Sex Differences in the Prognosis of Congestive Heart Failure
Results From the Cardiac Insufficiency Bisoprolol Study (CIBIS II)
Background—Whether female sex is associated with a better prognosis in patients with congestive heart failure (CHF) remains uncertain. The Cardiac Insufficiency Bisoprolol Study (CIBIS) II showed that bisoprolol reduced all-cause mortality and morbidity rates in CHF patients treated with diuretics and ACE inhibitors. We examined whether survival was different in men (n=2132) and women (n=515) enrolled in CIBIS II.
Methods and Results—Women differed from men with regard to age, NYHA functional classification, primary cause of CHF, and risk factors such as left bundle-branch block. After adjustment for baseline differences, the probability of all-cause mortality was significantly reduced by 36% in women compared with that in men (hazard ratio 0.64, 95% CI 0.47 to 0.86, P=0.003). Women also had a 39% reduction in cardiovascular deaths (hazard ratio 0.64, 95% CI 0.45 to 0.91, P=0.01) and a 70% reduction in deaths from pump failure (hazard ratio 0.30, 95% CI 0.13 to 0.70, P=0.005) compared with men. Kaplan-Meier survival analysis revealed a significant reduction in all-cause mortality among women treated with bisoprolol compared with men (6% versus 12% P=0.01) but not among women treated with placebo (13% versus 18%, P=0.10). However, this sex/β-blocker effect was not significant in multivariate analysis.
Conclusions—These results indicate that regardless of β-blocker treatment and baseline clinical profile, female sex is a significant independent predictor of survival in patients with CHF.
Despite declining age-adjusted incidence and mortality rates for coronary heart disease during the past 3 decades, the incidence of heart failure continues to increase while its prognosis remains poor.1 Congestive heart failure (CHF) is the leading principal cause for hospitalizations among elderly patients in the United States.2 Epidemiological studies, although relatively sparse, have suggested sex-related differences in the incidence and the prognosis of CHF.3 4 However, this was not the case in the Studies of Left Ventricular Dysfunction (SOLVD)5 data register.
The origin of this sex difference remains controversial mainly because of the number of biases inherent to case-control and cohort studies. In particular, it is not clear whether it reflects differences in the patient’s referral, stage of disease, management, and investigations or a specific and independent biological factor. Most of these observations were made before the use of ACE inhibitors in patients with CHF. Moreover, the difficulties in classification of outcomes in observational studies make these findings questionable.
The Cardiac Insufficiency Bisoprolol Study (CIBIS) II trial6 was the first large-scale, randomized, double-blind, placebo-controlled European trial to show a 34% reduction in all-cause mortality rates with bisoprolol, a β1-selective adrenergic antagonist, added to standard treatment with ACE inhibitors and diuretics among patients with CHF (class III and IV). The main purpose of the present study was to evaluate possible sex differences in the clinical profile, the circumstances of death, and the morbidity outcomes of patients who were included in CIBIS II. In addition, the influence of the underlying cause of CHF and the interaction between sex and β-blocker treatment on survival were compared in men and women.
Details of the study design and results of CIBIS II trial have been reported previously.6 Briefly, a total of 2647 patients with NYHA class III and IV CHF who were treated with diuretics and ACE inhibitors for ≥2 weeks were randomized into the study. Other vasodilators were allowed in case of intolerance to an ACE inhibitor (4%), whereas the use of digoxin was optional. Additional entry criteria included clinical stability for ≥6 weeks for CHF (3 months in case of CHF due to myocardial infarction [MI]) and a left ventricular ejection fraction (LVEF) of ≤35% obtained from either echocardiography or ventriculography.
CHF was attributed to (1) an ischemic cause in patients with documented coronary artery disease on the basis of either a confirmed history of MI or the presence in ≥1 coronary artery of stenosis of ≥70% on angiography, (2) a nonischemic cause in patients with idiopathic dilated cardiomyopathy who had normal coronary arteries on angiography, or (3) an undefined cause for all other patients, including those with hypertension or valvular heart disease together with suspected, but not documented, ischemic heart disease or cardiomyopathy.
Titration of the randomized study medication was performed without a run-in period with increasing doses during a maximum of 4 months according to tolerance.
The primary end point in CIBIS II was all-cause mortality. Secondary end points were cardiovascular mortality, noncardiovascular mortality, all-cause hospital admission, cardiovascular hospital admission, and premature treatment withdrawal.
Each outcome event, strictly defined before the patient’s enrollment, was reviewed and classified by members of the independent Critical Event Committee who were blinded to the treatment allocation. If necessary, additional information was provided by the investigator to support the initial report. Any disagreement between the members was resolved by review of the committee, who had the final decision for classification.
