Stochastic effects of radiation: Rad Tech

Stochastic effects of radiation are delayed, probability-based effects associated with ionizing radiation exposure. In diagnostic imaging, these effects matter because these exposures are often low dose; low linear energy transfer, meaning the x-rays deposit energy sparsely and in small amounts; and are delivered intermittently over time.

For radiation protection, the linear nonthreshold, or LNT, model assumes that any dose carries some risk. As dose increases, probability increases, but severity does not. As a radiologic technologist, understanding the epidemiologic evidence about stochastic effects helps inform radiation protection guidelines for medical imaging personnel and patients.

According to the LNT model, the stochastic dose-response relationship is linear and is represented by a straight line with no threshold. In other words, no radiation dose is considered completely risk-free. A higher dose may increase the chance that cancer develops, but it does not determine how aggressive that cancer will be. This differs from tissue reactions, such as radiation-related skin injuries. Tissue reactions have a dose threshold, and severity increases above it.

Because stochastic effects are uncommon, delayed, and indistinguishable from conditions arising without radiation, epidemiology studies are used to examine health patterns across populations.

Gathering epidemiologic evidence involves comparing exposed populations with similar comparison groups and following outcomes over time. Interpretation can be difficult because stochastic effects are rare and delayed; individual doses may be uncertain; and other risk factors can influence outcomes.

For example, after the Three-Mile Island nuclear reactor accident in 1979, estimated radiation doses to the surrounding population were low, making it difficult to detect a small increase in cancer risk. However, after the Chernobyl nuclear power plant accident in 1986, thyroid cancer was identified as radiation-associated, while other late effects remained uncertain. A series of reports on the Biologic Effects of Ionizing Radiation, or BEIR for short, synthesized evidence from many studies to inform radiation protection recommendations.

Because a radiation-induced effect usually can't be identified in one person, risk is estimated by comparing populations of exposed groups by calculating relative risk, excess risk, and attributable risk.

Relative risk compares the observed frequency of a radiation-induced effect in an exposed group with the expected frequency in a comparison group, or an unexposed group. A value of 1 means there’s no difference; a value above 1 means the effect is more common in the exposed group. A value below 1 does not prove a protective effect from low-dose radiation. Instead, it may reflect chance, a small sample size, or differences between study groups. Because protective effects from low-dose radiation have not been proven, imaging professionals continue to follow ALARA, or as low as reasonably achievable.

Now, to calculate relative risk, let’s use a teaching example based on an exposed population. Suppose there are 227 leukemia cases among 100,000 exposed people, compared with 150 cases in 100,000 people in the general unexposed population. To calculate the relative risk, the observed number is divided by the expected number, so 227 is divided by 150. This equals 1.51, which means the exposed group has about 1.5 times the risk of the unexposed group.

The same data can also be used to calculate excess risk, which is the observed number of cases minus the expected number of cases. In this example, 227 leukemia cases were observed, while 150 were expected. Subtracting 150 from 227 gives 77 excess cases per 100,000 people, meaning 77 cases above the expected number in the comparison population.

Attributable risk estimates the additional number of cases associated with a specific radiation dose in a defined population. For example, suppose three excess leukemia cases occur each year among 100,000 people who received an average dose of 20 mGy. Start by expressing the same rate for one million people. One million is 10 times larger than 100,000, so three cases become 30 cases. That gives 30 excess cases per one million people per 20 mGy each year. Next, express the risk per 10 mGy. Because 10 mGy is half of 20 mGy, and the calculation assumes a linear relationship, the estimated number of cases is also cut in half, from 30 to 15. The attributable risk is therefore 15 excess leukemia cases per one million people per 10 mGy each year.