Can screening effects explain increased risk of thyroid cancer among population living near nuclear power plants in Korea?

Following reports in Korea of an increased risk of thyroid cancer among those living near nuclear power plants (NPPs) [1], there has been debate regarding the interpretation of the findings. One suggested explanation is the screening effect on thyroid cancer risk. The basic assumption of this argument is that the screening process increases the number of detected thyroid cancer cases in a population, leading to an increase in incidence and relative risk. The Korea Hydro and Nuclear Power Company granted a project to support this assumption to gain an advantage in their legal litigation regarding thyroid cancer cases. However, these circumstances could occur in a limited condition and cannot always happen. Because the screening effect and causality are different issues, we must distinguish between the two different questions. Even if there is a screening effect, it does not mean that the observed increased risk of thyroid cancer does not result from living near NPPs. Therefore, I would like to make a few comments on this issue. Screening increases absolute risk by increasing the number of detected cancer cases, but it does not directly affect the value of relative risk. If it did not occur differently, the relative risk would not be changed. For example, if a twofold increased screening occurred in both the exposed and non-exposed population, then the relative risk would not be changed. Previously, it was reported that the slope of the dose–response relationship between radiation and thyroid cancer did not differ significantly before and after screening among patients who had received radiation therapy, indicating that the association between radiation and thyroid cancer may not be changed by intensive screening [2]. A study from the Chernobyl accident also reported screening to be a weak confounder for radiation dose and thyroid cancer [3]. Furthermore, only a certain proportion of detected cases from screening would ultimately be confirmed as cancer cases because the device used for screening (e.g., ultraso-nography) is not a diagnostic method. Therefore, the twofold screening rate difference, for example, does not directly indicate a twofold difference in incidence rate. However, when the screening occurs differently, distortion of the relative risk may be occurred. Therefore, we should determine whether populations living near NPPs experience more screening than other populations. If so, we need to examine how much this occurred differentially in relation to distance from NPPs. The next issue is to separately identify the proportion of observed relative …