Asymptomatic Cardiac Ischemia Pilot (ACIP) Study
Relationship Between Exercise-Induced and Ambulatory Ischemia in Patients With Stable Coronary Disease
Background We investigated whether the presence and frequency of asymptomatic ischemic episodes recorded during ambulatory ECG (AECG) monitoring could be predicted on the basis of clinical characteristics or exercise treadmill test (ETT) performance in patients with stable coronary disease and whether the estimate of ischemia severity was similar between the AECG and ETT.
Methods and Results Patients screened for the Asymptomatic Cardiac Ischemia Pilot (ACIP) study were selected for the current analysis if data were available from 48-hour AECG monitoring as well as from an ETT during which the patient developed ≥1-mm ST-segment depression. Exercise ECG data were available for 143 of the 910 patients without ischemic episodes and for 659 of the 910 patients with ischemic episodes during AECG monitoring. Angina was more frequent among patients with ambulatory ischemic episodes than among patients without such ischemia (P<.001). Patients with AECG ischemia had a consistently more marked ischemic response on the ETT than patients without AECG ischemia; patients likely to have AECG ischemia could be predicted on the basis of ETT performance characteristics. However, the correlation coefficients between the severity of ischemia estimated by ETT and by AECG were small.
Conclusions There are significant relations between ischemia detected by AECG monitoring and by ETT, but the relations are limited, indicating that the two tests are not redundant to characterize coronary patients. A larger study investigating the prognostic significance of the ischemia identified by each modality, with follow-up for clinical events, will be necessary to determine the most appropriate methods to evaluate patients with stable coronary disease.
The presence of asymptomatic ischemic episodes identified by AECG monitoring during routine daily activities in stable coronary patients is associated with an adverse cardiac outcome.1 2 3 4 5 High-risk subsets of patients with stable coronary disease can also be identified on the basis of ETT,6 yet it remains unknown whether the prognostic insight from each risk-stratifying modality is unique or redundant compared with the other. Recent studies demonstrate that the magnitude and perhaps the nature of the ischemia observed with different modalities such as ETT and AECG monitoring may be different,7 and accordingly, one would anticipate that the prognostic inferences from these test results would also be different. Exercise testing is currently a routine risk-stratifying technique, but AECG monitoring is not. Because only 20% to 50% of stable coronary patients exhibit asymptomatic ambulatory ischemic episodes,8 it would be desirable to identify clinical and ETT characteristics of the patient likely to exhibit such AECG ischemia. Efficient identification of such patients with asymptomatic ischemic episodes would optimize efficient use of medical resources because appropriate medical care would be directed to those patients who might be at high risk; such identification would also facilitate the performance of clinical trials to confirm the prognostic significance of AECG ischemia.
The ACIP study provides a unique database to compare clinical data, characteristics of AECG monitoring, and ETT performance in patients with stable coronary disease. Briefly, the major objectives of the ACIP study, a multicenter, international clinical trial funded by the NHLBI, were (1) to compare the safety and efficacy of three randomly assigned strategies to suppress ischemia and (2) to assess the feasibility of a larger trial to evaluate the effects of anti-ischemic treatment on mortality and morbidity, including ascertainment of the availability of patients, compliance with assigned therapies, and event rates at 1 year of follow-up.9
The purpose of the present database study was to determine whether the presence, frequency, and duration of asymptomatic ambulatory ischemia could be predicted on the basis of clinical characteristics or ETT performance and to determine whether the estimate of ischemia severity was similar between ETT and AECG monitoring.
