Topics Innovativeness of New Drugs and Price Premiums An Analysis Using Matched Samples from Japan, the U.S., and Germany

Printable PDF

Junichi Nishimura, Visiting Fellow, Pharmaceutical and Industrial Policy Research Institute, Professor, Gakushuin University
Sadao Nagaoka, Director, Pharmaceutical Industry Policy Institute Professor, Tokyo Keizai University

The drug price system is expected to play a central role in incentives for drug discovery innovation. This is because the price premium of new drugs relative to comparator drugs is expected to reflect the innovativeness of new drugs, thereby creating more innovative drugs. In this paper, we construct a new measure of innovativeness of new drugs and analyze how the Japanese drug pricing system has evaluated innovativeness relative to the U.S. and Germany, using matched data on the price premium of new drugs relative to comparator ^drugs1). Objective analysis of the relationship between price premiums and innovativeness of new drugs from an international perspective is important for considering the design of future systems.

In a previous study by Comanor et al. (2018 ⁾²⁾, they compared the price premiums of new drugs in the United Kingdom, where the cost-effectiveness of drugs by NICE and public procurement agencies play a major role in price determination, and the United States, where the evaluation of drug effectiveness and price negotiations are left to private agencies, and found that on average the two are almost They concluded that, on average, the two countries' price premiums were about the same and that pricing in the U.K. did not result in a significant difference from market-based pricing. On the other hand, "Price premiums for new drugs relative to comparators: an analysis using matched samples from Japan, the U.S., and Europe" (Policy Research Institute News No. 62) verified and extended this Comanor et al. (2018) study, finding that the level and variation of price premiums were smaller in Japan and Europe than in the U.S., but the factors behind this not examined. Furthermore, the preceding analysis assumes that the price premium for new drugs in the U.S. more effectively reflects the innovative nature of drugs, but does not examine this assumption. This paper will address these prior research issues. Note that we are currently unable to find any prior literature that evaluates differences in international price premiums depending on the innovativeness of drugs.

Price Data

This paper builds on the data set developed in "Price Premiums of New Drugs relative to Comparators: Analysis of Matched Samples in Japan, the U.S., and Europe" (Policy Research Institute News No. 62), and additionally collects data on innovativeness to construct the analytical data set. The analysis used IQVIA Pricing Insights to compile quarterly drug price information (substantiated by 2015 prices) for drugs launched between July 2010 and March 2019 in the five countries of Japan, the United States, Germany, the United Kingdom, and France, with generic ingredient names, international trade names, dosage forms, and specification units all being the same in each country. The analysis is based on the same methodology as described in the previous section. As in the previous analysis, this paper focuses on the top-selling drugs, with a population of 69 components for which price premium data are available. However, the actual number of observations in the analysis is less than 69 components due to deficiencies in the data collection for the innovativeness index and the patent protection period of drugs, which will be discussed later. A list of the data used and details of the procedure for constructing the analytical data are summarized in Appendix 1.

Price Premium Definition and Trends

In order to calculate the price ratio of a new drug to a comparator drug on a per treatment basis, this paper used information on the daily drug price of each drug from the drug price calculation data of the Central Social Insurance Medical Council (Chuikyo) to determine the price premium (logarithmic value of the price ratio, the rate of price increase of a new drug relative to a comparator drug) for new drugs in Japan, the US, and Germany.

Price premium = ln (daily drug price of new drug) - ln (daily drug price of comparator drug)

If the prices of both new drugs and comparative drugs are set to reflect their health benefits (based on ICER criteria), then this price premium would reflect the health improvement effect of the new drug relative to the comparative drug (see Policy Research Institute News No. 62). In addition, as can be seen from the definition formula, since the price difference is taken as a logarithm, intrinsic effects common to both the new drug and the comparator (e.g., indications, pharmacological effects, demand for the target disease, market competition, etc.) are eliminated.

Figure 1 shows the trend of this price premium in the U.S. In preparing the figure, the drug components used are limited to those for which price information is available for all new drugs and comparator drugs in the country and for which price premiums can be measured for all time periods. Therefore, the purpose of the study is to provide an overview of price trends in each country, and the composition of the sample is different from that of the Japan-U.S.-Germany common sample used in the later estimation.

