Patients with electrocardiographic (ECG) left ventricular hypertrophy (LVH) have repolarization abnormalities of the ST segment that may be confused with an ischemic current of injury. We analyzed the ACTIVATE-SF database, a registry of consecutive emergency department ST-segment elevation (STE) myocardial infarction diagnoses from 2 medical centers. Univariate analysis was performed to identify ECG variables associated with presence of an angiographic culprit lesion. Recursive partitioning was then applied to identify a clinical decision-making rule that maximizes sensitivity and specificity for presence of an angiographic culprit lesion. Seventy-nine patients with ECG LVH underwent emergency cardiac catheterization for primary angioplasty. Patients with a culprit lesion had greater magnitude of STE (3.0 ± 1.8 vs 1.9 ± 1.0 mm, p = 0.005), more leads with STE (3.1 ± 1.6 vs 2.0 ± 1.8 leads, p = 0.002), and a greater ratio of STE to R-S–wave magnitude (median 25% vs 9.2%, p = 0.003). Univariate application of ECG criteria had limited sensitivity and a high false-positive rate for identifying patients with an angiographic culprit lesion. In patients with anterior territory STE, using a ratio of ST segment to R-S–wave magnitude ≥25% as a diagnostic criteria for STE myocardial infarction significantly improved specificity for an angiographic culprit lesion without decreasing sensitivity (c-statistic 0.82), with a net reclassification improvement of 37%. In conclusion, application of an ST segment to R-S–wave magnitude ≥25% rule may augment current criteria for determining which patients with ECG LVH should undergo primary angioplasty.
Despite the central role of the electrocardiogram in the decision to perform primary percutaneous coronary intervention, changes in the ST segments on an electrocardiogram may lack sensitivity and specificity for an epicardial coronary occlusion. This diagnostic uncertainty is exacerbated in the presence of electrocardiographic left ventricular hypertrophy (LVH) because of repolarization abnormalities of the ST segment that may be confused with an ischemic current of injury. Accordingly, current American College of Cardiology/American Heart Association (ACC/AHA) electrocardiographic (ECG) criteria for ST-segment elevation myocardial infarction (STEMI) explicitly exclude LVH in the definition of significant ECG STEs. Because of the difficulty in characterizing significant ST-segment changes in patients with LVH, this setting is frequently associated with “false-positive” diagnoses of acute coronary syndrome and unnecessary activation of emergency cardiac care teams for thrombolysis or primary percutaneous intervention. We recently reported that ECG LVH was the most significant predictor of false-positive STEMI activations, even after multivariate adjustment. We hypothesized that additional ECG criteria for STEMI could improve overall classification by better predicting the presence of an angiographic culprit lesion in patients with ECG LVH.
Methods
We analyzed the ACTIVATE-SF database, a registry of consecutive emergency department (ED) STEMI diagnoses from 2 medical centers at University of California, San Francisco—a tertiary care university hospital and a county hospital and trauma center. At these hospitals, ED physicians make the autonomous decision to activate the emergency cardiac team for primary angioplasty. All ED physician-initiated STEMI activations from April 2008 to June 2011 were recorded in the ACTIVATE-SF registry. Among the 411 total STEMI activations during our study period, all patients who were brought to the catheterization laboratory were tracked, irrespective of angiographic findings. We identified 79 patients with ECG criteria for LVH based on the presenting electrocardiogram; these 79 patients formed the study population.
All clinical information on patient presentation was collected from the ED physician and nursing notes. The inciting ECG STEMI (i.e., the tracing that led to the decision to activate the STEMI team) was de-identified and the interpretation provided by the ECG equipment was removed. In most cases, this was the first electrocardiogram obtained within 10 minutes of arrival to the ED. The de-identified electrocardiograms were independently read for key variables by 2 cardiologists blinded to clinical outcomes and to the other cardiologist’s interpretation. In cases of disagreement on any variable beyond levels of concordance designated a priori, a third cardiologist adjudicated the ECG findings. Laboratory values and angiographic and echocardiographic data were collected from electronic medical records. Study data were collected and managed using Research Electronic Data Capture (REDCap) electronic data capture tools hosted at University of California, San Francisco.
Significant STE was defined as J-point elevation in ≥2 contiguous leads of ≥2 mm in lead V 1 , V 2 , or V 3 and ≥1 mm in other leads. ST-segment depression ≥1 mm in leads V 1 to V 3 was also considered STE. A single territory of STE was assigned for each patient based on the territory that displayed the greatest magnitude of STE: inferior territory was defined as STE in leads II, III, and aVF, anterior as STE in leads V 1 to V 3 , lateral as STE in leads V 4 to V 6 and/or I and aVL, and posterolateral as ST-segment depression in leads V 1 and V 2 . Traditional ACC/AHA criteria were used to define STEMI as ≥1-mm STE in 2 contiguous leads and 2 mm if STE was isolated to leads V 1 to V 3 .
