Infectious Disease Hospitalizations Among Older Adults in the United States From 1990 Through 2002 FREE

Adults 65 years or older (older adults) account for a disproportionate share of patients with infectious disease (ID)-related hospitalizations and all-cause hospitalizations in the United States.^1- 3 Rates of ID hospitalization increased among older adults from 1980 through 1994, while rates for the population younger than 65 years decreased during the same period.¹ The proportion of all-cause hospitalizations accounted for by older adults increased from 20% in 1970 to 38% in 2002,³ and ID mortality rates increased by 25% among older adults from 1980 through 1992.⁴

Currently, older adults represent approximately 13% of the US population, and the number of older adults is expected to increase from 35 million in 2000 to 69 million by 2030, accounting for approximately one fifth of the total US population.^5- 6 The growth of the older population represents an enormous challenge for the US health care system in terms of health expenditures and capacity. In 2003, Medicare expenditures exceeded $280 billion, and they are projected to be more than $700 billion by 2013.⁷ Given the increasing effect of older adults on US health care system resources in the near future, it is important to understand how IDs contribute to morbidity and mortality among older adults.

To better understand conditions associated with substantial morbidity among older adults, we describe hospitalization rates and trends for all IDs combined and for specific ID groups among older adults in the United States from January 1, 1990, through December 31, 2002.

The National Hospital Discharge Survey (NHDS) was used to generate hospitalization estimates from 1990 through 2002 for the general US population of older adults.⁸ The NHDS is a national probability survey that collects data from a sample of approximately 270 000 inpatient records obtained from about 500 hospitals annually by using a combination of manual hospitalization record abstraction and machine readable medical record data.⁹ The NHDS is designed to provide essentially unbiased national estimates of hospitalization characteristics occurring in nonfederal short-stay hospitals in the United States.

The ID hospitalizations were identified by means of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes listed on hospital discharge records.¹⁰ A maximum of 7 ICD-9-CM diagnoses are recorded on the NHDS discharge record. Unless otherwise noted, a hospitalization with an ID code as the first discharge diagnosis was considered an ID-related hospitalization. By using a comprehensive list of ICD-9-CM codes associated with IDs and related conditions,¹ we identified all ID-related hospitalizations and identified 13 specific ID categories for analysis (Table 1). The ID groups with too few sampled discharges for reliable US estimates (eg, human immunodeficiency virus and AIDS, viral central nervous system infection) were placed in the other infectious diagnoses ID category. Conditions such as infections and inflammatory reactions to prosthetic devices, which are not IDs per se, were included as IDs in our analyses.

The number of hospitalizations and hospitalization rates were evaluated according to year; according to age group (65-74, 75-84, and ≥85 years), sex, and US census region (Northeast, Midwest, South, and West) for patients; and according to 2 periods (January 1, 1990, through December 31, 1992, and January 1, 2000, through December 31, 2002). Aggregate estimates of discharges according to month were made by using records only from hospitals that fully reported to the NHDS for all 12 months of each calendar year and by adjusting the discharge weights for those hospitals to generate US estimates. Rates were expressed as the number of hospitalizations per 10 000 adults 65 years or older. Denominators for rate calculations were obtained by using US resident population estimates for 1990 through 2002.¹¹ Age-adjusted rates were calculated by means of the direct method, and the 2000 US resident population 65 years or older was used as the standard population.¹² Length of hospital stay was also determined and compared according to region, age group, and sex.

Rates were compared according to time and demographic characteristics by using 2-sided t tests. A t test based on ranks was used to compare median lengths of stay, and a weighted Cochran-Mantel-Haenszel χ² test was used to compare proportions.^3,13- 14 For all statistical comparisons, the standard error (SE) of estimates was incorporated before assessing statistical significance at P<.05.¹³ The SE is used as a measure of the sampling variability since the NHDS is a sample rather than a census of 100% of hospitalizations occurring in the United States.⁹

There were approximately 21.4 million (SE, 636 000) ID hospitalizations among older adults from 1990 through 2002. These hospitalizations with an ID as the primary diagnosis accounted for 48% (SE, 0.4%) of all hospitalizations with an ID diagnosis on the discharge record. Between 1990 through 1992 and 2000 through 2002, the ID hospitalization rate increased 13% from 449.4 to 507.9 hospitalizations per 10 000 older adults (P = .01) (Table 2 and Figure 1A). For the same 2 periods, the ID hospitalization rate for women increased significantly from 428.9 to 503.5 per 10 000 (P<.001), while the ID hospitalization rate for men did not change significantly (479.9 and 514.2 per 10 000) (Table 2). On a regional basis, only the ID hospitalization rate for the Northeast increased significantly, with rates of 434.2 and 560.8 per 10 000 older adults for 1990 through 1992 and 2000 through 2002, respectively (P<.001).

