Energy Reports 8 (2022) 1249–1259 Contents lists available at ScienceDirect Energy Reports journal homepage: www.elsevier.com/locate/egyr Research paper Geographical inequalities in energy poverty in a Mediterranean city: Using small-area Bayesian spatial models ∗ Marc Marí-Dell’Olmo a,b,c , , Laura Oliveras a,b,d , Carlos Vergara-Hernández e , Lucia Artazcoz a,b,c,d , Carme Borrell a,b,c,d , Mercè Gotsens a,b , Laia Palència a,b,c , María José López a,b,c,d , Miguel A. Martinez-Beneito f,g a Agència de Salut Pública de Barcelona, Pl. Lesseps 1, 08023, Barcelona, Catalonia, Spain b Institut d’Investigació Biomèdica Sant Pau (IIB Sant Pau), Sant Quintí 77, 08041, Barcelona, Catalonia, Spain c CIBER Epidemiología y Salud Pública (CIBERESP), Av. Monforte de Lemos, 3-5, 28029, Madrid, Spain d Department of Experimental and Health Sciences, Universitat Pompeu Fabra, Doctor Aiguader 88, 08003, Barcelona, Catalonia, Spain e Vaccine Research Unit, FISABIO, Av Cataluña 21, 46020, Valencia, Spain f Departament d’Estadística i Investigaciò Operativa, Dr. Moliner 50, 46100, Burjassot, Valencia, Spain g Unitat Mixta de recerca en mètodes estadístics per a dades biomédiques i sanitàries, UV-FISABIO, Dr. Moliner 50, 46100 Burjassot, Valencia, Spain article info a b s t r a c t Article history: Energy poverty (EP) is becoming an increasingly important problem in the urban contexts of southern Received 2 September 2021 Europe. In Barcelona, EP indicators are higher than those of the European Union and are strongly Received in revised form 30 November 2021 associated with poor health status and high use of health services and medication, becoming a major Accepted 7 December 2021 public health problem. EP is unevenly distributed in the population of Barcelona, according to axes Available online 30 December 2021 of social stratification. However, its geographic distribution at the small-area level remains unknown Keywords: because it cannot be directly estimated with the available information sources and commonly used Energy poverty methods. Therefore, the aim of this study was to analyze geographical inequalities in EP in Barcelona Fuel poverty by estimating reliable small-area EP indicators and a composite indicator (index). We used a novel Index method that allowed us to obtain 6 EP indicators for the 73 Barcelona neighborhoods and an EP index Geographical inequalities from a principal component analysis of these indicators. We found major geographical inequalities Multivariate analysis in the distribution of EP in Barcelona. Many neighborhoods had significantly higher EP than the city Bayesian model Spatial analysis average, and these areas made up 3 well-defined spatial clusters. Therefore, the estimated small-area Social determinants of health indicators and index allowed identification of the most affected neighborhoods. These results indicate the need to prioritize these areas for local interventions to alleviate EP, and could also be used for policy making. © 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction major public health problem (Bosch et al., 2019; Marí-Dell’Olmo et al., 2017; Marmot Review Team, 2011; Oliveras et al., 2020; Energy poverty (EP) is defined as the inability of a household Thomson et al., 2017b). to secure a socially and materially required level of energy ser- The distribution of EP in the population is uneven, following vices in the home (Bouzarovski, 2013). EP is currently a social the classical lines of social stratification. The phenomenon par- problem in Europe and recent publications have shown that the ticularly affects people with a more disadvantaged social class, most affected countries are those in southern and eastern Euro- those born in low- and middle-income countries and women pean Union (EU) (Bouzarovski and Tirado Herrero, 2017; Recalde aged 65 years and older (Carrere et al., 2020; Oliveras et al., et al., 2019). For example, in 2016, 9.0% of the EU population 2020). Geographical inequalities in EP at the small-area level have was unable to keep their home adequately warm during the also been found in various contexts. Some of the best studied cold months, while in Spain this percentage increased to 13.4% (Oliveras et al., 2021b). Several studies have shown that EP is regions are England (Baker et al., 2003; DECC, 2009; Fahmy et al., strongly associated with worse health and wellbeing, making it a 2011a; Fahmy and Gordon, 2007; HM Government, 2013; Robin- son et al., 2018; Sun and Sundell, 2011), Scotland (Morrison and ∗ Correspondence to: Agència de Salut Pública de Barcelona, Plaça Lesseps Shortt, 2008) and Portugal (Gouveia et al., 2019), with all showing 1, 08023 Barcelona, Spain. marked geographical inequalities. Some studies have found such E-mail address: mmari@aspb.cat (M. Marí-Dell’Olmo). inequalities