Economic Outcome of Cardiac CT-Based Evaluation and Standard of Care for Suspected Acute Coronary Syndrome In the Emergency Department: A Decision Analytic Model

Abstract

Rationale and Objectives

Cardiac computed tomography (CCT) in the Emergency Department may be cost saving for suspected acute coronary syndrome (ACS), but economic outcome data are limited. The objective of this study was to compare the cost of CCT-based evaluation versus standard of care (SOC) using the results of a clinical trial.

Materials and Methods

We developed a decision analytic cost-minimization model to compare CCT-based and SOC evaluation costs to obtain a correct diagnosis. Model inputs, including Medicare-adjusted patient costs, were primarily obtained from a cohort study of 102 patients at low to intermediate risk for ACS who underwent an Emergency Department SOC clinical evaluation and a 64 channel CCT. SOC costs included stress testing in 77% of patients. Data from published literature completed the model inputs and expanded data ranges for sensitivity analyses.

Results

Modeled mean patient costs for CCT-based evaluation were $750 (24%) lower than the SOC ($2,384 and $3,134, respectively). Sensitivity analyses indicated that CCT was less expensive over a wide range of estimates and was only more expensive with a CCT specificity below 67%. Probabilistic sensitivity analysis suggested that CCT-based evaluation had a 98.9% probability of being less expensive compared to SOC.

Conclusion

Using a decision analytic model, CCT-based evaluation resulted in overall lower cost than the SOC for possible ACS patients over a wide range of cost and outcome assumptions, including CT-related complications and downstream costs.

Background

In the United States, chest pain accounts for 5% of all Emergency Department (ED) visits and is a substantial economic and health resource burden¹. In 2008, the US spent approximately $12 billion dollars for the evaluation of acute chest pain¹. The current standard of care (SOC) for individuals presenting with possible acute coronary syndrome (ACS) often involves multiple tests over hours to days. Rest and stress SPECT is commonly used to test for myocardial ischemia with high sensitivity, but only after exclusion of myocardial damage. Multislice ECG-gated computed tomography can rapidly differentiate multiple causes of symptomatic vascular disease^{2, 3} including coronary artery disease (CAD)⁴ with high diagnostic sensitivity. As a result, use of ECG-gated thoracic CT as part of a rapid evaluation in patients at low to intermediate risk of ACS may be less expensive than the SOC^{5, 6}.

Previous economic model evaluations of cardiac CT (CCT) for evaluation of ACS suggested that CCT is less expensive than an SOC^6–8, but may be more costly with the inclusion of downstream costs associated with CT, such as iodinated contrast complications or incidental, non-cardiac CT findings⁹. However, no study to date has incorporated both actual ED patient costs and downstream costs associated with CCT.

We hypothesized that a CCT-based evaluation of ED patients at low to intermediate risk of ACS will be less expensive than SOC even when downstream costs of the CCT are considered. To test this hypothesis, we compared costs of CCT-based evaluation versus SOC using the results of a clinical trial in which all patients received a research CCT in addition to SOC and were followed up for 12 months. We developed a decision analytic model to compare the cost to obtain a correct diagnosis between CCT and SOC evaluations.

Methods

Study Design

We constructed a short-term (12 month), cost-minimization decision analytic model from the payer perspective. Model inputs were based primarily on a prospective clinical trial and the literature. All patients gave informed consent for clinical trial enrollment. The primary endpoint of this analysis was the cost to obtain a correct ACS diagnosis for CCT-based evaluation and SOC. We compared all costs for each strategy until the correct diagnosis was reached. We included downstream costs associated with the CCT, including CCT-related complications or work up of incidental, non-cardiac CCT findings.

We limited our model to a 12 month time horizon because we anticipated that longer term clinical outcome overall would be similar for CCT and SOC evaluations^{6, 7}. In our clinical study, there was one sudden death from an unknown cause in one patient without ACS. However, little data exist for longer term costs of treatments based on CCT findings, such as costs from non-obstructive coronary artery disease. We suspected that longer time horizons and other outcomes, including quality adjusted life year calculations, could be subject to more uncertainty than this short term evaluation.

