Introduction

In EU28, in 2017, about 63.4% of avoidable mortality was considered preventable, while about 36.6% was treatable. Preventable mortality is a valuable health outcome which differs across European countries, as shown in Graph A1, in the appendix. Preventable mortality is one of the components of (premature or) avoidable mortality; the other component is amenable or treatable mortality. Preventable and treatable mortality both concern to mortality among people under the age of 75. However, treatable mortality refers to causes of death that can be mainly avoided through timely and effective health care interventions, including secondary prevention and treatment, while preventable mortality refers to causes of death that can be mainly avoided through effective public health and primary prevention interventions (i.e., before the onset of diseases or injuries, to reduce incidence). Preventable mortality thus covers deaths from some infectious diseases (e.g., HIV/AIDS), some forms of cancer (e.g., lung cancer), some respiratory diseases (e.g., influenza), injuries, and alcohol and drug-related disorders (Eurostat 2023). To sum up, preventable mortality refers to what happens before the onset of the disease or injury, while treatable mortality occurs after the onset of disease or injury. The factors associated with preventable mortality may therefore differ from those associated with treatable mortality, despite some common determinants.

This work focuses on preventable mortality across countries in the European Union (EU) and its associated explanatory factors. Our aim is twofold: (i) determine the main factors associated with preventable mortality across EU countries; and (ii) test the relevance of two Hofstede cultural dimensions, namely, uncertainty avoidance and long-term orientation, as these two dimensions are strongly related to time and risk preferences. These economic features may play a role in the design of public health policy and primary prevention interventions while also potentially influencing individuals’ healthy behaviour. The contribution of this study arises from the fact that literature exploring the factors related to preventable mortality in Europe is scarce and studies considering the influence of cultural factors on health outcomes are even scarcer.

Preventable mortality is an important health outcome as it may be a good indicator of the effectiveness of social and health policies, which leads to significant benefits for the country. Reducing this rate of mortality could reflect a prioritization of prevention and premature intervention on socioeconomic, cultural, and commercial determinants of health (Dahlgren and Whitehead 1991; Kickbusch et al. 2016). Additionally, it contributes to the improvement of population health and well-being, the reduction in health care costs as well as work absenteeism, the increase in labour productivity, and boosts economic growth (Lyszczarz 2019). The reduction of preventable mortality is also aligned with Sustainable Development Goal 3 (SDG 3) which aims to ensure healthy lives and promote well-being for all at all ages, especially with the targets of reducing non-communicable diseases deaths, substance abuse, and road accidents. Identifying the factors that are most associated with the prevention of disease and promotion of health is thus key to designing successful policies. On the other hand, decisions concerning prevention may be grounded in economic preferences related to time and risk, which may be reflected in or by cultural values. Indeed, a failure to consider the intersection of culture with other structural and societal factors limits the potential benefits of policies and measures to improve health outcomes and reduce related costs (Napier et al. 2014). Moreover, given the increased drive towards the creation of a European Health Union, a better understanding is needed about the factors that are strongly associated with a reduction in preventable mortality.

Overview on previous studies

We begin by presenting previous studies that analyse the factors associated with avoidable mortality in Europe. While the literature on preventable mortality is limited, there is a wider focus on treatable mortality. Some of these studies focus on a single country (Nikoloski et al. 2021; Rásky et al. 2015), while others consider a set of countries. In the latter case, Treurniet et al. (2004) and Newey et al. (2004) found a significant trend of differences across countries in Europe between 1980 and 1997 due to potentially preventable causes of mortality. A negative relationship between treatable mortality and health spending was identified by Arah et al. 2005); Heijink et al. (2013); Alfonso-Sanchez et al. (2017) who concluded that treatable mortality is a good indicator of health expenditures per capita and effectiveness of health systems.

