Population Health Inequalities Across and Within European Metropolitan Areas through the Lens of the EURO-HEALTHY Population Health Index

The different geographical contexts seen in European metropolitan areas are reflected in the uneven distribution of health risk factors for the population. Accumulating evidence on multiple health determinants point to the importance of individual, social, economic, physical and built environment features, which can be shaped by the local authorities. The complexity of measuring health, which at the same time underscores the level of intra-urban inequalities, calls for integrated and multidimensional approaches. The aim of this study is to analyse inequalities in health determinants and health outcomes across and within nine metropolitan areas: Athens, Barcelona, Berlin-Brandenburg, Brussels, Lisbon, London, Prague, Stockholm and Turin. We use the EURO-HEALTHY Population Health Index (PHI), a tool that measures health in two components: Health Determinants and Health Outcomes. The application of this tool revealed important inequalities between metropolitan areas: Better scores were found in Northern cities when compared with their Southern and Eastern counterparts in both components. The analysis of geographical patterns within metropolitan areas showed that there are intra-urban inequalities, and, in most cities, they appear to form spatial clusters. Identifying which urban areas are measurably worse off, in either Health Determinants or Health Outcomes, or both, provides a basis for redirecting local action and for ongoing comparisons with other metropolitan areas.


Introduction
Health is a critical global development issue, especially in urban areas where the majority of the world's population lives [1][2][3][4]. The United Nations Sustainable Development Goals (SDG) put a focus on health promotion through a number of interconnected health-related targets like SDG-3 (good health and well-being), SDG-10 (reduced inequalities) and SDG-11 (sustainable cities and communities). These goals are achievable through multisectoral approaches, as stated in the New Urban Agenda [5][6][7]. There is ample evidence that contextual factors related with the social, physical and built urban environments affect health and are key drivers of health inequalities within cities: The access to green spaces and public places, the exposure to air pollution and noise, the access to affordable housing, the opportunity to use public transportation, to walk and bike, among others [6,[8][9][10][11][12][13][14][15].
The main factors influencing the overall health of the population are well illustrated by Dahlgren and Whitehead's model of social determinants of health [16,17], which describes the different layers of influence: Individual lifestyle factors, social and community networks, living working and conditions and general socioeconomic, cultural and environmental conditions in which the population lives. The determinants of health are shaped by individual and political decisions and can be either positive health factors (e.g., economic security, adequate housing), protective (e.g., social support, healthy diet) or risk factors (e.g., pollution, smoking) [17]. The recent study on environmental public health indicators in European urban areas, within the framework of the EURO-HEALTHY project, explored this through the association between the health impacts and the physical and built environmental risks in order to support the prioritisation of interventions that improve public health and reduce avoidable deaths [18].
There is a growing body of evidence showing a strong connection between the characteristics of the place of residence and health outcomes, even after accounting for individual risk factors [8,13,[18][19][20][21][22][23][24][25]. High levels of intra-urban inequalities are also visible as a result of the demographic, economic, environmental, and other societal challenges impacting cities, along with a greater population heterogeneity and different level of access to housing, amenities and services [2,10,13,26]. According to the literature, poor and vulnerable groups are often more at risk due to the concentration on disadvantaged and deprived neighbourhoods, usually in the outskirts of the city or in inner city areas [27][28][29][30]. The Atlas of Population Health in European Union Regions [31], shows that the capital region of each EU country ussually performs better in health determinants (e.g., economic and social conditions, healthcare resources) than in health outcomes. Capital regions are often affected by increased levels of crime and air pollution with negative impacts on health outcomes. Growing evidence demonstrates an excess on mortality and greatest burden of disease on urban areas compared to non-urban, namely related with lung cancer and cardiovascular disease [32,33], and of greater relative socioeconomic inequalities in mortality in Eastern and Northern European cities, although with variations in their magnitude [34,35].
A deeper understanding of the interconnection between compositional and contextual factors and how they affect health outcomes is required of local decision makers in order to better cope with the complexity of addressing health determinants that goes beyond traditional behavioural change approaches [2,13,20,22,36]. Most of the policy interventions largely concentrate on modifying the midstream determinants of health, the intermediate factors, such as individual health behaviours (smoking, physical activity, nutrition) and on targeting vulnerable groups to mitigate the negative impacts of disadvantage on health. Policies and actions need to be directed towards improving fundamental social and economic structures in order to remove barriers and allow people to achieve their full health potential [37][38][39][40]. Local governments, especially those from urban areas, occupy a unique leadership position on levelling up policies to tackle determinants of health, working across the upstream, midstream and downstream levels [1,25,40,41].
To effectively address the causes of health inequalities, spatially disaggregated data on different health determinants and better urban health metrics are needed [42,43]. A measure that summarises crucial data provides opportunities to understand the complexity of how much health differs within and across urban areas, given how it offers a comprehensive picture of health and health disparities. Nevertheless, the ability to include meaningful information into a single metric that captures the level of health, the intensity of health determinants, and the extent of disparities, is limited [44].
It was within the scope of the EU-funded project EURO-HEALTHY (shaping European policies to promote health equity) whose aim was to advance the knowledge on policies with the highest potential to promote health equity, that a multidimensional and multilevel index-the Population Health Index (PHI)-was built. This measure evaluates European population health across a wide range of areas of concern, dimensions and indicators of health determinants and health outcomes [31]. Its construction integrates the technical elements of a multi-criteria value model and the social elements of interdisciplinary and participatory processes by collecting the views of experts and stakeholders on which factors are relevant to evaluate health [45] and on how important it is to close the gap between indicators to improve population health [46].
More than 75% of the European population lives in urban areas, thus reflecting and encompassing a diversity of geographies and of physical, social, and economic environments. By using the EURO-HEALTHY Population Health Index (PHI), the aim of this paper was to identify inequalities in health determinants and health outcomes across and within nine European metropolitan areas.