Classification of deaths has been reported elsewhere.6 Briefly, death from pump failure was considered to be any death from cardiogenic shock, progressive deterioration of heart failure, or acute pulmonary edema. Noncardiovascular death was attributed if cardiovascular death was excluded. Unknown cause of death was any death without sufficient evidence of a cardiovascular or noncardiovascular cause.
Statistical analysis was performed with SAS computer software (SAS Institute). Complete follow-up data for outcome events were available for all patients. Survival curves were estimated according to the Kaplan-Meier method and compared by log-rank test. We used χ2 and nonparametric tests to assess the relationship between baseline characteristics and sex. Each significant predictor identified through this analysis was tested in a backward stepwise multivariate Cox proportional hazards model for time to death. Other variables considered likely to have an important prognostic factor and variables with a P value of ≤0.20 were also forced into the model.
For the latter analysis, adjustment was performed with the following variables: age, sex, body mass index (BMI), NYHA functional class, duration and underlying cause of CHF, smoking status, history of diabetes, atrial fibrillation, heart rate at inclusion, systolic and diastolic blood pressures, presence of left and right bundle-branch block, presence of atrioventricular block, abnormal ST-segment interval, LVEF, concomitant and randomized treatment, and the last tolerated dose before the event. The presence of an interaction between sex and each of the variables, selected as an independent factor with the multivariate Cox model, was also tested.
Backward stepwise multivariate Cox model stratified on the underlying cause of CHF was used to assess the influence of the cause on sex-related differences in survival.
Results are expressed as relative hazard (RH) and 95% CIs.
There were some noteworthy sex-related differences in the clinical profiles of men (n=2132) and women (n=515) at baseline (Table 1⇓). Women were significantly older (65±10 versus 60±11 years, P=0.001) with more severe NYHA classification (21% versus 16% in class IV, P=0.005). They were more likely than men to have a higher systolic blood pressure and left bundle-branch block. Symptoms and signs indicated a more advanced stage of the disease in women. They had a higher percentage of peripheral edema (32% versus 25%, P=0.001) and fatigue (89% versus 86%, P=0.025) and reported more dyspnea at rest (21% versus 16%, P=0.005) than men.
In contrast, there was a significantly lower rate of ever-smoking history among women (28% versus 70%, P=0.001). They were less often referred to cardiac angiography (30% versus 42%, P=0.037) and had a lower prevalence of documented ischemia as the underlying cause of CHF compared with men, whereas an undefined cause was more frequent among women. However, in patients with documented ischemia, the percentage with previous MI was similar for the 2 sexes (85%). Hypertension (35% and 24% in women and men, respectively) and primary cardiomyopathy (40% and 48% in women and men, respectively) were the leading suspected undefined causes of CHF.
A similar proportion of women and men were receiving digitalis, diuretics, and nitrates, whereas the use of ACE inhibitors (95% versus 97%, respectively; P=0.028), amiodarone (12% versus 16%, respectively; P=0.026), and aspirin (37% versus 43%, respectively; P=0.022) was less frequent in women.
Men and women did not differ with regard to other known prognostic factors, such as race, body mass index, LVEF, diabetes, atrial fibrillation, abnormal ST segment, heart rate at rest, and heart rate at the end of dose titration. The maximal dose of bisoprolol and the percentage of premature treatment withdrawal were not sex related.
Mortality and Morbidity Outcomes
The mean follow-up, 1.3 years, was similar in men and women. A total of 384 patients (331 men [16%] and 53 women [10%]) died during follow-up. There was a consistent pattern of a lower mortality rate among women (Figure 1⇓).
The risk of all-cause death was significantly lower in women after adjustment for all other significant or known predictors (RH 0.64, 95% CI 0.47 to 0.86, P=0.003). No significant interaction was found among sex, death, and the country of enrollment. Similarly, compared with men, the relative risk of death in women was 0.64 for deaths attributed to all cardiovascular causes (95% CI 0.45 to 0.91, P=0.013), 0.11 for noncardiovascular causes (95% CI 0.01 to 0.85, P=0.034), and 0.30 for pump failure (95% CI 0.13 to 0.70, P=0.005). The rates of fatal MI, sudden death, unknown cause of death, and hospital admissions did not significantly differ between men and women.
Predictors of Death
The independent predictors of death from all causes according to demographic and clinical characteristics are listed in Table 2⇓. Among all the variables entered into the model, sex, bisoprolol treatment, age, LVEF, NYHA classification, cause (nonischemic heart failure), last tolerated dose of the randomized treatment, and systolic blood pressure were predictors of death.
Table 3⇓ shows the incidence rate of death and the adjusted relative risk of all-cause mortality for the independent predictors of survival with regard to sex. For each predictor, the incidence rate of death was consistently lower in women than in men. However, no significant interaction was found between sex and these variables with regard to the risk of death.
