We developed a decision tree model to test the CE of three different strategies: (A) defer testing; (B) perform CACS, withholding further testing if CACS=0 and proceeding to CCTA if CACS>0; and (C) perform first-line CCTA in all cases, without prior CACS. In this simulation, CCTA was defined as the gold standard, yielding 100% diagnostic accuracy for obstructive CAD, and no non-diagnostic results. Nonetheless, to better reflect real-world clinical practice and costs, we considered that all patients with obstructive CAD on CCTA would ultimately undergo a confirmatory invasive coronary angiography (ICA) (Figure 1). Bayesian inference was then used to estimate the costs and effectiveness of each strategy according to the test characteristics.9

Figure 1.

Decision tree model for patients presenting with chest pain and low clinical pretest probability of suspected CAD (<15%): not testing (A), CACS as a gatekeeper for CCTA (B), and first-line CCTA (C). In strategy (B), all patients with CACS>0 would undergo CCTA to confirm or exclude obstructive CAD. The cost of a diagnostic ICA was later added to all patients with obstructive CAD on CCTA.

(0.19MB).

The base case population derives from a cross-sectional study performed at two tertiary centers in Lisbon, Portugal.10 Briefly, after excluding those with non-diagnostic results, a total of 1385 patients with non-acute chest pain and PTP <15%, who underwent CACS immediately followed by CCTA, were included.10 The PTP of obstructive CAD was estimated for each patient using the Juarez-Orozco tabular method endorsed by current societal guidelines.5,6

All scans were performed using an ECG-gated acquisition on dual-source 64-slice and 192-slice computed tomography scanners (Somatom Definition® and Somatom Force® Siemens Healthineers, Erlangen, Germany) in accordance with Society for Cardiovascular Computed Tomography guidelines.11 CACS was assessed on a non-contrast acquisition using the Agatston scoring method.12 Obstructive CAD was defined as the presence of maximal diameter stenosis ≥50% in any major epicardial vessel.

The baseline characteristics of the base case population are detailed in Supplementary Table 1. Key input data included the proportion of patients with obstructive CAD on CCTA (10.3%), the proportion of those with CACS=0 (57%), and the negative predictive value of CACS for obstructive CAD on CCTA (98.1%).

Comparison between each testing strategy was performed from the perspective of the society, including all attributable costs regardless of the payer. The CE of each strategy was defined as the cost per correct diagnosis (inclusion or exclusion of obstructive CAD). Accordingly, a decrease in that cost indicates improved CE. Mean costs, effectiveness (i.e., diagnostic performance), and cost-effectiveness ratios were derived. The incremental cost-effectiveness ratio (i.e., the added cost per additional correct diagnosis; ICER) was calculated for each strategy relative to the previous less expensive one, after excluding those subjected to strong dominance (less effective and more costly) and extended dominance (less effective with a higher ICER).

The overall cost of a diagnostic strategy includes both direct and indirect costs. Direct costs were calculated according to the current price list from the Portuguese National Health Service and are depicted in Table 1.13 In the CACS±CCTA strategy, a cost of €88.70 was assumed for patients undergoing CACS alone, whereas a cost of €299.40 was considered for those who also underwent CCTA (sum of the prices of CACS and CCTA). The indirect cost included (i) the cost associated with incidental findings during CACS and CCTA; (ii) the cost related to complications of ICA; and (iii) the cost of false negative tests:

CAC scoring and CCTA may uncover incidental non-cardiac findings of uncertain significance that warrant follow-up examination, most frequently lung nodules.14 Accordingly, we have assumed that 7% of patients undergoing those testing strategies would require a follow-up non-contrast chest CT scan, at an additional cost of €74.70.13,14

Because in this simulation all patients with a diagnosis of obstructive CAD on CCTA would undergo ICA, we assumed that the rate of major cardiovascular adverse complications (death, myocardial infarction, or stroke) of this procedure would be 0.083%, as reported in the literature.15 According to a previous study for the Portuguese reality, the cost of a typical ICA complication was estimated at €4300.16 The costs of complications related to noninvasive tests were not included in this simulation, as severe complications are extremely rare and would only marginally increase costs. Furthermore, costs regarding potential exposure to ionizing radiation were not included as this was a short-term CE study, and because the clinical impact of the radiation dosages used in these imaging tests is still uncertain.17

The costs of false negatives are remarkably difficult to estimate and include the cost of additional testing and of added risk for cardiovascular events (due to inappropriate management of CAD).16 In previous studies, these costs have been variably estimated as two to six times the cost of a false-positive.18–21 As such, a false negative was conservatively set at three times the cost of a false-positive test (€1178.96). In the defer testing strategy, the costs of false negatives were the only ones considered.

It is known that for the same age and chest pain characteristics, women have a lower PTP of obstructive CAD and a lower likelihood of having calcified atherosclerotic plaques compared to men.5,22 Accordingly, we performed sex-specific CE analyses, to test whether the cost-effectiveness of each strategy would vary among both sexes.

