Skip to main content
hpa logo
Topics A-Z:
Search the site:
Home › Topics › Radiation › Understanding Radiation › Frequently Asked Questions ›Epidemiology FAQs
Printer friendly page (opens in new window)

Epidemiology FAQs

 

What is epidemiology?

Epidemiology is the study of the distribution of disease in populations and of the factors that affect this distribution. In contrast to clinical medicine where the emphasis is on the individual, epidemiology involves the examination of patterns of disease in groups of individuals. While epidemiology originated from investigations of epidemics of infectious diseases in the 19th century, epidemiological research in western countries is now directed largely at chronic diseases, such as heart disease and cancer. Indeed, epidemiology has played a vital role in identifying and quantifying the health risks of cigarette smoking and exposure to agents such as asbestos, as well as radiation.

 

What effects can radiation have on health?

The adverse health effects associated with radiation exposure can be split into two categories. Deterministic effects generally arise shortly after exposure to a radiation dose, but only if this dose exceeds some threshold value. The severity of these effects, but not the probability of their occurrence, depends on the level of dose. Examples of deterministic effects include damage to body tissues such as the red bone marrow, gastrointestinal tract, central nervous system, lung and skin; at very high doses, these effects may lead to death within a short period. Deterministic effects that are generally non-fatal include vomiting and impaired lung function. For a given total dose, the severity of the effects tends to be less if the exposure is protracted than if it is received acutely.

In contrast to deterministic effects, the probability but not the severity of stochastic effects of radiation depends on dose. In particular, the probability appears to increase without a threshold in dose. The main types of stochastic effect, which may arise many years after radiation exposure, are the induction of cancer in persons exposed and of hereditary disease in subsequent generations.


 

What are the main sources of information on the late radiation effects?

Information on health effects induced years after radiation exposure arises from various studies of exposed groups. The most important of them is the follow-up of roughly 90,000 survivors of the atomic bombings of Hiroshima and Nagasaki, known as the Life Span Study (LSS). Particular strengths of this study are:

  • the large population, covering both males and females, and all ages.
  • the wide range of radiation doses received.
  • the exposure of the whole body, rather than principally of just a few body organs.
  • the ability to make individual estimates of organ-specific doses.

In view of considerations such as these, the LSS forms the main basis for estimates of late radiation risks derived by national and international bodies such as the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR), the International Commission on Radiological Protection (ICRP), and the US Committee on the Biological Effects of Ionising Radiation (BEIR), as well as the Health Protection Agency in the UK.

Other studies, such as of patients treated with radiation and of people exposed to radiation in the course of their work, are also informative. For example, studies of radiation workers provide information on the effects of protracted rather than acute exposure. In addition, whereas the Japanese survivors received predominantly low linear energy transfer (LET) exposure from gamma radiation, risks from exposure to high-LET radiation from alpha particles can be estimated, for example, from studies of persons exposed to radon in mines or in dwellings and from patients and workers with intakes of radium.

 

What are the major findings among the Japanese atomic bomb survivors?

The main radiation effect has been the induction of cancer. Up to 2000 – in other words, 55 years after the bombings of Hiroshima and Nagasaki – the excess number of deaths from all cancers other than leukaemia was estimated to be about 480 in a population of around 86,000; this represents about 8% of all such deaths among the, roughly, 48,000 survivors with a radiation dose greater than zero. In contrast, the excess number of leukaemia deaths in this group of survivors – estimated to be around 90 – represents over 45% of all leukaemia deaths. The magnitude of the excess increases with increasing dose. Analyses of such data have allowed the construction of models to describe how radiation-induced cancer risks vary according to dose, as well as factors such as age at exposure, time since exposure and sex.

Data on cancer incidence among the survivors (ie both fatal and non-fatal cases) have also been analysed, since they provide more information on cancers that are rarely fatal, such as those of the thyroid and skin and allowed examination of subtypes of leukaemia. In particular, it is known not only from this study but also from studies of people exposed for medical reasons that one particular subtype of leukaemia (chronic lymphatic leukaemia) does not appear to be induced by radiation.


