Article ID: CJ-22-0225
Background: The exercise stress test is a widely used noninvasive test for diagnosing ischemic heart disease. Patients with a “positive” result have a higher risk than those with a “negative” result. However, the outcomes of patients with an “inconclusive” result remain uncertain.
Methods and Results: We retrospectively collected the data of patients who underwent an ECG-based treadmill stress test between August 2009 and March 2020. Propensity score matching (PSM) was performed to adjust for confounders. Clinical outcomes were compared in terms of all-cause death and cardiovascular (CV) death. Subgroup analysis evaluated treatment interactions, including medication and examinations. In total, 25,475 patients were recruited, and after exclusion and PSM, 4,847 (1,621 with a positive result, 1,606 with a negative result, and 1,621 with an inconclusive result) remained. Compared with the negative group, the inconclusive group, but not the positive group, had a significantly worse outcome in terms of all-cause death (hazard ratio [HR]: 1.834, 95% confidence interval [CI]: 1.34–2.511 and HR: 1.327, 95% CI: 0.949–1.857, respectively); however, CV death was not significantly different in the inconclusive and positive groups (HR: 1.728, 95% CI: 0.413–7.232 and HR: 2.067, 95% CI: 0.517–8.264, respectively).
Conclusions: Clinicians must not underestimate the potential for worse outcomes in patients with an inconclusive stress test result.
The exercise stress test, also known as the ECG treadmill test (TMT), is a noninvasive test for diagnosing ischemic heart disease (IHD). According to the American Heart Association (AHA), patients should undergo a TMT if they have signs or symptoms and intermediate pretest probability of having coronary artery disease (CAD). The TMT can also assist in evaluating exercise-induced arrhythmia or functional capacity.1,2 The results of the TMT are generally summarized as “positive,” “negative” or “inconclusive” (also termed equivocal or indeterminate).3 Patients with a positive result have a higher risk of significant obstructive CAD than those with a negative result4 and require medications or further evaluation, such as coronary angiography (CAG), according to their individual risk and clinical response to treatments.5,6 However, the outcome of patients with an inconclusive result remains unclear, causing uncertainty in clinical decision-making.7
Approximately 22–39% of all patients who complete the TMT have an inconclusive result,8 and because of this relatively large percentage, some doctors suggest further noninvasive or invasive evaluations. Blankstein et al recommended that doctors use other modalities to identify the reason for an inconclusive result.7 For instance, if the inconclusive result is attributed to a submaximal heart rate, coronary computed tomography angiography (CCTA) or pharmacological-induced stress tests can be used; if it is attributed to equivocal ECG changes, the TMT should be repeated. Christman et al conducted a prospective cohort study to identify the frequency of downstream testing following different TMT results, and concluded that individuals with rapid recovery of ECG changes who have inconclusive and negative results are the least likely to benefit from downstream tests.9
Despite these recommendations, few studies have compared the long-term prognosis of patients with an inconclusive result with other patients in a large cohort. Therefore, in this retrospective study, we compared the clinical outcomes (all-cause and cardiovascular (CV) deaths) among positive, inconclusive, and negative result groups. Moreover, we investigated whether certain interventions could change the prognoses of the groups.
The study population comprised all patients who underwent a symptom-driven TMT at National Cheng Kung University Hospital (NCKUH), a medical center in southern Taiwan, between August 2009 and March 2020. In the case of repeat studies of single patients, we considered the results of only the first evaluation. We excluded patients aged under 18 years and those without regular follow-up of at least 6 months at the hospital.
Clinical DataWe retrospectively collected all available clinical information from the NCKUH CV electronic medical records databank, including demographic characteristics, laboratory and imaging data, medication and chart records, underlying comorbidities, and clinical events. For underlying comorbidities and clinical events, we manually performed validation randomly across years to ensure data correctness. CV death was defined as death associated with acute coronary syndrome, myocardial infarction, heart failure or no identifiable cause, which was regarded as sudden CV death.
