Utilization patterns of traditional medicine in Taiwan and South Korea by using national health insurance data in 2011.

BACKGROUND
The growing popularity of traditional medicine (TM) is reflected in the increasing trend for its use worldwide. Many people are turning to use TM as a complementary or integrative treatment. The aim of this study is to present the first nationwide report describing the use of TM in two countries (South Korea and Taiwan).


MATERIALS AND METHODS
To present the TM utilization patterns between South Korea and Taiwan, we analyzed data from the National Health Insurance cohorts in each country, each of which has approximately one million inhabitants.


RESULTS
In total, 261,478 (25.5%) of 1,025,340 people in South Korea and 260,529 (26.8%) of 970,866 people in Taiwan used TM services at least once under the National Health Insurance in 2011. Using multivariable logistic regression, TM users in South Korea were significantly more likely to be female, 61-80 years of age and individuals with a high income, and those in Taiwan were significantly more likely to be female, 21-40 years of age and individuals with a middle income. The two countries showed similar utilization patterns in visit seasons. People visited TM clinics more frequently than TM hospitals in both countries. The most common TM treatment in South Korea was acupuncture, whereas in Taiwan, various powdered Chinese herbal preparations were the most commonly used treatment. The most common diseases for people seeking TM services were musculoskeletal system and connective tissue diseases in South Korea and Symptoms, signs, and ill-defined conditions in Taiwan.


CONCLUSION
According to the National Health Insurance database, about one fourth of the NHI beneficiaries of South Korea and Taiwan had TM use in 2011. Different TM utilization patterns existed between South Korea and Taiwan, which might be due to the differences in insurance coverage between the two countries.


Materials and methods
To present the TM utilization patterns between South Korea and Taiwan, we analyzed data from the National Health Insurance cohorts in each country, each of which has approximately one million inhabitants.

Results
In total, 261,478 (25.5%) of 1,025,340 people in South Korea and 260,529 (26.8%) of 970,866 people in Taiwan used TM services at least once under the National Health Insurance in 2011. Using multivariable logistic regression, TM users in South Korea were significantly more likely to be female, 61-80 years of age and individuals with a high income, and those in Taiwan were significantly more likely to be female, 21-40 years of age and individuals with a middle income. The two countries showed similar utilization patterns in visit seasons. People visited TM clinics more frequently than TM hospitals in both countries. The most common TM treatment in South Korea was acupuncture, whereas in Taiwan, various powdered Chinese herbal preparations were the most commonly used treatment. The most common diseases for people seeking TM services were musculoskeletal system and PLOS

Introduction
Traditional medicine (TM) is a medical field that maintains health, prevent and treat illnesses based on its theories [1]. The growing popularity of TM is reflected in the increasing trend for its use worldwide [2]. Furthermore, the role of TM in Asian countries such as South Korea, Taiwan, Japan, and China is not only limited to an alternative therapy but is extended to a primary treatment for some diseases [1]. The health care systems and the National Health Insurance (NHI) systems are similar in Taiwan and South Korea. In Taiwan, most of the medical expenditures of residents have been reimbursed by the NHI since 1995 [3]; a year later, health coverage for TM by the NHI was also included [4]. At the end of June, 2014, over 99.9% of inhabitants in Taiwan were covered in the NHI [5]. In Taiwan, the TM services covered by the NHI included powdered Chinese herbal preparations, acupuncture and moxibustion, and manipulative therapy [4]. Specifically, 337 types of Chinese herbal formula and over 500 types of Chinese single herb preparations are covered by the Taiwan NHI. These TM services accounted for approximately 3.1% of the total medical expenditures in the Taiwan NHI program in 2011.
In South Korea, universal health coverage was achieved in 1987, and TM has been included in the health coverage by the NHI since the same year. At the end of 2012, 49,622,000 individuals (97.1% of the population) were enrolled in the NHI system. The TM services covered by NHI in South Korea include acupuncture, moxibustion, cupping therapy and powdered Korean herbal preparations. However, only 58 types of herbal formula and 68 single herb preparations are included in the NHI coverage. These TM services accounted for approximately 3.9% of the total medical expenditures in the Korean NHI program in 2013 [6].
Previous studies on the patterns of TM healthcare utilization were primarily characterized by small sample sizes or were based on sampling surveys. Since 1997, the Taiwan National Health Insurance Research Database (TNHIRD) from the National Health Insurance Research Institute has been released for academic research use. The TNHIRD includes more than 26 files, such as the Registry for Beneficiaries, Ambulatory Care Expenditures by Visit and Details of Ambulatory Care Orders. On the other hand, South Korea has also released the Korea National Health Information Database (KNHID) which contains the records of all NHI data from the Korea National Health Insurance service (KNHIS). In addition, the KNHIS has been releasing Korea National Health Insurance-National Sample Cohort (KNHIS-NSC) data since 2014, which include qualification data, medical treatment databases, health examination databases and clinic databases. Other countries, such as Japan and the Scandinavian countries, have similar NHI databases; however, their sizes are small, and the cross-linkage among the registries is not sufficient in Japan and Scandinavia databases [7][8][9][10].
To date, Taiwan and South Korea are the only two countries that have released detailed NHI data to researchers. The use of largely complete NHI databases allows us to reduce concerns about selection effects. Larger sample sizes also give more precision for estimates. However, only a few studies are available to describe the TM utilization patterns between these two countries. Therefore, this study employed national population-based databases from both Taiwan and Korea to describe the pattern of TM utilization between the two countries. The aim of this study is to present the patterns of TM treatment, diseases treated with TM use and the reimbursement system of TM of the two countries, respectively.

