1. 26th National Radon Training Conference
September 18–21, 2016
San Diego, CA

2. Risks from Radon in Homes (and Costs for Its Control)
What is new since BEIR VI? David Pawel, Ph.D. (EPA/RPD) Richard Krop, Ph.D. (CADMUS)

3. DISCLAIMER
The numbers given in this presentation are preliminary and do not reflect official new EPA estimates of risk for radon in homes.

4. “The more that things change,
the more they pretty much stay the same.”

5. “Current” Estimates of Risk from Radon in Homes
Radon in homes: “second-leading cause of lung cancer.”
Based on 1995 population data, EPA estimated that about 21K die each year from radon-induced lung cancer
Average cost per life saved (EPA 1992): $1.2 million (2016 $)
Current EPA estimates of radon risk (2003) are based on NAS (1999) BEIR VI models
Derived from data from 11 cohorts of underground miners
Risks from radon depend on baseline lung cancer rates
Excess risk from radon >> for smokers than non-smokers
Models applied to mortality data
What has happened since?

6. Ten to Twenty years later

7. Are Efforts to Control Radon Cost Effective?
Several more recent publications called into question EPA’s approach to controlling radon and its effects
Lantz et al. (2013) Radon, smoking and lung cancer: The need to refocus radon control policy
Also Mendez (2009) on impact of declining smoking on radon-related cancer
Ford et al. (1999) Radon and lung cancer : A cost-effectiveness analysis
About $5 million (2016$) per life saved !!!!
Based on BEIR IV (NAS 1988)
Used 4% discount rate for both benefits and costs

8. New Information on Risk from Radon
New underground miner study results
East German miners (WISMUT cohort)
Other study updates, e.g., Czech, French, Colorado Plateau
Direct information from pooled analyses of residential case-control studies
Thirteen European studies (Darby et al 2006)
Seven North American studies (Krewski et al 200?)
Iowa, NJ, Winnipeg, Missouri (2), Iowa, Conn., Utah/Idaho
Changes in Baseline Rates
Smoking prevalence has decreased

9. What Else Might Have Changed or is New?
Costs for mitigating homes?
How homes are tested ?
Average radon levels ?
Effectiveness of Mitigations ?
Housing patterns and demographics
Pennsylvania “Hot Spot” homes

10. Questions to Be Addressed
What effect might all this new information have on projections of risk for radon in homes?
What impact might it have on comparisons of estimated benefits vs. costs?
What are the most important gaps in knowledge?
Will NOT present new official EPA estimates of risks for radon in homes

11. EPA and BEIR VI Radon Risk Models and Projections

12. Pooled Underground Miner Data (11 Cohorts)
Number of Lung Cancers
113 (Non-exposed), 2674 (Exposed)
Cohorts with largest number of lung cancer cases
China (980), Czech (705), Colorado (336), Ontario (291), Newfoundland (118)
Mean exposure
164 WLM (miners)
vs. 14 WLM (lifetime residential)
Cohorts with largest number of cancers among miners exposed to < 50 WLM
Ontario (180), China (77), Radium Hill (52)
Mean exposure rate and duration
2.9 WL, 5.7 y
Cohorts with Smoking Data
China, Colorado, Newfoundland, 3 others

13. EPA Estimates of Risk for Radon in Homes for the U.S. Population
EPA formulas for calculating risk for the U.S. population:
5.4 × 10-4 lung cancer deaths per WLM
9.7 × 10-4 (WLM)-1 (ES); 1.7 × 10-4 (WLM)-1 (NS)
Average exposure of 0.18 WLM per year 1.25 pCi/L)
Life expectancy of about 75 y
Lifetime risk = 0.7% = 5.4 × 10-4 × 0.18 × 75 (All)
1.3% = 9.7 × 10-4 × 0.18 × 74 (ES)
0.2% = 1.7 × 10-4 × 0.18 × 76 (NS)
About 13% attributable to radon
Of 157.3K lung cancers (2010), 21K attributable to radon

14. Radon Risk Projections Using Updated (2010) Mortality Statistics
Increases in lung cancer death rates from radon exposure is roughly proportional to the baseline rates.
New information on all cause mortality, lung cancer rates, smoking prevalence
Cancer Intervention & Surveillance Modelling Network (CISNET)
Applied current EPA (BEIR VI) radon risk model to updated (2010) mortality data
An estimated 51% (males) & 37% (females) will (or have) smoked. Previous estimates: 53% (M), 41% (F)

