Trends in pharmaceutical expenditure in the Taiwan National Health Insurance database at different hospital levels

In all the original claims drug prescription data for all beneficiaries, the annual pharmaceutical expenditures from 2016 to 2018 in these 14 types increased from 2031.6 million to 2311.6 million USD in medical centers, from 1519.2 million to 1729.8 million USD in regional hospitals, from 473.1 million to 589.8 million in district hospitals and from 440.1 million to 452.7 million USD in clinics (Table 1). The average expenditure per hospital from 2016 to 2018 increased from 19.7 million to 22.5 million USD in regional hospitals and from 1.5 million to 1.9 million in district hospitals, but remained the same in medical centers and clinics over the period. We further divided the annual pharmaceutical expenditures among the different hospital levels into 14 categories (gastrointestinal, diabetes, nutritional supplement, blood disorders, cardiac and hypertension, urinary tract and sex hormones, systemic Hormones, anti-systemic infection, anti-tumor and immunologic, anti-inflammatory and anti-rheumatic, nervous system, respiratory system, ophthalmological and ontological and others) (Table 1). In medical centers, the anti-tumor and immunologic (679.6–813.5 million/year) showed the highest drug expenditure, followed by anti-systemic infections (317.2–396.6 million/year), blood disorders (236.9–243.7 million/year) and cardiac and hypertension (181.7–190.5 million/year). The drug prescription pattern in regional hospitals was similar to that in medical centers; the anti-tumor and immunologic (353.4–423.7 million/year) showed the highest drug expenditure, followed by Anti-systemic infection (233.4–372.7 million/year), cardiac and hypertension (205.7–207.7 million/year) and nervous system (182.6–187.1 million/year). The majority of the drug expenditure in district hospitals was cardiac and hypertension (96.3–101.9 million/year), followed by nervous system (78.4–86.4 million/year), anti-systemic infection (59.6–90.1 million/year) and diabetes (54.3–64.9 million/year). In clinics, cardiac and hypertension (149.1–156.2 million/year) showed the highest drug expenditure, followed by diabetes (61.4–67.9 million/year), nervous system (57.4–60.2 million/year) and respiratory system (46.8–48.3 million/year).

To understand the detailed patterns of drug prescription among different drug categories, we further divided the drug group into pharmacological subgroup (3rd level) according to the ATC code. The top 10 drug subgroups are shown in Figures 1–4. In medical centers, L01X (other antineoplastic agents) (311.6–389.3 million/year) showed the highest drug expenditure, followed by L04A (immunosuppressants) (186.6–224.6 million/year), J05A (direct acting antivirals) (111.7–181.4 million/year) and A16A (other alimentary tract and metabolism products) (113.8–135.3 million/year) (Figure 1). The drug prescription pattern in the regional hospitals was the same as that in the medical centers. L01X (other antineoplastic agents) (169.4–210.2 million/year) showed the highest drug expenditure, followed by J05A (direct acting antivirals) (77.6–170.5 million/year), L04A (immunosuppressants) (69.7–90.9 million/year) and A10B (blood glucose-lowering drugs, excl. insulin) (75.6–85.3 million/year) (Figure 2). The majority of the drug expenditure in district hospital was A10B (blood glucose-lowering drugs, excl. insulin) (35.6–47.9 million/year), followed by N05A (Antipsychotics) (33.8–35.0 million/year), J05A (direct acting antivirals) (8.5–33.9 million/year) and C10A (lipid-modifying agents, plain) (26.9–30.6 million/year) (Figure 3). In clinics, A10B (blood glucose-lowering drugs, excl. insulin) (51.4–56.6 million/year) showed the highest drug expenditure, followed by C10A (lipid-modifying agents, plain) (49.5–49.11 million/year), C09D (angiotensin II antagonists, combinations) (23.9–29.6 million/year) and C09C (angiotensin II antagonists, plain) (24.2–27.2 million/year) (Figure 4). Table 2 shown the ascending order and cost difference of drug expenditures for each medical facility type.

