Congestive heart failure (CHF) is a disease of the elderly, and the prevalence and incidence of CHF increases steeply with age.1 Given the demographics in the Western world, the population of elderly persons with CHF is rapidly increasing. Some previous studies have shown that age is an important negative prognostic factor in CHF,2,3 but others, even large studies, have not documented any effect of age on mortality.4-7 Several of these studies examined highly selected groups of patients, such as patients referred for transplant evaluation. Thus, it is not entirely clear whether age affects mortality in more unselected CHF populations. Furthermore, few previous studies have been dedicated to studying the effect of age in CHF and therefore little information is available about how advancing age modulates the effects of known risk factors in CHF. This information is of interest since several clinical parameters differ greatly among young and old patients with CHF. In particular, heart failure with preserved left ventricular systolic function (LVSF) is much more common in the elderly.8
The aim of the present study was to determine (1) if advanced age is a predictor of mortality in CHF, and (2) if age interacts with known predictors of risk in CHF, particularly LVSF.
Patients and methods
The study population consisted of patients screened for entry into the Danish Investigations of Arrhythmia and Mortality on Dofetilide (DIAMOND)-CHF study, which was a multicenter, randomized, double-blind, placebo-controlled trial of the efficacy of the class III antiarrhythmic agent dofetilide (Tikosyn) on mortality in patients with CHF. The drug trial included 27% of the screened population with CHF and showed no significant effect of dofetilide when compared with placebo9; consequently the current study did not distinguish between patients randomized or not randomized in the study.
A total of 5,548 consecutive patients hospitalized in 34 Danish hospitals with new or worsening CHF were screened for entry into the trial between November 1993 and July 1996; these patients comprise the DIAMOND-CHF screening registry. Patients with acute myocardial infarction (MI) within the preceding 7 days were excluded. At screening, a clinical history and a physical examination were performed together with a standardized echocardiogram. LVSF was assessed by calculation of wall motion index (WMI). An estimate of ejection fraction can be obtained by multiplying WMI by 0.3. Significant left ventricular systolic dysfunction was defined as WMI < 1.2 (LVEF approximately below 0.35).
Survival status was provided by the Danish Central Personal Registry in the fall of 2002, resulting in a follow-up time of 5 to 8 years. Survival status was available on 5,491 patients. The study was approved by the Central Danish Ethics Committee.
Differences in time to death between groups were analyzed by a two-sided log-rank test. Relative risks (RR) and 95% confidence intervals were calculated as hazard ratios obtained from Cox proportional-hazard models.
Baseline characteristics of the 5,491 patients are given in Table 1. Mean age was 71.7 ± 10.2 years with 13% being younger than 61 years, 27% between 61 and 70 years, 40% between 71 and 80 years, and 20% aged 81 or older. Younger patients were more commonly male. The prevalence of previous MI increased with age until the age group older than 80 years, in which the prevalence of ischemia declined and a similar trend was seen for several baseline
risk factors (hypertension, diabetes, and chronic obstructive pulmonary disease [COPD]). The degree of LV systolic dysfunction was greater in the younger patients. Elderly patients were less frequently treated with angiotensin-converting enzyme (ACE) inhibitors. This was true also when the analysis was restricted to those patients with significant LV dysfunction where ACE inhibition was clearly indicated (WMI ≤ 1.2). In the group of patients younger than 60 years, 84% were receiving an ACE inhibitor versus 60% of the patients older than 80 years (P < .001). Few patients were treated with beta blocking agents. Thirty days after the index hospitalization, 5% of the patients had died, ranging from 3% in the youngest patients to 10% in the oldest (Table 1).
A 10-year increase in age resulted in a highly significant increase in risk of death within the first 30 days (RR, 1.55 [1.36—1.76], univariate analysis). In a multivariate analysis containing age, sex, WMI, left ventricular end- diastolic diameter (LVEDD), history of ischemic heart disease or MI, COPD, diabetes, hypertension, valve disease, atrial fibrillation, smoking, creatinine clearance, and duration of heart failure, age was independently associated with 30-day mortality (RR per 10-year increase, 1.23 [1.04–1.47, P = .02]).
