Assessment of the Effectiveness of Radon Screening Programs in Reducing Lung Cancer Mortality
CHUQ—Centre de recherche du CHUL, Public Health Research Unit, Québec (Québec), Canada.
Abstract
The present study was aimed at assessing the health consequences of the presence of radon in Quebec homes and the possible impact of various screening programs on lung cancer mortality. Lung cancer risk due to this radioactive gas was estimated according to the cancer risk model developed by the Sixth Committee on Biological Effects of Ionizing Radiations. Objective data on residential radon exposure, population mobility, and tobacco use in the study population were integrated into a Monte-Carlo-type model. Participation rates to radon screening programs were estimated from published data. According to the model used, approximately 10% of deaths due to lung cancer are attributable to residential radon exposure on a yearly basis in Quebec. In the long term, the promotion of a universal screening program would prevent less than one death/year on a province-wide scale (0.8 case; IC 99%: –3.6 to 5.2 cases/year), for an overall reduction of 0.19% in radon-related mortality. Reductions in mortality due to radon by (1) the implementation of a targeted screening program in the region with the highest concentrations, (2) the promotion of screening on a local basis with financial support, or (3) the realization of systematic investigations in primary and secondary schools would increase to 1%, 14%, and 16.4%, respectively, in the each of the populations targeted by these scenarios. Other than the battle against tobacco use, radon screening in public buildings thus currently appears as the most promising screening policy for reducing radon-related lung cancer.
1. INTRODUCTION
Radon is a naturally occurring radioactive gas known to cause cancer in humans. It is classified as a Class A and Class 1 human carcinogen by the U.S. Environmental Protection Agency and the International Agency for Research on Cancer, respectively.(1,2) This status is based on the conclusions of several epidemiological studies conducted in mine workers confirming a higher incidence of lung cancer (LC) in individuals heavily exposed to radon.(3) In 1998, the members of the Sixth Committee on Biological Effects of Ionizing Radiations, the BEIR VI, concluded that the available data from epidemiological studies conducted in homes appeared to support a slight increase in LC risk compatible with extrapolations of the data from underground miners, but were insufficient at the time to accurately assess the magnitude of the risk caused by radon exposure in this context.(4) Since then, recent combined analyses of residential radon (RRn) studies in Europe and North America have provided direct evidence of an association between RRn and LC risk.(5,6)
A number of countries have implemented various screening programs aimed at reducing public exposure to radon.(7–11) These programs are based on lifetime risks that are one to several orders of magnitude higher than what would be considered acceptable for exposure to chemicals in the environment(12) as well as on the availability of techniques to reduce the concentrations of RRn.(13) Available information, however, reveals that persuading the population to carry out screening tests can represent an enormous challenge; the reasons evoked to justify the lack of intervention being their costs, skepticism regarding the risk associated with radon, and the difficulty in obtaining adequate information as to the remediation measures.(14) Moreover, a screening result showing elevated radon concentrations would not necessarily result in the implementation of mitigation measures, whether these measures were cost-free(15,16) or not.(17–19) In fact, there are no current studies enabling to assess whether programs implemented for controlling cancer risks associated with radon exposure have the capability to significantly reduce the incidence or mortality linked to LC. An assessment of the risk of dying from LC was thus conducted to evaluate the effectiveness of various screening scenarios in reducing the level of radon exposure in the population of the Province of Quebec, Canada.
2. MATERIALS AND METHODS
2.1. Assessment of Lung Cancer Risk from Exposure
The estimation of the risk of LC due to RRn exposure was performed according to the cancer risk model developed by the BEIR VI.(4) A Monte-Carlo-type model was used to integrate Quebec data on RRn exposure and tobacco use in adults as well as Canadian data on residential mobility and tobacco use in teenagers. Simulations were performed using the “@Risk ” software (Palisade Corp., Ithaca, NY, USA) from predetermined statistical distributions.(20)
Concentrations of radon in each home occupied by a person during lifetime were randomly selected from region-specific distributions obtained in the course of a province-wide Quebec survey conducted in 449 homes (i.e., bungalows, duplexes, row housing, and trailer houses, but excluding apartments) in regions other than those known for presenting high radon concentrations; i.e., regions encompassing over 99% of the Quebec population.(21) Concentrations measured on the ground floor were used. Where measurements were only carried out in the basement, ground floor concentration was estimated by dividing the basement concentration by 1.7; i.e., the ratio of values recorded in basements over those recorded on the ground floor for the entire province-wide study. The distribution of values obtained (arithmetic mean = 38.1 Bq m−3; standard deviation = 59.5 Bq m−3) was used to define exposure levels after moves outside the province.
