The provision of affordable, accessible, and innovation-promoting pharmaceutical care services is an essential element of a sustainable health policy (1). In most European countries, the main purchaser of pharmaceuticals is the public sector (2), and in these quasi-monopsony conditions, national health services’ negotiating power is enhanced as to the determination of pharmaceutical prices. To this respect, the vast majority of European countries employ external reference pricing (ERP), a price-regulation method whereby a government considers the price of a medicine in a specified basket of countries in order to set its price (3). The application of ERP differs among countries, with regard to, e.g., basket size, basket composition, and re-referencing rates. Yet, a country’s ability to adjust medicinal prices to its income and affordability is limited due the method’s endogenous structural and functional characteristics.

External reference pricing is a univariate algebraic process that incorporates information solely on prices, not accounting for other key socioeconomic factors, e.g., health-care resources, health-care structure, and demographic patterns (4), in the equation. In addition, the use of extensive ERP baskets results in the increase of unmatched observations on currencies and medicinal properties (brand names, packaging, etc.), encumbering the computational process, as ad hoc studies illustrate (5–10). As a consequence, ERP’s sub-optimal design has a profound impact on health policies’ efficiency: it produces outcomes that present low adaptability to economies’ national income, increase market risks (e.g., arbitrage, medicine launch delays, slashed patent holders’ profitability and competitiveness, etc.), raising barriers to existing therapies and to innovation as well (2). Nevertheless, the use of large baskets of countries remains a common European practice, given that 12 European Union (EU-28) countries refer to a minimum of 15 countries (11)¹.

Greece is a dynamic player of the European ERP system. The implementation of ERP pertains to all on-patent medicines sold in the country, directly affecting reimbursement. According to legislation, the prices of on-patent medicines are set as the average of the three lowest ex-factory prices in the EU-28. Any available price (e.g., ex-factory, wholesale, retail, hospital, insurance) is collected from official, published sources, and the necessary conversion of retail or wholesale to ex-factory price is made according to methodology and rates determined by the National Organization for Medicines (EOF) (12). Using ERP, pharmaceutical prices are reviewed (re-pricing) twice per year. Following each re-pricing, the Positive Reimbursement List is also reviewed, in order to update reimbursed prices accordingly. At the same time, Greece is included in the ERP basket of half of the EU-28 countries [14 out of 28 (3)] and is among the countries with the highest frequency of re-referencing (biannual) (2).

Since the beginning of the economic crisis in 2010, pharmaceutical expenditure has been placed in the center of fiscal consolidation and has become inextricably linked to the level of the continuously decreasing Greek gross domestic product (GDP) (13). In particular, according to the target set by the Memorandum, public pharmaceutical expenditure must not exceed 1% of GDP. For this reason, a number of cost-containment measures targeting the prices of pharmaceuticals, such as flat price reductions, the implementation of multiple rebates/discounts and ERP have been applied to obtain quick reductions in public pharmaceutical spending and thus achieve the fiscal targets. In this context, fitting flexible and unerring pharmaceutical pricing models that take into account national income and financial indicators is a matter of national necessity. The present study investigates new methodological pathways in the field of pharmaceutical pricing and reimbursement with aim to communicate additional evidence and inform health policy decision-making.

The next paragraphs provide point estimates of the differential basket and reimbursement revision effects on Greece’s prices. Quantitative results for the potential impact of the proposed model on the country’s external environment are additionally presented. The microeconomic, macroeconomic, and demographic indicators that compose the research conditions under which the present experiment was conducted are also discussed.

The formed baskets represent countries with similar demographic trends, as indicated by the data on life expectancy and death rates presented in Table 1. However, homogeneity between Greece and its baskets is reduced when health resources are taken into account. In particular, medical staff density in the population varies significantly between Greece and the baskets. Yet, this variation is independent from the baskets’ composition, considering Greece’s ranking above the OECD average with regards to this particular health-care parameter (16). The picture is completely reversed when the geographical dispersion of the nursing personnel is examined, with the number of practicing nurses per 1,000 people less than half in Greece compared to the baskets (Table 1). Another difference between Greece and the baskets concerns total per capita health expenditure. Greece’s per capita health expenditure is below the corresponding average of the baskets. The latter is consistent with the rich theoretic background on the positive relation between GDP and public health expenditures per capita in OECD (18) and the fact that Greece’s income was lower in comparison with the baskets.

