Connecting the dots

The National Cancer Institute soon will launch a project that uses digital mapping technology available on the market today to develop applications with an unprecedented capability to help researchers find links between cancer and environmental and social factors.

The pact, to be awarded as early as this fall, will provide NCI with the first geographic information system for epidemiological research by integrating two dozen new features into a commercially available GIS package. NCI plans to use the system to support a congressionally mandated study of possible links between breast cancer and pollution in Nassau and Suffolk counties on Long Island, N.Y., Schoharie County, also in New York, and Tolland County in Connecticut.

The Long Island Breast Cancer Study, as the project is known, was ordered in 1993 following reports that breast cancer rates in that region were among the highest in the country. Between 1990 and 1994, breast cancer rates in New York state were slightly higher than the national average, according to statistics supplied by the American Cancer Society.

Researchers suspect links between the cancer and environmental factors such as pollution but have found it hard to draw explicit connections between illness and long-term exposure to toxins.

While current GIS applications do help investigators picture these links, the ability of these systems to process complex statistical queries is limited. The NCI project will use 59 state, local and national data sets on population, pollution, health, birth and death records, and geography to help researchers find correlations.

"All this data has been collected for a good many years," said Bill Davenhall, manager of the health solutions group at Environmental Systems Research Institute Inc., a Redlands, Calif., GIS maker that is offering its software on several bids for the project. "This is the first project where NCI is saying, 'Let's connect the dots.' " Davenhall said he thinks state environmental and health organizations that want to do similar studies also will want to use the applications NCI develops.

Although epidemiologists already use GIS to draw connections between where people live and the occurrence of diseases or access to treatment, doing so is cumbersome, said Gerard Rushton, a University of Iowa geographer who helped NCI write the requirements for its system. Today, he said, researchers who want to study the links between health and geographic data "can't do [it] by clicking'' but need programming skills to marry disparate data sets.

"Right now, there is GIS, and there are statistical packages,'' said Joan Gardner, president of Boston-based Applied Geographics, which is bidding on the proj-ect.

"They take the data out of the GIS and put it into the statistical package and take it out of the statistical package and put it [back] into the GIS. The idea [of the NCI project] is to build those health functions into [the GIS] so it is seamless,'' Gardner said.

Virginia Lee, who heads the spatial analysis section at the federal Agency for Toxic Substances and Disease Registry, said "the technology has improved dramatically'' since her agency began using GIS nine years ago to study the health and demographics of people living near toxic waste sites. "I think it has great promise," she said.

Use of maps in public health research goes back more than a century, said Michael Goodchild, a geography professor at the University of California at Santa Barbara and chairman of the National Science Foundation-funded National Center for Geographic Information and Analysis. In 1854, British physician John Snow identified the cause of a cholera outbreak in London as a polluted drinking water source by pinpointing where 616 victims lived. "It would have taken a lot of time" to map those cases, Goodchild said.

A Desktop Tool

The NCI project is one example of a trend toward GIS becoming a desktop tool. "GIS tended to be its own discipline," said Bruce Jenkins, vice president of Daratech Inc., a Cambridge, Mass., market research firm. "Other workers prepared data input into a GIS that was manipulated by a specialist. Today it's evolving into usability by everyone."

Some vendors and industry observers anticipate that the NCI project, estimated to be worth $2.5 million to $3 million, may prime a new vertical market for spatial analysis software.

David Sonnen, president of Integrated Spatial Solutions Inc. and a senior consultant with International Data Corp., said custom public health applications are "not going to be the next killer app" for the $972 million GIS industry. But in the past six months, vendors have been introducing more modular versions of their software that "make it a lot easier" to add statistical functions or other data analysis tools.

The University of Iowa's Rushton said the new GIS, which NCI expects to have in two years, will not provide the reasons for the incidence of breast cancer on Long Island.

Any links the GIS illustrates will have to be explained by cancer experts. "You're really setting up a utility for researchers to use,'' he said.