ARTICLE IN PRESS

Journal of Environmental Management 81 (2006) 19–26

www.elsevier.com/locate/jenvman

Cost–benefit analyses of radon mitigation projects

Mads Lyngby Petersen, Thommy Larsen
Environmental Assessment Institute, Gammel Kongevej 5, 1 floor, Copenhagen V., Denmark
Received 13 May 2004; received in revised form 7 October 2005; accepted 11 October 2005
Available online 17 April 2006

Abstract

Radon is a radioactive gas that enters buildings and is known to cause lung cancer. Danish building regulation recommends simple
remedial measures for radon levels between 200 and 400 Bq m
3, and more effective methods for levels above 400 Bq m
3. The Danish
National Board of Health found that radon levels in 4.6% of the dwellings were above the reference level, and that the distribution of
radon levels in Danish one-family houses is lognormal with a geometric mean of 57.7 Bq m
3 and a geometric standard deviation of 2.28.
The purpose of the paper is to present a cost–benefit analysis of a possible future governmental intervention against radon in existing
Danish one-family houses. In other words, it seeks to answer the question: is it socio-economically worthwhile to avert lung cancer
deaths by reducing radon levels in the most exposed dwellings? The results indicate that an intervention based on the anti-radon measures
recommended by the Danish building regulation would, if implemented, avert less than 30 deaths each year. By comparison, a total of
300 people die from radon exposure annually in Denmark. The total costs exceed the valuated health benefits by h62 million (6%
discount rate). It is concluded that it is not socio-economically favourable to reduce radon levels in existing Danish one-family houses.
These results are based on a discount rate of 6%, a value of preventing a statistical fatality of h1.4 million, and a relatively high cost of
remediation in Denmark compared to other countries. Based on the sensitivity analysis, the conclusion will be different if a discount rate
of 4.3% or less is used, or if the value of preventing a statistical fatality is h2.54 million or higher, or if the cost of remediation is reduced.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Radon; Lung cancer; Cost–benefit analysis

1. Introduction be roughly identical to the relationship found in mines

(Lubin and Boice, 1997). Several factors complicate the
Radon is a natural radioactive gas that is formed in the causality, though, such as differences in sensitivity linked
ground from radium and uranium. The gas enters buildings to gender, age and smoking status. Among these, smoking,
from the ground and rises upwards through cracks in the combined with radon exposure, is believed to be the most
construction. When radon mixes with indoor air, part of it potent. For a comprehensive review of current knowledge
will be inhaled. Radon is not stable and decays into other on exposure and risk factor, consult the BEIR VI report
gaseous compounds, whereby it exposes the respiratory (National Research Council, 1999).
passages to a radiation. The connection between radon The statistical properties of domestic radon levels result
exposure and lung cancer is widely accepted, but the exact in lognormal distributions of number of dwellings as a
risk of developing lung cancer from domestic radon levels function of radon exposure as a global characteristic,
is very uncertain. A few studies have investigated the health though mean values and standard deviations vary from
effects of radon in indoor environments. The documenta- country to country (Miles, 1998). A Danish survey found
tion on miners, who are continuously exposed to high the radon levels in one-family houses to be roughly
radon levels, is far more comprehensive (Lubin et al., lognormal with a geometric mean of 57.7 Bq m
3 and a
1995). Some studies suggest the linearity between radon geometric standard deviation of 2.28 (Andersen et al.,
level and the cancer risk from domestic levels of radon to 2001). It is estimated that the exposure to radon in Danish
dwellings causes 300 [100–900] deaths from lung cancer
Corresponding author. Tel./fax: +45 72265812. each year in Denmark (Statens Institut for Strålehygiejne
E-mail address: [email protected] (M.L. Petersen). and Risø Helsefysik, 1987). This estimate was based on a

0301-4797/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2005.10.005
 ARTICLE IN PRESS
20 M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26

