You are here:
Hurricane exposure and county fetal death rates, utilization of a county environmental quality index for confounding control.
Citation:
Grabich, S., C. Gray, K. Rappazzo, L. Messer, J. Jagai, AND D. Lobdell. Hurricane exposure and county fetal death rates, utilization of a county environmental quality index for confounding control. Presented at Society for Pediatric and Perinatal Epidemiologic Research, Seattle, WA, June 23 - 24, 2014.
Impact/Purpose:
This study utilized the EPA created Environmental Quality Index as a measure to control environmental confounding in an epidemiologic investigation.
Description:
The effects of natural disasters on public health are a rising concern, with increasing severity of disaster events. Many disaster studies utilize county-level analysis, however most do not control for county level environmental factors. Hurricane exposure during pregnancy could influence fetal death through mechanisms related to access to care, nutrition, injury and disruption of infrastructure and health facilities. Among Florida women pregnant in 2004 during the hurricane season, we investigated the association between hurricane exposure and county fetal mortality rates. We categorized county hurricane exposure by maximum wind speed during each hurricane. Confounding and additive effect measure modification was assessed with linear models using the county-level Environmental Quality Index (EQI) developed by the Environmental Protection Agency. We adjusted for the overall EQI and by specific environmental domains (air, water, land, built, socioeconomic). The relationship between hurricane and fetal mortality was not consistent among hurricanes. However, analysis of the most severe hurricane, Charley, suggested increase in wind category was associated with increased fetal mortality rate (Rate Difference (RD) from 1.36(95% CI:-2.14, 4.85) to 1.65(-1.85, 5.38)) reliably among models. The addition of the EQI domains adjusted estimates towards the null value consistently across all models. The socioeconomic (SES) domain interacted antagonistically with hurricane estimates across all models (e.g. for Charley crude estimate hurricane RD (1.65), and estimate interaction with SES RD 1.39 (95% CI: -2.14, 4.85)). This analysis demonstrates the need to include ambient environment in modeling county level analysis. This abstract does not necessarily reflect EPA policy.