Author + information
- Received December 21, 1998
- Revision received March 25, 1999
- Accepted May 16, 1999
- Published online September 1, 1999.
- Scott A McHam, DOa,
- Thomas H Marwick, MD, PhD, FACCa,
- Fredric J Pashkow, MD, FACCa and
- Michael S Lauer, MD, FACCa,* ()
- ↵*Reprint requests and correspondence: Michael S. Lauer, Section of Heart Failure and Cardiac Transplantation Medicine, Department of Cardiology, Desk F-25, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio 44195.
This study was performed to determine whether a delayed decline in systolic blood pressure (SBP) after graded exercise is an independent correlate of angiographic coronary disease.
The predictive importance of the rate of SBP decline after exercise relative to blood pressure changes during exercise has not been well explored.
Among adults who underwent symptom-limited exercise treadmill testing and who underwent coronary angiography within 90 days, a delayed decline in SBP during recovery was defined as a ratio of SBPs at 3 min of recovery to SBP at 1 min of recovery >1.0. Severe angiographic coronary artery disease was defined as left main disease, three-vessel disease or two-vessel disease with involvement of the proximal left anterior descending artery.
There were 493 subjects eligible for analyses (age 59 ± 11 years, 78% male). Severe angiographic coronary disease was noted in 102 (21%). There were associations noted between a delayed decline in SBP during recovery and severe angiographic coronary disease (34% vs. 17%, odds ratio [OR] 2.59, confidence interval [CI] 1.58 to 4.25, p = 0.001). In multivariate logistic regression analyses adjusting for SBP changes during exercise and other potential confounders, a delayed decline in SBP during recovery remained predictive of severe angiographic coronary disease (adjusted OR 2.22, 95% CI 1.27 to 3.87, p = 0.005).
A delayed decline in SBP during recovery is associated with a greater likelihood of severe angiographic coronary disease even after accounting for the change in SBP during exercise.
Since Amon et al. (1)first reported that a delay in the decline of systolic blood pressure (SBP) after exercise was more accurate than ST segment depression for the diagnosis of coronary artery disease (CAD), the ratio of postexercise SBP to peak exercise SBP has received considerable attention (2–7). Previous studies have been limited, however, by relatively small sample sizes and have not systematically compared the predictive properties of recovery and exercise blood pressures independently and in combination. Additionally, SBP recordings during exercise may be inaccurate, with an error as much as 40 mm Hg at peak exercise (8). In this study, we sought to examine the association of the rate of blood pressure decline during recovery with angiographic coronary disease in a large sample of patients undergoing testing at a single center, while also considering the predictive impact of the increase of SBP during exercise.
Study population samples
Between September 1990 and December 1993, 9,608 consecutive adults underwent symptom-limited treadmill testing at the Cleveland Clinic Foundation. Ruling out CAD, following up patients with known CAD and preoperative evaluation were the most common indications for exercise testing. To be eligible for this study, patients had to have undergone coronary angiography within 90 days of exercise testing and had to have no history of prior invasive cardiac procedures, congestive heart failure, cardiomyopathy, valvular heart disease, congenital heart disease or preexcitation syndrome. An additional exclusion criteria was failure to augment SBP by at least 10 mm Hg above their resting standing blood pressure; this was designed to limit selection bias related to a greater likelihood of severe CAD associated with a hypotensive or markedly blunted blood pressure response to exercise. None of the patients excluded for this reason had a normal ST segment response. All subjects gave informed consent before treadmill testing.
Before treadmill testing, structured interviews and chart reviews yielded data on symptoms, medications, coronary risk factors, previous cardiac events and a number of cardiac and noncardiac diagnoses (9). Patients with a resting SBP >140 mm Hg or a resting diastolic blood pressure >90 mm Hg, or treatment with antihypertensive medications, were defined as having resting hypertension (10). Vital signs including weight were obtained by direct measurement. All these data, as well as stress test data, were entered prospectively on-line into a computerized database.