Study Area
We applied an ecological study to analyse the Health Determinants and Health Outcomes value-scores between and within metropolitan areas, taking 2014 as a reference year.
The indicators were collected at two levels defined by EUROSTAT [47]: Local administrative units (LAU) 1, corresponding to small areas, and LAU 2, corresponding to municipalities. The utilisation of both levels is explained by the diversified system of local governments in Europe and the pre-existent request from the EURO-HEALTHY project to employ the PHI to a political subdivision where a local government may implement interventions able to address health inequalities. Each metropolitan area specified different administrative levels, most of them corresponding to municipalities. Additional information on the LAU and delimitation of each metropolitan area is provided in Supplementary Materials S1.
The nine metropolitan areas analysed in this study where selected under the EURO-HEALTHY project framework. These areas represent the different EU geographic zones and populations (Northern, Southern, Central and Eastern Europe) reflecting Europe's diversity-in terms of contextual conditions (e.g., geographical, historical, political, cultural, social and economic) and impact of the financial and economic crisis (Table 1).

Applying the Population Health Index
The measure used in this study is the EURO-HEALTHY PHI, which was built to evaluate population health of the EU regions for the reference year 2014. Presenting a bottom-up hierarchical structure with several indices, the PHI measures health with respect to the components Health Determinants and Health Outcomes, both disaggregated into areas of concern, dimensions and indicators (Table 2) [31]. The Health Determinants component represents the contextual factors defined as the environmental conditions in which people live and which directly and/or indirectly influence health: Economic conditions, social protection and security, education, demographic change, lifestyle and health behaviours, physical environment, built environment, road safety, healthcare resources and expenditure, and healthcare performance. The Health Outcomes component refers to the severity and frequency of disease and/or death, including both mortality and morbidity indicators.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of m respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas m Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Fou emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pr Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outco

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4 presents the results of the pairwise co respect to Health Determinants and Health Outcomes Indi emerge: (1) Stockholm stands out with a significantly higher 87 on both components); (2) Athens, Barcelona

