Heart Failure Survival Among Older Adults in the United States:  A Poor Prognosis for an Emerging Epidemic in the Medicare Population FREE

DURING the past 2 decades, heart failure has emerged as a major chronic disease for older US adults.^1- 6 Almost 75% of nearly 5 million patients with heart failure in the United States are older than 65 years.¹ As the progressive aging of the US population continues during the next 40 years, the number of heart failure patients older than 65 years is expected to double.² Heart failure is already the leading principal hospital diagnosis among older adults.² Furthermore, more than 90% of deaths with heart failure as the underlying cause occur among adults 65 years or older.³ These statistics take on special importance to the Medicare plan and the aging population because hospitalizations and death rates for heart failure increased among older adults between 1980 and 1990,^3- 6 despite declines in death rates for ischemic heart disease (IHD)⁷ and cerebrovascular disease⁸ and improvements in the treatment of hypertension.⁹ It is not yet clear whether recent changes in the definition, diagnosis, and pharmacological management of heart failure^10- 16 have had any impact on national survival estimates.

Survival following a diagnosis of heart failure is bleak and may be worse than the prognosis for most types of cancer.¹⁷ In the 40-year Framingham Heart Study, only 25% of 331 men and 38% of 321 women with heart failure survived 5 years after diagnosis.¹⁷ Mean age at diagnosis was 70 years (SD, 10.8 years) and median survival was 1.7 years for men and 3.2 years for women.¹⁷ However, survival estimates from this small study sample may not be representative of the general population of older adults. Among participants 65 to 74 years who reported a history of heart failure in the First National Health and Nutrition Examination Survey (1971-1975), 41% of 229,504 women and 28% of 203,704 men survived 10 years.¹⁸ Results of this study may be biased because the study included only noninstitutionalized subjects who were healthy enough to participate. Furthermore, this national study did not include adults older than 74 years and could not examine race differences because of small case numbers at baseline.

Little information is available about the survival prognosis for adults with heart failure in the general US population and race differences, in particular. Black adults may be twice as likely as white adults younger than 60 years to have a first hospitalization for heart failure.^4,19 In the US population 65 years or older, death rates for heart failure were much higher for blacks than for whites between 1980 and 1990.^3,5 However, recent studies suggest that in-hospital mortality^6,20 in black patients with heart failure is similar to or slightly lower than that in white patients; 1-year mortality²¹ in this group was similar to that in white patients. Medicare beneficiaries provide one of the unique study populations with sufficient numbers of black adults to assess race differences in more long-term survival after hospitalization with heart failure. In this study, we determined the 6-year survival prognosis for heart failure in a national cohort of Medicare enrollees who were hospitalized and discharged with heart failure for the first time in 1986.

We obtained all Medicare hospital claims records for 1984 to 1986 and beneficiary enrollment data for 1986 to 1992 from the Health Care Financing Administration. Hospital claims included admission and discharge dates, race, sex, age, state or foreign country of residence, and 5 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)²² codes for disease diagnoses. Enrollment files contained records for each person eligible for Medicare benefits and included Medicare eligibility status, Medicaid eligibility status, health maintenance organization enrollment status, and date of death for a beneficiary who died during the year listed in the file or through March of the following year.

More than 1.2 million Medicare hospital claims were submitted in 1986 with heart failure (ICD-9-CM disease codes 428 or 428.0-428.9) as 1 of 5 diagnosis codes.⁶ We excluded 27% of these claims because they were rehospitalizations during 1986 for some patients. To identify patients who were probably hospitalized for the first time with heart failure in 1986, we excluded another 24% of patients who had prior hospitalizations for heart failure in 1984 or 1985. This information was available only for patients 67 years or older. From the remaining 655,498 patients with no prior evidence of the disease, we limited the study to 231,245 (35%) patients with heart failure as the principal (first-listed) diagnosis. Median survival was similar for patients with either a first-listed diagnosis or other-listed diagnosis of heart failure (21 vs 22 months, respectively). Additional patients were excluded for inconsistent dates of either admission or death (3%), enrollment in a health maintenance organization during follow-up (4%), residence outside either the 50 states or the District of Columbia (1%), other or unknown race (3%), ages younger than 67 years (6%), and deaths during the initial hospital stay (10%). Thus, the study cohort consisted of 170,239 patients 67 years or older who were admitted and discharged with a principal diagnosis of heart failure during 1986. The proportion of black patients in this heart failure cohort (9%) was comparable to that in the 1986 US population 65 years or older (8.1%).

