Original Studies
Socioeconomic and Racial Disparities of Pediatric Invasive Pneumococcal Disease After the Introduction of the 7-valent Pneumococcal Conjugate Vaccine Jennifer O. Spicer, MD, MPH,* Stephanie Thomas, MPH,† Amy Holst, MPH,† Wendy Baughman, MPH,‡ and Monica M. Farley, MD*‡ Background: Racial differences have been well described for invasive pneumococcal disease (IPD), but little information exists on how race interacts with community socioeconomic factors. Methods: The Active Bacterial Core surveillance/Emerging Infections Program performed active surveillance for IPD in the 20-county Metropolitan Atlanta area. All IPD cases among children younger than 5 years from 2001 to 2009 were geocoded and linked to census tract-level socioeconomic measures from the 2000 US Census. Race- and socioeconomic-specific average annual incidence rates per 100,000 population were calculated. Trends in IPD incidence were determined by χ2 tests for trend. Rate ratios (RRs) and 95% confidence intervals (CIs) were estimated using Poisson regression. Results: IPD incidence among the total population of children increased as percentage of household poverty increased (P = 0.002), as median household income decreased (P < 0.001), as wealth decreased (P = 0.018) and as percentage of individuals with less than a high school education increased (P = 0.023). After stratifying by race, there was no significant linear trend between socioeconomic characteristics and IPD incidence among white children; among black children, however, IPD incidence decreased as socioeconomic conditions worsened. Despite adjusting for sex and socioeconomic factors, the IPD rate remained higher among black children compared with white children (RR = 1.60; 95% CI: 1.39–1.84). Differences in RR of IPD associated with highest poverty and lowest wealth noted in 2001 [RR = 2.71 (95% CI: 2.17–3.39) and 1.80 (95% CI: 1.09–2.96), respectively] declined in 2009 [RR = 1.33 (95% CI: 0.90–1.96) and 0.76 (95% CI: 0.48–1.19), respectively]. Conclusions: Although socioeconomic disparities in IPD incidence exist among children, the association between socioeconomic characteristics and IPD rates may differ by race and may change over time. Community-level socioeconomic factors did not account for racial differences in IPD incidence. Key Words: Streptococcus pneumoniae, invasive pneumococcal disease, socioeconomic disparities, racial disparities (Pediatr Infect Dis J 2014;33:158–164)
Accepted for publication August 22, 2013. From the *Department of Medicine, Emory University School of Medicine, Atlanta, GA; †Department of Medicine, Atlanta Research and Education Foundation, Decatur, GA; and ‡Department of Medicine, Atlanta Veterans Affairs Medical Center, Decatur, GA. This study was funded in part by the Centers for Disease Control and Prevention’s Emerging Infections Program, Atlanta, GA. The authors have no conflicts of interest to disclose. This work was presented in part at the 52nd Annual Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, CA, September 2012. Address for correspondence: Monica M. Farley, MD, Department of Medicine, Atlanta VA Medical Center, 1670 Clairmont Rd, Decatur, GA 30033. E-mail: [email protected]. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (www.pidj.com). Copyright © 2013 by Lippincott Williams & Wilkins ISSN: 0891-3668/14/3302-0158 DOI: 10.1097/INF.0000000000000025
158 | www.pidj.com
S
