Patients, Materials, and Methods

Study data were obtained from 2 national systems, the Active Bacterial Core surveillance (ABCs; a component of the Emerging Infections Program of the Centers for Disease Control and Prevention [CDC]) and the National Health Interview Survey (NHIS). ABCs is an active, population-based surveillance program that has provided national data on pneumococcal disease and other specific bacterial diseases since 1995. In 1999 and 2000, ABCs personnel in 7 states identified cases of invasive pneumococcal disease and collected information on underlying conditions. Surveillance areas included California (3 counties), Connecticut (the entire state), Maryland (6 counties), Minnesota (7 counties), New York (15 counties), Oregon (3 counties), and Tennessee (5 counties). The population under surveillance was estimated by the US Bureau of the Census to be 14.7 million in 1999 and 15.1 million in 2000. A case of invasive pneumococcal disease was defined as an illness in conjunction with the isolation of Streptococcus pneumoniae from a normally sterile site, such as the blood or cerebrospinal, joint, or pleural fluid. Information on demographics, clinical presentation, underlying conditions, and outcome of disease was collected for each case. Underlying conditions on the ABCs case-report form included those listed as indications for polysaccharide pneumococcal vaccine by the Advisory Committee on Immunization Practices [13]. These conditions were individually marked as “yes” if they were noted to be present in the patient’s medical record at the time when treatment for pneumococcal disease was administered. If no underlying conditions were noted in the medical record, “none” was marked on the case-report form. Detailed information on the methods used by ABCs has been reported elsewhere (available at: http://www.cdc.gov/abcs/)

The present study included cases of invasive pneumococcal disease that occurred in 1999 and 2000 in persons ⩾18 years old; cases that occurred in patients in nursing homes or other chronic care facilities were excluded, so that the data would correspond to the NHIS population estimates. The number of cases of invasive pneumococcal disease reported was stratified by 2 race categories (white and black) and by 5 age categories (18–34 years, 35–49 years, 50–64 years, 65–79 years, and ⩾80 years). Cases for which race either was other than black or white or was unknown were excluded from the analysis, because of small sample sizes. Using the US Bureau of the Census data, we standardized, by age and race, the number of cases of invasive pneumococcal disease that occurred in the ABCs areas in adults with no reported underlying conditions and in adults with specific underlying conditions, and we used these data to estimate the number of cases of invasive pneumococcal disease that occurred in the entire US population. The specific underlying conditions evaluated included diabetes, chronic heart disease, chronic lung disease, solid cancer, HIV/AIDS, hematological cancer, and alcohol abuse (table 1). Hematological cancer included multiple myeloma, leukemia, and Hodgkin disease. Solid cancer was defined as any cancer except skin and hematological cancer

The NHIS is designed to address multiple health-related issues and has been conducted annually since 1957 by the National Center for Health Statistics of the CDC. The survey uses a complex multistage sample design, and survey questions assess the health status of noninstitutionalized civilian persons in the United States with respect to select medical conditions (table 1), along with the use of health services and such health-related behaviors as smoking and drinking (available at: http://www.cdc.gov/nchs/). The NHIS data are weighted, to account for the complex sample design and for nonresponse, and provide an estimated number of noninstitutionalized civilian persons with specific underlying conditions in the United States. In the present study, we defined persons who abused alcohol as women who consumed >16 drinks/week and men who consumed >20 drinks/week [14]. As for the ABCs data, leukemia, Hodgkin disease, and lymphoma were categorized as hematological cancer, and solid cancer was defined as any cancer except skin and hematological cancer. The NHIS data were used to estimate the total number of adults with the above conditions in the United States in 1999 and 2000. Although NHIS and ABCs data from 2001 and 2002 were available for use, we did not include these data in the present study, because of the potential effect of childhood conjugate vaccination on rates of invasive pneumococcal disease in adults

The NHIS question on HIV/AIDS asked whether the respondent was highly likely to have HIV/AIDS or had already received a diagnosis. Although the NHIS questionnaire asked about HIV infection and AIDS separately, the final publicly released data file combined the 2 answers. Nevertheless, the estimates of the number of persons with HIV/AIDS in the United States from the NHIS matched those from the Division of HIV and AIDS Surveillance of the CDC [15]. The estimate of the number of healthy adults was arrived at by removing persons who reported having the conditions we evaluated as well as persons who reported having conditions that this study did not consider, such as liver and kidney diseases