Additionally, to test the robustness of our model and to assess the extent to which the results depended on some of the variables under study, several sensitivity analyses were performed. Calculations were repeated using new assumptions, such as varying the cost of both CACS and CCTA from −50% to +50%, the cost of a false negative result from two to five times the cost of a false-positive, the prevalence of obstructive CAD from 5% to 14%, and the proportion of patients with noncalcified obstructive CAD from 0.5% to 5%. The fundamental definitions of comparative cost-effectiveness analysis can be found elsewhere.23

In our base case population, the prevalence of obstructive CAD was 12.5% for men and 9.7% for women and the proportion of patients with CACS=0 was 56.7% and 57.0%, respectively. The likelihood of having obstructive CAD without any coronary artery calcification was numerically higher for women compared to men (2.1% vs. 1.2%, p=0.46), yielding an overall diagnostic accuracy of the strategy using CACS±CCTA of 98.8% and 99.3%, respectively (Table 3). The ICER for strategy B vs. A was €943 in men and €1527 in women, whereas the ICER for strategy C vs. B was €3450 and €1972, respectively (Table 3).

The results of the sensitivity analyses are depicted in Table 4. Overall, increasing the cost of CAC scoring by 50% or decreasing the price of a CCTA by 50% would reclassify the CACS±CCTA strategy as non-cost-effective (strong dominance of stand-alone CCTA). First-line CCTA would also be the most cost-effective approach should the rate of false negatives in the strategy using CACS as a gatekeeper (i.e., obstructive CAD with CACS=0) increase to 5%. For a prevalence of obstructive CAD of 5%, not testing would be cost-effective at a CET of €3674.

The burden of non-acute chest pain in healthcare services remains remarkably high, with patients with a low pretest probability of CAD accounting for a large proportion.25 Whether, on the one hand, effective diagnostic strategies are necessary to detect CAD and eventually inform about secondary preventive measures, the low diagnostic yield of testing in this setting can result in a high rate of false-positive results, in an increased number of incidental findings of unknown significance, and in higher costs for the society.2,4 Along with the overall good prognosis of patients with low PTP, these arguments have been the rationale for recommendations to defer testing in this subset of patients.5 Notwithstanding this, it has been previously shown that CCTA referrals in patients with PTP < 15% can account for over 60% of all referrals due to suspected CAD, possibly because many are not willing to accept the diagnostic uncertainty of not testing.10

The selection of a diagnostic method, beyond its diagnostic accuracy, should also consider costs. As opposed to long-term CE studies – using quality-adjusted life years (QALYs) as a measure of outcome – a universally accepted CET to provide guidance on adopting or dismissing a certain diagnostic strategy does not exist. Of note, a previous study on the cost-effectiveness of different diagnostic strategies for CAD in Portugal identified a CET of at least €1010 for the Portuguese society,16 corresponding to the added costs of stress single photon emission computed tomography (SPECT) followed by ICA, one of the most used diagnostic pathways in our country. In that study, at a willingness to pay of €1000, CACS±CCTA has been shown to be cost-effective for the detection of obstructive CAD in symptomatic patients with a PTP <20–30%.16 No further studies have explored the cost-effectiveness of CACS as a gatekeeper in an all-comer population with suspected CAD, and to the best of our knowledge, our study is the first to access it specifically in symptomatic patients with a low PTP of obstructive CAD. On the other hand, in the setting of primary prevention in asymptomatic patients, CACS has been shown to be cost-effective in guiding the initiation of statin therapy.26 Although cost-effectiveness studies can inform clinical decision-making, the best strategy will ultimately depend on society's willingness to pay for an additional correct diagnosis and on the number of missed diagnoses stakeholders are willing to accept.

We have found that the CETs of each diagnostic method vary among both sexes. Using CACS as a gatekeeper for CCTA was the most cost-effective strategy between a willingness to pay for additional correct diagnosis of €1527 to €1972 for women and €943 to €3450 for men. The variable rate of false negatives from using this approach can help explain this difference. In our study, despite the prevalence of obstructive CAD being lower in women, and the proportion of patients without coronary artery calcifications being somewhat similar, female patients were approximately twice as likely to have obstructive CAD with CACS=0. It should be noted however that data from prior studies on plaque composition have shown heterogeneous findings for women. The CONFIRM registry, including over 5600 patients, reported a higher prevalence of plaques of every type in men than in women.27 In contrast, other studies, including one with 1050 matched patients found a higher proportion of non-calcified plaques in females.28,29

Of the scenarios assessed in the sensitivity analyses, the influence of the prevalence of obstructive CAD is particularly important. For very low probabilities of obstructive CAD (prevalence <5%), not testing would be a cost-effective approach for a wide margin of CET. In this context, CACS±CCTA would not be cost-effective, as it is extendedly dominated by the strategy using first line CCTA. Of note, the results remained similar when varying the cost of a false negative from two to five times the cost of a false-positive.

Some limitations of this study should be acknowledged. Firstly, our definition of obstructive CAD was based on CCTA and therefore we did not consider the costs of false negative CCTA results (having ICA as the reference). However, provided that the negative predictive value of CCTA in patients with low PTP ranges from 97% to 99%, the additional costs would be neglectable.30 Secondly, as false-positive results on CCTA are far more frequent in this setting, the proportion of missed diagnoses as well as the costs resulting from not testing are likely overestimated.4 Nevertheless, to reflect better real-world clinical practice, we considered that all patients with obstructive CAD on CCTA would undergo ICA. Thirdly, as this was a short-term CE analysis, we did not evaluate the impact of each strategy on subsequent therapeutic adjustments and their potential impact on quality-adjusted life years, even though there is currently evidence that using CACS/CCTA to guide therapy may improve clinical outcomes.31 Likewise, the economic impact of detecting non-obstructive CAD was not ascertained. Finally, the results of this CE study refer to patients with low PTP and to the costs currently practiced in the Portuguese National Health System and may not be applicable to other settings.