International Journal of Cardiology 177 (2014) 867–873
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Diagnostic performance and cost of CT angiography versus stress ECG — A randomized prospective study of suspected acute coronary syndrome chest pain in the emergency department (CT-COMPARE)☆,☆☆ Christian Hamilton-Craig a,b,d,⁎, Allison Fifoot a, Mark Hansen a, Matthew Pincus a, Jonathan Chan a,c, Darren L. Walters a,b, Kelley R. Branch d a
Heart and Lung Institute, The Prince Charles Hospital, Brisbane, Australia University of Queensland, Brisbane, Australia c Population and Social Health Research Program, Griffith University, Australia d University of Washington, Seattle, WA, United States b
a r t i c l e
i n f o
Article history: Received 26 August 2014 Received in revised form 7 October 2014 Accepted 18 October 2014 Available online 22 October 2014 Keywords: Coronary CT Exercise ECG Diagnostic performance Acute coronary syndrome
a b s t r a c t Background: Coronary CT angiography (CCTA) has high sensitivity, with 3 recent randomized trials favorably comparing CCTA to standard-of-care. Comparison to exercise stress ECG (ExECG), the most available and least expensive standard-of-care worldwide, has not been systematically tested. Methods: CT-COMPARE was a randomized, single-center trial of low–intermediate risk chest pain subjects undergoing CCTA or ExECG after the first negative troponin. From March 2010 to April 2011, 562 patients randomized to either dual-source CCTA (n = 322) or ExECG (n = 240). Primary endpoints were diagnostic performance for ACS, and hospital cost at 30 days. Secondary endpoints were time-to-discharge, admission rates, and downstream resource utilization. Results: ACS occurred in 24 (4%) patients. ExECG had 213 negative studies and 27 (26%) positive studies for ACS with sensitivity of 83% [95% CI: 36, 99.6%], specificity of 91% [CI: 86, 94%], and ROC AUC of 0.87 [CI: 0.70, 1]. CCTA (N50% stenosis considered positive) had 288 negative studies and 18/35 (51%) positive studies with a sensitivity of 100% [CI: 81.5, 100], specificity of 94% [CI: 91.2, 96.7%], and ROC of 0.97 [CI: 0.92, 1.0; p = 0.2]. Despite CCTA having higher odds of downstream testing (OR 2.0), 30 day per-patient cost was significantly lower for CCTA ($2193 vs $2704, p b 0.001). Length of stay for CCTA was significantly reduced (13.5 h [95% CI: 11.2–15.7], ExECG 19.7 h [95% CI: 17.4–22.1], p b 0.0005), which drove the reduction in cost. No patient had postdischarge cardiovascular events at 30 days. Conclusions: CCTA had improved diagnostic performance compared to ExECG, combined with 35% relative reduction in length-of-stay, and 20% reduction in hospital costs. These data lend further evidence that CCTA is useful as a first line assessment in emergency department chest pain. © 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/cc by-nc-nd/3.0/).
1. Introduction
Abbreviations: ACS, acute coronary syndrome; CCTA, coronary computed tomographic angiography; CPAS, chest pain assessment service; ED, emergency department; ExECG, exercise treadmill electrocardiography; MACE, major adverse cardiovascular events. ☆ Funding: The study was supported by grants from 1) the Queensland Emergency Medicine Research Foundation (#QEMRF-EMSS-2009-022), 2) the Smart Futures Fellowship Early Career Grant (Queensland State Government #ISF783), 3) the Washington–Queensland Trans-Pacific Fellowship fund and 4) Grant Number 5KL2RR025015-02 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research. The grant bodies had no input on study design, data analysis or writing. ☆☆ No conflicts of interest or relationships with industry. ⁎ Corresponding author at: Heart and Lung Institute, The Prince Charles Hospital, Brisbane, Australia.
Chest pain is a common cause for presentations to hospital emergency departments (EDs). The clinical investigation of undifferentiated chest pain must include the expeditious assessment for acute coronary syndrome (ACS). Many chest pain assessment pathways include serial electrocardiography and biomarkers followed by a provocative stress test to rule out myocardial ischemia [1,2]. In many institutions, treadmill exercise stress ECG (ExECG) is used to stratify intermediate risk patients due to the widespread availability and low cost, and is a Class IB indication in the AHA/ACC guidelines [3]. However, ExECG has relatively limited diagnostic performance with low sensitivity and specificity in unselected populations [4]. More recently, coronary computed tomographic angiography (CCTA) has been investigated as a rapid,
http://dx.doi.org/10.1016/j.ijcard.2014.10.090 0167-5273/© 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/cc by-nc-nd/3.0/).
868
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
noninvasive test with a high negative predictive value and reduced length of stay in low to intermediate risk patients with possible ACS [5–7]. These studies have suggested that CCTA-based care is also less expensive compared to provocative stress testing when coupled with imaging. However, cost analyses suggest that CCTA may be more expensive compared to ExECG-based care [8]. To date, there have been no large-scale clinical trials comparing CCTA-based care to ExECG-based care in possible ACS patients. The CT Coronary Angiography Compared to Exercise ECG (CTCOMPARE) study was a prospective randomized trial that compared dual source CCTA with ExECG as part of the standard of care in low– intermediate risk possible ACS patients presenting to the ED. The primary endpoints were the diagnostic performance measures and the hospital-based costs of CCTA-based care as compared to ECGbased care.
was available 24 h a day, and the subsequent testing (whether ExECG or CTCA) was performed as soon as available (described below). All subjects received 300 mg oral aspirin (unless contraindicated) and underwent continuous ST-segment ECG monitoring as part of the standard-of-care. Appendix 1 shows flow diagrams for patient recruitment in the ExECG and CCTA arms. Patients were randomized 24 h a day. 2.3. Exercise ECG procedure Treadmill ExECG was performed as part of an established 24-hour, 7 day per week chest pain assessment service, with testing available during and outside standard hours [1]. Subjects with a second negative 6-hour troponin for myocardial infarction (troponin I b 0.04 mg/dl) underwent the standard Bruce treadmill ExECG protocol (Marquette, GE Healthcare). A cardiology registrar/fellow performed continuous ECG and vital sign monitoring during ExECG testing. A cardiologist independently adjudicated the ExECG result using standard criteria for myocardial ischemia [11]. Subjects without ExECG evidence of myocardial ischemia were discharged. Subjects with positive or equivocal ExECG results were managed at the discretion of the treating cardiologist. 2.4. CCTA procedure
2. Methods 2.1. Study design CT-COMPARE was a randomized, prospective, non-blinded single-center study conducted in a large tertiary academic Australian hospital. Enrolled subjects were randomized to either CCTA or ExECG performed as part of an established chest pain assessment service [1,9]. The local Human Research and Ethics Committee approved the study (HREC/09/ QPCH/89) and all subjects were required to give informed consent prior to randomization. The study was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12614000630617). 2.2. Study subjects Males ≥30 and females ≥40 years of age presenting to the ED with acute undifferentiated chest pain were eligible for inclusion in the trial. Inclusion criteria included intermediate probability of coronary artery disease according to the Cardiac Society of Australia and New Zealand guidelines [10], initial 12-lead ECG without evidence of acute ischemia, TIMI (Thrombolysis in Myocardial Infarction) risk score b4 and a negative first serum sensitive troponin-I with a 99th centile at b0.04 ng/ml (Access 2 immunoassay, BeckmanCoulter) (Fig. 1). Exclusion criteria were previous diagnosis of coronary artery disease; confirmed pregnancy or lactating female; history of severe reactive airway disease or current exacerbation; allergy or contraindication to iodinated contrast or beta-blockade medications; current atrial fibrillation; and renal impairment (eGFR b 50 ml/min using the MDRD equation). To be eligible for randomization, subjects needed to be pain-free and potentially able to exercise on a treadmill. Using a computer-generated random sequence, randomization occurred after the first negative serum troponin result. Randomization
CCTA was performed after the first negative troponin, available daily from 0800–2200 h including weekends [9]. Subjects received 50–200 mg oral metoprolol tartrate, with a goal heart rate of b60 bpm. Further IV metoprolol, in 5 mg aliquots up to a total of 20 mg, was provided in CT if heart rate remained N60 bpm. Sublingual nitroglycerin 300 mg spray was administered 5 min prior to scanning (Somaton Definition 64 detector, or Definition Flash 128-detector; Siemens, Erlangen, Germany). 80 ml of non-ionic isoosmolar contrast (Iomeron 350, Bracco, Italy) was injected via an 18-gauge cubital fossa cannula at 6.0–7.0 ml/min augmented with a 50 ml normal saline flush at 6.0 ml/min. Radiation exposure was minimized by using prospective ECG triggered imaging for patients with a stable heart rate of b60 bpm (“adaptive sequential” mode), or dose-modulated retrospective ECG gating with automated pulse width (“min-dose auto”, Siemens, Erlangen, Germany), for patients with HR N 60 or N10% heart rate variability. Tube voltage was reduced to 100 kVp for subjects weighing b80 kg [12]. Calcium scoring was not performed in these symptomatic subjects. CCTA images were interpreted independently using a dedicated workstation (Leonardo, Siemens, Erlangen, Germany; or Brilliance, Philips, Best, Netherlands) by an expert radiologist and cardiologist (N5 years of CCTA experience each), blinded to subject history and clinical course. Discrepancies were resolved by consensus. Coronary artery lesions were categorized in an ordinal scale as normal (no disease), mild (1–49% diameter stenosis), moderate (50–69% stenosis) or severe stenosis (N70% stenosis), and analyzed on a per-patient and per-vessel basis. Negative (normal CCTA) subjects were discharged without repeat troponin (Appendix 1). Patients with mild CCTA disease had a 6-hour troponin before being discharged with a letter to their family physician. Moderate disease (50–70%) was admitted for a second troponin, and managed at the discretion of the treating cardiologist. Severe disease (N70%) subjects were admitted to the coronary care unit, treated as ACS according to current guidelines, and managed by the treating cardiologist with open access to CCTA data.
