Inequalities in medicine use in Central Eastern Europe: an empirical investigation of socioeconomic determinants in eight countries

We draw on cross-sectional data from the first wave of the European Health Interview Survey (EHIS), collected between 2006 and 2009 in 19 EU Member States [43]. This study focuses on CEECs. Countries in this region, due to their weaker economic situation compared to Western European countries, risk substantial socioeconomic inequalities. Eight (Bulgaria, Czech Republic, Hungary, Latvia, Poland, Romania, Slovenia, Slovakia) of the ten CEECs that were EU Member States at the time of the EHIS survey were included in our study (see Table 2 for an overview of the dataset); Estonia and Lithuania were excluded due to limited data and non-inclusion in the EHIS survey, respectively. Generally, the data focussed on the non-institutionalized population, but in the Czech Republic also the institutionalized population (living in nursing homes and convents/monasteries) was included in the survey [44]. Since this, however, only concerns less than an estimated 0.5 % of this country’s sample population (Š Daňková 2014, pers. comm., 6 October), we hardly expect this to affect our results. Finally, the EHIS dataset covers the population above age 15, however, we restrict our analysis to adults (aged 18 and over) as it is assumed that the development of an individual socioeconomic status may be expected from around this age onwards.

To determine the association between socioeconomic status and pharmaceutical consumption, we carried out multinomial multivariate logistic regression analyses (Stata 13.1). In these models, the probability was estimated of falling into certain categories compared to the reference group. To check the statistical significance of the calculated relative-risk ratios (RRR), z-statistics were used. Only results with p < 0.05 were considered statistically significant and thus discussed further. McFadden’s R² is reported as measure of model fit.

A detailed description of the variables included in the regression analysis can be found in Table 3. The dependent variable is based on the information on the consumption of non-prescribed medicines (only), consumption of prescribed medicine (only) and consumption of both types of medicines versus no medicine use. For an unbiased gender comparison, use of female-specific medicines (contraceptive pills; hormones for menopause; based on an additional survey question) was excluded from the analysis. As independent variables of interest, three socioeconomic indicators were included [45]. First, employment status was grouped into employed (reference group), retired, unemployed or disabled and in training or at home. Second, the highest level of education completed was measured based on the International Standard Classification of Education (ISCED) 1997 and distinguishes between ISCED ≤ 2, ISCED 3-4 and ISCED 5-6. Third, the income level was based on the range of the monthly net household income against the values of the national deciles’ limits and categorized into quintiles (reference group: first quintile). To control for demographic characteristics, we included age and sex, with the youngest and males as reference groups. Finally, since a socioeconomic gradient in health disadvantaging the poor is also well-established in CEE regions [46], we introduced dummy variables for chronic conditions (no chronic conditions as reference group) and self-assessed health (good as reference group).

An overview of the descriptive information for the individual countries is given in Table 4.

The analysis of pharmaceutical policies as potential co-determinant of inequalities in medicine use builds on an analytical framework based on the WHO approach for collective action [42] (see Addressing inequalities in medicine use and Table 1). We identified four key dimensions for the analysis, (1) sustainable funding, (2) affordability, (3) rational selection and use of medicines and (4) availability and accessibility. For each of the four dimensions, we developed a set of indicators (see Tables 1 and 5).

To fill these indicators with country specific information, we drew on several data sources: Background information and data for the policy analysis were taken from the literature, predominantly grey literature. Search terms included ‘medicines’, ‘pharmaceuticals’, ‘equity’, ‘access’ and the name of the country. We focussed on the period around the EHIS survey, and three years before, and also literature with overall relevant information, e.g. related to the transition process, was included. Country reports about national pharmaceutical systems were a key source of information, in particular the PPRI (Pharmaceutical Pricing and Reimbursement Information) Pharma Profiles (available for five of the eight CEECs), the Health Systems in Transition reports by the European Observatory on Health Systems and Policies as well as country reports produced by the Organisation for Economic Co-operation and Development (OECD) and by the Austrian Public Health Institute (references listed below Table 5 and below the additional table [see Additional file 1]). Since some of the reports did not refer to the exact years of the survey, unpublished information from the PPRI network was used as supplementary source [47, 48]. Health and pharmaceutical expenditure data were retrieved from the Eurostat database [49].