With the exception of women aged 65-74 years, the age-specific ID hospitalization rates did not increase significantly during the study (Table 2), and the age-adjusted ID hospitalization rate for older adults did not differ significantly between the 2 periods (475.2 and 505.0 per 10 000 in 1990 through 1992 and 2000 through 2002, respectively, P = .19). These findings indicate that the 13% increase observed in the overall older adult ID hospitalization rate resulted in part from the relative population growth of the older age groups (ie, 75-84 and ≥85 years) and the increasing influence of their higher ID hospitalization rates on the overall older adult ID hospitalization rate during the study. Between 1990 through 1992 and 2000 through 2002, the population of older adults aged 75 through 84 years increased by 22%, and the population aged 85 years or older increased by 39%, while the population aged 65 through 74 years increased by only 0.3%.¹¹

From 2000 through 2002, the ID hospitalization rate was highest among persons 85 years or older; the rate among this group was almost double that for persons aged 75 through 84 years and triple that for persons aged 65 through 74 years (1122.4 vs 587.6 and 304.9 per 10 000, respectively, P<.001 both comparisons) (Table 2). For 2000 through 2002, the ID hospitalization rates for men aged 75 through 84 years and 85 years or older were higher compared with those for their female counterparts: 649.9 vs 546.7 per 10 000 for those aged 75 through 84 years (P<.001) and 1262.8 vs 1063.2 per 10 000 for those 85 years or older (P = .004). According to region, the West had a significantly lower ID hospitalization rate among older adults than any of the other 3 US census regions (P<.001) (Table 2). The ID hospitalization rates were similar for the Northeast, Midwest, and South.

The all-cause hospitalization rate for older adults increased by nearly 6% between 1990 through 1992 and 2000 through 2002, but this increase was not statistically significant (P = .19) (Table 3 and Figure 1B). Of all hospitalizations among older adults, the proportion associated with ID did not differ significantly between the 2 periods (P = .09), with 13.4% (SE, 0.3%) during 1990 through 1992 and 14.3% (SE, 0.2%) during 2000 through 2002. From 2000 through 2002, ID hospitalizations composed 12% (SE, 0.2%), 14% (SE, 0.2%), and 19% (SE, 0.4%) of all hospitalizations among those aged 65 through 74 years, 75 through 84 years, and 85 years or older, respectively. These proportions were similar according to sex of patients.

The ID hospital fatality rate decreased from 8 deaths (SE, 0.3) per 100 hospitalizations in 1990 through 1992 to 7 deaths (SE, 0.2) per 100 hospitalizations in 2000 through 2002 (P<.001). During the last 3 years of the study, hospital fatality rates among older adults were similar according to sex. Hospital fatality rates from 2000 through 2002 were higher with increasing age group, with 5 deaths (SE, 0.4) per 100 ID hospitalizations among those aged 65 through 74 years, 7 (SE, 0.2) per 100 among those aged 75 through 84 years, and 9 (SE, 0.4) per 100 among those 85 years or older. The non-ID hospital fatality rate of 4 deaths (SE, 0.1) per 100 non-ID hospitalizations was significantly lower than the ID-related hospital fatality rate (P<.001).

The ID hospitalization rate trends for older adults varied according to ID group from 1990 through 2002. The hospitalization rates for lower respiratory tract infections (LRTIs) and kidney, urinary tract, and bladder (KUB) infections did not change significantly (Table 3); however, the rate for septicemia increased 22% from 50.4 to 61.7 hospitalizations per 10 000 older adults (P = .001) from 1990 through 1992 to 2000 through 2002. Other ID groups with hospitalization rates that increased significantly included infections of the heart, cellulitis, enteric infections, infections and inflammatory reactions to prosthetic devices, and postoperative infections. The most dramatic increases in ID hospitalization rates were for infections of the heart, infections and inflammatory reactions to prosthetic devices, and postoperative infections, with increases of approximately 240%, 130%, and 80%, respectively. Although the hospitalizationrates for infections of the heart increased significantly for all older adults, the largest increases were among the older age groups. Among older adults 85 years or older, the hospitalization rate increased by more than 400% from 4.1 to 21.2 per 10 000 (P<.001) and by nearly 250% for those aged 75 through 84 years from 3.1 to 10.8 per 10 000 (P<.001). Similarly, the postoperative hospitalization rates doubled for both those aged 75 through 84 years and those 85 years or older (5.8-11.6 and 5.2-10.6 per 10 000, respectively; P<.001 both age groups). In contrast, the increase in rates for infections and inflammatory reactions to prosthetic devices was highest among those aged 65 through 74 years, followed by those aged 75 through 84 years, and those 85 years or older (5.5-15.5 per 10 000 [P<.001], 8.6-17.1 per 10 000 [P<.001], and 10.8-18.3 per 10 000 [P = .04], respectively). The only group of IDs that decreased significantly during the study was upper respiratory tract infections (URTIs), which decreased by almost 40% from 5.9 to 3.6 hospitalizations per 10 000 older adults (P = .008).