simple calculation involving the assumption that all Danes Table 1

are exposed to 50 Bq m
3 from birth to death and a lifetime Effects and costs of the three methods selected (at 2003 prices)
relative risk of 8  10
5 (Bq m
3)
1. Method Reduction effect (%)a Costs (h)b
The present work is the first Danish cost–benefit analysis
of domestic radon mitigation. Several studies in other Active soil de-pressurization 80 6200
countries have investigated either the cost effectiveness or Passive soil de-pressurization 50 3100
Natural ventilation/vents to 20 400c
cost benefit of radon mitigation programmes. Among these
increase air change
are Snihs (1992) (Sweden), Colgan and Gutiérrez (1995)
a
(Spain), Kennedy and Gray (2001), Coskeran et al. (2002), The reduction effects are typical values from several studies (Henschel,
Denman et al. (2000), Denman et al. (2002), Kennedy et al. 1988, 1994; Swedjemark and Mäkitalo, 1990; Hoving and Arvela, 1993;
Andersen et al., 1997; US Environmental Protection Agency, 2003).
(2002) (UK), Ford et al. (1999), Lin et al. (1999) (USA) and b
Costs are from Andersen et al. (1997).
Stigum et al. (2003) (Norway). These studies use a c
The price is for five vents.
cost–efficiency or cost-analysis approach, with Colgan
and Gutiérrez (1995) as the only exception. The latter
presented a cost–benefit analysis, although benefits were experiments and the more extensive foreign knowledge.
not calculated directly from individual preferences. Costs are from Andersen et al. (1997), and include
Encouraged by the lack of cost–benefit analyses and the investments in building materials and skilled craftsmen to
differing approaches in existing cost–efficiency analyses, put up remedial measures. Costs play an important role
the present analysis aims at a detailed presentation of the and should be validated for the specific country where a
approach used for calculation of social benefits and costs of cost–benefit analysis is made. Further, the sensitivity of
radon remediation programmes. The approach is illu- the results to cost should be studied as part of a
strated by analysing a future Danish remediation policy. cost–benefit analysis. The costs in this paper are valid for
Danish legislation only contains a recommendation for Danish circumstances. A sensitivity analysis is presented in
householders to reduce radon concentrations above the Section 5.
action level. Reducing the radon concentration is left The list of methods mentioned in Table 1 is not
entirely to the householder’s own initiative. It might be complete. Other methods may apply to new buildings, or
feasible for society to establish a radon policy if radon to building methods not used in Denmark.
remediation is socio-economically worthwhile. A combined
information campaign and subsidy programme is one 2.2. Scenario
means of encouraging householders to remediate their
homes. Another option is to introduce compulsory tests Danish building regulation distinguishes between dwell-
followed by remediation, if required, in connection with ings, built before 1998, and ‘‘new’’ buildings, built after
sale of properties. Both options are expected to yield a slow 1998. The latter are required to have a cover between the
remediation rate as a result. The costs include tests for construction and the ground to prevent gases, including
radon levels, investments to reduce radon in existing one- radon, from entering the building. Hence, radon levels are
family houses and ongoing maintenance/operating costs. presumably low in new buildings, and are not considered in
The benefits comprise the quantified cost of illness and the present analysis. Buildings from before 1998 are not
averted death, based on willingness to pay (WTP). covered by the regulation, but it recommends simple and
less costly remedial measures to deal with radon levels just
2. Methods above 200 Bq m
3. For levels above 400 Bq m
3, it outlines
more effective, but expensive, measures. The recommenda-
2.1. Radon-reducing technologies tion forms a natural basis for the adopted scenario.
Suppose that a single-family house is tested to determine
The experience with radon-reducing measures in Den- its radon level. If the level exceeds the recommended
mark is very limited. Only one study has estimated the threshold value of 200 Bq m
3, some method must be
effects and costs of establishing interventions to reduce applied to reduce the level to below the threshold value.
radon in existing dwellings (Andersen et al., 1997). So, a Houses with levels above 400 Bq m
3 are provided with
detailed analysis of potential cost effectiveness requires active soil de-pressurization. Passive soil de-pressurization
foreign knowledge on technical issues, which is provided by is used for levels between 250 and 400 Bq m
3. Vents are
mitigation surveys carried out in the USA, Finland and used for levels between 200 and 250 Bq m
3.
Sweden (Henschel, 1988, 1994; Swedjemark and Mäkitalo, The choice of method and level intervals is made with
1990; Hoving and Arvela, 1993; US Environmental respect to the efficiencies from Table 1 to make sure that no
Protection Agency, 2003). Three methods were chosen for radon level exceeds 200 Bq m
1 after the intervention. As
their costs, efficiencies and suitability to be applied to yet, it is not technologically possible to reduce levels over
existing dwellings (see Table 1). In each case, the 1000 Bq m
1 to below 200 Bq m
1, as the impact of the
magnitude of the reduction effect is chosen as a typical most effective method is only 80%. However, the number
value from the composite experience of the few Danish of Danish dwellings containing radon levels above
 ARTICLE IN PRESS
M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26 21