Based on the information and data collected for the countries, we produced eight brief country posters [see Additional file 1]) and an overview table (Table 5). Following methods of policy analysis research [50], we used the country specific information and data to understand whether the identified socioeconomic (in)equalities in medicine use were impacted by the existing implementation degree of pharmaceutical policies. In the analysis, we interpreted ‘prescribed medicines’ as medicines funded, at least partially, by state (independently from their prescription status: both prescription-only and non-prescription medicines), whereas ‘non-prescribed medicines’ were considered as non-funded self-medication.

Overall, in the regression analyses we find that whenever there is a socioeconomic gradient in medicine consumption to be observed in a country, it will in virtually every case favour the well-off, as shown in Figs. 1 and 2, respectively. The full results including all variables of the country-specific multinomial logistic regression analyses are presented in an additional table [see Additional file 2].

For non-prescribed medicine use (versus no medicine use), individuals with higher education and/or income had a higher chance of taking such medicines in seven of the eight CEECs: In Hungary, for instance, individuals in the highest education group were estimated to be four times more likely to take non-prescribed medicines (RRR = 4.32, 95 % CI = 2.98–6.27) while the relative risk for people in high income households increases by a factor of 1.66 (95 % CI = 1.20–2.31), respectively, given the other variables in the model are held constant. In Poland, on the other hand, the observed gradient was comparably less pronounced (highest education: RRR = 1.79, 95 % CI = 1.55–2.07; highest income: RRR = 1.38, 95 % CI = 1.20–1.58). Only in Bulgaria, neither education nor income played any role for non-prescribed medicine use, but the unemployed and disabled were attributed a lower consumption likelihood.

For prescribed medicines, in contrast, a statistically significant socioeconomic gradient could be identified in three economies only, i.e. Latvia, Poland, Romania: The increased likelihood of the highest education groups in these countries (Latvia: RRR = 2.09, 95 % CI = 1.59–2.73; Poland: RRR = 1.92, 95 % CI = 1.62–2.27; Romania: RRR = 1.73, 95 % CI = 1.19–2.54) is similar in magnitude, whereas in terms of the highest income groups, Latvia showed the most pronounced effect with an RRR close to 2 (RRR = 1.99, 95 % CI = 1.17–3.37).

Regarding the influence of demographic characteristics on medicine use [see Additional file 2], females were more likely to take (any) medicines in general, whereas advanced age was particularly relevant for prescribed pharmaceutical consumption, which is consistent with the literature [51]. Such prescribed medicine use was also highly determined by health characteristics, especially by longstanding illnesses.

Details on the pharmaceutical policies in the CEECs are provided in Table 5 and an additional table [see Additional file 1]. Key features are as follows: Health and pharmaceutical expenditure, primarily funded by social health insurance contributions, was low compared to the Western European countries, and the share of health expenditure spent on medicines was rather high. In 2008, the EU-15 (i.e. 15 countries from Western, Northern and Southern Europe that joined the EU before 2004), spent an average of € 424 on medicines, corresponding to around 15.6 % of total health expenditure, whereas average figures equalled to € 180 and 24.4 % in the CEE-10 (the eight surveyed countries plus Estonia and Lithuania) [49, 52]. Within the eight surveyed CEECs, major discrepancies existed both in terms of the share of health expenditure spent on medicines and per capita pharmaceutical expenditure. Pharmaceutical expenditure was particularly low in Romania and Bulgaria, and also in Latvia and Poland. High increases in pharmaceutical expenditure were observed in Latvia (67 % during the three years before the survey), Romania (59 %) and Slovakia (57 %).

