Heart failure morbidity among older adults began increasing in the mid-1980s (1- 4), with advances in treatment for coronary heart disease (CHD), stroke and hypertension, and improved survival for persons with these conditions. Heart failure has become a major public health concern, especially for elderly Americans (5- 6), and has been described as a “new” or “emerging” epidemic for the 21st century (7- 8). Among all adults age 40 years, 1 in 5 will develop heart failure at some point in their lifetimes (9). Currently 5.2 million Americans (2.5%) are estimated to have heart failure (10). In 2007 the estimated direct and indirect costs of heart failure in the U.S. was $33.2 billion (10).

The rate of heart failure hospitalizations, particularly in developed countries with aging populations, has increased progressively over the past decades (1- 4,11- 15), making it the most common condition for hospital admission in elderly patients, followed by pneumonia, cerebrovascular disease, cancer, and coronary atherosclerosis (16). To update these trends into the 21st century, we analyzed the National Hospital Discharge Survey (NHDS) data from 1979 to 2004 to examine annual hospitalization rates and trends in characteristics for patients hospitalized with heart failure as either the first-listed diagnosis or additional (2nd to 7th) diagnosis. We also assessed the trends in other clinical conditions that lead to hospitalization among patients with any mention of heart failure.

We used the NHDS data, 1979 to 2004, which was conducted by the National Center for Health Statistics, Centers for Disease Control and Prevention. The NHDS collected annual data from a sample of in-patient records acquired from a national sample of hospitals (16). General or children's general hospitals and hospitals with an average length of stay of fewer than 30 days for all patients were included in the survey. Federal, military, and Veterans Affairs hospitals, as well as hospital units of institutions and hospitals with fewer than 6 beds staffed for patient use, were excluded. Prior to 1988, the NHDS was based on a 2-stage design. The survey was redesigned in 1988 (17). The new sample design used a modified, 3-stage design. Although there are a number of less significant differences between the original and new designs that have been documented, the changes do not affect the ability to conduct trend analysis (17).

The discharge records were selected within sampled hospitals using systematic random sampling. Because discharges were sampled, an individual patient might appear more than once in the sample if he/she had more than 1 hospitalization within the survey period. The medical record face sheet and discharge summary were primarily used to extract patient information onto the survey form. Variables collected in the NHDS conformed with the Uniform Hospital Discharge Data Set (18- 19). These variables included birth date or age, gender, race, marital status, admission and discharge dates, discharge status, expected source(s) of payment, and information on diagnoses and procedures.

Heart failure was defined with American College of Cardiology/American Heart Association heart failure performance measures International Classification of Diseases-Ninth Revision-Clinical Modification (ICD-9-CM) diagnosis codes (Table 1) (20) for first-listed diagnoses (principal diagnoses) or additional diagnoses (2nd to 7th listed diagnoses). Among hospitalizations with additional diagnoses of heart failure, the first-listed diagnoses were identified. These included CHD (ICD-9-CM 410 to 414); other diseases of the heart (ICD-9-CM 391 to 392, 393 to 398, 402, 404, 415 to 416, 420 to 427, and 429); other cardiovascular diseases (CVDs) (ICD-9-CM 390 to 459, exclusive of the previous codes for CHD and other diseases of the heart); respiratory diseases (ICD-9-CM 460 to 519), including pneumonia and chronic obstructive pulmonary disease; and all other non-CVD and nonrespiratory conditions. We also identified comorbidities of hypertension (ICD-9-CM 401 to 405) and diabetes (ICD-9-CM 250) by 2nd through 7th diagnoses.

Hospitalization rates were calculated for 1979 through 2004 using the estimated number of hospitalizations as the numerator and the U.S. civilian population (16) as the denominator. Annual age-standardized hospitalization rates (per 100,000) were calculated using the 2000 U.S. standard population. Age-specific rates and trends in characteristics were assessed for the following 5-year aggregated time periods: 1980 to 1984, 1985 to 1989, 1990 to 1994, 1995 to 1999, and 2000 to 2004. Relative percent changes (RPC) were estimated by the changes over the 2 periods divided by the first of the period. All analyses were conducted with SPSS, accounting for total weight to obtain estimates of U.S. hospitalization.

During the 26-year period, the estimated number of hospitalizations for heart failure as any of the 7 diagnoses was 1,274,000 in 1979 and tripled to 3,860,000 by 2004. The estimated number for heart failure as the first-listed diagnosis increased from 409,000 in 1979 to 1,166,000 in 2004. Of all heart failure hospitalizations, there were 69% whites, 11% blacks, 1% Asians/Pacific Islanders, 1% others, and 18% with race missing or unknown. Further analysis by race was not conducted because of a relatively high number of missing reports (21).