1000 Bq m
1 is too small to have any significance to the linearly from 25 years after investments stop, to reach
analysis. zero after a further 50 years. This assumption does not
Radon mappings are the first tool by which to determine significantly influence the result, as both benefits and costs
which dwellings are at risk. The present analysis used an appear in a distant future and are discounted to present
Internet-based program at Risoe National Laboratory for values.
this purpose. The advantage is that it is much cheaper than
testing all dwellings, and is a good first-level indicator of
2.3. The magnitude of the intervention
potential danger. Should a radon-mapped dwelling indicate
a concentration higher than the median level of 60 Bq m
1
Radon levels in one-family houses may roughly be
in Denmark, it is assumed necessary to perform a test to
described by a lognormal distribution, with a geometric
establish the precise level. Once this is known, the
mean of 57.7 Bq m
3 and a geometric standard deviation of
appropriate remedial measure will be implemented im-
2.28 (Andersen et al., 2001). Using this distribution as the
mediately.
basis of calculation, it is found that 2.8% of all dwellings
The scenario includes investments and effects within a
have levels between 200 and 250 Bq m
3, a further 2.8%
delimited time frame. It assumes that all radon tests and
have levels between 250 and 400 Bq m
3, and 0.9% have
interventions will be implemented gradually during the first
levels above 400 Bq m
3. The total number of dwellings
30 years. It also assumes that dwellings have a lifetime
built before 1998 is assessed at 1.4 million, according to
longer than 100 years, and that all anti-radon investments
Statbank Denmark (Danmarks Statistik, 2003b).
will stop abruptly after 100 years.
The analysis also assumes that the health effects of
reducing radon do not appear at once. It considers the time 2.4. Reduced health effects
from exposure to outbreak—or the latency period—of lung
cancer to be 25 years, after Nyberg et al. (2000), who found The risk of contracting lung cancer decreases as the
the latency period to be between 20 and 30 years. The radon level is reduced. The most widely accepted models
gradual implementation of interventions, combined with for risk calculation of radon-induced lung cancer are the
the latency period, will delay the full impact on health BEIR VI models (National Research Council, 1999). The
effects until 55 years after the first investments. Even after excess risk, i.e., the risk of lung cancer caused by radon
anti-radon investments stop, full positive health effects will exposure, is a function of exposure levels and duration of
be felt for a further 25 years—corresponding to the latency exposures in previous and present homes, time since
period. Hence, the positive full health impacts of the exposure and age of occupants. Using the model for the
intervention will be felt for 70 years (between 55 and 125 present purpose involves some questionable assumptions
years after the first investments, see Fig. 1). for these values. A less detailed model would be more
Some long-term health effects may be expected even suitable, as the exercise does not require the flexibility of
beyond 70 years, because people have spent time in the BEIR models. The BEIR VI committee also found that
dwellings with reduced radon levels. The exact long-term a constant relative risk model results in comparable
effect is unknown, and so it is assumed that benefits decline attributable risks (the proportion of deaths attributable
to radon exposure). A simple linear relation between the
relative risk RRr and radon level r is, therefore, used
instead, as suggested earlier, e.g., by Lubin and Boice
(1997). It may be expressed formally as
RRr ¼ 1 þ ar.
The value of the slope a adopted here is
0.0015 (Bq m
3)
1 (Stigum et al., 2003). The calculation
of a was based on a meta analysis of eight home studies
(Lubin and Boice, 1997). The meta analysis found a
relative risk RR equal to 1.14 at 150 Bq m
3. Provided
that the relation between RR and radon level is linear
Fig. 1. Time horizon of radon tests, implementation of radon-reducing
as described above, a is estimated to be (1.14–1)/
measures, maintenance of measures and reduced health effects. The most
important features are labelled with numbers. (1) All tests and remedial 150 ¼ 0.001 (Bq m
3)
1. A study conducted by Lagarde
measures are performed gradually for 30 years. (2) The health effects et al. (1997) found that exposure measurement errors bias
follow the investments but displaced by 25 years (the latency period). The the risk estimates towards lower values by a factor 1.5. This
full health effect of the intervention is reached after 55 years. (3) All is supported by Darby et al. (1997). The corrected value for
operations, including reinvestments, maintenance and power supplies, are
a is, therefore, 0.0015 (Bq m
3)
1.
discontinued after 100 years. (4) The full health effect continues for a
further 25 years, where it starts to drop-off. The health effects of radon are Smoking status is not included explicitly, as the linear
assumed to be back at pre-intervention level 175 years after the start of the relation is fitted to data without distinguishing between
remedial programme. smokers and non-smokers.
 ARTICLE IN PRESS
22 M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26