Fig. 1. CONSORT schematic diagram of study design and enrollment.
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873 2.5. Study endpoints Primary study outcomes were diagnostic performance for identification of ACS (sensitivity, specificity, positive and negative predictive values, receiver operator curves) and cost from a hospital perspective at 30 days. Secondary endpoints were time-todischarge from the hospital and from the ED, 30-day and 12-month major adverse cardiovascular events (MACE), hospital admission rates, downstream resource utilization, and repeat presentation for recurrent symptoms. 2.6. Cost The costs and length of stay were calculated using time stamps in trial data forms and from hospital systems. Total patient ED utilization costs for the ED and hospital stay, including 30 days after index admission, were tabulated for each patient. The costing methodology included direct costs associated with patient treatment for all inpatient and outpatient care related to the index admission. This encompassed labor cost (time per patient utilization for nursing and medical time), diagnostic imaging, pathology and pharmaceuticals cost, bed day costs (based on the fractional length of stay of the patient and ward location) and non-labor costs (including consumables). Average cost was then calculated for each trial arm and reported in 2012 $AUD values. Societal or opportunity costs were not included in this evaluation. 2.7. Adjudicated clinical diagnosis The cause of subjects' presenting symptoms was determined by adjudicated diagnosis using all available data including 30 day follow-up. Two board-certified cardiologists audited each patient chart and adjudicated the presence of ACS using case report forms based on Cardiac Society of Australia and New Zealand guidelines [13]. Non-ST segment elevation myocardial infarction (NSTEMI) was diagnosed in patients with sensitive troponin-I N 0.04 μg/l, Unstable angina was diagnosed in patients with 1) stress ECG with ≥1 mm ST depression in ≥2 consecutive leads, 2) a nuclear stress test with a summed difference score N3, 3) an echocardiographic stress test with new or worsening wall motion abnormality in at least 1 ventricular segment, or 4) a coronary stenosis N70% on invasive catheterization requiring revascularization or a coronary stenosis of 50–70% with fractional flow reserve b0.8. Differences in adjudication were resolved by consensus. 2.8. Follow-up All enrolled subjects provided consent for nurse follow-up by telephone interview at least 30 days after presentation and at 12 months using a structured questionnaire. Data on hospital presentations for chest pain, additional cardiac testing, diagnosis of ACS, and visits to physicians related to the index hospitalization were captured. 2.9. Statistics The study sample size was calculated based on a false negative rate of ExECG-based care up to 6% [14] and lower limits of CCTA-based care from ROMICAT 1 trial of 2%. Using a one sided two sample proportions test with a delta of 0.04 and an alpha = 0.05, there was 80% power to detect a difference in an estimated sample size of 592 patients. Variables are expressed as mean ±95% confidence interval (CI) or as number and percentages for binary and categorical variables. Analyses for continuous variables were compared using unpaired Student's t-test for parametric data and Mann–Whitney U test for non-parametric data. Binary data were compared by chi-squared testing. ROC area under the curve (AUC) was compared between trial arms using the c-statistic. Odds ratios were tabulated using logistic regression. All outcome data were considered as intentionto-treat analyses. Statistics were calculated using Microsoft Excel (Redmond, WA, USA) or STATA SE software (College Station, PA, USA). 2.10. Funding The study was supported by grants from the Queensland Emergency Medicine Research Foundation (#QEMRF-EMSS-2009-022), the Smart Futures Fellowship Early Career Grant (#ISF783), and the Washington–Queensland Trans-Pacific Fellowship fund. The grant bodies had no input on study design, data analysis or writing.
166 ± 19 cm, p = 0.003), no relevant statistical differences were found (illustrated in table in Appendix 2). Importantly, no ACS events occurred in these patients at 30 days or 12 months. Of the 562 remaining patients, 322 were randomized to the CCTA arm and 240 to the ExECG arm. More patients were excluded in the ExECG arm than the CCTA arm presumably due to loss of consent documentation during patient transport from the ED to the Chest Pain Center. The Human Research Ethics Committee did not approve re-consenting of these patients, thus they were excluded from the analysis. Baseline characteristics of the included cohort after randomization are shown in Table 1, and Appendix 2 shows no differences in the characteristics of the secondarily-excluded patients. 3.1. Primary outcomes 3.1.1. Diagnostic accuracy measures ACS prevalence in the study was 4.2%. In the CCTA arm, 17 (5.2%) had ACS, with 6 (1.8%) having non-ST elevation myocardial infarction and 11 (3.4%) had unstable angina. In the ExECG arm, 7 (2.9%) had clinically-adjudicated ACS with 3 (1.3%) having myocardial infarction and 4 (1.7%) having unstable angina. The diagnostic performance for the ExECG and CCTA arms to predict ACS is shown in Table 2. The ExECG arm had 27 positive studies. Of these, 9 subjects underwent invasive coronary angiography, with 4 true positive ExECG (3 subjects undergoing revascularization, 1 with N50% stenosis), and 4 normal coronary angiograms (false positive ExECG). The remaining 18 subjects with abnormal ExECG had downstream nuclear SPECT imaging, all of which were negative for ischemia (false positive ExECG). This led to a high specificity and negative predictive value, but poor sensitivity and poor positive predictive value (Table 2). The CCTA arm had 287 negative studies (93 with no coronary disease, 195 with mild disease), and 34 positive CCTA studies (N 50% stenosis). Of the 34 positive studies, 18 had at least one N70% stenosis, 16 with 50–69% stenosis, and 1 uninterpretable dataset due to motion artifact coded as “positive” for intention-to-treat analysis. A total of 26 CCTA arm subjects underwent invasive coronary angiography. Of those with severe (N70%) stenosis by CCTA, 16 had concordant N70% stenoses on invasive angiography for which 14 had revascularization (12 PCI, 2 CABG), 2 had side branch vessel disease that was treated medically, and 1 self-discharged prior to revascularization. Of 16 with moderate stenosis (50–69%), 9 had invasive angiography, 8 of which were concordant, and 1 had a false-positive CCTA with estimated 40% stenosis at cath. The 7 remaining positive CTs with moderate stenosis had inpatient stress SPECT scans that were all negative for ischemia. Two subjects with negative CCTA subsequently presented with recurrent chest pain and underwent invasive coronary angiography that confirmed no coronary artery disease (true negative CT). The diagnostic performance measures for CCTA (N50 and N70% stenoses) for ACS are reported in Table 2. Receiver operator characteristic (ROC) analysis showed an areaunder-the-curve (AUC) for the prediction of ACS of 87% for ExECG. Table 1 Baseline patient characteristics.
3. Results From January 2010 to April 2011, 717 low-to-intermediate risk patients with chest pain were enrolled in the study in a 1:1 ratio. Sixty one patients (8%) were subsequently excluded prior to testing due to not meeting strict enrolment criteria (Fig. 1). Eighty four patients (11%) were excluded at completion of the trial in consultation with the Human Research Ethics Committee due to misplacement of informed consent documentation. Data for included and secondarily excluded patients were compared for any systematic bias; other than a slight difference in height for excluded patients (171 ± 10 cm vs
869
Age (years) Male gender Weight (kg) Height (cm) Hypertension Dyslipidemia Diabetes Currently smoking Family history
All patients (n = 563)
CCTA arm (n = 322)
ExECG arm (n = 240)
52.3 ± 10.4 325 (58%) 86.0 ± 21 171 ± 10 173 (31%) 138 (25%) 38 (7%) 132 (23%) 186 (33%)
52.2 ± 10.7 182 (59%) 86.0 ± 19 171 ± 10 99 (31%) 81 (25%) 23 (7%) 77 (24%) 106 (33%)
52.3 ± 9.8 140 (58%) 86.0 ± 20 171 ± 11 74 (31%) 57 (24%) 15 (6%) 55 (23%) 80 (33%)
All data are presented as mean ± S.D. or N (%). CCTA = computed tomography coronary angiogram; ExECG = exercise ECG stress.
870
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
Table 2 Diagnostic performance for CCTA and exercise ECG-based care. Trial arm
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
ROC AUC
ExECG CCTA N 50% CCTA N 70%
83 (36, 100) 100 (82, 100) 94 (73, 100)
91 (86, 94) 94 (91, 97) 99 (98, 100)
19 (6, 38) 51 (34, 69) 90 (67, 99)
100 (97, 100) 100 (99, 100) 100 (98, 100)
0.87 (0.70, 100) 0.97† (0.96, 0.99) 0.97⁎ (0.92, 100)
PPV = positive predictive value, NPV = negative predictive value, ROC AUC = receiver operator characteristic area under the curve. † p = 0.22 compared to ExECG. ⁎ p = 0.26 compared to ExECG.