Details of the ACIP study design, recruitment, and methods have been published previously.9 10 11 In brief, stable coronary patients were eligible for participation in ACIP if they had evidence of ischemia on an ETT, a 48-hour AECG recording demonstrating ≥1 episode of asymptomatic ischemia, and angiographically documented obstructive coronary disease suitable for revascularization. The ETT and AECG were performed after all anti-ischemic medication except sublingual nitroglycerin had been discontinued for ≥2 days, except in postinfarct patients and those patients who were considered to require background anti-ischemic medication for adequate control of angina, who were permitted to continue taking atenolol 50 mg/d or a slow-release preparation of diltiazem 60 mg BID. Patients were selected for screening for ACIP on the basis of the presence of any of the following: evidence of ischemia on ETT or other studies conducted for clinical purposes, angiographic evidence of coronary artery disease, or a medical history suggestive of coronary artery disease.
Forty-eight–hour AECG recordings were performed by use of Applied Cardiac Systems AM cassette recorders. Electrodes were applied to reproduce the two ECG leads showing the greatest ST-segment deviation during the ETT. Before the patient left the clinical unit, an initial AECG recording was made in the left and right lateral decubitus, standing, supine, and sitting positions to ensure that artifactual ST-segment deviation did not occur. If artifactual ST-segment deviation was observed, the patient was excluded from further investigation. The recordings were analyzed with a CardioData Mk4 playback system with modified software12 13 in the AECG core laboratory at the Brigham and Women's Hospital, Harvard Medical School. Both the technician and the physician who reviewed the tapes were blinded to the patient's study status. An ischemic episode was defined as transient ischemic ST-segment deviation ≥1.0 mm that lasted ≥1.0 minute. The onset of each episode was defined as the time when the ST segment deviated from both the isoelectric line and the baseline ST-segment value by ≥1.0 mm, and the offset was defined as that point after the onset time when the ST segment no longer deviated by >1.0 mm compared with the baseline and the isoelectric line values. The duration of an episode, therefore, was the amount of time during which the ST-segment values deviated by ≥1.0 mm compared with the baseline and the isoelectric line values. After an episode of ischemia, the ST segment must have returned to the baseline value that preceded the episode of ischemia for ≥5.0 minutes before new ST-segment deviation was considered to constitute a new episode of ischemia. Patients were instructed to press the event marker at the onset of angina. Ischemic episodes accompanied by an event-marker notation were considered symptomatic and those not accompanied by an event-marker notation were considered asymptomatic. If the event marker was pressed without concomitant evidence of ischemia present on the AECG, ischemia was not considered to be present. For those patients who had <48 hours of AECG monitoring, the number of ischemic episodes and cumulative duration of ischemia were adjusted to conform to a standard 48-hour monitoring period.
Duplicate blinded readings of 146 pairs of 48-hour AECG recordings were performed for quality control. There was agreement for the presence of ischemia between the two readings in 144 (99%) of the 146 readings. The correlation coefficients and 99% CIs concerning the AECG variables of interest for the two blinded readings were as follows: number of ischemic episodes, r=.94 (.91, .96); minutes of ischemia, r=.79 (.69, .86); maximum ST-segment depression, r=.86 (.79, .91); baseline heart rate, r=.86 (.79, .91); heart rate 5 minutes before onset of ischemia, r=.82 (.74, .88); heart rate at onset of ischemia, r=.89 (.84, .93); maximum heart rate per ischemic episode, r=.92 (.88, .95); and mean heart rate per 24-hour period, r=.98 (.97, .99) (each P=.0001).
ETTs were performed according to the ACIP protocol,14 which is a linear protocol with two 1-minute warm-up stages followed by 2-minute stages at 3 mph designed to gradually increase workload by 1.5-MET increments. A modification of the protocol designed for elderly or short individuals permitted a walking speed of 2 mph with a steeper incremental increase in grade that resulted in similar estimated METs per unit of time.14 The ETTs were symptom limited, with standard criteria used for stopping, including serious dysrhythmias, severe angina pectoris, a drop in systolic blood pressure, or central nervous system complaints. The 12-lead torso ECG was acquired in the standing position before exercise, at each minute during exercise, at peak exertion, and at each minute after exertion or until normalization of exercise-induced ST-segment changes occurred with the patient in the sitting position for 5 minutes. Heart rate and blood pressure were recorded during each exercise stage and for each minute of recovery. Time to onset of angina, total exercise duration, rate of perceived exertion, and reasons for stopping exercise were recorded.
