2.1 Data Source

We obtained quarterly sales data for both the originator and biosimilar versions of adalimumab from October 2018 to October 2019 (four quarters) from the Multinational Integrated Data Analysis System (MIDAS)-IQVIA. The MIDAS-IQVIA database offers global sales information on medicines sold to retail and hospital pharmacies from manufacturers [47]. MIDAS-IQVIA data are based on collected sales volumes and generate the estimated sales values by multiplying the unit sold by prices from a variety of sources, including list prices, wholesaler transactions, and others [48, 49]. Previous studies have successfully utilized the MIDAS-IQVIA data for conducting cross-national and national studies, further validating its reliability and usefulness [37, 50,51,52,53].

We used the MIDAS-IQVIA database to obtain sales values in USD (converted from local currencies) and sales volumes measured using a standard unit (SU) of 40 mg per syringe, pen, or vial for adalimumab. The SU is defined as the number of standard ‘dose’ units sold by IQVIA. Multiple studies have been reported using MIDAS-IQVIA pertaining to SU to analyze medicine sales trends, including biologics [37, 50, 51, 53, 54]. Adalimumab 40 mg is considered the defined daily dose (DDD) for this medication, as established by the World Health Organization (WHO) [55]. Adalimumab is available in different dosage forms, such as 20 mg, 40 mg, and 80 mg. Among these options, adalimumab 40 mg accounted for approximately 97% of utilization in terms of SU in the fourth quarter (Q4) of 2018 across all the included countries. This percentage ranged from 91% in Japan to 100% in eight other countries. Also, adalimumab 40 mg has been used in previous studies to examine the dynamics of drug expenditure [42, 56, 57], suggesting the predominant use of the 40 mg dose of adalimumab and highlighting the consistency of its utilization across all included countries (Supplementary Table S1; see the ‘Supplementary Information’ in the electronic supplementary material).

2.2 Selection of the Biologics and the Countries

To conduct a cross-country comparison of adalimumab market dynamics, we identified biologics whose biosimilar had been launched, with varying availability across jurisdictions. Based on global sales values, we referred to TNF-α inhibitors (adalimumab, etanercept, and infliximab) and the launch dates of their respective biosimilars. The international launch dates for biosimilar etanercept and infliximab were relatively close, thus making it challenging to select a study period in which countries with and without biosimilars were evenly distributed. Consequently, adalimumab was the only biologic study drug whose originator (Humira®) ranked first in terms of the global biologics sales in 2018 [6].

We sought to include countries that encompassed geographical, economic, and demographic diversity: six European countries (Austria, France, Germany, Italy, Spain, and Sweden), five Asian Pacific countries (Australia, Japan, Korea, Singapore, and Taiwan), one North American (Canada), one Latin American (Brazil), and one African (South Africa) country. All of these countries are members of major international economic organizations, such as the Organisation for Economic Co-operation and Development (OECD), Asia-Pacific Economic Cooperation (APEC), and the Group of Twenty (G20). All included countries were classified based on the World Bank’s classification of economic level in 2019: (1) upper-middle-income economies (UMIEs) (Brazil and South Africa) and (2) high-income economies (HIEs) (Australia, Austria, Canada, France, Germany, Italy, Spain, Sweden, Japan, Korea, Singapore, and Taiwan) [58]. All included countries had already launched the adalimumab originator before the study period and maintained availability throughout the study period (from October [Q4] 2018 to October [Q4] 2019). If any biosimilar adalimumab was available throughout the study period for the specific country, it was classified as ‘yes’ for that country and ‘no’ otherwise. In conclusion, we included Australia, Brazil, Canada, Japan, Korea, Singapore, Taiwan, and South Africa for the control group.

Since a major indication for adalimumab is RA and RA activity can be influenced by seasonal changes [59, 60], we analyzed a minimum of four quarters for equal seasonal comparisons (Q4 2018 vs. Q4 2019). Although it would be ideal to analyze a longer time frame, our data source limited us to the available quarters. Fortunately, we were able to easily analyze seasonality due to the simultaneous launch of adalimumab biosimilars in all countries where they were available, which occurred on October 16, 2018 (Supplementary Table S2; see the ‘Supplementary Information’ in the electronic supplementary material) [12].

2.3 Descriptive Analysis2.3.1 Time Series Analysis

To examine whether the cost savings in the healthcare system were associated with the launch of adalimumab biosimilars, we compared the adalimumab expenditure and consumption by each country in Q4 2018 and Q4 2019. The financial impact of the biosimilar adalimumab was described by (1) the sales values (in USD) and (2) the sales volumes (in SU) of the adalimumab originator and its biosimilar using the MIDAS-IQVIA database.