Private funding of health care, particularly of medicines (on average 54 % in the CEE-10), was high compared to the EU-15 with 38 % [49, 52], and there was considerable variation between the surveyed CEECs: Bulgaria had the highest share of private pharmaceutical spending (82 %), followed by Latvia (62 %) and Poland (61 %). Apart from Latvia and Poland, the share of private pharmaceutical expenditure had grown in the three years before the EHIS survey, particularly in the Czech Republic (57 %) and Hungary (38 %). Private pharmaceutical expenditure resulted from full out-of-pocket payments in the case of self-medication (non-prescribed, non-reimbursable medicines) and from co-payments of reimbursable medicines (i.e. medicines whose cost were, at least, partially covered by public payers). Market shares for non-prescription and self-medication medicines were higher than in Western European countries [53‐55]. Whereas no co-payments were applied in the in-patient sector in the CEECs, they were charged for out-patient medicines: Some countries had prescription fees, and all CEECs applied co-payments as percentage of the price for those medicines that displayed a lower therapeutic value. De-listings (i.e. exclusion of medicines from reimbursement) were reported for some CEECs, and they contributed to increased private pharmaceutical expenditure.

The number of authorized medicines and of medicines on the market was usually lower in the CEECs than in the other European countries (for data on Europe see [9]). Except for Slovenia, pharmacy density was higher in the surveyed CEECs than in Western European countries (an average of 3,360 and 5,780 inhabitants per pharmacy in the EU-10 and the EU-15 respectively in 2005, [9]). But major intra-country discrepancies resulting in low accessibility to pharmacies in the rural areas existed in some CEECs (Bulgaria, Czech Republic, Poland, Romania). The sale of (specific) OTC medicines outside pharmacies was permitted in all surveyed CEECs but Latvia, Poland and Slovakia.

Pricing and reimbursement is a national competence in the EU Member States under the condition that overall procedural provisions, as laid down in the EU Transparency Directive [56] such as deadlines for pricing and reimbursement decisions, for instance, are complied with. The CEECs adapted their pricing and reimbursement systems to the ‘acquis communitaire’ before acceding to the EU in 2004 and 2007, respectively [57]. Some CEECs (Czech Republic, Slovenia, Slovakia) undertook additional changes in the years before the EHIS survey, to further strengthen their institutional framework and/or to react to new challenges, such as high-cost medicines [see country profiles in Additional file 1]. At the time of the survey, Poland was fundamentally changing its pharmaceutical pricing and reimbursement system following a European Commission’s infringement procedure related to the EU Transparency Directive that also addressed long decision times. Related to demand-side measures, measures to encourage rational prescribing and use of medicines, including a promotion of generics’ uptake, were in place in the CEECs, but limited enforcement was reported.

Overall, the literature review identified several key challenges and shortcomings in the pharmaceutical systems of the investigated CEECs: In all countries but Slovenia high co-payments and out-of-pocket payments as well a limited enforcement of measures to promote a more rational prescribing and use of medicines (including encouraging generics uptake) were an issue. Underfunding of the pharmaceutical system was reported from Bulgaria, the Czech Republic, Latvia and Romania, and informal payments on a major scale were identified in Bulgaria, Hungary, Latvia, Poland, and Romania. Bulgaria, Poland and Romania were criticized for inefficiencies in procedures and policy coordination. Slovenia was the only country out of the eight CEECs for which the above-mentioned deficits were not reported. However, Slovenia was struggling with limited availability of medicines, and so did Bulgaria, Latvia and Romania.

A major prerequisite for equitable access to medicines is sustainable funding [38, 42]. The comparably low pharmaceutical expenditure in Romania (€ 92 per capita) and Bulgaria (€ 110 per capita), but also Latvia and Poland (around € 135 per capita) points to possible underfunding. Insufficient investment in the field of medicines could be one explanation for the socioeconomic gradient in prescribed medicine use observed in Latvia, Poland and Romania. The more than 50 % increase in per capita pharmaceutical expenditure in Latvia, Romania, Slovakia and Bulgaria in the three years before the survey could be interpreted as countries’ efforts to invest into medicines, but it might also be an indicator that the countries spent more than their economic situation allowed and thus negatively impacted sustainability.