Patient Population and Emergency Department Evaluation

Patients presenting to a single academic medical center ED with possible ACS from July 2006 to May 2009 were enrolled. Male patients ≥30 years or female patients ≥45 years at low to intermediate clinical risk of ACS with at least one cardiac risk factor were considered for enrollment. Exclusion criteria included known CAD or a serum creatinine ≥1.8 g/dL. Detailed inclusion and exclusion criteria are described in an online supplement. Patient characteristics are presented in Table 1

Table 1.

Patient Characteristics (N=102)

Characteristic	Mean (95% CI) or N (%)
Age (years)	54 (51, 57)
Male (%)	60 (59%)
Caucasian (%)	78 (76%)
Weight (kg)	86.5 (83, 90)
BMI	28.0 (27, 29)
TIMI ACS risk score	0.9 (0.7, 1.1)
Risk Factors (%)
Hypertension	43 (42%)
Dyslipidemia	39 (38%)
Diabetes	9 (9%)
Family history of premature CAD	39 (38%)
Recent tobacco	18 (18%)
Obesity	40 (39%)
Sedentary lifestyle	45 (44%)
Presenting Symptoms*
Chest Pain	94 (92%)
Syncope	4 (4%)
Palpitations	2 (2%)
Shortness of Breath	1(1%)
Lightheadedness	1(1%)
Back pain	1 (1%)

^*Patients may have more than one symptom at presentation.

ACS = acute coronary syndrome, CAD = coronary artery disease, CI = confidence intervals, BMI = body mass index, N = number, TIMI – Thrombolysis in Myocardial Infarction.

This prospective cohort study compared a research CCT scan to an SOC evaluation that recommended stress testing with imaging. After informed consent, patient ED evaluation and treatment were guided by a SOC algorithm that included chest radiograph, serial ECG’s, serial troponin I. Stress myocardial perfusion imaging (MPI) using ^99mtechnetium tetrofosmin or ²⁰¹Thallium was performed in most patients and stress echocardiography was performed in a minority. Myocardial ischemia was diagnosed by stress imaging using established criteria¹⁰. Additional tests and treatments were performed at the discretion of the ED physicians, such as additional laboratory tests, or additional imaging studies. The billed tests for the SOC evaluation were tabulated for each patient and used to populate the patient data in Table 2.

Table 2.

Model Parameters: Clinical Probabilities and Costs for the Decision Tree

Variable Name	Baseline Value	Range	Distribution	Data Origin or References
ACS Prevalence	0.07	0–0.2	Normal	Patient data
CCT Sensitivity	1	0.85 1	Normal	Patient data,19,20,
CCT Specificity	0.88	0.65 0.9	Normal	Patient data,19,20,
Stress test sensitivity (weighted mean)	0.85	0.77–0.94
- Nuclear SPECT	0.86	0.81–0.96	Normal	12
- Echocardiogram	0.80	0.72–0.88		11
Stress test specificity (weighted mean)	0.76	0.68–0.83
- Nuclear SPECT	0.74	0.67–0.91	Normal	12
- Echocardiogram	0.86	0.77–0.95		11
Hospital admission and cardiac catheterization	0.07	0–0.2	Normal	Patient Data
Probabilities
SOC arm cardiac stress testing	0.77
- Nuclear SPECT	0.67	-	-	Patient data
- Echocardiography	0.10	-
MI or unstable in ED (SOC arm only)
- Patients with ACS	0.05	0–0.3	Beta	Patient data
- Patients without ACS	0.02	0–0.3
No SOC cardiac stress testing or catheterization	0.18	- -	-	Patient data
Incidental, non-cardiac findings on SOC imaging requiring further evaluation	0	0.0008–0.01	Beta	21
MI or death before diagnosis
- CCT Arm	0	0–0.5	Beta	Patient Data,22
- SOC arm	0.01	0–0.5
Cardiac re-evaluation for recurrent symptoms	0.07	0–0.5	Beta	Patient data
CCT contrast complication	0	0.001–0.2	Beta	Patient data,23
CCT non-cardiac finding
- CT non-cardiac findings requiring repeat CT	0.04	0.04–0.09	Beta	Patient data,13
- Positive repeat CT requiring PET	0.01	0–0.05
Costs ($USD)
Base case cost for SOC arm	$1,734	$1,672– $1,877	Normal	Patient data
Base case cost for CCT arm	$1,347	$1,077–$1,616	Normal	Patient data
Hospital admission and cardiac catheterization	$5,524	$4,420, $6,629	Normal	Patient data
Cardiac catheterization	$2,773	$2,218–$3,327	Normal	Patient data
Iodinated contrast complication	$897	$0–$76,317	Log normal	17
Incidental, non-cardiac CCT findings requiring repeat CT	$448	$358–$538	Normal	Patient data
- Positive repeat CT requiring further evaluation	$1,255	$1,004–$2,644
Cardiac re-evaluation for recurrent symptoms	$3,029	$2,474–$6,629	Normal	Patient data
MI or death before ACS diagnosis	$37,147	$7,776–$117,490	Normal	6, 24