The general social survey database prepared by Statistics Canada of Canadian citizens aged 15 or above in 1990 allowed to determine the number (mean: 14; minimum: 0; maximum: 43) and the probable distance of residential moves as a function of age (unpublished data). For individuals younger than 15 years of age, their residential mobility was considered to be similar to that of young adults of childbearing age. The location of the first residence as well as the destination after the move were chosen on a random basis amongst the entire sociodemographic regions of the province by allocating greater weight to more densely populated regions and by taking into consideration the distance and population size of targeted regions.
Life-table calculations were based on lung cancer mortality data according to smoking status among Caucasians in the United States.(4) The number of LC deaths due to RRn exposure was calculated from the BEIR VI exposure-age-concentration model with parameter estimates based on “updated” data.(4) For each year of life, the additional probability of dying from radon-related cancer was calculated by considering present and past exposure, gender, and tobacco-use status. The probabilities were then added to yield the additional risk of dying from LC over a life span of 80 years. This model was applied to various screening scenarios: (1) promotion of universal (province-wide) screening; (2) promotion of screening in the region with the highest concentrations; (3) promotion of screening on a local basis with financial assistance; and (4) systematic screening in primary and secondary schools. Primary prevention measures aimed at reducing the infiltration of radon during construction of the house, such as installing a protective barrier membrane against underground gases, were not considered.
The additional risk of dying from LC resulting from RRn exposure was calculated for nonsmokers, ex-smokers, and current smokers of both sexes by performing 10,000 simulations for each of the three groups. The number of deaths in each group was estimated by multiplying the calculated estimated risk by the number of individuals in that group. The number of individuals per tobacco-use categories in adults(22) and in the 12–19 year old age group(23) as well as demographic statistics(24) were from the most recent data available at the time of the study. For each population targeted by our scenarios, we calculated the annual number of deaths attributable to radon, the number of deaths prevented annually, and the reduction in radon-related mortality. The impact of each scenario on the overall provincial mortality by LC was also calculated.
2.2. Assessment of the Effectiveness of Interventions
2.2.1. Measurement and Effective Mitigation Conducted in All Quebec Houses
In order to serve as baseline, the maximal theoretical effectiveness was estimated by postulating that measurements were taken in all Quebec houses and that effective mitigation measures were implemented in every house where radon concentrations exceeded 150 Bq m−3, the lowest of all current guideline values.(1)
2.2.2. Promotion of Universal Screening
We estimated that 6% of homeowners would measure the concentration of radon on a province-wide scale on the basis of the results obtained from various surveys conducted in the United States.(15,25–27) The percentage of homeowners implementing mitigation measures was set at 12% for dwellings with concentrations between 150 Bq m−3 and 800 Bq m−3 and at 32% for dwellings with concentrations above 800 Bq m−3.(25) These values are comparable to those observed in a population in the Winnipeg region, Canada.(18) We assumed that mitigation measures would be implemented upon arrival of the occupants in their new home. Active soil depressurization was considered as the mitigation measure applied in 50% of cases, with less efficient corrective measures used in the remaining cases. This value was selected by taking into account the experience reported by householders who had implemented remedial measures in the United Kingdom.(28–30)
Concentrations after mitigation were projected by dividing the concentration initially measured in the house by a reduction factor chosen randomly from predetermined distributions. Truncated log-normal distributions were used for active soil depressurization (arithmetic mean: 14; standard deviation: 25; minimal–maximal values: 1–110) as well as for the other methods (arithmetic mean: 2.2; standard deviation: 2; minimal–maximal values: 1–24).(31) Concentrations following mitigation could not be inferior to 15 Bq m−3, a concentration found in outside air in certain U.S. areas.(32) It was assumed that once implemented, the effectiveness of remedial measures did not decrease with time.
