Author + information
- Received April 21, 2003
- Revision received July 7, 2003
- Accepted July 28, 2003
- Published online December 17, 2003.
- Terence Kavanagh, MD, FACC*,†,‡,* (, )
- Donald J Mertens, MD, MSc*,
- Larry F Hamm, PhD*,∥,
- Joseph Beyene, PhD§,
- Johanna Kennedy, RN*,
- Paul Corey, PhD§ and
- Roy J Shephard, MD, PhD‡,§
- ↵*Reprint requests and correspondence:
Dr. Terence Kavanagh, Room 306, Columbus Centre, 901 Lawrence Avenue West, Toronto, Ontario M6A 1C3, Canada
Objectives This study investigated the prognostic importance of measured peak oxygen intake (V̇O2peak) in women with known coronary heart disease referred for outpatient cardiac rehabilitation.
Background Exercise capacity is a powerful predictor of prognosis in men with known or suspected coronary disease. Similar findings are described in women, but fewer studies have utilized measured V̇O2peak, the most accurate measure of exercise capacity.
Methods A single-center design took data from 2,380 women, age 59.7 ± 9.5 years (1,052 myocardial infarctions, 620 coronary bypass procedures, and 708 with proven ischemic heart disease), who underwent cardiorespiratory exercise testing. They were followed for an average of 6.1 ± 5 years (median 4.5 years, range 0.4 to 25 years) until cardiac and all-cause death.
Results We recorded 95 cardiac deaths and 209 all-cause deaths. Measured V̇O2peakwas an independent predictor of risk, values ≥13 ml/kg/min (3.7 multiples of resting metabolic rate) conferring a 50% reduction in cardiac mortality (hazard ratio [HR] 0.5, p = 0.001). Considered as a continuous variable, a 1 ml/kg/min advantage in initial V̇O2peakwas associated with a 10% lower cardiac mortality. Adverse predictors were diabetes (HR 2.73, p = 0.0005) and antiarrhythmic therapy (HR 3.93, p = 0.0001).
Conclusions As in men, measured V̇O2peakis a strong independent predictor of cardiac mortality in women referred for cardiac rehabilitation.
The long-term prognosis of men with known or suspected coronary artery disease is strongly related to exercise capacity (1–3). The same relationship has been described in healthy women (4)and in women with ischemic heart disease (IHD) (5–8). However, most reports have expressed exercise capacity as a treadmill time or MET value (multiples of the resting metabolic rate) rather than measured peak oxygen intake (V̇O2peak), the most accurate measure of exercise capacity (9). We therefore examined the prognostic value of measured V̇O2peakin a substantial sample of women with proven coronary heart disease (CHD).
A total of 2,380 women (1,052 myocardial infarctions [MIs], 620 coronary artery bypass graft procedures [CABGs], and 708 documented cases of IHD, including 161 statistically similar patients who underwent a percutaneous coronary intervention) were referred for exercise testing 13.9 ± 4.1 weeks after the event, procedure, or, in the case of IHD, diagnosis. Referrals grouped in five-year intervals were as follows: 1973 to 1977, n = 72; 1978 to 1982, n = 235; 1983 to 1987, n = 394; 1988 to 1992, n = 549; 1993 to 1998, n = 1,130.
Health status was obtained from a questionnaire mailed at approximately 12- to 18-month intervals. Death certificates were examined for decedents resident in Ontario; for 15 decedents no longer resident in Ontario, cause of death was determined from information obtained from a physician, spouse, or other family members.
All patients underwent maximal cardiorespiratory exercise testing on the cycle ergometer, using the identical protocol we have previously employed in men (3). Measurements included continuous electrocardiographic monitoring, resting and exercise blood pressure, the rating of perceived exertion on the original Borg scale, and analysis of respiratory gases for V̇O2peakand ventilatory threshold.
Data were analyzed using univariate and multivariate Cox proportional hazard models. Survival time was defined as the day of the exercise test to cardiac death, all-cause death, or last contact. The V̇O2peakwas treated both as a continuous variable and then as a binary variable, with the optimal cutoff point determined by recursive partitioning (10,11). Hazard ratios (HRs) were calculated for individual risk factors, agreement between measured and predicted V̇O2peakassessed using kappa statistic, and paired comparison made using McNemar's chi-square test. Kaplan-Meier survival curves were also generated. SAS Software version 8.2 (SAS Institute Inc., Cary, North Carolina) was used for all statistical analyses.