other types of dwellings. It is assumed that the relative risk

of dying from lung cancer in a block of flats resembles the
risk in one-family houses. From Statbank Denmark, it is
known that 67.1% of the Danish population live in one-
family houses (Danmarks Statistik, 1998). Thus, the value
for O is equal to 67.1% of 3471. When these values are
entered into Eq. (1), it is found that a remedial programme
against radon reduced the annual incidence of lung cancer
deaths by 29.
BEIR VI calculated the attributable risks of their
preferred models. To make a comparison, it is necessary
to use the arithmetic mean of the radon level experienced
by the total population—not just for those who live in one-
family houses. For Denmark, this number is 59 Bq m
3
(Andersen et al., 2001). Reducing the mean by 59 Bq m
3
and now using the full 3471 annual lung cancer deaths for
O in the equation above gives 307 deaths attributable to
Fig. 2. Radon distribution after the remediation programme. The original radon in accordance with earlier results (Statens Institut
distribution is cut off at 200 Bq m
3. The new distribution below for Strålehygiejne and Risø Helsefysik, 1987). The attri-
200 Bq m
3 is a sum of the old distribution and the contributions butable risk is then calculated as 307=3471 ¼ 0:09 com-
previously above the cut-off. Note that the lower curve, corresponding to pared to 0.10–0.15 calculated for the BEIR VI models but
levels previously above 400 Bq m
3, extends beyond the cut-off. This is due
for US radon levels and death statistics. Though not
to the insufficient 80% reduction efficiency of active soil de-pressurization.
directly comparable, it indicates that the procedure
adopted here does not yield results that differ significantly
The distribution of radon in Danish one-family houses is
from the BEIR VI models. The number of deaths
approximately lognormal with an arithmetic mean of
attributable to radon includes deaths among the part of
81.0 Bq m
3. The interventions against levels higher than
the population exposed to less than 200 Bq m
3. The
200 Bq m
3 lower the mean of the distribution. Using the
relatively small number of avoided deaths (29) calculated
efficiencies listed in Table 1, all intervals between 200 and
above demonstrates that most deaths are ascribed to levels
250 Bq m
3 are decreased by 20%, intervals between 250
below the action level.
and 400 Bq m
3 are decreased by 50% and all intervals
from 400 Bq m
3 and upwards are decreased by 80%. This
3. Cost–benefit analysis
is illustrated graphically in Fig. 2.
After a final weighting, the new mean is found to be
In a cost–benefit analysis, all consequences of a project
72.8 Bq m
3 corresponding to a 10% decrease. The modest
need to be identified—both market and non-market costs
decrease is due to the 200 Bq m
3 action level. Homes with
and benefits. The economic consequences of the costs
levels below this value (comprising roughly 95% of all one-
include radon testing, investments to reduce radon,
family houses) are not mitigated.
reinvestments (due to a limited expectation of life for the
Stigum et al. (2003) found a relation between the change
investment) and operational expenditure. The benefits
in mean value and fewer deaths from lung cancer (their
comprise avoided illness and deaths.
Eq. (A8)):
A cost–benefit analysis of non-market societal impacts is
L ¼ Oaðr̄
r̄0 Þ, (1) based on individual preferences; i.e., the value that persons
accord to a particular cost or benefit (Freeman, 1993). This
where L is the number of reduced deaths, O the total value reflects the social opportunity costs of resources. It is
number of lung cancer deaths per year, and r̄ and r̄0 are the assumed that consumers’ optimization of their consump-
mean radon levels before and after the intervention, tion bundles is based on market prices. For a marginal
respectively. The equation is valid for small risks. Stigum change in consumption, it is possible to use observable
et al. (2003) showed its usability at Norwegian residential market prices as the value of goods. For goods not tradable
radon levels. As Danish levels are generally lower than on a market—such as fatality—a value can be calculated,
Norwegian levels, it makes sense to use the equation in based on WTP to reduce the risk of death.
Denmark and other countries with similar radon levels. The term ‘‘value of preventing a statistical fatality’’ is
The value of O is calculated on the basis of Statbank common in economic theory for this type of assessment,
Denmark (Danmarks Statistik, 2003c), taking a mean of confirming that it is possible to quantify and value a
the lung cancer mortality data for the years 1990–1999. The reduced risk of death (Johansson, 1995). Much controversy
number of deaths is 3471. However, since the analysis only is connected to valuing statistical fatality (Dolan, 2001).
considers one-family houses, the total number of lung A workshop held by EU DG Environment put
cancer deaths must be corrected to exclude deaths in all forward a value of between h0.9 million and h3.5 million,
 ARTICLE IN PRESS
M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26 23

