Previous studies have suggested that patients with dyspnea referred for stress testing have high mortality. However, it is not clear whether this is explained by high rates of ischemia. The aim of the present study was to evaluate the incidence of ischemia in patients with dyspnea compared with patients with chest pain referred for stress testing and assess the outcomes of such patients. We systematically searched the electronic databases, MEDLINE, PubMed, EMBASE, and the Cochrane Library, until December 2012 to identify studies of patients with known or suspected coronary artery disease undergoing stress testing. We extracted data on group-specific incidence of stress-induced ischemia and all-cause mortality. In our analyses, we identified and included 6 studies that evaluated a total of 5,753 patients with dyspnea and 24,491 patients with chest pain as the clinical indication for stress testing. There was no statistically significant difference in the incidence of ischemia on stress imaging in patients with dyspnea compared with patients with chest pain (37.4% vs 30.2%, odds ratio 1.43, 95% confidence interval 0.99 to 2.06, p = 0.06). However, during the follow-up period, patients with dyspnea had higher all-cause mortality rates compared with patients with chest pain (annual mortality 4.9% vs 2.3%), with odds ratio of 2.57 (95% confidence interval 1.75 to 3.76, p <0.001). In conclusion, in patients undergoing stress testing, those evaluated for dyspnea had a significant increase in all-cause mortality but did not have higher rates of ischemia compared with patients presenting with chest pain. Clinicians evaluating patients with self-reported dyspnea should be aware that these patients represent a high-risk group with increased risk of mortality.
Stress testing, both exercise and pharmacologic, has been shown to be an important risk stratification tool in a wide range of patients with known or suspected coronary artery disease (CAD). Patients with negative stress test result generally carry a favorable prognosis, whereas patients with stress-induced ischemia are at a greater risk of cardiac events and all-cause mortality. Although chest pain is the most common clinical manifestation of CAD, ischemia can manifest without chest pain. Previous studies using different stress methods are inconsistent in describing the incidence of ischemia in patients with dyspnea compared with patients with chest pain as the clinical indication for testing. Also, previous studies have suggested that dyspnea carries an adverse prognostic significance. The aim of the present study was to describe the incidence of ischemia in patients with dyspnea referred for stress testing compared with patients with chest pain. In addition, it aims to explore its prognostic significance by evaluating all-cause mortality.
Methods
We systematically searched the electronic databases, MEDLINE, PubMed, EMBASE, and the Cochrane Library for Central Register of Clinical Trials, using the MeSH terms, “exercise test,” “stress testing,” “stress echocardiography,” “myocardial perfusion imaging,” “chest pain,” “coronary disease,” “shortness of breath,” and “dyspnea.” We limited our search to studies in human subjects in peer-reviewed journals from 1980 to December 2012. Additionally, a manual search of all relevant references from the screened reports and reviews of cardiac or cardiopulmonary evaluations of dyspnea was performed for additional clinical studies. Studies in the abstract form without a published manuscript were excluded from this analysis.
Two investigators (EA and VA) independently extracted all study data in duplicate using a standardized protocol and reporting form. Differences with regard to any of the extracted data were resolved by reexamining the studies and by consensus. We extracted characteristics of each study, baseline patient information, and data on the incidence of ischemia and all-cause mortality in each subgroup. Eligible studies had to fulfill the following inclusion criteria: (1) patients with known or suspected CAD undergoing stress testing using stress echocardiography (both exercise and pharmacologic), myocardial perfusion imaging (both exercise and pharmacologic), or treadmill exercise electrocardiography and (2) subgroup-specific outcomes (stress-induced ischemia and/or all-cause mortality) reported by investigators for patients with dyspnea and chest pain. In studies defining types of chest pain, both overall chest pain group and typical angina subgroup were used for comparison. In cases of missing data, attempts were made to contact investigators of individual studies to obtain missing data.
We assessed the quality of the studies by the 9-point Newcastle-Ottawa Scale using 3 components: selection of study groups, comparability of study groups, and ascertainment of the study outcomes. The selected studies were assessed for quality, and consensus was obtained. The median score was 7, indicating an acceptable methodological quality ( Table 1 ).