2.2.3. Promotion of Screening in Regions Considered at Risk
The effectiveness of this scenario was evaluated herein by choosing the region showing the highest indoor radon concentrations (arithmetic mean on the ground floor = 71.9 Bq m−3) in the province-wide Quebec survey.(21) The probability of measuring RRn concentration was arbitrarily set at 12%, i.e., double the value assigned for the preceding scenario. It was assumed that more targeted information campaigns could result in greater participation levels. Likewise, the probability of implementation of mitigation measures was set at 24% (double the value assigned for the promotion of universal screening scenario) when concentrations were situated between 150 and 800 Bq m−3. Values assigned to the other variables of the model are the same as those allocated for the promotion of universal screening scenario.
2.2.4. Promotion of Screening on a Local Basis with Financial Support
The effectiveness of this scenario was assessed from data obtained in a sector of the municipality of Oka in Quebec, a locality with a high proportion of houses presenting elevated radon concentrations.(33) The average concentration of radon measured in dwellings therein was 781 Bq m−3 (arithmetic mean) with a standard deviation of 1,597 Bq m−3.(34) Following intensive information campaigns, 68% of homeowners in that sector measured the concentration of radon in their home; the proportion of those reporting mitigation measures were 18% for concentrations between 150 and 800 Bq m−3 and 53% for concentrations above 800 Bq m−3. These results were obtained after defraying the entire costs associated with radon testing as well as 75% of the first $ 5,000 (Canadian dollars) of eligible work carried out to significantly reduce radon levels in the home.(35)
2.2.5. Screening in Schools
In this scenario, radon concentrations in primary and secondary schools were estimated using the distribution of radon concentrations measured on the ground floor of residential dwellings in Quebec.(21) Exposure was calculated by taking into consideration that children between the ages of 5 and 18 years are exposed at a rate of 180 days/year and 8 hours/day according to an adapted scenario developed for estimating asbestos exposure in schools.(36) The prevalence of tobacco use in youths between 12 and 19 years old was established at 18.7%.(23) The number of cancers due to this exposure was calculated over a period of 80 years. Concentrations attained after mitigation were calculated based on the assumption that an active soil depressurization technique would be used in all buildings where concentrations would exceed 150 Bq m−3. In view of the lack of data for this specific age group, moving was not considered in this scenario.
3. RESULTS
According to the model used, 10.5% of deaths due to LC are attributable to residential exposure to radon in Quebec; that is, 430 of the 4,104 LC deaths in Quebec in 1998. Of those 430 deaths, 71 (16.5%) occur following exposure to concentrations equal to or greater than 150 Bq m−3 and could theoretically be prevented by exhaustive screening followed systematically by the implementation of mitigation measures. However, based on a realistic scenario for screening and mitigation, the implementation of a universal screening promotion program would allow to prevent on the long term less than one death/year on a province-wide scale (0.8 case; IC 99%: –3,6 to 5,2 cases/year), for an overall 0.19% reduction in radon-related mortality (Table I). This specific effectiveness would be increased to 1% by the promotion of a targeted screening campaign in high-risk regions while reaching 14% with a screening program on a local basis with financial support. In this latter scenario, the program would allow to prevent one death out of the seven radon-related LC deaths over a span of 20 years. As to the number of deaths from radon exposure in primary and secondary schools, it was estimated at 1,158 over a period of 80 years (including 112 in nonsmokers), i.e., a total of 14 per year. Radon screening and implementation of mitigation measures according to the scenario used would allow on the long term to reduce LC mortality due to radon exposure in these buildings by 2.36 per year, i.e., a reduction of 16% in radon-related mortality in the study population. It is estimated that 60% of radon-related deaths involve smokers, compared to 30% in former smokers and 10% in never smokers.
| Scenario | Targeted Populationb | Annual Number of Deaths Attributable to Radon | Number of Deaths Prevented Annually | Reduction in Radon-Related Mortality in the Targeted Population | Reduction in Lung Cancer Mortality in Quebecc |
|---|---|---|---|---|---|
| Status quo | 7,487,169 | 430 | – | – | – |
| Theoretical effectiveness | 7,487,169 | 430 | 71 | 16.5% | 1.7% |
| Promotion of universal screening | 7,487,169 | 430 | 0.8 | 0.19% | 0.02% |
| Promotion of screening in at-risk region | 96,929 | 9 | 0.09 | 1% | 0.002% |
| Promotion of screening on a local basis with financial support | 900 | 0.35 | 0.05 | 14.3% | 0.001% |
| Promotion of screening in schools | 1,183,077d | 14 | 2.36 | 16.4% | 0.06% |
- aWith an hypothetical guideline value of 150 Bq m−3.
