Correspondence
Radon Exposure and Lung Cancer
N Engl J Med 1994; 330:1684-1685June 9, 1994
- Article
To the Editor:
Pershagen et al. (Jan. 20 issue)1 were thorough in their measurements of radon exposure, but the incidence of smoking that they reported may have been seriously underestimated. If so, the relative risk of lung cancer associated with radon exposure would have been lower than they reported. The source of their underestimation of the incidence of smoking is the likely systematic underreporting of smoking when incidence is assessed by methods similar to those used in the Pershagen study. Smokers in the United States report only two thirds to three quarters of their actual use of tobacco2. Another study showed that light smokers were more likely to underreport tobacco use (many were actually moderate-to-heavy smokers)3. Most important, exsmokers being followed for a disease known to be linked to smoking provided the most inaccurate reporting4; self-reported exsmokers represent 8 percent of all subjects (108 of 1360) in the Pershagen study. This fact, together with the likely underreporting of smoking in general, may have introduced a serious systematic bias in the authors' calculations.
4 ReferencesWolffe Nadoolman, M.B.A.
Yale University School of Medicine, New Haven, CT 065041
Pershagen G ,Akerblom G ,Axelson O , et al. Residential radon exposure and lung cancer in Sweden. N Engl J Med 1994;330:159-164
Full Text | Web of Science | Medline2
Hatziandreu EJ ,Pierce JP ,Fiore MC ,Grise V ,Novotny TE ,Davis RM . The reliability of self-reported cigarette consumption in the United States. Am J Public Health 1989;79:1020-1023
CrossRef | Web of Science | Medline3
Perez-Stable EJ ,Marin BV ,Marin G ,Brody DJ ,Benowitz NL . Apparent underreporting of cigarette consumption among Mexican American smokers. Am J Public Health 1990;80:1057-1061
CrossRef | Web of Science | Medline4
Slattery ML ,Hunt SC ,French TK ,Ford MH ,Williams RR . Validity of cigarette smoking habits in three epidemiologic studies in Utah. Prev Med 1989;18:11-19
CrossRef | Web of Science | Medline
To the Editor:
Pershagen et al. have reported that exposure to radon is an important etiologic factor in lung cancer in the general Swedish population. We want to add data about a similar phenomenon.
We have observed an unexpectedly high rate of lung cancer in a small village in western Tirol, Austria (Umhausen, which has about 2600 inhabitants). This seems to be due to increased exposure to indoor radon in an area without mining activity. The source of radon is an 8700-year-old rock slide of granite gneiss, the largest of the alpine crystalline rocks. The rock slide has a strong emanating power, because its rocks are heavily fractured and have a slightly increased uranium content. Radon levels in the cellars of some houses in Umhausen were 500 times higher (200,000 Bq per cubic meter) than those recommended by the Commission of the European Community for existing buildings (400 Bq per cubic meter).
Umhausen is divided by a river into an eastern part (1000 inhabitants), where the much higher levels of radon were found, and a western part (1600 inhabitants) with average radon levels. Within the past 10 years, 29 people have died of lung cancer in Umhausen: 24 spent their lives in the eastern part of the village, whereas only 5 lived in the western part. These numbers point to a correlation between radon levels and lung cancer in homes.
With the average annual exposure to radon radiation in Europe -- for example, 20 Bq per cubic meter in the United Kingdom -- the associated mortality from lung cancer is 3.5 per 1000 inhabitants. Mortality from lung cancer in eastern Umhausen was 24 per 1000 persons during the observation period of 10 years. These findings led the Austrian government to support structural improvements in houses, schools, and kindergartens in Umhausen in order to reduce radioactivity.
Fritz Hoppichler, M.D.
Monika Lechleitner, M.D.
University of Innsbruck, Innsbruck 6020, AustriaTo the Editor:
In the study by Pershagen et al., a 100 percent sample of women and a 40 percent sample of men were merged, but there was no separate analysis for women. Women are likely to spend more time at home (and thus be exposed to residential radon longer) and to have less exposure to other confounders (e.g., occupational carcinogens) than men. According to an earlier report of the same data,1 the risk of lung cancer, if any, from radon exposure among women was less than the risk of lung cancer from radon exposure among men. There was no significant increase in the risk among women as radon levels increased.1
The authors present data in Table 5 showing that the relative risk of lung cancer does not increase with increased radon exposure in subjects who sleep near an open window. Unfortunately, it is not clear from the table whether the increased risk is among people who do not sleep near an open window or among those whose sleeping status is unknown. The data for the latter category (unknown) should be presented separately, as in Table 4. It would appear that Table 5, as it stands, is not particularly useful as a measure of the risk posed by exposure to radon.
