiveness of tests. As the cost of inappropriate hospital admission for a patient without ACI increases, the incremental CE ratio of tests with poor specificity increase, making them less cost-efficient.

For example, at a cost of inappropriate admission double that of the base case, the incremental CE ratio of ACI-TIPI compared with serial troponin T increases to approximately $33,000, almost 6 times higher than in the base case. When the cost of inappropriate admission for ACI increases fivefold, the CE ratio of ACI-TIPI increases 10-fold over the base case. Although it remains a very effective test for detecting patients with ACI, its cost-efficiency relative to other, less effective tests decreases.

Sensitivity analysis was also performed on the cost of inappropriate ED discharge of a patient with ACI, which may result in death. We varied the “cost of death” associated with missed ACI from a low of $600 for a return ED visit and resuscitation attempt, to the cost of a malpractice settlement ($2 million). ACI-TIPI retains its cost-effectiveness as the cost of inappropriate ED discharge increases, dominating all other strategies at costs over $400,000.

Low-risk Subgroup Analysis

This model estimates the total costs and effectiveness of applying tests to a low-risk population of ED patients with signs and symptoms of ACI in whom the initial ECG is normal or nondiagnostic. The prevalence rates of ACI and AMI in this population are 13% and 4%, respectively.

Triage Accuracy: Figure 4 shows the percentage of patients with ACI appropriately diagnosed by each of the tests. The results are similar to those for all ED patients. The biomarkers do not perform as well as other tests because they are not used to detect UAP. The imaging studies, ECG exercise testing, and the combination of troponin T-echocardiography have the best triage accuracy. Sestamibi imaging and exercise ECG testing perform nearly equally well, identifying slightly more than 85% of all patients with ACI.

Base Case Cost-effectiveness Analysis: Four different outpatient evaluation scenarios were used for the cost-effectiveness analysis for the low-risk subgroup to illustrate different outpatient evaluation strategies for those patients who undergo more costly and effective tests in the ED such as exercise testing and sestamibi imaging. The base case model assumes that all patients who return for outpatient evaluation have exercise ECG testing, except patients who have had negative exercise testing or sestamibi imaging in the ED. These patients do not have further diagnostic testing during the 30-day follow-up period.

The total costs and proportion of patients with ACI appropriately triaged for each test are shown in Figure 5. The most cost-efficient tests, compared with alternatives, are near or on the line: single myoglobin, CK-MB, and troponin T, serial myoglobin and troponin T, and ECG exercise testing. The slope of the line between single troponin T and exercise ECG testing is fairly steep, indicating a relatively low incremental cost-effectiveness ratio and high cost-efficiency for exercise ECG testing compared with other tests.

Table 7 shows the costs, effectiveness, and incremental cost-effectiveness ratios for each test. Six tests are more effective and less costly than the alternatives: single myoglobin and troponin T, serial CK-MB, myoglobin and troponin T, and exercise ECG testing. The serial biomarkers are eliminated by weak dominance. Exercise testing dominates all the imaging and ECG-based tests because it is less costly and more effective. Although exercise testing and sestamibi imaging are nearly equal in effectiveness, exercise testing costs $700 less per ED patient than sestamibi imaging. Its incremental CE ratio is $2,705 per additional appropriate triage for a patient with ACI compared with single troponin T.

Sensitivity analyses for
Low-Risk Subgroup

Variations in Cost of Follow-up: Changing the outpatient follow-up evaluation changes the cost of follow-up and therefore changes the incremental cost-effectiveness of tests (effectiveness of tests does not change). In the base case, all patients except those who had undergone exercise testing or sestamibi imaging in the ED undergo exercise ECG testing as part of their outpatient follow-up. If all patients who return for outpatient evaluation have no further diagnostic testing, exercise ECG testing is still the most cost-efficient test, but its incremental cost-effectiveness increases to nearly $7000 (from $2700 in the base case) compared with single troponin T.

If patients with a negative exercise test in the ED have stress sestamibi imaging, instead of follow-up exercise ECG, as part of their outpatient evaluation, the costs associated with a false negative exercise test increase. ECG exercise testing becomes more expensive than sestamibi imaging (about $2500 per ED patient), and its incremental CE ratio compared with single troponin T increases fourfold. Because exercise testing and sestamibi imaging have nearly the same effectiveness, both tests would have nearly identical CE ratios. If stress sestamibi is part of the outpatient evaluation for patients who have had sestamibi imaging in the ED, then sestamibi imaging once again becomes more costly than exercise testing, and exercise testing dominates. Thus, as the cost of a false-negative ECG exercise test increases, it becomes less cost-efficient than in the base case. However, it remains a cost-effective alternative due to its sensitivity in detecting AMI and relatively low cost compared with other tests.