treptococcus pneumoniae is the leading cause of pneumonia and meningitis in the United States, resulting in significant morbidity, mortality and healthcare costs.1 Known risk factors for pediatric invasive pneumococcal disease (IPD) include specific racial and ethnic groups,2–6 group childcare and a number of chronic medical conditions.7–9 The introduction of the childhood heptavalent pneumococcal conjugate vaccine (PCV)7 in 2000 resulted in a significant reduction in IPD among children. Although racial differences between black and white children also decreased,8,10–13 these differences continue to exist for unclear reasons. Socioeconomic factors, such as income, poverty, wealth, education and crowding, have been increasingly recognized in the literature as contributing to health disparities.14,15 It has been suggested that routine health reporting should analyze socioeconomic and racial health disparities jointly and separately to better understand health disparities.16 Until recently, socioeconomic variables were infrequently included in epidemiological analyses because most public health surveillance systems do not have the capacity to routinely collect socioeconomic information on all the cases that they identify. The Public Health Disparities Geocoding Project has addressed this issue by identifying community-level socioeconomic information routinely collected by the US Census Bureau that can be linked to cases in public health surveillance systems based on address information for the cases.17,18 Previous studies examining the association between socioeconomic factors and IPD rates found variable results across age groups and racial categories.6,11,13,14,19,20 Only 2 studies have presented specific results for pediatric populations.5,9 The first study was conducted prior to the introduction of PCV7 and found that race-adjusted IPD rates were highest among low-income children.5 The other study was a case-control study completed after the introduction of PCV7, and it showed that low income was associated with an increased risk of PCV7-type IPD among unvaccinated children but not among vaccinated children.9 Neither study analyzed the effect of socioeconomic factors within racial groups. Since the introduction of PCV7, there have been no studies describing the epidemiology of socioeconomic and racial health disparities in IPD among children. The objectives of this study were to describe the association between census tract-level socioeconomic measures and IPD in children younger than 5 years in the post-PCV7 era, to assess whether the association varies by race, to determine whether socioeconomic measures account for racial differences in IPD incidence and to establish whether socioeconomic disparities in IPD incidence have decreased since the introduction of PCV7.
METHODS Study Population The Centers for Disease Control and Prevention–sponsored Active Bacterial Core surveillance of the Georgia Emerging Infections Program conducted active population-based laboratory surveillance for IPD in the 20-county Metropolitan Atlanta area. Cases
The Pediatric Infectious Disease Journal • Volume 33, Number 2, February 2014
The Pediatric Infectious Disease Journal • Volume 33, Number 2, February 2014
were identified through weekly calls and periodic audits of clinical laboratories. Clinical and demographic information on each case was collected from medical records, clinical laboratories or the patient’s physician. Vaccination status was obtained through use of the Georgia Immunizations Registry and telephone interviews with parents and healthcare professionals as described elsewhere.21 For this analysis, the study population included children younger than 5 years with IPD from 2001 to 2009. The time period from 2001 to 2009 was chosen to evaluate socioeconomic influences after the introduction of PCV7 but prior to the introduction of PCV13. IPD was defined as isolation of S. pneumoniae from a normally sterile body site (blood, cerebrospinal fluid, pleural fluid, peritoneal fluid, pericardial fluid, joint, muscle, bone or other internal body site).
Classification of Pneumococcal Serotypes S. pneumoniae isolates were serotyped at the Centers for Disease Control and Prevention using the Quellung reaction. PCV7 serotypes included serotypes 4, 6A, 6B, 9V, 14, 18C, 19F and 23F. All other serotypes were classified as non-PCV7 serotypes.
