Economic inequality in satisfaction with healthcare in the Baltic countries during and after the economic crisis (2008–2014)

ABSTRACT

Public satisfaction with healthcare systems is an important dimension of healthcare legitimacy. The paper analyzes how satisfaction with healthcare systems depended upon the economic situation of our respondents during and after the economic recession in the Baltic states. The results show that there were no differences in public satisfaction with healthcare between better and worse off people in Estonia (except in 2009). In Lithuania, however, satisfaction among the economically better off was higher compared to other groups between 2008 and 2014. In Latvia, inequality in satisfaction between groups in different economic situations became significant from 2010.

KEYWORDS:

• Healthcare system
• satisfaction with healthcare
• Baltic states
• inequality

Introduction

The Baltic states are among the countries that were hit hardest by the economic crisis. This resulted in spending cuts in all the domains of social policy including healthcare. Although Estonia and Lithuania have similar healthcare systems (Social Health Insurance System) and Latvia has a different one (National Health Care System), each of the countries had similar and different reactions to the crisis. In all three countries, there were policies applied to reduce the price of medical goods, to improve the rational use of pharmaceuticals, and aiming to decrease inpatient care in favor of outpatient care (Karanikolos et al. 2013; Mladovsky et al. 2012). The socioeconomic inequality in health, as well as socioeconomic inequality in access to healthcare in the Baltic states was already much higher than in western Europe before the crisis (Mackenbach et al. 2008; Plug et al. 2012; Stirbu et al. 2010).

Healthcare satisfaction can be considered one of the most coherent indicators of the general subjective evaluation of healthcare, and public satisfaction with healthcare systems is an important dimension of healthcare legitimacy, regarding citizens’ evaluations of how their government has actually implemented healthcare services relative to what it promised (Rothstein 2001). This article studies the change in public satisfaction with the healthcare system during and after the economic crisis (2008–2014) in the Baltic countries, and analyzes how healthcare satisfaction depends upon a person’s economic situation. More specifically, how an individual’s economic situation is related to satisfaction with healthcare systems. During the crisis, several policy changes had a direct cost effect on patients, which might potentially have meant a greater cost barrier for the more economically vulnerable. We assume that the trends in distinctions with satisfaction with healthcare between groups in different economic situations are a good indicator of the adaptability of the healthcare system to the crisis in the three Baltic countries during the crisis years.

The economic crisis hit southern European countries (such as Greece, Spain, and Portugal) hard, as well as the Baltic countries (especially Latvia), but there have been many more studies on health and healthcare in relation to the economic crisis regarding the Southern European countries (e.g. Kentikelenis et al. 2014; Karanikolos et al. 2013; Kondilis et al. 2013; Simou and Koutsogeorgou 2014) than on the Baltic countries (an exception is Karanikolos et al. 2016). There are several studies about the impact of the economic crisis on health and socioeconomic inequality (Karanikolos et al. 2013; Kentikelenis et al. 2014; Kondilis et al. 2013; Porthé et al. 2017; Simou and Koutsogeorgou 2014) but satisfaction with healthcare has also not been studied much in relation to the economic crisis (some exceptions are AlSaud, Taddese, and Filippidis 2018; Porthé et al. 2017). This study fills the gap regarding the impact of the economic crisis on public satisfaction with healthcare and the differences between economic groups in public satisfaction with healthcare in the Baltic states.

This article first provides an overview of the concept and the determinants of public satisfaction with healthcare followed by an analysis of how the Baltic healthcare systems create inequalities, and the changes in healthcare during the economic crisis that could also be related to inequalities, and therefore influence the satisfaction of the different socioeconomic groups. Thereafter, the data (Eurobarometer 2008–2014) and methods are explained; after which the results are presented by country, different groups are compared in logistic regression models, and the results are discussed in the context of healthcare systems and policy changes during the recession.