- b Source: Quebec Institute of Statistics.
- cOf the 4,101 deaths from this type of cancer according to the Quebec death registry in 1998.
- dEventually applicable for the whole population.
4. DISCUSSION
The proportion of LC deaths attributable to radon in the Quebec population is roughly the same as that calculated for the American population (approximately 10%) after extrapolation of the dose-response relationship observed in underground miners to the concentration frequency distribution of radon measured in American dwellings.(4) This may appear somewhat surprising at first glance, considering the fact that radon exposure in Quebec is lower than levels observed in United States (arithmetic annual average radon concentration on the ground floor in single-family homes of 41.7 Bq m−3 in Quebec versus 48.5 Bq m−3 in the United States),(37) but can largely be explained by the higher prevalence of tobacco use for adults in the study model herein(22) compared to that used by the BEIR VI committee.(38)
Of the 430 LC deaths per year attributable to RRn in Quebec, 359 would occur following exposure to concentrations smaller than 150 Bq m−3. This can be explained by the nonthreshold dose-response relationship used to estimate LC risk and the fact that 97.5% of homes in Quebec have radon concentrations below 150 Bq m−3.(39) A number of 71 deaths would occur following exposure to concentrations equal to or greater than 150 Bq m−3, representing the maximal theoretically avoidable deaths by screening. In regard of the overall provincial mortality by LC (4,101 deaths in 1998), this public health strategy appears to have a very limited role in the fight against cancer in a population weakly exposed to this carcinogen yet comprising a high prevalence of smoking.
The promotion of universal screening would allow, based on published participation data, to prevent 0.8 deaths annually attributable to radon throughout the province. Although it is possible that the new Canadian guideline, recently updated to 200 Bq m−3,(40) or innovative approaches may increase public adherence, the result presented suggests that only marginal gains on mortality would result from any enhancement of this scenario.
It is precisely to maximize the effectiveness of interventions that many countries have chosen to act solely in zones considered most at risk. Assessments are based on measurements taken in homes or in public buildings, whether it be in random fashion or from geological mapping data.(41,42) According to our risk analysis, such a strategy would enable on the long term to reduce radon-related mortality in the present study population by 1%. It must be noted, however, that in the province-wide Quebec study, available geological indicators were only able to explain about 5% of the variations in radon concentrations found in residential dwellings.(39) Despite its apparent objectivity, this strategy is, nevertheless, not exempt of subjectivity. For example, in the United Kingdom, the threshold to determine at-risk zones is set at 1% of homes with concentrations above the baseline level of 200 Bq m−3 whereas in Finland, where mean exposure levels are higher, a zone is considered at risk when at least 10% of its dwellings exceed the baseline value of 400 Bq m−3.(43) Hence, the choice of levels considered to be of concern will always remain arbitrary. Moreover, identification of at-risk zones favors a sentiment of confidence and inertia in the population living outside these designated areas.(44)
Promotion of a screening program on a local basis with financial support would enable, on the long term, to reduce radon-related mortality by 14% in the present study population. Several countries have implemented aid programs in the areas most affected by radon.(45) On the basis of available data, however, it is difficult to assess to what extent citizens actually take advantage of available aid programs. It appears, however, that very few grants have been awarded,(16) the reduction of radon concentrations not being a priority for low-income families.(46) It is therefore possible that the probabilities selected in the present study for assessing this particular scenario likely represent maximal screening and mitigation values that can be obtained in such situations.