TMT Protocol and InterpretationThe symptom-limiting Bruce protocol10 was adopted for the TMT at NCKUH. The target heart rate was set at 85% of the maximal age-predicted heart rate (MAPHR), calculated as 220 minus the age in years.11 A positive result was defined as an upsloping ST-segment depression of ≥1.5 mm or downsloping ST-segment depression of ≥1 mm in at ≥2 consecutive leads and lasting >2 min. A negative result was defined as no symptoms being present upon reaching the target heart rate and no significant ST-segment depression on ECG during the TMT. An inconclusive result was defined as either not reaching the target heart rate or the production of uninterpretable ECG recording due to artifacts or preexisting conduction abnormalities such as left bundle branch block (LBBB).7,12 Reasons for not reaching the target heart rate included exercise intolerance and chronotropic incompetence. Therefore, we calculated the metabolic equivalent of task (MET) according to the Bruce protocol13 and regarded the MET as an more standardized indicator of exercise capacity compared with attained stage or exercise duration. In the inconclusive group, those who had a MET <7 were subdivided into the “exercise intolerance” subgroup, and those who had a MET ≥7 were assigned to the “chronotropic incompetence” subgroup. The reason we chose 7 as the cut point was because it separated patients in the inconclusive group into approximately equal size. The METs distribution in the inconclusive group, as well as the attained stage and exercise duration in each group are provided in Supplementary Tables 1,2.
The patients with exercise intolerance had inferior exercise capacity, and those with chronotropic incompetence had superior exercise capacity but could not achieve the maximum heart rate. Through our database search, we identified 4 and 28 patients among 1,621 patients in the inconclusive group with LBBB and missing MET data; they were subsequently classified into the “LBBB” and “unspecified” subgroups respectively. Patients’ hemodynamic responses, clinical symptoms, and equivalent workloads achieved were collected.
Statistical AnalysisThe demographic characteristics were compared between the negative, inconclusive, and positive groups. Continuous variables such as age are expressed as mean and standard deviation; they were assessed for normality using the Shapiro–Wilk test and compared between groups one-way analysis of variance. Categorical variables are presented as number and percentage and were compared using Fischer’s exact test. A 2-tailed P value <0.05 was considered statistically significant.
We conducted 1 : 1 : 1 PSM using a caliper width of 0.1 to balance the following confounders among the 3 groups: age, sex, and underlying comorbidities, including diabetes mellitus, hyperlipidemia, hypertension, chronic kidney disease (CKD), ischemic stroke, and cancer history. We also performed inverse probability of treatment weighting (IPTW) analysis to validate the 1 : 1 : 1 PSM result.
We used a Cox proportional hazards model to evaluate changes in mortality and to create survival plots. We calculated the hazard ratio (HR) and 95% confidence interval (CI) with the negative group serving as the control and compared the clinical outcomes between the positive and inconclusive groups. We also conducted subgroup survival analysis within inconclusive groups based on different New York Heart Association (NYHA) functional classes. Furthermore, we performed multivariate logistic regression to investigate the relationships between clinical outcomes and underlying characteristics in the inconclusive group. Finally, through subgroup analysis, we investigated whether different treatments or interventions could alter clinical outcomes. Interaction terms between the TMT results and different subgroups were evaluated through likelihood ratio tests. All analyses were performed using R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria). We used the “MatchIt” package, version 4.0.2, and “Twang” package, version 2.5, to perform PSM.14
From August 2009 to March 2020, 25,475 TMTs were performed. After exclusion of 3,249 repeat studies and 5,264 patients with inadequate follow-up, 16,295 patients were eligible for study enrollment. The study sample comprised 4,423 (58.8% men, mean age: 58.8±10.5 years) individuals in the positive group, 1,625 (54.5% men, mean age: 59.5±13.4 years) in the inconclusive group, and 10,247 (59.4% men, mean age: 54.1±12.7 years) in the negative group. After PSM was conducted, 1,606 (57.8% men, mean age: 59.57±10.9 years), 1,621 (54.5% men, mean age: 59.5±13.4 years), and 1,621 (55.2% men, mean age: 59.6±11.7 years) patients were assigned to the positive, inconclusive, and negative groups, respectively. Before PSM was performed, age, sex, and potential underlying comorbidities were all significantly different between groups (P<0.001); these parameters were comparable among the groups after matching. Compared with the negative and inconclusive groups, the positive group received more management for atherosclerosis and CV diseases, including antiplatelets (15.7% vs. 33.8% vs. 42.2%, respectively, P<0.001), statins (18.2% vs. 21.9% vs. 24.7%, respectively, P<0.001), further CAG examinations (2.2% vs. 10.7% vs. 18.9%, respectively, P<0.001), further CCTA examinations (2.7% vs. 3.4% vs. 9.2%, respectively, P<0.001), and further thallium scans (3.8% vs. 25.7% vs. 9.7%, respectively, P<0.001). These trends remained significant after PSM (Table 1, Figure 1).