Almost half (46% [SE, 0.7%]) of ID hospitalizations among older adults were associated with LRTIs from 2000 through 2002, and LRTIs were associated with 48% (SE, 1.6%) of ID-related hospital deaths (Figure 2). Pneumonia (ICD-9-CM codes 480-486) accounted for the majority of LRTI hospitalizations (93% [SE, 0.5%]) and LRTI-related deaths (97% [SE, 0.8%]) among hospitalized older adults. The next most common ID group was KUB infections, which were associated with 16% (SE, 0.5%) of ID hospitalizations. Women were disproportionately represented in the KUB ID group, accounting for 70% (SE, 1.0%) of these hospitalizations. The KUB infections accounted for 6% (SE, 0.7%) of ID-related hospital deaths. Septicemia accounted for 12% (SE, 0.4%) of ID hospitalizations and for 34% (SE, 1.5%) of ID-related hospital deaths.

In 2000 through 2002, ID hospitalizations associated with LRTIs among older adults peaked in January (Figure 3). In contrast, non–LRTI-related ID hospitalizations and non-ID hospitalizations were evenly distributed among the months. The seasonal patterns for ID hospitalizations were consistent according to age group and region.

The median length of stay for ID hospitalizations among older adults decreased from 7 days in 1990 through 1992 to 5 days for both men and women in 2000 through 2002 (P<.001). From 2000 through 2002, the median length of stay was also 5 days for each age group and for each region (except the Midwest, where it was 4 days). The median length of stay for non-ID hospitalizations among older adults was significantly shorter (4 days; P<.001) than that for ID hospitalizations from 2000 through 2002. In addition, 34.3 million hospitalization days (SE, 1 089 000) were associated with ID among older adults in 2000 through 2002, and this number represented about 16% of all hospitalization days among older adults during that period.

The ID hospitalization rates in the United States increased slightly among older adults during the 13-year study, and this increase contrasted with the decreasing ID rates among persons younger than 65 years from 1980 through 1994^1- 2 and among US infants from 1988 through 1999.¹⁵ The ID hospitalization rate was highest among persons 85 years or older, for whom the rate was nearly double that for persons aged 75 through 84 years and nearly triple that for those aged 65 through 74 years. The oldest age group also had the highest ID hospital fatality rate, a finding presumably reflective of age-related poorer overall health among hospitalized persons in this age group. Between 1990 through 1992 and 2000 through 2002, the population of older adults aged 75 through 84 years increased by 22%, and the population 85 years or older increased by 39%, while the population aged 65 through 74 years increased by 0.3%.¹¹ The steady ID hospitalization rates among the older age groups coupled with their population growth during the study contributed in part to the increase in the overall ID hospitalization rate because these age groups had much higher rates of ID hospitalization than those aged 65 through 74 years and accounted for an increasing proportion of the older adult population by the end of the study. This aging of the older adult population suggests that the proportion of ID hospitalizations, as well as all-cause hospitalizations, attributable to adults 75 years or older will increase rapidly during the coming years.

Almost half of the ID hospitalizations and the ID-related hospital deaths among older adults were associated with LRTIs. Consistent with previous study results describing LRTI hospitalization rates among all age groups,¹ LRTI hospitalization rates among older adults remained stable during our study. Septicemia was associated with 12% of hospitalizations but accounted for about one third of ID-related hospital deaths from 2000 through 2002. The KUB infections also accounted for a substantial percentage of ID hospitalizations and disproportionately affected women. We found that LRTI, KUB infections, and septicemia are common causes of ID morbidity and mortality among older adults, which suggests that infection-control measures should be focused on these ID groups.