Effects of β-Blocker Therapy on All-Cause Mortality
Figure 2⇓ shows survival curves in women and men according to the randomized treatment. Baseline characteristics were similar in women randomized to placebo or bisoprolol with the exception of BMI (27±5 versus 26±4 kg/m2, respectively, P=0.017).
Kaplan-Meier analysis revealed a lower percentage of death among women treated with bisoprolol compared with men (6% versus 12% at mean follow-up, respectively, P=0.01), whereas in the placebo group, the reduced rate of mortality in women was not significant (13% versus 18%, respectively, P=0.10).
The mortality rate per 100 person-years was 5.2±1.2 for women treated with bisoprolol versus 9.6±0.8 for men (Table 3⇑). However, when the sex/treatment effect was tested in the multivariate Cox model, the interaction was not significant.
Influence of the Underlying Cause of CHF
Cox stepwise analysis stratified on cause of CHF revealed sex differences, with men having poorer survival rates. Compared with men, the relative risk of mortality in women was 0.63 for ischemic CHF (95% CI 0.39 to 1.02, P=0.057) and 0.58 for undefined cause (95% CI 0.37 to 0.90, P=0.015). However, the mortality rate was not sex related in the nonischemic CHF patients (0.84 in women, 95% CI 0.32 to 2.25, P=0.734).
The results of this prospective clinical trial of 2647 patients with class III and IV heart failure show that female sex is a significant factor of survival regardless of baseline clinical profile and treatment.
Clinical Profile Differences
In line with previous findings, women at baseline were at a higher risk in terms of age, hypertension, and NYHA class, whereas men had a higher rate of smoking habit and a higher prevalence of coronary CHF. Women were at a more severe course of the disease as shown by the significantly higher rates of left bundle-branch block, symptoms, and NHYA classification.
Similar to SOLVD trials, coronary heart disease was less frequently identified as the underlying cause of CHF in women enrolled in CIBIS II.7 This might not solely be due to a lower rate of coronary artery disease among women with CHF but could be due in part to sex bias in patient management or investigations. Philbin and DiSalvo8 found that coronary angiography, exercise testing, and Holter monitoring were less likely to be performed in women. The present study is consistent with this finding. At baseline, referral of women to coronary angiography was significantly less likely than for men, whereas the frequency of previous MI was similar in the 2 groups.
A substudy of the SOLVD trial reported that the use of antiplatelet agent is associated with a 28% reduction in all-cause mortality in patients with left ventricular systolic dysfunction.9 Interestingly, there was a lower rate of the use of aspirin in the women CIBIS II compared with the men (37% versus 43%, P=0.02). The sex disparity for ACE inhibitor prescriptions found in CIBIS II is also consistent with previous findings that show women receive ACE inhibitors less often than do men, even in the absence of contraindication.10 11
Differences in Mortality Rates
Regardless of all of the baseline clinical differences, women were at lower risk for all-cause mortality, cardiovascular death, noncardiovascular death, and death from pump failure. The impact of female sex on survival is consistent with the 36% risk reduction observed in the Framingham Heart Study3 and the better outcome of women in NHANES-14 and the FIRST study12 but in contrast with SOLVD trial results.5 All patients in SOLVD had a reduced LVEF, whereas this measurement was not evaluated in the Framingham Heart Study and NHANES-1. It has been suggested that the higher rate of systolic dysfunction in men accounts for the sex differences in mortality rates. This was not the case in either CIBIS II or FIRST.
Experimental and clinic studies have suggested the presence of sex-specific differences in cardiac hypertrophy and function.13 14 A different pattern of left ventricular remodeling has been suggested in rats after MI.15 In patients who are ≥60 years old, the left ventricular adaptation to a similar degree of valvular aortic stenosis was better in women than in men.16 Similar sex differences in cardiac adaptation have been suggested in mild essential hypertension. The better systolic function and left ventricular adaptation were mostly pronounced before menopause and tended to disappear thereafter.17
Influence of the Cause of Heart Failure
Survival was significantly higher in women with an undefined cause and at the limit of significance in women with coronary heart disease. The classification used in CIBIS II was based on coronary angiogram data. This may in part explain the significantly higher frequency of unknown cause of heart failure in women. It is also possible that the better survival for undefined cause is due to the greater prevalence of hypertension in women and to sex differences in the cardiac response to an increase in afterload.
The survival rate in nonischemic CHF patients was not sex related. This finding is consistent with the 18-month follow-up of 238 men and 65 women with dilated cardiomyopathy18 but in contrast with those reported by FIRST investigators.19 The classification in FIRST, based on clinical assessment without angiogram data, and the small number of patients involved (109 men and 55 women) may explain these conflicting results.