Costs were based on or adjusted to 2007 Medicare-based reimbursement amounts except for cardiac imaging costs which were based on 2010 Medicare reimbursements. Baseline costs were derived from patient data when available. The patient and literature data provided the range of values used for sensitivity and probabilistic Monte Carlo analyses. All costs were in US dollars ($).

ACS = acute coronary syndrome; CT = computed tomography; CCT = cardiac computed tomography; ED = Emergency Department; MI = myocardial infarction, SOC = standard of care, SPECT = single photon emission computed tomography.

Each patient underwent an ECG-gated whole chest CT with a 64 slice GE LightSpeed VCT scanner (GE Healthcare, Chalfont St. Giles, UK) and 90–120 mL of iodixanol 320 contrast. Axial and multiplanar CT images were reviewed as a research scan by two experienced readers blinded to clinical data. Coronary artery findings were reported to the ED as negative (<30% stenosis in all coronary segments) or "not negative" (>30% stenosis), with no further information reported. To maximize sensitivity, we assumed patients had ACS by CCT if any coronary artery had a >50% stenosis.

For follow up, patients were contacted at 3, 6 and 12 months after the ED visit to detect clinical events or further testing that would contribute additional costs. The final, adjudicated clinical diagnosis was determined by two cardiologists based on patient data from the initial visit and 3-month follow-up without knowledge of the CCT results. Of the 102 patients enrolled, 7 (7%) were adjudicated to have ACS. All ACS patients were correctly identified by the ED SOC evaluation and by the CCT scan; there were no false negative stress tests or CCT scans for ACS.

Decision Analytic Model

Decision Tree

The decision trees for CCT-based and SOC-based evaluations are outlined in Figure 1. The SOC tree (Figure 1A) began with the ED evaluation, including serial testing during an ED observation period. Patients who became unstable or had documented myocardial infarction during the ED period were admitted to the hospital for cardiac catheterization. Cardiac catheterization is the gold standard for coronary artery disease and ACS diagnosis for the model. All other patients were sent for cardiac stress testing or discharged to home without stress testing following a period of evaluation in the ED. Patients sent for cardiac stress testing who had evidence of myocardial ischemia were admitted to the hospital for cardiac catheterization to determine ACS diagnosis. Patients without evidence of myocardial ischemia on stress testing were discharged home. Next, the rare patient with incidental, non-cardiac imaging findings on stress imaging, such as a thoracic mass, was sent for additional imaging. Patients with recurrent symptoms after the initial ED evaluation had further cardiology evaluation and cardiac catheterization, either as an outpatient or an inpatient. If the stress nuclear test was false negative (patient with ACS but a negative stress study), those patients may have incurred a myocardial infarction or death during follow up. Finally, the adjudicated ACS diagnosis for all patients completed the tree.

Figure 1.

Decision Analytic Trees for SOC-based (Figure A) and CCT-Based (Figure B) Evaluations.