Survival time was defined as number of months between date of admission for the first hospitalization for heart failure and the date of death, or December 31, 1992, for eligible survivors, or the last day of the month that Medicare eligibility ceased. Survival time ranged from 2 days after discharge to 83.8 months (almost 7 years). Among the heart failure cohort, 81.9% died, 18.1% survived, and Medicare eligibility ceased for 8 persons during follow-up through 1992.

Descriptive statistics included unadjusted frequencies and Kaplan-Meier survival estimates for groups defined by race, sex, age (67-74, 75-84, or ≥85 years), Medicaid eligibility, or comorbid conditions (an ICD-9-CM code entered in the second to fifth diagnosis fields), such as diabetes mellitus (250-250.9), hypertension (401-405.9), or IHD (410-414.9). Because there was no evidence that the proportionality assumption was violated by any variable of interest, multivariable Cox proportional hazards regression analyses²³ were performed to assess the association of selected characteristics with months of survival.

Black patients with heart failure were slightly younger than white patients at discharge (Table 1). Women were more likely than men to be 85 years or older. We assessed Medicaid eligibility as a potential surrogate for socioeconomic status that might affect survival. Black patients were 3 times more likely to be eligible for Medicaid than were white patients. Women were twice as likely to be eligible for Medicaid than were men. Median hospital stay for heart failure in 1986 was 6 to 7 days (Table 1). Diabetes, hypertension, and IHD were often reported as comorbid conditions during the initial hospitalization for heart failure (Table 1). Prevalence of diabetes was highest among black women. Hypertension was the most common comorbid condition among black patients; the prevalence was almost twice that in white patients. In contrast, IHD was the most common cormorbid condition among white patients with heart failure. There was very little overlap of hypertension and IHD. Among heart failure patients, both conditions were reported for only 7% of black men, 5% of white men, 9% of black women, and 7% of white women.

Compared with white patients with heart failure, black patients had lower mortality during follow-up, higher median months of survival, and slightly higher survival during each year of follow-up (Table 2). Men had lower probabilities of survival than did women. Six-year survival was 19% among black men, 16% among white men, 25% among black women, and 23% among white women.

The probability of survival decreased in successive age groups among men (Figure 1) and women (Figure 2). The proportion of patients who survived for 6 years was consistently higher among black men than white men for ages 67 to 74 years (24% vs 23%), 75 to 84 years (16% vs 14%), and 85 years or older (9% vs 6%). Median survival for black men in age groups 67 to 74, 75 to 84, and 85 years or older was 2.7, 2.1, and 1.3 years, respectively. White men had lower median survival for these age groups: 2.3, 1.7, and 1.1 years, respectively. After taking into account differences in the age distribution, white men had 10% greater mortality than did black men (age-adjusted risk ratio [RR], 1.10; 95% confidence interval [CI], 1.07-1.13).

There was little difference in survival probabilities between black women and white women in any age group (Figure 2). The proportions of black women and white women surviving 6 years were similar for ages 67 to 74 years (33% and 35%, respectively), 75 to 84 years (26% and 25%, respectively), and 85 years or older (13% and 11%, respectively). Median survival among women was 3.6 years for age 67 to 74 years, 2.6 years for age 75 to 84 years, and 1.6 years for age 85 years or older. Risk of dying during follow-up did not differ between black women and white women after adjustment for differences in age (P=.96). The mortality rate for men was 38% higher than for women after adjustment for age and race differences (adjusted RR, 1.38; 95% CI, 1.37-1.40).

Differences in the unadjusted 6-year survival probability and the risk of dying during follow-up after adjustment for age and sex differences were assessed for groups defined by selected characteristics (Table 3). Medicaid eligibility was associated with a poor survival prognosis among white patients (P<.05) but not among black patients. Among both race groups, heart failure patients with diabetes listed as a comorbid condition had less favorable survival than did heart failure patients with no evidence of diabetes (P<.05). For example, 17% of white patients with diabetes survived 6 years compared with 21% of white patients with no mention of this disease (adjusted RR, 1.24; 95% CI, 1.22, 1.25). In contrast, however, hypertension and IHD among heart failure patients were associated with more favorable likelihoods for survival (P<.05). We further assessed the 6-year survival probabilities for heart failure patients with no mention of either hypertension or IHD, hypertension only, IHD only, and both conditions. The respective survival probabilities were 20%, 25%, 22%, and 28% for black patients and 19%, 25%, 20%, and 25% for white patients.