Exercise ECGs were processed at the ACIP ECG core laboratory at St Louis University with the use of an off-line, quantitative, computerized exercise ECG analysis program, as previously described.15 16 Maximum depth of ST-segment depression during exercise or after exercise was measured by use of a high-resolution digitizing board in all leads with ≥0.5 mm of ST-segment deviation and compared with the baseline. An abnormal exercise ECG lead was defined as follows: (1) J-point and ST 80 depression ≥1.0 mm and ST segment horizontal or downsloping <1 mV/s or (2) ST 80 depression ≥1.5 mm and ST segment upsloping >1 mV/s compared with the rest tracing. The following exercise ECG variables were determined: (1) maximum depth of ST-segment depression in any lead; (2) number of abnormal exercise ECG leads; and (3) ECG location of ST-segment depression. Time to onset and offset of ischemic ST-segment depression was determined when any lead (excluding aVR) met the criteria for abnormality. Heart rate and systolic blood pressure were recorded at baseline, time to onset of ischemic ST-segment depression, time to onset of angina, and peak exertion. Peak exercise time was recorded, and the reasons for stopping exercise were noted.
Duplicate blinded readings of 89 pairs of ETT ECG recordings were performed for quality control. In 88 (99%) of 89 tests, there was agreement between duplicate readings for the presence or absence of ischemia. The correlation coefficients and 99% CIs of interest for the two blinded readings were as follows: maximum ST-segment depression, r=.97 (.95, .98); sum of ST-segment depression in abnormal leads, r=.99 (.98, .99); total number of abnormal leads, r=.96 (.93, .98); exercise time to onset of ischemia, r=.95 (.91, .97); offset time after ischemia, r=.85 (.75, .91); sum of ST-segment depression in lateral leads, r=.95 (.92, .97); sum of ST-segment depression in inferior leads, r=.97 (.95, .98); and sum of ST-segment depression in anterior leads, r=.99 (.98, .99) (each P=.0001).
Reproducibility of blinded duplicate readings of ETT ECGs and AECGs was assessed by use of Pearson correlation coefficients; CIs for these correlations were calculated by Fisher's z transformation.17 To adjust for variations in the number of hours recorded for each patient, the number of ischemic episodes was divided by the duration of the AECG and expressed in units of ischemic episodes per 48 hours. Patients were then grouped into ordered ischemia categories based on this outcome (0 episodes/48 hours, 0.1-1.9 episodes/48 hours, etc). Means of continuous measurements and percentages of categorical measurements were calculated within each of these ischemia categories. Differences in means or proportions by these ischemia categories were tested with ANOVA for continuous measurements and the Kruskal-Wallis test for ordered categorical data.18
Exercise ECG data were available for a smaller percentage of patients (143/910) screened for ACIP without AECG ischemia than for patients with AECG ischemia (659/910); accordingly, the average number of episodes and prevalence of AECG ischemia among patients with particular exercise ECG findings does not represent the corresponding values to be expected among all patients with similar exercise test findings. Thus, means and percentages for exercise ECG findings are presented by the presence of AECG ischemia or by the number of AECG ischemic episodes. The OR as a measure of association between exercise-induced ischemia and AECG findings is not affected by this unequal sampling.