The ROC AUC improved to 97% for CCTA N 50% stenosis, and 97% for CCTA N 70% stenosis, but did not reach statistical significance (Table 2, Fig. 2). 3.2. Cost analysis Costs from a hospital perspective, including the cost of downstream testing, hospital labor costs and management 30 days after admission, demonstrated a significantly lower per-patient total cost in the CCTA arm of $2193 (95% CI: $1997, $2389) compared to the ExECG arm cost of $2704 [(95% CI: $2555, $2853), p b 0.001]. Moreover, cost was further reduced in those subjects who could be discharged directly from the ED without ward admission [CCTA $1669 (95% CI: $1612, $1726) compared to ExECG $2459 (95% CI: $2397, $2521), p b 0.001]. The reduction in total per-patient cost from the hospital payer perspective was primarily driven by the reduced length of stay. 3.3. Secondary outcomes Length of stay was significantly reduced by 34% in the CCTA arm [13.5 h (95% CI: 11.2–15.7)] compared to the ExECG arm [20.7 h (95% CI: 17.9–23.1), p b 0.0001] (Table 3). Similarly, time to discharge from the ED was also significantly reduced (p b 0.0001). Hospital admission
rates were similar in both groups (CCTA 10.3% versus ExECG 10.8%; odds ratio for admission (OR) of 1.1 (95% CI: 0.6, 1.8); p = 0.8; Table 3). However, CCTA-based care increased rates of downstream cardiac testing compared to ExECG [13.4% versus 7.5%; OR 2.0 (95% CI: 1.1, 3.8); p = 0.02], This increase primarily resulted from a higher invasive coronary angiography rate [9.0% vs 4.2%, odds 2.3 (95% CI: 1.1, 4.7), p = 0.028]. Repeat presentation for recurrent chest pain or symptoms over 12 month follow-up was not statistically different between groups [12.7% vs 10.0%; OR 1.3 (95% CI: 0.8, 2.3); p = 0.30] (Table 3). 3.3.1. Radiation exposure The mean radiation exposure for CCTA was 3.8 mSv (95% CI: 3.5, 4.1 mSv, range: 0.63–16.9 mSv). The mean heart rate during CCTA acquisition was 59 bpm (95% CI: 58.6, 60.4 bpm). 3.4. Patient follow-up At 30 day follow-up, no major coronary events were identified (0% prevalence of ACS) and no deaths occurred. At 12-month follow-up, 4 subjects had events, 3 (0.9%) in the CCTA arm and one (0.4%) in the ExECG arm. One CCTA subject with mild coronary artery disease presented at 6 months with ACS and invasive angiography demonstrated a spontaneous coronary artery dissection of a 3rd marginal branch. There was no visible coronary disease at this location on the study CCTA. Two additional subjects in the CCTA arm died within 12 months; one for an unrelated admission for urosepsis, and one during a planned elective admission for aortic valve replacement complicated by multiorgan failure requiring extra-corporeal membrane oxygenation. The one subsequent death in the ExECG arm resulted from complications of cancer therapy. 4. Discussion For subjects presenting to the ED with acute chest pain, CCTA had better positive predictive value than ExECG, combined with a significantly reduced length of stay and reduced per-patient cost. In an outpatient population, CCTA has emerged as an accurate and rapid tool for the exclusion of coronary artery disease [15–19]. Recent randomized controlled trials in acute ED patients that compared CCTA to nuclear SPECT [7], or to a mixed standard-of-care [5,6] showed
Table 3 Study endpoints.
Fig. 2. Receiver-operator characteristic curves for diagnostic accuracy of ACS for ExECG and (A) CCTA N50% stenosis (B) and CCTA N70% stenosis.
Length of stay (hours, 95% CI) Inpatient admission (%) Downstream testing (%) Invasive angiography Echocardiography Re-presentation (%) Cost ($AUD, 95% CI) All comers Discharged from ED
CCTA (n = 322)
ExECG (n = 240)
p-Value
13.5 (11.2–15.7)
19.7 (17.3–22.0)
0.003
10.2 10.8 7.1 2.2 12.7
10.8 5.8 3.3 1.3 10.5
0.800 0.020 0.028 0.300 0.300
$2193 (1997–2389) $1669 (1612–1726)
$2704 (2555–2853) $2459 (2397–2521)
b0.001 b0.001
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
lower length of stay at lower cost. However, diagnostic performance measures were not included. ROMICAT-II, which used stress imaging as the comparator, is the largest diagnostic performance CCTA study to date and showed a high sensitivity with modest specificity, similar to our data. However, most international guidelines suggest that the least expensive and most widely available stress test, treadmill ECG, should be the first line test for patients at intermediate risk for acute coronary syndrome. The recent CT-STAT study noted that the lack of stress ECG as a comparator was a limitation of that large multicenter trial [20]. CCTA and ExECG-based care has not been directly compared to date. The current CT-COMPARE study represents the largest prospective, randomized trial of CCTA comparing to treadmill exercise ECG as part of the standard-of-care. Our data showed that CCTA-based evaluation is 35% faster and 20% less expensive than ExECG in patients at low– intermediate risk of acute coronary syndrome. These results suggest that CCTA appears to be a useful initial testing strategy for patients at low to intermediate risk for ACS. However, CCTA may unfavorably influence patient care. Diagnosis of acute coronary syndrome was statistically higher in the CCTA group, due to a higher rate of adjudicated unstable angina pectoris. Other randomized trials have also found increased rates of detection of coronary artery disease in the CCTA arm [6,7]. This may be a confounder effect from open, non-blinded access to imaging data as patients with observed coronary artery disease on CCTA may be more likely to be classified as unstable angina. This may also explain the increased downstream testing after CCTA, particularly invasive coronary angiography, in our study and others [6,21]. The decision to perform additional tests including invasive angiography was at the discretion of the treating cardiologist, and observed coronary artery disease may drive these decisions more than stress testing. However, this does reflect a real world approach to patient care. Importantly, the definition of “unstable angina” without biomarker positivity has been questioned, in the era of high-sensitivity troponins [22]; however, in the present study, standard-sensitivity troponins were used as per clinical care in the State of Queensland. Finally, CCTA is not suitable for all-comers; indeed, some have questioned the need for any imaging test in these lower risk patients, advocating accelerated protocols based on biomarkers alone [23]. Further studies into thresholds for invasive angiography and protocol driven decision-making appear warranted. Other data within the trial also deserve attention. The prevalence of coronary disease of 10.8% and ACS 4.2% is consistent for low to intermediate risk populations and suggests wide generalizability of these data. The prevalence is also within an ideal range for which CCTA is predicted to be cost-effective [24,25]. Between groups, the baseline risk appeared slightly higher for the CCTA group, with increased rates of hypertension, smoking and dyslipidemia. This may have increased the prevalence of coronary artery disease in the CCTA arm, but should be accounted for by randomization. Despite this, the overall per-patient cost, inclusive of invasive coronary angiography and interventions, remained 20% lower in the CCTA arm. Patient radiation exposure remains an important concern in diagnostic imaging. The mean effective dose in this study was 3.8 mSv (95% CI: 3.5–4.1 mSv). This exposure is substantially lower than the 11.5 mSv reported in CT-STAT [5], 13.9 mSv in ROMICAT-II [6], and 20.4 mSv in our previous observational study [9]. Mean heart rate during CT acquisition was 59 bpm which enabled effective use of CT dosereduction techniques [26]. In contrast, the ACRIN-PA study reported a substantial proportion (16%) of subjects randomized to CT who were unable to undergo the test due to persistently elevated heart rate [7].
871
primarily due to loss of paper documentation during patient transfers. This resulted in a disproportionate exclusion of patients in the ExECG group. An analysis between these excluded and study patients show no meaningful differences between baseline data (clearly outlined in Appendix 2), but primary outcomes could be affected if their data were included in the final analyses. Second, there is limited applicability of the CCTA strategy to subjects outside of the protocol (males b30 or females b 40 years of age, renal impairment or contraindication to heart rate control). Third, costs for this trial were from a hospital perspective, including labor costs, which is different than the payer perspective from other trials. We believe that this reflects a more realistic cost to the patient and care-givers especially when limited health care resources are considered. Lastly, this study was not powered to compare outcome data, which requires a much larger trial; due to the low event rates, as noted in the CT-STAT paper, the number needed to evaluate for missed MACE would be over 45,000 patients [5]. 6. Conclusions In symptomatic patients at low–intermediate risk for acute coronary syndrome presenting to the emergency department, coronary CT angiography is faster and less expensive, with improved diagnostic performance compared to exercise treadmill ECG-based care. These data add additional proof that protocols using CCTA rather than stress testing-based care in emergency departments should be considered. Further verification of these data in a multicenter randomized trial is warranted. Acknowledgments We thank the staff of the Prince Charles Hospital Medical Imaging Department, Liz Warburton, Steve Graves, Kathryn Arnett, and our patients. Appendix 1. Flow diagrams for CCTA and ExECG arms of the trial.