Screening and mitigation in schools would allow, on the long term, to reduce the mortality related to radon exposure in these buildings by 16%. Based on the analysis performed with regard to the control of radon exposure in schools, 188 deaths could be prevented over a span of 80 years if concentrations above 150 Bq m−3 were mitigated using the best technique available, representing an effectiveness three times superior to that estimated for the promotion of universal screening over the same time span. This screening option therefore represents the avenue with the most promise. Indeed, radon measurements were conducted systematically in schools in the United Kingdom,(11) Sweden,(7) Slovakia,(47) north-east Italy,(48) and in the Greek city of Kalamata,(49) as well as in a certain number of school buildings in Japan(50) and in British Columbia.(51)
It is important to keep in mind that the BEIR VI model used in this study to assess LC risks posed by indoor radon and cigarette smoking are subject to considerable uncertainty because of gaps in scientific knowledge of radon effects at low levels of exposure. Of the two risk models developed by the BEIR VI committee, the concentration-form of the model was preferred in the present assessment in order to be more conservative. Even if the duration-form model corresponds to the well-established inverse dose-rate effect for radon-induced LC,(52) the two models were equally preferred by the BEIR VI committee.(4)
The major shortcomings in the existing data relate to estimating LC risks near 148 Bq m−3 and below toward the average indoor level, especially the risks to never smokers.(4) Unfortunately, the two combined analyses of RRn studies in Europe and North America did not rule out all of the fundamental uncertainty relative to radon-associated elevation in LC risk at the lowest RRn exposures, particularly for nonsmokers. In fact, combined analysis of North American data sets contained no breakdown of odds ratios within specific RRn exposure groups.(53) In the combined analysis of European data sets, lower confidence bounds on estimated RRn-related adjusted relative risk within the seven categories of exposure considered exceeded one only for contiguous exposure categories greater than or equal to 400 Bq m−3 while the trend in estimated relative risk for nonsmokers in the first three exposure categories considered (<100 Bq m−3) was negative.(54) The combined analysis of European studies also indicates a magnitude of enhanced relative risk of radon-induced LC among smokers versus nonsmokers that is far greater than such a magnitude implied by the BEIR VI model.(5, 54)
Consequently, the number of LC deaths due to RRn exposure could have been overestimated in all of our scenarios. However, uncertainty about radon effects at low exposure levels does not bear the same importance on assessment of screening effectiveness; this strategy being oriented toward high exposure levels. Inversely, this may have introduced some underestimation of overall effectiveness considering that the objective of those who remediate may not be achieving the lowest possible concentrations if a threshold were to exist. Irrespective of the dose-response relationship at the lowest exposure levels, the BEIR VI committee's analysis showed that assumption of a threshold up to exposures of 148 Bq m−3 had little impact on the numbers of LC deaths theoretically preventable by mitigation of exposures above that level.(4) Of course, this does not imply that effectiveness of our realistic scenarios for screening and mitigation could not be overestimated. The number of deaths due to LC has been calculated by assuming a life-long exposure in individuals with a predetermined prevalence of tobacco use and born after the implementation of a remedial program. Consequently, the values put forth with respect to the reduction in LC deaths following the implementation of mitigation measures are the product of estimates for which results can be expected on the long term (several decades after implementation of the program) and basically represent the maximal expected values.
