Regional-Scale Assessment of N2O Emissions within the US Corn Belt: The Impac...
Spatial Analysis of Radon Risk in Schools
1. A Spatial Analysis of Geologic Conditions
and Radon Risk
Stephanie Foster, David Yeomans, Julia Bryant, Efomo Woghiren,
Brian Lewis, Andrew Dent
5th International Conference of Medical Geology
27 August 2013
Agency for Toxic Substances and Disease Registry
Division of Toxicology and Human Health Sciences
Geospatial Research, Analysis, and Services Program (GRASP)
9. Deaths per Year,United States
2,800
3,900
17,400
3,000
21,000
Home Fires Drownings Drunk Driving Lung Cancer
from
Secondhand
Smoke
Lung Cancer
from Radon
US EPA 2003 Assessment of Risksfrom Radon in Homes
CDC 2005-2006 National Center for Injury Prevention and Control Report
10. 3
20
32
62
120
150
18
0
20
40
60
80
100
120
140
160
0.4 1.3 2 4 8 10
RiskofCancer(per1,000people)
Lifetime Radon Exposure (pCi/L)
Smoker
Never
Smoked
Risk of Lung Cancerfrom Radon Exposure (per 1,000)
US EPA 2003 Assessment of Risksfrom Radon in Homes
11. 3
20
32
62
120
150
2 4
7
15
18
0
20
40
60
80
100
120
140
160
0.4 1.3 2 4 8 10
RiskofCancer(per1,000people)
Lifetime Radon Exposure (pCi/L)
Smoker
Never
Smoked
Risk of Lung Cancerfrom Radon Exposure (per 1,000)
US EPA 2003 Assessment of Risksfrom Radon in Homes
Average
indoor
level
EPA
action
level
Indoor radon has
been measured in
homes up to
3,200 pCi/L
Average
outdoor
level
12. 23 23.6
50.6
16.7
10.8
Prostate Breast Lung Colon Pancreas
Age-AdjustedCancerMortalityRatesper 100,000
NationalProgram of Cancer Registries,Centers for DiseaseControl and Prevention,2005-2009 Data
13. 23 23.6
50.6
16.7
10.8
Prostate Breast Lung Colon Pancreas
Age-AdjustedCancerMortalityRatesper 100,000
NationalProgram of Cancer Registries,Centers for DiseaseControl and Prevention,2005-2009 Data
14. Radon in Schools
Some states require testing in homes during real estate
transactions
Most states do not require
testing in public schools
Children are especially
susceptible to radon exposure
• They breathe proportionately
more air than adults
• Their lungs are still developing
15.
16. Problem Statement
School districts do not have the
means to test every school in
their jurisdiction on a regular
basis
Current publicly available
data is limited to
county-level averages
Provide guidance to
school districts for
focused radon
testing by
performing spatial
analyses on indoor
radon
measurements from
private homes
Objective
18. DATA
School and residential results from the Florida Radon
Program
Soil characteristics, geology, and uranium concentrations
from United States Geological Survey (USGS)
Temperature and precipitation from National Oceanic and
Atmospheric Administration (NOAA)
Housing and poverty data from the United States Census
(2010)
National Center for Education Statistics (NCES) school data
19. METHODS
Geocoding
• Florida residential and school addresses and NCES school
addresses using Centrus desktop software
• QA/QC of locational accuracy
Analysis
• Maximum radon test result per location
• Esri ArcGIS
o Buffers
o Area proportion/weighted averages
• EpiInfo 7 odds and risk ratio calculations
• Logistic regression (SAS 9.3)
23. FLORIDA DATA ANALYSIS – UNADJUSTED
205 993
299 3020
>4
<4
ONE MILE
School
>4 pCi/L <4 pCi/L
Residential
504 4013 4517
3319
1198
OR = 2.08 (1.72,2.52)
Mantel-Hansel x2 = 58.29 (p=0.0000)
RR = 1.90(1.61,2.24)
>4
<4
THREE MILES
School
>4 pCi/L <4 pCi/L
Residential
630 5546 6176
3654
2522301 2221
329 3325
OR = 1.37 (1.16,1.62)
Mantel-Hansel x2 = 13.99 (p=0.0002)
RR = 1.33 (1.14,1.54)
24. FLORIDA DATA ANALYSIS – LOGISTIC REGRESSION
Estimate P-value OR 95% CI
Max.Radon Residential 0.0327 0.0069 1.033 1.00-1.058
Summer Precipitation 0.4405 0.0002 1.553 1.229-1.964
Spring Precipitation 0.4019 0.0192 1.495 1.068-2.092
Fall Precipitation -0.8143 <0.0001 0.443 0.315-0.623
Permeability -0.2437 <0.0001 0.784 0.726-0.846
Thickness 0.0288 0.0001 1.029 1.014-1.044
Hydrologic Group -1.4744 <0.0001 0.229 0.159-0.330
Drain 0.6371 <0.0001 1.891 1.443-2.478
Slope -0.2673 0.0047 0.765 0.636-0.921
Annual Flood Frequency 0.4714 0.0156 1.602 1.094-2.347
Average Fall Temperature 0.2141 <0.0001 1.239 1.135-1.352
Average Spring Temperature 0.2649 <0.0001 1.303 1.158-1.467
Uranium (ppm) 0.1017 <0.0001 1.107 1.075-1.140