Preventable mortality has received much less attention from researchers than treatable mortality, despite the economic burden that the former poses for a country and the fact that it can be improved through several policies and measures at different levels (Lyszczarz 2019). Nolasco (2009) analysed the evolution of socioeconomic inequalities of preventable mortality in different urban areas in Spain, while Gavuova and Toth (2019) emphasized the importance of environmental factors in explaining regional disparities in Slovakia. In a study of 31 European countries, van den Heuvel and Olaroiu (2017) found that preventable mortality was strongly associated with expenditure on social protection, but factors related to lifestyles were also significant. Socioeconomic inequalities may explain differences in mortality across European countries, arising from differences in national income, quality of government, social transfers, and health care expenditure (Mackenbach et al. 2017a).

The relevance of cultural factors in determining health has long been recognised (Abel 2008; Caldwell and Caldwell 1991; Mackenbach 2014; Roudijk et al. 2017; Siegrist 2008). The consideration of cultural values as a potential explanatory factor for premature mortality has also been accounted for in previous analyses related to health determinants.

We take the wide concept of culture as defined by Macionis and Gerber (2010) that is, a way of thinking and acting which binds people together and shape their worldview and lifeway. Cultural values are thus broad preferences concerning the appropriate course of action that members of a society share and which underlie their norm of behaviour in specific situations (Nolan and Lenski 2004). In this work, we consider the cultural factors proposed by Hofstede dimensions (Hofstede 1980; 1991; 2010): power distance, individualism, masculinity, uncertainty avoidance, long-term orientation, and indulgence. From these six dimensions, we account for the two most closely related to economic preferences toward risk and time: uncertainty avoidance and long-term orientation (Wang et al. 2016; Rieger et al. 2015).

Uncertainty avoidance is defined as the extent to which the members of a culture feel threatened by ambiguous or unknown situations whereas long-term orientation stands for the fostering of virtues oriented towards future rewards (Hofstede 1980; 2010). On the one hand, higher uncertainty avoidance is associated with more risk-aversion in gains and more risk-seeking in losses (Rieger et al. 2015; Ruggeri et al. 2020). On the other hand, long-term orientation is associated with patience for future larger payoffs and avoidance of larger losses (Wang et al. 2016). These economic preferences for time and risk should not be taken as completely separable. These two preferences are intertwined, and it may be difficult to disentangle their effects on people’s decisions (Andreoni and Sprenger 2012; Lopez-Guzman et al. 2018; Tavares 2022b). Due to their close relationship, they may be expressed jointly in uncertainty avoidance and long-term orientation. Moreover, it may be argued that individual preferences may not coincide with societal preferences or policymaker preferences (Lawless et al. 2013; Moffett and Suarez-Almazor 2005). In general, within the framework of healthcare, it was established that framing gains are better suited to promote preventive behaviours which are perceived as low-risk, while framing losses are better for detection behaviours which are perceived as high-risk (Gisbert-Pérez et al. 2022). In addition, high discount rates and a preference for the present can contribute to governmental emphasis on acute health care, rather than preventive health care (Lawless et al. 2013).

There are very few studies that take into account cultural factors as determinants of health outcomes. The value of well-being or self-expression, proposed by Inglehart (1990), was found to be strongly associated with mortality as found by Mackenbach (2014) and Mackenbach et al. (2017), when focusing on European countries and the factors explaining disparities in mortality. Based on Hofstede’s (2010) dimensions, Touboul-Lundgren et al. (2015) aimed to investigate the use of antibiotics in primary care in Europe and to pinpoint cultural factors that influence it. Additionally, the life expectancy at 65 years of age in European countries was shown to be correlated with the cultural clusters based on Hofstede dimensions (Tavares 2022a).

