How the Bacon You Eat Can Affect Your Neighbors
By Oluwadamilola Fayanju, MD, MA, MPHS
To do good work.
It’s a goal that seems both lofty and concrete, broad but focused. It’s the central tenet of both my personal and professional lives — and it’s why I’ve committed my career to improving the equity and efficacy with which healthcare is delivered, both to the patients I treat as a surgical oncologist and through my research on disparities and value in medicine.
I’m thrilled to be joining the leadership team at Duke Forge, where I hope to harness the rising tide of data science to lift all boats. Along the way, I also want to highlight and celebrate the remarkable work of researchers across Duke who are using the power of big data to address a wide variety of health disparities.
When I sat down to think of a topic for my first monthly blog for the Forge, the work of two of my colleagues in the Department of Surgery immediately came to mind. Dr. H. Kim Lyerly and Dr. Julia Kravchenko have worked together for more than a decade on projects that examine associations between environmental factors and health outcomes including cancer. For example, in a 2014 paper, they demonstrated that passage of North Carolina’s 2002 Clean Smokestacks Act was associated with a significant reduction in deaths related to pulmonary disease.
Now, the team’s most recent publication has earned a high degree of national attention, including mentions in the New York Times, on Fox News and the Huffington Post, and in the Washington Post. Dr. Lyerly and Dr. Kravchenko had hypothesized that persistently poor health outcomes among people living in southeastern NC might be due to converging demographic, socioeconomic, behavioral, and access-to-care factors including residents’ proximity to large-scale industrial farming operations known as concentrated animal feeding operations, or CAFOs. There are currently more than 1500 CAFOs in NC, more than 900 of which are devoted to raising hogs.
Earlier studies have shown that people who work at CAFOs have worse performance than those who don’t for several important health indicators. Compared with similar groups of people who don’t work in industrial agriculture, CAFO workers have higher rates of anemia and chronic kidney disease due to toxin exposure. They are also more likely to suffer miscarriage, have children who are low weight at birth, and have more exposure to antibiotic-resistant organisms. Dr. Lyerly and Dr. Kravchenko, however, were able to build on this foundation by devising an unusually robust approach to controlling for confounding factors while asking a different but equally important question: whether living near large hog CAFOs (and potentially being exposed to the waste and other contaminants they produce) could contribute to adverse health outcomes in nearby residents.
Or, in other words, after controlling for other factors, is it possible that simply living near a hog farm could be enough to make you sicker?
As they set out to investigate potential associations between living near a CAFO and health outcomes, Dr. Lyerly and Dr. Kravchenko first had to identify and link multiple datasets with a large set of variables to appropriately adjust for potential confounders. To explore the key health outcomes of interest — disease-specific mortality, emergency department visits, and hospital admissions — the team accessed data from the State Center for Health Statistics and the Healthcare Cost and Utilization Project’s State Emergency Department Database, and combined it with socioeconomic, environmental, and behavioral data from other databases (table).
Dataset Type of Data Covariates American Community Survey Socioeconomic Median household income; education levels North Carolina Division of Water Resources Environmental Geographic location of CAFOs; number of swine/CAFO Area Health Resources Files (AHRF) Access-related No. of primary care providers*100,000 residents; percent uninsured Behavioral Risk Factor Surveillance System (BRFSS) Behavioral Smoking prevalence
The team then applied a battery of statistical methods including logistic regression, generalized estimating equations, propensity-score matching, and one tool that the team developed specifically for the study: Distance from Source of potential Contamination (DiSC), a methodology they used to evaluate changes in health outcomes relative to distance from a CAFO.
The results of this painstaking analysis were profound: among the nearly 2.3 million residents in the 221 zip codes where there was a hog CAFO, as well as among the subset of 400,000 residents living in the 56 zip codes within the highest 25thpercentile for hog density (> 215 hogs/km2), rates of all-cause mortality, infant mortality, and mortality secondary to anemia, kidney disease, tuberculosis, or septicemia were significantly higher when compared with residents who lived at a greater distance from CAFOs. In addition, both emergency department visits and hospital admissions were more frequent in these CAFO-proximate areas.
The authors ultimately concluded there are “poor indicators of health that are not solely due to the impact of converging demographic, socioeconomic, behavioral, and access-to-care factors, but are also due to the additional impact of multiple hog CAFOs located in this area,” although they cautioned that a causal relationship between outcomes and specific exposures from hog CAFOs had not been established. The importance of these findings are clear, especially in the wake of an unusually severe 2018 hurricane season that led to large-scale flooding and feces-filled run-off from CAFOs and ruptures of accompanying waste-holding facilities.
We at Duke are fortunate to be able to draw upon the deep expertise and knowledge of data science experts such as Dr. Kravchenko, clinician-scientists such as Dr. Lyerly, and the constellation of scholars, staff experts, and resources at the Nicholas School of the Environment and the School of Medicine to illuminate the complexities of potentially causal relationships between the food we eat and the ways we raise it. In the meantime, it’s worth noting that living near a CAFO is associated with many factors related to access to care — including insurance and transportation — that are potentially modifiable, are unrelated to the environment per se, and reflect the many inequities that can be found along the rural-urban continuum. While we continue to explore CAFOs and their effects on health outcomes, we must also use the tools already available to us to address the injustices in our midst.
I’m looking forward to discussing in future blog posts the various ways in which we can and should leverage big data for the sake of doing good work — for ourselves, for our neighbors, and for our planet.
Originally published at https://forge.duke.edu/blog/how-bacon-you-eat-can-affect-your-neighbors on April 8, 2019.
