Predicting Lead Hazards
A new research project from the Center for Data Science and Public Policy is helping Chicago’s public health officials reduce the risk of lead poisoning.
If there’s anything we all should have learned from the recent contamination of drinking water in Flint, Michigan, it’s this: lead poisoning is no joke. It’s irreversible. It primarily affects the population that’s most vulnerable: children younger than six. And it’s not just confined to Michigan; it affects every city built on a 19th-century infrastructure, including Chicago. Hundreds of thousands of children in the United States are poisoned by lead every year, and the health effects are severe.
Now a group of researchers from the Center for Data Science and Public Policy (DSaPP), a collaboration between the Harris School of Public Policy and the Computation Institute that uses mathematical models to solve social problems, is trying to find a way to stop it.
In theory, children in Chicago should have their blood tested for lead periodically throughout their early years of life. If high lead levels are detected, the Chicago Department of Public Health sends inspectors to the child’s home to identify lead-based paint or other hazards and work with the property owner to remove those hazards.
But here’s the big problem with that approach, says Eric Potash, a postdoctoral researcher at DSaPP: “It’s a bit like screening for cancer. But unlike some cancers, which are actually curable in their early stages, the damage of lead poisoning is permanent. All you can do is try to stop it from getting worse.”
Potash and Joe Walsh, his fellow researcher at DSaPP, have a better idea: they want to stop lead poisoning before it happens. To do that, they’ve developed a mathematical model to identify the children in Chicago who are most likely to be poisoned so that the hazards can be removed before they become a problem. This endeavor began as a Data Science for Social Good summer fellowship project two years ago and was selected by DSaPP to be expanded into a year-round initiative.
Potash and his fellow researchers want to stop lead poisoning before it happens. To do that, they’ve developed a mathematical model to identify the children in Chicago who are most likely to be poisoned so that the hazards can be removed before they become a problem.
Although the United States banned lead-based paint in 1978, 90 percent of the housing stock here was constructed before the ban took effect, and many houses and apartments, particularly in poor African-American neighborhoods, have not been rehabbed and are poorly maintained, meaning dangerous lead-based paint remains. As this paint ages, it disintegrates into lead-laden dust that young children (through normal hand-mouth behavior) can ingest. In 2013, the Illinois Department of Health reported that 10,361 children in Chicago age five or younger had blood lead levels above 5 micrograms per deciliter, the level the Centers for Disease Control consider the threshold for lead poisoning.
“Lead poisoning is irreversible and really damaging,” says Lindsay Knight, DSaPP’s project and development manager. “It’s a long-term arc. We don’t know the full implications. It causes a decrease in cognitive abilities and impulse control and an increase in crime and truancy. In the 21st century, using a child as a canary in a coal mine to find lead hazards is ludicrous.”
And as Potash points out, lots of children are falling through the cracks, usually those who are most at risk. These are kids who don’t get regular medical care or who move around a lot in neighborhoods where residents aren’t likely to report housing code violations. There are more than 40,000 children born every year in the city, and of those, about 2,000 will eventually test positive for lead poisoning. Finding those children, Potash says, is like looking for needles in a haystack. The DSaPP model works like a magnet to pick out those needles.
Potash and Walsh began with several data sets from various city agencies: the addresses where children have been poisoned by lead in the past; the results of lead home inspections; and Women, Infants, and Children program (WIC) enrollment data. They also had access to public databases that contain information about city housing, such as which buildings have been reported to have housing code violations.
The data aren’t perfect. There are misspellings and typos in the names and addresses. Because the city largely depends on citizens to report housing violations, the database is far from complete. But Potash hopes the data he and Walsh have are reliable enough, and that the model they’ve built is good enough, to provide some notion of who, exactly, is most at risk. Then it’s time for the practical part of the project because, as Potash points out, “it’s only useful if there’s an intervention.”
This month, inspectors from the Chicago Department of Public Health (CDPH) will begin conducting inspections of homes of young children. Homes with lead hazards will need to be remediated, with some of the costs subsidized by a $3.5 million grant the CDPH received from the U.S. Department of Housing and Urban Development. The project will move slowly at first, with 30 inspections a month for a year. If it the model works, DSaPP plans to expand the scale of the project in Chicago and into other cities.
“The main point of every DSaPP project is that it can be applied by our partners and exportable to different users,” says Knight. “It’s about tangible social impact. In most places in a university, theory and research are primary and practice is secondary. For us, practice and impact are the primary motivations.”
— Aimee Levitt