Backward stepwise logistic regression,19 with P≤.01 used as a criterion to remain in the model, was used to select a limited set of exercise test outcomes that predicted the presence of ischemia during AECG monitoring. The model selected was used to calculate predicted ORs for the presence of ischemia. Because the results of stepwise variable selection procedures among correlated predictors may depend on chance features of the data, we performed 10 replications of the selection process using data sets of the same size as the original data constructed by random sampling with replacement from the original data.20
A nonparametric regression technique (robust locally weighted regression, or LOWESS21 ) was used to draw a line estimating the mean number of episodes per 48-hours as a function of each ETT outcome. The LOWESS technique does not assume a particular form (eg, a straight line) for this relation and avoids predicting negative numbers of episodes at certain values of ETT outcomes, as could be the case if ordinary linear regression were used (eg, with minutes to onset). Spearman's correlation coefficient (r) is displayed on each figure together with a probability value for a linear trend in number of ischemic episodes according to ETT outcome.
To take into account the multiple hypotheses tested in secondary analyses such as those in the present report, the ACIP protocol specifies that in such analyses, P<.01 shall be considered evidence of differences and P<.001 strong evidence.
A total of 802 participants who underwent screening, performed an ACIP protocol ETT that indicated the presence of ischemia, and underwent AECG monitoring are included in this database study (Fig 1⇓). As previously reported,24 patients from one ACIP clinical unit have been excluded from all analyses, including 60 of 618 patients enrolled in ACIP and an additional 89 patients screened with AECG monitoring but not enrolled. Of patients with exercise ECG findings, 143 had no episodes of ischemia during AECG monitoring and 659 had ≥1 episode of AECG ischemia. Among the 659 patients with AECG ischemia, 69 (10%) had at least one symptomatic episode during the AECG recording; 2932 (97%) of the 3033 ischemic episodes were asymptomatic. Of the 802 patients screened for ACIP and included in this database study, 57 (7%) received β-blocker therapy within 48 hours and 45 (6%) received calcium channel blocker therapy within 24 hours of ETT and AECG.
Angina was more frequent among patients with AECG ischemia than among those without AECG ischemia (68% versus 55%; P=.004), and the prevalence of angina increased with the number of AECG ischemic episodes observed: 55% among those without AECG ischemia, 61% among those with 1.0 to 1.9 ischemic episodes per 48 hours, 66% among those with 2.0 to 2.9 episodes per 48 hours, 63% among those with 3.0 to 4.9 episodes per 48 hours, and 76% among those with ≥5 episodes per 48 hours (P<.001). There were no other significant differences between patients with ≥1 episode of ambulatory ischemia and those without ischemic episodes nor were there significant differences according to the frequency of ischemic episodes with respect to sex, age, prior coronary bypass grafting, or resting heart rate and blood pressure.
Relations Between Exercise Treadmill Outcomes and Ischemia During AECG Monitoring
Patients with AECG ischemia had a more marked ischemic response during the ETT than patients without AECG ischemia, and the degree of ischemia manifested by the ETT increased with the frequency of AECG ischemia (Table⇓s 1 and 2⇓). Patients with AECG ischemia more frequently displayed ST-segment depression in the anterior or lateral leads but less often displayed ST-segment depression in the inferior leads unaccompanied by anterior lead depression (P<.001). Compared with patients without AECG ischemia, those with AECG ischemia had a greater number of leads with ST-segment depression, an earlier onset of ST-segment depression with a later resolution of ST-segment changes after exercise, and a greater depth of ST-segment depression, and they developed ST-segment depression at a lower heart rate than did patients without AECG ischemia (P<.001 for each comparison).
Using backward stepwise logistic regression, we found three ETT variables to be associated with the presence of AECG ischemia: maximum ST-segment depression (P=.003), heart rate at onset of ST-segment depression (P<.001), and time to ST-segment depression (P<.001). The odds for presence of AECG ischemia increased by a factor of 1.5 (99% CI, 1.06 to 2.18; P=.003) with each millimeter of ST-segment depression, decreased by a factor of 0.8 (99% CI, 0.71 to 0.95; P<.001) for every 10 bpm in heart rate at onset, and decreased by a factor of 0.8 (99% CI, 0.75 to 0.93; P<.001) with each minute to onset. Angina on the ETT was the last variable eliminated in stepwise selection. After adjustment for the other three predictors, angina was associated with a relative increase in the odds of AECG ischemia of 1.7 (P=.02; 99% CI, 0.96 to 2.88). Fig 2⇓ shows the change in the relative odds (OR) of AECG ischemia as the three selected predictors varied. The ORs displayed in Fig 2⇓ were calculated relative to patients with 1 mm of ST-segment depression after 1 minute of exercise at a heart rate of 110 bpm. Other variables considered for this model included the number of leads with ST-segment depression, systolic blood pressure at onset of ST-segment depression, total duration of exercise, minutes after exercise to offset of ST-segment depression, and location of ST-segment depression in the inferior leads.