Census Tract Socioeconomic Variables Area-based socioeconomic measures for each census tract were downloaded from the US Census 2000 Summary 3 File,22 which contains population and household data from the census long form questionnaire. Because this information for our time period was only collected in the decennial census, data from the 2000 US Census was used for all cases in this analysis. After 2000, the US Census Bureau only collected housing, financial and socioeconomic information in the annual American Communities Survey rather than the decennial census, so socioeconomic information was not available from the 2010 US Census for use in this analysis. To determine whether changes in socioeconomic characteristics from 2001 to 2009 could have affected the results of our analysis, socioeconomic and population data were obtained from the compiled 2005–2009 American Community Survey 5-year estimates23 to provide more recent socioeconomic data. Five-year compilations are required to provide accurate estimates for small geographic areas, such as the census tracts. Rate ratio (RR) calculations using both the 2000 US Census data and the 2005–2009 American Community Survey socioeconomic and population denominators were included for comparison to detect any changes in socioeconomic determinants in the surveillance population from 2001 to 2009 (shown in Table 3). Variables were created to represent census tract-level measures of poverty, wealth, income, education and crowding using categories recommended by the Public Health Disparities Geocoding Project.24 Census tract poverty was defined as the percentage of households in the census tract with incomes below the federally defined poverty level and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. Census tract wealth was defined as the percentage of owner-occupied homes worth ≥400% of the US median value of owned homes and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. Census tract median household income was categorized as quintiles: $0– $35,416, $35,416–$44,329, $44,330–$52,935, $52,935–$67,946 and >$67,946. Census tract education was defined as the percentage of persons in each census tract of age ≥25 years with less than a 12th grade education, including those who completed 12th grade but did not receive a degree, and was divided into 4 categories: 0%–14.9%, 15.0%–24.9%, 25.0%–39.9% and >40%. Census tract crowding was defined as the percentage of households in each census tract with ≥1 persons per total number of rooms in the residence and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. © 2013 Lippincott Williams & Wilkins
Social Factors and Pneumococcal Disease
Data Analysis Demographic and clinical characteristics of IPD cases of black race and other race were compared with cases of white race using the χ2 test or Fisher exact test. The race of each case was obtained from medical records and classified as white, black or other. Other race included Asian, American Indian/Alaskan Native and Native Hawaiian/Other Pacific Islander races. For those cases missing race (12.3%), race was assigned according to the race that represented many children younger than 5 years within the case’s census tract because analyses excluding individuals with missing race yielded similar results to the final analyses. The methods for data analysis were based on those of the Public Health Disparities Geocoding Project.24 Using addresses obtained from the Georgia Emerging Infections Program case report forms, each case in this analysis was linked to a census tract from the 2000 US Census using ArcView/ArcGIS 10.1 (ESRI, Redlands, CA). Average annual incidence rates were calculated by dividing the average number of cases per year from 2001 to 2009 by the population of children younger than 5 years in the surveillance area according to the 2000 US Census, and rates were stratified by race and socioeconomic categories. The χ2 test for trend was used to analyze trends in incidence for each socioeconomic variable. Poisson regression was used to calculate RRs and 95% confidence intervals (CIs). For Poisson regression, the dependent variable was census tract IPD rates, and the independent variables were sex, race and 1 socioeconomic variable per model. All statistical analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary, NC).