and recommended the use of h1.4 million as the best Test kits to measure radon can be bought on the Internet
estimate for a statistical fatality (European Commission, for roughly h60. Tests are done gradually, over 30 years, in
2001). This value is also recommended by WHO (1999). dwellings suspected to be at risk, reaching a total cost of
Thus, h1.4 million is used as a basic value for a statistical h20 million.
fatality throughout this analysis. Reinvestments have to take place since a limited lifetime
Strictly speaking, it is necessary to perform a few of the equipment used to control radon is assumed. In this
corrections to this basic value. As age increases, VPF analysis, it is assumed that active and passive soil
decreases. This has been established theoretically, and is de-pressurization has a lifetime of 30 years, while the air
confirmed empirically by the EU workshop mentioned change vents last for 15 years. Furthermore, active soil
above (European Commission, 2001). According to Stat- de-pressurization requires an electric fan, which has an
bank Denmark, the average age of Danes who die of lung assumed lifetime of 10 years.
cancer is approximately 70 years (Danmarks Statistik, In the case of active soil de-pressurization, operating
2003a). Using the relationship between age and WTP expenditures for the electric fan must be included. The
established by the (WHO, 1999, their Fig. 3-3) reduces the welfare cost is estimated at h0.05 kWh
1, calculated as the
basic value for preventing a fatality by 20%. social cost of electricity produced by a power plant driven
Another adjustment is required based on the difference by coal (Møller et al., 2000). According to Andersen et al.
in Danish purchasing power relative to the rest of the EU. (1997), the electric fans use approximately 600 kWh a year.
Danish purchasing power is higher than that of other The total calculated costs are shown in Table 2. The total
European countries’. The Danish Ministry of Transport cost is dominated by the investments in radon-reducing
has calculated the correction factor to be 122. Hence, the measures.
basic value of h1.4 million increases by 22% (Trafikminis- Total benefits are comprised primarily of the fatalities
teriet, 2003). Thus, the effects of age and purchasing power avoided. The total number of averted deaths within the
approximately counterbalance each other for Danish time horizon of 175 years is 3174.
circumstances, leading to the use of h1.4 million in the As is clear from Table 3, the benefits are dominated by
analysis. The corrections have to be made for other the value of preventing a statistical fatality.
countries. A comparison between the costs and benefits of the
It is assumed that all lung cancer deaths occur after a project, as presented in Table 4, reveals the socio-economic
preceding course of illness. The value of a statistical fatality potential.
does not include the value of illness, so this is valued The net present value is negative. Thus, investments to
separately. The present analysis uses a sum of hospital reduce radon levels in existing dwellings turn out to be a
admissions and loss of production, because no suitable negative ‘‘bargain’’ for Danish society.
WTP surveys exist. A study focused on social costs of
smoking in Denmark conducted by Reindahl (1998)
estimated the sum of these two figures to be h47,785 per Table 2
case of lung cancer. Taken together, each averted case of Calculated costs of the radon remediation programme
lung cancer due to radon remediation corresponds to a Tests to measure the radon level (million h) 20
benefit of h1.4 million+h47,785. Investment and reinvestment (million h) 118
All costs and benefits that occurred before 2003 are Operating expenditures (million h) 3
inflated to the base year, 2003, using the consumer price Total (million h) 141
index. To compare today’s costs with those of coming
years, future costs and benefits are discounted to present
value. Discounting means that consumption in the future is
weighted lower than at the present. There is considerable Table 3
discussion, though, on what the ideal discount rate should Estimated benefits of the radon remediation programme
be. The Danish Ministry of Finance advices a rate of 6% Value of preventing statistical fatalities (million h) 76
(Finansministeriet, 1999), and this has been used in the Cost of illness (million h) 3
present analysis. A project is considered to be socially
Sum (million h) 79
advantageous if the discounted benefits exceed the dis-
counted costs.

4. Results
Table 4
Net present value of the radon remediation programme
The project is assessed in terms of the net present value
of costs and benefits. The results of the cost analysis are Costs (million h) 141
presented first, followed by the results of the benefit Benefits (million h) 79
Net present value (million h)
62
calculation. Finally, a comparison is made between the
Benefit/cost ratio 0.56
total costs and benefits.
 ARTICLE IN PRESS
24 M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26

