RN EPIDEM STUDIES. UNCERTAINTIES IN ASSESSING HEALTH RISKS FROM NAT. RAD.
RADON EPIDEMIOLOGIC STUDIES. UNCERTAINTIES IN ASSESSING HEALTH RISKS FROM NATURAL RADIATION.
Alexandra Flore, Constantin Cosma
“Babes-Bolyai” University, Faculty of Physics, Cluj-Napoca, 3400, Romania.
Epidemiological studies of human health and natural radiation may provide a direct way of evaluating the estimates validity of risk at low doses that are based on groups with generally higher radiation doses. In the absence of direct quantitative evidence, risks have to be estimated by extrapolation from higher exposures where direct effects can be observed, sometimes received over a fairly short period of time.
Indoor radon is now believed to be the most important source of ionizing radiation in the environment.
However, uncertainties can effect the interpretation of these studies. Even though natural radiation forms the main source of radiation exposure for most people, obtaining reliable information from epidemiological studies of these exposures is not easy, because of low statistical power, confounding factors, combined effects with other agents, uncertainties in assessing exposure. This paper reviews some of these uncertainties, presents different types of epidemiologic studies, comparing their advantages and disadvantages, giving particular attention to residential radon and lung cancer.
Introduction
During the last two decades, potential lung cancer risks due to inhalation of radon daughter products have been the subject of wide concern, not only because of the potential impact on the well-being of people living or working in high-natural environments, but also because of the high costs of radon remediation programs, for example, of uranium minig areas and private homes. Besides obvious economic, social, and psychological implications, the assessment of potential residential radon risks became an important test case for the validity of the linear-no-threshold LNT hypothesis, the related collective dose concept, and the value of epidemiological studies.
Indoor radon is now believed to be the most important source of ionizing radiation in the environment. Radon’s adverse effects were first observed in miners, especially uranium miners, and now increased attention is being directed to its implications for the general population, in order to correctly estimate its health hazards. Based on miners’ and some case control epidemiological studies, a large number of additional lung cancer in the population are due to residential radon, but also various recent studies (with never-smoking women in high-radon areas, as well as animal and cell experiments) indicate thresholds, perhaps even biopositive effects including the reduction of other types of cancer in areas with increased residential radon levels.
In the absence of direct quantitative evidence, risks have to be estimated by extrapolation from higher exposures where direct effects can be observed, sometimes received over a fairly short period of time. Particular examples are underground miners exposed to high levels of radon and the Japanese atomic bomb survivors. Epidemiological studies of human exposures to radiation, including residential radon, may provide means of assessing the validity of risk estimates based on the above extrapolation. But there are a lot of confounding factors and other agents that interfere because the subjects are placed in real life situation in a complex and changing human environment. In order to estimate the effects of radiation on population there have been developed models for low doses range from linear no-threshold to large thresholds, or considering hormesis [1], or the bystander effects [2].
2. Different types of epidemiological study
In order for a study of natural radiation to be informative, it is necessary to maximize the statistical power [3] and to minimise the potential for bias or confounding factors. Bias represent systematic errors in the design or conduct of the study, while a confounding factor is the effect or one or more variables that are correlated with both the exposure and the disease outcome of interest.
2.1 Case-control study
A case-control is an investigation that compares the exposure received by a group of persons with a disease (the cases) with a group of persons without the disease (the controls). It is not necessary to follow a large number of people for many years in order to identify the cases, and usually it is possible to obtain individual data on exposures by measurements in dwellings of radon or gamma radiation and the confounders by questionnaires. The bias may exist because of the retrospective nature of this type of epidemiological study, such as different participation rate in cases and controls, quality of information. Also the statistical power of one case-control study is not high.
Residential radon case-control studies have some advantages over other types of case-control studies in determining retrospective exposure, because a significant proportion of the radon exposure occurs in the home and the radon concentrations can be measured at some later date. Uncertainty in the estimating retrospective radon exposures increases when certain time periods in the 15 to 20-year time period before study are missing. For example some case-control studies in the USA [4] and Sweden], in order to reduce the bias due to estimating past exposure, measurements from glass items that were located in current and previous homes were performed. Only using the data containing integrated radon exposure over a longer period through glass measurements resulted in a statistically significant association between the radon gas and the risk of lung cancer.
