Changes in launch delay and availability of pharmaceuticals in 30 European markets over the past two decades

We extracted the first international launch date and national launch dates of prescription drugs from the IQVIA (formerly IMS Health) Database for the following 30 European countries: Austria, Belgium, Bosnia-Herzegovina, Bulgaria, Croatia, Czechia, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Norway, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, and the United Kingdom (UK). The dataset has been used for distantly related research before [5, 18, 19].

We restricted our sample to prescription drugs that were launched internationally between 2000 and 2017 in the retail or hospital market. Our follow-up period for national launch dates was from 2018 until the second calendar quarter of 2020. We distinguished between four time intervals to examine how the values of our indicators changed over time: 2000–2004, 2005–2009, 2010–2014, and 2015–2017.

To avoid confounding (a) by pharmaceuticals launched very late in some countries after already having been taken from the market in other countries, and (b) peculiarities in being counted as a (prescription) drug, two researchers independently reviewed all pharmaceuticals that were launched in fewer than five countries to verify that the international launch date given in the database was correct. Furthermore, we excluded vaccines from the analyses because these were marked as newly launched in the database in cases where a booster of an existing vaccine had been launched.

We operationalized access to newly discovered pharmaceuticals in three ways: First, we calculated the length of time between the first international launch date of each drug and its corresponding national launch date in each country and defined this as the launch delay. Second, we ranked each country in our sample according to the average launch delay across all new pharmaceuticals in that country, and we analysed changes in the ranking order of countries by year and across the four time intervals defined above. Third, we measured differences in the overall availability of new pharmaceuticals across the countries in our sample over time. We did so by determining the number of pharmaceuticals launched internationally during each of the four time intervals and calculating the percentage of these pharmaceuticals that were available in each country during that interval. For this last indicator, we excluded Estonia, Greece, and Luxembourg from our analyses because the IQVIA collects data only for the retail market in these countries.

To investigate whether pharmaceuticals classified as essential medicines by WHO (EMLs) [20] follow similar trends in launch delay and whether availability of pharmaceuticals differs compared to all other drugs, we analysed launches of EMLs in a subgroup analysis. In our dataset we found 46 launches to be part of the 482 medicines listed on EMLs (9.5%).

Furthermore, we sought to understand whether our measure of launch delay and the availability of newly discovered pharmaceuticals differed between (a) drugs that covered a potentially urgent need (i.e., first-in-class innovations) and (b) drugs that were similar to, and without substantial therapeutic advantage over existing products (i.e., so-called me-too products). To do so, we examined pharmaceuticals to treat diabetes (DPP-4 inhibitors) and psoriasis (monoclonal antibodies) more specifically in a subgroup analysis. We differentiated between sitagliptin and all other DPP-4 inhibitors, and adalimumab and all other monoclonal antibodies for psoriasis. We chose diabetes and psoriasis as therapeutic areas, as data was most robust for these. I.e., in diabetes and psoriasis, we observed most launches in our time period.

Furthermore, we extracted annual data on the gross domestic product (GDP) per capita of all 30 countries in order to correlate these with the launch delay and availability of pharmaceuticals [21]. Because the investigated variables were not normally distributed (all p-values were highly significant using the Shapiro-Wilk test: launch delay (p < 0.000), availability of pharmaceuticals (p < 0.000), GDP (p < 0.000)), we employed correlation analysis according to Spearman (for launch delay) and a point biserial correlation (for availability).

All analyses were performed using Stata SE 16.

Our final sample included 492 different molecules, of which 124 were launched between 2000 and 2004, 120 between 2005 and 2009, 158 between 2010 and 2014, and 90 between 2015 and 2017. Across the whole sample of 30 countries, there was a mean launch delay of 35.96 months in the first period (2000–2004), 27.52 months in the second period (2005–2010), 22.69 months in the third period (2010–2014), and 14.80 months in the fourth period (2015–2017). When looking at the mean launch delay across all pharmaceuticals in the five biggest European pharmaceutical markets, France was ranked as the country with the 12th shortest launch delay, Germany with the 5th shortest launch delay, Italy with the 11th, Spain with the 6th, and the UK with the 3rd in 2000. In 2017, however, France was ranked as the country with the 13th shortest launch delay, Germany with the 3rd shortest launch delay, Italy with the 14th, Spain with the 18th, and the UK with the 4th. Lastly, of the pharmaceuticals that were launched internationally during the four periods of our analysis, 69.86% were available in our sample of applicable countries in the first period (2000–2004), 65.21% in the second period (2005–2009), 66.50% in the third period (2010–2014), and 52.63% in the fourth period (2015–2017).

Generally, we saw a dynamic towards a decrease in launch delay across all investigated countries in our sample, with this delay falling from 37.2 months in 2000 to 11.8 months in 2017. From the first period (2000–2004) to the second period (2005–2009), the mean launch delay decreased by 8.9 months. From the second period (2005–2009) to the third period (2010–2014), the launch delay decreased on average by another 4.9 months. From the third period (2010–2014) to the fourth period (2015–2017), the launch delay decreased on average by 7.7 months. For example, for the five biggest European pharmaceutical markets, the launch delay decreased (from the first to the third period) from 33.0 to 24.1 months in France, from 19.4 to 11.5 months in Germany, from 30.7 to 20.7 months in Italy, from 26.4 to 23.5 months in Spain, and from 20.4 to 10.9 months in the UK. For the five smallest European pharmaceutical markets, the launch delay decreased from 99.5 to 45.2 months in Bosnia-Herzegovina, from 71.9 to 45.5 months in Serbia, from 61.3 to 33.7 months in Croatia, from 44.8 to 24.4 months in Slovenia, and from 51.9 to 30.9 months in Romania. The largest decrease in launch delay, of 36.46 months, was seen for Bosnia-Herzegovina from the first period to the second period. However, this decrease should be interpreted with caution given the restricted length of follow-up.