The ABCs and NHIS data from 1999 and 2000 were combined to calculate the incidence rates for invasive pneumococcal disease in healthy adults and in persons with any of the conditions evaluated. Data analyses were performed by use of SAS (version 8.2; SAS Institute) and SUDAAN software [16]. Incidence rates were calculated by use of Excel (version 2000; Microsoft). The overall age- and race-specific incidence rates for healthy adults and for persons with each condition evaluated were calculated on the basis of the estimated data for the entire US population, with the NHIS estimates as denominators and the ABCs estimations as numerators. SE estimates for the NHIS estimates were used to calculate the 95% confidence intervals (CIs) for the incidence rates. In addition, Poisson regression was used to estimate adjusted and unadjusted relative risks (RRs) and their respective 95% CIs. The projected population with each condition evaluated was used as an offset in the models. Because the number of persons racially classified as other was small, they were excluded from the univariate and multivariate analyses. Hence, to compute the RRs for adults with each condition evaluated versus healthy adults, the cumulative number of projected cases of invasive pneumococcal disease for white adults and black adults was used in univariate analysis. Multivariate analysis was used to compute the RRs for adults with each condition evaluated versus healthy adults, controlling for race, age, and the other conditions

Results

In 1999 and 2000 combined, 1570 cases of invasive pneumococcal disease were identified in healthy adults, and 2,765 cases were identified in adults with any one of the conditions evaluated (table 2). In 1999 and 2000, >50,000 healthy adults, representing 326 million Americans, participated in the NHIS (table 2), and 9597 adults with 1 or more of the high-risk conditions evaluated participated in the NHIS

Rates of invasive pneumococcal disease in healthy adults  The overall incidence rate was 8.8 cases/100,000 persons in healthy adults (⩾18 years old). Age-specific incidence rates were 3.3, 6.0, 9.9, 23.3, and 71.2 cases/100,000 persons in healthy adults 18–34, 35–49, 50–64, 65–79, and ⩾80 years old, respectively (figure 1). Compared with that in healthy adults 50–64 years old, incidence rates were twice as high in adults 65–79 years old (9.9 vs. 23.3 cases/100,000 persons) and 7-fold higher in adults ⩾80 years old (9.9 vs. 71.2 cases/100,000 persons)

Rates of invasive pneumococcal disease in adults with underlying conditions Incidence rates were higher in adults with chronic illnesses than they were in healthy adults. The rate ratios (healthy vs. ill) varied in magnitude among the conditions evaluated. The incidence of invasive pneumococcal disease ranged from 46.2 cases/100,000 persons in adults with diabetes, to 62.9 cases/100,000 persons in adults with chronic lung disease, and to 93.7 cases/100,000 persons in adults with chronic heart disease (table 2). Compared with that in healthy adults, the incidence rate in adults who abused alcohol was ∼11-fold higher (100.4 cases/100,000 persons). The risk of invasive pneumococcal disease in adults with immunocompromising conditions, such as HIV/AIDS or solid or hematological cancer, was considerably higher, with estimated incidence rates ranging from 300.4 to 503.1 cases/100,000 persons

Incidence rates in persons with chronic illnesses also varied by age (figure 1). For adults with diabetes, chronic heart disease, chronic lung disease, and solid cancer, a higher rate was seen in older adults than in young adults. For adults with chronic heart disease, the incidence rate was significantly lower in those in the 18–34-, 35–49-, and 50–64-year-old age groups than in persons in the older age groups (P<.001)

The incidence rate in adults with diabetes increased with advancing age, reaching 61.5 (95% CI, 57.5–66.0) cases/100,000 persons in those 65–79 years old and 133.8 (95% CI, 118.5–154.0) cases/100,000 persons in those ⩾80 years old. The incidence rate in adults with chronic lung disease was 16.3 cases/100,000 persons (95% CI, 15.0–17.8 cases/100,000 persons) in those 35–49 years old and increased significantly to 57.2 (95% CI, 52.7–62.5), 125.8 (95% CI, 116.4–136.9), and 233.4 (95% CI, 205.5–270.2) cases/100,000 persons in adults 50–64 years old, 65–79 years old, and ⩾80 years old, respectively (P<.001)

