Income differences in screening, incidence, postoperative complications, and mortality of thyroid cancer in South Korea: a national population-based time trend study

The study used data from the Korea Community Health Survey (KCHS) conducted in 2010, 2012, and 2014 by the Korea Centers for Disease Control and Prevention, data from the National Health Information Database (NHID) of National Health Insurance Service (NHIS) between 2001 and 2015, and cause-of-death statistics from Statistics Korea between 2006 and 2015. The KCHS is a nationally representative community-based cross-sectional health survey that targeted adults over 19 years of age [17]. The screening rate of TC was drawn from the KCHS items regarding TC screening. The incidence, mortality, and postoperative complication rates of TC, as well as the incidence and mortality rates of the negative control outcomes, were derived from the NHID and cause-of-death data. The NHID covers whole Korean population and includes income-based insurance premiums, demographic variables, information on the date of death, and information on healthcare utilization, thereby making it possible to analyze the incidence of diseases, medical care utilization, and mortality [18]. The study analyzed the mortality rate using individual linkage between the NHID data and cause-of-death data. Individual linkage of the raw data was conducted within the NHIS, and researchers utilized aggregate data without personal identifiers for the analysis.

This study was approved by the National Health Insurance Service of Korea (NHIS-2018-1-191) and the Seoul National University Hospital Institutional Review Board (IRB No. E-1805-047-944). Informed consent was waived by the board.

TC examinees were defined as those who responded “yes” to an item including “thyroid cancer” among questions regarding all types of cancer screening (including national cancer screening and opportunistic screening) in the most recent 2 years from the KCHS data. All survey participants in KCHS, regardless of their baseline health status, responded to the question. To calculate the incidence rate of TC in each year, the study targeted patients who were first diagnosed with TC (ICD-10 code: C73) as a primary or secondary diagnosis every year between 2006 and 2015. The first day of treatment was considered as the date of diagnosis each year, and the study targeted patients who were admitted for TC once or more in a year or who received outpatient medical services for TC more than twice. However, patients with a history of any type of treatment for TC in the 5 years before the date of diagnosis (e.g., any history between January 1, 2001 and December 31, 2005 for an incident case on January 1, 2006) were considered to have a prior history of TC and were therefore excluded from the calculation of the incidence rate for that year. The number of cases and the age-standardized incidence rates between 2006 and 2015 derived from the NHID using the method described above were very close to those based on the Korea Central Cancer Registry (KCCR) (Supplementary Table S1). Since information on income, medical care utilization, and complications cases was not available in the KCCR, we used the NHID for this study. TC Mortality was defined as instances of people over 20 years of age with TC (ICD-10 code: C73) as the cause of death by year between 2006 and 2015.

TC surgery-specific complications are defined as those specifically due to TC surgery, including hypoparathyroidism/hypocalcemia and vocal cord/fold paralysis [19]. General complications include postoperative fever, local complications (e.g. infection, hemorrhage, and hematoma), cardiopulmonary complications (e.g. pneumonia), and myocardial infarction, which commonly appears within 30 days after surgery [19]. The incidence of postoperative complications was calculated using patients with TC surgery-specific complications. Diagnostic and surgical procedure codes were confirmed in the healthcare utilization database of the NHID; thus, for this study, patients aged 20 or over who first underwent surgery for TC between 2006 and 2015 were included. Among those patients, postoperative complications were considered to have occurred in individuals who were hospitalized more than once or who used outpatient medical services once or more due to hypoparathyroidism (ICD-10 code: E20.9, E89.2) or vocal cord/fold paralysis (ICD-10 code: J38.0) within 31 to 365 days from the TC surgery date. However, the postoperative complication rate did not include those who used medical services due to the aforementioned complications within only 30 days after surgery, because such instances of complications were considered to have been transient symptoms in accordance with previous studies [19]. Additionally, the postoperative complication rate did not include those who had undergone any treatment for the aforementioned complications during 5 years before surgery, because such patients were considered to have a past medical history. The annual rates of TC surgery by income were also calculated. Thyroid cancer surgery included radical operations on malignant thyroid tumors, total thyroidectomy, and subtotal thyroidectomy, as well as the excision of cervical lymph nodes and neck lymphatic dissections that were performed as part of combined-modality treatment due to cervical lymph node metastasis.

The negative control outcomes were lung cancer and stroke since they were assumed to be less susceptible to overdiagnosis. The incidence and mortality of both lung cancer and stroke were defined similarly to those of TC. In the annual healthcare utilization database of NHID, incident cases of lung cancer were defined as individuals 20 years old or over who were first diagnosed with lung cancer (ICD-10 code: C33-C34) between 2006 and 2015. This study targeted patients who were hospitalized for lung cancer at least once or had outpatient visits twice or more for a year from the diagnosis date, as for cases of thyroid cancer. Then, patients with a past history – those who have been treated for lung cancer during the five years before the diagnosis date – were excluded. The number of cases and the age-standardized incidence rates of lung cancer from the NHID were also very close to those based on the KCCR (Supplementary Table S2). Using the annual healthcare utilization database between 2006 and 2015, incident cases of stroke were defined by identifying patients 30 years of age or over who had received their first diagnosis of stroke (ICD-10 code: I60-I64) in that year and including those were hospitalized with stroke more than once in each year. In addition, patients who had admitted for stroke the prior 5 years in each year between 2006 and 2015 were excluded because they were considered to have a past medical history. As with mortality from thyroid cancer, cases of mortality from lung cancer and stroke were defined based on linkage between the cause-of-death data from Statistics Korea and the NHID data. Those 20 years of age or older with an ICD-10 cause-of-death code of C33-C34 and those 30 years of age or older with an ICD-10 cause-of-death code of I60-I64 were defined as cases of mortality from lung cancer and stroke, respectively.

Household income from the KCHS data and national health insurance (NHI) premiums from the NHID data were used as an income index. Many previous studies in Korea have used the NHI premiums as an indicator of income [20, 21]. For the income index of TC examinees in the KCHS data, this study utilized converted values of equivalized income, in which items related to annual or monthly household income were adjusted for household size. The income index of the incidence, mortality, and postoperative complications of TC and the negative control outcomes was calculated using equivalized income adjusted for household size using NHI premium data from the NHID. In addition, each equivalized income parameter drawn from the KCHS or NHID was divided into income quintiles by year, sex, and 5-year age group [22].

The changing patterns of the age-standardized screening, incidence, surgery, postoperative complication, and mortality rates of TC were analyzed by sex and income level between 2006 and 2015. For age standardization, subjects were divided into 5-year groups, and the 2010 mid-year population from Statistics Korea was used as the reference. Unlike other types of cancer, TC has a very low mortality rate; thus, the annual age-standardized mortality rate by income quintile was derived by combining the data between 2006 and 2010 and the data between 2011 and 2015, respectively, to ensure the stability of TC mortality rates. The incidence and mortality rates of lung cancer and stroke (the negative control outcomes) were derived using age-standardized rates for each income quintile by calendar year between 2006 and 2015. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).