We checked the brand and generic drug prescription in the highest drug expenditure item among the top ten drug subgroups. In the medical centers, brand coagulation factor VIII (B02BD02) (44.5–49.6 million/year) showed the highest drug expenditure, followed by brand trastuzumab (L01XC03) (36.2–42.2 million/year), brand sofosbuvir/ledipasvir (J05AP51) (39.8 million/year) and brand agalsidase alfa (A16AB03) (36.9–39.5 million/year). The generic drug usage in these high-expenditure items among the top 10 drug subgroups was 3.89% in 2018 (3.18% in 2017 and 3.19% in 2016) (Table 3). The highest drug expenditure in the regional hospitals was brand sofosbuvir/ledipasvir (J05AP51) (54.8 million/year), followed by brand trastuzumab (L01XC03) (36.6–37.4 million/year), brand clopidogrel (B01AC04) (19.1–19.7 million/year) and brand pemetrexed (L01BA04) (15.8–16.6 million/year). The generic drug usage in these high drug expenditure items among the top 10 drug subgroups was 11.04% in 2018 (13.94% in 2017 and 12.88% in 2016). In the district hospitals, brand sofosbuvir/ledipasvir (J05AP51) (16.1 million/year) showed the highest drug expenditure, followed by brand valsartan and amlodipine (C09DB01) (9.2–10.4 million/year), brand coagulation factor VIII (B02BD02) (10.0 million/year) and generic rosuvastatin (C10AA07) (9.7–9.9 million/year). The generic drug usage in these highest drug expenditure items among the top 10 drug subgroups was 32.35% in 2018 (45.50% in 2017; 47.51% in 2016), which was significantly higher than that in medical centers and regional hospitals. In clinics, generic valsartan and amlodipine (C09DB01) (5.5–40.2 million/year) showed the highest drug expenditure, followed by generic atorvastatin (C10AA05) (20.9–23.5 million/year), generic amlodipine (C08CA01) (13.2–17.9 million/year) and brand valsartan/amlodipine (C09DB01) (7.3-9.4 million/year). More than half of these highest drug expenditure items among the top 10 drug subgroups in clinics were generics (78.14% in 2018, 75.67% in 2017 and 77.21% in 2016).

This study is one of the few retrospective studies examining the medical expenditure among different hospitals and drug categories using a national sample in a non-Western country. Using a large nationwide health insurance database, we discerned that medical expenditure in Taiwan increased over time in all hospitals and that it was higher in high-level hospitals (Table 1). Anti-tumor and immunologic agents showed the highest medical expenditure in medical centers and regional hospitals. Among these agents, brand trastuzumab was associated with top medical expenditures in the medical centers and regional hospitals. Cardiac and hypertension agents showed the highest medical expenditure in district hospitals and clinics. Among these agents, genetic valsartan/amlodipine was associated with top 1 medical expenditure in district hospitals and clinics.

The Ministry of Health and Welfare approved higher reimbursement service fees and incentives to encourage medical centers and regional hospitals to provide services for severe illness and emergency care [3]. This study revealed that medical centers as well as regional hospitals spend more on anti-tumor/immunologic and anti-systemic infection agents (Table 1), both of which belong to severe illness and emergency care. The main objectives of the district hospitals and clinics are chronic disease treatment and health consultations/referrals. We found that district hospitals use glucose-lowering drugs and treat psychosis with antipsychotics (Figure 3). In contrast, clinics mostly use bold glucose-lowering drugs and treat dyslipidemia with lipid-modifying agents (Figure 4). These drugs help in managing chronic diseases, and if patients need more aggressive treatment, they can be referred to higher-level hospitals.