The CCT-based decision tree (Figure 1B) was similar to the SOC tree except that patients had a CCT as the initial step. Patients with a >50% CCT coronary stenosis were presumed to have ACS and were admitted for cardiac catheterization. Patients with less than a 50% CCT stenosis were discharged to home without further testing. The CCT-based tree also included the incremental costs of CCT contrast complications, such as anaphylaxis and renal failure, which were not present in the SOC tree.

Inputs and Cost Parameters

Data used for the model are in Table 2. Probabilities for each node were derived from patient-level trial data or the literature as indicated in Table 2. Costs for the model were derived from trial participants’ current procedural terminology (CPT) or ambulatory payment classifications (APC) billing codes. Patients admitted to the hospital for cardiac catheterization used only a diagnostic related group (DRG) 124 charge. From these codes, model inputs reflect Medicare reimbursement for technical and professional fees. Probabilities and costs for clinical evaluations after the ED stay are summarized in Table 2. The clinical trial began in 2006, so costs are derived from the 2007 Medicare fee schedule which is largely similar to present reimbursement rates. However, we used 2010 costs for cardiac imaging to account for the disproportionate decline in Medicare reimbursement for these studies over time (Table 2). Literature-based costs completed the model when patient data were not available. When patient care after the ED visit occurred outside our institution, we used Medicare payment codes to estimate the most likely patient costs based on the patient’s description of the care provided.

Table 3 illustrates common billed items, billing codes and itemized costs. Further cost details are outlined in the online supplement. The base-case cost for SOC was calculated as the total costs from the ED stay. Of the 102 patients sampled for this study, only 77% had stress testing (Table 2). Stress tests were not performed in 18 (18%) patients at the physician’s discretion or due to patient non-compliance, financial concerns, or leaving against medical advice. The stress test sensitivity and specificity for the SOC tree were weighted by the proportion of patients who had either nuclear or echocardiographic imaging stress tests and the sensitivity or specificity of each test derived from the literature^{11, 12}. For example, mean stress sensitivity = (proportion with nuclear stress) × (nuclear stress sensitivity) + (proportion with stress echocardiogram) × (stress echocardiography sensitivity).

Table 3.

Common Billed Items Used to Determine Base Case Costs

Common ED Billed Items	CPT Codes	N	Mean per Patient*	Patients Billed for at Least One Item (%)	Medicare Adjusted Cost†
Complete blood count	85025, 85027	135	1.2	98%	$9
Comprehensive metabolic panel	80053	60	0.7	58%	$12
Basic metabolic panel	80048	75	0.9	47%	$9
Troponin-I	84484	283	2.4	100%	$14
D-Dimer	85379	53	0.7	53%	$14
Chest Radiograph	71010, 71020	109	1.0	99%	$66
Transthoracic echocardiogram	93307, 93320, 93325	5	0.1	4%	$421
ED level of service 4	99284	9	--	9%	$670
ED level of service 5	99285	93	--	93%	$902
Cardiac Testing‡
Stress nuclear perfusion test	78464, 78465	67	--	67%	$918
Stress echocardiogram	93350	10	--	10%	$476
Invasive cardiac angiography	93545	7	--	7%	$2,773
No stress test	--	18	--	18%	$0
Cardiac CT	0148T	102	1.0	100%	$410

^†2007 and ^‡2010 Medicare-adjusted costs including professional fees.

^*The mean number of items per patient, when present.

Comprehensive metabolic panel = basic metabolic panel plus liver function tests, ECG= electrocardiogram.

To determine the base case cost of the CCT evaluation, the CCT cost plus all other ED costs within six hours of ED admission were summed. The six hours was based on our median time to CCT evaluation of 5.5 hours. This relatively long time to CCT evaluation resulted from time to obtain informed consent and beta blockade, and the CCT availability only from 7am to 6pm. For each “significant” or “indeterminate” non-cardiac finding on imaging¹³, such as a pulmonary nodule, we added the costs of a follow up non-gated contrast chest CT scan and a new patient (Level 4) primary care outpatient clinic visit. If further evaluation was needed after the primary clinic visit, we added the incremental cost of a new patient specialist clinic visit plus a thoracic positron emission tomography (PET) scan. For patients with recurrent ACS symptoms, we added the costs of a new level 4 outpatient cardiology visit and a cardiac catheterization. Outpatient costs unrelated to the ED visit, CCT scan, or stress tests were not included.