Although the data in Table 4 show that the risk of lung cancer increases with the radon level and amount of smoking, a large proportion of the increase is due to smoking. None of the trends of an increased risk of lung cancer with a higher level of exposure to radon are significant.
Information on passive smoking was collected, but the results were not presented. Other important risk factors for lung cancer (e.g., diet and a family history of cancer) were not included in the study, and this may have had considerable confounding effect on the results.
With respect to radon exposure, measurement over a three-month period is not adequate. A year-long measurement is necessary to account for seasonal variability. The use of radon measurements from previously occupied homes is problematic, because of changes with different ownership and different characteristics of the house (e.g., changes in the heating system).
Because of the design and analytic limitations of this study, one should be cautious about interpreting the results to indicate either that radon in homes causes lung cancer or that the risk associated with radon exposure is multiplicative when exposure is combined with smoking. This study should not be used in making policy decisions about radon-exposure levels, particularly in the United States.
1 ReferencesJohn S. Neuberger, Dr.P.H.
University of Kansas Medical Center, Kansas City, KS 66160-7313To the Editor:
Pershagen et al. do not state the incidence of lung cancer in the population under study. It is therefore difficult to know what to make of their relative-risk figures. Although exposure to radon appears to increase the risk of lung cancer, is the risk of practical importance? As it is reported, this study is of no value to me as a primary care physician.
Howard T. Chatterton, M.D.
Marshfield Clinic, Ladysmith, WI 54848Author/Editor Response
The authors reply:
To the Editor: In epidemiologic studies of etiologic factors in lung cancer, it is important to have smoking information of adequate quality. We used questionnaire data obtained from the subjects in our study or their next of kin. As we indicated in our article, the method has been validated, and the risk estimates for different smoking categories were within the expected range. Some misclassification of detailed smoking habits is probable. However, since smoking was negatively associated with residential radon exposure, any residual confounding effect would be expected to contribute to an underestimation of the risk of lung cancer related to radon exposure. It is unlikely that risk factors for lung cancer not considered in our study would cause considerable confounding of the results.
An earlier report based on the same data suggested that the association between estimated radon exposure and lung cancer was stronger in men than in women, although the difference was not significant.1 In our study the relative increase in risk related to radon exposure seemed to be larger in the smokers than in the nonsmokers. Together with the higher smoking rates among the men, this radon-related risk factor contributed to the stronger association among the men. The annual incidence of lung cancer in Sweden from 1980 to 1984 was 71.0 and 24.2 per 100,000 men and women, respectively, between the ages of 35 and 74 years.
Information on sleeping near an open window was available for 88.3 percent of the study subjects through questionnaire responses. The excess relative risk of lung cancer per 100 Bq per cubic meter (2.7 pCi per liter) of time-weighted average exposure to radon was 0.17 (95 percent confidence interval, 0.05 to 0.54) for those subjects sleeping near closed windows. For the subjects who did not provide this information, the corresponding excess risk was 0.18 (95 percent confidence interval, 0.05 to 0.64); consequently, the two categories were combined in the analysis. It is likely that a minority of the subjects who did not provide this information slept near an open window.
Errors will occur when residential radon exposure is assessed retrospectively. Sources of error include the inability to measure radon in all dwellings of the study subjects, short measurement periods, errors in measurement, alterations to homes, and variation in the time spent at home. To the extent that the errors are independent of disease status and confounding variables, they will tend to attenuate the relation between radon exposure and lung cancer. Errors in the exposure assessment deserve attention in estimating the risk of lung cancer related to residential radon exposure. This will be addressed in subsequent reports on our study, as will interactions with passive smoking among nonsmokers and population-attributable risks.
1 ReferencesGoran Pershagen, M.D.
Frederic Lagarde, B.Sc.
Karolinska Institute, S-171 77 Stockholm, Sweden
