On the trail of agent orange: measuring risk with GIS. (Science Selections).

As with many environmental health questions, uncovering the true health effects of the herbicides used in the Vietnam War has been limited by problems with assessing exposure. In this issue, Jeanne Mager Stellman of Columbia University's Mailman School of Public Health and colleagues describe a new research tool: software that incorporates relational database technology, geographic information system (GIS) principles, and refined mathematical models to create "exposure opportunity" scores from military data on spray missions [EHP 111:321-328].

Between 1961 and 1971, U.S. forces sprayed nearly 19.5 million gallons of herbicide in Southeast Asia, mostly from fixed-wing aircraft. These herbicides had been contaminated during production with minute amounts of dioxins, by-products of the manufacturing process. Dioxins have been linked with Hodgkin disease, non-Hodgkin lymphoma, soft-tissue sarcoma, and more tentatively with type 2 diabetes mellitus and other conditions.

The GIS divides Vietnam and parts of Laos into a grid of 0.01-degree blocks (about 1.2 kilometers on a side) and contains data on the locations of villages, roadways, bridges, military bases, airfields, and targets, the known movements of U.S. military units, and the 9,141 airborne spraying missions of the U.S. herbicide campaign. The relational database uses mathematical modeling to calculate an exposure opportunity index for a military unit or a location on any date during the war. The index is a relative, not absolute, measure of exposure, says coauthor Steven Stellman; higher scores reflect more gallons being sprayed in an area, being closer to the sprays, and spending a longer time in a sprayed area. The GIS is flexible; researchers can insert other mathematical models to reflect different assumptions, such as how fast the contaminants degrade.

Jeanne Stellman says the new software was developed under contract to the National Academy of Sciences in response to a call by the Agent Orange Act of 1991 for better research. "The academy had found that no systematic study has been done on Vietnam because there was no agreed exposure methodology," she says. "We've been able to refine models and take advantage of new database technologies and GIS concepts [to create an approach that standardizes exposure opportunity]." It will be up to future research efforts to actually apply the new methodology to the health effects of herbicides used in Vietnam.

Nobody is ever likely to calculate exact exposures that occurred more than 30 years ago, so the exposure opportunity index represents a major advance in the quest to understand the effects of Agent Orange. Because some parts of Vietnam were heavily sprayed, while most were never sprayed, the range of relative exposure was huge. Jeanne Stellman says the greatest exposures were about six orders of magnitude higher than those of people who were not near spray zones.

A key to using the new database will be choosing the study population wisely, she says, because location data are better for some military units than others. "If you select a group to study randomly, it will be hard, because it's difficult to reconstruct locations after so many years. If you go for the `hot spots' and military units with good records [for inputting into the GIS], you'll be able to do a pretty good study."