15. Change in Mortality Statistics Relatively Minor Impact on “BEIR VI” Model Projections
General Population
Ever Smokers
Never Smokers
Risk per 104 WLM
4.2
5.4
7.8
9.7
1.4
1.7
Lifetime Risk
@ 4 pCi/L
1.9%
2.3%
3.4%
4.1%
0.6%
0.7%
Etiologic Fraction
12%
13%
11%
20%
26%
# Attributable to Radon per year
18K (2010)
21K (1995)
18K (1995)
3K (1995)
90% UI for 1995
8K-45K

16. How Many Deaths are Avoided from Mitigations in the U.S.?
Proportional to avg. radon level reduction from mitigations
Depends on radon levels in target populations
Depends on the testing “protocol”
Effectiveness of mitigations
Assume radon levels can be reduced to 2 pCi/L
Alternative assumptions could be made
“Our” estimate of average reduction = 3.3 pCi/L
Out of 100K homes that are targeted (267K people):
6.3K mitigations; 16.7K people affected; (IF) mitigations last 25y
Deaths Avoided (one estimate!) = 82
= 16.7K × (3.3 × 25y × WLM/y) × 4.2×10-4 WLM-1

17. BUT When Would the Deaths Have Occurred?

18. Deaths Avoided Due to Mitigations (3
Deaths Avoided Due to Mitigations (3.3 pCi/L reduction for 25 or 75y; million people)

19. How Should Benefits Be Counted?
Lung cancers avoided would otherwise occur 5 to >100 years after radon levels are reduced
EPA/NCEE recommends discounting benefits (at 3%).
Decisions on discounting have a huge impact on benefit/cost results.

20. Discounting of Deaths Avoided
When economic analysis are conducted using discounting, a death avoided that would have occurred today counts more than a death avoided that would occur later.
At a 3% discount rate: the number of avoided deaths that would otherwise occur 30y from now are divided by the factor = =2.4 (or multiplied by about 41%).
Suppose the “value of a statistical life” is $8.8 million.
Value of a life “saved” 30 y from now = 0.41*$8.8 M = $3.6 M
Value of a life “saved” 50 y from now = 0.23*$8.8 M = $2.0 M
Value of life “saved” 50 y from = 0.14*$8.8 M = $1.2M

21. Deaths Avoided After Discounting (100K Homes Targeted)
Discount Rate
Effective for 25 y
Effective for 70 y
None 82
231 1% 54
123 3% 26 44 4% 19 29 5% 14 20

22. Projections Based on Alternative Radon Risk Models
New underground miner study results
East German miners (WISMUT cohort)
Other study updates, e.g., Czech, French, Colorado Plateau
Direct information from pooled analyses of residential case-control studies
Thirteen European studies (Darby et al 2006)
RR models: increase in lung cancer death rates from radon exposure is roughly proportional to baseline rates
Applied radon risk models derived for these alternate sources to updated (2010) mortality data

23. Underground Miner Studies WISMUT (Bismuth)
East German Wismut uranium mines supplied the Soviet Union atomic industry from Uranium from occupied Germany > amount from Soviet Union (Walsh et al. 2014)
The largest study of health risks associated with exposures to underground miners
About 59K members in cohort
About 3K lung cancer cases: About equal to the total in the pooled analysis of underground miners for BEIR VI .
An estimated 1.4K attributable to radon (Walsh et al. 2010)
Large size allows for estimates of how risk depends on factors such as age, exposure rate, time-since-exposure

24. WISMUT (Limitations) Incomplete smoking information
Nested case-control study (Schnelzer et al 2009)
About 700 cases; 1.4K controls
High proportion of smokers: Cases (95%); Controls (75%)
Estimates of excess risk for exposure to radon did not depend greatly on whether smoking is taken into account.
Other exposures: diesel fumes, asbestos
Ascertainment of cause of death < 60% before 1970
Estimates based on pre-1970 vs. post-1970 data are similar
Most miner exposure rates >> exposure rates in homes

25. Pooled Analysis of European Residential Case-Control Studies
Darby et al (2006) reported results from a joint analysis of 13 European residential case-control studies
Provided a “direct” estimate of risk for radon in homes and “firm evidence that residential radon acts as a cause of lung cancer in home.” (p = )
Large, influential, impressive scientific achievement
A total of 7K persons with lung cancer (cases) and 14K controls.