 

Can radiation exposure lead to diseases other than cancer many years later?

Diseases other than cancer have been studied both through analyses of mortality in the Life Span Study of Japanese A-bomb survivors and via clinical examinations of a subset of these survivors. Based on follow-up to the end of 1998, the clinical studies have demonstrated clear associations with radiation dose in the incidence of thyroid disease, uterine myoma, hypertension, and chronic liver disease and cirrhosis. The last of these findings may reflect in part a high level of hepatitis virus infection in Japan. These studies have additionally indicated increased risks of cataract and decreased risks of glaucoma in relation to radiation exposure.

Mortality data for the Japanese survivors have also suggested associations between radiation and atherosclerosis, specifically heart attack and stroke, as well as with diseases of the digestive and respiratory systems. On the assumption that the risk of these diseases varies in direct proportion to the dose received, the estimated number of excess non-cancer deaths up to the end of 1997 among the Japanese survivors has been estimated to be about one-half of the corresponding number of excess cancer deaths. However, the basis for this assumption is not strong; much of the evidence for raised risks of non-cancer disease mortality arises at high and medium doses, and it is unclear whether or not risks are increased at doses below about 0.5 Sv. The evidence from other studies for associations between non-cancer disease mortality and radiation, particularly at low doses, has tended to be variable. Research into this topic is continuing.

Study has also been made of the offspring of the Japanese survivors and of those exposed in utero. Studies of offspring conceived after the bombings have not shown excesses of congenital abnormalities, or of mortality or cancer incidence when followed through to the late 1990s. In contrast, A-bomb radiation exposure in utero has been related to an increased incidence of cancer in adulthood, and to small head size and mental retardation.


 

How are the risks of cancer resulting from radiation exposure assessed?

In general, the types of adverse health effects that can arise over a prolonged period following radiation exposure are clinically indistinguishable from effects that arise in the general population. However, radiation can lead to an increased incidence of these diseases, and possibly also affect the times at which they occur. For example, not only were the observed numbers of leukaemias among the Japanese atomic bomb survivors and some medically exposed groups (for example, patients in the UK treated for ankylosing spondylitis) greater than that those expected, but these excesses were concentrated largely within the first 25 years or so after exposure.

Since late health effects due to radiation exposure cannot be identified clinically, it is necessary to use epidemiological approaches to assess the potential role of radiation in inducing such effects.

Radiation risks can be estimated on the basis of epidemiological findings, such as those of the Japanese atomic bomb survivors. The level of risk can depend on a number of factors. Among them, the size of the radiation dose is very important. Generally the risk varies in direct proportion to the dose, although in some circumstances the trend in risk with dose is more complex. For example, data on the Japanese survivors indicate that the risk of leukaemia varies according to both dose and the square of dose. In this instance, the risk per unit dose is smaller at lower doses than at higher doses.

As well as dose, radiation risks can be affected by factors such as sex and the age at which the exposure occurred. For example, for leukaemia and for many solid cancers, data from the Japanese survivors and other studies indicate that the relative increase in risk is greater for exposure in childhood than in adulthood. Also, the risk of leukaemia tends to be higher soon after exposure than at later times whereas – at least for exposures in adulthood – the relative increase in the risk of solid cancers is fairly stable over time. The degree to which the dose was protracted may also need to be taken into account when calculating the radiation risks, since various animal studies have suggested that – for a given total dose – risks may be higher if received acutely rather than over a prolonged period.


How can we tell whether someone's cancer was caused by radiation?

By their nature, the conclusions drawn from epidemiological studies apply to groups rather than to specific persons. However, information on levels of risk associated with radiation exposure can be used in assessing the probability that a disease suffered by a given person was induced by such an exposure. For example, the US National Institutes of Health have used data on the Japanese atomic bomb survivors to compile tables on the 'probability of causation' relating an individual's cancer to a prior radiation dose, and has recently updated its tables. This topic has also been reviewed by the International Atomic Energy Agency.