Variations in ACI Prevalence Rates: As in the analyses of all ED patients, as prevalence of ACI increases, the costs, and the effectiveness of all strategies increase linearly, and the cost-effectiveness of the tests decrease exponentially. The relative CE among the tests changes little as prevalence of ACI increases. Exercise testing and sestamibi imaging are the most effective strategies at all prevalence rates. Because exercise testing is less costly than sestamibi imaging, it also dominates sestamibi imaging at all prevalence rates. Its incremental CE ratio does not change substantially at ACI prevalence rates above that in the base case, remaining around $1800.

Discussion

This cost-effectiveness analysis attempts to incorporate all costs associated with use of diagnostic tests in the ED. Considering only the cost of a diagnostic test neglects the effect of the test on patient triage. Generally, more effective tests, such as imaging studies, lead to higher total costs than less effective tests because: 1) the tests themselves cost more (eg, sestamibi imaging vs biomarkers), and 2) for patients with ACI, the costs of hospitalization exceed those for discharge home and outpatient follow-up. However, more-effective tests also lead to fewer inappropriate hospitalizations for patients without ACI, so the ratio of total costs to the cost of a test decreases exponentially as test effectiveness increases. Thus, the more effective (and more costly) tests lead to proportionately lower total costs than less effective (and less costly) tests.

The results of the decision analysis indicate that the biomarkers have the lowest triage accuracy for patients with ACI, primarily because their diagnostic performance in patients with UAP is poor. The ECG-based tests, algorithms, and combinations of tests perform better. Among all ED patients, ACI-TIPI has the best triage accuracy for ACI and is a cost-efficient test with a relatively low incremental CE. Among low-risk patients, sestamibi imaging and ECG exercise testing are very effective for triage of patients with ACI, but exercise testing is substantially less costly, and thus more cost-efficient. As ACI prevalence or pretest likelihood for ACI increases, incremental CE ratios decrease, making both ACI-TIPI and exercise ECG testing more cost-efficient than in the base case.

Sestamibi imaging has excellent diagnostic accuracy for ACI but has not been tested in ED patients at moderate or high risk for ACI. Exercise testing may not be applicable to the majority of ED patients presenting with possible ACI, specifically patients at moderate to high risk for ACI. Sestamibi imaging may be a safer choice for these patients. Although we evaluated sestamibi imaging only in the low-risk subgroup, it may be applicable to a more general population of ED patients because of its low-risk of complications. The cost-effectiveness of exercise ECG testing and sestamibi imaging in a more general ED population requires further evaluation.

There are several limitations to the decision analysis. Little data exist on the diagnostic performance of most tests for UAP, thus the values for UAP sensitivity used in the decision model are estimates, which add uncertainty to the cost-effectiveness analyses. Also, we did not model the possible complications arising from waiting for test results in the ED (such as serial biomarkers) because of lack of data.

All tests were compared equally in the decision analysis despite differences in their application. Exercise tests and imaging studies require specialized equipment and trained personnel to administer and interpret results. These tests may not be available in all EDs or on a 24-hour basis as is possible with serum tests. Additionally, some tests, such as exercise ECG testing and sestamibi imaging, may be used to predict prognosis,52 and thus may provide information to clinicians beyond detection of ACI. For example, sestamibi imaging may be able to detect the “sickest” patients with ACI. This may be reflected in decreased mortality rates compared with other tests (ie, biomarkers) in patients with ACI but negative test results (ie, patients inappropriately discharged from the ED).

Finally, test diagnostic performance values used in the analyses were obtained from published reports and may not reflect ED physicians’ decisions on patient triage. Data on how results from diagnostic tests influence physicians’ triage decisions are lacking for most of the tests other than ACI-TIPI, and reliance on published data of test performance may not accurately reflect actual triage decisions in the ED. Test diagnostic performance varies among different patient populations, as shown in Table 1. Furthermore, triage decisions are often determined in the context of a patient’s pretest likelihood of ACI, and not based solely on diagnostic test results.

Despite these caveats, our study has several strengths. The analyses focused on how test diagnostic performance affects patient triage in the ED. We did not include long-term management and outcomes to prevent obscuring the effects of these tests in the ED. Our outcome measure was appropriate triage for ACI and not quality of life outcomes because the focus of the analysis was detection of ACI in the ED. The short (30-day) time horizon reflects this as well.

We attempted to make the decision model generalizable and reflect reality. Thus, data on diagnostic test performance were obtained from a recent extensive systematic review and meta-analysis of all diagnostic tests for ACI in the ED. We also varied test diagnostic performance to reflect how tests would perform among specific patient populations, such as those with non-diagnostic ECG changes. We used national data for costs and for most transition probabilities for patient disposition and outcomes.