5. Sensitivity analysis and discussion regarded as typical for US price level, only. To analyse the
sensitivity of the analysis to costs, it is nonetheless useful to
The calculations presented in this paper are based on an adopt the US installation costs. The costs given in Table 1
action level of 200 Bq m
3, and include one-family houses are, therefore, replaced by h1500 and h1200 for active and
only. It is found that 29 lung cancer deaths will be averted passive soil de-pressurization, respectively. The cost of
annually. This should be compared to the total annual natural ventilation is maintained unchanged because it is
number of lung cancer deaths attributed to radon in similar to the costs stated by the US Environmental
Danish homes, namely 300, and the number of all-cause Protection Agency (2003). The result is a positive net
lung cancer deaths which is 3471 year
1 on average. All present value of h12 million. The outcome of the analysis
essential costs and benefits were discounted by 6% and is, therefore, highly sensitive to a lowering of the cost level.
summed up, within the time horizon of 175 years from the Similar analyses performed for other countries may yield
first radon test. The result is a negative present value of different results. Several of the parameters involved depend
h62 million. on national circumstances, e.g., purchasing power, dis-
It must be noted that the analysis contains some count rate, cost of remedial measures and their efficiencies.
continuing points of uncertainty. A sensitivity analysis The latter is influenced by differences in construction of
was, therefore, performed for the most uncertain para- buildings, hence national costs and efficiencies should be
meters. The uncertainty of the risk factor is adopted to be used whenever possible. In addition to the parameters
780% (Stigum et al., 2003). The sensitivity of the net mentioned above, the net present value is also sensitive to
present values to the risk factor is found to be [h125 mil- the assumed time horizon. Most of the studies mentioned
lion, h2 million]. The high end of the uncertainty interval in the introduction use a time horizon of 40 years. Yet, the
results in a positive, though insignificant, net present value. time horizon in the present analysis is significantly longer.
This leaves the overall conclusion unchanged, namely that This is because it opted to extend the intervention period to
a nationwide radon mitigation project is economically 30 years and the latency period to 25 years, to make
unfavourable. scenarios as realistic as possible. Lengthening the time
The level of the discount rate for use in cost–benefit horizon ensures that all essential costs and benefits are
analyses is widely debated. The choice of discount rate included.
plays a significant role in projects with very long time The action level is set to 200 Bq m
3 to match the Danish
horizons. In this analysis, 6% is used as a basic value in recommendations. This accords with the action levels
accordance with most studies. The sensitivity analysis set by Norway, Sweden and some European countries.
shows that the discount rate should be less than 4.3% to The US uses an action level of 148 Bq m
3, while several
provide a positive net present value. Several countries EU countries employ 400 Bq m
3 (Åkerblom, 1999). It is
suggest a discount rate at this level. important to note that the action level is not chosen based
The value of preventing a statistical fatality is also on an expectation of a sudden increase in risk above the
uncertain. The European Commission (2001) recommends level. On the contrary, the risk is expected to increase
a value of h1.4 million as the best estimate for a basic value. linearly, and only slightly, at each subsequent radon level.
However, an ‘‘upper’’ value of h3.5 million and a ‘‘lower’’ The action level is set on the basis of the number of
value of h0.9 million are provided for use in sensitivity dwellings with radon concentrations above it and on the
analyses. The present analysis finds that the value of a extent to which consequent health effects are avoidable. A
statistical fatality has to be higher than h2.54 million to lower action level would brand many dwellings as being
give a positive net present value. This value is within the potentially at risk, which could be politically undesirable.
interval given by the European Commission. In a cost–benefit context, a lower action level implies high
An important element in the analysis is the costs. When benefits and also higher costs. The present work did
comparing the costs in Table 1 with costs given elsewhere, not intend to find the optimum benefit/cost ratio as a
e.g., US Environmental Protection Agency (2003), it is function of action level, whereas Stigum et al. (2003) found
clear that they are quite high. There may be several reasons 125 Bq m
3 to be the optimum action level in Norway.
for this, and it is difficult to separate out the most Smoking status is not included explicitly in the study. As
important factor. First of all, it must be noted that the in Stigum et al. (2003), it is implicitly assumed that the
costs used here are based on the very limited Danish lung cancer risk for smokers exposed to radon is a multiple
experience with radon mitigation. A large-scale remedia- of both risks. The assumption brings some further
tion programme could reduce costs, as competition uncertainty into the analysis. The linear relation between
between contractors becomes increasingly important. High excess risk and radon level was fitted to data without
wages and a correspondingly high price level in Denmark distinguishing between smokers and non-smokers (Lubin
are other plausible explanations. According to the US and Boice, 1997). The uncertainty of the slope (risk factor)
Environmental Protection Agency (2003), the typical range is, therefore, partly due to smoking. However, this is
of installation costs when using contractors is between accounted for by a sensitivity analysis of the risk factor,
$800–2500 for active soil de-pressurization and $550–2,250 presented below. The number of smokers affects the risk
for passive soil de-pressurization. These figures may be factor: if the number of smokers decreases, e.g., as a result
 ARTICLE IN PRESS
M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26 25

