Presentation on theme: "Use of Area-based Poverty as a Demographic Variable for Routine Surveillance Data Analysis CT and NYC CSTE Annual Conference June 10, 2013 J Hadler CT."— Presentation transcript:
Use of Area-based Poverty as a Demographic Variable for Routine Surveillance Data Analysis CT and NYC CSTE Annual Conference June 10, 2013 J Hadler CT EIP, NYC DOHMH
Outline Rationale Public Health Disparities Geocoding Project (PHDGP) Recommended standard Area-based SES measure Connecticut EIP Influenza hospitalizations, bacterial foodborne pathogens, HPV New York City Workgroup formation and recommendations All cause mortality, TB Conclusions
Rationale 1 Describing health disparities and monitoring progress in reducing them has been a national priority (HP 2010 and 2020). Major variable used to describe health disparities has been race-ethnicity. Use of race/ethnicity as a major means to describe disparities has some severe limitations –not always available –>20 official race/ethnic groups –difficult to interpret – disparities are only sometimes genetic or cultural; mostly race-ethnic disparities reflect SES differences
Rationale for use of area-based SES (ABSES) measure for data analysis US has no recommended SES measure for routine collection, analysis and display of surveillance data – race-ethnicity is a very unsatisfying surrogate. Geocoding accessibility and ease have made it possible to use area-based SES measures where have street address or ZIP code. PHDGP already laid groundwork
Public Health Disparities Geocoding Project 1 Harvard-based lead by Nancy Krieger, ~1998 - 2004 Recognized potential in public health data for analysis using ABSES Explored wide range of health outcomes using MA and RI data from 1990 using different area sizes and SES indices Found ABSES measures described disparities as big or bigger than those by race/ethnicity and usually described disparities within race/ethnic groups.
Public Health Disparities Geocoding Project 2 Recommended use of census tract level percentage of residents living below federal poverty level for routine data analysis. – 20% “Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: The PHDGProject”. Am J Public Health 2005; 95: 312-323. http://www.hsph.harvard.edu/thegeocodingproject/
Connecticut Objectives Gain experience using census tract poverty level to describe health disparities Began to analyze surveillance data routinely as part of the EIP in ~2009. –Invasive pneumococcal disease* –Influenza hospitalizations (pediatric*, adult**) –Cervical cancer precursors (CIN 2,3; AIS)* –Foodborne bacterial pathogens (campylobacter**, STEC, salmonella) * published; ** submitted
Incidence of influenza-associated hospitalizations by census tract poverty level, Children 0-17 years, NH County, CT, 2003/04 -2009/10 Incidence per 100,000 person-years Census tract poverty level <5% 5-9.9% 10-19.9% 20+% AJPH 2011;101:1785
Ratio of highest to lowest census tract-level poverty incidence of influenza-associated hospitalizations by year, Children 0-17 yrs, CT, 2003/04 – 2009/10 Incidence ratio H1N1 AJPH 2011;101:1785
Age-adjusted incidence of influenza- associated hospitalizations of adults 18+ yrs by selected ABSES measures, NH County, CT, 2005-2011 Incidence per 100,000 person-years
Ratio of highest to lowest census tract-level poverty incidence of influenza-associated hospitalizations by year, Adults 18+ yrs, CT, 2005-2011 Incidence ratio H1N1
Age-adjusted incidence of influenza- associated hospitalizations of adults 18+ yrs by poverty level* and race/ethnicity, NH County, CT, 2005-2011 Incidence per 100,000 person-years non-Hispanic
Incidence of Cervical Intraepithelial Neoplasia Grade 2+ by census tract poverty level, Women 20-39 years, NH County, CT, 2008-2009 Incidence per 100,000 person-years Census tract poverty level <5% 5-9.9% 10-19.9% 20+% AJPH 2012;103:156
Incidence of CIN2+ by census tract poverty and age group, Women 20-39 years, NH County, CT, 2008-2009 Incidence per 100,000 person-years AJPH 2012;103:156
Foodborne bacterial pathogen age-adjusted incidence by census tract poverty level and pathogen, CT, 1999-2011 Incidence per 100,000 person-years CampylobacterSalmonellaSTEC
Foodborne bacterial pathogen risk in children by census tract poverty level, CT, 1999-2011 Incidence per 100,000 person-years Age Group
Implications of identified SES disparities Influenza – target efforts to improve vaccination rates to neighborhoods with high rates of neighborhood poverty HPV vaccination – needed for all, not just a subset of the population. Very high rates of cervical cancer precursors in neighborhoods with low poverty levels. Bacterial foodborne pathogens – Focus prevention and prevention research efforts on high SES populations. –More research needed to understand risk factors in children – why children in high poverty neighborhoods have higher risk of campy/salmonella but not STEC.
