This was an observational, cross-sectional study performed at a single hospital center serving an urban population of 900 000 inhabitants in Lisbon, Portugal. The study population consisted of all patients referred for elective ICA for evaluation of chest pain symptoms between January 2006 and November 2010. Patients’ referral for ICA and the decision to perform previous noninvasive testing, including the testing modality, were left to the discretion of attending physicians. Noninvasive testing was performed mostly at private practice facilities.

The modalities of noninvasive testing were exercise electrocardiogram (ECG) stress testing, stress myocardial single-photon emission computed tomography (SPECT), stress echocardiography and coronary computed tomography angiography (CCTA). ‘Ischemic changes’ on the resting ECG were not considered noninvasive testing. The following exclusion criteria were applied sequentially: non-elective setting (acute coronary syndrome), previously known CAD (defined as previous acute coronary syndrome, revascularization procedure or documented coronary stenosis ≥50% on previous ICA), preoperative evaluation, presenting symptom other than chest pain, negative noninvasive test result and incomplete information on patients’ clinical characteristics or ICA result.

Data on demographic characteristics, cardiovascular risk factors, type of noninvasive testing and results of coronary angiography were prospectively collected in the ongoing ACROSS (Angiography and Coronary Revascularization On Santa cruz hoSpital) registry, approved by the local ethics committee. The diagnoses of hypertension, hypercholesterolemia and diabetes (regardless of type, duration or current treatment) were assigned if indicated in the patients’ referral letter or if the patient was being treated with antihypertensive or lipid-lowering drugs, oral antidiabetics or insulin. To avoid underdiagnosis, obstructive coronary was defined as a ≥50% reduction in vessel diameter as compared to a nondiseased proximal segment. This broad definition of obstructive CAD was not used as a criterion for revascularization.

Data are presented as counts (%), medians (interquartile range) or means ± standard deviation. Categorical variables were compared using Fisher's exact test. Continuous variables were compared by means of the t test. Patients with and without obstructive CAD were compared for differences in age, gender, body mass index, prevalence of cardiovascular risk factors and use of noninvasive testing. Variables that showed significant association with obstructive CAD (p<0.10) in univariate analysis were included in a binary logistic regression model to identify independent predictors. Temporal differences during the study period in the prevalence of obstructive CAD and the use of noninvasive testing were assessed using the chi-square test for trend. A two-sided p-value of less than 0.05 was considered to indicate statistical significance. All analyses were performed using the statistical package SPSS® version 17.0 (SPSS, Inc., Chicago, IL).

In our European, urban clinical setting, less than 57% of patients referred for elective ICA for evaluation of chest pain symptoms had obstructive lesions (defined by a broad criterion of ≥50% luminal stenosis), despite the fact that four out of five patients had undergone previous testing. Noninvasive testing was frequently used but was only a weak independent predictor of obstructive CAD (OR 1.34, p=0.028). This apparently low performance of noninvasive tests as gatekeepers for ICA has several possible explanations. One is that the performance of these tests in the “real world” is worse than that reported in the literature from large experienced centers. Another explanation would be a low pretest probability of obstructive CAD in this population, resulting in a relatively large absolute number of patients without obstructive disease undergoing ICA.7 A third hypothesis, supported by increasing evidence,6,8 is that the pretest likelihood of angiographically significant CAD may be overestimated when calculated on the basis of age, gender and chest pain characteristics in accordance with the seminal work of Diamond, Forrester and Pryor.9,10

The yield of any diagnostic test depends on the pretest likelihood of the patients in whom it is used and on the way the test modifies that probability. Ideally, a positive noninvasive test should increase the probability of obstructive disease to a level that justifies performing ICA, and a negative test should reduce that likelihood to a level at which obstructive CAD can be safely ruled out. While ICA will always be performed on some patients without coronary lesions, the 2011 standards for catheterization laboratory accreditation from the Accreditation for Cardiovascular Excellence organization suggest that the incidence of non-obstructive disease in elective patients should be <40%. In the interests of individual patients and of overall healthcare cost-effectiveness, extreme rates are undesirable. Recently, Genders et al.6 reported a rate of obstructive CAD of 58% (ranging from 39.4% to 75.5%) in a multicenter study involving 11 European hospitals. In the USA, Patel et al.5 reported an overall rate of 41% of patients with obstructive CAD in the National Cardiovascular Data Registry, although this varied significantly between different centers, from 23 to 100%.11 Taken together, these studies suggest that better gatekeepers are needed. Our findings are in line with both these studies, underlining the relatively low prevalence of obstructive CAD on ICA in a population with a high frequency of noninvasive testing.

We were also able to assess differences between noninvasive tests. It should be noted that our findings are mainly the result of using functional tests, particularly exercise ECG and stress SPECT, which accounted for more than 90% of testing, as gatekeepers for ICA. Although the proportion of obstructive CAD was higher in the CCTA group than for functional tests, it is uncertain whether the overall results would have been different if anatomic tests had been used more frequently. There is some evidence that CCTA may be a useful and cost-effective gatekeeper for ICA (particularly in patients with intermediate to low pretest probability), reducing the number of patients without obstructive CAD referred for invasive testing.12–19 Recent guidelines for the management of patients with chest pain from the United Kingdom's National Institute for Health and Clinical Excellence (NICE) recommend choosing tests according to the pretest probability of CAD. Functional imaging tests are preferred for patients with 30–60% pretest probability of disease, whereas CCTA (preceded by calcium score) is the preferred method for patients with 10–29% pretest probability.20 According to these guidelines, ICA should be offered as the first test to patients with pretest probabilities over 60%. Currently, there is disagreement over which type of test should be used as first line.21 Results from the US National Institutes of Health-sponsored PROMISE study (a clinical endpoint-driven randomized study comparing functional studies with CCTA for the evaluation of patients with suspected CAD) will hopefully shed more light on this matter.22,23

Several limitations of this study should be acknowledged. Since the characteristics of chest pain were not systematically assessed and recorded for each patient, it was not possible to calculate the pretest probability of CAD. Although the median age and prevalence of risk factors are compatible with a typical CAD risk population, it is not possible to ascertain whether the weak predictive power of noninvasive testing is related to its application to a population with low pretest probability. Another pitfall is related to the dichotomized classification of noninvasive tests as positive or negative for obstructive CAD. In most tests there is a continuum of ‘positivity’ which is difficult to address with this study design. It should also be emphasized that this was not a randomized trial of noninvasive testing and, as such, direct comparisons of testing vs. no testing and comparisons between noninvasive modalities should be interpreted with caution. The diagnostic performance of noninvasive testing is dependent on the pretest probability of disease, and the decision to perform noninvasive testing and the choice of the test itself depend on the physician's perception of pretest probability, which may have differed between the different diagnostic modalities applied.

The authors declare that no experiments were performed on humans or animals for this study.

The authors declare that they have followed the protocols of their work center on the publication of patient data and that all the patients included in the study received sufficient information and gave their written informed consent to participate in the study.

The authors declare that no patient data appear in this article.