Our definition of test “effectiveness” was its sensitivity for ACI, and did not include its actual use in clinical practice (except for ACI-TIPI) because data on clinical effectiveness for most of the evaluated tests was exceedingly sparse. Furthermore, a test’s sensitivity and specificity, may represent its “true” triage accuracy. The contribution of clinical decision-making is uncertain as it may sometime lead to incorrect diagnosis and triage. However, in an attempt to capture aspects of triage decision-making, we varied ACI prevalence to model pretest likelihood.

The results of the cost-effectiveness analyses are not intended to direct clinical recommendations for individual patients because the decision models apply to populations of ED patients. The most “effective” or “cost-effective” tests may not be appropriate for a particular patient. Furthermore, age-related mortality rates were not explicitly modeled. Additionally, the tests evaluated in the decision analysis may not be applicable even in the patient population in which they were evaluated. For example, stress tests may be highly restricted even in the low-risk patient population, The results of the decision analysis should be used for understanding the factors that are involved in and as an aid in decision-making for triage of patients with ACI in the ED. Prospective trials on the effect of individual diagnostic tests on ED patient triage and outcomes are required before definitive conclusions can be made.

References

1. Lau J, Ioannidis JPA, Balk EM, et al: Evaluation of technologies for identifying acute cardiac ischemia in emergency departments. Evidence Report/Test Assessment Number 26. (Prepared by New England Medical Center Evidence-based Practice Center under Contract No. 290–97–0019.) AHRQ Publication No. 01–E006, Rockville, MD: Agency for Healthcare Research and Quality, May, 2001.

2. Lau J, Ioannidis JPA, Balk EM, et al: Diagnosing acute cardiac ischemia in the emergency department: A systematic review of the accuracy and clinical effect of current technologies. Ann Emerg Med 37:453–60, 2001.

3. Balk EM, Ioannidis JPA, Salem MD, et al: Accuracy of biomarkers to diagnose acute myocardial infarction in the emergency department: A meta-analysis. Ann Emerg Med 37:478–94, 2001.

4. Ioannidis JPA, Salem MD, Chew PW, Lau J: Accuracy of imaging technologies in the diagnosis of acute cardiac ischemia in the emergency department: A meta-analysis. Ann Emerg Med 37:471–77, 2001.

5. Selker HP, Beshansky JR, Griffith JL, et al: Use of the acute cardiac ischemia time-insensitive predictive instrument (ACI-TIPI) to assist with triage of patients with chest pain or other symptoms suggestive of acute cardiac ischemia: A multicenter controlled clinical trial. Ann Intern Med 129:845–55, 1998.

6. Pope JH, Ruthazer R, Beshansky JR, et al: Clinical features of emergency department patients presenting with symptoms suggestive of acute cardiac ischemia: A multicenter study. J Thromb Thrombol 6:63–74, 1998.

7. Brogan GX Jr, Friedman S, McCuskey C, et al: Evaluation of a new rapid quantitative immunoassay for serum myoglobin versus CK-MB for ruling out acute myocardial infarction in the emergency department. Ann Emerg Med 24:665–671, 1994.

8. Hedges JR, Gibler WB, Young GP, et al: Multicenter study of creatine kinase-MB use: Effect on chest pain clinical decision making. Acad Emerg Med 3:7–15, 1996.

9. Hedges JR, Rouan GW, Toltzis R, et al: Use of cardiac enzymes identifies patients with acute myocardial infarction otherwise unrecognized in the Emergency department. Ann Emerg Med 16:248–252, 1987.

10. Hamm CW, Goldmann BU, Heeschen C, et al: Emergency room triage of patients with acute chest pain by means of rapid testing for cardiac troponin T or troponin I. N Engl J Med 337:1648–1653, 1997.

11. Laurino JP, Bender EW, Kessimian N, et al: Comparative sensitivities and specificities of the mass measurements of CK-MB2, CK-MB, and myoglobin for diagnosing acute myocardial infarction. Clin Chem 42:1454–1459, 1996.

12. Montague C, Kircher T: Myoglobin in the early evaluation of acute chest pain. Am J Clin Pathol 104:472–476, 1995.

13. Gornall DA, Roth SN. Serial myoglobin quantitation in the early assessment of myocardial damage: A clinical study. Clin Biochem 29:379–384, 1996.

14. Kennedy RL, Harrison RF, Burton AM, et al: A system for diagnosis of acute myocardial infarction (AMI) in the accident and emergency department: Evaluation and comparison with serum myoglobin measurements. Comput Methods Programs Biomed 52:93–103, 1997.

15. Levitt MA, Promes SB, Bullock S, et al: Combined cardiac marker approach with adjunct two-dimensional echocardiography to diagnose acute myocardial infarction in the emergency department. Ann Emerg Med 27:1–7, 1996.