Data sources
The first database used in this study is the Taiwan Longitudinal Health Insurance Database (LHID2000) released by the TNHIRD. A total of one million individuals were randomly selected from the Registry for beneficiaries in 2000, and all their claims information from 1996 to 2013 were recorded in the data labelled LHID2000. These are stratified sampling cohort data which represent Taiwanese NHI beneficiaries nationwide (23.75 million). The age, sex and economic level data distributions showed no significant differences between the sampling data and the entire TNHIRD. In this study, we used the Registry for Beneficiaries (ID), Registry for Contracted Medical Facilities (HOSB), ambulatory care expenditures by visits (CD), and details of ambulatory care orders (OO) files from LHID2000. We extracted the reimbursement claimed records in 2011 in this analysis.
The other data used in this study were retrieved from the National Sample Cohort 2002-2013 (NHIS-NSC 2002-2013) of the KNHIS in South Korea. This dataset also included approximately one million inhabitants who were randomly sampled in 2002, with no significant differences in age, sex and economic level compared to the entire population. In this study, the Qualification data, T120t data (Statement file of western medicine), T220t data (Statement file of dental and oriental medicine), T130t data (Details of treatment file of western medicine) and T230t data (Details of treatment file of dental and oriental medicine) in the medical treatment database and Clinic database in the National Health Insurance Service-National Sample Cohort 2002-2013 (NHIS-NSC 2002-2013) were used. We extracted the reimbursement claimed records in 2011 in this study.
All data used in this study were anonymized to protect patient privacy prior to release from the National Health Institute Research Institutes. This study received approval and certification from the Institutional Review Board (IRB) of Kyung Hee University in South Korea (Approval # KHSIRB-15-057 (EA) and the IRB of National Yang-Ming University in Taiwan (YM105098E).

Study process
The dataset from Taiwan (LHID2000) in this study was a randomly selected cohort from the Registry for Beneficiaries in 2000. Similarly, the dataset from South Korea (NHIS-NSC 2002-2013) used in this study was a randomly selected cohort from the Registry for Beneficiaries in 2002. The extracted reimbursement claimed records of the two countries showed that there were 1,025,340 and 970,866 beneficiaries in South Korea and Taiwan, respectively. The age and sex distribution of people selected from the 2011 ID files were not significantly different from the National Statistics of both countries. After excluding the records of the patients who visited dental clinics and the patients who did not use medical service covered by NHI, there were 909,646 and 779,337 people who had at least one claim data point for WM or TM in South Korea and Taiwan, respectively (Fig 1).
The subjects were categorized into three groups as follows: 1) WM-only users who only visited the WM institutes at least once in 2011, 2) TM-only users who sought only the TM service at least once in 2011, and 3) others who visited both WM institutes and TM institutes, and were categorized as WM & TM users. The TM users in this study included both TM-only users and WM & TM users, which encompassed all patients who received TM services at least once regardless of whether or not they used WM (Fig 1).
Demographic characteristics, including sex, age, and economic level, were analyzed for each group. Economic level was categorized into four groups according to monthly income (10 levels) in Taiwan and the level of insurance (10 levels) in South Korea: low economic level (a monthly income lower than 17,880 new Taiwan dollars [NTD] in Taiwan and an insurance amount lower than 23,980 Korean Won in South Korea), low-middle economic level (a monthly income ranging from 17,880 to 28,800 NTD in Taiwan and an insurance amount ranging from 23,980 to 28,590 Korean Won in South Korea), middle economic level (a monthly income ranging from 28,800 to 45,800 NTD in Taiwan and an insurance insured amount ranging from 28,590 to 41,730 Korean Won in South Korea), and high economic level (a monthly income over 45,800 NTD in Taiwan and an insurance amount over 41,730 Korean Won in South Korea).
Disease diagnoses were categorized according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) in Taiwan and the ICD-10 in South Korea.

