Estimation of health risk due to ionising radiation

If ionising radiation strikes the human body, it can cause damage in individual cells or tissues. In terms of radiation damage, a fundamental distinction is made between deterministic and stochastic damage. Deterministic radiation damage (e.g. reddening of the skin or hair loss) occurs when a person has received a radiation dose in excess of approx. 500 millisieverts (mSv). Stochastic radiation damage, on the other hand, can occur even below this threshold and includes diseases (e.g. cancer) and damage that only has a certain probability of occurrence.

This article describes how these probabilities – also referred to as “risks” in the field of epidemiology – can be estimated. One major challenge is that it is not possible based on the symptoms alone to differentiate these radiation-related diseases (e.g. cancer) from diseases that occur spontaneously or due to other causes. It is therefore only possible to establish a causal relationship with radiation if there is a statistically significant increase in incidence in people exposed to radiation and a consistently higher incidence across various groups of people relative to unexposed control groups, and if a relationship can be demonstrated between the dose and the level of disease risk (dose-effect relationship).

Estimation of cancer risk

In order to determine the radiation-related cancer risk, significant epidemiological studies have been carried out, particularly in the following groups of people:

• survivors of the atomic bomb detonations over Hiroshima and Nagasaki
• patients exposed to radiation for the purposes of diagnosis and therapy (e.g. the Canadian fluoroscopy cohort)
• occupationally exposed individuals (e.g. the Wismut uranium miners cohort)
• people living in the vicinity of nuclear facilities (e.g. Hanford (USA), Mayak (Russia))
• people living in the vicinity of damaged nuclear power plants (Chornobyl (Russian: Chernobyl) and Fukushima) and individuals deployed in clean-up operations
• individuals affected by above-ground nuclear weapon tests (e.g. people living in the vicinity of the former Semipalatinsk Test Site (Kazakhstan)).

When it comes to estimating the radiation-related cancer risk, the most important data is that obtained in relation to Japanese atomic bomb survivors. This group was exposed to a high dose rate (the entire dose in a fraction of a second), but the dose was only high in a small percentage of those affected.

The cancer risk can be estimated based on the above study populations. This risk comprises two components: the “spontaneous” cancer risk in a population – that is, the general risk of developing cancer without radiation exposure – and the radiation-induced cancer risk. The latter refers to cases of cancer that would not have occurred in the absence of radiation exposure. Models are adopted and estimated for both components, typically assuming a linear model without a threshold in order to estimate the dose-effect relationship. In other words, it is assumed that an increase in radiation dose will be associated with a proportional increase in cancer risk and that there is no threshold below which radiation is not dangerous.

Often, the objective is to make statements regarding the radiation risk not only for a study population (e.g. atomic bomb survivors) but also for other populations (e.g. the German population). In this case, the radiation risk determined in a study population must be applied to the radiation risk for the target population. Given the relatively low levels of radiation exposure that occur in the environment and the workplace today, it is necessary to further extrapolate from the findings obtained for the Japanese atomic bomb survivors: the epidemiological findings, which principally relate to high dose rates, are applied to the exposure situations with low doses and chronic exposure.

There are various ways of doing this: in relation to low doses and chronic exposure, the ICRP recommends dividing the risk coefficients by a factor of 2. Specifically, the ICRP assumes that a dose spread over a longer period has a lesser effect than an equally high dose resulting from short-term exposure. This is intended to take account in particular of the repair and recovery capacity of irradiated cells at low dose and dose-rate values. This reduction does not result directly from the observational data for cancer cases in humans and is based on model assumptions that build upon insights from laboratory experiments. In the view of the BfS, there is insufficient scientific justification for this reduction of the risk coefficients for low doses and chronic exposure.

Risk estimates are fundamentally subject to uncertainties. There are various reasons for this: on the one hand, a study population only comprises a limited group of people, who need not necessarily be representative of the target population of interest. On the other hand, there are many assumptions involved in the models and risk transfers. Furthermore, the recording of radiation dose is often subject to considerable uncertainties.

You can find more information on radiation-induced cancer and the risk of developing such diseases in the article Cancer.

Estimation of risk of diseases other than cancer

At present, it is not possible to reliably assess the risk of developing diseases other than cancer. Evaluations regarding survivors of the atomic bombs dropped in Japan, exposed population groups in the former Soviet Union, and radiotherapy patients indicate that cardiovascular diseases can occur not only at a dose of 0.5 grays or above as a form of late deterministic radiation damage, as was long assumed to be the case, but also at lower doses. The assumption that cataracts (clouding of the lens of the eye) are a type of deterministic radiation damage is also currently being called into question. Here, too, new findings have been obtained that suggest that cataracts already occur at doses 10 times lower than previously assumed (0.5 grays versus 5 grays). Discussions centre around the fact that there may be no threshold dose for these diseases and that, like malignant neoplasms, they are therefore to be viewed as stochastic radiation damage.

Estimation of risk of genetic damage

No reliable findings have yet been obtained in relation to genetic radiation damage in humans. In Hiroshima and Nagasaki, no increase has been identified in the rate of hereditary radiation damage in the offspring of radiation-exposed atomic bomb survivors relative to the rest of the Japanese population. However, it is known from animal studies that radiation can trigger genetic changes, known as mutations, in germ cells. The estimations of genetic radiation risk for humans therefore stem from these animal studies.

You can find out more about radiation-induced genetic damage and the risk of such damage in the article Hereditary radiation damage.

Risk assessment

The information set out above shows how radiation risks can be determined based on individual studies. However, it is almost impossible to carry out a well-founded risk assessment based on a single animal experiment or on a single epidemiological study of humans. In order to assess the health risks of radiation, it is necessary to consult the results of multiple studies and evaluate them within the framework of a general overview. The assessment of health risks is discussed in detail in an issue of Standpoint on Radiation Protection (in German) from the Federal Office for Radiation Protection.