The rates for the ID groups of infections of the heart, infections and inflammatory reactions to prosthetic devices, and postoperative infections increased dramatically during the study. Technological and medical advances have been made during the past several years in invasive surgical procedures, indwelling medical devices, and cardiac surgery, resulting in concomitant increases in the performance of these procedures.^16- 18 For example, the rate of invasive coronary artery surgeries among all older adults was 1.5 times higher in 2002 compared with the rate in 1995 and 2.5 times greater among persons 85 years or older for the same 2-year comparison.^19- 20 The rate of knee replacement surgery in 2002 was more than 40% higher compared with the 1995 rate among all older adults and was almost 70% higher during 2002 among persons 85 years or older compared with the rate in 1995.^19- 20 More frequent performance of such procedures and the infection risk inherent with them likely have led, in part, to increased rates of occurrence for infections in these ID categories. With the increasing frequency of these procedures, future efforts should focus on identifying those at highest risk for such infections and identifying areas in which standard preventive and treatment measures could be improved. In addition, the increasing prevalence of antimicrobial-resistant organisms, both within the community and in health care settings, may have contributed to the increases in these ID groups, particularly in the postoperative and prosthetic infections groups.^21- 22 The prevention and treatment of health care–acquired and postoperative infections remain an important goal in ID control.

In contrast to the hospitalization rate for other ID groups, the rate for URTIs decreased during the study. The reasons for the decrease in hospitalization rates for URTIs among older adults are unknown but could include changes in diagnostic coding and reimbursement issues, the result of better vaccine coverage for vaccine-preventable respiratory infections, or the moving of care for these conditions to the outpatient setting. However, there is evidence contrary to movement of care for URTIs to the outpatient setting, since the rate of physician visits with URTIs as a primary diagnosis among older adults was 9.1 per 100 older adults in 1995 and 9.2 per 100 older adults in 2002.²³

We used a large robust national sample of hospitalizations to generate estimates of ID hospitalizations among older adults in the United States. First-listed ID hospitalizations accounted for almost half of all hospitalizations with an ID diagnosis listed on the discharge record. By using the first-listed diagnosis to identify ID hospitalizations, we can be reasonably certain that an ID was the primary cause of hospitalization. Interpretation of hospitalizations with an ID as a secondary diagnosis is problematic due to the inability to determine if the ID contributed to the patient’s admission or if the ID was acquired during hospitalization. Future studies should focus on specific ID groups in more detail to determine the extent of their effect on hospital admission and subsequent discharge, as well as their effect on the overall health of older adults. In addition, because of the inability to track hospitalizations among individuals, we were unable to estimate the number of ID-related readmissions among older adults. Assessing the magnitude of readmissions associated with ID among individual older adults should be a priority for further study.

The population of older adults is expected to account for 20% of the US population by 2030; the oldest age group (ie, those aged ≥85 years) is expected to double by 2030.⁶ All-cause hospitalization rates remained stable, and ID hospitalization rates increased slightly during the study. Because of the anticipated population growth among older adults, steady or increasing hospitalization rates among older adults in the near future will result in a substantial increase in the sheer number of older adults being hospitalized. The IDs account for a substantial proportion of all hospitalizations among older adults. In particular, LRTIs compose the largest proportion of ID hospitalizations and ID hospitalization deaths, and continued efforts to increase the coverage of influenza and pneumococcal vaccines, as well as the development of new and improved vaccines for respiratory pathogens, could reduce ID morbidity among older adults.^24- 29 In addition, improving current antimicrobial therapy strategies and developing novel antimicrobial agents to address antimicrobial resistance could also reduce morbidity and mortality due to IDs.^22,30- 31 Finally, by encouraging preventive measures such as improving diet and nutrition, increasing physical activity, and better control of chronic medical conditions such as cardiovascular disease, cerebrovascular disease, and diabetes mellitus, risk factors predisposing some older adults for increased ID morbidity and mortality could be reduced.^32- 36 The reduction of ID hospitalization rates among older adults could help attenuate the anticipated increase in the number hospitalizations among older adults and should be a high priority given the projected population growth among older adults in the United States.

Correspondence: Aaron T. Curns, MPH, Centers for Disease Control and Prevention, 1600 Clifton Rd, Mail Stop A-39, Atlanta, GA 30333 (acurns@cdc.gov).

Accepted for Publication: July 6, 2005.

Financial Disclosure: None.

Author Contributions: Mr Curns had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Acknowledgment: We thank Claudia Chesley, BA (Division of Viral and Rickettsial Diseases, Centers for Disease Control and Prevention), for manuscript review and editing assistance.