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Note: The symbols and identify the metropolitan areas where scores were found to be statistically different. By way of example: Brussels presents mean scores that are statistically different from Stockholm, Turin, Lisbon, Prague and Berlin-Brandenburg (with higher scores: ) and from Barcelona (with lower scores: ). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). Note: The symbols and identify the metropolitan areas where scores were found to be statistically different. By way of example: Brussels presents mean scores that are statistically different from Athens, Barcelona, and Lisbon, London and Turin (with higher scores: ) and from Stockholm (with lower scores: ).
The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Note: The symbols and identify the metropolitan areas where scores were found to be statistically different. By way of example: Brussels presents mean scores that are statistically different from Stockholm, Turin, Lisbon, Prague and Berlin-Brandenburg (with higher scores: ) and from Barcelona (with lower scores: ). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). Note: The symbols and identify the metropolitan areas where scores were found to be statistically different. By way of example: Brussels presents mean scores that are statistically different from Athens, Barcelona, and Lisbon, London and Turin (with higher scores: ) and from Stockholm (with lower scores: ).
The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Note: The symbols and identify the metropolitan areas where scores were found to be statistically different. By way of example: Brussels presents mean scores that are statistically different from Stockholm, Turin, Lisbon, Prague and Berlin-Brandenburg (with higher scores: ) and from Barcelona (with lower scores: ). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geo variation in the distribution of the value-scores was revealed across metropolitan area components. Overall, almost all the municipalities registered value-scores above 50 (the P from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan a respect to Health Determinants and Health Outcomes Indices. Four groups of metropol emerge: (1) Stockholm stands out with a significantly higher mean score in both componen 87 on both components); (2) Athens, Barcelona and Lisbon present lower values Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcome 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). Note: The symbols and identify the metropolitan areas where scores were found to be statis different. By way of example: Brussels presents mean scores that are statistically different from A Barcelona, and Lisbon, London and Turin (with higher scores: ) and from Stockholm (with lower score The symbols and only display lower or higher differences (respectively), although not statis significant. NA = No group was found. Note: The symbols and identify the metropolitan areas where scores were found to be statis different. By way of example: Brussels presents mean scores that are statistically different from Stockholm, Lisbon, Prague and Berlin-Brandenburg (with higher scores: ) and from Barcelona (with lower scores: symbols and only display lower or higher differences (respectively), although not statistically signi NA = No group was found. Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) a variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower analyse each metropolitan area, higher internal variability was identified for Brussels and Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of me respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas me Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Fou emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outcom   Figure 2 presents a 12% variation in the Health Determinants variation in the Health Outcomes Index (CV = 0.10). Although th analyse each metropolitan area, higher internal variability was ident Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to t variation in the distribution of the value-scores was reve components. Overall, almost all the municipalities registere from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4 presents the results of the pairwise c  respect to Health Determinants and Health Outcomes Ind  emerge: (1) Stockholm stands out with a significantly highe 87 on both components); (2) Athens, Barcelona and Li Determinants (below 62.7); (3) Barcelona and Turin present 83); and (4) Lisbon and Prague present lower scores in Heal

Results
When the EURO-HEALTHY PHI was appl variation in the distribution of the value-scores w components. Overall, almost all the municipalities from 0 to 100), with 31% attaining 75 and above in b Tables 3 and 4 presents the results of the pa respect to Health Determinants and Health Outco emerge: (1) Stockholm stands out with a significant 87 on both components); (2) Athens, Barcelona Determinants (below 62.7); (3) Barcelona and Turin 83); and (4) Lisbon and Prague present lower scores

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geo variation in the distribution of the value-scores was revealed across metropolitan area components. Overall, almost all the municipalities registered value-scores above 50 (the P from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan a respect to Health Determinants and Health Outcomes Indices. Four groups of metropol emerge: (1) Stockholm stands out with a significantly higher mean score in both componen 87 on both components); (2) Athens, Barcelona and Lisbon present lower values Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcome 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) a variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower analyse each metropolitan area, higher internal variability was identified for Brussels and Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of me respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas me Health Determinants Index.   Figure 2 presents a 12% variation in the Health Determinants Index (C variation in the Health Outcomes Index (CV = 0.10). Although the variati analyse each metropolitan area, higher internal variability was identified for B Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same va Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Fou emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outcom   Figure 2 presents a 12% variation in the Health Determinants variation in the Health Outcomes Index (CV = 0.10). Although th analyse each metropolitan area, higher internal variability was ident Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to t variation in the distribution of the value-scores was reve components. Overall, almost all the municipalities registere from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4 presents the results of the pairwise c  respect to Health Determinants and Health Outcomes Ind  emerge: (1) Stockholm stands out with a significantly highe 87 on both components); (2) Athens, Barcelona and Li Determinants (below 62.7); (3) Barcelona and Turin present 83); and (4) Lisbon and Prague present lower scores in Heal

