Respiratory syncytial virus (RSV) is an important cause of acute lower respiratory tract disease among the elderly, but national estimates of the burden of this disease have not been made. To estimate the morbidity, mortality, and medical costs of RSV-associated pneumonia among US elderly, national hospital discharge data, vital statistics, etiologic studies of adult pneumonia hospitalizations, and Medicare cost data were reviewed. In the United States, 687,000 hospitalizations and 74,000 deaths caused by pneumonia occur annually among the elderly; ∼2%–9% of these are caused by RSV. At a cost of $11,000 per RSV pneumonia hospitalization, the estimated annual cost of RSV pneumonia hospitalizations is $150–$680 million. Exacerbations of congestive heart failure and other chronic conditions may also contribute substantially to RSV disease burden among the elderly. The total RSV disease burden is probably great enough to justify development of an RSV vaccine for use in this group.
Respiratory syncytial virus (RSV) is the single most important cause of serious lower respiratory tract disease in infancy and early childhood worldwide . In adults, RSV is also increasingly recognized as a cause of pneumonia and exacerbations of chronic pulmonary or cardiac disease [2–7]. One study has demonstrated that RSV is among the three pathogens most commonly detected in winter in adults with community-acquired pneumonia , and other studies have suggested that among the elderly, the rate of RSV-associated disease may be similar to that of influenza [2–4, 8, 9]. Adults of advanced age may be at particular risk for RSV disease; in outbreaks of RSV among nursing home elderly, for example, 5%–50% of infections result in pneumonia and 2%–20% in death [10–15]. This recognition of RSV disease among the elderly has raised the possibility that RSV vaccines (e.g., a subunit vaccine), which may be efficacious in adolescents with cystic fibrosis, may be useful in the elderly . The objective of this analysis was to estimate the number and direct medical costs of hospitalizations for RSV pneumonia among elderly and nursing home elderly in the United States, as a basis for future evaluation of the costs and benefits of immunizing these groups with RSV vaccines currently being developed.
To estimate the burden of RSV disease among the elderly, we determined the annual number and rates of pneumonia hospitalizations and deaths in this group, the percentage of these hospitalizations and deaths associated with RSV, and the direct medical costs of RSV-associated pneumonia hospitalizations among the elderly. To do this, we reviewed three data sources: national hospital and nursing home discharge data and mortality statistics, all from the National Center for Health Statistics; etiologic studies of adult pneumonia hospitalizations; and Medicare hospitalization cost data. “Elderly” was defined as being aged 3≥65 years. Nursing homes were defined as facilities with three or more beds in which nursing and personal care services were provided. We considered all pneumonias that were assigned ICD-9-CM (International Classification of Disease, 9th revision, Clinical Modification) codes of 480–486; these include viral, bacterial, and mycotic pneumonias and pneumonia and bronchopneumonia caused by unspecified organisms.
National hospitalization and mortality data
To determine current rates of pneumonia hospitalizations and deaths among all elderly, we reviewed the 1995 National Hospital Discharge Survey (NHDS)  and 1995 national mortality statistics , respectively. We compared these rates with rates in 1985 [19, 20]. Both NHDS and mortality statistics databases code illnesses according to ICD-9-CM diagnostic categories. The NHDS data were derived from a systematic random sample of discharge diagnoses from a sample of all US general short-stay hospitals, excluding military, federal, and Department of Veterans' Affairs hospitals. National mortality statistics were based on all death certificates filed in the United States in 1995. We considered only hospitalizations with a first-listed discharge diagnosis of pneumonia and only deaths for which pneumonia was the underlying cause of death.
To determine pneumonia hospitalization and death rates among nursing home elderly, we reviewed data from the 1985 National Nursing Home Survey (NNHS) . Discharge data from the 1995 NNHS are not yet available. NNHS data were derived from a sample of discharge diagnoses, coded by ICD-9-CM categories, from a sample of nursing homes in the contiguous United States. We considered only persons with a primary discharge diagnosis of pneumonia, who were either discharged to short-stay hospitals or discharged dead, and multiplied these rates by 96%, the percentage of nursing home residents discharged with pneumonia who were ≥65 years of age. We calculated the number of pneumonia deaths occurring among nursing home elderly by adding the number of pneumonia deaths that occurred in the nursing homes and the number of transfers of nursing home elderly to short-stay hospitals for pneumonia, multiplied by 14%, the mortality rate of all elderly patients hospitalized with pneumonia in 1985 .
