The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #60039

David K. Ahn, PhD, Catherine Cubbin, PhD, Ying-Chih Chuang, PhD, Namita Oswal, and Marilyn A. Winkleby, PhD, MPH. Stanford Center for Research in Disease Prevention, Stanford University School of Medicine, 1000 Welch Road, Palo Alto, CA 94304-1825, (650)725-5320, dahn@stanford.edu

The question of whether neighborhood social environments influence health is gaining increasing attention in epidemiological research. Studies that examine this question commonly use data from the U.S. Census to describe neighborhood social environments. These data are then linked to individual-level health and socioeconomic data for analyses. The large number of different neighborhood-level variables in the Census, and their availability only every ten years, present challenges in computing appropriate measures of neighborhood social environments over time. As part of a study on neighborhood influences on health outcomes with data on 8,419 adults who participated in the Stanford Heart Disease Prevention Program (1979-1990), we defined participants' neighborhoods (82 neighborhoods in four cities) using a combination of census tracts and block groups. We then sought to create appropriate measures to describe the social environments of these neighborhoods from the 1980 and 1990 census. This presentation will discuss methods for: 1) constructing composite measures that efficiently summarize information from multiple variables using a principal component analysis, 2) handling time trends when using data from two or more censuses; and 3) computing valid measures for a census variable when the desired neighborhoods are a combination of census block groups or tracts (e.g., two different median incomes from two different block groups). Recommendations will be made about appropriate methodology to use in studies that seek to examine relationships between neighborhood social environments and health outcomes using census data.

Learning Objectives:

• i) Identify how to construct composite measures from multiple variables in census data to summarize neighborhood social environments
• ii) Identify how to compute census measures from multiple censuses and multiple census tracts

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.