Changes in the ranking order of countries by launch delay occurred over the entire observation period. For example, for the five biggest European pharmaceutical markets, we found that France went from rank 12.8 in 2000–2004 to rank 17.0 in 2015–2017, Germany went from rank 4.6 to rank 3.8, Italy from rank 13.4 to rank 14.4, Spain from rank 9.0 to rank 18.3, and the UK from rank 5.0 to rank 3.4, with smaller ranks indicating shorter mean launch delays. The three countries that were able to improve their rank (i.e., reduce the launch delay) the most from 2000 to 2004 to 2015–2017 were Latvia (− 9.15 ranks), Slovenia (-7 ranks), and Luxembourg (-6.5 ranks). The 3 countries whose ranks worsened the most (i.e., due to an increase in launch delays) were Ireland (+ 10.65 ranks), Spain (+ 9.25 ranks), and Switzerland (+ 5.2 ranks).

For EML-drugs launch delay seems to be roughly the same compared to all new pharmaceuticals in 2004–2010. Thereafter (2011–2014) launch delay was shorter for EML-drugs compared to all other pharmaceuticals. Data for 2015–2017 could not be analysed as only a single drug listed on EML was launched (see appendix, Table S2).

For our subgroup analysis looking at diabetes and psoriasis, we observed similar launch delays for first-in-class innovations and me-too products. The mean launch delay across all 30 European countries was 24.05 months for first-in-class innovations and 25.10 months for me-too products.

Correlation analyses using Spearman’s rank correlation coefficient showed that launch delay was negatively correlated with a country’s GDP per capita (-0.67, p < 0.000), i.e., the higher the GDP, the shorter the launch delay (see Table S1, Appendix).

Figure 1 gives an overview of launch delays in 30 European markets in 2000 versus 2015. Table 1 reports the descriptive statistics for the yearly launch delay over time. Changes in the ranking order of launch delays over time can be seen in Fig. 2.

In general, the availability of pharmaceuticals that had been newly launched internationally varied significantly across the countries in our sample. The availability of such pharmaceuticals was quite stable from 2000 to 2014, but decreased afterwards, potentially due to the short length of follow-up time which results in less time to observe a launch. For the five biggest pharmaceutical markets in the EU, we found that the availability of new pharmaceuticals decreased from 72.6% in 2000–2004 to 61.1% in 2015–2017 in France, increased from 85.5 to 91.1% in Germany, decreased from 81.5 to 74.4% in Italy, decreased from 82.3 to 70.0% in Spain, and increased from 82.3 to 83.3% in the UK.

Looking at mean values across the whole period from 2000 to 2017, Germany had the highest availability of new pharmaceuticals (85.7%), followed by the UK (83.1%) and Norway (82.9%). The three countries with the lowest availability of new pharmaceuticals over the same period were Bosnia-Herzegovina (24.8%), Serbia (36.4%), and Latvia (43.5%). The three countries in which the availability of new pharmaceuticals decreased the most during this period were Turkey (-16.8% points from 2000 to 2004 to 2010–2014), the Netherlands (-14.0% points), and Bosnia-Herzegovina (-12.0% points).

Availability was generally higher for EML-drugs compared to all other drugs. (see Table 2). For our subgroup indications diabetes and psoriasis, we saw that the availability for first-in-class innovations vs. me-too products differed substantially. The average availability across all 30 European countries was 98.14% for first-in-class innovations and 79.16% for me-too products. Compared to the availability observed across all products (64.49%), availability for the products in our subgroup analysis was quite high. This also implies that at least one DPP-4 inhibitors and at least one monoclonal antibody for psoriasis were available in all markets.

The point biserial correlation analyses showed that GDP was positively correlated with the availability of pharmaceuticals – that is, the higher the GDP, the higher the availability of pharmaceuticals (+ 0.19, p < 0.000) (see Table S1, Appendix).

Table 2 reports the descriptive statistics for the availability of new pharmaceuticals over time.

Our analyses have several limitations. First, IQVIA’s methods for collecting data on launch dates differs slightly among countries. For example, launch dates for Estonia, Greece, and Luxembourg are calculated based on reports from the retail sector, whereas launch dates for all other countries are calculated based on both the retail and hospital sectors. However, the potential bias this might have introduced to our measure of launch delay and the ranking order of launches is probably small because launch dates from the retail and hospital sectors do not differ significantly. Nevertheless, we had to exclude Estonia, Greece and Luxembourg three countries in our analysis of the availability of pharmaceuticals.

Second, our calculations of launch delay can be based only on available data. Especially in cases where new products have not yet gained market access in all countries in which a launch is planned, the launch delays could be longer than is suggested by the results of our analysis. This is particularly the case for the 4th time interval (2015–2017) in our observation period. However, except for Bosnia-Herzegovina, Bulgaria, Croatia, Estonia, France, Hungary, Lithuania, Serbia, and Turkey, the average launch delay in 2015 was already shorter than two years, minimizing the impact on our analysis of having only three years of follow-up for most countries in our sample.

Third, in some countries (e.g. Germany), a new drug can be accessed immediately at launch and is reimbursed for all indications by payers. However, this is not the case in every country. In some countries, only parts of the indication are reimbursed or reimbursement is restricted to specific groups. Thus, our measure of launch delay -based on launch dates and not actual use - have certain limits when it comes to evaluating patient access.

Lastly, two researchers independently reviewed all pharmaceuticals that were launched in fewer than five countries to verify some of the launch dates in the database. Although carried out to the best of our abilities, manual checks are always accompanied by uncertainties. However, because two reviewers independently performed the check, errors should have been reduced to a minimum.