We used values obtained from currently available published data for estimation of the organ doses attributable to various types of medical radiation. We used local data on organ doses attributable to various types of radiography, where available.78 Unless the data are available locally, we used the data quoted by the previous study in UK6; as for positron emission tomography (PET)-computed tomography (CT), we used the organ doses presented by Huang et al.9 A suggested list of 25 items for the types of radiographic scanning which includes the panoramic radiography proposed by the previous study6 and PET-CT was used (Table 1). Further, we utilized the results of the study conducted by Kim et al.10 for individual organ doses received from a medical checkup at private clinics.

We divided into 4 the sources from which we can obtain data on how much the entire Korean population has been exposed to medical radiation for a year.

The first source is medical treatments that are covered by the National Health Insurance Service (NHIS), which can be obtained from the Health Insurance Review & Assessment-National Patient Sample (HIRA-NPS) of the Health Insurance Review & Assessment (HIRA) Service updated in 2013.11 These data are a collection of medical records from about 1.4 million patients, which accounted for 3% of the total patients of the year. Because the data are generated with stratified sampling method by which data was extracted from each stratified group (32 strata of sex and 16 5-year age groups), we could assume HIRA-NPS represents the whole Korean population. The medical bill section of this data set contains such information items as treatment start date, main diagnosis, and sub diagnosis, which can be elaborated by the use of the medical records each of which details such information items as medical services, drugs, and treatment materials provided by the hospital. From these data, we checked the codes related to diagnostic radiation and calculated the frequencies of radiographic services rendered to the patients during the year in terms of sex and 5-year age group. In 2013, 90.0% of the entire population visited hospitals for medical treatment and checkup; accordingly, the above-mentioned frequencies were multiplied by 0.90 to re-calculate the frequencies of insurance benefits to the entire population.

The second source is health examinations performed nationwide by the NHIS. The adults of interest in this study undergo annually or biennially general health checkups or transitional age health checkups at the ages of 40 and 66 during which they are exposed to medical radiation at the time of X-ray scanning of the chest. As the cancer screening service is rendered on national level, they have their stomach and breasts checked for cancer once every 2 years from the age of 40 and have their colon examined for cancer once every year from the age of 50. For examination of the stomach, one has an option for either upper gastrointestinal series or gastroscopy; for the colon, one has a choice for double contrast barium enema or colonoscopy after the patient is subjected to a stool guaiac test. And, for the breasts, mammography is used. These data are obtained from the Korean Statistical Information Service12 and used to calculate the frequencies of examinations by sexes and 5-year age groups.

The third source is private health examinations. It is quite difficult to systematically obtain detailed information on the health examinations performed on private level. Based on the recent study of 296 hospital surveys nationwide,10 we calculated average radiation dose for each organ by gender, for one time private health examination: using the information of the frequency of medical radiation use in private health examination (72% CT, 16% gastro-colonography, 9% PET, and 3% X-ray) and the contributing proportion of CT to each organ (abdomen 39%, thorax 32.2%, cardiac angiography 19.5%, brain 2.5%, lumbar spine 2.5%, low dose lung 1.3%, and cervical spine 1.1%). Next, we obtained the information of private health examination prevalence by gender and 5-year age strata based on the 2013 Survey on National Health and Nutrition (Korea National Health and Nutrition Examination Survey [KNHANES]) conducted by the Center for Disease Control and Prevention.13 We utilized the frequency of answer “Yes” to a question “Have you ever had a general health check privately during the last 2 years?” in the questionnaire of 2013 KNHANES. The numbers were divided by 2 and assumed to be cross-sectional frequencies for sexes and 5-year age groups.

The fourth source is conscript health examinations performed by the Military Manpower Administration (MMA) that targets males aged 19 residing in Korea.14 In 2014, there were 351,075 individuals age 19 or more eligible for such examination; and they were included into the 15–19-aged male statistics assuming that they all received simple chest X-ray scans.

All the doses attributable to diagnostic medical radiation that were identified from those sources were added up; that is to say, the numbers of radiographs obtained from the medical bill payments, general health checks, and transitional ages health checkups of the NHIS, cancer screenings, conscript health examinations from the MMA, and private general medical examinations were calculated in terms of sexes and 5-year age groups, which were then multiplied by estimated organ doses and summed up. Using these data, we calculate the exposed organ doses attributable to total diagnostic medical radiation in terms of sexes and age groups.