2.2 Correlation study
Known also as an “ecological” or ‘geographical” study, the correlation study is based on aggregated data for disease rates, exposures and possibly confounding factors. In the case of radon correlation study we look for a correlation between the lung cancer rate in certain geographical areas and the mean radon exposure in these areas. Since it is often possible to include large populations, the statistical power of correlation studies can be high. Smoking is an important bias/confounding factor in assessing the risk of lung cancer and also the differences between individual and aggregated levels versus aggregated exposure and confounders are important in the case of residential radon. [5]
2.3 Cohort study
These studies are based on identifying a specific group of individuals and following them over time (prospectively or retrospectively) in order to determine mortality and disease incidence. The statistical power rises only if the cohort is followed for many years, but in the context of natural radiation, this study is not efficient. An important example is the study that has been conducted in China [6].
3. A discution on statistical power
In order to estimate at best the relationship between radon exposure and lung cancer mortality the statistical power should be increased. A possible approach could be the increase of the number of cases in a study by having a lot of persons with high/low exposure. Another one could be to quantify the results of existent studies, or by using their data to interpret them in different statistical models.
The meta-analysis method is based on the summary of different published studies in order to obtain an overall estimate of risk. The results of individual studies range from positive to negative correlations as shown in fig.1. For example, Cohen [5] observed that the higher exposure to domestic radon correlated with the lower incidence of cancer (negative correlation), whereas Lubin and Boice [7] have reported a positive one, in agreement with the extrapolated data from the underground miners.
Fig.1. – Summary relative risk from meta-analysis of indoor radon studies [8]
Also is used another method, called combined or parallel analysis, which consists in interpreting the raw data obtained in different studies and comparing them, in order to examine the importance of various factors and their relevance such as smoking, age, gender.
The most serious problem appears to be the retrospective evaluation of smoking habits, and the underestimation of use could invalidate some of the well-known case-control studies. There is a substantial evidence for a threshold, or a U-shaped response curve [9] that can explain the beneficial effects attributed to radon in balneology.
Also, at the molecular level new assumptions and effects are studied in order to get a better perspective of low dose response. The bystander effect [2], used originally to develop oncogenis transformation in vitro, has been extended to low dose rates. Radiation response is considered as a superposition of bystander and the linear direct effects, and these data are fitted to the data from miners exposed to radon. The results suggest that, at low radon exposure, the risk could be dominated by bystander effects. There are new techniques and considerations to be made in order to better understand the significance of the epidemiological data thus increasing the statistical power of the studies. The environmental biodosimetry can give an estimate of cumulative lifetime dose, improving the accuracy and reducing the uncertainties in ecological risk assessment.
Implementing these methods to enhance the validity of risk estimates from the studies of natural radiation would provide in the future a better estimate of the lung cancer risks from residential radon.
Bibliography
[1] JawarowskI, Z. Beneficial Effects of Radiation and Regulatory Policy. Hormesis and Radioadaptive Response Web Site Australian Physical & Engineering in Medicine, 20, 1997
[2] Brenner, D.J., Sachs, R. K. Bystander Effects May Dominate Domestic Radon RiskBut Curent Risk Estimates are Probably Okay. Radiat. Res., 158, 2002, 790-791
[3] Muirhead, C. R. Uncertainties in assessing health risks from natural radiation, including radon. Elsevier. International Congress Series 1225,231-237.
[4] Field R. W, A Review of Residential Radon Case-Control Epidemiologic Studies Performed in the US, Reviews on Environmental Health, vol. 6, no.3, 2001.151-167.
[5] Cohen, B.L. Test of the linear-no threshold theory of radiation carcinogenesis for inhaled radon decay products. Health Physics 68, 1995, 157.
[6] Tao, Z., Zha Y., et al., Cancer mortality in high background radiation area of Yangjiang, China, 1979-1995. Natl. Med. J. China 79, 1999, 487-492.
[7] Lubin J. H. and Boice, J. D, Ling Cancer Risk from residential Radon: Meta-analysis of Eight Epidemiological Studies. J. National Cancer Inst. 89,1997,49-57.
[8] Lubin, J .H. Rejoinder: Cohen’s response to “On the discrepancy between epidemiologic studies in individuals of lung cancer and residential radon and Cohen’s ecologic regression”. Health Phys., 75, 1998, 29-30.
[9] Becker, K. residential radon and the LNT hypothesis. Elsevier.International Congress Series 1225, 2002,259-266