Satisfaction with the healthcare system and its determinants

According to the theory of contingent consent (Rothstein 2001), satisfaction with healthcare systems is considered an important dimension of the popular legitimacy of healthcare systems (Missinne, Meuleman, and Bracke 2013), as perceived legitimacy depends on public approval of the way a certain policy is implemented in practice (procedural justice). Thus, satisfaction refers to people’s evaluation of the performance of an existing healthcare system (Wendt et al. 2009). Public satisfaction with healthcare systems is influenced by many different factors, for example, experiences in healthcare (Missinne, Meuleman, and Bracke 2013; Wendt et al. 2009; Bleich, Özaltin, and Murray 2009; Footman et al. 2013), expectations about the organization of healthcare (Rothstein 2001; Missinne, Meuleman, and Bracke 2013), economic factors (Missinne, Meuleman, and Bracke 2013), the historical background of the country, as well as the wider political and institutional context (Footman et al. 2013; Aidukaite 2009), personal health, wellbeing, and socioeconomic status (Bleich, Özaltin, and Murray 2009; Footman et al. 2013; Missinne, Meuleman, and Bracke 2013; Wendt et al. 2009; Munro and Duckett 2016; Ainsaar and Nahkur 2017). In the following section, we provide an overview of the factors that influence public satisfaction with healthcare systems and focus on socioeconomic factors.

Several studies have found that economic factors are related to public satisfaction with healthcare systems. People who are less satisfied with their household economic situation (Footman et al. 2013) or for whom it is difficult to get by with their household income (Wendt et al. 2009), those who have below average income (Blendon et al. 2002), and those who do not have health insurance (Munro and Duckett 2016), tend to be less satisfied with healthcare systems. In contrast to the impact of income, people with lower education tend to be more satisfied with healthcare because they might have lower expectations (Bleich, Özaltin, and Murray 2009; Footman et al. 2013; Missinne, Meuleman, and Bracke 2013). Moreover, there is some evidence that men (Missinne, Meuleman, and Bracke 2013), older people (Bleich, Özaltin, and Murray 2009; Munro and Duckett 2016) and people living in rural areas (Footman et al. 2013) are, in some countries, more satisfied with healthcare systems, due to lower expectations of healthcare. In contrast to Western countries, in countries of the former Soviet Union, younger people were more satisfied with healthcare systems compared to older people (Footman et al. 2013). Another important factor is self-rated health. People with better health tend to be more satisfied with the healthcare system (Ainsaar and Nahkur 2017; Bleich, Özaltin, and Murray 2009; Footman et al. 2013; Missinne, Meuleman, and Bracke 2013; Munro and Duckett 2016; Wendt et al. 2009). In a recent study, Ainsaar and Nahkur (2017) found that lower satisfaction with healthcare among less healthy people was mainly explained by the lower incomes of the same group in Estonia.

Footman et al. (2013) also found that in former Soviet countries, trust in political institutions was a strong determinant of healthcare system satisfaction. Ideological beliefs (Missinne, Meuleman, and Bracke 2013), institutional characteristics like the level of government expenditure on health (Missinne, Meuleman, and Bracke 2013), or the density of general practitioners (GPs) (Wendt et al. 2009), and media (Munro and Duckett 2016) also influence public satisfaction with healthcare systems.

In relation to economic crisis it has been found that satisfaction with healthcare declined during the crisis and this decrease was the greatest in the countries, where the crisis had the most impact, but in the countries where the national per capita income was higher and where healthcare expenditure was higher, people were more satisfied with their healthcare (AlSaud, Taddese, and Filippidis 2018).

Healthcare in the Baltic countries and the impact of economic recession

In the Baltic countries, there are several problems related to healthcare: inequality, corruption, and low levels of satisfaction (see Aidukaite 2009, 306) in combination with unmet needs (Aaviksoo and Sikkut 2011). The lower level of satisfaction in eastern European countries compared to western European countries can be explained by the expectations of the government’s role in providing healthcare, reflections on demographic pressures, and especially the perception of the performance of the healthcare system (efficiency, equality, and health outcomes) (Schneider and Popic 2018). The following section gives an overview of the healthcare systems in the Baltic states, focusing on access to healthcare for different population groups that might influence the satisfaction with healthcare of these groups. During the studied period (2008–14), Latvia used a national healthcare system (NHS), where the whole population was theoretically equally covered, but it also had substantial user fees even in the public system and additional out of pocket payments to shorten waiting times (Mitenbergs et al. 2012, 49). Therefore, despite the NHS, the user fees created a considerable degree of inequality in Latvia. Estonia and Lithuania had a social health insurance system (SHI) (Lai et al. 2013, 59; Waters et al. 2013), where coverage was obtained through employment based contributions, but the government also covered some special groups (those eligible for any kind of pension or social assistance, children, students, pregnant women, women on maternity leave, the registered unemployed, disabled people and their carers, people on military service, etc.) (Eesti Haigekassa 2017; Lai et al. 2013; Murauskiene et al. 2013; National Health Insurance Fund 2014). As a result of these entitlement rules, population coverage was 95% in Estonia and 91% in Lithuania (Habicht 2016; Kacevicius 2016; Lai et al. 2013, 68). Despite the different initial types of health insurance systems, Estonian and Lithuanian systems worked similarly to the NHS because of the high coverage, but the Latvian system, despite being an NHS, created a high level of inequality because of the high user fees.