RSV pneumonia hospitalization data
We reviewed the medical literature published between 1966 and 1997 for etiologic studies of adult pneumonia hospitalizations. We identified eight studies in which adults with pneumonia were systematically tested for RSV infection by serologic tests (EIA or complement fixation assay on acute- and convalescent-phase serum samples) or by virus isolation or antigen detection on respiratory secretion specimens. From each study, we determined the percentage of pneumonia hospitalizations caused by RSV and, when possible, the percentage of hospitalizations for influenza. We compared the results of these studies, considering differences in study population, study size, inclusion criteria, years and seasons of study, and sensitivity of diagnostic tests used. Since studies during winter months overestimate the year-round contribution of RSV to pneumonia hospitalizations, we considered results from these studies separately. When possible, we also derived the percentage of pneumonias associated with RSV year-round from the percentage of pneumonias associated with RSV in winter. We did this by assuming that all RSV-associated pneumonia hospitalizations occurred during the winter study months and that the proportion of yearly pneumonia hospitalizations occurring during the winter study months for each study was similar to that reported by the NHDS .
Medicare hospitalization cost data
To determine current costs of pneumonia hospitalizations among US elderly, we reviewed 1996 hospitalization cost data from the Medicare health insurance program (Health Care Financing Administration, Office of Strategic Planning, unpublished data). Medicare is a federally administered insurance program enrolling > 38 million persons in the United States, including 95% of all US elderly . About 88% of Medicare enrollees are elderly, 12% are nonelderly persons with disabilities, and < 1% are nonelderly persons with end-stage renal disease (ESRD) . In 1996, ∼12million hospitalizations occurred among Medicare enrollees. Charges for each hospitalization (including charges for medical services, surgical procedures, intensive care, meals, routine nursing care, medications, and diagnostic tests) were recorded for each ICD-9-CM diagnosis. We examined the mean charges for hospitalization for “pneumonia, organism unspecified” (ICD-9-CM code 486), calculated from the cumulative charges made to all patients discharged with this diagnosis, since this is a likely diagnosis for undiagnosed RSV pneumonia. We then compared these charges with the mean total charges for hospitalizations for other types of pneumonia.
In 1985, among 28.4 million elderly in the United States, there were 400,000 hospitalizations and 57,000 deaths from pneumonia (table 1). In 1995, among 33.5 million elderly in the United States, there were 687,000 hospitalizations and 74,000 deaths from pneumonia. The hospitalization rate increased by 50% between 1985 and 1995 (1400 and 2100 hospitalizations/100,000 population, respectively), while the mortality rate remained almost unchanged (200 deaths/100,000 population in both years). Much but not all of the increase in hospitalization rates over this time period probably can be attributed to aging of the population; the hospitalization rates increase by 11% to 25% for the age strata 65–69 years, 70–74 years, 75–79 years, 80–84 years, and ≥85 years.
In 1985, among 1.3 million nursing home elderly, there were 33,200 transfers to short-stay hospitals for pneumonia (table 1). About 24,300 pneumonia deaths occurred among nursing home elderly: 19,700 deaths in nursing homes and ∼4600 deaths after transfer to a hospital. In 1985, the pneumonia hospitalization rate among nursing home elderly was 1.8 times greater than that among all elderly (2500 vs. 1400 hospitalizations/100,000 population), and the mortality rate was 9 times greater (1800 vs. 200 deaths/100,000 population per year).
The features and results of the eight hospital-based etiologic studies of adult pneumonia are summarized in table 2. The first three studies listed in the table were conducted year-round in Europe, and diagnosis of RSV infection was made by serologic assays [25–27]. Fransen et al.  studied only elderly persons but included persons hospitalized with either pneumonia or bronchitis. Hers et al.  studied only young adults (military recruits). Vikerfors et al.  studied 2400 adults of all ages hospitalized at any point over 10 years but did so retrospectively as part of a study describing clinical RSV disease. In these three studies, 2%–6% of adult pneumonia hospitalizations occurring year-round were found to be associated with RSV infection. The next three studies were small-scale (each testing ≤s100 persons), included adults of all ages, and were conducted during a single winter only [4, 9, 28]. In the studies by Kimball et al.  and Zaroukian et al. , all diagnoses of RSV infection were made by virus isolation, whereas Melbye et al.  diagnosed RSV infection serologically, by a 4-fold rise in IgG antibody titer or by a single high IgG titer. In these three studies, 2%–14% of winter adult pneumonia hospitalizations were associated with RSV. In the last two studies, Dowell et al.  and Falsey et al.  prospectively studied several hundred elderly pneumonia hospitalizations each, using serologic assays to diagnose RSV infections (Dowell S, Falsey A, personal communications). Dowell et al.  found that 5% of winter pneumonia hospitalizations of elderly were due to RSV, or 3% of hospitalizations year-round. Falsey et al.  found that 14% of winter pneumonia hospitalizations of the elderly were associated with RSV; from these data, we estimate that 9% of year-round pneumonias were associated with RSV (assuming that all RSV-associated pneumonia hospitalizations and 60% of yearly pneumonia hospitalizations occurred in the study months of November–April). Despite the wide variation in these eight studies, in study setting, design, and methods, the percentages of pneumonia hospitalizations associated with RSV were similar, ranging from 2% to 9% year-round and from 2% to 14% in winter. The percentages of hospitalizations associated with influenza were similar to those associated with RSV (table 2).