Income correlations constitute fundamental characteristics of this experiment where differential IP effects hinge on. The baskets represented countries of a higher income, on average, compared to Greece (Table 1). Basket 1 is placed below the European average (€25,500 and €28,300 in EU-28 and the Euro-area, respectively) (17), while basket 2 exceeds the EU-28 average. The application of IP reduces the level of current home prices by 6.49–7.47% (Scenarios 1 and 2, respectively) (Table 3), causing though no severe impact abroad, considering the small effects that vary per product and overall between −1.71% (medicine “M₆,” Scenario 2) and −0.64% (medicine “M₁₀,” Scenario 1), as Table 4 shows.

Regarding the performance characteristic, Greek prices were lower, compared to the mean price of the baskets, yet they presented higher variability, as indicated by the comparative central tendency and variation measures presented in Table 1. There were no statistical differences among the price distributions of the Scenario 1 (eight countries), except for the pair “France-Slovak Republic,” where French prices were lower (p-value = 0.05). In the smaller country set of the Scenario 2, Ireland appeared as an outlier, having statistically higher prices compared to France, Greece and Portugal (p-values 0.017, 0.022, and 0.024, respectively). This scenario yielded statistically lower Greek prices by 7.61% (p-value = 0.05) compared to the mean of the basket. CBI effects are in accordance with these statistical results, retaining however, their small range (<1%) across scenarios, considering that basket downsizing to five to eight countries caused Greek prices’ increase by 8.24% at the most (Table 3).

External reference pricing, a common policy tool used worldwide to control prices of pharmaceuticals, is – in its present form – no longer considered sufficient in terms of operational efficiency and adaptability to payers’ available income, and has raised a lot of discussion and concerns among policy-makers, academia, and the industry. In this context, the present analysis produces new evidence regarding changes that could be incorporated in the ERP tool, to restore its efficacy.

A non-randomized experiment was conducted in order to examine the impact from the implementation of an ERP model that is flexible and adaptable to health systems’ affordability level. This experiment used Greek pharmaceutical prices as a key node of the European ERP network and tested the impact of ERP basket reconstruction and price-model fitting based on fiscal parameters. Results are examined with regard to the impact of differential price effects on both the Greek and other EU pharmaceutical markets, assessing the proposed model in terms of its theoretic value and contribution.

Regarding the insurance price proposition, the model yielded small price effects in the two examined scenarios, despite the identified statistical differences between the involved price populations. Small price reductions though not only are consistent with the budgetary targets of the country but also deter collateral financial damages of restrictive pharmaceutical policies, such as public revenue losses, which is currently a serious fiscal issue in Greece (15).

Moreover, with its small price decreases, the proposed IP model poses a disincentive for parallel exports of pharmaceuticals, which in the past has been an intense phenomenon (19), restricting thus the role of arbitragers and thereby reducing the risk of pharmaceutical market shortages and barriers to pharmaceutical care (20). Considering barriers to medical care a topical health-care issue in Europe (16), this model property is very important from a general health policy perspective, and utilizable in Greece, where unmet health needs currently affect 9% of the population on average, while specific patient populations encounter even higher rates (e.g., 13% for multiple sclerosis patients) of barriers to medication (21).

In this context, the application of a bivariate pricing model that takes into account not only prices but also countries’ affordability level adjusts prices in the best interest of the public sector and the beneficiaries. Beyond these encouraging results, the model’s moderate effects also restrict unwanted price effects and potential damage in other countries to low percentages, i.e., around 1% (in absolute terms).

In support of these findings, which are consistent with other published studies (2), the estimated values of external cost indices (EI₁ and EI₂) are in fact overestimations. The mean size of EU baskets that include Greece equals 21 (11),⁵ whereas this work examined considerably smaller basket sizes (≤8 countries), which decrease the denominator of the quantity "pGP", thus increasing the value of indices, “EC₁” and “EC₂” (Eqs 15 and 16).

In terms of the question of theoretic contribution, the presented model proposed the use of smaller baskets in ERP practicing. Taking into account the complex system of Greece with multiple sources of data, we conclude that structures that are more flexible bypass bureaucracy, enable computational transparency, and increase homogeneity of referenced countries. Furthermore, as the moderate increasing effects (<10%) of the basket-downsizing model on the Greek prices indicate, this improvement is achieved without causing serious price and market volatilities.

Technological progress, which shifts quality and prices of pharmaceutical products upwards (18, 22, 23), as well as the rapid change in the demographic and epidemiological profile of the population (24), create a dynamic grid of long-term challenges for health-care systems, which are facing major pressures on their budgets. In light of the above, health-care decision makers must not stay inert. Advancing the quantity and quality of information incorporated in pharmaceutical pricing methodology, as suggested by the present analysis, is a road that is worth being paved for all: the economy, the health-care system, and the health status of the population.

All authors contributed equally to the conduct of the study and the development of this manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.