16. Mohler ER 3rd, Ryan T, Segar DS, et al: Clinical utility of troponin T levels and echocardiography in the emergency department. Am Heart J 135:253–260, 1998.

17. Green GB, Beaudreau RW, Chan DW, et al: Use of troponin T and creatine kinase-MB subunit levels for risk stratification of emergency department patients with possible myocardial ischemia. Ann Emerg Med 31:19–29, 1998.

18. Kontos MC, Jesse RL, Anderson FP, et al: Comparison of myocardial perfusion imaging and cardiac troponin I in patients admitted to the emergency department with chest pain. Circulation 99:2073–2078, 1999.

19. Sabia P, Afrookteh A, Touchstone DA, et al: Value of regional wall motion abnormality in the emergency room diagnosis of acute myocardial infarction: A prospective study using two-dimensional echocardiography. Circulation 84(3 Suppl):I85–92, 1991.

20. Peels CH, Visser CA, Kupper AJ, et al: Usefulness of two-dimensional echocardiography for immediate detection of myocardial ischemia in the emergency room. Am J Cardiol 65:687–691, 1990.

21. Kontos MC, Arrowood JA, Jesse RL, et al: Comparison between 2-dimensional echocardiography and myocardial perfusion imaging in the emergency department in patients with possible myocardial ischemia. Am Heart J 136(4 Pt 1):724–33.1998.

22. Sasaki H, Charuzi Y, Beeder C, et al: Utility of echocardiography for the early assessment of patients with nondiagnostic chest pain. Am Heart J 112:494-497, 1986;

23. Trippi JA, Lee KS, Kopp G, et al:  Dobutamine stress tele-echocardiography for evaluation of emergency department patients with chest pain. J Am Coll Cardiol 30:627–32, 1997.

24. Trippi JA, Kopp G, Lee KS, et al: The feasibility of dobutamine stress echocardiography in the emergency department with telemedicine interpretation. J Am Soc Echocardiogr 9:113–116, 1996.

25. Stewart RE, Dickinson CZ, Weissman IA, et al: Clinical outcome of patients evaluated with emergency centre myocardial perfusion SPECT for unexplained chest pain. Nucl Med Commun  17:459–62, 1996.

26. Kontos MC, Jesse RL, Schmidt KL, et al: Value of acute rest sestamibi perfusion imaging for evaluation of patients admitted to the Emergency department with chest pain. J Am Coll Cardiol 30:976–982, 1997.

27. Hilton TC, Thompson RC, Williams HJ, et al: Technetium-99m sestamibi myocardial perfusion imaging in the emergency room evaluation of chest pain. J Am Coll Cardiol 23:1016–1022, 1994.

28. Kirk JD, Turnipseed S, Lewis WR, Amsterdam EA: Evaluation of chest pain in low-risk patients presenting to the emergency department: The role of immediate exercise testing. Ann Emerg Med 32:1–7, 1998.

29. Lewis WR, Amsterdam EA, Turnipseed S, Kirk JD: Immediate exercise testing of low risk patients with known coronary artery disease presenting to the emergency department with chest pain. J Am Coll Cardiol 33:1843–1847, 1999.

30. Tsakonis JS, Shesser R, Rosenthal R, et al: Safety of immediate treadmill testing in selected emergency department patients with chest pain: A preliminary report. Am J Emerg Med 9:557–559, 1991.

31. Goldman L, Weinberg M, Weisberg M, et al: A computer-derived protocol to aid in the diagnosis of emergency room patients with acute chest pain. N Engl J Med 307:588–596, 1982.

32. Goldman L, Cook EF, Brand DA, et al: A computer protocol to predict myocardial infarction in emergency department patients with chest pain. N Engl J Med 318:797–803, 1988.

33. Poretsky L, Leibowitz IH, Friedman SA: The diagnosis of myocardial infarction by computer-derived protocol in a municipal hospital. Angiology 36:165–170, 1985.

34. Hedges JR, Young GP, Henkel GF, et al: Serial ECGs are less accurate than serial CK-MB results for emergency department diagnosis of myocardial infarction. Ann Emerg Med 21:1445–1450, 1992.

35. Gibler WB, Runyon JP, Levy RC, et al: A rapid diagnostic and treatment center for patients with chest pain in the emergency department. Ann Emerg Med 25:1–8, 1995.

36. Justis DL, Hession WT: Accuracy of 22-lead ECG analysis for diagnosis of acute myocardial infarction and coronary artery disease in the emergency department: A comparison with 12-lead ECG. Ann Emerg Med 21:1–9.1992.

37. Zalenski RJ, Cooke D, Rydman R, et al: Assessing the diagnostic value of an ECG containing leads V4R, V8, and V9: the 15-lead ECG. Ann Emerg Med 22:786–793, 1993.