Along with universal or near-universal coverage, the role and functioning of primary care can also be a factor that contributes to creating equality.¹ Detollenaere et al. (2017) found that strong primary care systems equalize income inequality in access to healthcare. Strong primary care is a priority for all three Baltic countries, and it is more efficient than in many other European countries (Kringos et al. 2015; van Ginneken et al. 2012, 1). In this case, Latvia once again has a slightly less equal system. Unlike in Estonia and Lithuania, in Latvia the ‘gatekeeping’ role of the family doctors is not very strong (van Ginneken et al. 2012, 1–2; Groenewegen et al. 2013).²

The economic crisis in 2008/2009 was a challenge for every social policy domain, including the healthcare sector. It resulted in a decrease in per capita healthcare expenditure in all Baltic countries (Eurostat 2017). The decrease in per capita public spending in the medical sector occurred in all three Baltic states in 2009 and 2010 (Thomson et al. 2014, 9), although total health expenditure as percentage of Gross Domestic Product increased in Estonia and Lithuania (World Health Organization 2016). In Latvia, the per capita reduction in healthcare financing was the greatest among European countries (comparing the 2012 level to 2007) (Thomson et al. 2014, 11).

Despite the great effect of the economic crisis on healthcare system expenditures in the Baltic states, the economic crisis did not change the general character of the healthcare systems in the Baltic states. In addition, all Baltic countries also managed to avoid sharp rises in out of pocket expenditures by using different strategies during the years of the crisis (van Ginneken et al. 2012). Out of pocket payments even decreased in Estonia by around 15%, but increased by around 7% in Latvia and Lithuania (Thomson et al. 2014, 10). Estonia and Lithuania were better prepared for the crisis because they also started applying some policy measures before the crisis, which included holding reserves ringfenced for healthcare funding (Karanikolos et al. 2013, 1324).

There were also major changes in the Latvian and Lithuanian hospital sectors. This restructuring involved closures, mergers, and centralization, there was a shift towards outpatient care and, in addition, Lithuania’s primary care coordination was improved (Mladovsky et al. 2012, 17–18, 26; Thomson et al. 2014, 25–26). In Estonia, dental care reimbursements were lowered (Mladovsky et al. 2012, 17–18) and waiting times for all medical treatments were increased (Thomson et al. 2014, 26). In Lithuania, salaries in the medical sector were reduced, which caused greater emigration of medical specialists to other countries (Karanikolos et al. 2013, 1325; Stuckler et al. 2010, 304).

An indicator that shows access to healthcare is unmet need. A study of unmet need in the Baltic states between 2005 and 2012 showed that Latvia had a considerably higher proportion of unmet medical need during that period compared to other Baltic countries (Karanikolos et al. 2016). Of the specific reasons for unmet need, costs have been the greatest problem in Latvia over 2005–12 (Thomson et al. 2014, 32), peaking at 2011 (15%) and decreasing after that until reaching the lowest level in 2016 (5%) (Eurostat 2017). In Estonia and Lithuania, this problem has been minimal, rather waiting lists have been the greatest problem in these two countries (Thomson et al. 2014, 32). The problem is incomparably greater in Estonia, steadily increasing from 3% in 2009 to 14% in 2016 (in Lithuania it was around 2% throughout this period) (Eurostat 2017). The unmet need of the poorest quintile due to costs was already the highest in Latvia in 2008, but had greatly increased by 2012 (Thomson et al. 2014, 32). Eurofound’s European Quality of Life Survey shows that when comparing 2007 and 2011 in Estonia, the problem of costs, as well as waiting times as an access barrier increased during the economic crisis for both bottom and top income earners, but only for bottom income earners in Lithuania (Dubois and Anderson 2013).

