|
Last fall, Durham County health director Gayle Harris faced a thorny question: How do you get a confused and suspicious public inoculated against H1N1 influenza? The much-feared flu strain had already arrived in Durham, forcing Duke to cancel several summer youth camps. Nationwide, H1N1 had infected 22 million people, killing 3,900, according to estimates by the U.S. Centers for Disease Control and Prevention. Still, persuading Durhamites to get vaccinated was proving tough.
“The H1N1 messaging from federal down to state came in bits and pieces,” Harris says. “First we said, ‘We have the outbreak, but we don’t have enough vaccine.’ Then we said, ‘Oh, we’ll let everybody have it. We have more than enough.’ What the public heard was: ‘We have something we don’t know how to deal with.’ ” What’s more, she says, certain populations appeared to be missing from the city’s vaccination clinics. As she visited clinics around the city, she noticed good numbers of Hispanics in the flu-shot lines, but very few African Americans.
Harris, who is African American, hypothesized that some black residents were suspicious of getting a live vaccine for a new strain. “Those people who have the memory of Tuskegee don’t want to be engaged with this experimental kind of thing,” she says, referring to an infamous forty-year research project in Tuskegee, Alabama, in which black men infected with syphilis unknowingly went untreated.
Harris didn’t want to rely solely on her own impressions. To reach those who needed the vaccine the most, she needed data—and she needed it fast. The health director turned to one of her most trusted colleagues, Marie Lynn Miranda ’85, director of the Children’s Environmental Health Initiative (CEHI), an interdisciplinary program housed at Duke’s Nicholas School of the Environment.
“Can you create a map for me showing the locations of Durham County residents who have presented in [local emergency rooms] with influenza-like illnesses?” Harris wrote Miranda in an e-mail message. “I would like to use it to help target locations for mass clinics.”
Miranda knew she could produce more than a simple map. CEHI’s forte is creating what’s called spatial data architecture—using Geographic Information Systems (GIS) to turn raw data into multilayered maps that reveal much more about a public-health problem than lower-tech tools can uncover by themselves. Miranda has used her mapping skills to tackle problems as widespread as childhood lead poisoning in Durham’s older neighborhoods; the rise in asthma rates after Hurricane Floyd; and the high rate of Lou Gehrig’s disease among some Gulf War veterans.
Miranda started with Duke’s patient database, which told her who had already sought treatment through the university’s health system for flu symptoms. “What became apparent was that we did not know where people who had been vaccinated lived relative to where people who were sick lived,” she says. The Durham County Health Department’s
immunization records were handwritten; its patient consent forms had not been computerized. Knowing that timeliness was crucial, Miranda hired six high-school students, trained them in medical privacy issues, and paid them to enter thousands of records into CEHI computers during their December holiday break.
The resulting maps showed that East Durham’s Fayetteville Street corridor, which is predominantly African American, had a particularly low immunization rate. Until then, most of the health department’s vaccination clinics were located in West Durham, where the population is whiter, and in the downtown commercial district. As soon as Harris reviewed the maps, she decided to match the geography to the need—and offer clinics in some of the most underserved areas, including a school and church in the Fayetteville Street corridor.
Moreover, Miranda’s mapping system provided a tool for the health department to improve its strategic planning in the future. The spatial data architecture now contains many layers of information: the locations of pediatric patients for whom influenza presents the greatest danger (children with asthma, for example); demographic information like race, age, sex, and neighborhood characteristics; and community resources like churches and schools, which show health planners where they can set up clinics. Knowing who’s getting sick and who’s not getting immunized will allow the health department to target the populations in greatest need during upcoming flu seasons.
“We need to reach out to those communities to say, ‘Help us understand what happened. Why did you not take advantage? What could we have done differently?’” Harris says.
Over the past decade, Miranda has become a national leader in using GIS technology to tackle complex public-health problems, particularly those affecting minorities and the poor. She does this not only by crunching data. She also works directly with community members to figure out how to use that data to meet their needs.
“She terms herself as a geek, but she is a very unusual geek,” says Harris, the county health director. “She is very much into social justice. She is passionate about what she does. In spite of her being highly intelligent and highly into technology, when she talks about it, she can put faces on it.”
Miranda does this work under the auspices of CEHI, the research, education, and outreach program she founded, which established partnerships with experts throughout the university. CEHI’s “most mature” project, as Miranda calls it, is designed to address lead exposure in children. Lead—found in paint, drinking water, and even soil tracked into homes—can cause brain damage, learning disabilities, behavior problems, and slowed growth. Children are particularly prone to its effects, both because they ingest paint in older houses and because their bodies absorb the element more easily.Children’s nervous systems are also more sensitive than adults’ to lead’s effects.
When they began working with CEHI researchers about ten years ago, Durham County health officials already knew that lead poisoning was a problem, especially in poorer communities. But “we had not done anything to go out and identify those kids who could be at risk,” Harris says. “We were just reacting to the ones that were diagnosed by physicians.” A citizen group called Durham CAN (Congregations, Associations and Neighborhoods) had been lobbying the department to address the issue more aggressively. But county officials said they lacked the money for universal screening and had no precise way to locate the children most in need of testing.
Maps that identified broad risk factors—the oldest houses or the poorest communities—wouldn’t add much value. “You see a lot of public-health data displayed with GIS these days,” Miranda says. “A lot of it is done at the county or ZIPcode level. If I said to the director of the health department, ‘Boy, you really need to worry about ZIP code 27704,’ she would look at me and say, ‘Well, tell me something I don’t know already.’ ”
Instead, Miranda and her staff met with community leaders to find out what information would be most helpful. They then created a model that zoomed down to the level of individual properties. They introduced every possible variable that could influence lead exposure, whether the information existed at the level of an individual property or tax parcel (for example, tax value, year of construction, and owner vs. renter occupancy) or a block or group of blocks used for compiling census data (racial demographics, median family income, childhood poverty rate). CEHI also negotiated confidential access to lead-screening data: the addresses, ages, and blood levels of the youngsters who had been tested and whether their families qualified for Medicaid and WIC. Plugging these GIS data into a statistical program, Miranda and her staff calculated how much each factor actually influenced lead levels. They then transferred their findings back into the GIS program—and assigned a risk category to every one of Durham County’s 62,000 residential parcels.
In the resulting color-coded map, the most worrisome 10 percent of properties are shaded dark blue, the next 10 percent bright green, and so on. “So now, if the health director says to me, ‘I have the resources to screen 1,000 kids; where should I go?’ I don’t have to tell her, ‘Go to ZIP code 27704,’ ” Miranda says. “I can tell her, very specifically, ‘Go to these houses.’ ”
Not only that. The model can map the nearest facilities to set up community-based test sites. “You could say, ‘Here are all the churches where you can do blood-lead screening,’ ” says Miranda. “Or you could do it at the pharmacies. If you wanted me to map all the tiendas to reach the Hispanic community, well, I can map that for you.” The maps also make it possible to select houses for cleanup and to target communities for education.
article continues on page two. |