38. Zalenski RJ, Rydman RJ, Sloan EP, et al: Value of posterior and right ventricular leads in comparison to the standard 12-lead electrocardiogram in evaluation of ST-segment elevation in suspected acute myocardial infarction. Am J Cardiol 79:1579–1585, 1997.

39. Spadafore JC, Lieber JG, Vasilenko P: Variance cardiography for emergency department evaluation of chest pain patients. Acad Emerg Med 3:326–332, 1996.

40. Baxt WG: Use of artificial neural network for the diagnosis of myocardial infarction [published erratum appears in Ann Intern Med 116:94, 1992]. Ann Intern Med 115:843–848, 1991.

41. Baxt WG, Skora J: Prospective validation of artificial neural network trained to identify acute myocardial infarction. Lancet 347:12–15, 1996.

42. Kontos MC, Anderson FP, Hanbury CM, et al: Use of the combination of myoglobin and CK-MB mass for the rapid diagnosis of acute myocardial infarction. Am J Emerg Med 15:14–19, 1997.

43. Kontos MC, Anderson FP, Schmidt KA, et al: Early diagnosis of acute myocardial infarction in patients without ST-segment elevation. Am J Cardiol 83:155–158, 1999.

44. Gillum RF, Fortmann SP, Prineas RJ, Kottke TE: International diagnostic criteria for acute myocardial infarction and acute stroke. Am Heart J 108:150–158, 1984.

45. Pope JH, Aufderhide TP, Ruthazer R, et al: A multicenter study of missed diagnoses of acute myocardial infarction and unstable angina in the emergency department. N Engl J Med 342:1163–1170, 2000.

46. National Cooperative Study Group: Unstable angina pectoris: National Cooperative Study Group to Compare Medical and Surgical Therapy. IV. Results in patients with left descending coronary artery disease. Am J Cardiol 48:517–524, 1981.

47. Mulcahy R, Awadhi AH, de Buitleor M, et al: Natural history and prognosis of unstable angina. Am Heart J 109:753–758, 1985.

48. Stuart RJ Jr, Ellestad MH: National survey of exercise stress testing facilities. Chest 77:94–97, 1980.

49. Physicians’ Fee Reference, ed 16. Wisconsin: Yale Wasserman Medical Publishers; 1999.

50. DRG Guidebook. Reston, VA: David Schultz; 1999.

51. Patterson RE, Eisner RL, Horowitz SF: Comparison of cost-effectiveness and utility of exercise ECG, single photon emission computed tomography, positron emission tomography, and coronary angiography for diagnosis of coronary artery disease. Circulation 91:54–65, 1995.

52. Kuntz KM, Fleischmann KE, Hunink MG, Douglas PS: Cost-effectiveness of diagnostic strategies for patients with chest pain. Ann Intern Med 130:709–718, 1999.

53. Personal communication, Paul Elko, GE Marquette Corporation.

 

Table 1. Values for Test Diagnostic Performance*

 

                Acute Myocardial Infarction    Unstable Angina      Non-Acute Cardiac Ischemia

 

                Base Value                             Base Value                             Base Value

Test         Sensitivity                Source    Sensitivity                Source    Specificity              Source

 

Serum Biomarkers

CK-MB single           All: 0.41   Meta-analysis3        0.05         Estimated7–11         0.95         Estimated8

                Low-risk: 0.41

 

CK-MB serial           All: 0.55   Meta-analysis3        0.07         Estimated from Hedges8         0.95         Hedges8

                Low-risk: 0.8

 

Myoglobin single      All: 0.5     Meta-analysis3        0.05         Kennedy14              0.95         Estimated from Levitt,15

                Low-risk: 0.4                                                                           Kennedy,14 Laurino11

               

Myoglobin serial      All: 0.82   Meta-analysis3        0.20         Estimated from Brogan7          0.95         Estimated from Levitt,15

                Low-risk: 0.86                                         and Kennedy14                      Kennedy,14 Laurino11

 

Troponin T single     All: 0.4     Meta-analysis3        0.20         Mohler,16 Hamm10  0.98         Mohler,16 Green17

                Low-risk: 0.5

 

Troponin T serial     All: 0.9     Meta-analysis3        0.30         Mohler,16 Hamm10  0.98         Estimated from Green17

                 Low-risk: 0.9          Mohler16

 

 

Imaging Studies

Rest         All: 0.93   Sabia19; Peels20    0.35         Sasaki,22 Mohler16 0.85         Estimated from

echocardiography   Low-risk: 0.95         and Kontos21                                                          meta-analysis4 and Kontos21

                                estimated from                       

 