In 1996, the mean cost of 414,000 hospitalizations of Medicare enrollees for pneumonia from unspecified organisms was $10,700 per hospitalization (table 3). This cost, which was chosen to represent the cost of hospitalization for undiagnosed RSV pneumonia, was similar to the mean costs of hospitalizations for all pneumonias (ICD-9-CM codes 480–486), bronchopneumonia, organism unspecified (485), influenza pneumonia (487.0), and obstructive chronic bronchitis (491.2). On the other hand, the 85 hospitalizations for pneumonia specifically caused by RSV (480.1) were more expensive than those for other pneumonias. These hospitalizations probably include only cases of adult RSV illness in which the diagnosis of RSV was sought and made. Such cases are probably atypical and, most likely, involve a high proportion of patients with high-risk conditions (e.g., bone marrow transplant recipients) or unusually severe illness, whose hospitalizations are expected to be more expensive.
We estimate that between 2% and 9% of pneumonia hospitalizations among the elderly each year are associated with RSV infection, or 14,000–62,000 pneumonia hospitalizations among all elderly persons and 700–3000 among nursing home elderly. The rates of RSV-associated pneumonia hospitalizations among all elderly and nursing home elderly are 40–180 and 50–230 hospitalizations/100,000 persons per year, respectively. Assuming that 2%–9% of pneumonia deaths among the elderly are also associated with RSV infection, we estimate that 1500–6700 deaths each year are associated with RSV infection among all elderly (5–20 deaths per 100,000 population) and 500–2200 deaths among nursing home elderly (40–170 deaths/100,000 population). At an estimated cost of $11,000 per hospitalization, the total annual cost of hospitalizations of all elderly for RSV pneumonia is between $150 and $680 million.
In this study, we demonstrate that RSV-associated pneumonia imposes a substantial medical and economic burden among the elderly in the United States: ∼14,000–62,000 RSV-associated pneumonia hospitalizations of elderly occur annually, at rates of 40–180/100,000 population and at an estimated cost of $150 to $680 million. RSV-associated hospitalization and death rates are even higher among nursing home elderly. The number of pneumonia hospitalizations associated with RSV may possibly be similar to that associated with influenza, as suggested by several of the etiologic studies of adult pneumonia hospitalizations. The number of pneumonia hospitalizations associated with RSV may also be similar to the number associated with Streptococcus pneumoniae (32,000 hospitalizations in 1995), as suggested by NHDS data . Among Medicare enrollees, direct medical costs of hospitalizations for RSV-associated pneumonia are similar to those for S. pneumoniae pneumonia ($421 million/year), emphysema ($232 million/year), and asthma ($652 million/year) (Health Care Financing Administration, Office of Strategic Planning, unpublished data).
There are several ways in which we may have overestimated or underestimated the burden of RSV-associated disease among the elderly, although we believe that, overall, we have underestimated it substantially. Sources of potential overestimation are as follows.
First, nonelderly Medicare enrollees with ESRD and ESRD-associated immunocompromised conditions are included in the calculation of mean hospitalization costs, and these persons may have more severe pneumonias and more expensive hospitalizations than do elderly persons in general. However, non-elderly persons with ESRD account for < 1% of Medicare enrollees and would not be expected to alter significantly the mean hospitalization charges for pneumonias.
Second, in reviewing the etiologic studies of adult pneumonias, we assumed that RSV was responsible for hospitalizations of patients dually infected with RSV and other pathogens. However, even if we were to assume that other pathogens were solely responsible for hospitalizations of dually infected patients, dual infections are too few (occurring in ∼14% of RSV-infected persons) to alter our results substantially .
Third, we may have overestimated RSV-associated mortality by assuming that the mortality rate for RSV-associated pneumonia is the same as that for all pneumonias. The actual mortality rate among the elderly hospitalized with RSV pneumonia may be lower, as suggested by one study that found a mortality rate of 6% among patients serologically diagnosed with RSV infection compared with a rate of 9%–11% among all elderly patients hospitalized with pneumonia between 1992 and 1995 [17, 30–32] (Falsey A, personal communication). However, since serologic diagnosis of RSV infection requires collection of a convalescent-phase serum specimen, some fatal RSV infections will be missed, and the actual RSV mortality rate is likely to be higher than that found by Falsey et al.
Even with these potential sources of overestimation, we believe that our estimates of the burden of RSV disease among the elderly are likely to be low for several reasons.