Results
When the EURO-HEALTHY PHI was appl variation in the distribution of the value-scores w components. Overall, almost all the municipalities from 0 to 100), with 31% attaining 75 and above in b Tables 3 and 4 presents the results of the pa respect to Health Determinants and Health Outco emerge: (1) Stockholm stands out with a significant 87 on both components); (2) Athens, Barcelona Determinants (below 62.7); (3) Barcelona and Turin 83); and (4) Lisbon and Prague present lower scores

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geo variation in the distribution of the value-scores was revealed across metropolitan area components. Overall, almost all the municipalities registered value-scores above 50 (the P from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan a respect to Health Determinants and Health Outcomes Indices. Four groups of metropol emerge: (1) Stockholm stands out with a significantly higher mean score in both componen 87 on both components); (2) Athens, Barcelona and Lisbon present lower values Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcome 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of me respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas me Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Fou emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outcom

Results
When the EURO-HEALTHY PHI was applied to t variation in the distribution of the value-scores was reve components. Overall, almost all the municipalities registere from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4

Results
When the EURO-HEALTHY PHI was appl variation in the distribution of the value-scores w components. Overall, almost all the municipalities from 0 to 100), with 31% attaining 75 and above in b Tables 3 and 4 presents the results of the pa respect to Health Determinants and Health Outco emerge: (1) Stockholm stands out with a significant 87 on both components); (2) Athens, Barcelona Determinants (below 62.7); (3) Barcelona and Turin 83); and (4) Lisbon and Prague present lower scores

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7).   Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a variation in the distribution of the value-scores was revealed across metropolitan a components. Overall, almost all the municipalities registered value-scores above 50 (th from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of m respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas m Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Fou emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pr Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outco

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of m respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas m Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.  Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geo variation in the distribution of the value-scores was revealed across metropolitan area components. Overall, almost all the municipalities registered value-scores above 50 (the P from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of me respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metr components. Overall, almost all the municipalities registered value-scores ab from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise comparisons of m respect to Health Determinants and Health Outcomes Indices. Four groups emerge: (1) Stockholm stands out with a significantly higher mean score in bo 87 on both components); (2) Athens, Barcelona and Lisbon present low Determinants (below 62.7); (3) Barcelona and Turin present high scores in Hea 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (belo Table 3. Pairwise comparisons of the differences between metropolitan areas m Health Determinants Index.

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-s from 0 to 100), with 31% attaining 75 and above in both components.
Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Four emerge: (1) Stockholm stands out with a significantly higher mean sc 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high scor 83); and (4) Lisbon and Prague present lower scores in Health Outcom

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ra from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components.    Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061). Health 2019, 16, x

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan variation in the distribution of the value-scores was revealed across metr components. Overall, almost all the municipalities registered value-scores ab from 0 to 100), with 31% attaining 75 and above in both components.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).
identify the metropolitan areas where scores were found to be statistically different.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).
) and from Stockholm (with lower scores:

Results
When the EURO-HEALTHY PHI was applied to the met variation in the distribution of the value-scores was revealed ac components. Overall, almost all the municipalities registered value from 0 to 100), with 31% attaining 75 and above in both component Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metrop variation in the distribution of the value-scores was revealed acros components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4     . Pairwise comparisons of the differences between metropolitan areas mean scores from the Determinants Index. . Pairwise comparisons of the differences between metropolitan areas mean scores from the Outcomes Index. 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% the Health Outcomes Index (CV = 0.10). Although the variation is lower when we only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI r from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metrop variation in the distribution of the value-scores was revealed acros components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4 presents the results of the pairwise co respect to Health Determinants and Health Outcomes Indic emerge: (1) Stockholm stands out with a significantly higher 87 on both components); (2) Athens, Barcelona

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI r from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metrop variation in the distribution of the value-scores was revealed acros components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI r from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metrop variation in the distribution of the value-scores was revealed acros components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a variation in the distribution of the value-scores was revealed across metropolitan components. Overall, almost all the municipalities registered value-scores above 50 (t from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropoli variation in the distribution of the value-scores was revealed across m components. Overall, almost all the municipalities registered value-score from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4