Sestamibi Low-risk: 0.96         Estimated from        0.8           Estimated from Hilton,27          0.77         Kontos,26 Stewart,25

imaging                    Stewart25 and Kontos21,26                   Kontos,21,26 and Stewart25                  and Hilton27

 

 

Table 2. Transition Probabilities

 

Variable                                                                 Base Value                                     Source

Prevalence of                                                                                               
ACI                                                                All ED patients: 18%;                Selker,5 Pope45
                                                                     low-risk patients: 13%;
                                                                     (range for sensitivity
                                                                     analysis: 1%–90%)

                                                                                                                      

AMI                                                               All: 8.4%; low-risk: 4%             Selker,5 Pope6

 

UAP                                                              Both models: 9%                      Pope45 and analysis of

                                                                                                                       ACI-TIPI data from Selker5

 

Subsequent hospitalization rate
within 30 days for

Missed AMI                                                   72%                                          Pope45

Missed UAP                                                  50%                                          Pope45

 

30-day survival rate for
hospitalized patients with

AMI                                                               90%                                          Pope45

UAP                                                              98%                                          Pope45

 

30-day survival rate for                              

patients discharged from the ED

with AMI                                                       89%                                          Pope45

UAP                                                              95%                                          Pope45

 

Percentage of patients with                         All ED patients: 9%;                  Estimate based on

untreated UAP who develop                        low-risk subgroup: 5%             expert opinion and

AMI within 30 days                                                                                        National Cooperative Study

                                                                                                                       Group46 and Mulcahy47

 

Subsequent hospitalization rate                   9%                                            Analysis of ACI-TIPI data

for patient without ACI discharged

from the ED

 

Value of appropriate triage                          1                                               Decision analysis

(hospitalization) for patients with ACI

 

Value of inappropriate triage                        0                                               Decision analysis

(discharge) for patients with ACI

 

Value of triage for patients with                  0                                               Decision analysis

non-ACI

 

ACI = acute cardiac ischemia; ACI-TIPI = acute cardiac ischemia time-insensitive predictive instrument; AMI = acute myocardial infarction; ED = emergency department; UAP = unstable angina pectoris.

 

 

Table 3. Possible Patient Dispositions Within 30 Days of Presentation to the ED

 

Condition of Patient        Possible Disposition

Patients with AMI          

Hospitalized                     Survives; dies

Discharged                      Survives, returns for hospitalization; survives, outpatient
evaluation; dies

 

Patients with UAP

Hospitalized                     Survives; dies

Discharged                      Survives, develops AMI, returns for hospitalization; survives, UAP continues,
returns for hospitalization; survives, UAP continues, outpatient evaluation; dies

 

Patients with non-ACI

Hospitalized                     Survives

Discharged                      Survives, returns for hospitalization;
survives, outpatient evaluation

 

AMI = acute myocardial infarction; Non-ACI = non acute cardiac ischemia; UAP = unstable angina.

 

 

Table 4. Factors Involved in Calculating Total Costs (Reimbursements)

 

Patient Disposition                                                        Factors Involved in Total Cost Calculation

 

Patient with ACI appropriately triaged (admitted)            Costs of test + hospitalization*

Patient with ACI inappropriately discharged, dies          Costs of: test +  ED visit†  +  “missed ACI”‡

Patient with ACI inappropriately discharged, survives   Cost of: test + ED visit + subsequent
hospitalization or outpatient evaluation

Patient with UAP inappropriately discharged,                 Costs of:  test + ED visit + subsequent
develops AMI, survives                                                  hospitalization for AMI*

Non-ACI patient admitted for ACI treatment                    Costs of:  test + “23-hr observation” admission§

Non-ACI patient discharged from ED                              Costs of: test + ED visit + subsequent
hospitalization or outpatient evaluation

 

*The cost for hospitalization of a patient with UAP is the same as for a patient with AMI to prevent penalizing appropriate admission for AMI and to provide for the possibility of cardiac angiography for a patient with UAP. This assumption is also based on the similar hospital reimbursements for hospitalization for AMI without arterial bypass surgery and for UAP with cardiac angiography.

†The reimbursement rate for an ED visit for a patient with suspected ACI was obtained from average reimbursements for such ED visits, which do not take into account intensiveness of treatment given in the ED.

‡Costs associated with a death from misdiagnosis of ACI ranged from a minimum of a return ED visit in the base case ($600) to a maximum of a malpractice settlement or cost of a trial in sensitivity analyses ($2 million).

§Cost of “observation” status for a patient on “rule-out MI” protocol. Does not include costs of treatment for non-ACI conditions.

ACI = acute cardiac ischemia; AMI = acute myocardial infarction; ECG = electrocardiogram; ED = Emergency department; UAP = unstable angina pectoris.

