Eastern Massachusetts Pollution Study
Using data from multiple personal exposure studies, plus several other studies that have collected pollution data at multiple locations in Eastern Massachusetts, we will fit GIS regression to develop predictive models of pollution concentrations as a function of
population density, distance to roads, estimated traffic count, land use data, etc. We are obtaining from the Massachusetts Department of Public Health data from birth certificates for all live births in the area for the last five years, as well as all deaths. These will be geocoded to latitude and longitude. We will then merge in block group and tract level data on socio-economic factors. We will examine whether estimated pollution exposure is associated with birthweight, gestational age, small for gestational age, and low birthweight using the birth data. We will also examine the role of contextual level SES variable in explaining the same outcome, the independent explanatory effect of block group vs tract data, and the extent to which measures of physical environment (pollution) explain some of the social gradient in outcome. Using the death data, we will investigate whether rates of death for causes are related to pollution exposure, and again whether pollution explains some of the social gradient in death rate. Finally, we will select controls matched on age and sex, geocode them and estimate their exposure, and repeat the analysis as a case-control study.

Normative Aging Study
The normative aging study is a long term prospective cohort study of approximately 2500 persons who were free of disease in the early 1970’s, and who have been followed since then. Using the geocoded exposure data discussed above, and address history, we will geocode their address and estimate cumulative exposure in each participant. We will estimate the association with myocardial infarction, COPD, and death on followup.

Area-based socioeconomic measures for health research (also known as the Public Health Disparities Geocoding Project)
See the The Public Health Disparities Geocoding Project for more details Currently, most US public health surveillance systems lack socioeconomic data, thereby precluding monitoring of socioeconomic disparities in health, as well as their contribution to racial/ethnic disparities in health. One possible solution is to use geocoding and area-based socioeconomic measures (ABSMs), to characterize both the cases and population in the catchment area, thereby enabling computation of rates stratified by the area-based measure of socioeconomic position. Yet, it is unknown which ABSMs, at which level of geography, would be most apt for monitoring US socioeconomic inequalities in health, overall and within diverse racial/ethnic-gender groups.

We accordingly launched the Public Health Disparities Geocoding Project to ascertain which ABSMs, at which geographic level (census block group [BG], census tract [CT], or ZIP Code [ZC]), would be suitable for monitoring US socioeconomic inequalities in the health. Drawing on 1990 census data and public health surveillance systems of 2 New England states, Massachusetts and Rhode Island, we analyzed data for: (a) 7 types of outcomes: mortality (all cause and cause-specific), cancer incidence (all-sites and site-specific), low birth weight, childhood lead poisoning, sexually transmitted infections, tuberculosis, and non-fatal weapons-related injuries, and (b) 18 different ABSMs. We conducted these analyses for both the total population and diverse racial/ethnic-gender groups, at all 3 geographic levels.

Our key methodologic finding was that the ABSM most apt for monitoring socioeconomic inequalities in health was the census tract (CT) poverty level, since it: (a) consistently detected expected socioeconomic gradients in health across a wide range of health outcomes, among both the total population and diverse racial/ethnic-gender groups, (b) yielded maximal geocoding and linkage to area-based socioeconomic data, and (c) was readily interpretable to and could feasibly be used by state health department staff.
PUBLICATIONS OF THE PUBLIC HEALTH DISPARITIES GEOCODING PROJECT

Socioeconomic trends in breast cancer incidence
To date, despite suggestive evidence about possible changes in the socioeconomic patterning of breast cancer, no studies have investigated whether the socioeconomic gradient in breast cancer among women in the United States is changing over time. Our proposed study seeks to test the hypothesis that: (a) this gradient is decreasing, because (b) incidence rates are rising more quickly among women residing in less affluent areas. Moreover, to aid interpretation of study findings, we propose to test the same hypothesis among men, because if patterns are similar it would suggest that factors driving the patterns are not only reproductive but non-reproductive, e.g., anthropometric or environmental.

To test our hypotheses, we will obtain cancer data from the Northern California Cancer Center (NCCC), covering 5 counties in the San Francisco Bay Area (1973-2002; n ~ 70,000 females and ~ 490 male cases), the Los Angeles Cancer Surveillance Program (1973-2002; n ~ 129,400 female and ~ 940 male cases); and the Massachusetts Cancer Registry (1982-2002; n ~ 93,100 female and ~ 760 male cases), along with relevant 1970 to 2000 US census data. We will then geocode each cancer registry record, based on each patient’s residential address, to its corresponding latitude, longitude and census tract, generate state-of-the-art census-derived area-based socioeconomic measures, and link each geocoded record to the relevant area-based socioeconomic measures. We will likewise generate denominators stratified by the area-based socioeconomic measures, using data from the 1970-2000 decennial censuses.

Developing & Evaluating Objective Measures of Outdoor Recreational Areas
An interdisciplinary research team from the fields of physical activity/public health, geography, landscape architecture, and parks and recreation will develop and evaluate objective measures of outdoor recreational settings where walking, bicycling and other linear activities can be performed. Selection of measures will be based on ecological models and social cognitive theory, current evidence from public health/physical activity studies, practice guidelines from landscape architecture, and formative work that the research team will conduct. Existing GIS databases, observation/auditing of facilities, and other methods will be used to develop objective GIS measures of facility characteristics.

The physical activity data, measured by accelerometer and self-report, allow us to identify active and sedentary periods by time of day for 251 students. We are assessing factors such as density, mix of land uses, and completeness of the sidewalk network, using GIS data, aerial photographic maps, and site visits for each school and its surrounding neighborhood. We will also be looking at the layout of the schools themselves including collecting information on the size of the campus and the presence of stairs.

Description of using GIS to study local features and physical activity

Among the questions we will be asking is whether the pedestrian environment in the neighborhood is related to physical activity for students on weekends, and if active-school environments are associated with more activity during the school day. The database also includes information on weather conditions, and the researchers will see if the neighborhood environment influences whether children walked or biked to school in bad weather.

Trucking Industry Particle (TrIP) Study
Researchers from the Channing Laboratory will be using GIS to develop exposure models to explore the relationship between workplace exposure to diesel exhaust and other particle exposures and lung cancer in a large cohort of trucking industry employees.

Nurses Health Study
Geocoding of large cohort to investigate health effects of air pollution using EPA databases. This will involve use of GIS regression approaches to estimate exposure at residential addresses.

Home Allergens and Asthma Study
In a birth cohort, we geocoded participants' addresses and measured serum IgE in the mothers of participants. We then utilized kriging methods to map serum IgE levels and found that within the Boston City limits, mothers had higher serum IgE levels compared with levels from mothers living in the suburbs. With the help of our colleagues at the Harvard School of Public Health, Department of Biostatistics, we will link the health outcomes in children from this birth cohort with exposures in the environment.

The overall focus of the new research will be developing and extending statistical methods to best quantify a spatial region to be an area with a disproportional incidence of disease, otherwise known as a hotspot. This type of research can be utilized to prove, or disprove, an existence of a disease (e.g. asthma) cluster or other adverse health effects in regions that can be linked to environmental hazards. In particular, for the Boston Home Allergens study, one could use this research to statistically test if there is an increase in incidences of asthma occurring in areas closer to highways. Therefore these methods could be used as a way to relate previously non-linkable sources of environmental hazards to increases in adverse health outcomes.

The research problem will first be approached by extending the currently most popular method, the Scan Statistic, from binary to censored outcomes and to also allow for other known predictors. The second approach would be to extend cumulative residual methods that are currently applied to failure time outcomes to include a spatial component that would be able to detect significant clustering.