of anti-smoking campaigns, so does the risk factor Andersen, C., Ulbak, K., Damkjær, A., Graversen, P., 2001. Radon i
(National Research Council, 1999). This may overestimate danske boliger. Sundhedsstyrelsen, Statens Institut for Strålehygiejne,
the health effects and consequently the net present value. GEUS.
Colgan, P.A., Gutiérrez, J., 1995. Justification and optimisation in the
The probability of radon exposure is assumed to be choice of reference levels for radon in existing Spanish dwellings.
independent of smoking status. The connection between Journal of Radiological Protection 15 (4), 289–301.
smoking and radon exposure still is a matter of debate. Coskeran, T., Denman, A.R., Philips, P.S., Gillmore, G.K., 2002. A
Studies on miners suggest something between multiplica- critical comparison of the cost-effectiveness of domestic radon
tive and additive, but closer to the former as a multi- remediation programmes in three counties of England. Journal of
Environmental Radioactivity 62, 129–144.
plicative relation fit the data much better (National Danmarks Statistik, 1998. Statistisk Årbog 1998. Danmarks Statistik,
Research Council, 1999). Copenhagen.
Danmarks Statistik, 2003a. Statistikbanken, Befolkning og Valg, Fødsler
og dødsfald. Danmarks Statistik, Copenhagen. Available at: /http://
www.statistikbanken.dkS (Accessed 5 November 2003).
6. Conclusion
Danmarks Statistik, 2003b. Statistikbanken, Byggeri og boligforhold.
Danmarks Statistik, Copenhagen. Available at: /http://www.statis-
Roughly 300 persons in Denmark die from lung cancer tikbanken.dkS (Accessed 5 November 2003).
due to radon exposure each year. The Danish legislation Danmarks Statistik, 2003c. Statistikbanken, Døde efter amt, dødsårsag,
recommends simple and cheaper remedial measures for alder og køn. Danmarks Statistik, Copenhagen. Available at: /http://
www.statistikbanken.dkS (Accessed 5 November 2003).
houses with radon levels just above 200 Bq m
3 and more
Darby S.C., Whitley E., Reeves G.K., 1997. The importance of accounting
effective, and also more expensive, measures for levels for uncertainties in the determination of radon concentrations in
above 400 Bq m
3. The present cost–benefit analysis of studies of the carcinogenic effects of residential radon exposure. In:
domestic radon mitigation in existing houses is the first Health Effect of Low Dose Radiation. BNES, London, pp 96–99.
attempt to evaluate the economic consequences of a Denman, A.R., Philips, P.S., Tornberg, R., 2000. A comparison of the
costs and benefits of radon remediation programmes in new and
nationwide radon programme in Denmark. The methods
existing houses in Northamptonshire. Journal of Environmental
and techniques used in the present article may inspire other Management 59, 21–30.
countries to evaluate or reinvestigate policies for reducing Denman, A.R., Philips, P.S., Tornberg, R., 2002. The costs and benefits of
radon. radon remediation programmes in existing homes: case study of action
It is estimated that less than 30 of the 300 deaths level selection. Journal of Environmental Radioactivity 62, 17–27.
Dolan, P., 2001. Output measures and valuation in health. In:
are avoidable by means of anti-radon remediation. The
Drummond, M., McGuire, A. (Eds.), Economic Evaluation in Health
avoidable fraction of radon-induced lung cancer deaths is Care, Merging Theory with Practice. Oxford University Press, Oxford,
quite low, mainly because most deaths are due to exposures pp. 46–67.
below the threshold level. The cost–benefit analysis showed European Commission, 2001. Recommended Interim Values for the Value
that reducing the radon level in existing Danish dwellings of Preventing a Fatality in DG Environment Cost Benefit Analysis.
to below 200 Bq m
3 is not socio-economically favourable. European Commission. Available at: /http://europa.eu.int/comm/
environment/enveco/others/recommended_interim_values.pdfS (Ac-
The benefits in terms of avoidable deaths are not high cessed 5 November 2003).
enough to outweigh the considerable costs. Finansministeriet, 1999. Vejledning i udarbejdelse af samfundsøkono-
The results presented in this paper are connected with miske konsekvensvurderinger. Finansministeriet.
some uncertainty. The sensitivity analyses showed that Ford, E.S., Kelly, A.E., Teutsch, S.M., Thacker, S.B., Garbe, P.L., 1999.
the conclusion is sensitive to the discount rate, the Radon and lungcancer: a cost-effectiveness analysis. American Journal
of Public Health 89 (3), 351–357.
cost of remediation measures and the value of preventing Freeman, A., 1993. The measurement of environmental and resource
a statistical fatality. The conclusion is less sensitive to the values. Theory and Methods: Resources for the Future. Washington,
uncertainty of the risk value. DC.
Even though a Danish radon mitigation project is not Henschel, D.B., 1988. Radon reduction techniques for detached houses.
socio-economically worthwhile, the opposite may be true Technical Guidance, US Environmental Protection Agency.
Henschel, D.B., 1994. Analysis of radon mitigation techniques used in
for individual householders, as their risk of contracting existing US houses. Radiation Protection Dosimetry 56 (1–4), 21–27.
lung cancer may be decreased considerably. Therefore, the Hoving, P., Arvela, H., 1993. Effectiveness of ventilation improvements as
authorities should inform householders about tests, risks a protective measure against radon. Proceedings of the Indoor Air 4,
and remediation measures. 615–620.
Johansson, P.O., 1995. Evaluating Health Risks. An Economic Approach.
Cambridge University Press, Cambridge.
Kennedy, C., Gray, A.M., 2001. Cost effectiveness of radon remediation
References programmes. The Science of the Total Environment 272, 9–15.
Kennedy, C., Gray, A.M., Denman, A.R., Philips, P.S., 2002. The cost-
Åkerblom, G., 1999. Radon legislation and national guidelines SSI. effectiveness of residential radon remediation programmes: assump-
Report No. 99. Swedish Radiation Protection Institute. tions about benefits stream profiles over time. Journal of Environ-
Andersen, C., Bergsøe, N., Brendstrup, J., Damkjær, A., Graversen, P., mental Radioactivity 59, 19–28.
Ulbak, K., 1997. Radon-95: en undersøgelse af metoder til reduktion Lagarde, F., Pershagen, G., Akerblom, G., Axelson, O., Baverstam, U.,
af radonkoncentrationen i danske enfamiliehuse. Forskningscenter Damber, L., Enflo, A., Svartengren, M., 1997. Residential radon and
Risø. Prepared for Bygge- og boligstyrelsen og Statens Institut for lung cancer in Sweden: risk analysis accounting for random error in
Strålehygiejne. the exposure assessment. Health Physics 72, 269–276.
 ARTICLE IN PRESS
26 M.L. Petersen, T. Larsen / Journal of Environmental Management 81 (2006) 19–26