Sensitivity and Probabilistic (Monte Carlo) Analysis

To allow for uncertainty for the model inputs, we derived data ranges for each variable from 95% confidence intervals or ± 20% of the mean value based from study data or from the literature. We used the lowest and highest values from either patient data or the literature to define the range for each input (Table 2). Using the ranges of data in Table 2, one-way sensitivity analyses were performed for all variables to test for thresholds where CCT became more expensive. To identify variables that had large cost effects in the model, we also created a Tornado diagram, which graphs the range of cost-minimization values for each variable against the average cost savings from the model. Any variables in the diagram that had visible effects on model costs also underwent threshold sensitivity analyses using extreme values (outside the ranges in Table 2). For any variables with visible cost effects for both probability and costs, such as with contrast complications, we performed two-way sensitivity analyses to determine if there were cost or probability thresholds where CCT became more expensive.

We then determined the effects of other “scenarios” which may affect costs. To determine the cost impact of differing ACS prevalence, we tested models with no (0%), low (7%), and intermediate (10% and 20%) ACS prevalence (Table 2). In addition, we determined the cost effects of exclusively using the less expensive stress echocardiogram or the more expensive nuclear stress test in the 77% of patients who had stress testing. Sensitivity and specificity particular to each type of stress test were weighted as described previously.

We also tested the scenario where a CCT scan was performed on very low risk patients who normally would have limited SOC evaluation and who would not usually undergo cardiac stress testing. We estimated a base case ED cost of $798 from our lowest risk patients and ACS prevalence of 0. By sensitivity analysis, we determined the incremental number of similar very low risk patients needed to make CCT more expensive than SOC.

Finally, we performed a Monte Carlo probabilistic sensitivity analysis of 10,000 simulations using the range of values for the variables and distributions in Table 2.

This HIPAA-compliant study was approved by the Institutional Review Board at our home institution.

Results

Mean estimated cost for obtaining a correct diagnosis using the base case values was $3,127 for the SOC compared to $2,384 for CCT-based evaluation, which is a $750 savings (24%) for CCT-based evaluation (Table 3). The cost savings from CCT evaluation were predominantly due to the difference between reimbursement for a nuclear stress test and a CCT ($508).

Sensitivity, Threshold, and Probabilistic Analyses

Sensitivity analyses performed for all variables using the data ranges in Table 2 suggested that CCT became more expensive only if CCT specificity fell below 67.4%. For the range of parameter values tested, no other single variable had a threshold where CCT became more expensive. Two-way sensitivity analysis that incorporated both the probability and the cost of contrast complications did achieve thresholds where CCT became more expensive (Figure 2). CT was more expensive with either a minimum probability of contrast complications of 0.1% with a per-patient cost of $76,000 or a minimum cost of $3800 with a corresponding probability of over 19%. In contrast, CCT was not more expensive by two-way sensitivity analysis of death or myocardial infarction prior to ACS diagnosis.

Figure 2.

Two-Way Sensitivity Analyses for CCT Contrast Complications. To determine cost thresholds where CCT evaluation is more expensive for correlated variables, probability ranges used in the model are on the x axis and compared to either costs on the Y axis. CCT evaluation is less expensive within the shaded area.

The impact on cost over the range of variables from Table 2 was graphed using a Tornado diagram (Figure 3). Using extreme values for variables with visible impact on costs, sensitivity analyses suggested that CCT became more expensive when 1) ACS prevalence exceeded 84%, 2) stress testing specificity exceeded 99.4%, 3) the cost for non-cardiac findings on CCT exceeded $19,000 per patient, or 4) the base case ED costs were less than $691 for SOC or more than $2263 for CCT.

Figure 3.

Tornado Diagram for Modeled Costs. The range of possible incremental costs for each variable is centered at the mean cost savings for CCT evaluation ($750). Only variables with visible effects on modeled costs were included for this graph. Only CT specificity had values where SOC became less expensive than CCT evaluation. ACS= acute coronary syndrome, CT = computed tomography, SOC = standard of care.

