Behind the Metrics: Lead Risk
Aug. 14, 2018
Choosing an Analytic Approach: Vox Media, Washington State Department of Health, and Us
Exposure to lead affects the health of hundreds of communities across the United States. The CDC reports about 500,000 children nationwide have dangerous levels of lead in their blood. Because of this, the Dashboard team wanted to include a measure of lead exposure on the site. However, it’s quite a challenge to access high-quality city-level information about lead exposure across the United States.
After months of background research (and a bit of luck), our team found an analytic approach comprehensive enough to describe cities across the country and useful to Dashboard users. After extensive investigation, we chose to replicate the strategy developed by researchers at the Washington State Department of Health and their colleagues at Vox Media.
The Dashboard presents two lead-related measures: the percentage of housing in a city of census tract at risk for lead and an overall score of relative lead risk. The data for these metrics comes from the 2016 American Community Survey (part of the US Census).
One Topic, Two Metrics
The housing with potential lead risk metric is the percent of housing at risk for lead in a given city or census tract, adjusted by the age of housing only. This metric takes one component of the Vox/Washington State Department of Public Health analysis and displays it on its own. Housing built before 1940 and between 1940 and 1959 is considered to be at highest risk for lead. Housing built from 1960 to 1999 has an attenuated risk, and housing built after 2000 is considered to be the safest.
The lead exposure risk index uses information about poverty and age of housing to describe the overall risk of lead exposure, with information about older housing stock (described above) and the percentage of people living below 125% of the federal poverty line contributing to a higher score. The lead exposure risk index is a ranking (1 is lowest risk; 10 is highest risk) of this score relative to the other geographies on the Dashboard (not the entire US).
The Dashboard represents risk in both ways because although the two metrics are related, they are informative in different ways. While the age of housing is very informative about the likelihood of lead exposure, taking economic factors (like poverty) into account accounts for the quality of the housing stock. For example, communities with more economic resources will likely have better maintained homes that—despite housing age—pose a lower risk of lead exposure, compared to less privileged communities.
Interestingly, in Miami, Florida, only 22.2% of housing is at risk for lead but the city has a lead exposure risk index value of 9. While the index tells us that Miami residents have a relatively high risk of lead exposure, a relatively small percentage of housing in Miami has potential lead risk. This pattern suggests that poverty, not housing, is the main driver of Miami's score on the lead exposure index.
The lead exposure risk index users see on the Dashboard mirrors Vox Media’s calculations for its lead exposure map. Users in a location not included on the Dashboard might find Vox’s resource particularly interesting – enter any street address in the US to find the lead risk score associated with it.
For a detailed explanation of how the Dashboard adapted Vox and the Washington State Department of Health’s methods to describe the percentage of housing with potential lead risk, and how our calculation of the risk index differs slightly from Vox’s, check out the “Notes on analysis” section for each metric in our Technical Document, Part 1.
We’re grateful to the Vox/Washington State Department of Health team for sharing their code on Github.