During the entire follow-up period, a total of 3,955 patients (72%) died. The risk ratio for death during long-term follow-up associated with a
10-year increase in age was 1.55 (1.50—1.61, P < .001, Figure 1). In a multivariate analysis similar to the
one described for short-term out-come, age interacted significantly with WMI (P = .003) and thus separate multivariate models were generated for patients with LV systolic dysfunction (WMI ≤ 1.2) and those with WMI greater than 1.2. Age was a significant predictor of mortality in both models. For patients with LV systolic dysfunction, the RR (10-year increase in age) was 1.29 (1.19—1.39) and for patients with WMI greater than 1.2 the RR was 1.57 (1.43–1.72). Adding the presence or absence of treatment with diuretics, beta blockers, ACE inhibitors, or digoxin at discharge to the multivariate model did not significantly alter the RR for the effect of age (data not shown). To give a more clinically useful impression of what to expect in terms of survival for hospitalized CHF patients, the median survival time was calculated for the various age groups (Table 2). Figures for the patients younger than 60 years should be interpreted with caution, because less than 50% in this group had died at follow-up.
This study confirms that older patients are more likely to be female and less likely to have LV systolic dysfunction and dilatation.1,10,11 The presence of risk factors and comorbidity was clearly dependent on age, but the association was not linear. Similar results regarding the distribution of hypertension, ischemic heart disease, and diabetes have been reported from a study in the Digitalis Investigation Group (DIG) population. There is no simple explanation for the apparent decline in hypertension, diabetes, and ischemic heart disease in the oldest patient group. It could relate to underdiagnosis of these comorbidities or to a selection process implying that patients with these conditions simply were less likely to survive until the
age of 80 or older. The proportion of elderly patients being treated with ACE inhibitors was considerably lower than in the younger patients, even after controlling for the difference in the proportion of patients with LV systolic dysfunction. Lower use of ACE inhibitors in elderly CHF patients has been reported previously10,12 and may be related to a higher frequency of contraindications or fear of adverse effects in the geriatric pop-ulation. The study does not provide useful information about beta blockade in the elderly because patients were included in the registry several years before the general recommendation for beta blocker use in systolic heart failure.
The present study shows that advancing age increases short- and long-term mortality in hospitalized heart failure patients. The effect of age on mortality was substantial and persisted after controlling for several potential confounders. Most previous studies have reported increased mortality with advancing age.2,13,14 In a number of studies, however, no effect of age on mortality has been found. A complex effect of age was reported from the large Department of Veterans Affairs Cooperative Vasodilator-Heart Failure Trials (V-HeFT) database.4 Even though a small effect of age on mortality was detected in V-HeFT-I, no effect was found in V-HeFT-II and consequently there was no overall independent effect of advancing age.
Possibly the absence of an effect of age in the V-HeFT study could be explained by the fact that only men aged 75 years and younger were included. Similarly, other studies in which no effect of age has been found may have been undersized or may have included mainly younger patients, reducing the chance of discovering an effect.5-7 Indeed, some studies have demonstrated an effect of age only in the oldest patients.15 In our study, an effect of age could be demonstrated in all age strata. Overall, it seems reasonable to conclude that in studies that do not exclude elderly patients (either by age per se or by criteria correlating with advanced age), increasing age appears to have a clear negative impact on survival in patients with CHF. This effect appears to be greater in patients with preserved LVSF. The reason for this interaction is not clear. However, since normal LVSF is a predictor of survival, it could be speculated that in patients with preserved LVSF, the aging process, and the cardiac and noncardiac diseases associated with it, might play a greater role.
Elderly patients are undertreated with ACE inhibitors. Although age appears to be of greater importance in patients with nonsystolic heart failure than in those with low ejection fraction, advancing age is a potent, independent risk factor for short- and long-term mortality in all hospitalized CHF patients.