First, we considered only NHDS hospitalizations with a primary diagnosis of pneumonia. In doing so, we excluded 383,000 hospitalizations in which pneumonia was a secondary diagnosis ; some of these hospitalizations were likely to have been caused by RSV infection or prolonged by nosocomial RSV infection.
Second, we used nursing home discharge data from 1985, which do not capture the increases, between 1985 and 1995, in the number of elderly in nursing homes and the rate of pneumonia hospitalizations among all elderly.
Third, several of the etiologic studies examined RSV infection among persons with admission, rather than discharge, diagnosis of pneumonia. It is possible that the percentage of RSV infections among those diagnosed with pneumonia at admission might be lower than that among those diagnosed at discharge, since patients admitted with the diagnosis of pneumonia may be discharged with noninfectious diagnoses, such as pulmonary neoplasm or coronary artery disease [2, 3]. However, in the studies by Falsey et al. and Dowell et al. (personal communications), the frequency of RSV infection was very similar between patients with pneumonia diagnosed at admission and those diagnosed at discharge.
Fourth, RSV disease was underestimated because RSV serology, although quite sensitive when true baseline serum samples are available for comparison with convalescent-phase samples, can fail to detect infection in 15%–33% of persons with RSV diagnosed by culture or antigen detection [3, 33]. Documentation of a 4-fold rise in serum IgG antibody titer may be difficult because all adults have had previous RSV infections and will likely mount rapid antibody responses to reinfection. The study by Dowell et al.  illustrates this problem. These researchers defined RSV infection by a 4-fold rise in IgG antibody titer and identified 57 persons who met this definition; however, an additional 15 persons did not meet this definition, in spite of having RSV IgM antibodies and a single high RSV IgG titer. Antigen detection and isolation studies appear to be even less sensitive than serology [3, 33, 34].
Perhaps the most significant source of underestimation of RSV disease burden is the exclusion of several potentially expensive outcomes of RSV infection, the costs of which are poorly defined.
First, we did not account for the costs of continued medical care following hospitalization for pneumonia. In one study, 14% of elderly patients hospitalized for RSV disease required a higher level of care at discharge than at admission, but the costs associated with this care were not determined .
Second, we did not account for the costs of outpatient visits for RSV-associated upper respiratory tract illnesses. We estimate these visits to cost about $2–$9 million, assuming that 3.4 million annual outpatient visits are made by elderly patients for such illnesses, each visit costs $70, and 1%–4% of these illnesses are associated with RSV infection [35–39].
Finally, and most important, we did not account for the costs of RSV-associated exacerbations of chronic conditions, such as congestive heart failure and chronic obstructive pulmonary disease. Falsey et al. (personal communication) enrolled elderly patients hospitalized with any acute cardiopulmonary illness between November and April over 3 years and found RSV infection (serologically confirmed by 4-fold rise in IgG antibody titer) in 19 (8%) of 249 winter admissions for congestive heart failure (ICD-9-CM codes 401.01, 402.11, 402.91, 428) and in 32 (22%) of 146 winter admissions for chronic obstructive pulmonary disease or bronchitis (ICD-9-CM codes 466, 490, 491, 496). In 1995, a total of 756,000 and 403,000 elderly were hospitalized for these conditions, respectively, at a cost of about $10,000 per hospitalization [3, 29] (Health Care Financing Administration, Office of Strategic Planning, unpublished data). In addition, Falsey et al. (personal communication) found RSV infection in 6 (17%) of 35 elderly hospitalized in the winter with aspiration pneumonia (ICD-9-CM code 507.0), 3 (14%) of 21 with respiratory failure (ICD-9-CM code 518.81, 518.82), and 2 (11%) of 19 with “influenza” (ICD-9-CM code 487). If further study confirms that RSV is associated with these illnesses, then estimates of RSV-associated medical costs would increase significantly.
We did find that the rate of pneumonia hospitalizations increased over the study period, although mortality rates did not. Some but not all of this increase appears to be attributable to aging of the population; the rate of pneumonia in the elderly increases with increasing age. Further study is needed to more carefully define factors that may be contributing to these increased rates of pneumonia hospitalization.
In summary, although our estimate of the burden of RSV disease among the elderly remains incomplete, we can conclude that this burden is substantial and justifies continued development of an RSV vaccine for use in the elderly. Additional studies that examine RSV infection among hospitalized and outpatient elderly with any of several cardiopulmonary illnesses are needed to refine these preliminary estimates.
We gratefully acknowledge Ann Falsey and Scott Dowell for providing data for this analysis; Christine Brower, Robert Holman, David Shay, A. Marshall McBean, and Bill Barker for contributing to this analysis and manuscript; and John O'Connor for editorial review.