Finally, we refer to empirical analyses which use a panel data approach to identify the factors associated with health outcomes expressed by an aggregate health production function (Zweifel et al. 2009). Mortality rates, life expectancy and potential years of life lost were explained by different factors such as GDP per capita, health expenditure, education, unemployment, Gini index, protein intake, fat and sugar consumption, tobacco, physician density and democracy (Arah et al. 2005; Joumard et al. 2008; Mackenbach et al. 2019; Or 2000; Spijker 2005; Tavares 2022a). Additionally, Heijink et al. (2013) investigated the relationship between avoidable mortality and healthcare expenditure in 14 western countries between 1996 and 2006, using a panel data approach; Nikoloski et al. (2021) used a panel data for Mexico states between 2000 and 2015 to find the relationship between health system inputs and amenable mortality; and lastly, Mackenbach et al. (2017) study amenable mortality in 17 European countries between 1980 and 2010.

Study design

Data and variables

Data used in this study is collected from Eurostat, for the period 2011–2019, before the COVID-19 pandemic, for 27 countries (Table SM1 in Supplementary Material lists the considered countries). Additionally, data concerning Hofstede cultural values were collected from ‘Geerthofstede website’ and obtained from the Dimension data matrix (Hofstede 2023).

Dependent variable

The dependent variable is the standardized preventable mortality death rate (per 100,000 inhabitants) and its abbreviation is “Preventable”. Graph A1 in Appendix shows the differences in the preventable mortality rate across 27 European countries in 2019.

Independent variables

We considered the following time-varying independent variables (and abbreviations):

  1. i)

    natural logarithm of GDP per capita in purchasing power parities (GDPpc),

  2. ii)

    percentage of the population aged 15–64 with a tertiary education (levels 5–8 International Standard Classification of Education (ISCED 2011)) (Education),

  3. iii)

    unemployment rate as a percentage of the labour force (Unemploym),

  4. iv)

    percentage of persons at risk of poverty or social exclusion (Risk_pov),

  5. v)

    number of immigrants per capita (Immigrants),

  6. vi)

    health expenditure in preventive function as a percentage of current health expenditure (Heath_Exp),

  7. vii)

    social protection benefits as a percentage of GDP (SocSec_Exp),

  8. viii)

    air emissions measured by annual emission of particulates < 2.5 μm (Particul2.5).

Additionally, we considered two time-invariant cultural variables: uncertainty avoidance and long-term orientation. In Supplementary Material Table SM2 the scores for these indexes are presented for each country. Both indexes vary from 0 to 100. Lower values express weaker uncertainty avoidance and short-term orientation while higher values relate to stronger uncertainty avoidance and long-term orientation.

Analytical strategy

Our econometric approach, performed in STATA.15, is compatible with the consideration of cultural values fixed across time. We used two basic approaches to analyse the relationship between preventable mortality and explanatory factors: (1) Least square dummy variable (LSDV) regressions with fixed time effects with robust standard errors, and (2) panel data regressions using different methods of estimation: AR(1), Driscoll-Kraay, Prais-Winstein, and Hausman-Taylor.

A brief explanation of these methods is next provided. The LSDV is a common OLS regression which considers dummy variables for each year and standard errors are corrected to prevent heteroscedasticity. Panel data models consider the analysis across time and countries simultaneously. These models can be estimated in several different ways. The AR(1) method is used to fit cross-sectional time-series regression models when the disturbance term is first-order autoregressive (command xtregar). This means that the error term follows a first-order autoregressive process and so by accounting for this process, the AR(1) method provides consistent and asymptotically unbiased estimates. The Driscoll-Kraay estimation (command xtscc) produces standard errors for the coefficient estimated when the error structure is heteroskedastic and autocorrelated between panels. This procedure is identical to the estimation of a pooled OLS with country fixed effects, but the standard errors are robust to general forms of cross-sectional and temporal dependence. The Prais-Winstein regression (command xtpcse) is used for panel-corrected standard errors estimation, and it assumes that the disturbances are heteroskedastic and contemporaneously correlated across panels. In this case, estimated coefficients are adjusted by standard errors which account for these panel-specific structures and become more robust estimates. Finally, Hausman-Taylor estimation (command xthtaylor) fits random-effects model in which some of the covariates are correlated with unobserved individual-level random effects. This estimator is designed to consider time-varying and time-invariant variables and to address potential endogeneity by using instrumental variables. These instruments are some of the time-varying exogenous variables. For the Hausman-Taylor estimation, we considered time-varying independent endogenous variables, health expenditures (Health_Exp) and social security expenditures (SocSec_Exp).