A
Intermediate risk pain No known CAD
Contraindication to ExECG or CCTA NO
YES
Able to be randomised to ExECG vs CCTA
Excluded from study
Consents to Study
NO
Refused Consent Thus follow standard care pathway
YES
Randomised to ExECG vs CCTA
ExECG Group
CCTA Group
5. Limitations Limitations of this study need to be addressed. First, of the 21% of eligible patients not enrolled, 11% had misplaced informed consent
1A. CAD = coronary artery disease, CPAS = chest pain assessment service, CCTA = coronary CT angiography, and ExECG = exercise treadmill electrocardiography.
872
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
B
Intermediate risk pain No known CAD Successfully recruited and randomised CCTA
ExECG
CCTA
ExECG
Negative serial ECG’s, enzymes enzymes and stress test
Inconclusive stress test result
Low risk Immediate discharge from traditional chest pain assessment service.
Low risk Discharge to hiome
or stress
Intermediate risk
Further testing (e.g. MPS or Cath) Normal Abnormal
Positive ECG, test
Suspect high risk ACS Admitted to ward/CCU
Further testing (e.g. MPS or Cath)
Abnormal
Presumed ACS Ward/CCU Admission. Invasive angiography unless already performed or contraindicated.
Normal
Low risk Consider other diagnosis of noncoronary cardiac damage such as myocarditis.
Presumed ACS Ward/CCU Admission. Invasive angiography unless already performed or contraindicated.
1B. Flow diagram. CAD = coronary artery disease, CPAS = chest pain assessment service, ACS = acute coronary syndrome, CCTA = coronary CT angiography, CCU = coronary care unit, ExECG = exercise treadmill electrocardiography, and MPS = stress nuclear myocardial SPECT.
Appendix 2. Comparison of patients excluded due to lack of consent documentation.
Age (years) Male gender (%) Weight (kg) Height (cm) Hypertension, n (%) Dyslipidemia Diabetes Smoking Family history
Trial patients (n = 563)
No consent (n = 82)
p value
52.3 ± 10.4 325 (58%) 86.0 ± 21 171 ± 10 173 (31%) 138 (25%) 38 (7%) 132 (23%) 186 (33%)
54.2 ± 10.6 38 (46%) 86.0 ± 23 166 ± 19 24 (29%) 22 (27%) 9 (10%) 20 (24%) 30 (37%)
0.13 0.06 0.94 0.003 0.79 0.65 0.35 0.82 0.50
References [1] C.N. Aroney, H.L. Dunlevie, J.H. Bett, Use of an accelerated chest pain assessment protocol in patients at intermediate risk of adverse cardiac events, Med. J. Aust. 178 (2003) 370–374. [2] W.A. Parsonage, L. Cullen, J.F. Younger, The approach to patients with possible cardiac chest pain, Med. J. Aust. 199 (2013) 30–34. [3] R.J. Gibbons, G.J. Balady, J.T. Bricker, et al., ACC/AHA 2002 guideline update for exercise testing: summary article. A report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines (committee to update the 1997 exercise testing guidelines), J. Am. Coll. Cardiol. 40 (2002) 1531–1540. [4] R. Gianrossi, R. Detrano, D. Mulvihill, et al., Exercise-induced ST depression in the diagnosis of coronary artery disease. A meta-analysis, Circulation 80 (1989) 87–98.
[5] J.A. Goldstein, K.M. Chinnaiyan, A. Abidov, et al., The CT-STAT (Coronary Computed Tomographic Angiography for Systematic Triage of Acute Chest Pain Patients to Treatment) trial, J. Am. Coll. Cardiol. 58 (2011) 1414–1422. [6] U. Hoffmann, Q.A. Truong, D.A. Schoenfeld, et al., Coronary CT angiography versus standard evaluation in acute chest pain (ROMICAT-II), N. Engl. J. Med. 367 (2012) 299–308. [7] H.I. Litt, C. Gatsonis, B. Snyder, et al., CT angiography for safe discharge of patients with possible acute coronary syndromes (ACRIN-PA), N. Engl. J. Med. 366 (2012) 1393–1403. [8] K.R. Branch, B.W. Bresnahan, D.L. Veenstra, et al., Economic outcome of cardiac CT-based evaluation and standard of care for suspected acute coronary syndrome in the emergency department: a decision analytic model, Acad. Radiol. 19 (2012) 265–273. [9] M. Hansen, J. Ginns, S. Seneviratne, et al., The value of dual-source 64-slice CT coronary angiography in the assessment of patients presenting to an acute chest pain service, Heart Lung Circ. 19 (2010) 213–218. [10] Cardiac Society of Australia and NZ — guidelines for the management of acute coronary syndromes 2006, Med. J. Aust. 184 (2006) S9–S29. [11] A.W. van't Hof, A. Liem, M.J. de Boer, F. Zijlstra, Clinical value of 12-lead electrocardiogram after successful reperfusion therapy for acute myocardial infarction. Zwolle Myocardial infarction Study Group, Lancet 350 (1997) 615–619. [12] G.Y. Liew, M. Feneley, S.G. Worthley, Noninvasive coronary artery imaging: current clinical applications: Cardiac Society of Australia and New Zealand guidelines, Heart Lung Circ. 20 (2011) 425–437. [13] D.P. Chew, C.N. Aroney, P.E. Aylward, et al., Addendum to the National Heart Foundation of Australia/Cardiac Society of Australia and New Zealand Guidelines for the management of acute coronary syndromes (ACS) 2006, Heart Lung Circ. 20 (2011) 487–502. [14] S. Goodacre, J. Nicholl, S. Dixon, et al., Randomised controlled trial and economic evaluation of a chest pain observation unit compared with routine care, BMJ 328 (2004) 254. [15] A. Arbab-Zadeh, J.M. Miller, C.E. Rochitte, et al., Diagnostic accuracy of computed tomography coronary angiography according to pre-test probability of coronary artery disease and severity of coronary arterial calcification. The CORE-64 (Coronary Artery Evaluation Using 64-Row Multidetector Computed Tomography
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
[16]
[17]
[18]
[19]
[20]
[21]
Angiography) international multicenter study, J. Am. Coll. Cardiol. 59 (2012) 379–387. S. Leschka, P. Stolzmann, L. Desbiolles, et al., Diagnostic accuracy of high-pitch dualsource CT for the assessment of coronary stenoses: first experience, Eur. Radiol. 19 (2009) 2896–2903. A.L. Vavere, A. Arbab-Zadeh, C.E. Rochitte, et al., Coronary artery stenoses: accuracy of 64-detector row CT angiography in segments with mild, moderate, or severe calcification—a subanalysis of the CORE-64 trial, Radiology 261 (2011) 100–108. M.J. Budoff, D. Dowe, J.G. Jollis, et al., Diagnostic performance of 64-multidetector row coronary computed tomographic angiography for evaluation of coronary artery stenosis in individuals without known coronary artery disease: results from the prospective multicenter ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) trial, J. Am. Coll. Cardiol. 52 (2008) 1724–1732. W.B. Meijboom, A.C. Weustink, F. Pugliese, et al., Comparison of diagnostic accuracy of 64-slice computed tomography coronary angiography in women versus men with angina pectoris, Am. J. Cardiol. 100 (2007) 1532–1537. M. Salerno, J.M. Bourque, G.A. Beller, Coronary angiographic evaluation of low-risk chest pain in the emergency department CT-STAT, or maybe not quite that fast? J. Am. Coll. Cardiol. 58 (2011) 1423–1425. E. Hulten, C. Pickett, M.S. Bittencourt, et al., Outcomes after coronary computed tomography angiography in the emergency department: a systematic review and
[22]
[23]
[24]
[25]
[26]
873
meta-analysis of randomized, controlled trials, J. Am. Coll. Cardiol. 61 (2013) 880–892. L.T. Burgers, S.T. Nauta, J.W. Deckers, J.L. Severens, W.K. Redekop, Is it cost-effective to use a test to decide which individuals with an intermediate cardiovascular disease risk would benefit from statin treatment? Int. J. Cardiol. 176 (3) (Oct 20 2014) 980–987. M. Than, L. Cullen, S. Aldous, et al., 2-Hour accelerated diagnostic protocol to assess patients with chest pain symptoms using contemporary troponins as the only biomarker: the ADAPT trial, J. Am. Coll. Cardiol. 59 (2012) 2091–2098. J.M. May, W.P. Shuman, J.N. Strote, et al., Low-risk patients with chest pain in the emergency department: negative 64-MDCT coronary angiography may reduce length of stay and hospital charges, AJR Am. J. Roentgenol. 193 (2009) 150–154. V.L. Priest, P.A. Scuffham, R. Hachamovitch, T.H. Marwick, Cost-effectiveness of coronary computed tomography and cardiac stress imaging in the emergency department: a decision analytic model comparing diagnostic strategies for chest pain in patients at low risk of acute coronary syndromes, JACC Cardiovasc. Imaging 4 (2011) 549–556. C. Hamilton-Craig, R. O’Rourke, A.J. Wesley, J. Ayres, M. Molan, Temporal Trends in Cardiac CT Radiation Exposure and the Impact of the PROTECTION trials in an Australian Academic Center, J. Cardiovasc. Comput. Tomogr. 6 (July/August 2012) 4S.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Diagnostic performance and cost of CT angiography versus stress ECG — A randomized prospective study of suspected acute coronary syndrome chest pain in the emergency department (CT-COMPARE)☆,☆☆ Christian Hamilton-Craig a,b,d,⁎, Allison Fifoot a, Mark Hansen a, Matthew Pincus a, Jonathan Chan a,c, Darren L. Walters a,b, Kelley R. Branch d a
Heart and Lung Institute, The Prince Charles Hospital, Brisbane, Australia University of Queensland, Brisbane, Australia c Population and Social Health Research Program, Griffith University, Australia d University of Washington, Seattle, WA, United States b
a r t i c l e
i n f o
Article history: Received 26 August 2014 Received in revised form 7 October 2014 Accepted 18 October 2014 Available online 22 October 2014 Keywords: Coronary CT Exercise ECG Diagnostic performance Acute coronary syndrome