Methodology

Although satisfaction with healthcare systems is considered an important dimension of their legitimacy, the studies of satisfaction with healthcare are characterized by the lack of a universally accepted definition or measure, and by a dual focus: while some researchers focus on patient satisfaction and the quality of healthcare services received, others focus on public satisfaction with the healthcare system more generally (Bleich, Özaltin, and Murray 2009). In this paper, we focus on public satisfaction with healthcare systems, measuring it with a question ‘How would you judge the current situation in healthcare provision in x country’ from the Eurobarometer.³ Although public satisfaction with healthcare systems has been previously measured by directly asking about satisfaction with a country’s healthcare system (Footman et al. 2013; Wendt et al. 2009), other measures have been used as well. For example, it is also measured by asking about the way healthcare is run (Bleich, Özaltin, and Murray 2009; Kotzian 2009; Munro and Duckett 2016), the need for changes in the healthcare system (Blendon et al. 2002; Wendt et al. 2009), the overall state of health services (Ainsaar and Nahkur 2017; Missinne, Meuleman, and Bracke 2013), and the efficiency of healthcare provision (Missinne, Meuleman, and Bracke 2013) in a respondent’s country.⁴ Our methodology corresponds to this list well.

We are interested in the variance of healthcare satisfaction associated with the economic situation of a respondent’s household. We measured this using a question about the difficulty of paying bills, with the categories: most of the time, from time to time, almost never/never.⁵ We use Eurobarometer data from 2008, 2009, 2010, 2011, and 2014 on Estonia, Latvia, and Lithuania to study the differences between economic groups in public satisfaction with healthcare systems during the economic recession in the Baltic states. Unfortunately, the data on public satisfaction with healthcare was not measured by Eurobarometer in 2012 and 2013, and therefore we have a gap for these years. Table A1 gives an overview of the sample sizes by year and country, as well as the proportions of the respondents in the categories of the variable of difficulty of paying bills by year and country. Eurobarometer is a study carried out by Eurostat in member and candidate states of the European Union. It is based on a multi-stage (stratified by metropolitan, urban and rural areas, i.e. proportional to the population size, for total coverage of the country and population density) probability sample of the national, resident population above 18 years of age (Survey Methods 2013; Gesis 2017).

Binomial logistic regression models were run separately for every country and every year (2008, 2009, 2010, 2011, and 2014). As there have been indications that binomial logistic regression coefficients might be biased and not suitable for comparing between different samples (like different countries and years) we controlled our results with linear probability models and average marginal effects as suggested by Mood (2010). All of our conclusions hold in the case of these models.

In our models, we controlled for gender, age (we also added age squared to cover the nonlinear relationship between age and healthcare satisfaction), the age of finishing education (proxy for education), type of municipality respondents lived in (large city, small town, rural area), and assessment of the current state of public administration (very good, rather good, rather bad, very bad; coded to binomial variable with values good and bad). We controlled attitudes about the current state of public administration because it has been previously found that trust in political institutions (Footman et al. 2013) and general satisfaction with the government (Ainsaar and Nahkur 2017) were strong determinants of healthcare system satisfaction.

Results

As the relationship between economic situation and satisfaction with healthcare provision is the main interest of this article, we focus on this relationship in presenting our results. Figure 1 shows the trends in satisfaction with healthcare provision in all three Baltic states from 2008 to 2014.

Figure 1. Proportion of respondents who think that the state of healthcare provision in their country is very good or rather good in 2008–14.

Figure 1 shows that satisfaction with healthcare has been the highest in Estonia throughout the whole period. The proportion of people in Estonia who considered healthcare provision rather good or very good has increased from 52% in 2008 to 63% by 2010 but slightly decreased after that (to 57% by 2014). In Latvia, this percentage was the lowest of the Baltic countries and it further decreased during the period under study from 34% in 2008 to 28% in 2014. The proportion of people considering healthcare provision in Lithuania good or very good (35%) was very similar to Latvia in 2008; it increased by 5% points during 2009, was quite stable during the greatest years of the crisis, 2009–11, and increased to a similar level with Estonia by 2014 (52%).

Figure 2 presents the relationship between satisfaction with healthcare and the economic situation of the respondents.

Figure 2. Proportion of people who consider healthcare provision in their country very good or rather good, depending on their economic situation (difficulty paying bills) in 2008–14.