The Bruce or modified Bruce standard protocols were used for symptom-limited treadmill testing during which leaning on handrails was explicitly not allowed (11,12). With each stage of exercise and recovery (first 3 min), data on symptoms, rhythm, heart rate, blood pressure (by indirect arm-cuff sphygmomanometry), workload in metabolic equivalents (METs) and ST segment changes were collected and entered on-line. Blood pressures were measured by mercury column sphygmomanometry with values recorded to the nearest millimeter. Because each patient had only one exercise test, reproducibility could not be formally tested. Exercise capacity in METs was estimated from standard published tables (12). Patients were credited with the appropriate time-based proportion of the estimated METs, if the final stage of exercise was not completed. Per laboratory protocol, exercise could be stopped for marked (>2.5 mm) ST segment depression, exercise SBP >250 mm Hg and ventricular tachycardia.
Coronary angiograms were interpreted semiquantatively without exercise hemodynamic data available to those reading the coronary angiograms. Any coronary disease was defined as >50% diameter stenosis in a proximal or middle coronary artery or major branch. Severe coronary disease was defined as: 1) >50% diameter stenosis of the left main coronary artery; 2) three-vessel disease with >70% diameter stenosis in each major coronary artery system; or 3) two-vessel disease with >70% diameter stenosis of the proximal left anterior descending coronary artery.
The decline in SBP during recovery was assessed by calculating the ratio of SBPs at 3 min of recovery to peak exercise (D/B in Fig. 1): a value above the 75th percentile for the population (which corresponded to a ratio >0.95) was considered abnormal. The increase in exercise blood pressure was considered abnormally low if the ratio of peak SBP to resting SBP (B/A in Fig. 1) fell beneath the 25th percentile for the population (which corresponded to a ratio <1.22). Odds ratios (ORs) and Cochran-Mantel Haenszel confidence intervals (CIs) were calculated relating abnormal recovery and exercise blood pressure values to the presence of any or severe angiographic coronary disease. Sensitivities, specificities and positive and negative predictive values were calculated using standard definitions. Stratified analyses according to beta-blocker usage were performed, with potential interactions tested using the Breslow-Day test for heterogeneity of ORs.
It has been noted that blood pressure measurement obtained by indirect sphygmomanometry during exercise may be subject to a high degree of error, even as much as 40 mm Hg at peak exercise (8). Therefore, we also assessed the decline in SBP by calculating the ratio of SBP at 3 min of recovery to systolic blood pressure at 1 min of recovery (D/C in Fig. 1): a value above the 75th percentile for the population (which corresponded to a ratio >1.0) was considered abnormal. This measure would have the theoretical advantage of considering only “resting” blood pressures; that is, no measure is taken during exercise.
Multivariate logistic regression analyses were used to assess the associations of recovery and exercise blood pressure variables to severity of angiographic coronary disease after adjusting for each other and for potential confounders. These included age, gender, weight, chest pain history, history of known coronary disease, resting SBP, smoking, diabetes, use of lipid lowering medications, use of cardiovascular medications including beta-blockers and exercise capacity in METs.
All analyses were performed using Version 6.12 of the SAS statistical package (13); p < 0.05 was considered significant.
Baseline and exercise characteristics
There were 493 patients eligible for analyses. The most common reasons for terminating exercise were fatigue (81%) and test-terminating angina (13%), with only 6% of tests stopped due to an exercise SBP >250 mm Hg. The median value for the ratio of SBPs at 3 min of recovery to peak exercise was 0.87 (25th to 75th percentiles, 0.79 to 0.95). The median values for the ratio of SBPs at 3 min of recovery to 1 min of recovery was 0.91 (25th to 75th percentiles, 0.85 to 1.00) and for the ratio of peak exercise SBP to resting SBP was 1.37 (25th to 75th percentiles, 1.22 to 1.51).
In Table 1, the patient population was divided into two groups according to the ratio of SBP at 3 min of recovery compared with SBP at peak exercise: <0.95 (n = 371) and >0.95 (n = 122). Subjects with a recovery ratio >0.95 were older, were less likely to be female, had higher resting systolic and diastolic blood pressures, were more likely to be taking antihypertensive medications and were more likely to have diabetes. Exercise characteristics of subjects with a recovery ratio >0.95 revealed a lower maximum SBP accompanied by a smaller change in SBP during exercise as well as more frequent ischemic ST segment changes (Table 2).