The purpose of the study is to analyze how price premiums reflect innovativeness, and the analysis focuses on price premiums during periods of exclusivity protection for both new drugs and comparator drugs. Once the exclusivity protection period for a comparative drug ends, the drug price becomes much less relevant to the efficacy of the drug due to competition from generic drugs, etc., and the price premium reflects this. On the other hand, focusing only on initial drug prices does not allow for an appropriate assessment of the changes in drug prices over time in the degree to which drug prices reflect innovation in each country, such as changes in drug prices depending on market evaluation of therapeutic efficacy and the impact of drug price regulations on new drugs. In the following, we review trends in drug prices and price premiums for new drugs and comparative drugs in the U.S. in order to confirm the appropriateness of the analytical framework in which price premiums during the period of exclusivity protection for both new drugs and comparative drugs are subject to analysis.

Figure 1 shows the trend of price premiums for 37 drug components in the U.S. (left axis), as well as the prices of the new drugs and their corresponding comparator drugs (right axis). The average price premium in the U.S. tends to be around 0.5 before the exclusivity protection period of the comparator drug. This means that a new drug has a price premium of about 50% compared to the comparator drug. At the median, the price premium tends to increase over time from 0.2 to around 0.6, suggesting that there are large differences in price premium trends among drugs. Regarding the trends in drug prices for new drugs and comparative drugs, both prices of comparative drugs were on an increasing trend over time before the period of monopoly protection for new drugs, but the extent of the increase was greater for new drugs. For new drugs, prices may rise as their value is gradually recognized in the market. As for comparative drugs, patent protection expires around 2015, but it is also characteristic that in the U.S., the increase in drug prices is rather large after the expiration. Thus, it was confirmed that patent protection of comparator drugs has a significant impact on the price premium of new drugs, especially in the United States. In the following, we construct an index of innovativeness, focus on the price premiums of new drugs and comparator drugs during the patent protection period using econometric methods, and analyze its determinants in a matched sample of Japanese, U.S., and German pharmaceutical ingredients.

Indicators of Innovativeness of New Drugs

Four new drug innovativeness indices are newly developed in this paper. The first indicator was created based on the Chuikyo's "Drug Classification for Selection of Similar Drugs " ³⁾.

Next, a group of patents (substance patents, crystal patents, and use patents) protecting each pharmaceutical ingredient was identified from the SANEI Report, and a group of scientific papers that were cited in the bibliographic information (front page) of the corresponding U.S. patent group was collected from Web of Science to create the following index of innovativeness. Since citation information is inherently subject to truncation problems and the propensity of pharmaceutical patents to cite scientific and technical papers changes over time, when using the following indicators for estimation, we control for the ATC drug effect area and year of launch of the new drug as fixed effects.

The validity of the innovativeness index used in this paper is discussed from the following three points. First, it is assumed that drugs with novel mechanisms of action are based on new scientific knowledge and have small citation lags as described above, while drugs with existing mechanisms of action have long citation lags. In fact, there was a positive correlation (0.271, P value = 0.095) between innovativeness (novelty) and innovativeness (speed of science utilization) (47 components, from cross-section data).

Second, it is assumed that the higher the level of innovativeness, the stronger the competition, and the higher the rate of drug price increase (initial value1) over time for a new drug. For example, in the U.S., there is a positive correlation (0.048, P value = 0.078) between the rate of increase in drug prices and innovativeness (novelty), a positive correlation (0.053, P value = 0.052) between the rate of increase in drug prices and innovativeness (number of scientific papers), a positive correlation (0.070, P value = 0.015) between the rate of increase in drug prices and innovativeness (number of citations), a positive correlation and innovativeness (speed of science utilization) were positively correlated (0.071, P-value = 0.012) (43 components, 1240 observations, from panel data).

Third, in Policy Research Institute News No. 55, drug components that cited scientific papers in their patents were more likely to contribute to the cure rate of hospitalized patients, and were also positively correlated with the drug contribution as judged by physicians. It is assumed that pharmaceuticals that make greater use of scientific knowledge have a higher economic impact. Based on the above, we believe that the innovativeness index used in this paper reflects, even partially, the innovativeness of pharmaceuticals. The degree of innovativeness of a drug that cannot be captured in this paper (e.g., ease of administration, degree of side effects, etc.) is treated as a missing variable in the model. Details are explained in the theoretical background in Appendix 2.

Analytical Model

To analyze the relationship between the innovativeness of a new drug and the price premium calculated from the drug prices of its comparator drugs, panel data are used. The theoretical background of the following estimation model is summarized in Appendix 2. The following model (1) is used to estimate the relationship between innovativeness and price premiums in Japan, the U.S., and Germany.