Statistical analysis
Data in the text and tables were expressed as the frequency, mean ± standard deviation, median and range. Descriptive statistics were used in the patients' demographic characteristics, WM & TM use, and use of medical services. This study used the Chi-square test to examine the relationship between each variable and the differences among the WM-only, TM-only, and WM & TM users. To analyze the predictive variables of TM users, we used univariable and multivariable logistic regression, with TM use as the dependent variable and age, sex, and economic level as independent variables. Multivariable logistic regression models included sex, age, and economic level. All data were analyzed using SAS version 9.3 (TS1M2). Two tailed tests were used, and a P-value < 0.05 was considered to be significant.

Demographic characteristics of WM-only users, TM-only users and those who used both WM and TM
In total, 261,478 (25.5%) of the 1,025,340 people in South Korea and 260,529 (26.8%) of the 970,866 people in Taiwan used TM services at least once under the NHI in 2011. Table 1 shows the characteristics of the WM-only users, TM-only users, and WM & TM users in South Korea and Taiwan. In both countries, significant differences existed in sex, age and economic level between the WM-only users, TM-only users, and WM & TM users (all P < 0.001). Univariable and multivariable logistic regression analyses of the TM users (include both TMonly and WM & TM users) are listed in Table 2. Using univariable logistic regression, being female was associated with higher odds of TM use. A pattern of increasing odds for TM use with greater age, peaking at 61-80, but then declining for those aged over 80 in South Korea. There was evidence of decreasing use for high economic level in South Korea. On the other hand, being female was also associated with higher odds of TM use in Taiwan. A pattern of decreasing odds for TM use was seen with greater age since 21-40 onwards. There was Table 1

Utilization patterns of TM visits in South Korea and Taiwan
There was a mean of 5.7 ± 7.5 TM visits (median: 3, ranging from 1 to 137) among individuals who used TM services in Taiwan in 2011, which is lower than the mean of 8.9 ± 14.0 TM visits (median: 4, ranging from 1 to 892) in South Korea. The frequencies of the TM visits in both countries are shown in Fig 2. Most people had 1-3 TM visits in one year in both countries.  Taiwan   Table 4 and Table 5 shows the disease categories of TM use in South Korea and Taiwan. Among the TM visits in South Korea in 2011, the most common ailments were musculoskeletal system and connective tissue diseases, which accounted for 45.1% of the total TM visits in 2011, followed by diseases of the digestive system (18.2%) and injury, poisoning and certain other consequences of external causes (16.5%). Among the TM visits in Taiwan in 2011, the most common ailments were Symptoms, signs, and ill-defined conditions which accounted for 22.1% of the total TM visits, followed by respiratory system diseases (16.2%) and musculoskeletal system and connective tissue diseases (13.4%). Table 6 shows the prevalence of specific TM treatments used in South Korea and Taiwan. In South Korea, acupuncture accounted for 59.7%, followed by cupping (12.8%), moxibustion