High rates of invasive pneumococcal disease occurred in adults with solid cancer or who abused alcohol after the age of 35 years. Incidence rates in adults who abused alcohol were ⩾5-fold higher in those ⩾35 years old than in those 18–34 years old (107.9–168.6 vs. 22.8 [95% CI, 20.4–25.8] cases/100,000 persons). The incidence rate in adults with solid cancer increased from 183.9 (95% CI, 148.4–241.6) cases/100,000 persons in those 35–49 years old to 536.4 (95% CI, 433.1–704.6) cases/100,000 persons in those ⩾80 years old

Incidence in adults 18–34 years old was highest in those with hematological cancer and in those with HIV/AIDS—136.8 (95% CI, 85.2–346.9) cases/100,000 persons and 215.6 (95% CI, 178.3–272.4) cases/100,000 persons, respectively. Incidence was even higher in adults with hematological cancer after age 35 (457.9–673.4 cases/100,000 persons). The peak incidence in adults with HIV/AIDS occurred between the ages of 35 and 49 years (673.7 [95% CI, 571.3–820.7] cases/100,000 persons)

Racial differences in rates of invasive pneumococcal disease In healthy adults <65 years old, incidences were 4–7-fold higher in black adults than in white adults. In adults ⩾65 years old, incidences were approximately twice as high in black adults than they were in white adults (figure 2). The disparity in incidence was noted consistently for all conditions evaluated (table 3). Compared with white adults with the respective conditions, black adults with solid cancer or hematological cancer had a 2–3-fold higher incidence, black adults who abused alcohol had a 5-fold higher incidence, and black adults with HIV/AIDS had a 7-fold higher incidence. Racial differences in incidence were smaller in adults with chronic heart or lung disease or with diabetes

Rates of invasive pneumococcal disease in persons with multiple conditions Incidence rates increased with the number of conditions present (table 2). Incidence rates increased from 52.3 (when HIV/AIDS and hematological cancer were not included) and 87.5 (when HIV/AIDS and hematological cancer were included) cases/100,000 persons in adults with ⩾1 condition to 104.5 and 128.4 cases/100,000 persons, respectively, in adults with ⩾2 conditions. Inclusion of HIV/AIDS and hematological cancer had more of an effect on incidence rates in black adults than in white adults; when these conditions were included, rates increased (1) from 47.4 to 75.0 cases/100,000 persons in white adults versus 109.6 to 216.6 cases/100,000 persons in black adults for those with ⩾1 condition and (2) from 105.6 to 120.8 cases/100,000 persons in white adults versus 132.3 to 234.6 cases/100,000 persons in black adults for those with ⩾2 conditions

Multivariate analysis The effects of chronic illnesses on invasive pneumococcal disease were determined by adjusting for age, race (black or white), and the other underlying conditions evaluated in the present study. Compared with that in healthy adults, RRs for invasive pneumococcal disease were 3–6-fold higher in adults with diabetes or with chronic heart or lung disease, 11-fold higher in adults who abused alcohol, and 23–48-fold higher in adults with solid cancer, hematological cancer, or HIV/AIDS (table 2). RRs were 4–7-fold higher in adults with ⩾1 condition and were 8–10-fold higher in adults with ⩾2 conditions, compared with that in healthy adults. In addition, black adults had a 2.8-fold (95% CI, 1.9–4.3-fold) higher rate of invasive pneumococcal disease than white adults (P<.0001), and adults ⩾50 years old had a 2.9-fold (95% CI, 1.9–4.3-fold) higher rate than adults <50 years old (P<.0001)