 

 

Table 5. Reimbursement Costs

 

Tests & Combinations                                                Factors Involved in

of Tests                                                                        Calculating Costs                         Costs ($)*

 

Admission for ACI¶                                           Average of an average admission           4,400

                                                                          for an uncomplicated AMI ($4,627)

                                                                          and admission for UAP with

 

Initial and return ED visits**                                Includes all services provided by              600

                                                                          the ED, including resuscitation

 

 

 

Outpatient visit, no diagnostic testing                Professional fee only                                 130
(low ACI likelihood follow-up)

 

 

 

Table 5. Reimbursement Costs, continued

 

Tests & Combinations                                                Factors Involved in

of Tests                                                                        Calculating Costs                         Costs ($)*

 

Cost of death from missed diagnosis of ACI     In sensitivity analysis ranges             Base case: 0;
(inappropriate discharge)                                  from an ED visit for attempt at            sensitivity analysis:

                                                                          resuscitation to settlement of             600–2,000,000

                                                                          malpractice suit

 

*Median reimbursements for tests as reported in the Physicians’ Fee Reference 199949 and average 1999 Medicare reimbursements based on DRG codes.50 In US$ 1999.

†Physicians’ Fee Reference data not available; cost for test at New England Medical Center, Boston.

‡Cost of ACI-TIPI varied in sensitivity analyses.

§Does not include costs of treatment for non-ACI conditions.

¶The cost for hospitalization of a patient with UAP is the same as for a patient with AMI to prevent penalizing appropriate admission for AMI and to provide for the possibility of cardiac angiography for a patient with UAP. This assumption is also based on the similar hospital reimbursements for hospitalization for uncomplicated AMI ($4,627) and that for UAP with cardiac angiography ($4,155) admission [50].

**The reimbursement rate for an ED visit for a patient with suspected ACI was obtained from actual ED patient visits to a Boston tertiary care hospital between October, 1998 and May 1999 and was based on average DRG reimbursements for ED visits for a patients with suspected ACI.

ACI-TIPI = Acute Cardiac Ischemia Time-Insensitive Predictive Instrument; CK-MB = creatine kinase MB; ECG = electrocardiogram; ECHO = echocardiogram. ACI = acute cardiac ischemia; AMI = acute myocardial infarction; ED = Emergency department; UAP = unstable angina pectoris.

 

 

Table 6. All ED patients: Cost per Patient, Number of Patients With ACI Appropriately Triaged, and Incremental Cost-effectiveness Ratios of Diagnostic Tests

 

                                                                                                              Number of

                                                                                   Number of            additional                                                                                               patients with                                    patients

                                                                                        ACI                 with ACI             Incremental

Test*                          Cost per       Incremental      appropriately      appropriately               Cost-

                                ED patient†          Cost‡               triaged§               triaged¶           effectiveness**

 

Myoglobin, single         $1,677                                          46                                                      

 

CK-MB, single              $1,684                $7                      39                                              Dominated

 

Troponin T, single        $1,691               $14                     53                        7              Weakly dominated

 

CK-MB, serial               $1,760               $70                     74                       21             Weakly dominated

 

Myoglobin, serial          $1,780               $20                     94                       21             Weakly dominated

 

Troponin T, serial         $1,796               $16                    107                      13                      $1,944

 

CK-MB &                      $1,817               $21                     80                                              Dominated

Myoglobin single

 

Goldman                      $1,829               $33                     76                                              Dominated

 

ECHO rest                    $2,175              $379                   112                       5              Weakly dominated

 

Troponin T & ECHO      $2,202               $27                    122                      10             Weakly dominated

 

ACI-TIPI                        $2,269               $67                    168                      46                      $7,860

 

*Ordered by increasing cost.

†Costs are total costs (in 1999 $) of applying test to an ED patient with symptoms suggestive of ACI.         

‡Difference in cost between test and previous test. Calculations directly from table may be off due to rounding.

§Among 1,000 ED patients in whom prevalence of ACI is 18%. A test with 100% sensitivity for ACI would lead to appropriate triage for 180 patients with ACI.

¶Compared with previous non-dominated test.

**Incremental CE ratios were calculated by dividing the difference in costs by the difference in effectiveness between a test and the previous most effective and less costly non-dominated test. Tests that were less effective and more costly than another test were eliminated by “simple” dominance. Tests that were less effective and had a higher incremental CE ratio were eliminated by “weak” dominance. Calculations directly from table may be off due to rounding.

ACI = acute cardiac ischemia; ACI-TIPI = Acute Cardiac Ischemia Time-Insensitive Predictive Instrument; CK-MB = creatine kinase MB; ECG = electrocardiogram; ECHO = echocardiogram; ED = Emergency department. 