RESULTS Case Characteristics From 2001 to 2009, 935 cases of IPD were identified in children younger than 5 years in the Metropolitan Atlanta area; 905 (96.8%) cases were successfully geocoded. The remaining cases (n = 35) were not geocoded because of an invalid address (69%) or use of a PO box (31%). The racial distribution among children with IPD included 49.9% white, 45.9% black and 4.2% of other races. The majority of the cases were men (60.1%), and the mean age of the cases was 20.2 months (standard deviation = 14.6 months; see Table, Supplemental Digital Content 1, http://links.lww.com/ INF/B699, which contains descriptive characteristics for the cases). White children presenting with IPD were younger than black children (white = 18.9 months, black = 21.2 months, P = 0.024). Vaccination status was known for 92.5% of cases; for those with known vaccination status, 628 of 837 cases (69.4%) were known to have received at least 1 dose of the PCV7 vaccine, and most of the cases were infected with non-PCV7 S. pneumoniae serotypes (84.1%). A diagnosis of pneumonia was more common among black children with IPD than among white children (35.2% vs. 26.2%; P = 0.006), whereas meningitis was a more common presentation among white children than black children with IPD (10.0% vs. 5.5%; P = 0.016). Most of the cases did not have any known underlying diseases predisposing them to IPD (69.6%). Among those with underlying diseases, asthma was the most prevalent (n = 82), followed by immunosuppressive therapy (n = 23), sickle cell disease (n = 21) and solid organ malignancies (n = 17). Conditions that were significantly more prevalent in black children with IPD compared with white children included asthma (n = 49 vs. n = 29; P = 0.006) and sickle cell disease (n = 21 vs. n = 0; P
Socioeconomic and Racial Disparities of Pediatric Invasive Pneumococcal Disease After the Introduction of the 7-valent Pneumococcal Conjugate Vaccine Jennifer O. Spicer, MD, MPH,* Stephanie Thomas, MPH,† Amy Holst, MPH,† Wendy Baughman, MPH,‡ and Monica M. Farley, MD*‡ Background: Racial differences have been well described for invasive pneumococcal disease (IPD), but little information exists on how race interacts with community socioeconomic factors. Methods: The Active Bacterial Core surveillance/Emerging Infections Program performed active surveillance for IPD in the 20-county Metropolitan Atlanta area. All IPD cases among children younger than 5 years from 2001 to 2009 were geocoded and linked to census tract-level socioeconomic measures from the 2000 US Census. Race- and socioeconomic-specific average annual incidence rates per 100,000 population were calculated. Trends in IPD incidence were determined by χ2 tests for trend. Rate ratios (RRs) and 95% confidence intervals (CIs) were estimated using Poisson regression. Results: IPD incidence among the total population of children increased as percentage of household poverty increased (P = 0.002), as median household income decreased (P < 0.001), as wealth decreased (P = 0.018) and as percentage of individuals with less than a high school education increased (P = 0.023). After stratifying by race, there was no significant linear trend between socioeconomic characteristics and IPD incidence among white children; among black children, however, IPD incidence decreased as socioeconomic conditions worsened. Despite adjusting for sex and socioeconomic factors, the IPD rate remained higher among black children compared with white children (RR = 1.60; 95% CI: 1.39–1.84). Differences in RR of IPD associated with highest poverty and lowest wealth noted in 2001 [RR = 2.71 (95% CI: 2.17–3.39) and 1.80 (95% CI: 1.09–2.96), respectively] declined in 2009 [RR = 1.33 (95% CI: 0.90–1.96) and 0.76 (95% CI: 0.48–1.19), respectively]. Conclusions: Although socioeconomic disparities in IPD incidence exist among children, the association between socioeconomic characteristics and IPD rates may differ by race and may change over time. Community-level socioeconomic factors did not account for racial differences in IPD incidence. Key Words: Streptococcus pneumoniae, invasive pneumococcal disease, socioeconomic disparities, racial disparities (Pediatr Infect Dis J 2014;33:158–164)
Accepted for publication August 22, 2013. From the *Department of Medicine, Emory University School of Medicine, Atlanta, GA; †Department of Medicine, Atlanta Research and Education Foundation, Decatur, GA; and ‡Department of Medicine, Atlanta Veterans Affairs Medical Center, Decatur, GA. This study was funded in part by the Centers for Disease Control and Prevention’s Emerging Infections Program, Atlanta, GA. The authors have no conflicts of interest to disclose. This work was presented in part at the 52nd Annual Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, CA, September 2012. Address for correspondence: Monica M. Farley, MD, Department of Medicine, Atlanta VA Medical Center, 1670 Clairmont Rd, Decatur, GA 30033. E-mail: [email protected]. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s website (www.pidj.com). Copyright © 2013 by Lippincott Williams & Wilkins ISSN: 0891-3668/14/3302-0158 DOI: 10.1097/INF.0000000000000025