Clinical and physiologic characteristics
The mean age at the time of testing was 59.7 ± 9.5 years. We were unable to contact 103 patients (4.3% of sample). Patients tended towards obesity (27% had a body mass index ≥30; mean 34.4 ± 4.0 kg/m2), and approximately one-half were taking a platelet inhibitor (most commonly aspirin) or a beta-blocker (Table 1).
The average V̇O2peakvalues were typical of those we and others have reported on postcoronary women in their late 50s (12,13). Complex ventricular arrhythmias, significant ST-segment depression (≥0.2 mV), and exertional angina occurred in 9.1%, 7.7%, and 10.4% of patients, respectively.
The average follow-up time was 6.1 ± 5 years (median 4.5 years, range 0.4 to 25 years). During follow-up, 95 cardiac deaths (59 MI, 23 CABG, 13 IHD) and 209 all-cause deaths (4% and 8.8%, respectively, of the sample) were recorded. The average time from the exercise test to cardiac death was 7.3 ± 4.6 years (median 6.8 years) and to all-cause death 8.3 ± 3.2 years (median 7.7 years).
The Kaplan-Meier curves for MI, CABG, and IHD patients showed that the survival profiles for cardiac and all-cause death were similar for the three diagnostic categories (log rank test, p > 0.05) and, therefore, the data for all three diagnoses were pooled.
Predictors of mortality
Univariate and multivariate analyses of predictors of cardiac death are shown in Table 2. Variables not achieving statistical significance on univariate analysis were age, obesity (BMI ≥30 kg/m2), ST segment depression ≥0.2 mV or anginal symptoms on test, exertional ectopy (Lown 3, 4, or 5), current smoking or history of smoking, hypercholesterolemia (total cholesterol ≥240 mg/dl, 6.2 mmol/l), exertional hypotension, and ventilatory threshold, as well as taking aspirin, a beta-blocker, a calcium antagonist, a nitrate, a statin, or an anticoagulant. Three of the six variables achieving statistical significance after univariate analysis remained on multivariate analysis. Taking a V̇O2peakof <13 ml/kg/min (∼ 4 METs) as the referent, values at or above this level conferred a 50% reduction in cardiac death. When V̇O2peakwas treated as a continuous variable, for each 1.0 ml/kg/min advantage in V̇O2peaksubjects gained a 10% lower cardiac mortality (HR 0.90, 95% confidence interval [CI] 0.85 to 0.96, p = 0.001) within the range of the observed sample. Therapy with antiarrhythmic drugs and a history of diabetes were significant adverse predictors of cardiac death.
Age did not emerge as a significant predictor of cardiac death in the univariate analysis, and, therefore, we conclude that the protective effect of a high V̇O2peakas well as the adverse effects of antiarrhythmic therapy and of diabetes are not influenced by age.
In terms of all-cause death, V̇O2peakvalues ≥13.0 ml/kg/min conferred a 29% reduction in mortality, but because almost one-half of the deaths were cardiac in origin, it seemed likely that the association was heavily influenced by this fact (Table 3).
We also considered the prognostic value of V̇O2peakpredicted from the peak power output on the cycle ergometer (14). As we found in men (3), the predicted V̇O2peaktended to overestimate aerobic power, with the result that in 21% of the total sample there was a discrepancy between predicted and measured V̇O2peak. This differential discordance between the predicted and the measured V̇O2peakvalues was statistically significant (McNemar's test p value <0.0001). In fact, at this cutoff, the agreement between the predicted and measured values was moderate at best (kappa coefficient = 0.47, 95% CI 0.43 to 0.51).
The Kaplan-Meier survival curves for cardiac and all-cause deaths are shown in Figure 1. When data were grouped by initial V̇O2peak, there was a marked protective effect with regard to cardiac death (Fig. 2) and for all-cause death (Fig. 3) for those at or above V̇O2peak13 ml/kg/min. Cardiac survival prospects were substantially worsened by a history of diabetes (15-year survival: 64.4% vs. 88.2%, p < 0.0001) and by antiarrhythmic therapy (15-year survival: 71.4% vs. 87.6%, p < 0.0001).