After descriptive statistics on the variables, we performed some econometric tests and analyses, such as pairwise correlations, multicollinearity check using VIF information, Breusch-Pagan / Cook-Weisberg test for heteroskedasticity, Wooldridge test for autocorrelation, and finally, linktest for specification testing.

We considered time-varying independent variables lagged one year so that they are not contemporaneous with preventable mortality rate. Thus, the estimated model is as follows:

$$\begin{gathered}Preventabl{e_{t,i}} = \,c\, + {\text{ }}{b_1}GDPp{c_{t - 1,i}}^{\left( - \right)} + \,{b_2}Educatio{n_{t - 1,i}}^{\left( - \right)} \hfill \\+ \,{b_3}Unemploy{m_{t - 1,i}}^{\left( + \right)} + {b_4}Risk\_Po{v_{t - 1,i}}^{\left( + \right)}\, \hfill \\+ {b_5}Immigrant{s_{t - 1,i}}^{\left( - \right)} + {b_6}Health\_{\text{ }}Ex{p_{t - 1,i}}^{\left( - \right)} \hfill \\+ {b_7}SocSec\_Ex{p_{t - 1,i}}^{\left( - \right)} + {b_8}Particul{2.5_{t - 1,i}}^{\left( + \right)} \hfill \\+ {b_9}Uncert\_Avoi{d_i}^{\left( ? \right)} + {b_{10}}Long\_Ter{m_i}^{\left( ? \right)} + {\text{ }}{\varepsilon _{t,i}}. \hfill \\ \end{gathered}$$

The sign (+) hypothesizes a positive correlation, (−) a negative correlation and (?) an unknown expected sign correlation with preventable mortality rate.

Results

The descriptive statistics for all variables are presented in Table A1 in appendix.

Concerning the cultural characteristics, it may be stated that the countries with the lowest value for the uncertainty avoidance index are Denmark and Sweden, while the highest scores are found in Portugal and Greece; the lowest value for the time orientation score is registered in Ireland and Portugal and the highest score for long-term orientation is found in Estonia and Germany.

Graphs SM1 (A and B) and SM2 (A and B), in Supplementary Material, show the distribution of countries across cultural values, uncertainty avoidance and long-term orientation, and preventable mortality (graph A) and the linear relationship with a 95% confidence interval (graph B). While there is no significant statistical correlation for uncertainty avoidance distribution, there is a significant positive correlation (equal to 0.35) for the long-term orientation case.

Table A2 in the Appendix shows the pairwise correlations between time-varying independent variables. The strongest significant correlations (above 50%) are found between the following pairs of variables and corresponding signs:

$$\begin{gathered}\left( {GDPpc,SocSec\_Exp\left( - \right)} \right),{\text{ }}\left( {GDPpc,Education\left( + \right)} \right), \hfill \\\,\left( {GDPpc,Risk\_Pov\left( - \right)} \right),{\text{ }}\left( {GDPpc,Immigrants\left( + \right)} \right),{\text{ }} \hfill \\\,\left( {Risk\_Pov,Unemploym\left( + \right)} \right),{\text{ }}and{\text{ }}finally{\text{ }}\left( {GDPpc,Preventable\left( - \right)} \right). \hfill \\ \end{gathered}$$

The pairwise correlation between uncertainty avoidance and long-term orientation is equal to 0.034 but has no statistical significance. In Graph SM3, in Supplementary Material, we show the countries’ positioning relative to these two dimensions and the pairwise correlation without Ireland (IE) and Denmark (DK) is statistically significant and equal to (– 0.309).

