You are here:
Reliability testing across the Environmental Quality Index and national environmental indices.
Citation:
Grabich, S., C. Gray, J. Jagai, Y. Jian, L. Messer, K. Rappazzo, AND D. Lobdell. Reliability testing across the Environmental Quality Index and national environmental indices. Epidemiology Congress of the Americas 2016, Miami, FL, June 21 - 24, 2016.
Impact/Purpose:
The purpose of this study was to explore the reliability of the EPA created Environmental Quality Index (EQI). Exploration of domain specific indices were used since the EQI is unique and no other index captures the same information.
Description:
One challenge in environmental epidemiology is the exploration of cumulative environmental exposure across multiple domains (e.g. air, water, land). The Environmental Quality Index (EQI), created by the U.S. EPA, uses principle component analyses combining environmental domains (air, water, land, social, and built environment) to construct a county-level environmental metric for the United States. Although challenging, it is important to assess consistency across indexes to verify appropriate exposure measurement. Since there is no similar total index covering the full U.S., we compared the EQI’s air, water, and land domains to the California Communities Environmental Health Screening Tool (CES). We first assessed the Pearson correlation coefficient (rho) between county index values. Then, we used generalized linear models to estimate the county low birth weight (LBW) rate difference per 1000 births (RD [95%CI]) available through CES with continuous index scores. In preliminary analysis comparing the EQI to the CES, moderate correlations were observed across all comparable domains (rho=air 0.6, land 0.5, and water 0.4). Results of regression analysis showed that the CES was associated with LBW (216.8 [100.1, 333.6]), and the EQI water domain was associated with LBW (114.3 [64.2, 164.5]), but the EQI air and land domains were not associated with LBW. Overall the moderate correlation between county indices provides some validity between these environmental measures, although differences could be due to differing number of data sources included in the CES (e.g. EQI air domain n=87 data sources and CES air variables n=3). We plan to examine consistency further by comparing the EQI’s sociodemographic domain to the Neighborhood Deprivation Index in future analyses. The EQI gives a more total picture of the U.S. county environmental exposure by including more domains and sources of exposure then other similar indices. This abstract does not necessarily reflect EPA policy.