Discussion

The present findings quantify the magnitude of risk of invasive pneumococcal disease in adults with various underlying conditions and demonstrate that these conditions are strongly associated with the occurrence of disease. Rates of invasive pneumococcal disease in adults with chronic illnesses and in those who abused alcohol ranged from 51.4 to 100.4 cases/100,000 persons; in adults with immunocompromising conditions, the rates ranged from 300.4 to 503.1 cases/100,000 persons. The risk of disease increased with advancing age, and it was also increased in older persons with diabetes, chronic lung disease, or solid cancer. The present study evaluated rates of disease in healthy adults and the specific effects of age, race, and underlying conditions on the incidence of disease, and our data have reinforced previous findings showing that there are race and age differences in the epidemiologic profile of invasive pneumococcal disease [2, 3]

Our findings indicated that adults with diabetes, chronic heart disease, or chronic lung disease had a 3–6-fold increased risk of invasive pneumococcal disease, compared with healthy adults (this finding was true regardless of whether age, race, and the other conditions evaluated were controlled for). In general, the effect of alcohol abuse on the risk of invasive pneumococcal disease was more dramatic after the age of 35 years. The incidence in adults who abused alcohol was 11-fold higher than it was in healthy adults. However, a case-control study conducted in immunocompetent adults 18–64 years old found that the risk of invasive pneumococcal disease was ∼3-fold higher in persons with diabetes, heart failure, or chronic obstructive pulmonary disease and was ∼7-fold higher in adults who abused alcohol than in healthy adults, without controlling for other factors [17]. Chronic heart disease had a stronger effect on rates in young adults and in very old adults. The effect of diabetes on the incidence of invasive pneumococcal disease was observed to gradually increase with increasing age. The role played by chronic lung disease in the risk of invasive pneumococcal disease was evident after the age of 50, with a greater effect in persons in older age groups. The incidence of invasive pneumococcal disease found in the present study was lower than previous findings for those with chronic lung disease (503 cases/100,000 persons) [1] and for those with chronic heart disease or diabetes (195 and 176 cases/100,000 persons) [18], likely because of differences in databases and study methods. Higher rates of disease in older adults with diabetes, chronic heart disease, or chronic lung disease suggest that the effect of these illnesses on the risk of disease was much stronger in adults in the older age groups, perhaps because of increasing severity of the underlying condition over time. Given how common diabetes, heart disease, and alcoholism are in the US population, these conditions are major contributors to pneumococcal disease burden. In the present study, having >1 chronic illness increased the risk of disease, suggesting that multiple conditions have a cumulative effect

Adults with severely immunocompromising conditions, including solid or hematological cancer or HIV/AIDS, had very high incidences of invasive pneumococcal disease—23–48-fold higher than the risk of disease in healthy adults. In the present study, the overall incidence in adults with HIV/AIDS was 422.9 cases/100,000 persons, with the highest incidence in adults 35–49 years old (673.7 cases/100,000 persons). Other studies have found higher rates for persons with HIV infection or AIDS. In a study from the era before highly active antiretroviral therapy (HAART) was available, it was found that persons with AIDS had an incidence of pneumococcal bacteremia of 940 cases/100,000 persons [6]; in another study, the incidence of invasive pneumococcal disease in those with HIV infection was found to be 587 cases/100,000 persons in 1995 [1]. A study that spanned the introduction of HAART found incidences of 802.9 cases/100,000 persons in patients with AIDS and 15.7 cases/100,000 persons in those without known HIV infection [9]. We also found that the effect of HIV infection on rates of invasive pneumococcal disease was higher in black adults than in white adults, which is consistent with earlier findings from San Francisco [9]. The true incidence in adults with HIV/AIDS could be somewhat higher than what we measured, because the number of adults with HIV/AIDS in the NHIS included both those who were already infected with HIV and those with a high chance of becoming infected. A recent study also found higher rates of invasive pneumococcal disease in persons with nonhematological cancer (216 cases/100,000 persons) and in persons with hematological cancer (734 cases/100,000 persons) [18]. Our data, along with the increase in the number of persons with HIV infection in recent years, emphasize the disproportionate risk of invasive pneumococcal disease in persons with immunocompromising conditions