within cities such as Oberhausen (Germany) (März, https://doi.org/10.1016/j.egyr.2021.12.025 2352-4847/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 2018), London (United Kingdom) and Madrid (Spain) (Martín- The aim of this study was to analyze geographical inequalities Consuegra et al., 2019; Sanchez-Guevara et al., 2019). Other stud- in EP in Barcelona by estimating reliable EP indicators and a ies have found marked geographical inequalities in EP at the composite indicator (index) at the neighborhood level. To the best medium- to large-area level, such as districts, in cities such as Va- of our knowledge, this is the first study to analyze EP in this small lencia and Barcelona (Ajuntament de Barcelona, 2018a; Gómez- sub-division of the city, making inequalities on this scale visible Navarro et al., 2021; Tirado Herrero, 2018). However, the geo- for the first time. Furthermore, we employ a novel methodology graphical inequalities found within cities may be oversimplified that permits obtaining reliable EP indicators at small-area level, by the size of these areas, leading to the danger of falling into while taking into account both the potential dependence between the modifiable areal unit problem (MAUP) which could lead, EP indicators and the spatial dependence of the values of each of for example, to misinterpretations by policymakers and local the EP indicators (between adjacent neighborhoods). The EP indi- authorities (Openshaw, 2016). Unfortunately, in these cities, EP cators and index estimates can be finally represented graphically distribution at the smaller area level such as neighborhoods or in the form of maps, thus allowing, among other things, the study census tracts is still unexplored. of its geographical patterns, and the detection of neighborhoods Most EP analyses are undertaken at the country level or even on a larger scale such as the EU level. However, estimates of EP at with significantly greater or lower EP. In addition, the EP indica- the small-area level are a versatile and powerful tool that could tors used in this study are among those recommended by the EU be used for the following (Baker et al., 2003; Morrison and Shortt, Energy Poverty Observatory to study and monitor EP. Finally, we 2008; Wilson et al., 2012): (1) identifying the most affected make available the code used to perform the statistical analyses areas and prioritizing them in the fight against EP (developing in order to facilitate the replicability of the study in other settings. evidence-based local public policies prioritizing the most affected The paper is organized as follows. Section 2 specifies the meth- areas, including effective targeting of EP programs or developing ods used in this study: the design, the sources of information, the local affordable warmth and cold strategies); (2) identifying more study population, the variables studied, and the novel statistical contextual causes of EP, such as deficiencies in the housing stock approach that we propose. Section 3 illustrates the results of the or in the distribution network; (3) raising social and political study, including the estimated values of the EP indicators and awareness of EP as an issue of concern and providing civil society of the index obtained for each of the Barcelona neighborhoods, with tools for advocacy (maps are a very powerful tool for this as well as their geographic distribution. Section 4 incorporates purpose) and (4) advancing research, e.g. exploring the relation- the discussion of our main results, where we state the main ship between EP and health inequalities and other health and findings of our research, interpret the results and we comment deprivation indicators. the main limitations and strengths of the study. Finally, the EP EP is also becoming an increasingly important problem in the policy implications derived from the study, possible future work urban contexts of southern Europe. A clear example is the city and the conclusions of the study are given in Section 5. of Barcelona (Spain), where several EP indicators show values above those of the EU and EP has become an explicit prior- ity for the city council (Tirado Herrero, 2018). For instance, in 2. Methods 2016, 9.4% of the population in Barcelona could not afford to keep their homes adequately warm and 13.9% were in arrears 2.1. Design, information source, units of analysis and study popula- in utility bills in the last 12 months. In this city, EP is also a tion major public health problem. There is a proven strong association between EP and worse health status, as well as with a higher use We performed a cross-sectional study using EP indicators and of health services and medication, which substantially impacts sociodemographic variables at the individual level and sociode- health and health services at