a b s t r a c t Background: Coronary CT angiography (CCTA) has high sensitivity, with 3 recent randomized trials favorably comparing CCTA to standard-of-care. Comparison to exercise stress ECG (ExECG), the most available and least expensive standard-of-care worldwide, has not been systematically tested. Methods: CT-COMPARE was a randomized, single-center trial of low–intermediate risk chest pain subjects undergoing CCTA or ExECG after the first negative troponin. From March 2010 to April 2011, 562 patients randomized to either dual-source CCTA (n = 322) or ExECG (n = 240). Primary endpoints were diagnostic performance for ACS, and hospital cost at 30 days. Secondary endpoints were time-to-discharge, admission rates, and downstream resource utilization. Results: ACS occurred in 24 (4%) patients. ExECG had 213 negative studies and 27 (26%) positive studies for ACS with sensitivity of 83% [95% CI: 36, 99.6%], specificity of 91% [CI: 86, 94%], and ROC AUC of 0.87 [CI: 0.70, 1]. CCTA (N50% stenosis considered positive) had 288 negative studies and 18/35 (51%) positive studies with a sensitivity of 100% [CI: 81.5, 100], specificity of 94% [CI: 91.2, 96.7%], and ROC of 0.97 [CI: 0.92, 1.0; p = 0.2]. Despite CCTA having higher odds of downstream testing (OR 2.0), 30 day per-patient cost was significantly lower for CCTA ($2193 vs $2704, p b 0.001). Length of stay for CCTA was significantly reduced (13.5 h [95% CI: 11.2–15.7], ExECG 19.7 h [95% CI: 17.4–22.1], p b 0.0005), which drove the reduction in cost. No patient had postdischarge cardiovascular events at 30 days. Conclusions: CCTA had improved diagnostic performance compared to ExECG, combined with 35% relative reduction in length-of-stay, and 20% reduction in hospital costs. These data lend further evidence that CCTA is useful as a first line assessment in emergency department chest pain. © 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/cc by-nc-nd/3.0/).
1. Introduction
Abbreviations: ACS, acute coronary syndrome; CCTA, coronary computed tomographic angiography; CPAS, chest pain assessment service; ED, emergency department; ExECG, exercise treadmill electrocardiography; MACE, major adverse cardiovascular events. ☆ Funding: The study was supported by grants from 1) the Queensland Emergency Medicine Research Foundation (#QEMRF-EMSS-2009-022), 2) the Smart Futures Fellowship Early Career Grant (Queensland State Government #ISF783), 3) the Washington–Queensland Trans-Pacific Fellowship fund and 4) Grant Number 5KL2RR025015-02 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research. The grant bodies had no input on study design, data analysis or writing. ☆☆ No conflicts of interest or relationships with industry. ⁎ Corresponding author at: Heart and Lung Institute, The Prince Charles Hospital, Brisbane, Australia.
Chest pain is a common cause for presentations to hospital emergency departments (EDs). The clinical investigation of undifferentiated chest pain must include the expeditious assessment for acute coronary syndrome (ACS). Many chest pain assessment pathways include serial electrocardiography and biomarkers followed by a provocative stress test to rule out myocardial ischemia [1,2]. In many institutions, treadmill exercise stress ECG (ExECG) is used to stratify intermediate risk patients due to the widespread availability and low cost, and is a Class IB indication in the AHA/ACC guidelines [3]. However, ExECG has relatively limited diagnostic performance with low sensitivity and specificity in unselected populations [4]. More recently, coronary computed tomographic angiography (CCTA) has been investigated as a rapid,
http://dx.doi.org/10.1016/j.ijcard.2014.10.090 0167-5273/© 2014 The Authors. Published by Elsevier Ireland Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/cc by-nc-nd/3.0/).
868
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
noninvasive test with a high negative predictive value and reduced length of stay in low to intermediate risk patients with possible ACS [5–7]. These studies have suggested that CCTA-based care is also less expensive compared to provocative stress testing when coupled with imaging. However, cost analyses suggest that CCTA may be more expensive compared to ExECG-based care [8]. To date, there have been no large-scale clinical trials comparing CCTA-based care to ExECG-based care in possible ACS patients. The CT Coronary Angiography Compared to Exercise ECG (CTCOMPARE) study was a prospective randomized trial that compared dual source CCTA with ExECG as part of the standard of care in low– intermediate risk possible ACS patients presenting to the ED. The primary endpoints were the diagnostic performance measures and the hospital-based costs of CCTA-based care as compared to ECGbased care.
was available 24 h a day, and the subsequent testing (whether ExECG or CTCA) was performed as soon as available (described below). All subjects received 300 mg oral aspirin (unless contraindicated) and underwent continuous ST-segment ECG monitoring as part of the standard-of-care. Appendix 1 shows flow diagrams for patient recruitment in the ExECG and CCTA arms. Patients were randomized 24 h a day. 2.3. Exercise ECG procedure Treadmill ExECG was performed as part of an established 24-hour, 7 day per week chest pain assessment service, with testing available during and outside standard hours [1]. Subjects with a second negative 6-hour troponin for myocardial infarction (troponin I b 0.04 mg/dl) underwent the standard Bruce treadmill ExECG protocol (Marquette, GE Healthcare). A cardiology registrar/fellow performed continuous ECG and vital sign monitoring during ExECG testing. A cardiologist independently adjudicated the ExECG result using standard criteria for myocardial ischemia [11]. Subjects without ExECG evidence of myocardial ischemia were discharged. Subjects with positive or equivocal ExECG results were managed at the discretion of the treating cardiologist. 2.4. CCTA procedure
2. Methods 2.1. Study design CT-COMPARE was a randomized, prospective, non-blinded single-center study conducted in a large tertiary academic Australian hospital. Enrolled subjects were randomized to either CCTA or ExECG performed as part of an established chest pain assessment service [1,9]. The local Human Research and Ethics Committee approved the study (HREC/09/ QPCH/89) and all subjects were required to give informed consent prior to randomization. The study was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12614000630617). 2.2. Study subjects Males ≥30 and females ≥40 years of age presenting to the ED with acute undifferentiated chest pain were eligible for inclusion in the trial. Inclusion criteria included intermediate probability of coronary artery disease according to the Cardiac Society of Australia and New Zealand guidelines [10], initial 12-lead ECG without evidence of acute ischemia, TIMI (Thrombolysis in Myocardial Infarction) risk score b4 and a negative first serum sensitive troponin-I with a 99th centile at b0.04 ng/ml (Access 2 immunoassay, BeckmanCoulter) (Fig. 1). Exclusion criteria were previous diagnosis of coronary artery disease; confirmed pregnancy or lactating female; history of severe reactive airway disease or current exacerbation; allergy or contraindication to iodinated contrast or beta-blockade medications; current atrial fibrillation; and renal impairment (eGFR b 50 ml/min using the MDRD equation). To be eligible for randomization, subjects needed to be pain-free and potentially able to exercise on a treadmill. Using a computer-generated random sequence, randomization occurred after the first negative serum troponin result. Randomization
CCTA was performed after the first negative troponin, available daily from 0800–2200 h including weekends [9]. Subjects received 50–200 mg oral metoprolol tartrate, with a goal heart rate of b60 bpm. Further IV metoprolol, in 5 mg aliquots up to a total of 20 mg, was provided in CT if heart rate remained N60 bpm. Sublingual nitroglycerin 300 mg spray was administered 5 min prior to scanning (Somaton Definition 64 detector, or Definition Flash 128-detector; Siemens, Erlangen, Germany). 80 ml of non-ionic isoosmolar contrast (Iomeron 350, Bracco, Italy) was injected via an 18-gauge cubital fossa cannula at 6.0–7.0 ml/min augmented with a 50 ml normal saline flush at 6.0 ml/min. Radiation exposure was minimized by using prospective ECG triggered imaging for patients with a stable heart rate of b60 bpm (“adaptive sequential” mode), or dose-modulated retrospective ECG gating with automated pulse width (“min-dose auto”, Siemens, Erlangen, Germany), for patients with HR N 60 or N10% heart rate variability. Tube voltage was reduced to 100 kVp for subjects weighing b80 kg [12]. Calcium scoring was not performed in these symptomatic subjects. CCTA images were interpreted independently using a dedicated workstation (Leonardo, Siemens, Erlangen, Germany; or Brilliance, Philips, Best, Netherlands) by an expert radiologist and cardiologist (N5 years of CCTA experience each), blinded to subject history and clinical course. Discrepancies were resolved by consensus. Coronary artery lesions were categorized in an ordinal scale as normal (no disease), mild (1–49% diameter stenosis), moderate (50–69% stenosis) or severe stenosis (N70% stenosis), and analyzed on a per-patient and per-vessel basis. Negative (normal CCTA) subjects were discharged without repeat troponin (Appendix 1). Patients with mild CCTA disease had a 6-hour troponin before being discharged with a letter to their family physician. Moderate disease (50–70%) was admitted for a second troponin, and managed at the discretion of the treating cardiologist. Severe disease (N70%) subjects were admitted to the coronary care unit, treated as ACS according to current guidelines, and managed by the treating cardiologist with open access to CCTA data.