The prognosis following the first hospitalization for heart failure is grim. Only 19% of black men, 16% of white men, 25% of black women, and 23% of white women survived 6 years in this national cohort of Medicare patients. Almost one third of patients died within the first year after discharge. These survival estimates and those reported in the literature^{17- 18,21,24- 28} suggest that the prognosis for patients with heart failure has not improved during recent decades. Comparisons with other study populations may be problematic because of differences in the definition of heart failure, in ascertainment as an incident or prevalent case, and variations in the distributions of age and other underlying characteristics. However, the particular strength of this study is the large number of black and white adults hospitalized with heart failure during the past decade who are more representative of contemporary older US patients.

Important findings include the greater survival for women than men following a hospitalization for heart failure since women in the Medicare population tend to have less favorable survival than men following hospitalizations for stroke,²⁹ acute myocardial infarction,^30- 31 and invasive procedures³² such as angioplasty and bypass surgery. Heart failure patients with IHD or hypertension reported as comorbid conditions had a better survival prognosis than did patients who did not have these comorbid chronic conditions reported on their records. This finding is consistent with studies assessing in-hospital mortality.^33- 34 There is some speculation that heart failure patients with other comorbid conditions may have more frequent contact with health care providers, who are in a position to identify and treat the earlier stages of heart failure.³⁴ This may also explain sex differences in survival if women are more likely to see physicians and adhere to treatment strategies. It is also possible that the increasing use of angiotensin-converting enzyme (ACE) inhibitors and β-blockers following myocardial infarction or as part of hypertension management during the past decade^9,21,35- 38 may have reduced the disease severity among heart failure patients with these chronic conditions. However, coding bias may be another possible explanation for this finding. Studies with 1-year follow-up suggest that heart failure patients who are severely ill and in the process of dying have more severe acute complications that may take precedence in coding over chronic conditions.^33- 34 Because this Medicare study has a 6- to 7-year follow-up, however, this potential bias (ie, systematic underrecording of chronic conditions among severe heart failure cases at baseline) is far less likely to affect the study results.

Patients with IHD had less favorable 6-year survival than did patients with hypertension only or both hypertension and IHD. This suggests that hypertension as a precursor to heart failure may result in a less severe disease or that some aspect of the pharmacological management of hypertension confers an advantage. Furthermore, hypertension was the most common comorbid condition among black patients with heart failure, while IHD was most common among white patients—a finding consistent with the Studies of Left Ventricular Dysfunction (SOLVD).²¹ Black-white differences in comorbidity may account for greater survival in black than white men, although there was no race difference in survival among women. These findings warrant further investigation.

An important limitation of this study is that Medicare files do not contain data on functional status or severity of disease. These administrative files provide only ICD-9-CM codes. In addition, dates for first hospitalization for heart failure do not necessarily correspond to diagnosis dates if patients with less severe cases have been treated on an outpatient basis. Furthermore, unlike epidemiological surveys that obtain physiologic measurements, we were unable to examine the effect of physiologic indicators on survival prognosis. For example, ejection fractions quantify disease severity and help distinguish between left ventricular systolic dysfunction (ie, ejection fractions ≤40%) and left ventricular diastolic dysfunction (ie, normal ejection fractions but clinical evidence of pulmonary venous hypertension and congestion).¹¹ This identification influences treatment of the 2 variations^11- 14 and may be an important determinant in the prognosis of heart failure. It would be helpful if ICD-9-CM coding could be revised in the future to distinguish between these 2 types of heart failure. Another limitation of using the Medicare data is the inability to determine the accuracy of administrative or physician coding.³⁹ However, a record review of hospital claims submitted in 1985 reported a sensitivity of 85% for heart failure as the principal diagnosis and higher sensitivity values for diabetes and hypertension as secondary diagnoses.⁴⁰