Angiographic CAD and exercise and recovery SBPs
Subjects were divided into three groups by angiographic findings of no CAD, any CAD and severe CAD, with the number of subjects being 194 (39%), 197 (40%) and 102 (21%), respectively. As shown in Figure 1, compared with patients with no coronary disease and any coronary disease, patients with severe coronary disease had higher resting SBP, lower peak exercise SBP and a slower decline of SBP during the first 3 min of recovery.
Angiographic CAD and SBP recovery ratio
Subjects with a high ratio of SBPs at 3 min of recovery to peak exercise were more likely to have any CAD, as well as severe CAD (Table 3). The sensitivity, specificity, positive predictive value and negative predictive value for any CAD and severe CAD were 29%, 82%, 72% and 43%, and 36%, 78%, 30% and 82%, respectively. Similarly, an abnormally high ratio of SBP at 3 min of recovery to 1 min of recovery was also associated with any CAD and severe CAD (Table 3). The sensitivity, specificity, positive predictive value and negative predictive value for any CAD and severe CAD were 29%, 86%, 77% and 43%, and 38%, 81%, 34% and 83%, respectively.
Angiographic CAD and change of blood pressure during exercise
Subjects who had an attenuated blood pressure response to exercise (ratio of peak to resting SBP <1.22) were more likely to have any, as well as severe, CAD (Table 3). The sensitivity, specificity, positive predictive value, negative predictive value for any CAD and severe CAD were 31%, 85%, 76% and 44%, and 41%, 79%, 34% and 84%, respectively.
When patients were stratified according to beta-blocker usage, no interactions were noted regarding the ability of exercise or recovery blood pressure patterns to predict any or severe angiographic CAD (all Breslow-Day p > 0.10).
The associations of angiographic CAD with the ratio of SBPs at 3 min of recovery to peak exercise and SBPs at 3 min of recovery to 1 min of recovery were stratified according to exercise SBP response (Fig. 2). After this stratification, an abnormal 3- to 1-min systolic blood pressure ratio still predicted any CAD (adjusted OR 2.20, CI 1.31 to 3.68, p = 0.002) and severe CAD (adjusted OR 2.2, CI 1.31 to 3.7, p = 0.002). Similarly, after adjusting for the ratio of SBPs at 3 min of recovery to 1 min of recovery, an abnormally low ratio of SBP at peak exercise compared with rest was also predictive of any CAD (adjusted OR 2.01, CI 1.24 to 3.27, p = 0.005) and severe CAD (adjusted OR 2.12, CI 1.29 to 3.5, p = 0.003). As shown in Figure 2, recovery and exercise SBP changes provided additive predictive value. Breslow-Day analyses showed no interactions between exercise and recovery SBPs for predicting any or severe CAD.
After adjusting for exercise SBP increase, an abnormal ratio of SBPs at 3 min of recovery to peak exercise only tended to predict any CAD (adjusted OR 1.58, CI 0.99 to 2.52, p = 0.05) and severe CAD (adjusted OR 1.58, CI 0.97 to 2.56, p = 0.08).
After adjusting for age, gender, resting SBP and other potential confounders using multivariable logistic regression (Table 4), the presence of any CAD was predicted by the ratio of SBP at 3 to 1 min of recovery, but not by the change in SBP during exercise. Both recovery and exercise SBP measures were independently predictive of severe CAD.
In a separate set of logistic models, exercise and recovery blood pressure ratios were included together, along with potential confounders (Table 5). The ratio of SBPs at 3 min of recovery to 1 min of recovery remained independently predictive of any and severe CAD; on the other hand, the ratio SBPs at 3 min of recovery to peak exercise was no longer predictive.