Figure 2 shows that, in Estonia, differences in satisfaction with healthcare provision between groups in different economic situations did not change in 2008, in 2009 the richest were more satisfied, and by 2014 the poorest group had lower satisfaction compared to those who had no difficulty paying bills. In Latvia, the group with difficulties paying bills most of the time had lower levels of satisfaction from 2011. In Lithuania, in 2009–14 the richest were more satisfied with healthcare system compared to other groups, similar to Estonia, the group in the worst economic situation (difficulties paying bills most of the time) became less satisfied compared to other groups by 2014 and the level of satisfaction with healthcare provision was similar to Estonia for all groups in 2014. In Lithuania, in 2014, 36% of those who had difficulties paying bills most of the time considered healthcare provision rather good or very good, as did 50% of those who had occasional difficulties and 59% of those who never had difficulties paying bills.

Satisfaction trends can be shaped by the composition of the population structure in different countries, and therefore, we next ran binomial logistic regression models controlled for age, place of residence, gender, satisfaction with public administration, and age at graduation of highest educational level. Tables 1–3 present binomial logistic regression models separately for all the years (2008, 2009, 2010, 2011, and 2014), respectively for Estonia (Table 1), Lithuania (Table 2), and Latvia (Table 3).

Table 1. Logistic regression models of Estonia for assessment of good or very good healthcare provision in the country (reference bad or very bad).

		2008		2009		2010		2011		2014
Independent	Variables	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig
Difficulty of paying bills	From time to time	1.08		1.31		1.39		1.07		1.69
(ref: most of the time)	Almost never	1.35		1.95	*	1.66		1.25		1.55
Age of	Finishing education	1.01		1.04		1.03		1.01		1.06	**
Place of	Small town	1.14		1.78	**	0.97		1.28		0.95
Residence (ref: large city)	Rural	1.15		2.13	***	1.33		1.04		0.89
Gender	Female	0.96		1.12		0.84		0.93		0.89
Age		0.98		0.95		0.96		0.97		0.98
Age squared		1.00		1.00		1.00		1.00		1.00
State of public administration (ref: bad)	Good	3.27	***	2.59	***	2.29	***	3.11	***	3.56	***
Intercept		0.69		0.38		0.97		0.77		0.18
R^2	(Nagelkerke)	0.49		0.42		0.42		0.54		0.56

Notes: ***p < 0.001; **p < 0.01; *p < 0.05.

Table 2. Logistic regression models of Lithuania for assessment of good or very good healthcare provision in the country (reference bad or very bad).

		2008		2009		2010		2011		2014
Independent	Variables	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig	Exp (B)	Sig
Difficulty of paying bills	From time to time	1.03		1.40		1.02		1.12		1.59
(ref: most of the time)	Almost never	2.57	**	1.84	*	1.90	***	1.62	*	2.25	***
Age of	Finishing education	0.92	*	1.01		1.00		1.01		1.07	**
Place of	Small town	1.13		1.02		1.08		1.09		0.7
Residence (ref: large city)	Rural	1.18		0.80		1.35		1.56		0.95
Gender	Female	1.06		0.77		1.15		0.87		0.97
Age		0.89	***	0.93	*	0.94	*	0.95		0.92	**
Age squared		1.01	***	1.01	*	1.01	*	1.00		1.01	**
State of public administration (ref: bad)	Good	3.30	***	3.42	***	2.42	***	1.91	***	3.49	***
Intercept		20.04	**	1.21		1.23		0.94		1.05
R^2	(Nagelkerke)	0.74		0.65		0.51		0.58		0.65

Notes: ***p < 0.001; **p < 0.01; *p < 0.05.

Table 3. Logistic regression models of Latvia for assessment of good or very good healthcare provision in the country (reference bad or very bad).