158 | www.pidj.com
S
treptococcus pneumoniae is the leading cause of pneumonia and meningitis in the United States, resulting in significant morbidity, mortality and healthcare costs.1 Known risk factors for pediatric invasive pneumococcal disease (IPD) include specific racial and ethnic groups,2–6 group childcare and a number of chronic medical conditions.7–9 The introduction of the childhood heptavalent pneumococcal conjugate vaccine (PCV)7 in 2000 resulted in a significant reduction in IPD among children. Although racial differences between black and white children also decreased,8,10–13 these differences continue to exist for unclear reasons. Socioeconomic factors, such as income, poverty, wealth, education and crowding, have been increasingly recognized in the literature as contributing to health disparities.14,15 It has been suggested that routine health reporting should analyze socioeconomic and racial health disparities jointly and separately to better understand health disparities.16 Until recently, socioeconomic variables were infrequently included in epidemiological analyses because most public health surveillance systems do not have the capacity to routinely collect socioeconomic information on all the cases that they identify. The Public Health Disparities Geocoding Project has addressed this issue by identifying community-level socioeconomic information routinely collected by the US Census Bureau that can be linked to cases in public health surveillance systems based on address information for the cases.17,18 Previous studies examining the association between socioeconomic factors and IPD rates found variable results across age groups and racial categories.6,11,13,14,19,20 Only 2 studies have presented specific results for pediatric populations.5,9 The first study was conducted prior to the introduction of PCV7 and found that race-adjusted IPD rates were highest among low-income children.5 The other study was a case-control study completed after the introduction of PCV7, and it showed that low income was associated with an increased risk of PCV7-type IPD among unvaccinated children but not among vaccinated children.9 Neither study analyzed the effect of socioeconomic factors within racial groups. Since the introduction of PCV7, there have been no studies describing the epidemiology of socioeconomic and racial health disparities in IPD among children. The objectives of this study were to describe the association between census tract-level socioeconomic measures and IPD in children younger than 5 years in the post-PCV7 era, to assess whether the association varies by race, to determine whether socioeconomic measures account for racial differences in IPD incidence and to establish whether socioeconomic disparities in IPD incidence have decreased since the introduction of PCV7.
METHODS Study Population The Centers for Disease Control and Prevention–sponsored Active Bacterial Core surveillance of the Georgia Emerging Infections Program conducted active population-based laboratory surveillance for IPD in the 20-county Metropolitan Atlanta area. Cases
The Pediatric Infectious Disease Journal • Volume 33, Number 2, February 2014
The Pediatric Infectious Disease Journal • Volume 33, Number 2, February 2014
were identified through weekly calls and periodic audits of clinical laboratories. Clinical and demographic information on each case was collected from medical records, clinical laboratories or the patient’s physician. Vaccination status was obtained through use of the Georgia Immunizations Registry and telephone interviews with parents and healthcare professionals as described elsewhere.21 For this analysis, the study population included children younger than 5 years with IPD from 2001 to 2009. The time period from 2001 to 2009 was chosen to evaluate socioeconomic influences after the introduction of PCV7 but prior to the introduction of PCV13. IPD was defined as isolation of S. pneumoniae from a normally sterile body site (blood, cerebrospinal fluid, pleural fluid, peritoneal fluid, pericardial fluid, joint, muscle, bone or other internal body site).
Classification of Pneumococcal Serotypes S. pneumoniae isolates were serotyped at the Centers for Disease Control and Prevention using the Quellung reaction. PCV7 serotypes included serotypes 4, 6A, 6B, 9V, 14, 18C, 19F and 23F. All other serotypes were classified as non-PCV7 serotypes.