Lin, C.Y., Gelman, A., Price, P.N., Krantz, D.H., 1999. Analysis of Nyberg, F., Gustavsson, P., Järup, L., Bellander, T., Berglind, N.,
local decisions using hierarchical modeling, applied to home Jacobsson, R., Pershagen, G., 2000. Urban air pollution and lung
radon measurement and remediation. Statistical Science 14 (3), cancer in Stockholm. Epidemiology 11 (5), 487–495.
305–337. Reindahl, S., 1998. De samfundsøkonomiske omkostninger ved tobaks-
Lubin, J.H., Boice Jr., J.D., 1997. Lung cancer risk from residential radon: rygning. Master of Public Health, Aarhus University.
meta-analysis of eight epidemiologic studies. Journal of the National Snihs, J.O., 1992. Swedish radon programme. Radiation Protection
Cancer Institute 89 (1), 49–57. Dosimetry 42 (3), 177–184.
Lubin, J.H., Boice, J.D.J., Edling, C., Hornung, R.W., Howe, G.R., Statens Institut for Strålehygiejne, Risø Helsefysik, 1987. Naturlig stråling
Kunz, E., Kusiak, R.A., Morrison, H.I., Radford, E.P., Samet, J.M., i danske boliger. Sundhedsstyrelsen.
Tirmache, M., Woodward, A., Yao, S.X., Pierce, D.A., 1995. Lung Stigum, H., Strand, T., Magnus, P., 2003. Should radon be reduced in
cancer in radon-exposed miners and estimation of risk from indoor homes? A cost-effect analysis. Health Physics 84 (2), 227–235.
exposure. Journal of the National Cancer Institute 87, 817–827. Swedjemark, G.A., Mäkitalo, A., 1990. Recent Swedish experiences in
Miles, J., 1998. Mapping radon-prone areas by lognormal modeling of 222Rn control. Health Physics 58 (4), 453–460.
house radon data. Health Physics 74 (3), 370–378. Trafikministeriet, 2003. Partikelredegørelse. Trafikministeriet.
Møller, F., Andersen, P., Grau, P., Huusom, H., Madsen, T., Nielsen, J., US Environmental Protection Agency, 2003. Consumer’s Guide to Radon
Strandmark, L., 2000. Samfundsøkonomiske vurderinger af miljøpro- Reduction. 402-K-03-002, Revised February 2003. /http://www.epa.gov/
jekter: Danmarks Miljøundersøgelser; Miljøstyrelsen; Skov- og Nat- radon/images/consguid.pdfS.
urstyrelsen. Available at: /http://www.dmu.dkS. WHO, 1999. Health costs due to road traffic-related air pollution. An
National Research Council, 1999. Health Effects of Exposure to Radon impact assessment project of Austria, France and Switzerland.
(BEIR VI). National Academy Press, Washington, DC Available at: Economic Evaluation, Technical Report on Economy. World Health
/http://www.nap.eduS. Organization.