The multicollinearity was checked by using VIF values which are presented in Table A3, in the Appendix, and it shows that there is no such problem as the mean VIF value equals 2.09. The Breusch-Pagan / Cook-Weisberg test for heteroskedasticity results in X2 = 23.17 with Prob > X2 smaller than 0.001, which calls into question the existence of heteroskedasticity. The specification test ‘linktest’ for single-equation models shows hatsq estimated coefficient without statistical significance (p-value equals 0.225) which means that the model is well-specified. Finally, the Wooldridge test for autocorrelation in panel data raises the possibility of the existence of autocorrelation (F statistic = 9.123 and p-value equals 0.0056). In sum, we conclude that there is no multicollinearity but there is potential for heteroskedasticity and autocorrelation between panels. For these reasons, we considered robust standard errors and the mitigation of potential temporal dependence in the panel data estimated models.

The results obtained for the estimation of the least squares dummy variable regressions are shown in Table 1. Except for GDP per capita and Uncertainty avoidance, all the remaining variables are statistically significant when considering time-fixed effects.

Table 1 Least squares dummy variable regressions
Full size table

Table 2 shows the estimation results for AR(1), Dirscoll & Kraay, Prais-Winstein and Hausman-Taylor procedures for panel data.

Table 2 Panel data regressions
Full size table

The results obtained from the different econometric procedures are generally consistent. We found a negative association between preventable mortality rate and GDP per capita, social security expenditures and uncertainty avoidance; and a positive association between air pollution measured air particles < 2.5 μm and long-term orientation. We also found that immigrants and health expenditure contribute to the reduction of preventable mortality, but this can only be observed in LSVD and Driscoll-Kray estimation procedure. Finally, we found that the education, unemployment, and risk of poverty variables resulted in different estimated signs in Driscoll-Kray and Hausman-Taylor procedures, failing to confirm the initial hypotheses.

Discussion

Preventable mortality is an indicator of the population’s health and the effectiveness of the public economic, social and health policy affecting health determinants and people’s behaviour. Understanding the factors associated with preventable mortality contributes to better improved and designed policies to reduce those causes of death, improve overall population health and well-being, and add to sustainable economic growth. In fact, reducing preventable mortality is considered within Sustainable Development Goal 3 (SDG 3): Ensuring healthy lives and promoting well-being for all at all ages. Specifically, targets 3.4, 3.5, and 3.6 are focused on the reduction of mortality from non-communicable diseases, the prevention of substance abuse, and the reduction of road accidents.

This work aimed to identify the main drivers for preventable mortality in 27 European countries between 2011 and 2019. The main contribution of this analysis comes from the consideration of cultural factors related to economic preferences. Hofstede cultural dimensions of uncertainty avoidance and long-term orientation have been considered in the set of explanatory factors of preventable mortality together with other socioeconomic factors.

The most relevant finding is the importance of the cultural dimensions of uncertainty avoidance and long-term orientation in explaining the preventable mortality rate. Countries that are more prone to uncertainty avoidance have lower preventable mortality rates, while countries that are more focused on the long-term tend to register higher preventable mortality rates.

Other relevant findings include the absence of the significant role of health expenditures but a significant beneficial effect of social security expenditures in reducing preventable deaths. Additionally, GDP per capita and education also contribute to the decrease of this type of mortality, while the emission of particulates < 2.5 μm contributes to additional premature mortality.

First, our key findings point to a significant relationship between cultural dimensions, uncertainty avoidance, and long-term orientation, and preventable mortality rates. These deaths occur before the onset of the disease or injury, usually because of the absence of some preventive behaviour or intervention. So, in countries with high uncertainty avoidance and intolerance to ambiguity in society, there are more formal and informal rules that may contribute to higher predictability of events, which in turn mitigate the likelihood of premature death.

