Health Disparities Interest Group March 2009 Seminar
- Dr. Nancy Breen called the meeting to order and asked participants to introduce themselves.
- Dr. David Berrigan introduced the guest speaker, Dr. Ann Carroll Klassen, who presented "Validating Area-Level Profiles for Cancer Research." This research is ongoing.
Summary of Dr. Klassen's Presentation/Discussion
- To date, more behavioral than social issues have been addressed in cancer control research, yet some of the strongest patterns in cancer disparities are reflected in social class. Researchers now recognize the need to understand the influence of social stratification on cancer biology.
- An approach using sociological theory, including epidemiologic sociology, offers an understanding of the influence of nonmaterial resources, social tensions, and social framing of issues not interpretable from a quantitative perspective.
- The counter-argument to bringing a more sociological perspective into cancer control posits that cancers have numerous complex biological mechanisms that broad sociologic theory may not explain accurately. Additionally, social science focuses on relationships that are difficult to manipulate and test.
- More data are needed on behavioral and social information, as surveillance data in the United States lack this information. Ideally, data would offer insight on individuals as well as their environments.
Area-Level Information in Explaining Cancer Patterns
- Historically, evidence of social influences on health has been strong, and social resources have been shown to be important as both absolute and comparative influences.
- In the United Kingdom, well-developed mechanisms exist for measuring area-level social influences and to inform both research and policy. In the United States, however, ethnic disparities make assignment to social class more difficult.
- Measuring social factors includes demographic, economic, cultural, and behavioral attributes of communities.
- Geodemographics provide descriptive characteristics of populations in area groupings, but there are questions about their validity as an indicator of individual behavior.
- In the United States, sources include the U.S. Census and consumption data from the Consumer Expenditure Survey by the U.S. Department of Labor.
- Area-level behavior examines, for example, whether an individual reads the newspaper as well as how many in his/her area read the newspaper. Its use in cancer control is to separate contributions made by census data from lifestyle data measured through spending.
- Lifestyle segmentation data extend census data. In cancer control, spending data possibly are related to cancer burden, such as total, household, and per capita spending on food, fruit and vegetables, meat, tobacco, alcohol, reading material, sports and recreation, medical spending, supplemental spending, and charitable and political contributions.
- Data quality, however, can suffer because data often are proprietary, and methods and coverage are difficult to verify. Additionally, historical data are scarce, household size and composition vary, and spending does not always indicate the volume of a commodity consumed. Maryland is a good choice of locale to study these issues as it mirrors the United States in that it contains concentrated areas of wealth and poverty and has a longstanding, substantial African-American population in all areas and levels of social class. Its cancer rate exceeds that of the United States, but it is not a high-cancer state. Preliminary results of the study now are being geocoded.
- Validating area-level behavioral profiles for cancer disparities research:
- The measures will be tested by statistically modeling their similarity to high-quality individually reported data from the CLUE studies of Washington County, MD, residents.
- Analysis demonstrated that area-level measures do predict individual answers adjusted for individual predictors such as age, gender, and education level.
- Area-level measures predict aggregate answers and blockgroups' average of CLUE responses if aggregates are weighted by number of respondents per blockgroup.
- Data from the year 2000 on breast, colorectal, prostate, and lung cancer now are being studied to determine how area-level profiles add to the explanation of cancer risk.
US Army Synergistic Idea Award: Spatially Informed Investigations of Race-Specific Social Gradients in Breast Cancer Disparities
- This study explored social class-related disparities in cancer burden in Maryland's African-American and white women diagnosed with breast cancer in 1992 to 2003. Data from the study were used to build community-level profiles for socioeconomic resources, income and education, and consumer behaviors. The goal was to determine (1) the area-level characteristics that are most associated with breast cancer that lead to poor outcomes, and (2) the social gradients that differ across race, age, and rurality.
- African-American women have lower breast cancer incidence than white women in the United States but are younger at the time of onset and have more aggressive tumor biology and worse outcomes. Within racial groups, incidence is highest among higher social classes, but outcomes are worse for women of lower social classes.
- Analysis will be completed in two phases: multilevel analysis and spatial analysis.
- This approach:
- may identify priority areas or populations for planning or intervention.
- could serve as an inexpensive, accessible tool for applied activities.
- still remains to be tested to determine if results differ from what is known about advantaged/disadvantaged populations.
- A five-item index that predicts healthy or unhealthy communities might be useful.
- Neighborhood effects on behavior are small but can be interpolated to the public health context by examining overall effects on populations rather than the interaction between the individual and the environment. Overall effects are small and might mask effects on individuals or vulnerable groups within an environment.
- Aggregate or contextual analyses around individuals are useful because it is unrealistic to try to collect data from individuals. These analyses have added value, but further testing must be conducted targeting specific communities.
- Dr. Klassen's group would like to build models that examine Caucasian women similar to the regression analyses conducted on African-American women.
- The next HDIG meeting is scheduled for Monday, July 13, and will include a showing and discussion of additional parts of the movie Unnatural Causes. The meeting is scheduled for 1:00 to 2:30 p.m. in Room 405, 6116 Executive Boulevard, Rockville, MD.
Last Modified: 03 Sep 2013