The Monte Carlo probabilistic analysis suggested that CCT evaluation had a 98.9% probability of costing less than SOC evaluation over the range of variables.

Finally, additional scenarios which may alter costs were considered. In the first scenario, increasing the ACS prevalence narrowed the cost differential between SOC and CCT (Table 4). In the second scenario, using the less-expensive stress echocardiography exclusively narrowed the cost difference between CCT and SOC. In the third scenario, CCT-based evaluation became more expensive when the proportion of very low risk patients in the population was greater than 44%.

Table 4.

Cost for Correct Diagnosis for the Standard of Care and Cardiac CT-based Evaluation

	Decision Analytic Model
Protocol	Mean Cost	Mean Savings	% Savings
CCT Evaluation	$2,384	$750	24%
SOC Evaluation	$3,134	--	--
	Probabilistic (Monte Carlo) Sensitivity Analysis
Protocol	Mean Cost (95% CI)	Mean Savings (95% CI)	% Savings (95% CI)
CCT Evaluation	$2,385 ($1,679, $3,807)	$750 ($681, $819)	24% (22%, 27%)
SOC Evaluation	$3,315 ($2,490, $4,462)	--	--

Discussion

In this decision analytic model study, the evaluation of low to intermediate risk ED patients with suspected acute coronary syndrome was $750 (24%) less expensive with CCT evaluation than SOC evaluation, even when downstream costs of a CCT scan were included. Only a CCT specificity of less than 67% made CCT evaluation more expensive than the SOC. These modeled cost savings were largely due to the difference in cost between CCT scan and stress imaging.

Previous studies have suggested that CCT evaluation is less expensive for ED patients. In a prospective randomized study of CCT versus nuclear stress testing in 197 low risk ED patients, Goldstein et al reported a 15% ($289) per patient savings for CCT⁵. Chang et al. presented actual patient costs for 4 different evaluation protocols of low risk ED patients, including SOC and CCT with and without serial biomarkers and observation¹⁴. In that retrospective analysis, CCT evaluation without ED observation was the least expensive protocol whether or not stress testing was used as part of the SOC. However, Takakuwa, et al suggested that imaging costs alone may be more expensive based on the use of stress nuclear testing after CT angiography¹⁵.

Economic modeling has also reported similar cost savings suggesting CCT saved $433 to $580 compared to stress echocardiography or $798 to $990 compared to nuclear stress testing. Otero and Rybicki used a literature-based cost model to report a $433 to $990 savings for CCT evaluation compared to SOC stress testing, using either stress echocardiography or stress nuclear myocardial perfusion⁸. In a microsimulation cost model for a theoretical 55 year old man and woman, Ladapo et al reported mean savings for a CCT evaluation of $380 in women yet a cost increase of $200 in men⁷. Using a cost effectiveness model derived from a clinical database of low risk ED patients, Khare et al reported a $580 to $798 savings for CCT evaluation compared to stress ECG or stress echocardiography and similar long term outcomes⁶. However, all of these studies either fail to include ranges of data (for broader applicability) or they rely on theoretical or retrospective patient data as opposed to actual patient data used in our study. More importantly, only two cost studies have included incremental costs associated with CCT imaging in the model, one in stable angina patients⁹ and another in the ED population¹⁶. Both studies suggested that overall costs for CCT-based evaluation were higher, QALY’s were marginally better.

In contrast to these prior studies, our model incorporated actual data from ED patients who underwent both CCT and SOC evaluation as well as included the downstream incremental costs of CCT. In our study, CCT remained less expensive than SOC evaluation across a wide range of variables and for most (98.9%) probable scenarios using the range of data in Table 2. CCT evaluation did become more expensive at low CCT specificity due to the increase in cardiac catheterization for false positive results. CCT was also more expensive at certain thresholds including those related to the costs of IV contrast-related complications or renal failure (Figure 2)¹⁷. However, achieving these thresholds is unlikely to occur in clinical practice, except in selected high risk populations. Further, SOC and CCT evaluation costs narrowed with increasing ACS prevalence and use of less expensive stress echocardiography (Table 4). While we acknowledge these potential limitations, the cumulative data suggest that CCT evaluation is less expensive compared to the current SOC for the majority of patients at low to intermediate risk for ACS.