Apart from Slovakia, all above mentioned countries had shares of private pharmaceutical expenditure of more than 55 % (Bulgaria as high as 82 %). The high patients’ contributions in Latvia, Romania and Poland are likely a major reason for the non-use of non-prescribed, privately funded medicines, but they might also explain the socioeconomic gradient in prescribed medicine use observed in these countries since patients were frequently required to co-pay for out-patient prescribed medicines. Further co-payments in health care could be additional barriers for patients with a lower socioeconomic background to visit a doctor and get a prescription: In the CEECs co-payments for attending a physician were, generally, in place, supplemented by ‘under-the-table payments’.

Overall economic developments are key drivers for policy-makers to curb publicly financed expenditure, even more with the global financial crisis. The Czech Republic, for instance, introduced co-payments to doctors’ visits and significantly increased the value-added tax (VAT) on medicines in 2008, and the share of the private pharmaceutical spending had grown by 57 % within three years. This could explain the socioeconomic gradient in non-prescribed use, and it might also lead to socioeconomic inequalities in prescribed medicine use though this was not confirmed by the EHIS results on the Czech Republic.

Within the CEECs Latvia, Hungary and Romania were particularly impacted by the global financial crisis [59]. Latvia responded through several cost-containment measures [60, 61]. This probably aggravated existing socioeconomic inequalities whose existence in health care (not medicines) had been identified as extensive in Latvia, also in comparison to the other Baltic countries [62]. In Romania, accessibility to prescribed, reimbursable medicines was strongly limited at the time of the survey due to high co-payments for out-patient medicines, and particularly due to the existence of pharmaceutical budgets for pharmacies: Once a monthly financial threshold had been reached a pharmacy was no longer permitted to fill prescriptions [63]. Thus, patients could not access prescribed medication unless they were willing, and able, to pay for it out-of-pocket. In 2007, after years of galloping growth rates in pharmaceutical expenditure, Hungary launched a reform to drastically curb public pharmaceutical expenditure. As a result, the share of private pharmaceutical expenditure (52 %) was rather high at the time of the survey (2009), and it had considerably grown (by 38 %) within the previous three years. This may cause concerns about limited affordability for lower socioeconomic groups [28]. Despite this austere cost-containment strategy no socioeconomic gradient in prescribed medicine use was found. A possible explanation is provided by a study on Hungary [64] according to which affordability of medicines was not negatively impacted by the cost-containment measures. It was argued that the policies were designed in a way to particularly target market participants (industry, distribution actors) rather than private households. Medicine prices were drastically cut, for instance, but this resulted in lower (percentage) co-payments for the patients.

Medicine prices should be affordable for those who pay for them – public payers and private households. There is lack of knowledge about prices of OTC medicines in general [65], and these prices have not been surveyed in the CEECs either. The socioeconomic gradient in non-prescribed medicine use that was found in all eight CEECs potentially implies that patients with a lower socioeconomic background considered medicines, which had to be paid out-of-pocket, as not affordable. In this context, attention should be paid to the value-added tax (VAT) on medicines. In many low- and middle-income countries world-wide, taxes and duties account for a considerable share of the medicine prices, thus contributing to increasing medicine prices up to an unaffordable level for the majority of the population [66]. In the CEECs, the only relevant tax related to the price of a medicine is the value-added tax. In the CEECs the VAT rates on medicines (amounting to 10 % or less) tended to be lower than the standard VAT rates [9, 67] and lower compared to the other EU countries. Nonetheless, they had risen in some countries (Czech Republic, Poland, Romania, Slovenia) from usually 5 % in 2004 [7] to their current rates of between 7.5 and 10 % since an increase in VAT rates was a common cost-containment policy [60]. While it might not appear high, it might still result in a medicine price of non-funded medicines that lower socioeconomic groups were not able, or not willing, to afford.