Fig. 1. CONSORT schematic diagram of study design and enrollment.
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873 2.5. Study endpoints Primary study outcomes were diagnostic performance for identification of ACS (sensitivity, specificity, positive and negative predictive values, receiver operator curves) and cost from a hospital perspective at 30 days. Secondary endpoints were time-todischarge from the hospital and from the ED, 30-day and 12-month major adverse cardiovascular events (MACE), hospital admission rates, downstream resource utilization, and repeat presentation for recurrent symptoms. 2.6. Cost The costs and length of stay were calculated using time stamps in trial data forms and from hospital systems. Total patient ED utilization costs for the ED and hospital stay, including 30 days after index admission, were tabulated for each patient. The costing methodology included direct costs associated with patient treatment for all inpatient and outpatient care related to the index admission. This encompassed labor cost (time per patient utilization for nursing and medical time), diagnostic imaging, pathology and pharmaceuticals cost, bed day costs (based on the fractional length of stay of the patient and ward location) and non-labor costs (including consumables). Average cost was then calculated for each trial arm and reported in 2012 $AUD values. Societal or opportunity costs were not included in this evaluation. 2.7. Adjudicated clinical diagnosis The cause of subjects' presenting symptoms was determined by adjudicated diagnosis using all available data including 30 day follow-up. Two board-certified cardiologists audited each patient chart and adjudicated the presence of ACS using case report forms based on Cardiac Society of Australia and New Zealand guidelines [13]. Non-ST segment elevation myocardial infarction (NSTEMI) was diagnosed in patients with sensitive troponin-I N 0.04 μg/l, Unstable angina was diagnosed in patients with 1) stress ECG with ≥1 mm ST depression in ≥2 consecutive leads, 2) a nuclear stress test with a summed difference score N3, 3) an echocardiographic stress test with new or worsening wall motion abnormality in at least 1 ventricular segment, or 4) a coronary stenosis N70% on invasive catheterization requiring revascularization or a coronary stenosis of 50–70% with fractional flow reserve b0.8. Differences in adjudication were resolved by consensus. 2.8. Follow-up All enrolled subjects provided consent for nurse follow-up by telephone interview at least 30 days after presentation and at 12 months using a structured questionnaire. Data on hospital presentations for chest pain, additional cardiac testing, diagnosis of ACS, and visits to physicians related to the index hospitalization were captured. 2.9. Statistics The study sample size was calculated based on a false negative rate of ExECG-based care up to 6% [14] and lower limits of CCTA-based care from ROMICAT 1 trial of 2%. Using a one sided two sample proportions test with a delta of 0.04 and an alpha = 0.05, there was 80% power to detect a difference in an estimated sample size of 592 patients. Variables are expressed as mean ±95% confidence interval (CI) or as number and percentages for binary and categorical variables. Analyses for continuous variables were compared using unpaired Student's t-test for parametric data and Mann–Whitney U test for non-parametric data. Binary data were compared by chi-squared testing. ROC area under the curve (AUC) was compared between trial arms using the c-statistic. Odds ratios were tabulated using logistic regression. All outcome data were considered as intentionto-treat analyses. Statistics were calculated using Microsoft Excel (Redmond, WA, USA) or STATA SE software (College Station, PA, USA). 2.10. Funding The study was supported by grants from the Queensland Emergency Medicine Research Foundation (#QEMRF-EMSS-2009-022), the Smart Futures Fellowship Early Career Grant (#ISF783), and the Washington–Queensland Trans-Pacific Fellowship fund. The grant bodies had no input on study design, data analysis or writing.
166 ± 19 cm, p = 0.003), no relevant statistical differences were found (illustrated in table in Appendix 2). Importantly, no ACS events occurred in these patients at 30 days or 12 months. Of the 562 remaining patients, 322 were randomized to the CCTA arm and 240 to the ExECG arm. More patients were excluded in the ExECG arm than the CCTA arm presumably due to loss of consent documentation during patient transport from the ED to the Chest Pain Center. The Human Research Ethics Committee did not approve re-consenting of these patients, thus they were excluded from the analysis. Baseline characteristics of the included cohort after randomization are shown in Table 1, and Appendix 2 shows no differences in the characteristics of the secondarily-excluded patients. 3.1. Primary outcomes 3.1.1. Diagnostic accuracy measures ACS prevalence in the study was 4.2%. In the CCTA arm, 17 (5.2%) had ACS, with 6 (1.8%) having non-ST elevation myocardial infarction and 11 (3.4%) had unstable angina. In the ExECG arm, 7 (2.9%) had clinically-adjudicated ACS with 3 (1.3%) having myocardial infarction and 4 (1.7%) having unstable angina. The diagnostic performance for the ExECG and CCTA arms to predict ACS is shown in Table 2. The ExECG arm had 27 positive studies. Of these, 9 subjects underwent invasive coronary angiography, with 4 true positive ExECG (3 subjects undergoing revascularization, 1 with N50% stenosis), and 4 normal coronary angiograms (false positive ExECG). The remaining 18 subjects with abnormal ExECG had downstream nuclear SPECT imaging, all of which were negative for ischemia (false positive ExECG). This led to a high specificity and negative predictive value, but poor sensitivity and poor positive predictive value (Table 2). The CCTA arm had 287 negative studies (93 with no coronary disease, 195 with mild disease), and 34 positive CCTA studies (N 50% stenosis). Of the 34 positive studies, 18 had at least one N70% stenosis, 16 with 50–69% stenosis, and 1 uninterpretable dataset due to motion artifact coded as “positive” for intention-to-treat analysis. A total of 26 CCTA arm subjects underwent invasive coronary angiography. Of those with severe (N70%) stenosis by CCTA, 16 had concordant N70% stenoses on invasive angiography for which 14 had revascularization (12 PCI, 2 CABG), 2 had side branch vessel disease that was treated medically, and 1 self-discharged prior to revascularization. Of 16 with moderate stenosis (50–69%), 9 had invasive angiography, 8 of which were concordant, and 1 had a false-positive CCTA with estimated 40% stenosis at cath. The 7 remaining positive CTs with moderate stenosis had inpatient stress SPECT scans that were all negative for ischemia. Two subjects with negative CCTA subsequently presented with recurrent chest pain and underwent invasive coronary angiography that confirmed no coronary artery disease (true negative CT). The diagnostic performance measures for CCTA (N50 and N70% stenoses) for ACS are reported in Table 2. Receiver operator characteristic (ROC) analysis showed an areaunder-the-curve (AUC) for the prediction of ACS of 87% for ExECG. Table 1 Baseline patient characteristics.
3. Results From January 2010 to April 2011, 717 low-to-intermediate risk patients with chest pain were enrolled in the study in a 1:1 ratio. Sixty one patients (8%) were subsequently excluded prior to testing due to not meeting strict enrolment criteria (Fig. 1). Eighty four patients (11%) were excluded at completion of the trial in consultation with the Human Research Ethics Committee due to misplacement of informed consent documentation. Data for included and secondarily excluded patients were compared for any systematic bias; other than a slight difference in height for excluded patients (171 ± 10 cm vs
869
Age (years) Male gender Weight (kg) Height (cm) Hypertension Dyslipidemia Diabetes Currently smoking Family history
All patients (n = 563)
CCTA arm (n = 322)
ExECG arm (n = 240)
52.3 ± 10.4 325 (58%) 86.0 ± 21 171 ± 10 173 (31%) 138 (25%) 38 (7%) 132 (23%) 186 (33%)
52.2 ± 10.7 182 (59%) 86.0 ± 19 171 ± 10 99 (31%) 81 (25%) 23 (7%) 77 (24%) 106 (33%)
52.3 ± 9.8 140 (58%) 86.0 ± 20 171 ± 11 74 (31%) 57 (24%) 15 (6%) 55 (23%) 80 (33%)
All data are presented as mean ± S.D. or N (%). CCTA = computed tomography coronary angiogram; ExECG = exercise ECG stress.
870
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
Table 2 Diagnostic performance for CCTA and exercise ECG-based care. Trial arm
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
ROC AUC
ExECG CCTA N 50% CCTA N 70%
83 (36, 100) 100 (82, 100) 94 (73, 100)
91 (86, 94) 94 (91, 97) 99 (98, 100)
19 (6, 38) 51 (34, 69) 90 (67, 99)
100 (97, 100) 100 (99, 100) 100 (98, 100)
0.87 (0.70, 100) 0.97† (0.96, 0.99) 0.97⁎ (0.92, 100)
PPV = positive predictive value, NPV = negative predictive value, ROC AUC = receiver operator characteristic area under the curve. † p = 0.22 compared to ExECG. ⁎ p = 0.26 compared to ExECG.
