International Archives of Int Arch Occup Environ Health ( 1987) 59:123-131 itipt Acii 2 nd ;ll " tl Springer-Verlag 1987 Indoor exposure to radon from the ground and bronchial cancer in women Christer Svensson l, Gunnar Eklund l, and G 6 ran Pershagen 2 'Department of Cancer Epidemiology, Radiumhemmet, Karolinska Institute, Box 60500, S-10401 Stockholm 2 Department of Epidemiology, National Institute of Environmental Medicine, Box 60208, S-10401 Stockholm, Sweden Summary A case-referent study on the possible association between radon emanating from the ground and bronchial cancer was carried out on 292 female lung cancer cases and 584 matched population referents Both groups had lived for at least 30 years in the city of Stockholm, Sweden The cases were diagnosed during 1972 to 1980 with oat-cell and other types of anaplastic pulmonary carcinomas A sample of about 10 % of the dwellings where cases and referents had lived was selected for measurements of radon and radon daughters There was a relative risk of 2 2 (P = 0 01) for lung cancer associated with living in dwellings close to the ground in areas with an increased risk of radon emanation Smoking habits did not appear to be a major confounding factor for this association, although a detailed evaluation was not possible The measurements indicated increased radon daughter concentrations in ground level dwellings within radon risk areas where lung cancer cases had lived, suggesting that this exposure was of etiologic importance. Key words: Case-referent study Lung cancer Oat cell carcinoma Radon Introduction Several studies have shown an association between exposure to ionizing radiation emanating from the progeny of radon-222 (Rn), the radon daughters (Rn D), and bronchial cancer among miners l 1, 3, 10, 18 l In conjunction with data from animal experiments, this has led to a general acceptance of the association as causal. In recent years interest has focused upon radon as an etiologic factor for bronchial cancer also among the general population The radon levels in dwellings are generally much lower than in the mines where increased cancer risks have been observed This and other factors connected with radon exposure in Offprint requests to: Ch Svensson at the above address
124 C Svensson et al. mines, e g concomitant exposure to dust and diesel exhausts, give rise to uncertainties in the risk extrapolation from miners to the general population. Epidemiological studies from Sweden show a statistically significant association between bronchial cancer and estimated exposure to radon in dwellings l 4, 8, 16 l The studies are small, however, and the risk estimates are uncertain A US study on lung cancer incidence in Washington County, Maryland, and exposure to radon as estimated on the basis of basement construction and building material did not show any association l 19 l A contributing reason to the negative result could be the lower than average content of uranium, the progenitor of radium and radon, in the soil in this area. Rn in dwellings can emanate from the building material, tap water or from the ground In Sweden the ground is the most important Rn source in dwellings with the highest concentrations l 2 l The city of Stockholm is partly built on Rnemitting ground types, i e glacial eskers and granite l 11 l, which may lead to elevated indoor concentrations The aim of this study was to investigate whether an increased lung cancer risk could be detected among subjects who had lived in dwellings on these types of ground Only women were included in the study as they were less likely to be exposed to carcinogens in the work environment that could act as confounding factors l 17 l and they generally spent more time at home. Materials and methods The study was performed according to the case-referent (case-control) technique From the local cancer register in Stockholm the names and birth data were obtained for 391 women, who were diagnosed during the period from 1972 to 1980 with bronchial cancers belonging to the unspecified epithelial group This is a conglomerate of histological types where the small- ("oat-")cell anaplastic and large-cell cancers dominate The study was restricted to this group of cancers since earlier studies have suggested that an increased incidence of small cell anaplastic carcinomas can be detected at lower exposure levels than for other histological types l 1, 10 l The study group consists of those 292 women who had not lived outside the Stockholm city limits for more than two years during the 30 years prior to diagnosis Their mean and median age at time of diagnosis was 70 years with a range of 27 to 97 years Two hundred and fifty-two of the diagnoses were based on histological and 40 on cytological evidence. Two referents were drawn for each case from the population register in Stockholm This is a register of all present and past inhabitants of Stockholm from 1885 onwards It contains names, birth data and all addresses in Stockholm where a person has