It has been shown (eg, by randomly dividing data sets into two subsamples20 ) that the models selected by stepwise regression procedure can vary considerably depending on the correlations among the predictors considered and the presence of influential observations. To assess whether the choice of ETT predictors of AECG ischemia from the ACIP data was affected by such factors, 10 replications were performed with the use of data sets constructed by random sampling with replacement from the original data. Time to ST-segment depression was chosen in all replications; heart rate at onset of ST-segment depression and maximum depth of ST-segment depression were chosen in 7 of 10 replications, with all three of these predictors and no other variables being selected in 4 of 10 replications. Angina during exercise was selected in 4 of 10 replications, with no other variable being selected more than twice.
Correlation Between Severity of Ischemia on the ETT and Frequency and Duration of Ambulatory Ischemia
There were also significant relations between the frequency and duration of ischemic episodes during AECG monitoring and the severity of the ischemic response during ETT (Table 3⇓; Fig 3⇓). Spearman's rank correlation coefficients were calculated between ETT outcomes and both the number of ischemic episodes per 48 hours and the total minutes of ischemia per 48 hours during AECG monitoring. There were small but statistically significant correlations between the two AECG outcomes and the number of leads with ≥1-mm ST-segment depression, minutes of exercise until the onset of ≥1-mm ST-segment depression, duration of ≥1-mm ST-segment depression during recovery, maximum depth of ST-segment depression, heart rate, systolic blood pressure, and rate-pressure product at onset of ≥1-mm ST-segment depression.
Because the majority of studies3 4 5 indicate that the presence of asymptomatic cardiac ischemic episodes identified by AECG monitoring is independently associated with an adverse cardiac outcome, it would be important to identify these patients who are considered at high risk. However, not all patients with stable coronary disease exhibit these asymptomatic manifestations,8 and it may be inefficient to monitor all stable coronary patients to search for those with AECG ischemia. Our results indicate that certain clinical characteristics may help to identify stable coronary patients with AECG ischemia but that ETT variables are particularly useful. Compared with patients without AECG ischemia, patients with AECG ischemia exhibit a lower heart rate at onset of exercise-induced ST-segment depression, a greater number of leads with ST-segment depression, a shorter exercise time to onset of ST-segment depression, a longer recovery time to offset of ST-segment depression, and more marked ST-segment depression. However, the correlation between severity of the ischemia by AECG measures and by ETT measures is weak, which suggests that the severity of ischemia is reflected differently in each test or that different mechanisms may be responsible for the manifestations of ischemia detected by the two different methods.
Relation Between Clinical Characteristics and the Presence and Frequency of Asymptomatic Ambulatory Ischemic Episodes
Many previous studies have attempted to identify clinical characteristics, such as the nature of the clinical presentation, presence of prior coronary manifestations, risk-factor profile, age, sex, and use of anti-ischemic medication, that could distinguish stable coronary patients who manifest episodes of asymptomatic AECG ischemia during routine daily activities from those who do not exhibit such episodes. Although we observed that the presence of symptomatic ischemia (ie, angina) at the time of the screening tests was associated with an increased incidence of asymptomatic AECG ischemia, none of the prior studies25 26 found any characteristic that could identify the patient with asymptomatic ischemic episodes. The differences in results between the present study and prior studies probably are due to our substantially larger sample size. Patients included in this ACIP database study were selected on the basis of ischemia on the exercise ECG; this selection could potentially account for differences in associations between clinical characteristics and presence of ischemia on the AECG reported here and in results from other studies. Although our observed differences in clinical characteristics are statistically significant, the absolute differences are quite small. Thus, the clinical value of using these characteristics to distinguish patients with versus those without AECG ischemia remains slight.