Census Tract Socioeconomic Variables Area-based socioeconomic measures for each census tract were downloaded from the US Census 2000 Summary 3 File,22 which contains population and household data from the census long form questionnaire. Because this information for our time period was only collected in the decennial census, data from the 2000 US Census was used for all cases in this analysis. After 2000, the US Census Bureau only collected housing, financial and socioeconomic information in the annual American Communities Survey rather than the decennial census, so socioeconomic information was not available from the 2010 US Census for use in this analysis. To determine whether changes in socioeconomic characteristics from 2001 to 2009 could have affected the results of our analysis, socioeconomic and population data were obtained from the compiled 2005–2009 American Community Survey 5-year estimates23 to provide more recent socioeconomic data. Five-year compilations are required to provide accurate estimates for small geographic areas, such as the census tracts. Rate ratio (RR) calculations using both the 2000 US Census data and the 2005–2009 American Community Survey socioeconomic and population denominators were included for comparison to detect any changes in socioeconomic determinants in the surveillance population from 2001 to 2009 (shown in Table 3). Variables were created to represent census tract-level measures of poverty, wealth, income, education and crowding using categories recommended by the Public Health Disparities Geocoding Project.24 Census tract poverty was defined as the percentage of households in the census tract with incomes below the federally defined poverty level and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. Census tract wealth was defined as the percentage of owner-occupied homes worth ≥400% of the US median value of owned homes and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. Census tract median household income was categorized as quintiles: $0– $35,416, $35,416–$44,329, $44,330–$52,935, $52,935–$67,946 and >$67,946. Census tract education was defined as the percentage of persons in each census tract of age ≥25 years with less than a 12th grade education, including those who completed 12th grade but did not receive a degree, and was divided into 4 categories: 0%–14.9%, 15.0%–24.9%, 25.0%–39.9% and >40%. Census tract crowding was defined as the percentage of households in each census tract with ≥1 persons per total number of rooms in the residence and was divided into 4 categories: 0%–4.9%, 5.0%–9.9%, 10.0%–19.9% and >20%. © 2013 Lippincott Williams & Wilkins
Social Factors and Pneumococcal Disease
Data Analysis Demographic and clinical characteristics of IPD cases of black race and other race were compared with cases of white race using the χ2 test or Fisher exact test. The race of each case was obtained from medical records and classified as white, black or other. Other race included Asian, American Indian/Alaskan Native and Native Hawaiian/Other Pacific Islander races. For those cases missing race (12.3%), race was assigned according to the race that represented many children younger than 5 years within the case’s census tract because analyses excluding individuals with missing race yielded similar results to the final analyses. The methods for data analysis were based on those of the Public Health Disparities Geocoding Project.24 Using addresses obtained from the Georgia Emerging Infections Program case report forms, each case in this analysis was linked to a census tract from the 2000 US Census using ArcView/ArcGIS 10.1 (ESRI, Redlands, CA). Average annual incidence rates were calculated by dividing the average number of cases per year from 2001 to 2009 by the population of children younger than 5 years in the surveillance area according to the 2000 US Census, and rates were stratified by race and socioeconomic categories. The χ2 test for trend was used to analyze trends in incidence for each socioeconomic variable. Poisson regression was used to calculate RRs and 95% confidence intervals (CIs). For Poisson regression, the dependent variable was census tract IPD rates, and the independent variables were sex, race and 1 socioeconomic variable per model. All statistical analyses were performed in SAS version 9.3 (SAS Institute Inc., Cary, NC).
RESULTS Case Characteristics From 2001 to 2009, 935 cases of IPD were identified in children younger than 5 years in the Metropolitan Atlanta area; 905 (96.8%) cases were successfully geocoded. The remaining cases (n = 35) were not geocoded because of an invalid address (69%) or use of a PO box (31%). The racial distribution among children with IPD included 49.9% white, 45.9% black and 4.2% of other races. The majority of the cases were men (60.1%), and the mean age of the cases was 20.2 months (standard deviation = 14.6 months; see Table, Supplemental Digital Content 1, http://links.lww.com/ INF/B699, which contains descriptive characteristics for the cases). White children presenting with IPD were younger than black children (white = 18.9 months, black = 21.2 months, P = 0.024). Vaccination status was known for 92.5% of cases; for those with known vaccination status, 628 of 837 cases (69.4%) were known to have received at least 1 dose of the PCV7 vaccine, and most of the cases were infected with non-PCV7 S. pneumoniae serotypes (84.1%). A diagnosis of pneumonia was more common among black children with IPD than among white children (35.2% vs. 26.2%; P = 0.006), whereas meningitis was a more common presentation among white children than black children with IPD (10.0% vs. 5.5%; P = 0.016). Most of the cases did not have any known underlying diseases predisposing them to IPD (69.6%). Among those with underlying diseases, asthma was the most prevalent (n = 82), followed by immunosuppressive therapy (n = 23), sickle cell disease (n = 21) and solid organ malignancies (n = 17). Conditions that were significantly more prevalent in black children with IPD compared with white children included asthma (n = 49 vs. n = 29; P = 0.006) and sickle cell disease (n = 21 vs. n = 0; P