		2008		2009		2010		2011		2014
Independent	Variables	Exp(B)	Sig	Exp(B)	Sig	Exp(B)	Sig	Exp(B)	Sig	Exp(B)	Sig
Difficulty of paying bills	From time to time	1.47		0.93		1.54		1.55	*	2.11	*
(ref: most of the time)	Almost never	1.24		1.39		1.47		1.8	**	2.41	**
Age of	Finishing education	0.99		1.00		1.00		1.03		1.02
Place of	Small town	1.51	*	1.49	*	0.58	*	1.03		0.86
Residence (ref: large city)	Rural	1.69	**	1.17		1.21		1.26		1.07
Gender	Female	0.86		0.68	*	1.21		1.16		0.90
Age		0.98		1.06		0.90	**	0.97		0.88	***
Age squared		1.00		1.00		1.01	**	1.00		1.01	**
State of public administration (ref: bad)	Good	2.89	***	3.68	***	5.19	***	1.96	**	4.38	***
Intercept		0.43		0.17	*	2.26		0.26		1.69
R^2	(Nagelkerke)	0.46		0.39		0.41		0.38		0.49

Notes: ***p < 0.001; **p < 0.01; *p < 0.05.

The models show that difficulties paying bills was related to satisfaction with healthcare in all three Baltic states, but there were different trends depending on the country. In Estonia, the economic difficulty differentiated people in healthcare satisfaction only in 2009, when the crisis reached its full impact. Those who never had difficulties in paying bills were almost two times more likely to consider healthcare provision in Estonia rather good or very good compared to those who had difficulties most of the time. In the following years the difference between economic groups was not statistically significant. In 2009, people in rural areas and small towns were approximately two times more likely to consider healthcare provision very good or rather good compared to large cities in Estonia. The 2009 effect can be related to the increase in maximum waiting times. They increased from four weeks to six weeks (Thomson et al. 2014, 26; Lai et al. 2013, 150). The reason people in large cities were less content could be that they were more influenced by longer waiting times because of the higher population density, whereas in the countryside and small towns not much changed.

Although Lithuania had a very similar healthcare system to Estonia (Lai et al. 2013; Waters et al. 2013), although with somewhat lower coverage (Habicht 2016; Kacevicius 2016; Lai et al. 2013), the richest group differed from the other groups during the whole period in Lithuania (2008–14). This difference already existed prior to the economic crisis year of 2008. The people who almost never had difficulties in paying bills compared to those who had difficulties most of the time were about two and half times more likely to consider the healthcare system rather good or very good in 2008, this difference decreased a bit during 2009–11 by two times but increased by 2014 almost back to the 2008 level.

Looking at the relationship between the economic situation of the respondents and their healthcare satisfaction in Latvia, we see of emergence, and clear increase, in inequality from 2010.⁶ In 2010, those who almost never had difficulties paying bills and those who had difficulties from time to time were both 50% more likely to assess healthcare provision in Latvia as rather good or very good compared to those who had difficulties paying bills most of the time. By 2014, the difference between those who never had difficulties paying bills and those who had difficulties most of the time, was close to three times, and the difference between those who had difficulties from time to time and those who had difficulties most of the time was about two times.

All models also demonstrate that satisfaction with the running of public administration in the country was closely related to satisfaction with healthcare provision in all the years and all Baltic countries. People who were more satisfied with public administration were also more satisfied with the healthcare system. Those who considered the state of public administration in their country rather good or very good were at least two times more likely to be rather or very satisfied with the healthcare provision in their country in all of the years. This finding is in line with the results of a previous study conducted in former Soviet countries (Footman et al. 2013). Other variables did not have a very clear relationship in satisfaction with healthcare provision over time in the models and were added as controls.

Discussion and conclusions

Our research contributes to the discussion about the impact of economic recession on groups with different economic resources and their satisfaction with healthcare systems between 2008 and 2014 in the Baltic countries. In all three Baltic countries, the level of public satisfaction with their healthcare systems was different and the reaction to the crisis was also different. In Estonia, the proportion of people who considered healthcare provision rather good or very good was the highest of all of the Baltic countries and, surprisingly, it even continued to rise during the first years of the crisis but fell again afterwards. In Latvia, the level of satisfaction with healthcare was the lowest and fell further during the crisis, this trend continued until 2014. In Lithuania, the proportion of people who considered healthcare provision in their country good or very good did not change during the crisis and increased afterwards, reaching the Estonian level. In the European context, Latvian and Lithuanian results comply with the general trend of worsening satisfaction during the crisis (in Estonia it was not the case). Furthermore, in accordance with previous findings that the more severe the impact of the economic crisis the lower the satisfaction with healthcare (AlSaud, Taddese, and Filippidis 2018), this was also evident in Latvia’s case, which was the Baltic state hit hardest by the economic crisis. Additionally, in agreement with the research of AlSaud, Taddese, and Filippidis (2018), in Estonia, where the healthcare expenditure per inhabitant has been the highest (Eurostat 2017), satisfaction with healthcare was the highest, followed by Lithuania in terms of healthcare expenditure, as well as satisfaction.