Age-adjusted hospitalization rates (per 100,000) increased from 219 in 1979 to 390 in 2004 (RPC +79%; 95% confidence interval [CI]: 67% to 93%) for the first-listed diagnosis of heart failure and more than doubled from 461 in 1979 to 937 in 2004 (RPC +103%; 95% CI: 92% to 116%) for heart failure as an additional diagnosis. For both first-listed and additional diagnoses of heart failure, the age-adjusted hospitalization rates were higher in men than in women (Figure 1).

To determine age and gender impact on heart failure hospitalization, we measured annual age-specific (<65, 65 to 74, and ≥75 years) heart failure hospitalization rates by gender for both first-listed diagnosis and additional diagnosis. From 1979 to 2004, in all age and gender groups, there were steady increases in heart failure hospitalization rates. (Table 2) summarizes the hospitalization rates and RPC from 1979 to 2004. Although those younger than 65 years had the lowest hospitalization rates, they had the highest relative increases over the period. Men had higher hospitalization rates than women in all age groups. Among patients age 65 years and older, the RPCs were similar between men and women among first-listed heart failure hospitalization. However, among hospitalization with heart failure as an additional diagnosis, women had a higher increase over the period.

We compared patient characteristics across 5-year aggregated time periods for hospitalizations with any mention of heart failure (Table 3). During these time periods, more than one-half of the patients were women. More than 80% of hospitalizations for heart failure occurred among the Medicare-aged population (>65 years), except during the most recent 2000 to 2004 time period (78%). In fact, more than 80% of these hospitalizations were covered by Medicare and/or Medicaid. The age distribution of patients hospitalized for heart failure has changed during the past 2 decades so that the proportion of patients older than 85 years increased from 17.9% in 1980 to 1984 to 23.7% by 2000 to 2004, and the proportion of patients age 65 to 74 years decreased from 27.9% to 21.5%. Patients with secondary diagnoses of hypertension and diabetes increased over the years. There were increases in the proportion of hospitalizations resulting in transfers to short- and long-term care facilities, but a decrease in the proportion discharged home. The median hospital stay and the proportion of in-hospital deaths declined during this time period.

Among hospitalizations with any listed diagnosis of heart failure, approximately 30% had a first-listed diagnosis of heart failure, regardless of time period. The majority of hospitalizations were primarily due to diseases other than heart failure. Distributions of first-listed diagnoses in (Table 3) indicate that there was a decline in the proportion of CHD and other CVD as the first-listed diagnosis, as well as an increase in proportion of respiratory and other non-CVD, nonrespiratory diseases from 1980 to 1984 to 2000 to 2004. In fact, over the 26-year period, the RPC for first-listed conditions were: CHD, −50% (95% CI: −24% to −106%); other CVD, −39% (95% CI: −11% to −120%), and heart failure, −6% (95% CI: −4% to −10%). Of note is the increased proportion for first-listed diagnoses of respiratory diseases (including pneumonia) (RPC +52%; 95% CI: 24% to 88%) and other non-CVD, nonrespiratory diseases (RPC +48%; 95% CI: 28% to 80%).

For 1980 to 1984 and the more recent 2000 to 2004 time periods, we compared patient characteristics between hospitalizations having first-listed diagnoses of heart failure with those having additional (2nd to 7th) diagnoses of heart failure (Table 4). Regardless of time period, there were greater proportions of patients older than 85 years, more transfers to long-term care facilities, and more in-hospital deaths among those with additional diagnoses of heart failure compared with those with heart failure as first-listed diagnosis. Over the 26 years, there were consistent declines in in-hospital mortality (RPC −44%; 95% CI: −18% to −117%), and the decline was larger among those with heart failure as the first-listed diagnosis (RPC −60%; 95% CI: −18% to −199%) than those with heart failure as the additional listed diagnosis (RPC −38%; 95% CI: −15% to −95%).

Heart failure-related hospitalizations increased between 1979 and 2004. This increase was consistent with other earlier reports of national hospitalization data (1- 4,11- 13); however, our report provides the most recent data. This increase was greater for hospitalizations among patients with heart failure as an additional diagnosis than among those with heart failure as the first-listed diagnosis. Moreover, among hospitalizations with any mention of heart failure, the proportion with a first-listed diagnosis of CHD and other CVD declined. However, the proportions of hospitalizations with first-listed diagnoses of respiratory or other non-CVD increased and surpassed CHD as the most common first-listed diagnosis. During the same period, in-hospital mortality declined among patients hospitalized with heart failure. This decline was greater among hospitalizations with heart failure as the first-listed diagnosis than among those with an additional diagnosis of heart failure.