lived The referents were matched to the cases according to year of birth and were alive at the time of diagnosis of their respective cases They also met the same follow-up criteria. All addresses where cases and referents had lived for at least two years up to a date preceding the time of diagnosis for the respective case with five years were classified into either of two categories: Radon risk (+) or no radon risk (-) Two criteria had to be met for the positive classification: ( 1) The ground should be of a geological type which has been shown to emit radon in significant amounts The classification was done with the use of geological maps l 21 l. ( 2) The person should have lived close to ground level, i e in a detached house or on the bottom floor of a multi-family house. To get a validation of the geological classification, indoor measurements of Rn and Rn D were made for a sample of close to 10 % of the total number of addresses ( 1654) All the 45 addresses classified as radon positive and 110 of the negative addresses were selected for measurement The sample of negative addresses was obtained by randomly drawing a number of
Indoor exposure to radon and bronchial cancer in women 125 cases and one of their referents, and selecting all the addresses where they had lived during the time-period in question. In multi-family houses it was not possible to pinpoint the exact apartment in which a person had lived Seventy-seven measurements were made in apartments classified as geologically Rn negative, and in 65 of these the measurements were made on the floor where the individual had lived At 23 addresses, or 15 % of the total sample, no measurements were made since the present dwellers did not let the measurement crew in or because the buildings had been torn down or altered to such an extent that measurements were not considered meaningful. The measurements were made by the Stockholm Municipal Environment and Health Administration according to their standard procedure This is a filter method (grab-sampling), utilizing an EDR-RDA 200 measuring instrument l 20 l With this instrument both Rn and Rn D concentrations can be measured The values for the Rn D concentrations are expressed in Bq/m 3 EER 1 In order to obtain representative and reproducible results, measurements were made only at ambient temperatures in the range -20 to + 10 0 C and wind velocities below 6 m/s The dwellers were asked to restrict ventilation and vacuum-cleaning during the 12 h preceding measurements. Data on smoking habits of the study subjects were not obtained In order to get an estimate of smoking habits in the source population, data from a national survey in 1963 on smoking habits was utilized l 6 l The sample included 3208 women living in Stockholm In this group the percentage of women who were smokers in each parish was computed The percentages were standardized for age so that they could be compared to the present material. The results of the geological classification were treated statistically according to the method described by Miettinen l 12 l for a matched pair study with two referents per case, using testbased confidence limits l 13 l. The results from Rn and Rn D measurements were analyzed by use of the Mann-Whitney U-test Since the life-time residential addresses of an individual cannot be assumed to be statistically independent of one another, the results of the radiation measurements were evaluated with regard to persons rather than to dwellings All p-values were calculated with twotailed tests. To get an estimate of the dose-response relationship, a weighted logistic regression analysis of the results from the radiation measurements was performed with the aid of the computer program GLIM l 5 l Cumulative exposure to Rn D, i e Rn D-concentration x years of residency at the address, was used as dose variable The weighting was done by assigning a weight factor to each of the four categories cases/referents, exposed/not exposed to radon according to the geological classification The weights were calculated as the quotients between the total number of individuals in a category and the number of individuals in this category for whom measurements were made Confidence intervals were computed according to a method described by Woolf l 23 l. Results The Rn-risk areas according to the geological classification are shown in Fig 1. These areas primarily consists of an esker running from north to south through the center of Stockholm and of scattered areas of granite bedrock in the northern parts. In Table 1 the distribution of cases and referents with respect to the geological classification is shown The matching between cases and referents into trip- 'EER, the Equilibrium Equivalent Concentration of Radon, of a non equilibrium mixture of short-lived radon daughters in air is that acitivity concentration of radon in radioactive equilibrium with its short-lived daughters which has the same potential a-energy concentration as the actual non equilibrium mixture.