Price premium in Japan, the U.S., and Germany = α1 + β1 (innovativeness) + control (1)

The observation unit is the pharmaceutical component, which is quarterly data. The observation period of price data is a little more than eight years, and some drugs have no initial price observed. Therefore, by introducing the number of years since the launch of a new drug as a control variable, we estimate the price premium for each year in terms of the initial price. At the same time, the number of years since launch is removed from the estimation equation, and the price data observable during the exclusivity protection period of the comparison drug is used in the estimation. In this case, the estimation results are based on the average price during the exclusivity protection period.

The four aforementioned indicators are simultaneously included in the model as innovativeness variables. In consideration of the multicollinearity problem among the four indicators, the variables were replaced step by step, but the coefficient values did not change significantly. As control variables, we also included year dummies, quarterly dummies, new drug launch year dummies, and ATC drug effect category dummies. By including a cohort effect through the new drug launch year dummies, we take into account the tendency for recent new drugs to have smaller citation numbers in the papers they are cited in. The least squares method is used in the estimation. In addition, all new drugs analyzed in this paper are priced using price data prior to the expiration of patent protection in each country. As already mentioned, in order to analyze the impact of innovativeness of new drugs on price premiums, the efficacy of new drugs and comparative drugs must be appropriately reflected in their prices, and therefore, only data before patent expiration were used for both new drugs and comparative drugs in the estimation. Furthermore, for the analysis based on the matched sample of Japan, the U.S., and Germany, we used 30 components of drugs for which data exist for both new drugs and comparative drugs in common in the three countries.

Next, we estimated the difference between the U.S. price premium and that of each country as the explained variable, using the U.S. price premium as the standard. Model (2) is as follows.

U.S. price premium - Japanese and German price premiums = α2 + β2 (innovativeness) + control (2)

In this estimation, the effect of innovation on the price premium can be measured without bias because the intrinsic effect specific to the combination of new drugs and comparator drugs (which cannot be captured by the innovativeness indicator) can be removed by taking the international difference. As control variables, we include the number of years since the launch of a new drug in the U.S. and Japan/Germany, year dummies, quarterly dummies, new drug launch year dummies, and ATC drug effect category dummies.

We discuss the possibility of bias in our estimation model. The new drugs in our analysis are the top 300 drugs in terms of global sales, are highly successful drugs, and are expected to have generally high price premiums. As a result, drugs are selected at the level of the price premium, and this bias acts to reduce (underestimate) the estimated value of the coefficient of innovativeness in the price premium equation.

This problem becomes particularly acute when scientific papers play an important role but lack citation information. In this case, our constructed innovativeness measure would be zero, but since it is a top 300 drug and yields a high price premium, there is a risk that the coefficient of innovativeness is substantially underestimated. To avoid this, we exclude from our estimation drugs for which the identified U.S. patents do not cite any scientific papers (in developing our measure of speed of scientific utilization, drugs with zero scientific papers are necessarily eliminated) ⁴⁾.

Estimation results: innovativeness and price premium

Table 1 shows the results of the estimation of innovativeness and price premiums for model (1) for each country. Equations (1) through (3) include the number of years since the launch of a new drug in the model, and look at the effect of innovativeness on the initial price premium after eliminating the effect of elapsed years. Equations (4) through (6) exclude the number of years after the launch of a new drug, which confirms the robustness of the model and estimates the average relationship between price premium and innovativeness over the entire observation period (life cycle). First, Table 1 shows that the innovativeness indices (novelty, number of scientific papers, number of citations, and speed of scientific utilization) are all positive and significant in all three countries (Japan, the U.S., and Germany). The coefficient value for the U.S. is the largest. For example, since the estimation model is a two-log model, a 10% increase in the number of citations of scientific papers on which the new drug relies or a 10% increase in the speed of utilization of the results of scientific papers in equation (2) means that the price premium of the new drug relative to the comparator drug in the U.S. would, on average, be 3.7% and 2.4% higher, respectively. higher. Looking at the coefficient values, the coefficients for the U.S. are the largest except for the degree of novelty, and the differences are also large. These results indicate that in the U.S. market, the innovativeness of a new drug is best reflected in its price premium.