25. FLORIDA DATA ANALYSIS – LOGISTIC REGRESSION
Estimate P-value OR 95% CI
Summer Precipitation 0.4514 <0.0001 1.571 1.246-1.980
Spring Precipitation 0.3827 0.0001 1.466 1.050-2.048
Fall Precipitation -0.7966 0.02 0.451 0.322-0.631
Permeability -0.2559 <0.0001 0.774 0.717-0.835
Thickness 0.0297 <0.0001 1.03 1.015-1.045
Hydrologic Group -1.5541 <0.0001 0.211 0.147-0.303
Drain 0.6813 <0.0001 1.976 1.512-2.583
Slope -0.2601 0.0054 0.771 0.642-0.926
Annual Flood Frequency 0.4839 0.0129 1.622 1.108-2.376
Average Fall Temperature 0.2159 <0.0001 1.241 1.137-1.354
Average Spring Temperature 0.2584 <0.0001 1.295 1.152-1.455
Uranium (ppm) 0.1021 <0.0001 1.108 1.076-1.140
26. APPLICABILITY OF MODEL
Preliminary work not successful
Connecticut
• Preliminary results unsuccessful
• Summer and spring precipitation uniform
Ohio and Colorado to be processed
27. LIMITATIONS
Radon measurements are dependent upon:
• Time of day
• Time of year
• Length of test
• Building construction
• HV/AC systems
Residential tests are voluntary
28. STRENGTHS
First assessment of potential radon risk using maximum
result per location
First analysis to use proximity of residential radon test
results and spatial association with schools nearby
First project to show potential utility of residential testing
results to inform testing programs for public schools
29. NEXT STEPS
Refine model
• Revisit the soil coding
• Explore interaction effects
• Include additional soil characteristics
Predictive Value Testing
Spatial analysis over time
Partner with other Federal, State, and local Agencies
30. Thank you!
Jorge Laguna & Clark Eldridge/ Florida
Department of Health,Radon Program
Francesca Provenzano, Connecticut
Department of Health,Lead/Radon/Healthy
Homes
Dr.Ashok Kumar, University of Toledo and the
Ohio Department of Health Radon Program
31. Stephanie L. Foster, MPH, MA
For more information please contact Agency for Toxic Substances and Disease Registry
4770 Buford Hwy, NE Chamblee, GA 30341
Telephone: 1-800-CDC-INFO (232-4636)/TTY: 1-888-232-6348
Visit: www.atsdr.cdc.gov | Contact CDC at: 1-800-CDC-INFO or www.cdc.gov/info
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the
Centers for Disease Control and Prevention.
Agency for Toxic Substances and Disease Registry
Division of Toxicology and Human Health Sciences
SLFoster@CDC.GOV
770.488.3870
4770 Buford Highway, Mailstop F-09
Atlanta, Georgia 30341
Editor's Notes
Brief introduction including brief summary of ATSDR and GRASP
Talk briefly about how the project developed
Extension of work to explore areas for increased outreach to physicians in Michigan for clinical preventive services in environmental health. One of the hazards of interest was radon.
We noticed discrepancies between EPA radon risk counties and areas with high radon test results.
Points out the areas of discrepancies. Wondering what is going on.
This is the most up-to-date radon risk information from the EPA. These data were collected in the late 1980s and early 1990s and display the potential for elevated indoor radon levels.
These data have not been updated since then and have been questioned in relation to testing results that are beginning to be accumulated by state radon programs.
Curious about what is happening in my area, my home, my daughter’s school. But first, had to learn about radon. Most of you in the audience probably know more about this than me…
Sections 307 and 309 of the Indoor Radon Abatement Act of 1988 (IRAA) directed EPA to list and identify areas of the U.S. with the potential for elevated indoor radon levels. EPA's Map of Radon Zones assigns each of the 3,141 counties in the U.S. to one of three zones based on radon potential.
What do the colors mean?
Zone 1 counties have a predicted average indoor radon screening level greater than 4 pCi/L (picocuries per liter) (red zones)Highest Potential
Zone 2 counties have a predicted average indoor radon screening level between 2 and 4 pCi/L (orange zones)Moderate Potential
Zone 3 counties have a predicted average indoor radon screening level less than 2 pCi/L (yellow zones)Low Potential
Many caveats for what this map is not and what is should not be used for, including in lieu of testing during real estate transactions and all homes should be tested regardless of geographic location.