ECG LVH was defined using the Cornell criteria, the Sokolow-Lyon criteria, or the presence of an R-wave amplitude >11 mm in lead aVL. Meeting any 1 of these criteria was considered evidence of LVH. Magnitude of STE was measured from the base of the isoelectric T-P segment. QRS height was defined as the positive deflection measured from the T-P segment to the peak of the QRS interval; in leads V 1 to V 3 QRS height was measured as the entire magnitude from peak to nadir (R-S–wave magnitude). Percent STE was defined as the ratio of STE in millimeters to QRS magnitude in millimeters in the same lead. For ease of clinical application, we chose a value of 25% as a cutoff for a significant ratio, representing the median ratio in patients with an angiographic culprit lesion and the 90th percentile for ST/R-S–wave ratio in patients without an angiographic culprit lesion.
Reciprocal ST-segment depression was defined as ST-segment depression ≥1 mm in a territory distinct from leads with STE (e.g., ST-segment depression in leads II, III, and aVF in a patient with STE in leads V 1 to V 3 ). T-wave inversions were defined as a predominately negative T-wave deflection in the lateral or anterior leads or a positive T wave with a negative terminal component. ST-segment concavity was determined by drawing a line from the J-point to the apex of the T wave. The structure was labeled concave if the ST segment subtended an area below this line and convex if the ST segment subtended an area above this line.
The main outcome measurement, an angiographic culprit lesion, was defined as the presence of a total or subtotal occlusion of a coronary artery at the time of cardiac catheterization that was consistent with causing the patient’s clinical presentation. Angiograms of each case were reviewed and scored by a cardiologist blinded to the presenting electrocardiogram.
Summary statistics for the overall sample were constructed using frequencies and proportions for categorical data and means ± SDs, medians, and interquartile ranges for continuous variables. These summary statistics were stratified by whether patients had an angiographic culprit lesion. Univariate logistic regression was used to determine the unadjusted relation (odds ratio) between predictor variables and presence of an angiographic culprit lesion. Postmodel estimation was applied to determine receiver operator curves of different diagnostic criteria for prediction of an angiographic culprit lesion. All univariate and receiver operator analyses were performed using STATA 11 (STATA Corp., College Station, Texas). A p value <0.05 was considered statistically significant. Net reclassification index was calculated as previously described.
Recursive partitioning, which creates a branching decision tree that uses predictor variables to divide subjects into groups with and without the outcome of interest, was then used to identify clinical decision-making rules for prediction of an angiographic culprit lesion in patients with LVH and chest pain. All ECG variables associated (alpha <0.20) with the primary outcome were inputted into a recursive partitioning algorithm using C4.5 software. The class variable was the presence or absence of an angiographic culprit lesion. Ninety percent of the dataset was randomly selected to be the training set, and the remaining 10% was used as a validation cohort. Maximum number of branches was set at 6 to decrease the likelihood of an overly fractured classification scheme. After a rule set was created, the program was run 10 more times, using a different randomly selected validation cohort, to confirm the integrity of the classification scheme. This final categorization recurred 82% of program runs, suggesting a strong rule set. The entire dataset was then used to determine the accuracy of the rule set. The rule was then cross validated by applying it to the entire dataset and the actual outcomes of each patient. Final sensitivity and specificity analyses reflect application of the rule to the entire dataset.
All authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the report as written.
Results
In an entire cohort of 411 STEMI activations, 79 patients (18%) had LVH on their presenting electrocardiogram. ECG LVH was significantly associated with a higher rate of false-positive activation for presumed STEMI compared to the overall cohort (72% vs 36%, p <0.001). Compared to patients with LVH and absence of a culprit lesion, patients with LVH and a culprit lesion at angiography had a higher body mass index, were less likely to have a history of heart failure, and were less likely to have a history of drug use ( Table 1 ).
| Variable | Culprit Lesion | p Value | |
|---|---|---|---|
| No | Yes | ||
| (n = 57) | (n = 22) | ||
| Age (years) | 59.2 ± 14.6 | 60.6 ± 10.5 | 0.7 |
| Race/ethnicity | |||
| White, non-Hispanic | 11 (20%) | 4 (18%) | 0.1 |
| African-American | 25 (45%) | 5 (23%) | |
| Asian | 12 (21%) | 9 (41%) | |
| White Hispanic | 2 (4%) | 3 (14%) | |
| Pacific islander | 6 (11%) | 1 (5%) | |
| Men | 41 (72%) | 15 (68%) | 0.7 |
| Body mass index (kg/m 2 ) | 24.1 ± 5.1 | 26.3 ± 4.2 | 0.07 |
| Coronary artery disease | 27 (47%) | 8 (36%) | 0.4 |
| Diabetes mellitus | 11 (19%) | 5 (23%) | 0.7 |
| Hypertension ⁎ | 34 (60%) | 16 (73%) | 0.3 |
| Dyslipidemia ⁎ | 21 (37%) | 8 (37%) | 1.0 |
| Peripheral vascular disease | 5 (9%) | 0 | 0.2 |
| Congestive heart failure | 7 (13%) | 0 | 0.08 |
| Smoker | 22 (42%) | 7 (37%) | 0.7 |
| Illicit drug use | 22 (39%) | 1 (5%) | 0.005 |
⁎ Hypertension and dyslipidemia were defined as present if they were identified as part of the medical history by the responsible emergency physician.