Little can be said about whether medicine prices were affordable for public payers in the CEECs given the lack of price studies in this region. One study [68] on 23 European countries that also included six CEECs showed that the pharmacy retail price (i.e. price at pharmacy) in the CEECs, except for Slovenia, was considerably lower than in the other European countries. However, unpublished analyses from the Pharma Price Information service of the Austrian Public Health Institute [69] indicated that at the time of the EHIS survey prices for specific new, high-cost medicines (e.g. cancer medicines, orphan medicines) were comparably high. These medicines were typically fully funded by public payers, and this ensured access to higher-priced medicines for the patients. This might explain why a socioeconomic gradient for prescribed medicine use was found in only three of the eight CEECs included in our study.

Limited medicine availability was an issue in the CEECs. However, availability deficits were apparently not linked to the socioeconomic background of the patients, but targeted the whole population, as in the case of Slovenia and Latvia with a low number of medicines authorized and/or on the market. These countries are considered as ‘small markets’ that might not be sufficiently attractive for the pharmaceutical industry to be supplied [70]. Limited availability also resulted from medicine shortages, i.e. that medicines were not supplied in sufficient quantity to pharmacies and hospitals. This was reported to be a major problem in Romania [71]. Furthermore, we found indications of regional inequalities: Despite the generally high pharmacy density in the CEECs (except for Slovenia), major intra-country discrepancies existed in some CEECs (Bulgaria, Czech Republic, Poland, Romania) and resulted in low accessibility to pharmacies in the rural areas. Regarding the regional variation of OTC retailers, based on the experience from Western European countries about a clustering of these dispensaries in towns with already good accessibility [65], it can be assumed that the easier access to OTC medication was likely to primarily serve well-off people in the urban areas who were willing and able to pay for these non-funded medicines. This would contribute to regional inequalities within countries, but depending on the regional distribution of socioeconomic groups can also contribute to the socioeconomic gradient identified in this paper.

A major need for improving the pricing and reimbursement procedure was reported from Bulgaria and Romania whose processes were considered unclear and intransparent. The design of a pricing and reimbursement system is not simply a technical issue but it can considerably impact accessibility and affordability: Unclear and intransparent processes encourage corruption and an irrational selection of medicines into reimbursement, and this, eventually, can limit sustainable public funding. Cumbersome long processes, as observed for Poland, might result in delays in access to needed medicines for patients.

The CEECs have been criticized for not paying enough attention to promoting a rational use of medicines [57] that aims to ensure that patients receive the appropriate medicine at the right dose and at the time they require. The instruments to encourage rational prescribing existed in the CEECs but were apparently not fully enforced. Knowing this, however, we cannot draw conclusions as to whether the medicine use reported in the EHIS survey was appropriate, or not.

Given the limits in funding by both public payers and private households, promoting the uptake of generics is considered as a recommended policy option [72]. The CEECs already had comparably high generic market shares, since local generic industries played a strong role historically, even before the transition process [73], and the use of prescribed generics was promoted by tools such as INN prescribing and generic substitution. Still, high reluctance versus the quality of generics was reported [12]. Thus, awareness-raising campaigns to inform about generics could be a policy to reduce in the observed socioeconomic gradient in non-prescribed use, by motivating people to use lower priced generics.