Our data demonstrate that, as in men, the prognosis of women with coronary artery disease is strongly linked to their exercise capacity, and this observation is in keeping with prior work (5,6,8). Using techniques identical to those we had previously used in men (3), we found that the magnitude of the advantage of well-conserved aerobic function is relatively independent of gender. Thus, the V̇O2peakcutoff point above which there is a marked benefit in prognosis (13 ml/kg/min in women versus 15 ml/kg/min in men), as well as the 1 ml/kg/min advantage in V̇O2peakwhen treated as a continuous variable (10% lowering of cardiac mortality in women versus 9% in men), is similar to that we had previously noted in men (3).
Other significant multivariate predictors in our analysis were diabetes and antiarrhythmic therapy (for cardiac deaths), and a history of hypertension, diabetes, or antiarrhythmic or digoxin therapy (for all-cause deaths). The increase in risk associated with the taking of an antiarrhythmic drug is likely due to the fact that 40% of these patients were taking proarrhythmic agents (quinidine, disopyramide, procainamide), which were commonly prescribed in the early years of the follow-up period.
Exercise-induced ST-segment depression is less sensitive in women than in men (15,16), and, therefore, it is not surprising that in our data this variable was not a significant predictor of death.
Our conclusions are based on relatively young women (<70 years) with stable CHD who were referred for cardiac rehabilitation, and may not apply across the spectrum. No patients were in overt heart failure, and severe comorbidities such as peripheral vascular disease, chronic pulmonary disease, and disabling arthritis were rare. Also, our sample contained few subjects from ethnic minority groups, being predominantly white and from the middle classes. Finally, secular trends in behavior and environmental factors, inevitable in a long follow-up, as well as shifts in clinical practice, may have influenced the shape of the mortality curves.
From a clinical viewpoint, in men as in women with proven coronary artery disease, measured V̇O2peakis a strong independent guide to prognosis, is not influenced by age, and is more valid than a predicted V̇O2peak.
The authors are indebted to the Human Performance Laboratory staff: Drs. A. Elzawi, B. Evans, V. Matosevic, A. Qureshi, and R. Rudner (physicians); P. Baskey, BScN, R. Belliard, K. Bernsdorff, J. D'Elia, S. Qureshi, MSc, Barb Smith-Perrin, and E. Strohm, MSc (technical staff); M. Bhagi, BE, for her help with data collection; and J. Will for administrative assistance and manuscript preparation.
☆ This work was supported by a grant from Canadian Cardiac Rehabilitation Foundation, Toronto, Ontario, Canada.
- coronary artery bypass graft
- coronary heart disease
- confidence interval
- hazard ratio
- ischemic heart disease
- multiples of resting metabolic rate
- myocardial infarction
- peak oxygen intake
- Received April 21, 2003.
- Revision received July 7, 2003.
- Accepted July 28, 2003.
- American College of Cardiology Foundation
- Vanhees L.,
- Fagard R.,
- Thijs L.,
- Staessen J.,
- Amery A.
- Kavanagh T.,
- Mertens D.J.,
- Hamm L.F.,
- et al.
- Roger V.L.,
- Jacobsen S.J.,
- Pellikka P.A.,
- Miller T.D.,
- Bailéy K.R.,
- Gersh B.J.
- Snader C.,
- Marwick T.H.,
- Pashkow F.J.,
- Harvey S.A.,
- Thomas J.D.,
- Lauer M.S.
- Arruda-Olson, Juracan EM, Mahoney DW, McCully RB, Roger VL, Pellikka PA. Prognostic value of exercise echocardiography in 5,798 patients: is there a gender difference? J Am Coll Cardiol 2002;39:625–31
- Fletcher G.F.,
- Balady G.J.,
- Amsterdam E.A.,
- et al.
- Zhang H.,
- Singer B.
- Brieman L.,
- Friedman J.H.,
- Olshen R.A.,
- Stone C.J.
- Kavanagh T.,
- Hamm L.F.,
- Shephard R.J.,
- Mertens D.J.,
- Kennedy J.,
- Beyene J.
- Gibbons R.J.,
- Balady G.J.,
- Beasley J.W.,
- et al.
- Okin P.M.,
- Kligfield P.