By evaluating the risk of disease in healthy adults, we were able to evaluate the influence of age and race without the confounding effects of chronic illnesses. Previous studies that did not separately evaluate incidence in healthy and chronically ill adults showed that the incidence of invasive pneumococcal disease increased after the age of 65 years [19]. Our findings demonstrate that age itself is an independent risk factor for the disease, likely because of the decreased function of the immune system [20]. It also has been reported that black adults in the United States have higher rates of invasive pneumococcal disease [1, 2] and are more likely to die of pneumococcal disease than white adults, regardless of underlying illnesses [3, 21]. As has been suggested by a previous case-control study [17], we found that rates of disease in healthy adults were 3–4-fold higher in black adults 18–49 years old than in white adults in the same age groups. Although racial differences in the incidence of disease became narrower after the age of 65, the gap in incidence between black adults and white adults persisted. Whether genetic susceptibility or socioeconomic, environmental, or behavioral risk factors contribute to this difference in incidence is unknown. The introduction of a 7-valent pneumococcal conjugate vaccine for children is reducing the racial disparity in pneumococcal disease [22]

The present study used a unique combination of data from 2 national databases to assess the relationships between specific conditions and the incidence of invasive pneumococcal disease in adults in the United States. A few limitations exist and merit consideration when our results are interpreted. National estimates derived from NHIS and ABCs are comparable, and it is likely reasonable to use them to measure the incidence of disease for several conditions. However, these 2 systems differ in their approach to data collection and in the wording of the questions on these conditions. Thus, the numerators and denominators may not match exactly for some conditions, such as alcohol abuse, cancer, or HIV/AIDS. Such differences could lead to some over- or underestimation of incidence rates. For some potentially important conditions, such as smoking and asthma, we were unable to estimate risk because of the substantial differences in definitions between the 2 systems

Previously, smoking was found to be independently associated with the risk of invasive pneumococcal disease in adults 18–64 years old [17]. Because we were not able to adjust for smoking in our analysis, it could be a confounding factor in our observed associations. In addition, the present study did not account for the potential effects of variations in coverage of pneumococcal polysaccharide vaccine and its effectiveness in the studied conditions

The 23-valent pneumococcal polysaccharide vaccine is underutilized in high-risk persons who may benefit from it [23]. The coverage of vaccine in persons with such chronic illnesses as diabetes, HIV infection, or AIDS is limited [7, 23–24]. It has been documented that polysaccharide pneumococcal vaccine is 65%–84% effective against invasive disease in persons with chronic illnesses [25], but studies have suggested that its protective effect is limited in persons with severe immunocompromising conditions [24 –27]. Nevertheless, the efficacy of polysaccharide pneumococcal vaccine remains controversial [28]. The extraordinarily high rates of invasive pneumococcal disease in persons with immunocompromising conditions highlight the need for more-effective prevention strategies. Conjugate pneumococcal vaccine produces somewhat stronger immune responses than does polysaccharide vaccine in patients with HIV infection or sickle-cell disease [29] and was recently reported to have 65% efficacy against invasive disease in HIV-infected infants in South Africa [30]. Thus, conjugate vaccines may have the potential to reduce the incidence of disease in this high-risk group. In the United States, there was a 15%–46% reduction of invasive disease in adults ⩾20 years old after the introduction of 7-valent conjugate pneumococcal vaccine [31], suggesting that widespread use of this vaccine in young children offers benefits to adults through reduced transmission. Whether these benefits are being seen in the highest-risk group, adults with immunocompromising conditions, is unknown. For now, vaccination against invasive pneumococcal disease remains an important part of preventive strategies for adults with chronic conditions

Active Bacterial Core Surveillance Team

Anne Schuchat (Respiratory Diseases Branch, Division of Bacterial and Mycotic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, GA), Ruth Lynfield (Minnesota Department of Health, Minneapolis), James Hadler (Connecticut Department of Public Health, Hartford), Paul R. Cieslak (Oregon Department of Human Services, Health Division, Portland), Lee H. Harrison (Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD), Nancy M. Bennett (Monroe County Department of Health and University of Rochester, Rochester, NY), Arthur Reingold (School of Public Health, University of California at Berkeley, Berkeley), and Allen Craig (Tennessee Department of Health, Nashville)