Second, the coefficients for the elapsed time since the launch of a new drug are not statistically significant in any of the countries. Equations (1) through (3) and (4) through (6), which exclude the number of years since launch, show no significant change in the coefficient value of innovativeness.

Table 2 shows the results of the estimation of international differences in innovation and price premiums in model (2). Equation (1) is estimated by subtracting the Japanese price premium from the U.S. price premium, and equation (2) is estimated by subtracting Germany from the U.S. The estimated coefficient values indicate the extent to which the innovativeness index reflects smaller in Japan and Germany compared to the U.S. Similar to Table 1, the estimation excluding the number of years elapsed since the launch of a new drug is also done using equations (3) and (4). As mentioned earlier, international differencing reduces the bias in the estimates.

Looking at the coefficients of the innovativeness indices, the number of scientific papers on which pharmaceutical patents rely and the speed of scientific utilization are positive and significant in all equations. In Table 1, we have already pointed out that the coefficients of these indicators are smaller in Japan and Germany than in the U.S. However, the difference-in-differences estimation shows that the degree to which innovativeness is reflected in the price premium is significantly lower in Japan and Germany than in the U.S., even after controlling for estimation bias. Furthermore, the results of equations (3) and (4), which exclude the number of years elapsed since the launch of a new drug, show that the coefficient values of the innovativeness indicators (number of scientific papers and speed of scientific utilization) tend to be larger than in equations (1) and (2). This suggests that innovation is more highly valued in the U.S. than in Japan and Germany during the life cycle of a pharmaceutical product, and this is reflected in the price premium in the U.S. Although such a trend was not observed in Table 1, it is thought to be an estimation bias due to the missing variable in Model (1).

While Tables 1 and 2 were estimated using individual innovativeness indicators, Tables 3 and 4 create a composite indicator of the four innovativeness indicators and perform the same estimation. The composite indicator is obtained by multiplying the coefficient value of the U.S. innovativeness indicator in equation (2) of Table 1 by the respective innovativeness indicator. This takes into account that the innovativeness index is most reflected in price premiums in the U.S. market. The basic results are the same when a composite indicator is constructed using the average of the coefficient values of the innovativeness indicators in Japan, the U.S., and Germany.

The estimated results in Tables 3 and 4 are consistent with those in Tables 1 and 2. The innovativeness (synthetic indicator) is positive and significant in all three countries (Japan, the U.S., and Germany) and also positive and significant in the difference model. In the difference model, the coefficient value is also larger when the number of years since the launch of a new drug is excluded. The large coefficient values also indicate that innovativeness is less likely to be fully reflected in the price premium in Japan compared to Germany. Since the difference between the estimated values of the U.S. and Japanese composite indicators in equations (4) and (5) in Table 3 is almost equal to the estimated value in equation (3) in Table 4, there is no significant bias in the estimated values in equations (4) and (5) in Table 3. This is also true for the U.S. and Germany. Therefore, there is almost no bias in the estimates of equations (4) through (6) in Table 3, and these coefficient values indicate that the extent to which the price premium for new drugs in Japan reflects innovativeness is about 30% of that in the U.S. and about 60% of that in Germany.

This result is consistent with the analysis in "Price Premiums for New Drugs relative to Comparator Drugs: Analysis Using Matched Samples for Japan, the U.S., and Europe" (Policy Research Institute News No.62). In the previous analysis, we analyzed the relationship between the price premiums in Japan, Europe, and the U.S. and found that only about 30% of the price premium reflected in the U.S. is reflected in Japan, while about 70% is reflected in Germany (for example, a 30% price premium in the U.S. would result in a 9% price premium in Japan). Taken together, these results suggest that differences in the degree to which the innovativeness of a new drug is reflected in its price premium may explain, to a large extent, the differences in price premiums between Japan, the U.S., and Germany.

Conclusion

In this paper, we construct a new measure of innovativeness of new drugs and analyze how the Japanese, U.S., and German drug pricing systems have assessed innovativeness using matched data on the price premiums of new drugs relative to comparator drugs. The analysis extended Comanor et al. (2018) and "Price premiums for new drugs relative to comparators: an analysis using matched samples from Japan, the U.S., and Europe" (Policy Research Institute News No. 62) to examine the factors behind the smaller levels and fluctuations of price premiums in Japan and Europe compared to the U.S. in the price premiums of new drugs from the innovation The analysis examined the factors behind the lower levels and fluctuations of price premiums in Japan and Europe compared to the U.S. from the perspective of the evaluation of innovativeness in the price premium of new drugs. The use of a matched sample greatly reduces the likelihood that missing variables at the disease level (factors we cannot control for in our unavailable data) or in the combination of new drugs and comparator drugs that act on drug prices will distort the estimates. Key findings and implications are as follows.

