In recent years, there has been growing interest regarding the relationship between political ideology and health status. Strong associations between voting patterns, social deprivation and mortality were reported in studies conducted in England and Ireland ^{1– 3} . These studies, in general, observed reduced mortality rates in areas where the majority voted for the conservative party. This effect has been explained in relation to socio-economic status; people with better socio-economic status tend to be more conservative and thus, support a party that will improve their existing privileged condition ¹ . Studies conducted in the USA have also reported similar findings. Kondrichin and Lester ⁴ analysed data from the 1980 election that pitted Ronald Reagan against Jimmy Carter and found an inverse correlation between mortality and the percentage voting for the Republican Party. Studies assessing political affiliation at an individual level also suggest that political ideology correlates with the degree of the importance survey correspondents attach to health care ^{5, 6} . These studies have indicated that Republicans are healthier and also assign a lower priority to issues related to health care than Democrats. It could also be possible that a reverse association exists whereby individuals with poor health are more likely to be influenced by ‘leftist’ ideology; however, this investigation requires longitudinal data on health and political ideology which is not available ⁷ .

Contemporary Poland lacks a clear social cleavage that could determine the shape of a stable party system because of its unique history ⁸ . The Polish political scene has been somewhat unstable since the collapse of communist rule in 1989. It has been regularly exposed to political shocks, splits and transformations. It has been postulated that as a consequence, the political preferences of voters have not been very stable, with evidence demonstrating significant flows between individual parties and coalitions ⁹ . Moreover, since the re-establishment of parliamentary democracy in 1989, none of the governing coalitions have managed to win a second election after being in power until 2011. This instability does not necessarily mean that there is irregularity in the voting preferences of particular social groups. However, there is a paucity of research investigating the voting behaviour of the Polish population in relation to their health. As a result of this lack of empirical evidence, a robust discussion regarding the potential relationship has been largely amiss. This study, therefore, is the first to examine the relationship between political preference and mortality in Poland.

The findings of this study are based on the results of three general elections held in 2001, 2005 and 2007. A proportional voting system is applied in the case of the lower chamber of the Polish Parliament and a majority rule system (plurality-at-large – multi-seat constituencies; first-past the post since 2011) in the case of the upper chamber of Parliament. Only the elections for the lower chamber of Parliament were considered in this study. Each political party was assigned a political "programme factor" varying from 3L (extreme left), through 2L (moderate left), 1L (centre-left), 1R (centre-right), 2R (moderate right) to 3R (extreme right). The economic programmes of the political parties in Poland are not usually coherent and stable. Hence, the assignment of a "programme factor" was based on other postulates such as concepts that refer to morals, the state’s world-view on neutrality, and opinions on the church, religion and individualism. The population indicator for political preference (left or right) was constructed based on the percentage of votes gathered by each party in an election. The results were then unified using a weighted mean, where the results of each party had the weight equal to its "programme factor". Finally, a higher positive number indicated more conservative views dominating in the examined population, whereas a higher negative number demonstrated more liberal views. It is worth noting that some of the parties have been excluded from the analysis, as their political programmes were impossible to be reconstructed, or were ambiguous in the relevant political spheres. These were generally small parties with an extremely narrow electoral support. The lists of the parties that have been included and excluded in the analysis have been listed in the supplementary material. We then analysed the results of the general elections on the level of the poviat, which represent the second tier of local self-government administration in Poland, equivalent to a county in many countries. In 2010, there were 379 poviats with population sizes varying between less than 30,000 to more than 200,000 ¹⁰ .

Data on mortality, education and income levels from 2002–2007 was obtained from the Polish Central Statistical Office. Standardised mortality ratios (SMR) were calculated in each poviat using the overall age-specific death rates for Poland for the period under consideration. We analysed the data using SPSS (statistical software, version 14). A correlation matrix was computed between SMR, voting preference, secondary education and income. Finally, a multiple linear regression analysis was conducted to examine the influence of political preference on SMR, controlling for socio-economic factors and election year.