The ROC AUC improved to 97% for CCTA N 50% stenosis, and 97% for CCTA N 70% stenosis, but did not reach statistical significance (Table 2, Fig. 2). 3.2. Cost analysis Costs from a hospital perspective, including the cost of downstream testing, hospital labor costs and management 30 days after admission, demonstrated a significantly lower per-patient total cost in the CCTA arm of $2193 (95% CI: $1997, $2389) compared to the ExECG arm cost of $2704 [(95% CI: $2555, $2853), p b 0.001]. Moreover, cost was further reduced in those subjects who could be discharged directly from the ED without ward admission [CCTA $1669 (95% CI: $1612, $1726) compared to ExECG $2459 (95% CI: $2397, $2521), p b 0.001]. The reduction in total per-patient cost from the hospital payer perspective was primarily driven by the reduced length of stay. 3.3. Secondary outcomes Length of stay was significantly reduced by 34% in the CCTA arm [13.5 h (95% CI: 11.2–15.7)] compared to the ExECG arm [20.7 h (95% CI: 17.9–23.1), p b 0.0001] (Table 3). Similarly, time to discharge from the ED was also significantly reduced (p b 0.0001). Hospital admission
rates were similar in both groups (CCTA 10.3% versus ExECG 10.8%; odds ratio for admission (OR) of 1.1 (95% CI: 0.6, 1.8); p = 0.8; Table 3). However, CCTA-based care increased rates of downstream cardiac testing compared to ExECG [13.4% versus 7.5%; OR 2.0 (95% CI: 1.1, 3.8); p = 0.02], This increase primarily resulted from a higher invasive coronary angiography rate [9.0% vs 4.2%, odds 2.3 (95% CI: 1.1, 4.7), p = 0.028]. Repeat presentation for recurrent chest pain or symptoms over 12 month follow-up was not statistically different between groups [12.7% vs 10.0%; OR 1.3 (95% CI: 0.8, 2.3); p = 0.30] (Table 3). 3.3.1. Radiation exposure The mean radiation exposure for CCTA was 3.8 mSv (95% CI: 3.5, 4.1 mSv, range: 0.63–16.9 mSv). The mean heart rate during CCTA acquisition was 59 bpm (95% CI: 58.6, 60.4 bpm). 3.4. Patient follow-up At 30 day follow-up, no major coronary events were identified (0% prevalence of ACS) and no deaths occurred. At 12-month follow-up, 4 subjects had events, 3 (0.9%) in the CCTA arm and one (0.4%) in the ExECG arm. One CCTA subject with mild coronary artery disease presented at 6 months with ACS and invasive angiography demonstrated a spontaneous coronary artery dissection of a 3rd marginal branch. There was no visible coronary disease at this location on the study CCTA. Two additional subjects in the CCTA arm died within 12 months; one for an unrelated admission for urosepsis, and one during a planned elective admission for aortic valve replacement complicated by multiorgan failure requiring extra-corporeal membrane oxygenation. The one subsequent death in the ExECG arm resulted from complications of cancer therapy. 4. Discussion For subjects presenting to the ED with acute chest pain, CCTA had better positive predictive value than ExECG, combined with a significantly reduced length of stay and reduced per-patient cost. In an outpatient population, CCTA has emerged as an accurate and rapid tool for the exclusion of coronary artery disease [15–19]. Recent randomized controlled trials in acute ED patients that compared CCTA to nuclear SPECT [7], or to a mixed standard-of-care [5,6] showed
Table 3 Study endpoints.
Fig. 2. Receiver-operator characteristic curves for diagnostic accuracy of ACS for ExECG and (A) CCTA N50% stenosis (B) and CCTA N70% stenosis.
Length of stay (hours, 95% CI) Inpatient admission (%) Downstream testing (%) Invasive angiography Echocardiography Re-presentation (%) Cost ($AUD, 95% CI) All comers Discharged from ED
CCTA (n = 322)
ExECG (n = 240)
p-Value
13.5 (11.2–15.7)
19.7 (17.3–22.0)
0.003
10.2 10.8 7.1 2.2 12.7
10.8 5.8 3.3 1.3 10.5
0.800 0.020 0.028 0.300 0.300
$2193 (1997–2389) $1669 (1612–1726)
$2704 (2555–2853) $2459 (2397–2521)
b0.001 b0.001
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
lower length of stay at lower cost. However, diagnostic performance measures were not included. ROMICAT-II, which used stress imaging as the comparator, is the largest diagnostic performance CCTA study to date and showed a high sensitivity with modest specificity, similar to our data. However, most international guidelines suggest that the least expensive and most widely available stress test, treadmill ECG, should be the first line test for patients at intermediate risk for acute coronary syndrome. The recent CT-STAT study noted that the lack of stress ECG as a comparator was a limitation of that large multicenter trial [20]. CCTA and ExECG-based care has not been directly compared to date. The current CT-COMPARE study represents the largest prospective, randomized trial of CCTA comparing to treadmill exercise ECG as part of the standard-of-care. Our data showed that CCTA-based evaluation is 35% faster and 20% less expensive than ExECG in patients at low– intermediate risk of acute coronary syndrome. These results suggest that CCTA appears to be a useful initial testing strategy for patients at low to intermediate risk for ACS. However, CCTA may unfavorably influence patient care. Diagnosis of acute coronary syndrome was statistically higher in the CCTA group, due to a higher rate of adjudicated unstable angina pectoris. Other randomized trials have also found increased rates of detection of coronary artery disease in the CCTA arm [6,7]. This may be a confounder effect from open, non-blinded access to imaging data as patients with observed coronary artery disease on CCTA may be more likely to be classified as unstable angina. This may also explain the increased downstream testing after CCTA, particularly invasive coronary angiography, in our study and others [6,21]. The decision to perform additional tests including invasive angiography was at the discretion of the treating cardiologist, and observed coronary artery disease may drive these decisions more than stress testing. However, this does reflect a real world approach to patient care. Importantly, the definition of “unstable angina” without biomarker positivity has been questioned, in the era of high-sensitivity troponins [22]; however, in the present study, standard-sensitivity troponins were used as per clinical care in the State of Queensland. Finally, CCTA is not suitable for all-comers; indeed, some have questioned the need for any imaging test in these lower risk patients, advocating accelerated protocols based on biomarkers alone [23]. Further studies into thresholds for invasive angiography and protocol driven decision-making appear warranted. Other data within the trial also deserve attention. The prevalence of coronary disease of 10.8% and ACS 4.2% is consistent for low to intermediate risk populations and suggests wide generalizability of these data. The prevalence is also within an ideal range for which CCTA is predicted to be cost-effective [24,25]. Between groups, the baseline risk appeared slightly higher for the CCTA group, with increased rates of hypertension, smoking and dyslipidemia. This may have increased the prevalence of coronary artery disease in the CCTA arm, but should be accounted for by randomization. Despite this, the overall per-patient cost, inclusive of invasive coronary angiography and interventions, remained 20% lower in the CCTA arm. Patient radiation exposure remains an important concern in diagnostic imaging. The mean effective dose in this study was 3.8 mSv (95% CI: 3.5–4.1 mSv). This exposure is substantially lower than the 11.5 mSv reported in CT-STAT [5], 13.9 mSv in ROMICAT-II [6], and 20.4 mSv in our previous observational study [9]. Mean heart rate during CT acquisition was 59 bpm which enabled effective use of CT dosereduction techniques [26]. In contrast, the ACRIN-PA study reported a substantial proportion (16%) of subjects randomized to CT who were unable to undergo the test due to persistently elevated heart rate [7].
871
primarily due to loss of paper documentation during patient transfers. This resulted in a disproportionate exclusion of patients in the ExECG group. An analysis between these excluded and study patients show no meaningful differences between baseline data (clearly outlined in Appendix 2), but primary outcomes could be affected if their data were included in the final analyses. Second, there is limited applicability of the CCTA strategy to subjects outside of the protocol (males b30 or females b 40 years of age, renal impairment or contraindication to heart rate control). Third, costs for this trial were from a hospital perspective, including labor costs, which is different than the payer perspective from other trials. We believe that this reflects a more realistic cost to the patient and care-givers especially when limited health care resources are considered. Lastly, this study was not powered to compare outcome data, which requires a much larger trial; due to the low event rates, as noted in the CT-STAT paper, the number needed to evaluate for missed MACE would be over 45,000 patients [5]. 6. Conclusions In symptomatic patients at low–intermediate risk for acute coronary syndrome presenting to the emergency department, coronary CT angiography is faster and less expensive, with improved diagnostic performance compared to exercise treadmill ECG-based care. These data add additional proof that protocols using CCTA rather than stress testing-based care in emergency departments should be considered. Further verification of these data in a multicenter randomized trial is warranted. Acknowledgments We thank the staff of the Prince Charles Hospital Medical Imaging Department, Liz Warburton, Steve Graves, Kathryn Arnett, and our patients. Appendix 1. Flow diagrams for CCTA and ExECG arms of the trial.