First, the variation in the price premium of new drugs in the U.S. more strongly reflects the degree of innovativeness of new drugs than in Germany or Japan. According to our estimates, the degree to which the price premium for new drugs in Japan reflects innovativeness is about 30% of that in the U.S. and about 60% of that in Germany. Thus, one of the causes of the higher price premium in the U.S. is that it reflects the innovativeness of new drugs more. However, this is only one of the causes.

Second, our newly constructed measure of innovativeness of new drugs from the data of scientific paper citations of drug patents significantly explains the variation in price premiums in the U.S., Germany, and Japan. Science is an important driver of drug discovery innovation, and the use of scientific advances to invent extensively and at a high rate is important for innovative drug discovery.

Third, a comparison of the degree to which innovativeness is reflected in the initial price premium and the degree to which innovativeness is reflected in the price premium over the life cycle shows that the latter tends to be greater in the U.S. than in Japan and Germany. This suggests that in the U.S., there is a mechanism by which the innovativeness of a new drug, which is recognized after its launch, is reflected in its price.

Although there are many issues for future research, we believe the following are important. First, the sample of target drugs available for this estimation is small, and it is important to expand it. It is necessary to set up an appropriate control group of drugs whose prices are determined based on the cost accounting method, rather than the comparable drug effect method, to be included in the analysis. Second, the analysis should include the UK and France as additional European data. Third, it is necessary to investigate the reasons why innovativeness is not reflected in drug prices under the Japanese drug price system as it is in the U.S. and Germany. This can be examined based on data on the usefulness addition ratio and the foreign average price adjustment ratio, but this will require an expansion of the sample size. Fourth, further examination of innovativeness indicators is also necessary. Although this paper focused mainly on the use of science in research, it is also important to utilize the results of clinical trials and their publication data. In particular, clinical trials may directly show differences in drug efficacy from comparative drugs. Finally, it is also necessary to elucidate why science utilization speed is important among the innovativeness indicators. While speed of utilization may contribute to the acquisition of first-in-class and other leading advantages, verification of this will be necessary in the future.

Appendix 1: List of Data Used and Procedures for Constructing the Analytical Data

List of data used

The following data is available: 1) IQVIA Pricing Insights (licensed under Grant-in-Aid for Scientific Research B), 2) patent data from the San-Ei Report and information on protection periods and reexamination periods, 3) drug patent article information from Clarivate (licensed under Grant-in-Aid for Scientific Research B), and 4) drug price information created by the Pharmaceutical Industry Policy Institute, constructed from drug price calculation information from the Chuikyo Pharmaceuticals Association. (e.g., addition ratio, foreign average price adjustment ratio, etc.); (5) daily drug prices of new drugs and comparative drugs collected from the NHI drug price calculation information of the JCIA; (6) drug classification for selecting similar drugs by the JCIA; (7) GDP deflator and purchasing power parity information (2015 prices) from the OECD; (8) sales of IQVIA World Review Analyst 2018 Top Drug Information.

Data Construction Procedure

IQVIA Pricing Insights, which we license from IQVIA, provides monthly pricing information for drugs launched in Japan, the United States, Germany, the United Kingdom, and France from July 2010 through March 2019 (with the exception of Japan, for which pricing information is missing for 2010). IQVIA Pricing Insights contains price information at each stage of distribution, but this analysis uses pharmacy purchase prices (PPP), which are assumed to be the closest to actual prices in each country.

From this basic database, we created a database by matching the price information of each country for each drug unit that shares the same generic ingredient name, international trade name, dosage form, and specification unit, and then simply averaged the price data on a quarterly basis for each country. We then obtained information on GDP deflators and purchasing power parity as of 2015 from the OECD, realized the data at 2015 prices, and converted the data to purchasing power parity to produce dollar-based price data.

Next, a list of the top 300 drugs in terms of sales from IQVIA World Review Analyst 2018 and a list of comparative drugs corresponding to the top drugs identified from the drug price calculation information from the Japanese Central Medical Association were created and connected to the database constructed from IQVIA Pricing Insights. The list was then connected to a database built from IQVIA Pricing Insights.