Finally, if the use of CCT expands to include very low risk patients, who would otherwise have been discharged without further evaluation nor stress testing, the cost savings for CCT evaluation may decrease. Our analysis suggests that CCT evaluation may become more expensive if greater than 44% of patients in the tested population are very low risk. The impact of expansion of CCT to these patients is unclear, but may impact overall payer costs.

Based on these findings, the most substantial cost savings for CCT evaluation would be in lower risk patients who would otherwise have a SOC cardiac evaluation that included ED observation and stress testing. While the cost savings associated with CCT is reduced with inclusion of populations at higher risk for ACS or downstream expenses from the CT, such as older patients or those with additional cardiac risk factors, our study suggests that CCT is less expensive for the majority of low to intermediate risk patients. In addition, our study likely underestimates other benefits of CCT evaluation in this population because we do not account for the opportunity costs of decreasing the time in the ED. This would almost certainly benefit patients and hospitals and should be investigated with future analyses.

Limitations

There are several important limitations to this analysis. First, the base case CCT cost estimates are extrapolated from billed services incurred within our average six hour time to CT diagnosis for the ED admission. CCT costs might have been different if the patient’s ED course had been based on the CCT findings. However, we chose a six hour window to bias costs against CCT evaluation to account for hospitals without 24 hour access to cardiac CT. Second, we assume that patients with a >50% CCT coronary artery stenosis will undergo further testing. Selecting a lower stenosis threshold for further cardiac testing (e.g., 30% stenosis) would increase CCT false positive rates, decrease specificity, and increase overall CCT cost. Third, CCT detection of “non-obstructive” CAD (<50% stenosis) will likely result in incremental long term payer costs due to additional visits to outpatient physicians, further cardiac testing, and/or medications. The effects on cost and outcomes of these interventions, both positive and negative, are not clear and therefore not included in our study. Fourth, the patient population on which we base our assumptions is relatively small and from a single center. However, we modeled realistic ranges of data from the patient data and literature to incorporate uncertainty and increase generalizability to other populations. Finally, although we anticipate similar long term outcomes for CCT and SOC evaluations⁷, this short-term cost-minimization analysis does not take into account other time horizons nor perspectives. These may include the small lifetime cancer risks from CCT radiation¹⁸, longer term outpatient costs, hospital or broader social costs, and lost productivity.

Conclusions

Modeled CCT-based evaluation of ED patients at low to intermediate risk of ACS was $750 (24%) less expensive at one year than a SOC evaluation with no expected differences in health outcomes, even when downstream costs of CCT complications were considered. CCT evaluation had a 98.6% probability of being less expensive by probabilistic analysis over a wide range of variables. Estimated CCT cost savings appear predominantly due to differences in cost between CCT and stress testing. Additional cost analyses in larger patient populations and from different perspectives may be warranted.

Table 5.

Modeled Costs for SOC and CCT Evaluations by ACS Prevalence and Type of Stress Test

	Mean Cost	Savings ($)	CCT Savings (%)
ACS Prevalence 0%
CT arm	$2,146	--	--
SOC	$2,964	$818	28%
- Stress Nuclear	$3,043	$897	29%
- Stress Echo	$2,413	$267	11%
ACS Prevalence 7%
CT arm	$2,384	--	--
SOC arm	$3,134	$750	24%
- Stress Nuclear	$3,208	$722	23%
- Stress Echo	$2,617	$233	9%
ACS Prevalence 10%
CT arm	$2,486	--	--
SOC arm	$3,207	$721	22%
- Stress Nuclear	$3,278	$792	24%
- Stress Echo	$2,617	$131	5%
ACS Prevalence 20%
CT arm	$2,846	--	--
SOC arm	$3,449	$603	17%
- Stress Nuclear	$3,514	$668	19%
- Stress Echo	$2,996	$150	5%

Stress testing was performed in 77% of patients for all SOC cost calculations.