The growing burden of heart failure can probably be explained by multiple converging factors, including the aging of the U.S. population and the widespread availability of revascularization procedures leading to improved survival of patients following acute myocardial infarction (22- 25). However, many of these survivors develop heart failure later in life (26). Therefore, despite an overall decline in age-adjusted mortality from CHD in the U.S. (10), heart failure incidence and hospitalization rates have increased (1- 4,7- 8,10- 13,27). As mortality from myocardial infarction and other CHD continues declining and aging of the U.S. population continues, it is likely that increases in the incidence and prevalence of heart failure in the U.S. will also continue. Further improvement in medical therapy for heart failure is likely to result in declining mortality and case fatality but increasing morbidity and economic costs.

Although overall heart failure-related hospitalizations increased, the increase was greater among hospitalizations that included heart failure listed as an additional diagnosis. Interestingly, among hospitalizations with any diagnosis of heart failure, the percentages of hospitalizations with CHD, other heart disease, and other CVD as the first diagnosis also declined. By contrast, the proportion of hospitalizations with non-CVD as the first-listed diagnosis increased over the period, suggesting that noncardiac chronic conditions are becoming more common in heart failure-related hospitalizations. These noncardiac conditions, such as respiratory diseases, were more prevalent in hospitalizations of older patients with heart failure and are strongly associated with adverse clinical outcomes leading to patient hospitalization (28). According to Braunstein et al. (28), the most common noncardiac conditions that led to heart failure-related hospitalization are chronic obstructive pulmonary disease, bronchiectasis, lower respiratory disease, asthma, and acute and chronic renal failure.

The increased reporting of non-CVD conditions as the first-listed diagnosis among hospitalizations of heart failure patients may be explained by underutilization of effective heart failure therapies in the presence of other conditions that contraindicate the use of these therapies. For example, using beta-blockers in asthma or angiotensin-converting enzyme inhibitors in renal insufficiency could have a negative impact on these conditions and lead to hospitalization (29- 30). By contrast, with advanced prevention and treatment for CVD over the past decades, it is possible to control heart failure adequately and minimize heart failure hospitalizations caused by other CVD. Additionally, patients with heart failure are more likely to experience illness from influenza and pneumonia, which can lead to hospitalizations (29,31). The opportunity to reduce hospitalizations among heart failure patients includes managing and treating patients with noncardiac diseases appropriately rather than controlling for cardiac disease alone.

Other possible explanations for increasing additional diagnoses of heart failure are the “upcoding” of heart failure discharge diagnosis for quality assurance. Patients with heart failure pose a challenge for hospital care (32- 33). The impact of heart failure as a comorbidity includes more medications and higher risk of adverse interactions, more frequent doctor visits and higher health care costs, impaired quality of life, and higher mortality (34). Furthermore, the U.S. Centers for Medicare and Medicaid Services reimburses hospitals for Medicare patients on the basis of diagnosis-related groups. For other primary diagnoses, the presence of complications increased the level of reimbursement (35). Because heart failure qualifies as a complication, there was economic incentive for coding heart failure as a complication (36).

Between 1979 and 2004, in-hospital deaths among heart failure hospitalizations declined substantially. During this time, increased use of angiotensin-converting enzyme inhibitors, beta-blockers, diuretics, and digoxin; declines in the use of potentially harmful drug therapies (e.g., calcium channel blockers); and increased use of beneficial invasive procedures (e.g., revascularization procedures) all may have had an impact on in-hospital mortality (37- 40). Although results here indicate that in-hospital mortality declined during the past 2 decades, it is possible that mortality among heart failure patients has merely shifted to another setting, such as chronic rehabilitation hospitals or skilled nursing facilities. In fact, our data showed that discharges to long-term care facilities increased during the period. Unfortunately, NHDS does not include information on postdischarge mortality.

The strengths of NHDS include its large size, representative quality, standardized methodology, and ability to examine long-term trends in hospitalizations (16). However, major limitations of this administrative database include the inability to validate diagnoses or procedure use, assess racial or state variations, or determine postdischarge mortality. In addition, the data provides minimal clinical detail to assess disease severity. Furthermore, the analyses are based on hospital admissions rather than the patients. Finally, without access to sampling design weight, we were unable to estimate standard errors. However, these limitations should not affect the internal accuracy of this study.

In summary, during the past 26 years, U.S. hospitalizations for heart failure increased, particularly among patients with heart failure as an additional diagnosis. With aging of the U.S. population, advanced therapeutic interventions and improved survival for patients with acute myocardial infarction and other CHD, it is expected that heart failure hospitalizations could continue increasing in the future. This will substantially increase health care costs paid by Medicare/Medicaid. To reduce heart failure hospitalizations, therapeutic strategies should include preventive treatment for noncardiac conditions to decrease the number of hospitalizations.