Finally, we extracted the 69 components of drugs that have been launched in Japan, for which NHI prices have been determined on a comparative basis, and that are also available in the United States and one of the three European countries (Germany, the United Kingdom, or France) (comparative drug data corresponding to each of the 69 new drugs calculated using the Comparable Drug Efficacy Method). In addition, all of the drugs in the analysis had to share the same generic ingredient name, international trade name, dosage form, and specification unit in each country.

In addition, in order to calculate the price ratio of new drugs to comparator drugs on a per treatment basis, this paper used information on the daily drug price of each drug from the NHI drug price calculation data of the Central Medical Council to determine the price premium (logarithmic value of price ratio, the rate of price increase of a new drug relative to comparator drugs) for new drugs relative to comparator drugs in each country.

In addition, as an indicator of the innovativeness (novelty) of each component, whether it is a new mechanism of action or not, or how many of the existing mechanisms of action have been launched, the "Drug Classification for Selection of Similar Drugs" of the Chuikyo was used as a basis. This document is "a classification of ethical drug ingredients created to make the selection of similar drugs for the similar drug effect comparison method more transparent, and is the basic data for determining pharmacological action similar drugs in the NHI drug price calculation. Furthermore, as an indicator of innovativeness, we identified a group of patents (substance patents, crystal patents, and use patents) protecting each drug ingredient from the San-Ei Report, and by connecting to Clarivate's patent article database, we collected scientific papers (registered in Web of Science) cited by the relevant patent group (U.S. patents). The scientific papers cited by the patents (U.S. patents) were collected. Then, the following data were collected: the extent to which each drug component cites scientific papers (number of scientific papers), the average number of times the scientific paper is cited in other papers (number of citations), and the extent to which the scientific paper is utilized early (for each component, the average difference between the publication date of the scientific paper and the Japanese filing date of the patent that cites it: month (for each component, the average difference between the publication date of the scientific paper and the filing date of the cited patent in Japan: calculated on a monthly basis and converted to an annual basis).

Finally, we used information on the patent protection period of each pharmaceutical ingredient in each country from IQVIA Pricing Insights. Information on price premiums was available for 69 components, but only a limited number of components had innovation indicators and information on patent protection periods for new drugs and comparator drugs, as well as drug price data for periods when both of these patents are in effect. In addition, the estimation process uses drug components common to Japan, the U.S., and Germany for international comparisons, so the actual estimation process falls short of 69 components.

Appendix 2 Theoretical Background of the Analysis

Suppose that the drug price p of a certain new drug i is proportional to the country's willingness to pay for health improvement w, and is also determined as follows, reflecting the level of innovativeness of the drug in question (the drug's therapeutic effect) q, and other factors α. All of these factors are assumed to depend on the combination of disease and drug and do not differ by country.

p _i = wα _i (q _i ) ^β (1.1)

Correspondingly, the drug price of the comparative drug,

p _o,i = wα _0,i (q _o,i ) ^β (1.2)

where β ≥ 0. The degree to which prices reflect innovativeness varies from country to country. If β = 1, then the drug price is set equally and proportionally to the effect of innovativeness; if β = 0, then the price does not reflect innovativeness.

（Taking the logarithm of both sides of equations (1.1) and (1.2) to obtain the log difference between the new drug and the comparator, the price premium (the percentage increase in price relative to the comparator) for the new drug is as follows

premium _i = lnp _i -lnp _o,i = β ln(q _i /q0 _{, i} ) + ln (α _i /α0 _{, i} ) (2.1)

Assuming that the drug price is determined in the same way in the reference country (the U.S.)

premium _i^{* = lnp} _i^lnp _o,i^{∗ =β∗} ln (q _i /q0 _{, i} )+ln (α _i /α0 _{, i} ) (2.2)

It follows that

ln (q _i /q _o,i ) is the rate of increase in the therapeutic effect of the drug, which is the progress in scientific knowledge between the two drug inventions (sci _i ), which depends on the use of the drug and other factors. Factors independent of scientific advances increase the therapeutic effect of the pharmaceuticals, θ _i is defined as θ The residuals in the treatment effect that cannot be explained by scientific progress are defined as θ _i and assume that the two are independent.

ln(q _i /q _o,i ) = δlnsci _i +θ _i (3)

In this case, the domestic price premium can be calculated from equation (2.1) as

premium _i = β(δln(sci _i ) + θ _i ) + ln(α _i /α0 _{, i} ) (4.1)