| Before matching | After matching | |||||||
|---|---|---|---|---|---|---|---|---|
| Negative (n=10,247) |
Inconclusive (n=1,625) |
Positive (n=4,423) |
P value | Negative (n=1,621) |
Inconclusive (n=1,621) |
Positive (n=1,606) |
P value | |
| Characteristic | ||||||||
| Age, mean (SD) | 54.05 (12.7) | 59.54 (13.4) | 58.75 (10.5) | <0.001 | 59.61 (11.7) | 59.50 (13.43) | 59.65 (10.9) | 0.95 |
| Male | 6,091 (59.4) | 885 (54.5) | 2,602 (58.8) | <0.001 | 894 (55.2) | 884 (54.5) | 928 (57.8) | 0.14 |
| Diabetes | 1,544 (15.1) | 505 (31.1) | 1,101 (24.9) | <0.001 | 506 (31.2) | 501 (30.9) | 500 (31.1) | 0.98 |
| Hyperlipidemia | 4,928 (48.1) | 879 (54.1) | 2,741 (62) | <0.001 | 883 (54.5) | 879 (54.2) | 872 (54.3) | 0.99 |
| Hypertension | 3,665 (35.8) | 957 (58.9) | 2,198 (49.7) | <0.001 | 970 (59.8) | 953 (58.8) | 955 (59.5) | 0.82 |
| Stroke | 159 (1.6) | 55 (3.4) | 109 (2.5) | <0.001 | 60 (3.7) | 54 (3.3) | 52 (3.2) | 0.74 |
| Cancer | 1,414 (13.8) | 331 (20.4) | 599 (13.5) | <0.001 | 326 (20.1) | 330 (20.4) | 328 (20.4) | 0.97 |
| Chronic kidney disease | 736 (7.2) | 331 (20.4) | 599 (13.5) | <0.001 | 355 (21.9) | 362 (22.3) | 359 (22.4) | 0.94 |
| Exercise capacity during treadmill test | ||||||||
| METs <7 | 1,082 (10.6) | 789 (48.5) | 700 (15.8) | <0.001 | 243 (16.4) | 657 (41.7) | 258 (17.6) | <0.001 |
| Management | ||||||||
| Antiplatelet | 1,305 (12.7) | 550 (33.8) | 1,887 (42.7) | <0.001 | 294 (15.3) | 548 (33.8) | 679 (42.2) | <0.001 |
| Anticoagulant | 69 (0.7) | 38 (2.3) | 52 (1.2) | <0.001 | 15 (0.9) | 38 (2.3) | 21 (1.3) | 0.003 |
| ACEI/ARB | 956 (9.3) | 274 (16.9) | 703 (15.9) | <0.001 | 248 (15.3) | 273 (16.8) | 285 (17.7) | 0.1681 |
| DHPCCB | 879 (8.6) | 265 (16.3) | 557 (12.6) | <0.001 | 256 (15.8) | 264 (16.3) | 240 (14.9) | 0.5699 |
| Diuretic | 184 (1.8) | 120 (7.4) | 134 (3) | <0.001 | 56 (3.5) | 120 (7.4) | 66 (4.1) | <0.001 |
| Statin | 1,350 (13.2) | 356 (21.9) | 1,184 (26.8) | <0.001 | 295 (18.2) | 355 (21.9) | 397 (24.7) | <0.001 |
| Subsequent CAG exam | 149 (1.5) | 175 (10.8) | 833 (18.8) | <0.001 | 35 (2.2) | 174 (10.7) | 304 (18.9) | <0.001 |
| Subsequent CCTA | 255 (2.5) | 60 (3.7) | 556 (12.6) | <0.001 | 44 (2.7) | 56 (3.4) | 148 (9.2) | <0.001 |
| Subsequent thallium scan | 316 (3.1) | 464 (28.6) | 443 (10.0) | <0.001 | 63 (3.8) | 416 (25.7) | 155 (9.7) | <0.001 |
Differences in the characteristics between groups became nonsignificant after PSM. ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blocker; CAG, coronary angiography; CCB, calcium-channel blocker; CCTA, coronary computed tomography angiography; CI, confidence interval; DHP, dihydropyridine; MET, metabolic equivalent of task; PSM, propensity score matching; SD, standard deviation.
Patient selection and study flowchart. TMT, treadmill test.