In a population sample of adults referred for graded treadmill exercise testing and coronary angiography, a delayed decline in SBP after exercise was independently predictive of any and severe CAD, even after taking into account the increase in blood pressure during exercise. Two methods were used to assess the decline in blood pressure during recovery: the ratio of SBPs at 3 min of recovery to peak exercise and the ratio of SBPs at 3 min of recovery to 1 min of recovery. The latter variable has the advantage of involving blood pressure measures being obtained only in a recovery state, which avoids the inherent inaccuracy associated with exercise blood pressure measurement (8). Indeed, once exercise blood pressure and clinical confounders were considered, only the 3- to 1-min ratio of SBP remained independently predictive of angiographic CAD.
Although the sensitivities of both exercise and recovery blood pressure measures for detection of severe CAD were low, the specificities were quite high, with all being about 80%. Thus, in terms of accuracy, exercise blood pressure measures were comparable with traditional ST segment measures, as has been noted by others (1).
A delayed recovery of SBP has been noted by others to be associated with angiographic severity of CAD (2–4), the presence of perfusion abnormalities on thalium-201 scintigraphy (5)and the presence of CAD in patients with left ventricular hypertrophy and chronic heart failure (6,7). The current study extends upon these previous findings in several important respects. First, the decline in blood pressure after exercise predicts angiographic CAD even after considering the change in blood pressure during exercise. Second, using two measures of blood pressure at different points of recovery, rather than comparing a recovery to an exercise measure, results in better prediction. Finally, the sample studied was the largest single-center experience reported to date.
Mechanism of abnormal SBP response
Blood pressure is determined by a complex interplay between cardiac output, which is related to left ventricular systolic function, and peripheral vascular resistance. Previous detailed studies of cardiovascular hemodynamics during and immediately after exercise have shown that patients with severe coronary disease develop left ventricular dysfunction during exercise (2,14). Immediately after exercise, there is rapid amelioration of left ventricular asynergy (14), leading to an improvement in cardiac output. This would be expected to increase blood pressure, or slow down the decrease in blood pressure, after exercise. Furthermore, exercise itself, along with the occurrence of ischemia, promotes increased levels of circulating catecholamines, which would increase peripheral vascular resistance during early recovery (2,14). Peripheral vasoconstriction may also occur during exercise as a compensatory response to ischemic-induced left ventricular systolic dysfunction; this compensatory vasoconstriction may well persist during the first few minutes of recovery. The combination of a rapid improvement in left ventricular systolic function and increased levels of circulating catecholamines may well explain why patients with severe coronary disease have higher blood pressures during early recovery than those without disease.
Interpreting any study relating exercise to angiographic findings yields a possible “work-up bias,” in which the results of the exercise test determine in part whether coronary angiography will be performed (15). Subjects who did not augment SBP by at least 10 mm Hg were excluded from analysis because a hypotensive or blunted response to exercise is known to be associated with a high risk of CAD (16). There was no reason to believe that recovery blood pressure behavior would have influenced clinicians’ decisions to pursue coronary angiography. Another potential limitation was the use of indirect arm-cuff sphygmomanometry for SBP measurements. The potential for error was most likely to occur with measurements during exercise. Although it is possible that had all blood pressures (resting, exercise and recovery) been obtained directly, our results may have been different, the current study reflects real life practice.
A delayed blood pressure decline after graded exercise may be a useful aid in identifying patients likely to have significant angiographic CAD. We found that a delayed SBP decline predicts CAD independent of, and in addition to, a blunted blood pressure increase during exercise. Further research will be needed to optimally define how blood pressure changes during recovery should be incorporated into routine graded stress exercise treadmill testing interpretation.
- coronary artery disease
- confidence interval
- metabolic equivalents
- odds ratio
- systolic blood pressure
- Received December 21, 1998.
- Revision received March 25, 1999.
- Accepted May 16, 1999.
- American College of Cardiology
- Amon K.W,
- Richards K.L,
- Crawford M.H
- Lauer M.S,
- Pashkow F.J,
- Harvey S.A,
- et al.
- Fletcher G.F,
- Froelicher V.F,
- Hartley L.H,
- et al.
- ↵SAS Institute. SAS/STAT User’s Guide, Vol. 2. 4th ed. Cary, NC: SAS Institute, 1990.
- Rozanski A,
- Elkayam U,
- Berman D.S,
- et al.
- Bruce R.A,
- DeRouen T.A,
- Hammermeister K.E