Despite limitations of the Medicare data, our findings convincingly demonstrate a dismal survival prognosis for older adults with heart failure. Although we find significant advantages for subgroups with selected characteristics, the harsh reality is that the probability of surviving heart failure for 6 years is poor for all older adults. Median survival after an initial hospitalization with heart failure for 67- to 74-year-olds was 3.6 years among black women and white women, 2.3 years among white men, and 2.7 years among black men. The contrast of these averages to 1990 estimated life expectancies between 9.5 and 17.5 years for men and women in this same age range⁴¹ is considerable. Unfortunately, this study cannot address the survival prognosis for younger adults with heart failure. Two clinical trials with younger heart failure patients (mean age, 60-61 years) were conducted during almost the same time (1986-1991) as this cohort follow-up.^42- 43 The SOLVD treatment trial found that 60% of the placebo group and 65% of the enalapril group survived 4 years.⁴² The second Veterans Administration Cooperative Vasodilator–Heart Failure Trial found that 46% of the enalapril group and 52% of the hydralazine hydrochloride–isosorbide dinitrate group survived 5 years.⁴³ This contrasts with 4-year survival probabilities of 27% to 37% and 5-year survival probabilities of 21% to 31% for this older Medicare cohort.

ACE inhibitors are underused in the treatment of heart failure in the United States, although use of the drug for heart failure therapy is increasing.^15- 16,44 National survey estimates show that ACE inhibitor use during office visits for heart failure patients increased from 24% in 1989 to 31% in 1994.¹⁶ This is consistent with ACE inhibitor use among 32% of almost 6000 patients entering the SOLVD registry from 1988 to 1989⁴⁵ and among 35% of Medicare patients with heart failure as the principal diagnosis on admission to hospitals in 1993-1994 in 10 states.⁴⁴ Clinical studies, especially those conducted in cardiology units, may show higher use. Cardiologists are much more likely than primary care physicians to prescribe ACE inhibitors, digitalis compounds, and loop diuretics for heart failure patients.¹⁵ These drugs are all important components of the management strategies recently recommended for heart failure patients with left ventricular systolic dysfunction by the Agency for Health Care Policy and Research^11- 12 and a joint task force of the American College of Cardiology and the American Heart Association.¹⁴ Thus, national educational initiatives similar to the National High Blood Pressure Education Program are needed to ensure that clinical practice guidelines for the evaluation and care of patients with heart failure are followed appropriately by all physicians to improve survival through consistent pharmacological management of this chronic condition.

As the population of older adults at risk for developing heart failure continues to increase, physicians and public health professionals should be mindful that heart failure is an emerging epidemic in the United States.^1- 6,18 In 1986, there were more than 1.2 million Medicare claims submitted for patients with any mention of heart failure. By 1993, that number had increased to 1.9 million.⁶ More than half of these claims represent multiple hospitalizations for some heart failure patients either during the same or the previous 2 years. Severely ill patients with heart failure will continue to make a considerable impact on national health expenditures and resources in the 21st century.

Adults with myocardial infarction, coronary artery disease, diabetes, and hypertension are at increased risk for developing heart failure as they age.^17,21,25,46 Continued refinement of our ability to both detect and treat these predisposing conditions, as well as the early symptoms of heart failure, is vital if we wish to improve the survival prognosis for heart failure during the next decade. Aggressive use of secondary prevention strategies following myocardial infarction may be warranted to prevent the development of heart failure. This should include use of β-blockers, aspirin, lipid-lowering and anticoagulation therapies, and coronary revascularization in appropriate patients.⁴⁶ As a complementary approach to secondary prevention, strategies must also continue to encourage the prevention of these predisposing conditions as the essential first step in the primary prevention of heart failure.

Accepted for publication May 18, 1998.

This work was presented in part at the National Conference on Cardiovascular Health: Coming Together for the 21st Century, San Francisco, Calif, February 19, 1998.

The authors gratefully acknowledge the technical assistance provided by Joan K. Miller.

Corresponding author: Janet B. Croft, PhD, Cardiovascular Health Branch, National Center for Chronic Disease Prevention and Health Promotion, Mailstop K-47, Centers for Disease Control and Prevention, 4770 Buford Hwy NE, Atlanta, GA 30341-3724 (e-mail: jbc0@cdc.gov).