126 C Svensson et al. Fig 1 Map of the city of Stockholm Shaded areas represent ground types with an increased probability of radon emanation Table 1 Distribution of 292 triplets, each containing a female case of bronchial corcinoma and two matched referents, according to estimated exposure to radon from the ground in the homes of the study subjectsa Case Referents Both exposed One exposed Both unexposed and one unexposed Exposed 0 2 20 Unexposed 0 19 251 a An analysis with retained matching between cases and referents gives a relative risk of 2 2 for bronchial cancer associated with exposure to radon from the ground ( 95 % confidence interval: 1 2-4 0)
Indoor exposure to radon and bronchial cancer in women 127 Table 2 Radon daughter concentrations (Bq/m 3 EER) in dwellings of women with bronchial cancer and referents in relation to geological classification of exposure to radon from the ground Cases Referents Exposed Not exposed Exposed Not exposed (C+) (C-) (R+) (R-) No of individuals 15 28 13 29 Radon daughter concentration Arithmetic mean 55 0 24 8 38 7 33 4 Standard deviation 49 2 17 6 48 3 39 1 Geometric mean 39 3 19 7 23 0 21 9 Years of residency 15 8 18 2 16 4 16 7 (arithmetic mean) lets containing one case and two referents has been retained in the table. Twenty-two of the 292 cases ( 7 5 %) and 21 of the 584 referents ( 3 6 %) were exposed, corresponding to a relative risk (RR) of 2 2 and a 95 % confidence interval of 1 2 to 4 0 (X 2 = 6 45) The etiologic fraction, i e the proportion of cases attributable to the exposure, was 4 1 % or approximately 12 cases of bronchial cancer The analysis includes one case and two referents who had lived within Rn risk areas but for whom information on whether they had lived close to ground level was not available For these the case address was regarded as Rnnegative and the referent addresses as Rn-positive. The relative risk for living in a Rn-positive dwelling was the same in an unmatched analysis as in the matched and in the subsequent statistical calculations the matching was dissolved Seventy-three cases ( 25 %) and 115 referents ( 20 %) had lived in at least one dwelling close to ground level This difference in proportions is not statistically significant Among these subjects a greater proportion of the cases than of the referents had lived in radon risk areas The corresponding RR is 1 9, (P = 0 09) About 17 % of the cases and referents had at least one address without information as to the level The results were not altered even if those dwellings were all classified as close to ground. The results of the Rn D-measurements are presented in Table 2 Each individual is grouped into one of four categories; cases (C) and referents (R) with (+) or without (-) geological radon risk The mean Rn D-concentration is highest in the C+ category, but the difference is not statistically significant (P = 0 07 for C+ compared with the other categories combined) There is also a tendency for the R+ category to have higher concentrations than the R category. Table 2 includes 13 dwellings that were at least partly built of concrete made from alum shale, which has a high content of uranium and thereby emanates Rn For many years this type of concrete had widely been used as building material in Sweden, but production was stopped in 1975 l 22 l If dwellings containing aerated concrete are excluded from the analysis, the Rn D-concentrations in the C+ category are significantly increased compared with the concentrations in the other categories (P < 0 01).
128 C Svensson et al.., - Er Cumulative exposure of residency Fig 2 Relative risk (RR) of developing oat-cell cancer of the lung as a function of cumulative exposure to radon daughters The exposure is expressed in Bq/m 3 EER x years of residency, assuming an occupancy factor of 0 8, i e 7000 h/a The plotted curve represents a least square fit of RR to the US l 1 l and Czechoslovak l 10 l data, assuming a linear dose-response relationship Vertical bars indicate 95 % confidence limits In Fig 2 the results from the logistic regression analysis of the dose-response relationship is shown The cumulative exposure was separated into four groups ( 0-249, 250-499, 500-749 and > 750 (Bq/m 3 ) x years of residency), and the group with the lowest cumulative exposure was used as reference The arithmetic mean of the cumulative exposure in each group was used as exposure value in the figure The point estimates of the relative risks increase slightly with increasing cumulative exposure, but a test for linear trend does not show any statistical significance (P= 0 29) The confidence intervals are wide due to the limited sample size and none of the relative risks is significantly increased above 1.0. For comparison, Fig 2 also includes the results from two studies on oat cell carcinoma among US and Czechoslovak uranium miners l 1, 10 l A doseresponse curve fitted to the data in these studies is not inconsistent with the results from the present investigation. With regard to smoking, it was found that the age-adjusted smoking ratio among women in Stockholm as a whole was 25 % For the parishes with Rnpositive addresses where cases had lived the smoking ratio was 26 %. Discussion The study shows that a significantly larger proportion of the cases than of the referents had lived in dwellings which for geological reasons could be expected to have increased Rn-concentrations The results point in the same direction as