Results
When the EURO-HEALTHY PHI was appli variation in the distribution of the value-scores w components. Overall, almost all the municipalities r from 0 to 100), with 31% attaining 75 and above in b Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a variation in the distribution of the value-scores was revealed across metropolitan components. Overall, almost all the municipalities registered value-scores above 50 (t from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropoli variation in the distribution of the value-scores was revealed across m components. Overall, almost all the municipalities registered value-score from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4

Results
When the EURO-HEALTHY PHI was appli variation in the distribution of the value-scores w components. Overall, almost all the municipalities r from 0 to 100), with 31% attaining 75 and above in b Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan a variation in the distribution of the value-scores was revealed across metro components. Overall, almost all the municipalities registered value-scores abo from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both comp Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Four emerge: (1) Stockholm stands out with a significantly higher mean sco 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high score 83); and (4) Lisbon and Prague present lower scores in Health Outcom

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI r from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metro variation in the distribution of the value-scores was revealed acro components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise compariso respect to Health Determinants and Health Outcomes Indices. Four emerge: (1) Stockholm stands out with a significantly higher mean sco 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high score 83); and (4) Lisbon and Prague present lower scores in Health Outcom

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geograp variation in the distribution of the value-scores was revealed across metropolitan areas in components. Overall, almost all the municipalities registered value-scores above 50 (the PHI r from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   Note: The symbols and identify the metropolitan areas where scores were found to be statisticall different. By way of example: Brussels presents mean scores that are statistically different from Athens Barcelona, and Lisbon, London and Turin (with higher scores: ) and from Stockholm (with lower scores: The symbols and only display lower or higher differences (respectively), although not statisticall significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan area variation in the distribution of the value-scores was revealed across metropoli components. Overall, almost all the municipalities registered value-scores above 5 from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4

Results
When the EURO-HEALTHY PHI was applied to the metrop variation in the distribution of the value-scores was revealed acros components. Overall, almost all the municipalities registered value-sc from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparison respect to Health Determinants and Health Outcomes Indices. Four emerge: (1) Stockholm stands out with a significantly higher mean sco 87 on both components); (2) Athens, Barcelona and Lisbon pre Determinants (below 62.7); (3) Barcelona and Turin present high score 83); and (4) Lisbon and Prague present lower scores in Health Outcom

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4   The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical variation in the distribution of the value-scores was revealed across metropolitan areas in both components. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges from 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with respect to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas emerge: (1) Stockholm stands out with a significantly higher mean score in both components (above 87 on both components); (2) Athens, Barcelona and Lisbon present lower values in Health Determinants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around 83); and (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found.