In general, there are differences in public satisfaction with healthcare between groups in different economic situations, as also shown by previous studies (Wendt et al. 2009; Footman et al. 2013; Blendon et al. 2002). Yet, the economic resources and satisfaction with healthcare had a different relationship in all three Baltic countries throughout the economic crisis. Studies of other areas of inequality (e.g. the gender wage gap or income inequality) showed a decrease in inequality during the recession in Estonia and Lithuania (Karu et al. 2010; Roots 2011; Šilingienė and Radvila 2016). This trend was not fully confirmed in the case of satisfaction with the healthcare systems in Estonia and Lithuania. Estonia is remarkably equal in satisfaction with healthcare throughout the period and the differences between groups with different economic situations appeared only in 2009, the year, when the recession reached its deepest point. Looking at Figure 2, we see that compared to the previous year, in 2009 the level of satisfaction decreased for the worst off, increased for the best off, and did not change for the middle group in Estonia. It can be that the hardships of the economic crisis hit the worst off first (because they possessed fewer resources and had a less secure position on the labor market) and this is what we also see in the results. In Lithuania, inequality in satisfaction with healthcare provision with the worst and best off was persistent throughout the whole period (2008–14). When we take a closer look at the regression models of Lithuania (Table 2), although the differences between the best and the worst off were statistically significant throughout the entire period, the coefficients were smaller in 2009–11 than before or after the most severe crisis years, so it is possible to observe a decrease in inequality during the recession.⁷

In Latvia, there was growing economic inequality from 2010 in satisfaction with the healthcare system as it also happened in other areas of society. For example, gender wage gap increased during the crisis (Šilingienė and Radvila 2016). In Latvia, the satisfaction with healthcare of those whose economic situation was the worst decreased during the economic crisis. This could be explained by the interaction between healthcare system arrangements and economic recession. Previous studies have shown that in Latvia the cost barrier in healthcare utilization was the greatest due to high user fees (Habicht et al. 2009) and this barrier increased during the economic crisis (Karanikolos et al. 2016). In current analysis, we see that it has also influenced satisfaction with healthcare provision in Latvia because the differences in satisfaction with healthcare increased between groups who were in different economic situations. It was not the case in the other two Baltic countries.

In conclusion, we can say that we see the interaction between healthcare system and economic recession impacting economic inequality in public satisfaction with healthcare. In Estonia and Lithuania, where the healthcare system was more equal by design (with near universal coverage, without substantial user fees, and strong primary care), there were no lasting effects of economic recession that would have increased the inequality between groups in different economic situations. The situation in Latvia, however, was different, as the design of the healthcare system already created more inequality compared to other Baltic countries (substantial user fees and the lower gatekeeping role of the primary care system), the impact of the economic crisis seems to add to this process and we see an increase of inequality in satisfaction with healthcare between groups in different economic situations.

Limitations of the study

A major limitation of the comparative research of the Baltic countries is the lack of adequate comparative data for all Baltic countries. There is, for example, the question about the satisfaction with healthcare, and very good socio-demographic and other control variables in the European Social Survey, but, in many survey waves, the Latvian and Lithuanian data is missing. European Union Statistics on Income and Living Conditions (EU-SILC) include all Baltic countries over time, but misses the healthcare satisfaction variable.

The Eurobarometer is one of the rare studies that covers the three Baltic countries in many waves, as well as the trend of the healthcare satisfaction variable. The serious problem with Eurobarometer, however, is the meagre number of socioeconomic variables and often a lack of essential socioeconomic variables. Unfortunately, the study does not include important variables like directly measured income, directly measured education, migration background, or health status and healthcare utilization. For income, we had to use a proxy (difficulty of paying bills) and fortunately that was quite an adequate substitute for income.⁸ For education, age of graduation is used, but this variable is unfortunately not a good proxy; a universal education scale like the International Standard Classification of Education (ISCED), or the European Social Survey’s EISCED, would have been preferable. Lack of relevant data also presents some obstacles for our interpretation of results, which require more information to draw more robust conclusions.

Finally, for studying trends and processes, longitudinal data would, of course, have been needed, but currently there is no such dataset available for the Baltic countries that would allow for the study of public satisfaction with healthcare during the chosen period.