 

               

 

 

 

Table 7. Low-Risk Patients: Cost per Patient, Number of Patients With ACI Appropriately Triaged, and Incremental Cost-effectiveness Ratios of Diagnostic Tests

 

                                                                                                              Number of

                                                                                   Number of            additional                                                                                               patients with                                    patients

                                                                                        ACI                 with ACI             Incremental

Test*                          Cost per       Incremental      appropriately      appropriately               Cost-

                                ED patient†          Cost‡               triaged§               triaged¶           effectiveness**

 

Myoglobin, single         $1,535                —                      22                        —                          —

 

CK-MB, single              $1,548               $13                     19                        —                   Dominated

 

Troponin T, single        $1,554               $19                     34                       12                      $1,546

 

CK-MB, serial               $1,608               $54                     35                        1              Weakly dominated

 

Myoglobin, serial          $1,622               $14                     50                       15             Weakly dominated

 

Troponin T, serial         $1,637               $15                     61                       11             Weakly dominated

 

CK-MB &                      $1,657               $20                     30                        —                   Dominated
Myoglobin, single

 

Goldman protocol        $1,685               $48                     36                        —                   Dominated

 

Exercise ECG              $1,764              $127                   112                      50                      $2,705

 

CK-MB &                       $1,799              $35                     67                        —                   Dominated
Myoglobin, serial

 

Continuous/                 $1,820               $57                     34                        —                   Dominated
serial ECG

 

CK-MB                         $1,964              $200                    50                        —                   Dominated
& serial ECG

 

Troponin T & ECHO      $2,047              $283                    76                       —-                  Dominated

 

Echocardiography       $2,050              $286                    69                        —                   Dominated

 

Sestamibi imaging        $2,420              $656                   110                       —                   Dominated

*Ordered by increasing cost.

†Costs are total costs (in 1999 $) of applying test to an ED patient with symptoms suggestive of ACI.

‡Difference in cost between test and previous test. Calculations directly from table may be off due to rounding.

§Among 1,000 low-risk ED patients in whom prevalence of ACI is 13%, a test with 100% sensitivity for ACI would lead to appropriate triage for 130 patients with ACI.

¶Compared with previous non-dominated test.

*Incremental CE ratios were calculated by dividing the difference in costs by the difference in effectiveness between a test and the previous most effective and less costly non-dominated test. Tests that were less effective and more costly than another test were eliminated by “simple” dominance. Tests that were less effective and had a higher incremental CE ratio were eliminated by “weak” dominance. Calculations directly from table may be off due to rounding.

ACI = acute cardiac ischemia; CK-MB = creatine kinase MB; ECG = electrocardiogram; ECHO = echocardiogram;
ED = Emergency department.

               

 

 

Figure 1. Decision Tree. The decision model represents the possible dispositions and outcomes that occur for each patient presenting to the emergency department (ED) with signs and symptoms suggestive of acute cardiac ischemia (ACI). Shown is a decision node branch of the tree for one diagnostic test. The branches emanating from a chance node, depicted by a circle, represent the possible outcomes that could occur. The probabilities of AMI, UAP, and non-ACI are determined by the prevalence rates of these conditions in the population of patients presenting to the ED. The probabilities of the possible outcomes are determined by the probabilities of all choices that have occurred along the path to the outcome. The proportion of patients at each of the terminal nodes is determined by the probabilities of all choices that have occurred along the way to each node. OPFU = Outpatient follow-up evaluation; ACI = acute cardiac ischemia; ED = Emergency department.

 

Figure 2. Triage accuracy of tests for patients with acute cardiac ischemia (ACI) among all emergency department (ED) patients with symptoms suggestive for ACI. ECHO = echocardiograph; Goldman = Goldman chest pain protocol.

 

Figure 3. Diagnostic test costs per ED patient and percentage of patients appropriately triaged for acute cardiac ischemia (ACI) among all emergency department (ED) patients with symptoms suggestive of ACI. ACI-TIPI = Acute Cardiac Ischemia Time Insensitive Predictive Instrument; ECG = electrocardiograph; ECHO = echocardiograph; Goldman = Goldman chest pain protocol.

 

Figure 4. Triage accuracy of tests for patients with acute cardiac ischemia (ACI) among low-risk emergency department (ED) patients with symptom suggestive for ACI. ACI-TIPI = Acute Cardiac Ischemia Time Insensitive Predictive Instrument; ECG = electrocardiograph; ECHO = echocardiograph; ETT = exercise ECG test; Goldman = Goldman chest pain protocol.

 

Figure 5. Diagnostic test costs per emergency department (ED) patient and percentage of patients appropriately triaged for acute cardiac ischemia (ACI) among low-risk ED patients with symptoms suggestive for ACI. ECG = electrocardiograph; ECHO = echocardiograph; ETT = exercise ECG test; Goldman = Goldman chest pain protocol.