Thyroid fine-needle aspiration biopsy positively correlates with increased diagnosis of thyroid cancer in South Korean patients

The data used in this study were obtained from the NHIS National Sample Cohort 2002–2013 released in 2014 and include a nationally representative random sample of 1,025,340 individuals, approximately 2.2% of the entire NHIS population in 2002. The data were compiled by the NHIS using a systematic sampling method to generate a representative sample of 46,605,433 Korean residents. The database includes all medical claims filed from January 2002 to December 2013. To investigate the relationship between the number of thyroid fine-needle aspiration biopsies in each geographic region and newly diagnosed cases of thyroid cancer, we excluded patients who were diagnosed with thyroid cancer (ICD-10: C73) before 2003. We then identified patients who underwent a thyroid biopsy (EDI code: C8591) and aggregated this number as a unit of 253 basic administrative districts (Si-Gun-Gu; city-county-ward) of South Korea. Data used in this study consisted of 11,024,548 person-years of 1,122,456 individuals during 2003–2013.

Our outcome variable was the number of newly diagnosed cases of thyroid cancer during the study period, indicated by the first hospital visit during which thyroid cancer (ICD-10: C73) was the major diagnosis for each patient.

The primary independent variable was the number of fine-needle aspiration biopsies performed in each Si-Gun-Gu. We first identified whether patients received thyroid needle aspiration biopsies based on EDI Code and aggregated the number of biopsies as a unit of Si-Gun-Gu per each year. We then calculated the number of biopsies per 100,000 patients using the following formula:

We also adjusted other independent variables when analyzing the association between the number of biopsies per 100,000 people and cases of newly diagnosed thyroid cancer. Other independent variables included sex, age, income, type of insurance coverage, study year, region, and the financial independence rate of the local government. Ages were categorized as ≤19, 20–29, 30–39, 40–49, 50–59, 60–69, 70–79, and ≥80 years to reflect differences in diagnosis of thyroid cancer [18]. The types of insurance coverage were categorized as medical aid, NHI employee insurance, or NHI self-employed insurance based on NHI criteria. Those with NHI employee insurance included workers and employers in all workplaces, public officials, private school employees, continuously insured persons, and daily paid workers at construction sites. Beneficiaries of NHI employee insurance included spouses, descendants, siblings, and parents. Individuals with NHI employee insurance paid approximately 7% of their average salary in contribution payments, though these rates usually changed annually. The NHI self-employed insurance category included people whose contribution amount was set based on their income, property, living standard, and rate of participation in economic activities. Medical aid beneficiaries were patients with an income below the government-defined poverty level or who had a disability and were provided with free in- and outpatient care via government funds. Therefore, the type of insurance coverage represented each patient’s socioeconomic status [2]. We included this variable in the study to consider potential differences in the accessibility of thyroid cancer screening according to socioeconomic status. The financial independence rate of the local government was an index of the finance utilization capacity of a local government with independent discretionary power, which was calculated as: (local taxes + non-tax revenue)/local government budgets × 100 [19].

We first examined the frequencies and percentages of each categorical variable or the mean and standard deviation of each continuous variable at each patient’s baseline, respectively. We performed χ2 tests to analyze the distribution of person-years for each categorical variable by diagnosis of thyroid cancer and an analysis of variance (ANOVA) for each continuous variable by diagnosis during the study period. The tests were performed in all study subjects and patients who received thyroid biopsy, respectively. Finally, we performed Poisson regression analysis using generalized estimating equations (GEE) to investigate the relationship between the number of thyroid biopsies and cases of newly diagnosed thyroid cancer adjusting for sex, age, income, type of insurance coverage, study year, region, and financial independence rate of regional government. GEE models with link logit that included both patient- and region-level variables were analyzed, as data used in this study were hierarchically structured and had binary outcome variables. This model assumed proper distributions for each hospitalization case while taking into account the correlation among individuals within the Si-Gun-Gu. In this study, the correlation was an exchangeable correlation structure [20]. To identify whether thyroid biopsies were unnecessary for diagnosis of thyroid cancer, we also analyzed the relationship between the number of thyroid biopsies and newly diagnosed thyroid cancer cases only among patients who received thyroid biopsies. The goodness-of-fit for the GEE model was assessed using the quasi-likelihood under the independence criterion (QIC), whose lower value indicated the goodness-of-fit [21]. In addition, we performed sub-group analyses for Poisson regression analyses to compare differences in the association between the number of biopsies and cases of newly diagnosed thyroid cancer according to sex, income, the median number of thyroid fine-needle biopsies, and financial independence rate of the local government. All statistical analyses were performed using SAS version 9.4.