The results of the three general elections are displayed in Table 1. The results presented show that the electoral turnout was relatively low in all three elections, although the proportion of eligible people voting increased significantly between 2005 and 2007.

The right wing parties dominated in all three elections – both in terms of the electoral support, and the distribution of seats in Parliament. In addition, the results show an increasing trend of electoral support towards right wing parties over the years, with the proportion of votes won by right wing parties increasing from 55% to 85% between 2001 and 2007.

The correlation matrix between SMR and other socio-political factors is presented in Table 2. SMR showed a modest positive correlation with the liberal political views factor, a medium negative correlation with secondary education and a modest negative correlation with monthly income. The correlation between SMR and the conservative views factor was negative; although the result did not reach statistical significance. The conservative factor showed a medium negative correlation with secondary education and monthly income whereas the relationship was smaller and positive with the liberal factor (p<0.05 for all).

The contribution of the political views factor, together with socioeconomic variables, to the statistical explanation of variance in SMR is summarised in Table 3. Secondary education and income explained more variation in SMR than did the political factor. The conservative factor alone had no influence on SMR and the liberal factor could explain only 6.8% of the variation. In a multivariate regression analysis, the liberal factor and secondary education were significantly associated with the SMR (P<0.001 for both) (Table 4).

Here we report the results of the first study of voting behaviour in relation to mortality in Poland. In our study, we found a positive correlation between liberal political views and SMR. Conservative political views were inversely correlated with SMR, although the correlation did not reach statistical significance. An inverse correlation was also found between the conservative views factor and secondary education and income levels, whereas, the relationship with the liberal view factor was positive. In a multivariate regression analysis, liberal political views and secondary education significantly predicted the SMR.

Our results are in alignment with previous studies conducted in the UK and the USA, where conservative political views are negatively correlated with SMR ^{1– 2, 4} . However, the finding of an inverse correlation between conservative political views with education and income levels is opposite to the findings of a US study ⁶ . The most likely explanation for this difference is the very different social structure and political party landscape in a post-communist, catholic country. Another explanation is that conservative views are less defined by economic status or social class in Poland, but more by geographical differences. For example, the north-west of Poland usually manifests more liberal political attitudes, while the south-eastern areas have a more conservative orientation. This may again be partly determined by the economic characteristics of the regions with western regions being more industrialised and urbanised compared to the eastern regions which are mostly rural and agricultural.

One of the limitations of our study is that we were not able to adjust for socioeconomic variables other than secondary education and income which could have influenced the study results. Moreover, the results are based on regional socioeconomic indicators and so are prone to suffer from this ecological fallacy. Thus, the results from this aggregated study could potentially have differing health determinants and characteristics from individuals within the study. However, other recent studies investigating political views at both the individual and regional levels did not show major differences between individual and ecological level political preference data, suggesting that political ideology influences health through several pathways at the individual and contextual level ^{6, 11, 12} .

Important differences in political traditions, political agendas and political party labelling should be acknowledged when comparing studies on political views and health from different countries ^{13, 14} . While it is important to consider the evidence base when analysing and interpreting the results of studies regarding political ideology and health, it is critical that policy decision makers and others utilising such evidence tailor their policies and programmes to meet local population needs. Furthermore, the hypothesis that conservative values, ideologies and latent attitudes could potentially prove conducive to health promoting behaviours and lifestyles ^{6, 15} has to be treated with caution. Unlike in many western countries, the results from this study indicated that higher levels of education and income were commonly correlated with more liberal political attitudes of examined individuals and groups in Poland.

There has been a wide-ranging debate on the possibility for political ideology to influence health; however, the reverse association is also possible. Health could potentially influence political ideology. For example, those with poor health may be inclined to liberal values and ideology as liberal parties typically tend to value and support social equality which in turn favours disadvantaged populations ¹⁵ .

Overall, our findings are consistent with earlier studies conducted in western countries in showing a positive correlation between liberal political views and mortality, but differ in that an inverse relationship was observed between conservative political orientation and education and income. The importance of socioeconomic and geographical factors in relation to political ideology and health inequalities in Poland should be further explored in a longitudinal observational study.