Regarding the clinical outcomes analysis, before PSM both the positive and inconclusive groups had a significantly higher risk than the negative group of all-cause death (HR: 1.37, 95% CI: 1.10–1.17 and HR: 2.84, 95% CI: 2.23–3.63, respectively) and CV death (HR: 3.81, 95% CI: 1.39–10.49 and HR: 5.19, 95% CI: 1.587–17.03, respectively). After PSM, compared with the negative group, all-cause death (HR: 1.83, 95% CI: 1.34–2.51), but not CV death (HR: 1.73, 95% CI: 0.41–7.23), was significantly higher in the inconclusive group; neither was significant in the positive group (HR: 1.33, 95% CI: 0.95–1.86 and HR: 2.07, 95% CI: 0.52–8.26, respectively). The IPTW analysis revealed similar results. Only all-cause death remained significantly higher in the inconclusive group compared with the negative group (HR: 2.20, 95% CI: 1.66–2.94; Table 2, Figure 2A). Meanwhile, all-cause death between patients with inconclusive and positive results showed no statistically significant difference (HR: 1.38, 95% CI: 1.01–1.89; Supplementary Table 3).
| Negative | Inconclusive | Positive | Inconclusive vs. negative |
Positive vs. negative | |
|---|---|---|---|---|---|
| No. of events (%) | HR (95% CI) | ||||
| Unmatched | |||||
| All-cause death | 227/10,247 (2.2) | 91/1,625 (5.6) | 125/4,423 (2.8) | 2.84 (2.23–3.63) | 1.37 (1.10–1.17) |
| CV death | 6/10,247 (0.1) | 5/1,625 (0.3) | 10/4,423 (0.2) | 5.19 (1.587–17.03) | 3.81 (1.39–10.49) |
| Inverse probability of treatment weighting | |||||
| All-cause death | 227/10,247 (2.2) | 91/1,625 (5.6) | 125/4,423 (2.8) | 2.20 (1.66–2.94) | 1.29 (0.95–1.63) |
| CV death | 6/10,247 (0.1) | 5/1,625 (0.3) | 10/4,423 (0.2) | 2.91 (0.81–10.47) | 2.53 (0.88–7.22) |
| Propensity score matching | |||||
| All-cause death | 70/1,621 (4.3) | 91/1,621 (5.6) | 68/1,606 (4.2) | 1.83 (1.34–2.51) | 1.33 (0.95–1.86) |
| CV death | 3/1,621 (0.2) | 5/1,621 (0.3) | 6/1,606 (0.4) | 1.73 (0.41–7.23) | 2.07 (0.52–8.26) |
Matching through inverse probability of treatment weighting or propensity score matching produced similar results for all-cause or CV-related death among the groups. CI, confidence interval; CV, cardiovascular; HR, hazard ratio.
Comparison of outcomes between the treadmill test (TMT) groups and functional status. We applied the Cox proportional hazards test for the analysis, which revealed that, in the inconclusive group, the cardiovascular (CV) mortality outcome was similar to that in the positive group, but the all-cause mortality outcome was less favorable. The New York Heart Association (NYHA) functional classes correlated well with patient prognosis. In the inconclusive subgroup analysis, cardiovascular mortality was similar between the exercise intolerance and chronotropic incompetence subgroups, whereas the exercise intolerance group exhibited a higher all-cause mortality rate. (A) Among the 3 groups, the inconclusive group had the least favorable all-cause mortality survival. (B) For higher NYHA functional classes, the all-cause mortality survival curve exhibited a downward trend, especially for patients with NYHA classes III and IV. (C) The TMT result did not significantly affect CV-related mortality. (D) At higher NYHA functional class, the CV-related mortality survival curve exhibited a downward trend, especially for patients with NYHA class IV. (E) The exercise intolerance group had the worst all-cause mortality outcome of all inconclusive subgroups. (F) CV-related mortality was similar across all subgroups of the inconclusive group.
Furthermore, functional status correlated strongly with clinical outcomes in the inconclusive group, as in the other two groups. Compared with those with NYHA functional class I, all-cause death was significantly elevated in a stepwise manner in patients with NYHA classes II, III, and IV (HR: 2.27, 95% CI: 1.29–4.00; HR: 2.87, 95% CI: 1.71–4.82; and HR: 3.5, 95% CI: 1.65–7.41, respectively; Table 3, Figure 2C).
| NYHA 1 | NYHA 2 | NYHA 3 | NYHA 4 | NYHA 2 vs. 1 | NYHA 3 vs. 1 | NYHA 4 vs. 1 | |
|---|---|---|---|---|---|---|---|
| No. of events (%) | HR (95% CI) | ||||||
| All-cause death | 28/923 (3.0) | 21/290 (7.2) | 30/291 (10.3) | 9/87 (10.4) | 2.27 (1.29–4.00) | 2.87 (1.71–4.82) | 3.50 (1.65–7.41) |
| CV death | 2/923 (0.7) | 0/290 (0) | 1/291 (0.35) | 2/87 (2.3) | NA | 1.56 (0.14–17.25) | 11.11 (1.56–78.89) |
NA, not applicable; NYHA, New York Heart Association. Other abbreviations as in Table 2.