Results
When the EURO-HEALTHY PHI was applied to th variation in the distribution of the value-scores was revea components. Overall, almost all the municipalities registered from 0 to 100), with 31% attaining 75 and above in both com Tables 3 and 4    O-HEALTHY PHI was applied to the metropolitan areas, a geographical tribution of the value-scores was revealed across metropolitan areas in both l, almost all the municipalities registered value-scores above 50 (the PHI ranges 31% attaining 75 and above in both components. presents the results of the pairwise comparisons of metropolitan areas with eterminants and Health Outcomes Indices. Four groups of metropolitan areas lm stands out with a significantly higher mean score in both components (above nents); (2) Athens, Barcelona  sults When the EURO-HEALTHY PHI was applied to the metropolitan areas, a geographical tion in the distribution of the value-scores was revealed across metropolitan areas in both onents. Overall, almost all the municipalities registered value-scores above 50 (the PHI ranges 0 to 100), with 31% attaining 75 and above in both components. Tables 3 and 4 presents the results of the pairwise comparisons of metropolitan areas with ct to Health Determinants and Health Outcomes Indices. Four groups of metropolitan areas ge: (1) Stockholm stands out with a significantly higher mean score in both components (above n both components); (2) Athens, Barcelona and Lisbon present lower values in Health minants (below 62.7); (3) Barcelona and Turin present high scores in Health Outcomes (around nd (4) Lisbon and Prague present lower scores in Health Outcomes (below 66.7). The symbols and only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Table 4. Pairwise comparisons of the differences between metropolitan areas mean scores from the Health Outcomes Index.
only display lower or higher differences (respectively), although not statistically significant. NA = No group was found. Figure 2 presents a 12% variation in the Health Determinants Index (CV = 0.12) and a 10% variation in the Health Outcomes Index (CV = 0.10). Although the variation is lower when we analyse each metropolitan area, higher internal variability was identified for Brussels and Athens in Health Determinants (CV ≥ 0.074), whereas in Health Outcomes, the same variability was found in Turin and London (CV ≥ 0.061).  In addition, this figure reveals that 55% of the municipalities in this study perform better in Health Determinants than the region where they are located [31]. Exceptions are found in Berlin-Brandenburg, Brussels and Turin, where more than 60% of the municipalities present worse values. In Health Outcomes, the opposite was found. Berlin-Brandenburg and Brussels are the exceptions, where more than 70% of the municipalities present better values. Prague stands out since all the municipalities perform better than the region in Health Determinants and worse in Health Outcomes.
The distribution of Health Determinants is not homogeneous across and within the metropolitan areas (Figure 3a). There is a gradient from northern European to southern European countries, with higher scores being found in Stockholm, Berlin and Brussels and lower scores in Lisbon, Athens and Barcelona. When looking at the within-metropolitan areas inequalities, the geographical pattern differs when we compare scores from a centre-periphery model point of view. The metropolitan centre of Brussels and Berlin-Brandenburg present lower scores when compared with the municipalities located in the periphery. The opposite is found in Stockholm.
As for Health Outcomes, the north-south gradient is not evident (Figure 3b). Along with Stockholm, Turin and Barcelona registered higher scores. Lower scores were identified in Lisbon and Prague. The geographical variation in the distribution of the value-scores across municipalities is considerable when compared to Health Determinants, with no clear pattern being found in the majority of the metropolitan areas. However, it is visible that the centres from Brussels, Athens and Berlin-Brandenburg present lower value-scores than the periphery.  The value-scores are displayed by using classification by Equal interval, taking into account the PHI minimum and maximum scores (from 0 to 100). The colour coding of the classes used a gradation inspired by a traffic light system (from red to green). In the case of the metropolitan areas, the light green represents the municipalities with worse population health and dark green represent better scores. Figure 4 illustrates the presence of clusters in all metropolitan areas and in both components, apart from Barcelona, Turin and Berlin-Brandenburg where they were only identified for the Health Determinants Index. Almost 1/4 of the population being studied lives in municipalities located in the Low-Low clusters (concentration of lower value-scores) in Health Determinants while only 8% are living in clusters characterised by a concentration of higher value-scores (High-High). For the Health Outcomes Index, the rates are also relevant:11% are clustered in Low-Low and 4% are clustered in High-High. The analysis also revealed populations living in municipalities classified in the cluster Low-Low for both Health Determinants and Health Outcomes indices in Athens (17%), Prague (16%), London (13%), Brussels (7%) and Stockholm (2%). In Brussels (0.5%), London (2%), Athens (8%) and Stockholm (21%), there are also municipalities classified in the cluster High-High for both indices. , at the municipal level. Note: The value-scores are displayed by using classification by Equal interval, taking into account the PHI minimum and maximum scores (from 0 to 100). The colour coding of the classes used a gradation inspired by a traffic light system (from red to green). In the case of the metropolitan areas, the light green represents the municipalities with worse population health and dark green represent better scores. Figure 4 illustrates the presence of clusters in all metropolitan areas and in both components, apart from Barcelona, Turin and Berlin-Brandenburg where they were only identified for the Health Determinants Index. Almost 1/4 of the population being studied lives in municipalities located in the Low-Low clusters (concentration of lower value-scores) in Health Determinants while only 8% are living in clusters characterised by a concentration of higher value-scores (High-High). For the Health Outcomes Index, the rates are also relevant:11% are clustered in Low-Low and 4% are clustered in High-High. The analysis also revealed populations living in municipalities classified in the cluster Low-Low for both Health Determinants and Health Outcomes indices in Athens (17%), Prague (16%), London (13%), Brussels (7%) and Stockholm (2%). In Brussels (0.5%), London (2%), Athens (8%) and Stockholm (21%), there are also municipalities classified in the cluster High-High for both indices. Indices. Note: The figure represents the clusters identified for both the Health Determinants Index (backward diagonal shading) and the Health Outcomes Index (forward diagonal shading). Blue lines represent the municipalities with low value-scores that are surrounded by municipalities also with low value-scores (cluster Low-Low). Red lines represent the municipalities with high value-scores that are surrounded by municipalities which also register high value-scores (cluster High-High).

