Identifying Climate Thresholds for Heat-Related Illness in North Carolina

EPA Grant Number: F13D10708
Title: Identifying Climate Thresholds for Heat-Related Illness in North Carolina
Investigators: Kovach, Margaret Mae
Institution: University of North Carolina at Chapel Hill
EPA Project Officer: Lee, Sonja
Project Period: August 19, 2014 through August 19, 2016
Project Amount: $84,000
RFA: STAR Graduate Fellowships (2013) RFA Text |  Recipients Lists
Research Category: Academic Fellowships , Fellowship - Geography


The objective of this research is to identify climate thresholds that lead to increases in emergency department visits for heat-related illness. Climate thresholds represent climate values (e.g. temperature, heat index, humidity) beyond which heat-related illness emergency department admissions increase significantly. Climate thresholds will be identified for separate categories of age and gender for multiple populations across North Carolina to determine the differential impact of heat stress.


Heat-related illness is measured using all 113 emergency departments across the state of North Carolina, allowing for complete spatio-temporal cover from 2007 to 2012 at the ZIP code level. Heat-related illness emergency department visits also are linked to daily weather stations, census data and land cover data to investigate different locations (i.e., urban locations, areas with large amounts of crop land, locations with a high percentage of the population living poverty, etc.). Generalized additive models will relate heat-related illness to climate variables to smooth the data for climate threshold calculation. Climate thresholds will be calculated objectively for each climate variable, using an automated routine that compares the similarity of heat-related illness rates above and below climate values iteratively across a wide range of potential threshold values. Heat-related illness incidence also is mapped with demographic, land cover and socioeconomic variables to identify potential risk for heat-related illness. Risk factors are identified through a geographic information system framework and regression techniques, such as spatial regression and multilevel modeling.

Expected Results:

The research hypotheses proposed in this project suggest significant variability in heat-related illness across different populations and within different time periods. Areas with high heat-related illness incidence will be locations in which one or more demographic groups are unable to adapt to heat, either due to lower socioeconomic status, increased physical exposure to heat (e.g. high numbers of an outdoor work force) or a land cover that retains heat (e.g., developed land, certain agriculture fields). Climate thresholds will vary significantly across the state of North Carolina depending on the demographic, spatial location and season. For instance, the climate threshold for the elderly population will be lower than for younger demographics because of their inability to thermoregulate as efficiently. Additionally, the shape (i.e., linear vs. nonlinear) of these thresholds will vary, since slight temperature changes affect elderly more than the younger demographic.

Potential to Further Environmental/Human Health Protection

General circulation models of climate change project extreme heat events to increase in severity, duration and intensity, placing future populations at greater risk for adverse health effects. This motivates the need to understand risk locations and population-specific climate thresholds, which can be integrated into an early heat-warning system and targeted public health interventions. Such measures are needed since heat-related illness is easily preventable with adequate warning and appropriate education and response.

Supplemental Keywords:

early warning systems, heat health risk, urban climate

Progress and Final Reports:

  • 2015
  • Final