Relation Between ETT Performance and the Presence and Frequency of Asymptomatic Ambulatory Ischemic Episodes
It is known that patients with asymptomatic AECG ischemia generally have a more positive ETT response than patients without such ischemia.26 27 28 29 30 The most consistent association has been an inverse correlation between the number of ischemic episodes recorded during AECG monitoring and the exercise time to 1.0-mm ischemic ST-segment depression.26 27 28 29 Some investigators have also found that asymptomatic ischemia is more likely to occur in patients with exercise-induced ischemia occurring at a low peak heart rate,26 28 a low peak double product,26 or a more prolonged period of ST-segment depression in the recovery phase after exercise.28 Still other investigators have found little or no correlation between exercise test characteristics and the presence or absence of ambulatory ischemic episodes31 nor any correlation between the severity of exercise-induced ischemia and the severity of AECG ischemia.7 30 31 Our results indicate that there is a significant, consistent, and direct relation between indexes of ischemia by exercise testing (number of leads with ST-segment depression, exercise time to onset of ST-segment depression, persistence of ST-segment depression in the recovery phase, and maximum depth of ST-segment depression) and the presence and frequency of asymptomatic AECG ischemia (Table⇑s 1 and 2).
These relations also lend themselves to the use of a logistic regression model to estimate the odds that a patient will exhibit AECG ischemia. These calculations are based on the three ETT variables that were associated with the presence of AECG ischemia: exercise time to ischemia, heart rate at onset of ischemia, and maximum depth of ST-segment depression. The OR calculated from these three variables (Fig 2⇑) may be very useful to identify the patient likely to exhibit AECG ischemia who may be at high risk of an adverse outcome or who may need to be evaluated in further clinical investigation.
In a recently reported small study32 of 48 patients with a positive ETT (26 with and 22 without AECG ischemia), angina during exercise, total exercise time, time to onset of ≥1-mm ST-segment depression, and maximum ST-segment depression were found to be significant predictors of AECG ischemia, with time to onset of ≥1-mm ST-segment depression and maximum ST-segment depression selected as significant predictors in stepwise logistic regression. Those investigators reported no significant difference in heart rate or systolic blood pressure at onset of ≥1-mm ST-segment depression, variables that were significant univariate predictors of AECG ischemia among patients screened for ACIP. Possible reasons for the reported differences in the identified predictors of ischemia and the results of the current study include differences between the ETT protocols used in the two studies (Bruce versus ACIP, the latter of which uses a more gradual increase in workload), the larger sample size and consequently greater power to detect associations in the ACIP study, and chance variation in the outcome of stepwise selection procedures, as was seen in the bootstrap replications performed with the ACIP data.
Although there is an important concordance between the ETT and the AECG for the presence of ischemia, we have also demonstrated that the correlation between indexes of the severity of ischemia during the ETT and during AECG monitoring is nevertheless small. Borzak et al7 actually observed that there was no correlation between the number of ischemic episodes during AECG monitoring and the exercise time to ischemic ST-segment depression. Thus, although the ETT is useful to suggest which patient is likely to manifest AECG ischemic episodes, the frequency of ischemic episodes identified by AECG monitoring is not well reflected by the severity of ischemia gauged by ETT performance variables. Because the prognosis of stable coronary patients may depend on both the presence and duration of asymptomatic ambulatory ischemia, the ETT alone may not be sufficient to estimate a patient's prognosis; AECG monitoring may be necessary to provide additional prognostic insight.