Discussion
The objective of this study was to analyse health inequalities as measured by Health Determinants and Health Outcomes indices, across and within different European metropolitan areas.
The results contribute to deepening the knowledge about health at the urban level and follow previous work done across European regions [31] under the scope of the EURO-HEALTHY project. In the application of the Population Health Index (PHI), designed to evaluate health in two multidimensional components, Health Determinants and Health Outcomes, this study examines the results obtained from nine metropolitan areas (Athens, Barcelona, Berlin-Brandenburg, Brussels, Note: The figure represents the clusters identified for both the Health Determinants Index (backward diagonal shading) and the Health Outcomes Index (forward diagonal shading). Blue lines represent the municipalities with low value-scores that are surrounded by municipalities also with low value-scores (cluster Low-Low). Red lines represent the municipalities with high value-scores that are surrounded by municipalities which also register high value-scores (cluster High-High).

Discussion
The objective of this study was to analyse health inequalities as measured by Health Determinants and Health Outcomes indices, across and within different European metropolitan areas.
The results contribute to deepening the knowledge about health at the urban level and follow previous work done across European regions [31] under the scope of the EURO-HEALTHY project. In the application of the Population Health Index (PHI), designed to evaluate health in two multidimensional components, Health Determinants and Health Outcomes, this study examines the results obtained from nine metropolitan areas (Athens, Barcelona, Berlin-Brandenburg, Brussels, Lisbon, London, Prague, Stockholm and Turin) which represent different European regions and heterogeneous geographic, social and economic contexts.
Overall, it was found that: (i) Strong population health inequalities exist across metropolitan areas, with municipalities from Southern and Eastern countries presenting, in general, worse value-scores; (ii) metropolitan areas present better health, measured by Health Determinants, than the region where they belong, although some exceptions were found, and; (iii) Municipalities with worse Health Determinants scores tend to also perform worse on Health Outcomes.
Thus, the analysis of the distribution of the value-scores on both indices shows a high dispersion across metropolitan areas: In Health Determinants the range goes from 49 (Athens) to 92 (Stockholm) and in Health Outcomes, from 62 (Lisbon) to 99 (Stockholm). The fact that urban areas from North-Western countries present better health scores than the Eastern and Southern ones is not new and is aligned with results from previous studies on population health in Europe [31,35,50].
Simultaneous to a high difference in value-scores among metropolitan areas, there is a considerable variation within municipalities of the same metropolitan area. Of note are Brussels and Athens, which display a clear geographic variation in Health Determinants scores, and Turin and London in Health Outcomes. Previous studies also identified the presence of inequalities within these metropolitan areas, although at a more detailed scale [29,35,51]. For example, in the Lisbon case, geographic disparities between municipalities are not evident as expected, considering other studies on health inequalities at the small area level [19,21,27].
As the PHI model was previously applied to EU regions, it offered the possibility to compare the population health scores of metropolitan areas to those performed by the respective regions where they are located. In opposition to what it was identified for the regions, the municipalities from the metropolitan areas often perform better on Health Outcomes than on Health Determinants. Nonetheless, it was found that most municipalities performed better scores in the Health Determinants index when compared with the regional scores. Prague stands out as a paradigmatic case: All municipalities present significantly better scores than the respective region in Health Determinants, performing worse in Health Outcomes. These results may be understood as 'ambiguous', considering that the country and specifically the capital were emerging from a long-period of social and economic stagnation and recession, with negative impacts on health determinants [52,53]. The contrary occurs with Berlin-Brandenburg, Brussels and Turin, considering that these metropolitan areas present worse Health Determinants scores than their respective regions. One plausible explanation is that when they are compared with the larger administrative region, they perform worse in important health determinants, such as high levels of air pollution, ageing and crime, indicators used to build the PHI. According to another recent study from the EURO-HEALHTY project, using data from the same nine metropolitan areas, revealed that worse air quality is typically encountered in deprived European urban areas [54]. Still, those health determinants do not affect those metropolitan areas equally. Brussels, for instance, is younger than the rest of the country [55]. Unemployment and poor housing conditions provide a better explanation for this metropolitan area [56].