Acknowledgments

We are very thankful to the NORFACE project Healthdox for funding this research. We are very grateful to Jolanta Aidukaite and Sven Hort for their useful and helpful comments on the first version of this article.

Additional information

Funding

This work was supported by the NORFACE network under Healthdox project.

Notes on contributors

Ave Roots

Ave Roots is a Research Fellow of Social Policy in the Institute of Social Studies at University of Tartu. She has received her PhD in Sociology at the University of Tartu. Her main research interests are socioeconomic inequality in health and access to health care in different healthcare systems. She is also interested in topics on Labor Market, Gender Segregation, Social Mobility and Inequality.

Mare Ainsaar

Mare Ainsaar is a Senior Researcher in the Institute of Social Studies at University of Tartu. She has received her PhD in Social Sciences at the University of Turku, and worked later in different universities. She is a National Coordinator of European Social Survey in Estonia since 2003. Her main research interests are related to Demographic Behavior, Family Policy and Social Policy.

Oliver Nahkur

Oliver Nahkur is a PhD candidate in Sociology and Junior Research Fellow in the Chair of Social Policy, Institute of Social Studies at University of Tartu. His doctoral dissertation focuses on the construction, validation and application of a new social indicator - Societal Index of Interpersonal Destructiveness (SIID) – that could be regularly used to compare interpersonal violence in different societies. He is also interested in the public attitudes toward healthcare and its determinants.

Notes

1. Primary care is the first level of healthcare, the general level, where people present their health problems, and it is focused on the person as a whole, instead of on only one specific organ or health problem, general practice or family practice is often considered to be the core of primary care (Kringos et al. 2015).

2. The gatekeeping role of primary care means that it is the entry point to the healthcare system and it coordinates patients through the system (Detollenaere et al. 2017, 1).

3. The questionnaire measured answers on a four-category scale: very good, rather good, rather bad, very bad. We coded it into a dichotomous variable: good (including very good and rather good) and bad (including very bad and rather bad) (it was coded the same way in AlSaud, Taddese, and Filippidis 2018).

4. As the question was asked from users, as well as from non-users of healthcare services, the measure can be considered as a more general measure of public satisfaction with healthcare systems as opposed to patient satisfaction with the quality of healthcare services received.

5. The scale of the 2008 survey of this variable was: totally agree, tend to agree, tend to disagree, totally disagree. We coded totally agree equal to most of the time, tend to agree and tend to disagree equal to from time to time and totally disagree equal to almost never/never. Figures 1 and 2 show that this coding results in quite steady trends compared to other years, and seems to produce comparable outcomes.

6. We tested the inequality dynamics with the year fixed effects model, where the country level difference of the proportion of those who considered healthcare very good or good among those who had difficulties in paying bills all the time and those who had no difficulties of paying bills was the dependent variable, controlling for age, gender, age of finishing education, community size of the place of residence, and satisfaction with public administration. This model confirmed the increase in inequality.

7. We also tested the inequality dynamics with the year fixed effects model, where the country level difference of the proportion of those who considered healthcare very good or good among those who had difficulties in paying bills all the time and those who had no difficulties of paying bills was the dependent variable, controlling for age, gender, age of finishing education, community size of the place of residence, and satisfaction with public administration. This model confirmed the decrease in inequality in 2010 and 2011 in Lithuania.

8. We compared the relationship between the assessment of household income and satisfaction with healthcare with European Social Survey data with the data available and the general trends comply with our findings in this article.

Appendix

Table A1. Descriptive statistics of difficulty of paying bills by country and year.

	Difficulty of paying bills	2008	2009	2010	2011	2014
Estonia	Most of the time	11	9	9	6	8
	Time to time	49	24	28	25	21
	Never or almost never	40	67	63	69	71
	%	100	100	100	100	100
	N	1000	1007	1000	1001	1012
Latvia	Most of the time	25	21	24	25	19
	Time to time	59	36	39	35	35
	Never or almost never	16	43	37	40	46
	%	100	100	100	100	100
	N	1002	1012	1016	1019	1016
Lithuania	Most of the time	24	22	28	27	17
	Time to time	67	39	35	40	36
	Never or almost never	9	39	37	33	47
	%	100	100	100	100	100
	N	1011	1022	1036	1026	1013