On the other hand, countries with a stronger long-term orientation, where the preference for future rewards is well-accepted, tend to show higher preventable mortality rates. Maybe in countries characterized by a strong emphasis on perseverance; where the importance of leisure is low, while the importance of work status and market positions is high, the opportunity cost of prevention is considered too high, resulting in lower levels of compliance with public health prevention. This finding may differ from the one anticipated by Lawless et al. (2013) who predicted that countries with a short-term preference would have a lower preference for preventive care, and so a higher related mortality. However, it could be that Lawless et al. were unable to disentangle the influence of risk aversion of governments during the political cycle. Our result may be the effect of improving preventable deaths resulting from a sequential postponement of investment in preventive care in different prevention dimensions. In addition, accounting for the potential negative correlation between the two cultural dimensions across countries (Graph SM3 in Supplementary Material), the opposite estimated effect arising from the cultural dimensions on preventable mortality rate is expected.

On the other hand, considering that uncertainty avoidance and long-term orientation are correlated with economic preferences (Rieger et al. 2015; Wang et al. 2016), it may be inferred that countries with stronger risk aversion and higher time preference (short-term orientation) tend to have lower preventable mortality. This means that the need to control uncertainty with prevention of accident, injury, or disease, and the preference to obtain short-term results benefiting life and health. This may result in lower preventable mortality rates, despite the intertwined linkages between risk and time preferences (Andreoni and Sprenger 2012; Lopez-Guzman et al. 2018; Tavares 2022b).

Second, other results concerning the role played by GDP per capita, health prevention and social expenditures, and air pollution are aligned with previous results. While GDP per capita (Arah et al. 2005; Cutler et al. 2006; Guzel et al. 2021; He and Li 2018; Joumard et al. 2008; Mackenbach et al. 2017, 2019; Minagawa and Jagger 2020; Or 2000; Tavares 2017, 2022b; Spijker 2005; Zare et al. 2015), health expenditures (Arah et al. 2005; Bradley et al. 2011; de Meijer et al. 2013; Jaba et al. 2014; Joumard et al. 2008; Mackenbach et al. 2019; Or 2000; Rahman et al. 2018; Spijker 2005; Tavares 2017, 2022a) and social expenditures (Alexiou et al. 2021; Bergqvist et al. 2013; Bradley et al. 2011; Eikemo et al. 2008; Reeves et al. 2016; Reynolds and Avendano 2018; van den Heuvel 2017; Zare et al. 2015) contribute to the improvement of the population health, air pollution (Arah et al. 2005; Dominski et al. 2021; Joumard et al. 2008; Manisalidis et al. 2020; Or 2000; Tavares 2022a; WHO 2004, 2020b) has a negative effect. The influence of GDP per capita on preventable mortality can be observed in different channels of transmission, such as (i) wider and better information and cognitive interpretation of the information (for instance, better education or communication campaigns); (ii) better environmental and living conditions related to water, sewage, garbage, and also communications infrastructures (easier and better access to internet); (iii) more employment opportunities and safe working conditions, and (iv) poverty reduction by mitigating severe adverse living conditions. Although our data on health prevention and social expenditures are not strongly correlated with GDP per capita, the channels of transmission to minimize preventable mortality are identical. Air pollution has a well-established negative effect on people’s health by increasing the risk of cardiopulmonary diseases and reducing life expectancy.

Finally, we found less consistent results related to education, unemployment, risk of poverty, and immigration, even though these variables are strongly correlated with GDP per capita but not among themselves, in general. We begin by discussing education as an associated factor with preventable mortality. We started on the assumption of a negative correlation, which was in fact found in AR(1) and Hausman-Taylor models. In these models, we confirm previous results that relate education to improved health outcomes (Cutler et al. 2006; Bijwaard et al. 2015; Marmot et al. 2012; Mackenbach et al. 2019; Murtin et al. 2017; Zare et al. 2015). Higher levels of education contribute to better knowledge on healthy behavior, disease prevention, and management of chronic diseases.