We also compared the HR of the all-cause and CV-related mortality rates in the inconclusive group. Because of the missing data in the unspecified group and the absence of all-cause or CV-related death in the LBBB group, we mainly compared between the exercise intolerance and chronotropic incompetence subgroups. All-cause death (HR: 0.36, 95% CI: 0.23–0.57) but not CV-related death (HR: 0.51, 95% CI: 0.08–3.03) was significantly lower in the chronotropic incompetence subgroup compared with the exercise intolerance subgroup (Table 4, Figure 2E,F)
| Exercise intolerance |
Chronotropic incompetence |
Chronotropic incompetence vs. exercise intolerance |
|
|---|---|---|---|
| No. of events/total (%) | HR (95% CI) | ||
| All-cause death | 60/681 (8.8) | 28/908 (3.0) | 0.36 (0.23–0.57) |
| CV death | 3/681 (0.4) | 2/908 (0.2) | 0.51 (0.08–3.03) |
Abbreviations as in Table 2.
We further investigated the relationships between clinical outcomes and background characteristics in the inconclusive group. All-cause death was lower in women (odds ratio [OR]: 0.59, 95% CI: 0.36–0.97) and in those who was able to reach a MET >7 (OR: 0.39, 95% CI: 0.25–0.65). All-cause death was higher in the patients with cancer (OR: 8.42, 95% CI: 5.23–13.8) or CKD (OR: 2.34, 95% CI: 1.40–3.90). No background characteristic significantly affected CV-related death (Table 5).
| All-cause death | CV-related death | |||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P value | OR | 95% CI | P value | |
| Characteristic | ||||||
| Age | 1.01 | 0.99–1.03 | 0.28 | 1.14 | 1.00–1.27 | 0.06 |
| Female | 0.59 | 0.36–0.97 | 0.04 | 1.12 | 0.13–7.60 | 0.90 |
| Diabetes | 1.19 | 0.74–1.97 | 0.49 | 1.78 | 0.21–13.01 | 0.56 |
| Hyperlipidemia | 0.80 | 0.49–1.33 | 0.40 | 1.24 | 0.17–10.77 | 0.82 |
| Hypertension | 0.85 | 0.50–1.45 | 0.55 | 0.47 | 0.06–4.21 | 0.46 |
| Stroke | 1.33 | 0.41–3.54 | 0.59 | NA | NA | 0.99 |
| Cancer | 8.42 | 5.23–13.8 | <0.001 | 4.34 | 0.70–33.60 | 0.11 |
| Chronic kidney disease | 2.34 | 1.40–3.90 | 0.001 | 1.70 | 0.20–1.24 | 0.59 |
| Exercise capacity during treadmill test | ||||||
| METs >7 | 0.39 | 0.23–0.65 | <0.001 | 1.26 | 0.14–9.32 | 0.81 |
OR, odds ratio. Other abbreviations as in Tables 1,2.
Finally, in the subgroup analysis to investigate the management effects between the positive and negative groups and between the inconclusive and negative groups, for all-cause death, no significant interactions were observed in terms of medications, including antiplatelets, anticoagulants, renin-angiotensin-aldosterone system inhibitors, and statins, or management followed by early CAG examination. We noted a significant interaction in further thallium scans in the inconclusive group, as compared with the negative group (Figure 3).
Subgroup analysis of medications and subsequent coronary angiography (CAG). With the exception of the thallium scan interaction, no significant subgroup interaction between medications or management was detected in the positive and negative groups compared with the inclusive group. ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blocker; CCB, calcium-channel blocker; CCTA, coronary computed tomography angiography; CI, confidence interval; DHP, dihydropyridine.
Our study results indicated that patients with a positive TMT result were prescribed treatments based on their clinical presentation and risk stratification, a practice supported by the lack of significant differences in all-cause and CV death compared with the negative group after PSM. However, standardized management is lacking for the inconclusive group, which is evident from our finding of this group’s less favorable all-cause mortality rate, even after PSM or IPTW analysis. To our knowledge, ours is the largest cohort study to compare long-term prognosis based on different TMT results.