For a given exposure scenario (ie the size of the dose, the sex of the person exposed, the age of exposure, the time between exposure and the onset of disease, and the degree of dose protraction), it is possible using models such as those developed by the US BEIR V Committee to derive an estimate of the relative increase in risk. This can then be used to estimate the probability of causation (PC), which is normally quoted as a percentage. In instances where the dose is not known or where there are differing views about its magnitude, the calculation can be reversed to obtain the dose required to produce a given value for PC.

 

Have raised cancer risks been seen at low radiation doses?

The most recent findings for the Japanese atomic bomb survivors indicate that cancer risks decrease with decreasing dose over a wide dose range. It is not possible to identify increased risks at very low doses, owing to limitations on the statistical precision at these levels. However, a recent analysis of cancer incidence has excluded the possibility that risks are increased only at doses exceeding 60 mSv. In general, the results from this study are consistent with there being no dose threshold for cancer risks.

A pooled analysis of studies of thyroid cancer following external irradiation in childhood shows a linear trend in risk with dose, with raised risks identifiable down to about 100 mSv. Also, studies such as the Oxford Survey of Childhood Cancers have indicated a raised risk from prenatal x-ray exposures to the fetus of the order of 10-20 mSv.

 

What is known about the effects of low level radiation received over a long period?

Epidemiological data on chronic low-dose exposures are available from studies of radiation workers. For example, a combined analysis of data from the UK, USA, and Canada showed a statistically significant association between occupational dose and leukaemia risk. Similar findings arose from a large study by NRPB, now the Radiation Protection Division of HPA, of UK radiation workers. The estimates of risk per unit dose, both for leukaemia and all other cancers, are consistent with the A-bomb findings, although the corresponding statistical confidence intervals are somewhat wide. Continued follow-up of these workers will reduce the statistical uncertainties. Further information on the effects of dose protraction arise from a multitude of animal studies. These studies suggest that for a given cumulative low-LET dose, the cancer risk is lower at low than at high dose rates.

 

How confident are you in your risk estimates?

There are several uncertainties in estimates of the total radiation-induced cancer risk in a population of all ages and both sexes. They include:

  • the dose and dose rate effectiveness factor (DDREF) associated with extrapolating results from the Japanese atomic bomb survivors to low doses and low dose rates
  • the method of projecting of risks seen so far in the bomb survivors until the end of life.
  • sampling variation and possible errors in dosimetry for the A-bomb survivors.
  • the relative biological effectiveness (RBE) of alpha radiation relative to low-LET radiation.

The ranges of uncertainty for each of these are generally about a factor of 1.5-2, with the exception of the RBE for which this uncertainty factor is likely to be higher. However, while the DDREF may be either higher or lower by around a factor of 2, lower rather than higher values are likely to be associated with the other sources of uncertainty.

In addition to uncertainties in the total cancer risk, estimates of the risk of specific types of cancer may be influenced by the approach used to transfer risks seen in the Japanese bomb survivors to the UK, where the spectrum of risks in the general population is different. For example, stomach cancer is much more prevalent in Japan than in the UK, whereas the reverse is true for lung cancer. To use the Japanese bomb survivor data to predict risks in the UK requires a model for how radiation interacts with other factors, such as smoking, in inducing cancer. The models used by the Health Protection Agency generally assume that radiation acts multiplicatively on factors that determine baseline cancer rates. While there is a consequent uncertainty in calculations for certain cancer types, the impact on total cancer risk is much lower since the baseline rates for all cancers combined do not differ greatly between Japan and the UK.

It should be emphasised that uncertainties in risk estimates apply across all sources of radiation, whether natural or artificial.


Last reviewed: 21 December 2007