26. Pooled Analysis of European Residential Case-Control Studies (II)
Studies carefully chosen according to the criteria
Clear rules for selection of cases
Controls are representative of population that lung cancer cases had been drawn
Detailed residential histories for at least 15 y
Careful (long-term) measurement of radon gas concentrations
Data on smoking and other pertinent variables
Each had at least 150 cases and 150 controls
A complement to underground miner studies
Less precision to examine how risk depends on age and time-since-exposure

27. Projected Number of Deaths Avoided (100K homes tested; 6
Projected Number of Deaths Avoided (100K homes tested; 6.3K mitigations)
Length of Mitigation
Discount Rate
“BEIR VI”
WISMUT
(cohort)
Europe pooled
resident’l
New Official EPA
25 y 0% 82 45 46 ? 3% 26 15 18
70 y
231
125
128 44 24 31

28. Benefits of Mitigation
Deaths Avoided
About $8.8M value of life saved
Medical Care for Lung Cancer Patients
About $120 K per lung cancer
Other avoided costs not included
Societal costs such as from lost employment

29. Mitigation Costs
Installation of Active Soil Depressurization (ASD): $1460
Effective either 25 or 70 y.
Maintenance Costs: $166 per year
Operating Costs
Fan Replacement Every 10 y: $383
Warning 37 y: $121
Testing Costs: $106
Assumed 80% through real estate transactions
Only 20% of positive non-real estate tests result in retests
Discount Rate: 3%

30. Benefit-Cost Analysis Mitigations Effective 25 y; 100K homes targeted
BEIR VI
WISMUT
Europe Residential
Lives Saved
82 (26)
45 (15)
46 (18)
Costs (discounted)
$44 M
Installation
$9.2 M
Operating
$21 M
Testing
$14 M
Cost per Life Saved
$530 K
$980 K
$960 K
Cost per Life Saved (both discounted)
$1.7 M
$3.0 M
$2.4 M
I used a 25 year mitigation system life – I know this is conservative. I welcome your feedback. On the next slide I calculated it at 75 years.

31. Benefit-Cost Analysis Mitigations Effective 70 y; 100K homes targeted
BEIR VI
WISMUT
Europe Residential
Lives Saved
231 (44)
125 (24)
128 (31)
Costs (discounted)
$59 M
Installation
$9.2 M
Operating
$37 M
Testing
$14 M
Cost per Life Saved
$260 K
$470 K
$460 K
Cost per Life Saved (both discounted)
$1.4 M
$2.4 M
$1.9 M

32. Benefit-Cost Ratio Mitigations
Discount Benefits?
Mitigation Duration (y)
“BEIR VI”
WISMUT
Europe Residential No 25
16.8
9.1
9.3 70
34.8
18.9
19.3
Yes (3%)
5.4
3.0
3.8
6.6
3.7
4.7

33. Radon Resistant New Construction
Difficult to analyze benefits and costs without more information/data
How much are radon levels reduced through passive depressurization?
Where are the RRNCs, e.g., how many in Zone 1?
What proportion of RRNCs are tested?
How many that test positive are fitted with fans?

34. Information Wish List
Areas where better information is needed/desired:
RRNC !!!
How many mitigations/RRNC are done each year?
Fan data: how many are replacements vs. for new mitigations
Testing
How many are done through real estate transactions?
What is actually done, e.g., testing devices, placement within homes?
False-positive/negative rates?
How effective are mitigations in reducing radon levels?
How long are mitigations effective?
What are the lung cancer risks from exposure to radon in homes?

35. Summary
EPA’s current radon-in-home risk projections are based on risk models adapted from BEIR VI and 1990 mortality data
Changes (from 1990 to 2010) in mortality statistics have a relatively small impact on radon risk projections.
A “BEIR VI” model estimate suggests about 20K radon-induced lung cancer deaths in 2010
“New” information from residential case-control studies provides “direct” evidence that radon in homes causes lung cancer
Risk projections based on models derived from recently published analyses, e.g., from data from WISMUT and the residential case-control studies tend to be within a factor of about 2 of BEIR VI model projections
Decisions on discounting benefits can have a huge impact on results.
Preliminary calculations indicate that benefit-cost ratios for mitigation may be about 3 or larger (even with discounting)

36. Acknowledgments
We wish to thank Jani Palmer, Phil Jalbert, Dave Rowson, Bill Long and Angana Roy for their helpful comments, input and assistance.