According to the American College of Cardiology and AHA guidelines, patients with a positive TMT result have an intermediate-to-high risk of significant occlusive CAD and require follow-up confirmatory tests such as myocardial perfusion imaging.15 These patients may eventually benefit from active interventions such as cardiac catheterization.16 However, the clinical implications of an inconclusive result remain unclear; our study indicated a significantly increased risk of all-cause death for patients with an inconclusive result compared with those with a negative result.
As mentioned, a TMT result is regarded as inconclusive when the patient cannot achieve 85% of the MAPHR during the TMT. This is termed chronotropic incompetence and implies that the heart is incapable of supplying sufficient blood to the body during exercise. Therefore, this phenomenon is often regarded as a type of autonomic dysfunction.17 Consistent with our results, multiple studies have indicated that patients with chronotropic incompetence had higher all-cause mortality rates and more CV events compared with those with a normal heart rate.18,19 Analysis of our survival plots revealed that the inconclusive group had a significantly lower survival rate compared with the positive and negative groups after PSM was applied to adjust potential confounders (Figure 2A). In the inconclusive group, the HR of all-cause death increased with a decrease in functional capacity. There are 2 possible reasons for the discrepancy in all-cause death for patients with inconclusive and positive results compared with inconclusive results. First, more catheterized CAG and subsequent angioplasty were conducted for those with a positive result (Table 1). Second, in contrast to a positive result, there was no clear evidence to improve the outcomes of patients with inconclusive result.
We conducted a subgroup analysis of the inconclusive group, stratifying the patients who were unable to reach the maximum heart rate into exercise intolerance and chronotropic incompetence subgroups according to their METs. All-cause mortality was higher for those with inferior exercise capacity, which may be attributable to the presence of a severe underlying disease, such as congestive heart failure. Therefore, follow-up examinations must be performed on patients with relatively low MET levels. Additionally, the lack of a significant difference in CV-related mortality was likely because patients with inferior exercise capacity often receive timely CV drugs or subsequent examinations to prevent CV-related adverse effects.
Among all of the background characteristics, women tended to have lower all-cause mortality than men, perhaps because women monitor their health status more closely. In addition, if a patient exhibited high exercise capacity (MET >7), they were regarded as having a more favorable body condition and therefore had lower all-cause mortality than those with lower exercise capacity. Cancer and CKD are severe systematic diseases that affect multiple organs, so they had strong negative effects in terms of all-cause mortality.
The subgroup analysis revealed that in the inconclusive group, patients who were prescribed medications in roughly the TMT period or who underwent CAG or CCTA did not exhibit a significant interaction for clinical outcomes, although the trend indicated that patients who did not receive these interventions had a higher all-cause mortality rate. We should not conclude that these therapies were in vain; instead, the nonsignificant result of the interaction might be because numerous interventions were involved in the treatment of these patients, and thus no single drug could significantly reduce the HR of all-cause death. The thallium scan, which serves as a functional survey, was associated with a more favorable prognosis compared with structural surveys such as CCTA or CAG. However, this result might indicate a tendency to focus more on patients with a positive thallium scan result. Therefore, more evidence is required to validate this result.
Study LimitationsFirst, this was a retrospective study, which may have been affected by selection bias and lack certain data not yet standardized in the electronic health record format such as operation notes; however, because we applied cohort-based real-world data, the results may nevertheless accurately reflect clinical practice scenarios. Second, our data were collected over 10 years. The guideline for managing IHD may have evolved over time, which might have led to a discrepancy with current practice. Nevertheless, the comparison among the 3 groups still provided valuable information because the patients received corresponding modalities in parallel. Last, this was a cohort study of a single medical center, and the generalizability to other medical centers or hospitals requires further investigation.
Compared with patients with a negative TMT result, those with an inconclusive result had significantly higher all-cause mortality. Clinicians must not underestimate the possibility of worse outcomes in these patients and instead, should arrange more intense outpatient follow-up or further testing including stress echocardiography, nuclear myocardial perfusion imaging, CAG, and surveys for other systematic diseases.
This study was supported through grant MOST 111-2634-F-006-007 from Taiwan’s Ministry of Science and Technology and grant D111-G2512 from the Ministry of Education of Taiwan’s Higher Education SPROUT Project to Headquarters of University Advancement at National Cheng Kung University.
This study was approved by the Institutional Review Board of National Cheng Kung University Hospital (A-ER-110-095). All procedures in the study were conducted in accordance with the Declaration of Helsinki and the ethical standards of the responsible committee on human experimentation.
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-22-0225