On univariate analysis, electrocardiograms of patients with ECG LVH and a culprit lesion showed taller STEs (3.0 ± 1.8 vs 1.9 ± 1.0 mm, p = 0.005), were more likely to have Q waves (27% vs 9%, p = 0.06), have a larger number of leads with STE (3.1 ± 1.6 vs 2.0 ± 1.8, p = 0.002), and were more likely to have reciprocal ST-segment depressions (64% vs 28%, p = 0.02). Because patients with LVH may have J-point elevation in the absence of a current of injury, we also examined the ratio of STE to R-S–wave magnitude in the lead with the greatest STE ( Table 2 ). Subjects with an angiographic culprit lesion had a greater ratio of STE to R-S–wave magnitude (median 25% vs 9.2%, p = 0.003). Characteristics of the ST segment and T wave were also predictive of a culprit lesion. Patients with an angiographic culprit lesion were less likely to have a concave ST segment (50% vs 70%, p = 0.09). Presence or absence of T-wave inversions was not predictive of an angiographic culprit lesion on univariate analysis.
| Variable | Culprit Lesion | p Value | |
|---|---|---|---|
| No | Yes | ||
| (n = 57) | (n = 22) | ||
| Criteria for left ventricular hypertrophy ⁎ | |||
| Cornell | 30 (53%) | 9 (41%) | 0.4 |
| Sokolow-Lyon | 33 (58%) | 8 (36%) | 0.09 |
| R wave >11 mm in lead aVL | 8 (14%) | 8 (36%) | 0.06 |
| Leads with ST-segment elevation | <0.001 | ||
| V 1 –V 3 | 48 (84%) | 10 (45%) | |
| II, III, aVF | 4 (7%) | 6 (27%) | |
| V 1 –V 2 (depression) | 0 | 3 (14%) | |
| V 4 –V 6 , I, aVL | 0 | 3 (14%) | |
| ST-segment elevation (mm) | 1.9 ± 1.0 | 3.0 ± 1.8 | 0.005 |
| Percent ST-segment elevation/R-wave height, median (interquartile range) † | 9.2 (4–67) | 25 (5–86) | 0.003 |
| Presence of Q waves | 5 (9%) | 6 (27%) | 0.06 |
| Number of leads with ST-segment elevation, mean ± SD | 2.0 ± 1.8 | 3.1 ± 1.6 | 0.002 |
| Presence of any reciprocal ST-segment depression | 16 (28%) | 14 (64%) | 0.02 |
| Concave ST segment (%) | 40 (70%) | 11 (50%) | 0.09 |
| Presence of T-wave inversions, lateral leads | 26 (46%) | 7 (32%) | 0.3 |
| Presence of T-wave inversions, anterior leads | 14 (25%) | 2 (9%) | 0.2 |
| Corrected QT interval (ms) | 447 ± 37.7 | 445 ± 32.7 | 0.9 |
⁎ Columns for left ventricular hypertrophic criteria exceed 100% because each patient could meet multiple criteria for left ventricular hypertrophy.
† ST-segment elevation/R-wave height is reported as percent height of ST segment compared to absolute height of R-S wave in the lead with maximal ST-segment elevation.
Application of ACC/AHA criteria to diagnosis of STEMI in patients with ECG LVH had a sensitivity of 73% and a specificity of 58% for the presence of an angiographic culprit lesion ( Table 3 ). Increasing the cutoff of “significant” STEs to 2.5 mm in leads V 1 to V 3 increased the specificity of the electrocardiogram to 74%, but sensitivity was decrease to 68%. Integrating other features including presence of Q waves or morphologic analysis of the ST segment improved specificity, but in each case significantly decreased sensitivity ( Table 3 ).
| Criteria | Sensitivity (%) | Specificity (%) | Positive Predictive Value (%) | c-Statistic |
|---|---|---|---|---|
| 1-mm ST-segment elevation in any contiguous leads | 77 | 53 | 39 | 0.65 |
| 1-mm ST-segment elevation in any contiguous leads, 2 mm if leads V 1 –V 3 | 73 | 58 | 40 | 0.65 |
| 1-mm ST-segment elevation in any contiguous leads, 2.5 mm if leads V 1 –V 3 | 68 | 74 | 50 | 0.71 |
| 1-mm ST-segment elevation in any contiguous leads, 2 mm if leads V 1 –V 3 , and ST-segment/R-wave ratio >25% for leads V 1 –V 3 | 64 | 93 | 78 | 0.78 |
| 1-mm ST-segment elevation in any contiguous leads + presence of Q waves | 18 | 96 | 67 | 0.57 |
| 1-mm ST-segment elevation in any contiguous leads + ST segment flat or convex | 50 | 86 | 58 | 0.68 |
| 1-mm ST-segment elevation in any 2 contiguous leads but 3 contiguous leads if leads V 1 –V 3 | 68 | 80 | 56 | 0.74 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