The Map was developed using five factors to determine radon potential:
indoor radon measurements;
geology;
aerial radioactivity;
soil permeability; and
foundation type.
Radon potential assessment is based on geologic provinces. Radon Index Matrix is the quantitative assessment of radon potential. Confidence Index Matrix shows the quantity and quality of the data used to assess radon potential. Geologic Provinces were adapted to county boundaries for the Map of Radon Zones. To order the U.S. Geological Survey radon potential books, organized by region, with individual chapters for each state, see http://energy.cr.usgs.gov/radon/grpinfo.html
Radon is the result of the natural decay of uranium found in underlying bedrock. ‘When solid radium decays to form radon gas, it loses two protons and two neutrons. These two protons and two neutrons are called an alpha particle, which is a type of radiation. The elements that produce radiation are called radioactive. Radon itself is radioactive because it also decays, losing an alpha particle and forming the element polonium.’ (From USGS, “The Geology of Radon”) Radon gas can become part of the surrounding environment. It can be incorporated into water, if present, or can diffuse through the soil. In both cases it can ultimately release into the air.
‘Radioactivity is commonly measured in picocuries (pCi). This unit of measure is named for the French physicist Marie Curie, who was a pioneer in the research on radioactive elements and their decay. One pCi is equal to the decay of about two radioactive atoms per minute.’ (From USGS, “The Geology of Radon”)
If it is present it can become trapped inside homes through multiple potential pathways into a home. Cracks in a home’s foundation, through a sump pump, and through windows. If it is in the water we can be exposed to it when taking a hot shower with the particles being released through the mist providing opportunity for exposure by inhalation. Also, some evidence of potential health effects from ingesting contaminated water, as well.
Radon is the number one environmental cause of cancer mortality
Smoking increases the effects of radon exposure.
Stigma associated with lung cancer contributes to lack of lung cancer awareness in general, but it is especially reflective of the lack of awareness of lung cancer caused by radon exposure. Radon is a lot like this transparent ribbon. The fact that you can’t see, smell, or taste radon, and also that any effects of exposure have such a long-latency period all contribute to the fact that people just forget that this is an issue.
In general there is not much attention given to radon but in some places there is an awareness for testing of homes during real estate transactions.
Our exploration has uncovered that while 9 states have some type of legislation about testing for radon in schools, only Florida has a comprehensive testing program.
To date the only program with comprehensive school testing data acquired from Florida. Therefore, started analysis to explore spatial relationship between the residential and school results. Anticipating can use the residential results to understand potential risk in nearby schools and apply this relationship to residential results in other States.
Collected additional soil characteristic data including permeability, slope, drainage, thickness of sediment, and uranium concentration from USGS.
Research also indicates a relationship with temperature gradient (inside home temperature and ambient temperature). Since we do not have data from homes we use the ambient temperature and precipitation.
Also, since housing characteristics play an important role and once again since we don’t have individual housing data we tried using age of housing and measure of poverty from the US Census as surrogates for individual housing characteristics.
Centrus geocoding, selected only records with AS0, 100% street address match, for locational accuracy.
Many locations have multiple test results per location and across time. This is because multiple tests within a house at initial testing to account for multiple living spaces and multiple floors. Could also be the result of multiple tests over time; especially pre- and post- mitigation. Also, situations where multiple unit housing. Selected the maximum test result per location making the assumption that the underlying environmental conditions are driving the radon levels in the area – regardless of air movement/housing conditions that may affect multiple readings across time.
Created buffers of varying distances uses ESRI ArcGIS 10.1 and calculated area proportion/weighted averages for the characteristics within these buffers.
3,755 school tests
335 (8.92%) greater than 4 piC/L
There is a statistically significant and strong association between the residential test results and school radon test results. Our understanding that things that are closer together geographically tend to be more alike, therefore, holds true in this preliminary analysis. These preliminary results suggest there are almost 3x increased odds/chance of a school located within a ¼ mile of a residential radon test result above the EPA action to have a radon level also greater than the EPA action level. Additionally, there is a 234% increased risk of a school testing >4 pCi/L if there are homes testing >4 pCi/L within the ¼ mile.
As you move further away the relationship still holds true; however, as the distance increases the strength of the association decreases.
Out to three miles there is still a statistically significant increased odds and risk.
This holds true for these Florida data. Additional analyses with school data in other states in necessary to test whether this relationship holds true. However, assuming it does, how can we use this information in states without school testing programs? We use residential testing data in North Carolina and Georgia to explore the application of these preliminary findings.
This is preliminary and we don’t believe this is the final model and we need to explore parameter interactions and perhaps take another look at the way the soil data are coded. However, there are interesting results.
When we remove residential test results the fit of the model does not change and the estimates are modified slightly. Since we have had difficulty acquire residential test results and since housing characteristics vary which we don’t have want to see if we can explore/predict in schools without the housing data.