Next variable is unemployment and its influence on people’s health has been found to be negative (Gerdtham and Joannesson 2003; Leclerc et al. 2006) because it increases the risk of suicide as well as the risk of other diseases (but not cancer or cardiovascular illnesses). However, we only found this relationship using the Hausman-Taylor model. The third variable with lower consensual results is risk of poverty. Poverty is usually associated with poor health outcomes (Mackenbach 2017; Marmot 2005; Wang et al. 1997; WHO 2010) as it prevents people accessing health care, fosters low-quality nutrition, poor housing conditions, and chronic stress. Our results confirm this type of relationship in the case of LSDV and Driscoll-Kray models estimation. Lastly, immigration may also affect population health and impact positively on life expectancy (Giuntella and Mazzonna 2015; Hendi et al. 2021; Singh and Hiatt 2006) and our models also found this relationship, both LSDV and Driscoll-Kray models because immigrants tend to be young, healthy, and contribute to the workforce.

A final word on our estimated models, both Driscoll-Kray and Hausman-Taylor, resulted in unexpected signs for education, unemployment, and risk of poverty, which we cannot fully comprehend or explain. However, these results should be known and compared with the other models for consistency, and they still provide good and consistent results for the remaining explanatory variables. In Supplementary Material, Table SM3 presents the model estimation without the explanatory variable GDP per capita, and the results are identical in general.

The main strength of this work is the focus of the analysis on preventable mortality and its relationship with cultural factors. The results show that even in a future European Health Union, cultural differences must be considered when designing effective European interventions and policies. Moreover, the findings obtained in this study contribute to informing policymakers on the relevance of socioeconomic and cultural factors, when designing policies aimed at improving the prevention of illness and injuries.

Public health campaigns aimed at disseminating health information and health care literacy contribute to creating lower levels of ambiguity concerning health and illness prevention (Erman and Medeiros 2021). The creation of community health centres, where diverse preferences are accommodated without emphasising people’s status and hierarchies, may also be considered to overcome the excess burden created by the perseverance to achieve long-term rewards. Social support measures addressing social determinants of health, aiming to mitigate long-term health disparities and uncertainty created across social groups may contribute to the reduction of treatable mortality rates. Other public health policy challenges may be needed to overcome to improve treatable mortality rates accounting for uncertainty avoidance and long-term orientation values (IM 1988). On the one hand, improving the public image of public health improves credibility and reliance on public health initiatives. On the other hand, improving effective leadership, and strengthening interaction between technical and political aspects of decisions, contribute to lower levels of public uncertainty and to smooth market and status positions. These interactions with cultural values may foster behaviours that decrease treatable mortality. These public health policies aimed to decrease treatable deaths are well-aligned with SDG 3, targets 3.4–3.6, and our findings contribute to better-informed policymakers and to the global effort to achieve these goals.

The main limitation of this analysis is that it considers short panel data, specifically, the number of periods is small. In the future, a larger period may be considered, and large panel data econometric procedures may be performed to analyse short- and long-term effects, as well as a causality analysis. Another possible criticism of this study is the absence of a factor expressing and proxying lifestyles. However, there was no complete series of data for variables such as obesity, smoking, or alcohol intake. Future work may also face the problem of overfitting due to the complexity of the models with a high number of independent variables, years, and countries. For this case, other econometric procedures are possible to use, such as the lasso and ridge regressions, which will shrink the coefficients of the less relevant explanatory variables and improve the interpretability of the estimated model.

Conclusion

This work has identified some potential drivers of preventable mortality in European countries, which is a complex and multidimensional issue. The high percentage of preventable mortality of total avoidable mortality calls for more focused attention. We have found that socioeconomic factors, such as GDP, education, environmental factors (including air pollution), and social policies influence the number of preventable deaths per year in European countries. We also found that cultural factors expressed by uncertainty avoidance and long-term orientation play a role in determining preventable deaths. The nature of preventable mortality, which can be avoided through effective public health and primary prevention interventions, calls for public policies to improve socioeconomic and commercial determinants of health, reduce exposure to disease risk factors, improve occupational safety regulations, reduce alcohol, and drug abuse. All these preventive nature interventions require compatibility with cultural values to be effective in reducing the premature mortality rate and successfully contribute to the achievement of SDG 3.