Similarly, in the reference country (U.S.)

premium _i^{* =β*} (δln(sci _i ) + θ _i ) + ln(α _i /α0 _{, i} ) (4.2)

It follows that where sci _i coefficient reflects the progress of science (and all the effects it has) on the innovativeness of the drug. ln(α _{i/α0, i} ) is zero (no difference between the new drug and the comparator) or ln(sci _i ) is uncorrelated with β, we can obtain β without bias by estimating equation (4.1). If there is correlation, bias arises in the estimation of β.

ln(α _{i/α0, i} ) is determined by the type of disease, which is ln(sci _i ) correlates with ln(α i/α0,i), but if the former is determined by the characteristics of the drug independent of the country and is common, then the effect can be eliminated by differencing across countries. That is, if we take the difference between equations (4.1) and (4.2),

premium _i^premium i _i = (β∗ - ^β )(δln(sci _i ) + ln(θ _i )) (5)

The result is as follows. The difference in the two estimates allows us to estimate ( ^β*-β ) without any bias.

After a new drug is launched, the price premium changes according to the drug price revision of the new drug and the comparative drug. Assuming that the drug price at the time of launch is given by equation (4.1) and that the price premium changes in proportion to the elapsed year t

premium _i,t = β(δln(sci _i ) + θ _i ) + ln(α _i /α0 _{, i} )+γt (6)

It follows that. （The estimation of equation (6) allows us to estimate β by utilizing price data at multiple time points, along with a parameter γ that summarizes the variation of the price premium over time (more accurate if the variation over time is linear and γ is estimated correctly). By estimating international differences, we can also assess the extent to which prices reflect innovativeness at the initial point in time (Δβ = ^β*-β ) and differences in dynamics (Δγ = ^γ*-γ ). Assuming that the year of launch of the drug is equal in both countries,

premium _i^premium i _i = (Δβ)(δln(sci _i )+θ _i )+(Δγ)t (7)

Note that in the actual estimation model, we introduce a new drug launch year (cohort year) dummy, a calendar year dummy and an ATC drug effect category dummy. By introducing a launch year cohort year variable, the sci _i data base, we can control for the effect of changes in citation conventions and truncation of citation data from drug patents to the Web of Science, which is the basis of the data in the This is the increase in the propensity to cite scientific papers from the patent literature, or the presence of truncation in the number of citations to scientific papers, etc. In other words, the estimation uses only the variation among drugs launched in the same year. The calendar year dummies control for the fact that our drug price data is not a balanced panel, and as a result, fluctuating price premiums affect the dynamics of drug prices; the ATC drug classification dummies control for disease-specific demand and supply factors.

Appendix 3: Basic Statistics

References

Reference Figure 1 shows the mean and median price premiums for 31 drug components in Japan (left axis). It also shows the drug price of the new drug and the corresponding comparator drug (right axis). The average and median price premiums were stable at 0.15 and 0.08, respectively, before the exclusivity protection period for the comparator drug. In Japan, after the exclusivity protection period for comparator drugs, both the mean and median prices show a similar upward trend, but this is mainly due to a decline in the price of comparator drugs. While most of the new drugs are covered by the additional subsidy for new drug creation, and thus their prices are maintained without falling until a certain period of time, the prices of comparative drugs are on a downward trend due to the expiration of patent protection around 2012 and the corresponding changes in drug price regulations and competition, resulting in a large increase in price premiums.

Reference Figure 2 shows the price premiums of 40 drug components in Germany (left axis). It also shows the prices of the new drugs and the corresponding comparator drugs (right axis). The average price premium tends to increase slightly before the period of monopoly protection for the comparator drug, but remains almost constant between 0.23 and 0.24 for the entire period. The median price premium tends to increase slightly over time from 0.11 to 0.14 before the exclusivity protection period for comparator drugs. In Germany, drug prices for both new drugs and comparison drugs have tended to decline by the same amount over time in real terms, partly due to changes in the price level. For comparative drugs, patent protection expired around 2014, and prices have been declining at an accelerated pace since then, but the impact on price premiums has been minor compared to Japan and the U.S.

Reference figures 3 through 5 below show the relationship between innovativeness and price premiums by country, with innovativeness (synthetic indicator) on the horizontal axis and price premiums averaged over the estimation period on the vertical axis. These figures are consistent with the estimation results in the main text.

Return to News List

TOP