Summarizing, the policy analysis identified some factors that help explain the observed socioeconomic gradient in non-prescribed medicine use in the eight CEECs and in prescribed medicine use in three of them. These include a high and increasing share of private funding such as co-payments for prescribed out-patient medicines and co-payments beyond medicines (e.g. for attending doctors to get a prescription), limited investment in medicines, and long and intransparent processes regarding the inclusion of medicines into reimbursement. Further deficits such as limited availability of medicines, irrational prescribing and regional variation in the accessibility of pharmacies and OTC retailers are additional factors limiting an equitable access but it could not be determined whether, or not, lower socioeconomic groups are particularly affected. The analysis of the pharmaceutical policies in Latvia, Poland and Romania provides an indications that actually these policies contribute to the observed socioeconomic inequalities in prescribed medicine use in these countries. While the three countries cannot be seen as a very homogenous cluster in terms of pharmaceutical policies, they share an accumulation of policies that do not adequately address inequalities in medicines use. Similarly, for Bulgaria, the policy framework observed would suggest a socioeconomic gradient in prescribed medicine use. This, however, was not confirmed based on the EHIS data. Also, it might have been possible that cost-containment policies would have had a negative impact on equitable access to medicines in the Czech Republic and in Hungary. In this respect, note that the only two other studies [19, 20] on socioeconomic determinants in medicine use in CEECs, focussing on adolescents in Slovakia and on cardiovascular medicines in Czech Regions did identify a socioeconomic gradient. Though pharmaceutical policies were apparently successful in reducing socioeconomic inequalities in medicine use in some CEECs, further factors were likely to have an impact. These might be factors at the individual level of the patients (e.g. adherence to medicines) as well as overall political and economic factors, such as the stability of a health care system, patients’ trust into it and the impact of the transition processes as observed in the CEECs in the 1990-ties. Finally, the non-existence of a socioeconomic gradient in medicine use can still imply any under-use, over-use or misuse of medicines in general, and overall limited availability and accessibility of medicines for the patients of that country.

This study does not come without limitations. First, the quantitative analysis is based on self-reported information and hence comes with a risk of measurement errors. However, at the individual country level, where data was collected, several quality control steps were taken [74] and data comparability was also ensured by Eurostat, based e.g. on a standard questionnaire, conceptual guidelines and rationales as well as a standard translation protocol [75, 76]. Data was collected either via a stand-alone survey or as part of another (health or non-health) survey (i.e. national health interview survey, labour force survey, other household survey) based on face-to-face interviews, telephone interviews, self-administered questionnaires (or a combination of these means) [77]. Consistency and integrity checks on the national EHIS data were carried out by Eurostat, and quality reports were also disseminated [74]. Also, household surveys are an acknowledged standard tool in pharmaceutical consumption research according to the WHO [78]. Second, the EHIS household surveys goes beyond the regulatory definition of medicines in national legislations since ‘dietary supplements such as herbal medicines or vitamins’ were part of a question on medicine use (Table 3). To find out if this influences our results, sensitivity analyses were carried out excluding both the use of ‘vitamins, minerals or tonics’ or/and ‘some other type or medicine or supplement’, respectively. While the socioeconomic gradient was generally found to be less pronounced but still statistically significant in these sensitivity analyses, the qualitative conclusions regarding the socioeconomic gradient identified in the main analysis remained the same. Third, interviewees were asked about their use of ‘medicines prescribed or recommended by a doctor’ and ‘medicines not prescribed or recommended by a doctor’ (Table 3). Since we had no information about medicines recommended but not prescribed, we interpreted these two categories of ‘prescribed’ and ‘non-prescribed use’. We based this approach on the assumption that doctors tend to believe that patients expect a prescription [79]. Fourth, the analysis refers to different time periods between 2007 and 2009. We investigated the pharmaceutical policy framework at the time of the EHIS survey at national level, but for comparing national data (e.g. expenditure) versus a European or CEE average a specific year (2008) was taken as reference. This was, in some cases, not the year of the survey. Also, the survey comes with different national sample sizes. This seems especially relevant for Poland, where the analysis is based on a comparatively large study population, and thus even small differences between socioeconomic groups may become statistically significant. Finally, non-use of medicines can have numerous causes. We explored whether the pharmaceutical policy framework impacts, positively or negatively, on socioeconomic (in)equalities. However, as the wealth of literature on adherence rightly points out, there are several other reasons for not using medicines or stopping prescribed medication prematurely. To identify the relative importance of the various factors as co-determinants of the use or non-use of medicines, in-depth investigation for single countries is needed in future research.