Sex may interact with the treatment of heart failure. The survival benefit with ACE inhibitors in patients with heart failure seems to be lower in women than in men.19
Some studies have suggested possible sex-related differences in adrenergic receptor sensitivity or postreceptor effector activity.20 21 There also are data that suggest sex-related differences exist in the pharmacokinetics and pharmacodynamics of β-blockers, resulting in greater drug exposure in women.22 23 Maqbool et al24 recently reported 2 common variants of the β1-adrenoreceptor. However, it is not known whether this polymorphism affects the response to β-blocker therapy.
We found that women who were treated with bisoprolol had a higher unadjusted protective effect than did men for all-cause mortality. The reduction in mortality rates was not sex related with carvedilol.25 The interaction between sex and metoprolol therapy was not analyzed in the MERIT study; however, subgroup analysis showed a significant reduction in mortality rates for men treated with metoprolol but not for women.26
These controversial results may be due to the underrepresentation of women in clinical trials in general (898 of 3093, 515 of 2132, and 256 of 838 in the MERIT, CIBIS II, and carvedilol trials, respectively). The larger CI observed for the reduction in mortality rates in women treated with metoprolol can be solely due to the smaller number of women who were included. Further studies that include a meta-analysis of β-blocker trials in heart failure are needed to provide definite conclusions regarding the sex/β-blocker interaction effect.
We found that the risk of death was significantly related to the last tolerated dose of β-blocker or placebo (Table 2⇑). To the best of our knowledge, this is the first large-scale study to suggest a dose-effect relationship. However, this assumption, which is being further analyzed in the CIBIS II database, should be taken with great caution, because the patients were not randomized to receive different doses of β-blockers.
No registry was planned for CIBIS II. Patients who enter clinical trials may be different from those in large population cohorts, and this may represent a selection bias. Although our data are similar to observational data such as those of the Framingham Heart Study in terms of age, cause of CHF, and outcome, they may not be applicable to the general population of patients with class III and IV CHF.
Clinical trials are designed principally to evaluate drug efficacy and safety rather than to test hypotheses such as sex differences. The present study is limited by the post hoc nature of its analyses. Nevertheless, we believe that the strictly defined inclusion and noninclusion criteria, the blinded evaluation of all the outcomes by a critical event committee, and the similar management of the patients during follow-up reduce this limitation. Moreover, the better survival rate for women was present in multivariate analysis that took into account all of the baseline differences and the known prognostic factors.
A significantly higher percentage of women had an undefined cause of CHF due to the absence of coronary angiogram data. Given the possibility of misclassification, findings regarding the underlying cause of CHF should be taken with great caution. CIBIS II was not designed to collect data on menopause status or regarding hormone replacement therapy; therefore, the influence of menopause and hormone replacement therapy remains to be evaluated in studies of large groups of patients.
In CIBIS II, bisoprolol reduced the mortality rates for both men and women. Despite this beneficial effect, female sex is a significant independent predictor of survival in patients with severe heart failure.
- Received June 2, 2000.
- Revision received August 30, 2000.
- Accepted September 8, 2000.
- Copyright © 2001 by American Heart Association
Ho KK, Anderson KM, Kannel WB, et al. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation. 1993;88:107–115.
Bourassa MG, Gurne O, Bangdiwala SI, et al. Natural history and current practices in heart failure. J Am Coll Cardiol. 1993;22:14A–19A.
Adams KF Jr, Sueta CA, Gheorghiade M, et al. Gender differences in survival in advanced heart failure: insights from the FIRST study. Circulation. 1999;99:1816–1821.
Pfeffer JM, Pfeffer MA, Fletcher P, et al. Favorable effects of therapy on cardiac performance in spontaneously hypertensive rats. Am J Physiol. 1982;242:776–784.
Schaible TF, Malhotra A, Ciambrone G, et al. The effects of gonadectomy on left ventricular function and cardiac contractile proteins in male and female rats. Circ Res. 1984;54:38–49.
Carroll JD, Carroll EP, Feldman T, et al. Sex-associated differences in left ventricular function in aortic stenosis of the elderly. Circulation. 1992;86:1099–1107.
Garavaglia GE, Messerli FH, Schmieder RE, et al. Sex differences in cardiac adaptation to essential hypertension. Eur Heart J. 1989;10:1110–1114.
Petrie MC, Dawson NF, Murdoch DR, et al. Failure of women’s hearts. Circulation. 1999;99:2334–2341.
Riemer RK, Wu YY, Bottari SP, et al. Estrogen reduces beta-adrenoceptor-mediated cAMP production and the concentration of the guanyl nucleotide-regulatory protein, Gs, in rabbit myometrium. Mol Pharmacol. 1988;33:389–395.