The application of LISA to detect spatial concentrations of similar scores within the same metropolitan area revealed that the share of the population living in Low-Low clusters (concentration of lower value-scores) in Health Determinants is three times higher than those living in clusters characterized by a concentration of higher value-scores (High-High cluster). Also, in a considerable number of urban areas, clusters were revealed in both indices. This is the case of Brussels, London, Athens and Stockholm, with clusters of High-High and Low-Low value-scores for both indices. The municipalities that share worse health determinants and worse health outcomes (Low-Low clusters) should be pointed out as 'urban zones in alert' calling the attention of local policy makers to the need to address population health in an inter-sectoral and integrated way. Measuring health at the local level is complex since there are diverse and interconnected factors operating at different scales in the same place [57]. The result of the application of a single index to depict inequalities is not immune to criticism and should be interpreted considering some limitations. Though the greatest asset of having aggregated indices is the simplification of data and the possibility to compare different geographical units using a single measure, there is always an amount of information that is lost [58]. Thus, some aspects regarding the administrative delimitation of metropolitan areas and the type of indicators used in the PHI model may contribute to masking inequalities. The lack of consensus on the delimitation of the metropolitan areas led to the utilization of different functional definitions and administrative levels. Although this issue was already referred to in previous studies as a limitation of ecological studies in Europe, so far, there remains no solution [28,29,34,35]. Moreover, the population size of municipalities within the same metropolitan area (e.g., Barcelona has a municipality with more than 1.5 M inhabitants and other with less than 200,000) can introduce an important MAUP (Modifiable Areal Unit Problem) effect [59].
According to Rothenberg and others [44], inequality among countries is mirrored in the inequality within their regions and cities. Therefore, bottom-up approaches based on local data and knowledge is of high relevance to promoting equity-based policies [1,10,36,40,[60][61][62]. The type of indicators used in the construction of urban health indices is key to detect inequalities among municipalities. The Health Determinants and Health Outcomes indices analysed in this study are the result of an aggregation of multiple dimensions and indicators that were selected as relevant to evaluate population health at a regional level. This participatory process, conducted with stakeholders representing the countries of these metropolitan areas and the indicators selected, was analysed by Freitas and others [45]. Although indicators had been selected for a regional scale, they were considered as adequate and reliable proxies to measure health determinants and health outcomes at the metropolitan level. Yet, indicators focusing on specific urban characteristics (e.g., access to green spaces, transportation, social protection) and more sensitive to the local social, physical and built environment [6,[8][9][10] are not included in the PHI, not because they were not considered relevant, but due to the lack of availability and/or comparability across regions, two criteria required for an indicator to be included in the PHI [45,63].
The application of an urban health index to the municipalities of nine European metropolitan areas, very different in contexts and levels of data availability, brought some constraints and represented some risks, implying the need to apply a predefined and common protocol of data collection and harmonization. The lack of indicators at the administrative level of the municipality led to the use of data at coarser geographical resolution-regional or national. This indicates that besides the need to reinforce the data collection at the sub-national level, already identified in previous studies [31,63], there are different levels of capacity from national and municipal statistics to collect urban health-related data at the local level.
Regardless of the above-mentioned limitations, the strength of this study is to show that geographical analysis is needed when investigating health inequalities. These results should be understood as a point of departure to show that there are inequalities within the same region and metropolitan area and there is a need to examine and act locally to address any existing inequities. And though we focused only on aggregated indices of Health Determinants and Health Outcomes, it offers enough clues that a multi-sectoral commitment between the health sector and other sectors at the local level would be valuable when it comes to promoting population health.

Conclusions
This study adds evidence to the debate on the existence of health inequalities across Europe: Not only across countries and regions, but also among and within urban areas.
The application of a single measure-the Population Health Index-to evaluate health determinants and health outcomes in nine European metropolitan areas show that not only do they exhibit differences between them, but that municipalities within the same urban area face different population health profiles. Conversely, municipalities from these nine metropolitan areas have now a common tool to compare themselves with and share the lessons on how to tackle similar health problems.
The responsibility of promoting health does not lie exclusively in the health sector or with the national government. More and more, municipalities across Europe are demonstrating their responsability when it comes to adopting policies that improve the health and well-being of their citizens.