Concerning the scenarios related to dwellings, major concerns relate to mitigation techniques selected, notably their time of application, their effectiveness in a context where they would be implemented by the homeowners themselves rather than by professional contractors,(30,55) their applicability to the Quebec climate, and, finally, their durability. Moreover, studies suggest that those who screen or remediate are not representative of the population as a whole. The percentage of homes tested for radon rose with the respondents' increasing education levels (from 9% to 19%) or income.(56) People who remediate also appear to have attained higher educational levels, higher incomes, and perceive themselves as being healthier than the average person in the population. They are also more likely to be retired and not to have smoked in the house.(19,57) The health benefits accruing from radon screening or remediation is therefore lower than expected, largely because the occupants are older, have a smaller family size, and include fewer smokers than the population average.(58)
Such overestimation of remediation effectiveness is probably lesser for the screening-in-schools scenario. In cases of elevated concentrations, it would be difficult for public building managers not to proceed with applying mitigation measures, even in the absence of legal obligations. Mitigation actions performed in schools with mean radon concentration levels greater than 200 Bq m−3 have indeed demonstrated that it is possible to reduce these levels by 95% through active soil depressurization.(51)
It should be reminded that in the absence of exhaustive data in Quebec schools, it is the distribution of radon concentrations measured on the ground floor of residential dwellings in this province that was used as the measure of exposure, which could overestimated the risks associated with these environments. Indeed, concentrations in large public buildings are generally inferior to those measured in single-family dwellings. In a study measuring 375 schools, average radon levels in schools were lower than in the surrounding homes.(51) This likely reflects the larger room sizes or greater ventilation of classrooms.(59) Moreover, the average concentration of radon in public places during working hours is generally lower than the total average obtained during the full period of exposure, the ratio being 0.85 for schools.(60) Results from other studies indicate that remediation may preferentially reduce the high levels of radon at night. As a result, the reduction in dose, and therefore the improved health benefit, will be less than expected than the reduction in average radon levels for schools.(61)
It is also obvious that the cancer risk model used in this study is mainly based on case-control studies that did not include childhood exposure. Children may be more susceptible than adults to cancer induced by radiation for physiological reasons (rapidly dividing cells, higher breathing rates, lungs less efficient in removing foreign particles).(62) The BEIR VI nevertheless considers that there was not a clear indication of the effect of age at exposure.(4) A recent study on the risk of cancer occurring up to the age of 14 found an inverse association between radon exposure and risk of any cancers.(63) It was, however, pointed out that only radon measurements in the home at time of diagnosis or selection of control were used.(64) Globally, it is difficult to know to what extent this could represent a source of overestimation or underestimation in regard of this scenario.
It should be noted that only primary and secondary schools were taken into account by this scenario. Indeed, other public spaces could also be included, like kindergartens. In 2003, approximately 36.9% of Quebec children aged 0 to 5 years attended a day care center. In addition, almost half of these children (45.7%) were cared for in a family setting where they could have had access to home basements.(65) The number of cancer cases that could be avoided by an eventual screening program of public surroundings could thus be even higher than what is estimated herein. In fact, over half of European countries have defined a reference level for radon in the workplace and public buildings.(8) In most countries, these reference levels have a legal value.(66)
Theoretically, a universal preventive measure to reduce radon exposure applied during home construction could allow for a greater reduction in mortality because the entire population as a whole, even people exposed under 150 Bq m−3, would benefit from reduced exposure levels. However, such radon-resistant construction techniques may be unnecessary in low-risk zones if a threshold were to exist at low concentrations. Moreover, there is still an ongoing need to define truly effective anti-radon measures installed at the time of construction.(67)
Because most of the predicted radon-induced deaths due to LC occur in smokers or former smokers, the most effective method to reduce radon-related LC mortality remains to reduce smoking habits, particularly if the radon-related risk at low doses is confined to smokers. In our study, we assessed that an intervention program enabling to reduce tobacco use by only 1% would enable to prevent 30 LC deaths attributable to tobacco use and 6.2 radon-related LC deaths annually (calculated from model-estimated risk levels and by reducing the number of smokers by 1%, data not shown). Consequently, stopping smoking is by far the most effective (and cost-effective) method to reduce LC both relative to smoking-related and radon-related LC. Smoking reduction would also prevent other diseases such as heart diseases, neurological diseases, diabetes, etc., which often occur at younger ages than LC.
5. CONCLUSION
According to the model used, the promotion of universal screening in dwellings would reduce, over the long-term, radon-related mortality and total lung cancer mortality by only 0.19% and 0.02%, respectively, in the population of Quebec. Consequently, the role of this public health action in the fight against cancer appears to be very limited in a relatively mobile population weakly exposed to this carcinogen yet comprising a high prevalence of smoking. This does not necessarily mean that there should not be any control of radon levels in houses, only that there should be a discussion as to what is reasonable and cost effective. In the present assessment, financial support programs offered on a very localized basis would only be able to prevent a limited number of cases on a population scale. The number of deaths prevented annually by screening in schools could be up to three times higher than that expected from the promotion of universal screening of all Quebec dwellings. Therefore, aside from prevention and cessation of smoking, screening in public buildings appears to be the most promising alternative to prevent radon-induced LC in Quebec; part of the effectiveness of mitigation measures for new construction being based on the assumption of a nonthreshold relationship between radon exposure and LC risk.