However, due to significant variations in the absolute values of sales values and volumes among all included countries, direct comparisons were challenging. To address this issue, we adjusted for economic differences and population size. Namely, the quarterly sales value (representing adalimumab expenditure) per the gross domestic product (GDP) per capita was calculated to adjust for the economic differences of all included countries. Also, to account for the difference in population size, we calculated the quarterly sales volume per a 1000 people, as previously conducted in cross-country comparison studies [50, 61,62,63,64,65]. GDP per capita data and population figures of all countries, except Taiwan, were obtained from the World Bank and OECD, respectively [66, 67]. For Taiwan, National Statistics open data and websites that released global countries’ populations were referred to [68, 69]. Using these adjusted measures, we compared the expenditure and volume of adalimumab among countries where the biosimilars were available versus those where they were not.

2.3.2 Decomposition of Adalimumab Expenditure

To quantify the impact of adalimumab biosimilars on adalimumab spending, we compared price, quantity, and drug mix between the countries where adalimumab biosimilars were available and those where they were not, which are the components of changes in drug expenditures. The changes in adalimumab expenditure from Q4 2018 to Q4 2019 can be expressed as in Eq. (1), which was developed by Gerdtham et al. [43] and Gerdtham and Lundin [44, 70].

$$} = \frac}_ }_ }}}_ }_ }} = }_}} \times }_}} \times \varepsilon_}} = \frac}_ }_ }}}_ }_ }} \times \frac}_ }}}_ }} \times \frac}_ }_ \sum }_ } \right)}}}_ }_ \sum }_ } \right)}}$$

(1)

In this Eq. (1), Q0 and Q1 represent the quantities of adalimumab (in SU) at Q4 2018 and after four quarters (Q4 2019), respectively. P0 and P1 represent the prices of adalimumab (in USD) at Q4 2018 and after four quarters (Q4 2019), respectively. The prices were calculated by dividing the expenditure of adalimumab (in USD) by the quantity of adalimumab (in SU), which are the ex-factory prices. Pt represents the change in the price of adalimumab, which is calculated using the Laspeyres index. Qt is the change in the quantity of adalimumab, which captures the treatment expansion effect. εt is the drug-mix effect, which captures a shift in prescription patterns towards cheaper alternatives.

If any of these indices are less than 1 among countries where biosimilars are available, it indicates that the presence of adalimumab biosimilars contributed positively to the decrease in drug expenditure by reducing prices through competition, decreasing the demand for adalimumab, or the switch to more affordable alternatives.

2.4 Statistical Analysis

To examine the changes in adalimumab expenditure and utilization, we employed the difference-in-difference (DID) method, one of the most widely used study designs for evaluating the impact in various fields, including healthcare, by comparing treatment–control group and before–after difference [71,72,73,74]. We compared the two groups, one with adalimumab biosimilars present and the other with biosimilars absent, from Q4 2018 and after four quarters (Q4 2019).

We defined countries without adalimumab biosimilars as the control group and countries with adalimumab biosimilars as the treatment group, at the beginning of the study period (Q4 2018) and after four quarters (Q4 2019).

Our DID outcomes Y were modelled using the following regression framework:

$$Y = \beta_ + \beta_ \cdot post + \beta_ \cdot treatment + \beta_ \cdot post \cdot treatment + \varepsilon$$

(2)

where Y represents the absolute changes in adalimumab expenditure and utilization; ‘treatment’ is a dummy variable, which is equal to 1 for countries where adalimumab biosimilars were present and 0 otherwise; and ‘post’ is a dummy variable, which is equal to 1 for the period after four quarters (Q4 2019) and 0 otherwise.

In this Eq. (2), β0 captures the intercept. β1 captures the time trend between the entry of adalimumab biosimilars (Q4 2018) and after four quarters (Q4 2019) for those of the control group (countries without adalimumab biosimilars). β2 measures the treated group effect between the control group (countries without adalimumab biosimilars) and the treatment group (countries with adalimumab biosimilars) at the beginning of the study (Q4 2018). The coefficient of interest in our study is β3, which captures the changes in adalimumab expenditure and utilization for countries with adalimumab biosimilars compared to the control group after four quarters, as a result of adalimumab biosimilars’ benefit.

The outcomes with a p value of less than 0.05 were deemed statistically significant. Statistical analyses were done with SAS 9.4 (SAS Institute, Cary, NC) using PROC MIXED.

2.5 Sensitivity Analysis

It has been reported that drug consumption is associated with GDP [50], and we recognized the importance of controlling for economic differences. Therefore, in the sensitivity analysis, we narrowed our focus to HIEs only, excluding UMIEs (Brazil and South Africa), to reduce the potential confounding effects of GDP variations.