Demographic characteristics of health care institution users
Private TM clinics in South Korea and Taiwan outnumber those in other countries where TM is being practiced. According to the National Statistical Bureau, 16 TM hospitals and 3,411 TM clinics were located across Taiwan in 2011 [11]. In South Korea, there were 168 TM hospitals and 12,061 TM clinics in the same year [6]. In the present study, the ratio of patients who visited TM hospitals to patients who visited TM clinics were approximately 1:10 in Taiwan and 1:47 in South Korea, implying that people visited TM clinics more than TM hospitals when seeking TM treatment in both countries. Compared to the ratio of TM hospitals/clinics (1:213 in Taiwan and 1:72 in South Korea), there were more TM hospital visits in Taiwan than in South Korea. Taiwan and South Korea have similar NHI systems. Both are single-payer systems and provide universal coverage. Furthermore, most WM and TM hospitals and WM and TM clinics collaborate with the NHI system in both countries. Outpatient visits of TM services either in primary care clinics or in hospitals are reimbursed by the NHI of South Korea and Taiwan. However, inpatient TM services in hospital were reimbursed only in South Korea, but not in Taiwan.
The dual WM and TM health care systems have a long history in both Taiwan and South Korea. Although WM is the leading official health care system at present, the statuses of WM doctors and TM doctors are similar in both countries [3]. Most people in both countries perceive WM as having faster effects and with a well-founded evidence base, whereas TM is thought to treat patients as a whole person by immediately treating not only the disease itself but also adjusting to the whole-body situation [12]. Thus, combining the strengths of WM and TM through integrative use might be more in line with patient preferences. Some hospitals have established joint WM and TM treatment centers. Unfortunately, medical insurance did not cover both WM and TM treatment on the same day for the same disease in South Korea until recently. In Taiwan, no strict provisions existed that prohibit a combined WM and TM treatment on the same day. Furthermore, to promote combined treatment, extra subsidies are available from the NHI when a patient uses combined WM and TM treatment for cancer and cerebrovascular diseases in Taiwan [13].
In terms of the distribution of the utilization of TM services in this study using multivariable logistic regression, the number of TM users increased with age in South Korea; in Taiwan, the prevalence of TM use increased from age 0-20 to age 21-40, and then decreased for older age groups. The results were consistent with the Korean Health Panel results from 2008 to 2009, and with another previous study from Taiwan [14]. In order to elucidate the effect of age, sex and economic level on TM use, we have performed three different logistic regression models. When we only included sex and economic level in the multivariable model, we found the direction of the effect of both variables were consistent with univariable analysis. However, if we stratified by sex further, we found the effects of economic level between male and female were in opposite direction. Thus, we speculated that sex is a modifying factor between economic level and TM use. When we included age and economic level in the model, we found the effects of the economic level were changed. Thus, we speculated that age may have modifying or confounding effects. We stratified by age further (data not shown), and found that age is also a modifying factor. According to the aforementioned results, we speculated that sex and age are effect modifiers between economic level and TM use.

Disease categories for TM visits in South Korea and Taiwan
Approximately half of the TM visitors in South Korea were classified in the musculoskeletal disease category (45.1%), followed by the digestive disease category (18.2%). A previous study indicated that more than half of the TM visits were for diagnoses of musculoskeletal system disorders in South Korea [3]. This is consistent with our study results. Compared with South Korea, the common disease categories for TM visits were not as concentrated in Taiwan. Symptoms, signs, and ill-defined conditions (22.1%), respiratory system diseases (16.2%) and musculoskeletal system and connective tissue disorders (13.4%) were ranked the top three most common diseases for which TM treatment was sought in Taiwan. The first nationwide survey of TM visits published in 2007 indicated that respiratory diseases represented the most common disease category for TM visits from 1996 to 2001 [14]. According to a previous study, which focused on a specific age group, the most common reason for TM visits in children was also respiratory diseases [15]. Yang et al. [16] have also shown that the most common reason for TM visits was musculoskeletal system diseases in elderly patients in Taiwan.
Acupuncture was the most common TM treatment type in South Korea. Acupuncture has been shown to have effects on relieving chronic pain and improving the conditions of musculoskeletal disorders [14]. The differences of common types of treatment used by TM doctors between the two countries (acupuncture in South Korea and powered herbal preparation in Taiwan) may be due to the differences of diseases diagnoses commonly treated in both countries. There were significantly more patients with musculoskeletal system disorders seeking TM service in South Korea than in Taiwan. South Korea and Taiwan have similar health care systems. Both the WM and TM health care systems are reimbursed by the NHI. Due to the differences of diseases commonly treated in TM in the two countries, there were limitations when performing a comparative study. Moreover, TM doctors need to use the ICD system that is designed for WM for insurance reimbursement. According to the ICD-11 beta draft, the newly designed ICD-11 includes a TM diagnosis section. The TM diagnosis section was established due to TM use in China, South Korea and Japan. The limitations of the disease systems classification may be improved when the ICD-11 system is put into practice [17].