Limitations of the present study include the sampling procedure. The patients included are a subset of the patients screened for enrollment in the ACIP Study for whom both ETT and AECG data were available. The number of patients with and without AECG ischemia in the current data set does not reflect the overall proportions with and without AECG ischemia in the populations of patients with positive ETTs screened for this study. Thus, these data cannot be used to obtain estimates of the absolute probability of AECG ischemia associated with different ETT findings. However, the relative increase in the probability of AECG ischemia, as measured by an OR, can be estimated from the current data. All patients selected for this substudy had ≥1-mm ST-segment depression on the ETT. Estimates of associations between AECG and ETT outcomes in this study may differ from those found in a population not selected for a positive ETT.
Significant and separate relations have been observed between cardiac outcome and the presence of ischemia identified by ETT and by AECG monitoring. Although each risk-stratifying modality evaluates manifestations of coronary disease, it has not been known whether these two modalities reflect similar manifestations of coronary disease or to what extent they are independent or redundant in providing important prognostic information. Our results indicate that there are statistically significant relations between AECG monitoring and exercise testing to detect manifestations of coronary disease and that exercise test performance variables can predict which patient will likely exhibit AECG ischemia. However, the magnitude of ischemia identified by one modality does not appear to be a substitute for the magnitude of ischemia identified by the other. These observations suggest that the mechanism(s) responsible for ischemia identified by the two tests may be different.33 A larger study with follow-up for clinical events is required to investigate the relative prognostic importance of the coronary disease manifestations detected by each modality and their use in the monitoring of such patients.
Selected Abbreviations and Acronyms
|ACIP||=||Asymptomatic Cardiac Ischemia Pilot|
|bpm||=||beats per minute|
|ETT||=||exercise treadmill test|
This study was funded by the National Heart, Lung, and Blood Institute, Cardiac Diseases Branch, Division of Heart and Vascular Diseases, Bethesda, Md, by research contracts HV-90-07, HV-90-08, and HV-91-05 through HV-91-14. Study medications and placebo were donated by ICI Pharmaceuticals Group (Zeneca Pharma after April 1993), Marion-Merrell Dow, and Pfizer. Support for ECG data collection was provided in part by Applied Cardiac Systems, Marquette Electronics, Mortara Instrument Inc, and Quinton Instruments. Some units received partial support from General Clinical Research Center grants. We are grateful to John Loring for help in the preparation of the manuscript.
- Received November 15, 1995.
- Revision received April 9, 1996.
- Accepted April 16, 1996.
- Copyright © 1996 by American Heart Association
Rocco MB, Nabel EG, Campbell S, Goldman L, Barry J, Mead K, Selwyn AP. Prognostic importance of myocardial ischemia detected by ambulatory monitoring in patients with coronary disease. Circulation. 1988;78:877-884.
Deedwania PC, Carbajal EV. Silent ischemia during daily life is an independent predictor of mortality in stable angina. Circulation. 1990;81:748-756.
Yeung AC, Barry J, Orav J, Bonassin E, Raby KE, Selwyn AP. Effects of asymptomatic ischemia on long-term prognosis in chronic stable coronary disease. Circulation. 1991;83:1598-1604.
Pepine CJ, Cohn PF, Deedwania PC, Gibson RS, Hanberg E, Hill JA, Miller E, Marks RG, Thadani U. Effects of treatment on outcome in mildly symptomatic patients with ischemia during daily life: the Atenolol Silent Ischemia Study (ASIST). Circulation. 1994;90:762-768.
Chaitman B. Exercise stress testing. In: Braunwald E, ed. Heart Disease: A Textbook of Cardiovascular Medicine. 4th ed. Philadelphia, Pa: WB Saunders Co; 1992:161-179.
Borzak S, Fenton T, Glasser SP, Shook TL, MacCallum G, Young PM, Stone PH, for the Angina and Silent Ischemia Study Group (ASIS). Discordance between effects of anti-ischemic therapy on ambulatory ischemia, exercise performance, and anginal symptoms in patients with stable angina pectoris. J Am Coll Cardiol. 1993;21:1605-1611.