Cost is a leading policy challenge in the healthcare system. The high proportion of NHI expenses on drugs was identified as a problem in Taiwan's healthcare system [18]. Product price control, reference pricing and profit control are frequently used to control the drug expenditure [5,6,19]. The average annual drug expenditure per capita in Taiwan has been increasing over time (189.6, 199.5 and 215.5 USD in 2016, 2017 and 2018, respectively), which is similar to that in China (187.6 USD in 2016) [20,21]. Although the average annual drug expenditure per capita is similar to that in the neighbouring countries, the proportion of drug expenditure in some categories is still higher than what the government expected. Considering anti-tumor and immunologic drug as an example; the total medical and drug expenditures in cancer treatment was 2298.0–2704.5 and 1072.4–1292.8 million/year, respectively, in 2016–2018. The drug expenditure among all cancer treatment medical expenditures was 46.67–47.80%. This result indicates that drug expenditure reduction may be beneficial to the entire NHI system. In Taiwan, the NHI authorities low-priced drug items in the market for cost reduction, and most of them are generic. The generic drugs used in the top 10 categories of prescription drugs in medical centers and regional hospitals were only 3.19-3.89% and 11.04-13.94% during 2016–2018 (Table 3). These results indicate that product price control for low-priced drugs is of limited help to curtail the drug expenditure. Paying attention to expensive drugs with the intention of cost control may be the way to reduce the drug expenditure in higher-level hospitals. In another way, the generic drugs used in the top 10 categories of prescription drugs in district hospitals and clinics were significantly higher than those used in medical centers/regional hospitals during 2016–2018 (32.35–47.51% in district hospitals and 75.67–78.14% in clinics). As these generics offer cost control benefits, BNHI should provide incentives to encourage pharmaceuticals to sustain generic drug development.

Although the analyses of the national health insurance research database (NHIRD) have provided several benefits, such data are still fraught with limitations owing to the nature of the NHIRD design. First, being based on a claim data sets, the present study lacked self-pay medication information in detailed patient medical records. Another limitation is the time lag in obtaining the NHIRD since NHI needs nearly 1 year to update the claims database; hence, the most updated information was not available. Taiwan national claim data sets are available through formal application to Applied Health Research Data Integration Service provided by the BNHI of Taiwan by provide detail research proposal. We try to find the prescription pattern in Taiwan for both inpatient and outpatient and not distinguish them in the beginning of our study design. It will be difficult to separate inpatient and outpatient prescription in this study. According to previous Taiwan national health insurance statistic, outpatient prescription was much higher than inpatient prescription (6.07-fold in 2018, 5.94-fold in 2017 and 6.67-fold in 2016) (https://www.mohw.gov.tw/dl-72320-5da2775d-2c64-418d-b293-1f1ee56dea6c.html). Results in this study were mainly outpatient prescription. Due to outpatient prescription and inpatient prescription visit may dramatic decreased during COVID-19 pandemic, we will try to analysis the inpatient and outpatient prescription separately between year 2019 to 2021 (before and after COVID-19 pandemic).

In this study, we found that the annual pharmaceutical expenditures from 2016 to 2018 were increasing over time in all hospitals. The average expenditure per hospital depicted an increasing trend in regional and district hospitals but remained similar in medical centers and clinics from 2016 to 2018. Among the categories of drug prescription, anti-tumor and immunologic showed the highest drug expenditures in medical centers/regional hospitals. In contrast, the top 1 category of drug prescription was cardiac and hypertension in district hospitals and clinics. The dominant drug expenditure used for top 10 prescribed generic drugs used were higher in district hospitals/clinics than in medical centers/regional hospitals from 2016 to 2018.

This study is based on data from the Applied Health Research Data Integration Service provided by the BNHI. The interpretation and conclusions contained herein do not represent those of the BNHI or Ministry of Health and Welfare.

This work was supported by the Mackay Medical College (Sanzhi, Taiwan; grant numbers MMC-RD-109-CF-G1-01, MMC-RD-110-1B-P014, MMC-RD-110-CF-G001-01, MMC-RD-111-1B-P014 and MMC-RD-111-CF-G001-01). The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

The Mackay Medical College entrusted the MacKay Memorial Hospital Institutional Review Board, Taiwan, to review this clinical study. The MacKay Memorial Hospital Institutional Review Board Taiwan, R.O.C., approved the study protocol (protocol no. 19MMHIS188e)

The data underlying this study belong to the Applied Health Research Data Integration Service provided by the BNHI of Taiwan and cannot be made publicly available due to legal restrictions. However, the data are available through formal application to the Applied Health Research Data Integration Service provided by the BNHI of Taiwan and require a signed affirmation regarding data confidentiality. The authors have no special privilege of access to the database.

This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/