Maggie M. Kovach, Christopher M. Fuhrmann, Charles E. Konrad II Southeast Regional Climate Center University of North Carolina at Chapel Hill Conor Harrison.

Community Level Indicators of Heat Related Morbidity in North

Carolina

Maggie M. Kovach, Christopher M. Fuhrmann, Charles E. Konrad II

Southeast Regional Climate Center

University of North Carolina at Chapel Hill

Conor Harrison

Department of Geography

University of North Carolina at Chapel Hill

Previous Literature• What geographic locations are at greater risk

for heat-related illness?

– Urban areas are higher risk for heat illness due to higher temperatures (CDC, 2004), (Jones et al. 1982), (Harlan et al. 2006) (Reid et. al 2009)

• What specific populations are at risk?

– Young adults and working population experience higher rates of heat related illness in NC (Lippmann in review)

– Poor, minorities, socially isolated, elderly (CDC, 2004)

dailykos.com

USA Today

Previous Literature

• Are agricultural workers at greater risk for HRI?– In the US, North Carolina accounts for 57% of all heat related deaths

among crop workers from 1992 to 2006 (Luginbuhl et al. 2008)– African Americans, Latino workers (Richardson and Gregory 1997,

Richardson and Mirabelli 2002).

EPAAgricultural Worker Health Project

: David Baconers.usda.gov

Census 2000 Data Potential Relationship to HRI

Race: (Hispanic, Black, White) Populations most vulnerable to heat

Citizenship: (Naturalized, Non-Citizen, Spanish speakers) Agricultural workers/social isolation

Income: (food stamps, below $20,000, median household income) Wealth or poverty

Housing Type: (Mobile home, multihouse, rental occupancy) Wealth or poverty/Social isolation

Electricity source:(LPG, natural gas, electricity, heating oil) Rural or Urban/Poverty

National Land Cover Database (2008) Potential Relationship to HRI

Developed Land: High intensity, medium intensity, Low intensity Rural or Urban/Geographic Locations

Cultivated Crops: 30 total crops (e.g. tobacco, corn, apples, oats, peanuts)

Agriculture workers/Microclimate of fields

Forest: Evergreen, Mixed forest, deciduous forest, woodland Cooling potential from vegetation

Data Sources

North Carolina Disease Event Tracking and Epidemiologic Tool (NC DETECT)

Dates Available: 01/01/2007 – 12/31/2008ICD 9: 992

Methodology 1.) Transform data to a similar spatial scale.

2.) Evaluate relationship between heat-related hospital admissions and land cover & socioeconomic variables through Pearson correlations.

3.) Perform regression analysis

of risk factors associated

with heat-related illness.

Geographically Weighted Regression is a spatial regression technique that models spatially varying relationships. It generates a separate regression equation for each census tract based on the values of neighboring census tracts.

ED HRI admission per 100,000 people

N = 2590 ED Visits (Entire State)N = 2248 ED Visits (Piedmont and Coastal Plain)

ED heat admissions for North Carolina

Where is HRI geographically located?

Variables RDeveloped Land -0.34

Population Density -0.31Natural Gas (Urban) -0.27Median Year Built -0.26

Multi-house -0.25Renters -0.24

Evergreen Land 0.32LPG (Rural) 0.29Woodland 0.22

Developed Land

Evergreen Land Cover*p-values < 0.05

Rural populations of North Carolina are at increased risk for heat related illness compared to urban populations.

Urb

an

Ru

ral

Variables RMobile Homes 0.37

Mobile Homes

Is poverty associated with increased HRI?

With the exception of mobile homes, correlations are weak for HRI and other measures of poverty (i.e. food stamps, median income, home values below $10,000, incomes below $20,000).

Non-Citizens

Caucasian Population

Variables RCitizens 0.14

Caucasian 0.07Non Citizens -0.12Nationalized -0.11

Spanish speaking -0.11Hispanic -0.08

*p-values < 0.05

Are specific populations at greater risk HRI?

Correlations are weak for HRI and different minority populations.

Fruits and Vegetables

Wheat Crops

All Crops

Variables RAll Crops 0.20

Corn 0.17Soybean 0.15

Fruits &Vegetables 0.13Wheat Crops 0.12

Tobacco & Cotton 0.10*p-values < 0.05

Are specific farm laborers at higher

risk for HRI?

Of the 30 crops examined only a few were correlated with HRI.

Variables: Home values below $10,000, Rental Occupancy, Mobile Homes, Cropland (all crops)

Geographically Weighted Regression Analysis

Local R2 values:

Local R2 values: these values range between 0.0 and 1.0 and indicate how well the local regression model fits observed HRI admissions. In this model, the R2 predicts up to 0.700 in particular areas .

Corn Crops

Cotton Crops

Soybean Crops

CroplandCoefficient

Tobacco Crops

Geographically Weighted Regression Analysis

The positive relationship between crops and HRI is located in the Northern Piedmont and Northern Coastal Plain, where soybean, tobacco and cotton agriculture is located.

Geographically Weighted Regression Analysis

Home Values below 10,000 Coefficient

Rental Occupancy Coefficient

Mobile HomesCoefficient

GWR Coefficients

-200 - -100

-99 - 0

1 - 100

101 - 330

331 - 595

These maps display the relationship between the coefficients and HRI.

Reds are positive and blues are negative.

Summary• In North Carolina, heat related illness (HRI) is found predominately in rural areas with no

development, low population density, and locations with more “green space.”

• Mobile homes, a proxy for rural poverty, increase a community’s risk for heat-related illness. Other indicators for poverty such as food stamps, income below $20,000 or home value below $10,000 have less influence on HRI.

• No correlations were observed for minority populations and HRI. However, previous heat mortality research found that minority populations are less likely to seek care (Richardson and Mirabelli 2002).

• Agriculture is positively correlated with HRI in the Northern Piedmont and Northern Coastal Plain of North Carolina, where the tobacco, cotton and soybeans are the predominate cash crops.

• In the Sandhills and Southern Coastal Plain of North Carolina, socioeconomic factors such as income and mobile homes increase the likelihood of HRI.

Current Work

• Incorporate NC-DETECT data for 2009, 2010

• Examine heat wave, non-heat wave heat related ED heat admissions, ages of HRI ED patients.

• Incorporate climate information with individual and neighborhood risk factors to model heat risk.

Agricultural Worker Health Project : David Bacon

Acknowledgements: NC Division of Public Health

NC-DETECTSoutheast Regional Climate Center

The NC DETECT Data Oversight Committee does not take responsibility for the scientific validity or accuracy of methodology, results, statistical analyses or conclusions presented.

Contact: [email protected]