Indoor exposure to radon and bronchial cancer in women 129 some earlier studies l 4, 8 l A direct comparison of the risk estimates cannot be done, however, as the present study only included a histological subgroup of bronchial cancers. The results of the Rn and Rn D measurements are not as unequivocal The distribution of Rn D concentrations in the geologically Rn-positive dwellings where cases had lived is shifted towards higher values compared with the other categories, but the differences are not statistically significant. Radon from the building material may obscure differences in radon concentrations caused by emanation from the ground, and since our aim was to assess the lung cancer risk associated with radon from the ground it can be argued that the influence from other sources of radon should be minimized This may be achieved by excluding measurements made in dwellings built from material with a high content of uranium When such dwellings were excluded, the difference between C+ and the other categories was statistically significant. The Rn D-concentrations in geologically Rn-positive dwellings where referents had lived do not differ significantly from those in geologically Rn-negative dwellings, although there is some tendency in this direction The difference in Rn D-concentrations between the C+ and other categories could partly be explained by the fact that many dwellings classified as geologically radon positive do not have increased radon concentrations It has been shown that the Rnemanation from the ground can vary dramatically even within short distances l 14 l, and the building technique is also of importance for the amount of Rn that can reach the indoor atmosphere l 11, 15 l Therefore, one would expect the proportion of geologically positive dwellings having increased radon concentrations to be greater among the cases than among the referents, provided the exposure is of etiologic importance. Because of the limited number of measurements, the uncertainties of the point estimates of the relative risks at different cumulative exposures in the logistic regression analysis are very great, but there is a tendency for increased relative risks with increasing cumulative exposure (see Fig 2) Extrapolations from dose-response relationships with regard to oat-cell carcinomas in uranium miners indicate that the risks observed in the present study are somewhat greater than expected, although not inconsistent with the occupational data The uncertainties in the extrapolation and the comparison between the environments should be stressed It is not known in detail how differences in breathing patterns, particle content in the air and the fraction of unattached ions affect the risk per exposure unit in dwellings as compared to mines. To be a confounder in this study, smoking has to be correlated to living close to ground in radon risk areas The data from the national survey on smoking habits indicated that smoking among women in the relevant age groups was not more common in the parishes where Rn-positive cases had lived However, this is a crude estimate and one cannot entirely rule out the possibility of geographical differences in smoking habits within the parishes No information exists as to the relation between smoking habits and living close to ground It should be noted, however, that the point estimate of the relative risk for lung cancer associated with living in a radon positive dwelling is almost doubled, even if the comparison is restricted to ground level dwellings only Among other possible
130 C Svensson et al. confounders which could not be evaluated were general air pollution and occupational exposures, but it is unlikely that these factors were of major importance in the study group. The measurement program was primarily designed to evaluate the quality of the geological classification Regarding the validity of the Rn D-measurements as indicators of acquired radiation dose, it must be emphasized that the method used renders values of the concentrations at one point in time only The diurnal and seasonal variations in Rn-concentration are great l 7, 9 l, and these may affect both the accuracy and precision of the measurements For example all measurements were made during office hours, and since the Rn-concentration is highest during the night, it is possible that the values obtained underestimate the yearly average On the other hand, it is obvious that the pertinent exposures for the study subjects lie up to several decades before the time of measurements and the indoor radon levels were probably lower in the past l 22 l. We found an etiologic fraction of approximately 4 % for unspecified epithelial carcinoma of the lung that could be related to Rn emanating from the ground Thus, Rn from the ground does not seem to be a major health problem in the city of Stockholm One reason for this is that most people