Background 2010 NYC has had long-standing emphasis on describing and minimizing health disparities. Most programs used race/ethnicity; some programs used SES measures: income, neighborhood poverty No standardization of measures, neighborhood size, cut-points
Background (cont) Has cross-cutting “Data Task Force” as forum for discussion of data issues agency-wide Following presentation of PHDGP recommendations for standard area-based SES measure, workgroup set up to explore NYC-specific issues and make recommendations.
Challenges for a NYC standard Population distribution not the same as MA and RI “Neighborhoods” used have been UHF areas, not census tracts With higher cost of living than most of rest of US, is federal poverty level the best level to use?
Poverty Measure Workgroup 1 Poverty measure workgroup formed to explore these issues and develop recommendations re: a standard neighborhood SES measure. Composed of volunteers from Communicable disease, Epi Services, HIV, Immunizations, STD, TB, Vital Statistics
Poverty Measure Workgroup 2 Agreed early on to the following: –Important to have a standard measure that can be used and compared to other public health jurisdictions (cities, states) –Accept the background work of the PHDGP and use a neighborhood poverty measure –May need different neighborhood poverty cut points than those recommended based on work in MA & RI –Need to explore NYC data to determine best cut points and neighborhood size to use.
Percentage of population by census tract poverty level, NYC, 2000 & PHDGP 1990 Percentage of population Percent below poverty in census tract 46%
Percentage of population by % of residents in census tract, zip code and UHF area who live below poverty, NYC, 2000 Percentage of Population Percent below poverty in census tract
Age-adjusted Mortality Rate by % in census tract who live below poverty, NYC, 2000 Death Rate per 1000 Percent below poverty in census tract
Age-adjusted Mortality Rate by % in census tract who live below poverty by race/ethnicity, NYC, 2000 Death Rate per 1000 Percent below poverty in census tract
Age-adjusted Mortality Rate by % in census tract who live below poverty, NYC, 1990 and 2000 Death Rate per 1000 Percent below poverty in census tract
Age-adjusted TB Rate by % of residents in census tract who live below poverty, NYC, 2000 Rate of TB per 100,000 Percent below poverty in neighborhood
Age-adjusted TB Rate by % in census tract who live below poverty by race/ethnicity, NYC, 2000 Rate of TB per 100,000 Percent below poverty in census tract
Age-adjusted TB rate by % of residents in census tract who live below poverty, NYC, 2000 and 2008 Rate of TB per 100,000 Percent below poverty in census tract
Key Recommendations 1.All routinely collected surveillance data with geolocating info should be analyzed using neighborhood poverty as a standard variable 2.Standard Measure % in neighborhood who live below federal poverty level 6 categories for analysis: <5%, 5-9%, 10-19%, 20-29%, 30-39%, 40+% 4 categories as needed for small numerators or display: <10%, 10-19%, 20-29%, 30+% Use census tract when possible (rather than ZIP, UHF)
Conclusions 1.Analysis of data using census tract poverty (CTP) is a meaningful way to describe disparities for some diseases and provides new insights relevant to control –Find disparities within race/ethnic groups –Some diseases more common among those of higher SES –Can be used regardless of whether have race/ethnicity data –Targeting groups for intervention based on SES more attractive than based solely on race/ethnicity 2.Use of CTP level is gaining traction –Increasing experience using it, CSTE involved
Where do we go from here? 1.Up to state and local health dep’t epidemiologists and CSTE to bring SES measures to the data we collect – to take the lead. –We are the experts in analyzing and using the info we collect. –Academia has shown the way – is best suited to studying the mechanisms related to SES disparities. –CDC is interested, but is slower to move than state and local jurisdictions – and doesn’t have address data. continued ….
Where do we go from here? (cont) 2.Take advantage of the PHDGP work: –Begin to routinely include ABSES measures in surveillance data analyses, ideally, including the recommended “standard” –Help CSTE move ABSES, esp. census tract poverty level, into the national dialogue about measuring and addressing health disparities.