Poster - Impact of air pollution and hospitalizations in Metro Detroit region
Abstract
Purpose: To study the impact of air pollution on health outcomes in Detroit Metro, Michigan. Detroit, Michigan contributes to more than 80% of air pollution in Michigan
Method: This analysis involves 3 steps.
Use OpenAPIs provided by Environmental Protection Agency (EPA) and Medicare/Medicaid program to download air pollution data and hospitalization data respectively
Conduct spatial and temporal analysis of the data to identify the regions and dates of interest
Use Regression Analysis to identify the relationship between air pollution and hospitalization
Result:
Spatial and temporal analysis of 50,000 hospitalization records identified 3 regions of interest in Detroit Metro area.
Strong correlation was found between air pollution and the duration of stay in hospitals (p value= 0.000, R=0.71)
Mild correlation was found between air pollution and the number of hospital visits (p value = 0.01, R=0.5)
Software used: R (ggplot, dplyr, jsonlite, httr), Arcgis
Background
Air pollution is a significant health risk to our society. Air is considered polluted when it contains particulate matters (PM).
According to the Environmental Protection Agency (EPA), PMs range in diameter from course (>2.5µm and <10µm), fine (< 2.5 µm) and ultrafine (<100 nm),
PM linger in the body for a long time. They can penetrate soft tissues such as brain and lungs and disseminate throughout the body via blood vessels. PM’s have found to cause neurological, respiratory and integumentary diseases.
Purpose
Analyze the correlations between air pollution levels and health outcomes in metro Detroit region with the goal to educate the community of the health risks
Method
Gathered the EPA sensors from EPA Air Quality Open API for metro Detroit area with the following R code:
For each sensor location, gathered the annual PM 2.5 observed values. I removed the empty records. I used data from the years for which EPA had data.
For get the hospital visits and hospital duration, I used Medicare and Medicaid’s open apis. The dataset versions were listed for each year and so I had to combine the datasets into a dataframe.
I visualized the Medicare data and identified 5 zip codes for which there was also EPA air quality data. I conducted regression analysis and found the correlations between air quality and the number of hospital visits and air quality and the duration of stay.
Results
Figure 1: Quad chart with the results from analysis
Spatial analysis showed 8 EPA sensors in Detroit Metro region (Figure 1a)
Trend chart of duration of hospital stay shows 3 Detroit regions consistently have longer duration of stay (Figure 1b)
Strong correlation between air pollution and duration of stay (Figure 1c)
Mild correlation between air pollution and the number of hospital visits (Figure 1d)