The Biological Effects of Ionizing Radiation (BEIR) VII report15 uses the concept of LAR, where medical radiation LAR indicates the risk of cancer incidence or mortality of an individual or age group resulting from exposure to medical radiation over their lifetime from birth to a specific point of time (at the age of 85 in this study). The LAR can be obtained by using both excess absolute risk (EAR) model and excess relative risk (ERR) model. The beta coefficients obtained from the EAR and ERR models are used to understand how many more persons will be exposed to such risk by each unit exposure and how the rate changes against the baseline.

The LAR at the age of “a” for a gender (sex) “g” who is exposed to exposure “D” at an exposure age “e” can be expressed in the following expression:

Where L = minimal latent period (5 year for solid cancers, 2 year for leukemia), S_aj(a,g) = a survival rate for the gender “g” aged “a” unless the person suffers exposure to radiation, S_aj(e,g)/S_aj(e,g) = a conditional probability that a person survives from the exposure age “e” till an attained age, M(a,g) = baseline risk, ω = risk transport weight factor.

The background cancer incidence and mortality in 2013 were obtained from the Korea Central Cancer Registry at the National Cancer Center16 and the National Mortality Database provided by the Korean Statistical Information Service,17 respectively, and the probabilities of survival to age were also provided by the Korean Statistics Information Service.

Together with those data and the calculated organ doses per sex and 5-year age groups, by using the same cancer-specific parameters of the EAR and ERR models as are presented in the BEIR VII, we estimated the diagnostic medical radiation LARs in Koreans up to the age of 85 for cancers relating to the bladder, colon, liver, lung, stomach, thyroid, and breasts and leukemia.

The values obtained in the EAR and ERR model can be weighted to get LARs. As in the BEIR VII methodology, combined estimates are calculated from the weighted averages of the logarithmic scale of the estimates obtained from the EAR and ERR models, where relative risk transport was given a weight of 0.7 and absolute risk transport was given a weight of 0.3 (in the opposite way in case of lung cancer). For female breast cancer contrariwise, 100% EAR model was used. As for thyroid cancer, the estimates from the ERR model were used per se because those from the EAR model are not available. Finally, dose and dose-rate effectiveness factor (DDREF) 1.5 was applied to all the combined estimates.

The information of background incidence16 and mortality,17 and the cancer-specific LARs calculated by the use of the actual number of population in 201318 were used to get their fractions of LARs and presented population PAFs and the number of patients attributable to diagnostic medical radiation exposure for each type of cancers. Finally, the total PAFs summed up from 7 male-dominant cancers and 8 female-dominant cancers were obtained and applied to the total cancer incidence, from which the LARs for all types of cancer were estimated for the Korean population aged up to 85. The figure shows how to come up with the PAF (Fig. 1).

The R version 3.1.3 (R Foundation, Vienna, Austria; Comprehensive R Archive Network: http://cran.r-project.org) was used in data management and mathematical calculation.

The study protocol was approved by the Institutional Review Board of Sungkyunkwan University (SKKU 2014-06-008).

The annual average organ doses increase with increasing ages in both of the male and female age groups. In males aged up to 79, the dose in the lungs shows the highest with 57.73 mGy, while in females the dose in the stomach reads the highest with 51.57 mGy (Table 2).

The LARs for cancer incidence show 0.38 stomach cancer cases, 0.31 lung cancer cases, and 0.27 colon cancer cases per 100,000 males and 1.45 thyroid cancer cases, 0.60 lung cancer cases, and 0.95 breast cancer cases per 100,000 females. The LARs for cancer mortality reveals that lung cancer was the top cause of cancer deaths in both sexes (male, 0.25; female, 0.40 per 100,000) (Table 3).

The PAFs are distributed from 0.3% to 6.4% depending on sexes and cancers. The PAF for leukemia incidence is the highest with 2.6% in males whereas that for bladder cancer is the highest with 3.9% in females. In terms of cancer mortality, the PAF for leukemia is the highest with 2.1% in males while that for thyroid cancer is the highest with 6.4% in females. The PAFs that reflect 7 male-dominant cancers indicate that medical radiation accounts for 0.6% of the cancer incidence and mortality respectively in males out of the total population. They also show that, in females and 8 female-dominant cancers, medical radiation accounts for 1.2% of their cancer cases and 1.7% of their mortality cases, which are larger than those in males. The PAF is estimated to be 0.9% in the entire population. When we assume that ionizing radiation has causal effect on the incidence of all cancers, total PAFs for both sexes could be applied to the incidence and mortality numbers of total cancer. A total of 1,915 cases of cancer and 637 cancer deaths in 2013, in association with exposure to diagnostic medical radiation over their lifetime, were found (Table 4).