| Study (Follow-Up, Mean, Yrs) | Bernheim et al (5) | Bergeron et al (3.1) | Abidov et al (2.7) | Zellweger et al (NA) | Christopher Jones et al (4.3) | Balaravi et al (10) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Quality score | 7 | 7 | 8 | 6 | 5 | 7 | ||||||
| Stress mode | Exercise echocardiography | Dobutamine echocardiography | Exercise or pharmacologic MPI | Exercise or pharmacologic MPI | Exercise ECG | Exercise or pharmacologic MPI | ||||||
| Group | D | CP | D | CP | D | CP | D | CP | D | CP | D | CP |
| n | 1,363 | 2,630 | 443 | 2,033 | 1,091 | 11,356 | 151 | 760 | 841 | 5,848 | 1,864 | 1,864 |
| Age (yrs) | 70 (10) | 69 (12) | 65 (11) | 59 (13) | 71 (12) | 65 (13) | 72 (10) | 65 (11) | 58 (12) | 53 (12) | 66 (10) | 66 (10) |
| Men (%) | 759 (56) | 1,198 (46) | 237 (54) | 926 (46) | 544 (50) | 6,005 (53) | 70 (46) | 344 (45) | 486 (58) | 3,180 (54) | 974 (52) | 974 (52) |
| Known CAD (%) | 415 (30) | 843 (32) | 53 (12) | 145 (7) | 403 (37) | 3,582 (32) | 0 | 0 | 63 (7) | 609 (10) | 0 | 0 |
| Diabetes mellitus (%) | 336 (25) | 555 (21) | 45 (10) | 167 (8) | 227 (21) | 1,698 (15) | 151 (100) | 760 (100) | 91 (11) | 615 (11) | 424 (23) | 332 (18) |
| Hypertension (%) | 866 (64) | 1,628 (62) | 231 (52) | 846 (42) | 618 (57) | 5,464 (48) | 105 (70) | 515 (68) | 359 (43) | 2,251 (38) | 1,177 (63) | 1,063 (57) |
| Hyperlipidemia (%) | 725 (54) | 1,617 (61) | 223 (50) | 1,078 (53) | 438 (40) | 5,591 (47) | 64 (42) | 386 (51) | 67 (8) | 408 (8) | 1,011 (54) | 1,048 (56) |
| Smoker (%) | 903 (66) | 1,445 (55) | 52 (12) | 277 (14) | 148 (14) | 1,445 (13) | 23 (15) | 100 (13) | 156 (19) | 1,176 (20) | 959 (51) | 850 (46) |
| Body mass index (kg/m 2 ) | 29 (6) | 29 (7) | 28 (5) | 28 (7) | NA | NA | NA | NA | 29 (6) | 29 (6) | NA | NA |
| Lung disease (%) | 238 (17) | 235 (9) | 8 (2) | 36 (2) | NA | NA | NA | NA | 55 (7) | 268 (5) | NA | NA |
We compared the incidence of ischemic electrocardiographic changes and ischemia on imaging studies in patients with dyspnea and chest pain as the clinical indication for stress testing. We also studied the difference in all-cause mortality between the groups. Although 3 studies defined ischemic electrocardiographic changes as ≥1 mm of horizontal or downsloping ST-segment depression occurring 80 ms after the J point, 2 studies had no specific definition. Echocardiography-based studies defined myocardial ischemia as stress-induced new or worsening wall motion abnormalities. Myocardial perfusion imaging studies defined ischemia as the presence of stress-induced reversible perfusion abnormalities.
All analyses were performed by METAN command of Stata 10.1 (Stata Corp., College Station, Texas). The meta-analysis was performed in accordance with the recommendations from the Cochrane Collaboration and the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) statement. Heterogeneity was assessed with the I 2 statistic proposed by Higgins and Thompson, with I 2 <25% considered low and I 2 >75% considered high. The random-effects model as described by DerSimonian and Laird was used to calculate the effect sizes because of known clinical and methodological heterogeneity of the studies. Reported values are 2-tailed, and hypothesis-testing results were considered statistically significant at p <0.05. Small study effect, including publication bias, was tested using the regression intercept of Egger et al and visually by funnel plots. To determine whether individual studies had an undue influence on the overall results (because of size or magnitude of effect), we conducted a post hoc influence analysis using the METANINF command in Stata. In addition, meta-regression was used to evaluate the modifying effect of common clinical and demographic variables, including age, gender, known history of CAD, diabetes mellitus, hypertension, hyperlipidemia, smoking, and known history of underlying lung disease.
A previous study has suggested that in patients with underlying dyspnea without known coronary disease, the incidence of ischemia is low. Also, patients with typical anginal symptoms may carry a greater burden of obstructive CAD compared with other chest pain subgroups. Hence, we separately compared all-cause mortality and ischemia on imaging among patients without known CAD and patients with dyspnea compared with patients with investigator-specified typical angina.
Results
Based on the initial search criteria, we identified 404 studies ( Figure 1 ). Of these, 6 studies were included in our final analyses, which evaluated a total of 30,244 patients, including 5,753 patients (19%) with dyspnea and 24,491 patients (81%) with chest pain as the clinical indication for stress testing. All the studies were observational. Three studies examined patients referred for stress myocardial perfusion imaging (exercise or pharmacologic), 1 study used exercise echocardiography, and 1 study used dobutamine echocardiography ( Table 1 ). In addition, the study by Christopher Jones et al reported patients undergoing symptom-limited exercise treadmill testing.
The characteristics of the studies included in the final analysis are summarized in Table 1 . The mean follow-up duration was 5.5 years (range 2.7 to 10). On average, patients in the dyspnea group were older than patients in the chest pain group (mean age 67 vs 62 years, p <0.001). Also, patients with dyspnea carried a greater burden of hypertension (58% vs 48%, p <0.001) and diabetes mellitus (22% vs 15%, p <0.001). Although 3 studies reported lower mean left ventricular ejection fraction for patients with dyspnea, 2 studies also reported a greater prevalence of left ventricular enlargement on imaging studies among patients with dyspnea.
Five studies reported ischemic electrocardiographic changes during stress testing, and ischemia on imaging was reported by 5 studies. Incidence of ischemia at electrocardiography was similar in patients referred for dyspnea and chest pain (odds ratio [OR] 0.84, 95% confidence interval [CI] 0.55 to 1.28, p = 0.41; Supplement Figure 1 ). Interpretation of this finding is confounded by limited sensitivity and specificity of stress electrocardiography. There was a greater incidence of ischemia on stress imaging in patients with dyspnea compared with patients with chest pain (37.4% vs 30.2%), but this difference was not statistically significant in the random-effects analysis (OR 1.43, 95% CI 0.99 to 2.06, p = 0.06; Figure 2 ). Although there was significant heterogeneity among the studies, in the meta-regression analysis, none of the studied variables (age, gender, known history of CAD, diabetes mellitus, hypertension, hyperlipidemia, smoking, and known history of underlying lung disease) had an influence on the final results.