Pepine CJ. Is silent ischemia a treatable risk factor in patients with angina pectoris? Circulation. 1990;82(suppl II):II-135-II-142.
Pepine CJ, Geller NL, Knatterud GL, Bourassa MG, Chaitman BR, Davies RF, Day P, Deanfield JE, Goldberg AD, McMahon RP, Mueller H, Ouyang P, Pratt C, Proschan M, Rogers WJ, Selwyn AP, Sharaf B, Sopko G, Stone PH, Conti CR, for the ACIP Investigators. The Asymptomatic Cardiac Ischemia Pilot Study (ACIP): design of a randomized clinical trial, baseline data and implications for a long-term outcome trial. J Am Coll Cardiol. 1994;24:1-10.
Knatterud GL, Bourassa MG, Pepine CJ, Geller NL, Sopko G, Chaitman BR, Pratt C, Stone PH, Davies RF, Rogers WJ, Deanfield JE, Goldberg AD, Ouyang P, Mueller H, Sharaf B, Day P, Selwyn AP, Conti CR, for the ACIP Investigators. Effects of treatment strategies to suppress ischemia in coronary artery disease patients: 12-week results of the Asymptomatic Cardiac Ischemia Pilot (ACIP) study. J Am Coll Cardiol. 1994;24:11-20.
Shook TL, Valvo V, Hubelbank M, Feldman CL, Stone PH. Validation of a new algorithm for detection and quantification of ischemic ST segment changes during ambulatory electrocardiography. IEEE Proc Comput Cardiol. 1987;57-62.
Caralis DG, Wiens G, Shaw L, Younis LT, Haueisen ME, Wiens RD, Chaitman BR. An off-line digital system for reproducible interpretation of the exercise ECG. J Electrocardiol. 1990;23:1-7.
Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods. 2nd ed. Boston, Mass: PWS-Kent Publishing Co; 1988:88-91.
Conover WJ. Practical Nonparametric Statistics. 2nd ed. New York, NY: John Wiley & Sons; 1980:229-237.
Kleinbaum DG, Kupper LL, Morganstern H. Epidemiologic Research: Principles and Quantitative Methods. New York, NY: Van Nostrand Reinhold & Co; 1982:419-444.
Cleveland WS. The Elements of Graphing Data. Monterey, Calif: Wadsworth Advanced Books and Software; 1985:167-178.
SAS/STAT User's Guide, Version 6. 4th ed. Cary, NC: SAS Institute, Inc; 1990;1:851-875 and 1990;2:1071-1126.
AXUM: Technical Graphics and Data Analysis—Reference Manual. 3rd ed. Seattle, Wash: Trimetrix, Inc; 1993:317-319.
Conti CR, Knatterud GL, Sopko G. Correction. J Am Coll Cardiol. 1995;26:842.
Quyyumi AA, Mockns L, Wright C, Fox KM. Morphology of ambulatory ST-segment changes in patients with varying severity of coronary artery disease: investigation of the frequency of nocturnal ischemia and coronary spasm. Br Heart J. 1985;53:186-193.
Campbell S, Barry J, Rocco MB, Nabel EG, Mead-Walters K, Rebecca GS, Selwyn AP. Features of the exercise test that reflect the activity of ischemic heart disease out of hospital. Circulation. 1986;74:72-80.
Mody FV, Nademanee K, Intarachot V, Josephson MA, Robertson HA, Singh BN. Severity of silent ischemia on ambulatory electrocardiographic monitoring in patients with stable angina pectoris: relation to prognostic determinants during exercise stress testing and coronary angiography. J Am Coll Cardiol. 1988;12:1169-1176.
Stone PH. Unraveling the mechanism of ambulatory ischemia: how and why. Circulation. 1990;82:1528-1530.