Differences in reimbursed treatment and examination items by the NHI for TM treatment between the two countries
TM treatment and TM examination are reimbursed by NHI both in South Korea and Taiwan. Acupuncture, moxibustion, powered herbal preparation, electro-acupuncture and pulse diagnosis were all reimbursed in both countries. Cupping, hot/cold packs, meridian diagnosis and Yangdorak were only reimbursed in South Korea. However, manipulative therapy and tongue diagnoses were only reimbursed in Taiwan.
The most obvious differences in commonly used TM treatments in these two countries were in the most common procedures: acupuncture in South Korea and powdered herbal preparations in Taiwan. We also found that the rate of powdered herbal preparations use was much higher in Taiwan, perhaps because most TM doctors in Taiwan use powered herbal preparation for their primary treatment type rather than acupuncture, and for Korean TM doctors, the opposite is true. There were only 58 types of herbal formulas that were eligible for reimbursement in Korean NHI, but there were 337 types in Taiwan. Korean TM doctors might feel limited in their use of powered herbal preparations, and therefore, most of them choose to use acupuncture as their primary treatment type.
One of the important policies raised in the 3rd Foster Development Plan of Korean Medicine (2016-2020) was the expansion of the coverage range of Korean medicines in the NHI system. This policy called on scholars to perform more international comparative studies and evidencebased medicine studies to establish a more powerful background for the safety and effectiveness of Korean medicine [18]. Our prediction is that some changes will need to be made in the utilization of TM in South Korea in the coming years. Further research is also needed to follow up the clinical effectiveness of powdered herbal preparation use in both countries.
Among the countries where TM is a common part of the health system, South Korea and Taiwan are the only two countries that have released most of their NHI data for research studies. To date, nationwide studies have used the NHI database of a single country. This study was the first to present TM use in two countries using nationwide health insurance data. Our results provide a rough outline of the utilization patterns of TM. We believe that future studies can be performed using the framework of our study, while providing more detailed comparisons (e.g., changes in the process of the reimbursement system or the pattern of powered herbal preparation use) between the two countries. More comparative studies between different countries can be performed to discuss the use of TM among different races or ethnic groups. We believe that if the TM treatment can be proven to be beneficial in two or more races or ethnic groups through comprehensive NHI dataset, the results would be more convincing than those of an analysis performed in only one country.

Strengths and limitations of this study
To the best of our knowledge, this is the first cross-national study of utilization patterns of traditional medicine. Although our study has strength of large scale size and national population-based study, there are still some limitations in this study. First, the use of crude herbal drugs, an important TM treatment, could not be presented for both countries because they are not covered by the NHI. Second, because the insurance coverage was based on the use of the ICD diagnoses, we were forced to use the same system to classify patients in this study; however, the primary dialectic terms used by the TM doctors were the eight principles syndrome differentiation. Third, to maintain consistency between the two NHI databases in each country, we could only extract one year (2011) of data from the whole cohort database for the analysis. Beside, the data we used in this study was over six years old. However, there were no considerable policy changes in the NHI system in either country in recent years. Therefore, our data can still reflect the current situation. Fourth, except sex, age and economic level, there might be still some potential confounders such as education level and occupational status which were not included in this study, due to by lack of this information in the original NHI database. Detailed studies should be conducted by connecting the NHI database with other database such as The Korean Medical Use and Consumption Survey or National Health Pannel. Another limitation is that this study was conducted in two different countries, since the regulations of NHIRD could not allow the database to be transferred to other places. We could only analyze the data separately, and therefore restrictions remained for evaluating the country effect on TM use. For these reasons, this study may not have presented the entire pictures regarding differences in the use of TM services between the two countries.

Conclusion
Among the numerous countries that use TM, only South Korea and Taiwan provide detailed NHI data for academic purposes. According to the NHI database, about one fourth of the NHI beneficiaries of South Korea and Taiwan had TM use in 2011. Different TM utilization patterns existed between South Korea and Taiwan, which might be due to the differences in insurance coverage of the two countries.