in the area live in apartments situated above the ground level On the other hand, an RR of 2 2 implies that Rn emanating from the ground may be of importance in some areas Furthermore, if the misclassification of exposure is unbiased, the relative risks will be underestimated. In conclusion, the present study suggests that exposure to radon daughters in dwellings is related to an increased risk of unspecified epithelial carcinoma of the lung The results are in accordance with some earlier findings Further studies are recommended to elucidate the role of indoor exposure to radon and radon daughters in the etiology of lung cancer among the general population, especially with regard to dose-response relationships and to the interaction with smoking habits. Acknowledgement This study was supported by a grant from King Gustav the V:s Jubilee Fund in Stockholm We thank John Carstensen for assistance in some of the data analyses and the Stockholm Municipal Environment and Health Administration for carrying out the radon measurement program. References 1 Archer VE, Saccomanno G, Jones JH ( 1974) Frequency of different histologic types of bronchogenic carcinoma as related to radiation exposure Cancer 34:2056-2060 2 Akerblom G, Andersson P, Clavensjo B ( 1983) Soil gas radon a source for indoor radon daughters Swedish Geological Report ID-nr IRAP 84020, 38 pp Swedish Geological, Lule A, Sweden 3 Axelson O, Sundell L ( 1978) Mining, lung cancer and smoking Scand J Work Environ Health 4:46-52 4 Axelson O, Edling C, Kling H ( 1979) Lung cancer and residency a case-referent study on the possible impact of exposure to radon and its daughters in dwellings Scand J Work Environ Health 5:10-15 5 Baker RJ, Nelder ( 1978) The GLIM System, Release 3 Numerical Algorithms Group, Oxford, England
Indoor exposure to radon and bronchial cancer in women 131 6 Cederlf R, Friberg L, Hrubec Z, Lorich U ( 1975) The relationship of smoking and some social covariables to mortality and cancer morbidity A ten year follow-up in a probability sample of 55,000 Swedish subjects aged 18-69 Report of the Department of Environmental Hygiene Vol 1 Karolinska Institute, Stockholm, Sweden 7 Downard TR, Geiger EL, Millard JB ( 1984) Field evaluation of Eberlines radon daughter working level monitor In: Proceedings of the 3rd International Conference on Indoor Air Quality and Climate, vol 2 pp 99-104 B Berglund, T Lindvall and J Sundell (ed). Swedish Council for Building Research, Stockholm, Sweden 8 Edling C, Kling H, Axelson O ( 1984) Radon in homes a possible cause of lung cancer. Scand J Work Environ Health 10: 25-34 9 Harley JH ( 1981) Radioactive emissions and radon NY Acad Med 57:883-895 10 Horacek J, Placek V, Sevc J ( 1977) Histologic types of bronchogenic cancer in relation to different conditions of radiation exposure Cancer 40:832-835 11 Mikitalo A, Wilson C ( 1982) Radon measurements in Swedish dwellings A summary of results from measurements by local authorities Swedish Geological Report ID-nr BRAP 82116, 38 pp Swedish Geological, Lule A, Sweden (in Swedish) 12 Miettinen O ( 1970) Estimation of relative risk from individually matched series Biometrics 26:75-86 13 Miettinen O ( 1976) Estimability and estimation in case-referent studies Am J Epidemiol 103: 226-35 14 The National Board of Physical Planning and Building ( 1981) Radiation in buildings, Report 54 from the National Board of Physical Planning and Building: 119 pp National Board of Physical Planning and Building, Stockholm, Sweden 15 Nazaroff WW, Nero AV ( 1984) Transport of radon from soil into residence In: Proceedings of the 3rd International Conference on Indoor Air Quality and Climate, vol 2, pp 15-20 B Berglund, T Lindvall and J Sundell (ed) Swedish Council for Building Research, Stockholm, Sweden 16 Pershagen G, Damber L, Falk R ( 1984) Exposure to radon in dwellings and lung cancer: a pilot study In: Proceedings of the 3rd International Conference on Indoor Air Quality and Climate, vol 2, pp 73-78 B Berglund, T Lindvall and J Sundell (ed) Swedish Council for Building Research, Stockholm, Sweden 17 Pershagen G, Hrubec Z, Svensson C ( 1986) Passive smoking and lung cancer in Swedish women Am J Epidemiol (in press) 18 Samet JM, Kutwirt DM, Waxweiler RJ, Key CR ( 1984) Uranium mining and lung cancer in Navajo men N Engl J Med 310:1481-1484 19 Simpson SG, Comstock GW ( 1983) Lung cancer and housing characteristics Arch Environ Health 38:248-251 20 Stockholm Municipal Environment and Health Administration ( 1984) Investigations of radon in Stockholm Report of the Stockholm Municipal Environment and Health Administration, Technical Department, 29 pp, Stockholm Municipal and Health Administration, Stockholm, Sweden (in Swedish) 21 Swedish Geological ( 1969) Geological maps over Stockholm SGU Ser AE nr 1-4, Swedish Geological, Uppsala, Sweden 22 The Swedish Radon Commission ( 1983) Radon in dwellings Report from the Radon Commission, Ministry of Agriculture, SOU 1983:6, 145 pp Liber, Stockholm, Sweden (in Swedish) 23 Woolf B ( 1955) On estimating the relationship between blood groups and disease Ann Human Genet 19:251-253 Received May 23 / Accepted September 23, 1986