A
Intermediate risk pain No known CAD
Contraindication to ExECG or CCTA NO
YES
Able to be randomised to ExECG vs CCTA
Excluded from study
Consents to Study
NO
Refused Consent Thus follow standard care pathway
YES
Randomised to ExECG vs CCTA
ExECG Group
CCTA Group
5. Limitations Limitations of this study need to be addressed. First, of the 21% of eligible patients not enrolled, 11% had misplaced informed consent
1A. CAD = coronary artery disease, CPAS = chest pain assessment service, CCTA = coronary CT angiography, and ExECG = exercise treadmill electrocardiography.
872
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
B
Intermediate risk pain No known CAD Successfully recruited and randomised CCTA
ExECG
CCTA
ExECG
Negative serial ECG’s, enzymes enzymes and stress test
Inconclusive stress test result
Low risk Immediate discharge from traditional chest pain assessment service.
Low risk Discharge to hiome
or stress
Intermediate risk
Further testing (e.g. MPS or Cath) Normal Abnormal
Positive ECG, test
Suspect high risk ACS Admitted to ward/CCU
Further testing (e.g. MPS or Cath)
Abnormal
Presumed ACS Ward/CCU Admission. Invasive angiography unless already performed or contraindicated.
Normal
Low risk Consider other diagnosis of noncoronary cardiac damage such as myocarditis.
Presumed ACS Ward/CCU Admission. Invasive angiography unless already performed or contraindicated.
1B. Flow diagram. CAD = coronary artery disease, CPAS = chest pain assessment service, ACS = acute coronary syndrome, CCTA = coronary CT angiography, CCU = coronary care unit, ExECG = exercise treadmill electrocardiography, and MPS = stress nuclear myocardial SPECT.
Appendix 2. Comparison of patients excluded due to lack of consent documentation.
Age (years) Male gender (%) Weight (kg) Height (cm) Hypertension, n (%) Dyslipidemia Diabetes Smoking Family history
Trial patients (n = 563)
No consent (n = 82)
p value
52.3 ± 10.4 325 (58%) 86.0 ± 21 171 ± 10 173 (31%) 138 (25%) 38 (7%) 132 (23%) 186 (33%)
54.2 ± 10.6 38 (46%) 86.0 ± 23 166 ± 19 24 (29%) 22 (27%) 9 (10%) 20 (24%) 30 (37%)
0.13 0.06 0.94 0.003 0.79 0.65 0.35 0.82 0.50
References [1] C.N. Aroney, H.L. Dunlevie, J.H. Bett, Use of an accelerated chest pain assessment protocol in patients at intermediate risk of adverse cardiac events, Med. J. Aust. 178 (2003) 370–374. [2] W.A. Parsonage, L. Cullen, J.F. Younger, The approach to patients with possible cardiac chest pain, Med. J. Aust. 199 (2013) 30–34. [3] R.J. Gibbons, G.J. Balady, J.T. Bricker, et al., ACC/AHA 2002 guideline update for exercise testing: summary article. A report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines (committee to update the 1997 exercise testing guidelines), J. Am. Coll. Cardiol. 40 (2002) 1531–1540. [4] R. Gianrossi, R. Detrano, D. Mulvihill, et al., Exercise-induced ST depression in the diagnosis of coronary artery disease. A meta-analysis, Circulation 80 (1989) 87–98.
[5] J.A. Goldstein, K.M. Chinnaiyan, A. Abidov, et al., The CT-STAT (Coronary Computed Tomographic Angiography for Systematic Triage of Acute Chest Pain Patients to Treatment) trial, J. Am. Coll. Cardiol. 58 (2011) 1414–1422. [6] U. Hoffmann, Q.A. Truong, D.A. Schoenfeld, et al., Coronary CT angiography versus standard evaluation in acute chest pain (ROMICAT-II), N. Engl. J. Med. 367 (2012) 299–308. [7] H.I. Litt, C. Gatsonis, B. Snyder, et al., CT angiography for safe discharge of patients with possible acute coronary syndromes (ACRIN-PA), N. Engl. J. Med. 366 (2012) 1393–1403. [8] K.R. Branch, B.W. Bresnahan, D.L. Veenstra, et al., Economic outcome of cardiac CT-based evaluation and standard of care for suspected acute coronary syndrome in the emergency department: a decision analytic model, Acad. Radiol. 19 (2012) 265–273. [9] M. Hansen, J. Ginns, S. Seneviratne, et al., The value of dual-source 64-slice CT coronary angiography in the assessment of patients presenting to an acute chest pain service, Heart Lung Circ. 19 (2010) 213–218. [10] Cardiac Society of Australia and NZ — guidelines for the management of acute coronary syndromes 2006, Med. J. Aust. 184 (2006) S9–S29. [11] A.W. van't Hof, A. Liem, M.J. de Boer, F. Zijlstra, Clinical value of 12-lead electrocardiogram after successful reperfusion therapy for acute myocardial infarction. Zwolle Myocardial infarction Study Group, Lancet 350 (1997) 615–619. [12] G.Y. Liew, M. Feneley, S.G. Worthley, Noninvasive coronary artery imaging: current clinical applications: Cardiac Society of Australia and New Zealand guidelines, Heart Lung Circ. 20 (2011) 425–437. [13] D.P. Chew, C.N. Aroney, P.E. Aylward, et al., Addendum to the National Heart Foundation of Australia/Cardiac Society of Australia and New Zealand Guidelines for the management of acute coronary syndromes (ACS) 2006, Heart Lung Circ. 20 (2011) 487–502. [14] S. Goodacre, J. Nicholl, S. Dixon, et al., Randomised controlled trial and economic evaluation of a chest pain observation unit compared with routine care, BMJ 328 (2004) 254. [15] A. Arbab-Zadeh, J.M. Miller, C.E. Rochitte, et al., Diagnostic accuracy of computed tomography coronary angiography according to pre-test probability of coronary artery disease and severity of coronary arterial calcification. The CORE-64 (Coronary Artery Evaluation Using 64-Row Multidetector Computed Tomography
C. Hamilton-Craig et al. / International Journal of Cardiology 177 (2014) 867–873
[16]
[17]
[18]
[19]
[20]
[21]
Angiography) international multicenter study, J. Am. Coll. Cardiol. 59 (2012) 379–387. S. Leschka, P. Stolzmann, L. Desbiolles, et al., Diagnostic accuracy of high-pitch dualsource CT for the assessment of coronary stenoses: first experience, Eur. Radiol. 19 (2009) 2896–2903. A.L. Vavere, A. Arbab-Zadeh, C.E. Rochitte, et al., Coronary artery stenoses: accuracy of 64-detector row CT angiography in segments with mild, moderate, or severe calcification—a subanalysis of the CORE-64 trial, Radiology 261 (2011) 100–108. M.J. Budoff, D. Dowe, J.G. Jollis, et al., Diagnostic performance of 64-multidetector row coronary computed tomographic angiography for evaluation of coronary artery stenosis in individuals without known coronary artery disease: results from the prospective multicenter ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) trial, J. Am. Coll. Cardiol. 52 (2008) 1724–1732. W.B. Meijboom, A.C. Weustink, F. Pugliese, et al., Comparison of diagnostic accuracy of 64-slice computed tomography coronary angiography in women versus men with angina pectoris, Am. J. Cardiol. 100 (2007) 1532–1537. M. Salerno, J.M. Bourque, G.A. Beller, Coronary angiographic evaluation of low-risk chest pain in the emergency department CT-STAT, or maybe not quite that fast? J. Am. Coll. Cardiol. 58 (2011) 1423–1425. E. Hulten, C. Pickett, M.S. Bittencourt, et al., Outcomes after coronary computed tomography angiography in the emergency department: a systematic review and
[22]
[23]
[24]
[25]
[26]
873
meta-analysis of randomized, controlled trials, J. Am. Coll. Cardiol. 61 (2013) 880–892. L.T. Burgers, S.T. Nauta, J.W. Deckers, J.L. Severens, W.K. Redekop, Is it cost-effective to use a test to decide which individuals with an intermediate cardiovascular disease risk would benefit from statin treatment? Int. J. Cardiol. 176 (3) (Oct 20 2014) 980–987. M. Than, L. Cullen, S. Aldous, et al., 2-Hour accelerated diagnostic protocol to assess patients with chest pain symptoms using contemporary troponins as the only biomarker: the ADAPT trial, J. Am. Coll. Cardiol. 59 (2012) 2091–2098. J.M. May, W.P. Shuman, J.N. Strote, et al., Low-risk patients with chest pain in the emergency department: negative 64-MDCT coronary angiography may reduce length of stay and hospital charges, AJR Am. J. Roentgenol. 193 (2009) 150–154. V.L. Priest, P.A. Scuffham, R. Hachamovitch, T.H. Marwick, Cost-effectiveness of coronary computed tomography and cardiac stress imaging in the emergency department: a decision analytic model comparing diagnostic strategies for chest pain in patients at low risk of acute coronary syndromes, JACC Cardiovasc. Imaging 4 (2011) 549–556. C. Hamilton-Craig, R. O’Rourke, A.J. Wesley, J. Ayres, M. Molan, Temporal Trends in Cardiac CT Radiation Exposure and the Impact of the PROTECTION trials in an Australian Academic Center, J. Cardiovasc. Comput. Tomogr. 6 (July/August 2012) 4S.