the population level (Carrere et al., mographic variables at the neighborhood level. Individual-level 2020; Oliveras et al., 2021a, 2020). However, the maximum level variables were obtained from the 2016 Barcelona Health Survey of spatial disaggregation at which the EP situation is known in (BHS). This survey was conducted by interviewing 3519 individu- detail are the 10 districts of the city, mainly because surveys are als. The sample was representative of the city and the 10 districts usually representative at this level at most. In contrast, there are of Barcelona (median population 164,881 people) (Bartoll et al., important city-level strategies focusing on a smaller level than the district, specifically on the neighborhood level. This is the 2018). The aggregated values of sociodemographic variables for case, for example, of the Barcelona Neighborhoods Plan, which each of the 73 neighborhoods of Barcelona (median population aims to reduce social inequalities in different neighborhoods of 20,369 people) was drawn from the 2016 municipal register the city with actions on education, employment, public space (Barcelona city council) and from the 2011 Housing and Popula- and urban ecology, among other areas (Ajuntament de Barcelona, tion Census (National Statistics Institute). The study population 2016a). In addition, it is known that people living in EP are more consisted of non-institutionalized individuals aged 15 years or vulnerable to climate change due to their reduced ability to adapt older who resided and were included in the Barcelona municipal (Jessel et al., 2019). For example, living in poorly insulated houses register of inhabitants in 2016. To aid understanding of this with no means of cooling does not protect people from high day- article, a map is provided with the geographical location of the time and nighttime temperatures and heat waves. Consequently, neighborhoods, their identification number and their full name among the objectives of the Barcelona City Council Climate Plan (Fig. A.1). for 2030 is the elimination of EP with actions designed to improve knowledge of the phenomenon and its impact on health and to 2.2. Study variables and data analysis reinforce interventions that are already under way (Ajuntament de Barcelona, 2018a). Moreover, the Public Health Agency of Barcelona is working on the design of a system to monitor over The EP index and small-area estimates of the EP indicators time and space the effects of climate change on health including, comprising the index were obtained by using a novel method. among many other indicators, EP and its effects on health (Marí- To promote the reproducibility of the research conducted, the Dell’Olmo et al., 2019a; Mercuriali et al., 2021). Therefore, there statistical methods and the R scripts used for data analysis are is a clear need for a tool able to detect the small areas with the documented in detail in an RMarkdown document (Appendix B). highest levels of EP. In summary, the method was based on the following 5 steps: 1250 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Step 1: Variables of the BHS that allowed detection of a situ- calculate the indicators in each neighborhood, taking into account ation of EP in households were identified, henceforth termed EP all the available demographic and spatial information. indicators. These indicators were: (1) households that could not Step 5: Once the estimates of the EP indicators were ob- afford to keep the home at an adequate temperature during the tained at the small-area level (smaller than originally planned cold months (Temp_cold); (2) households that could not afford in the BHS), to obtain the EP composite indicator or index (at to keep the home at an adequate temperature during the hot the neighborhood level), a principal component analysis (PCA) months (Temp_hot); (3) households in arrears in utility bills in was performed on the posterior mean of the proportions (of the the last 12 months (Arrears); (4) dwellings with leaks, dampness 6 EP indicators) obtained in step 4. In particular, the logit of in walls, floors, ceilings or foundations, and/or rot in floors, win- the posterior mean of these indicators was used to normalize dow frames or doors (Conditions); (5) dwellings without means of their values. Finally, the index was constructed from the first heating or with central heating or room-heating appliances, but component of the PCA that explained 91% of the variance. The not used when necessary (Heating); and (6) dwellings without indicator with the highest weight was inadequate temperature an air conditioner or with an air conditioner, but not used when in cold months, while the indicator with the lowest weight was necessary (Air_cond). Other EP indicators related to energy ex- that related to the presence of leaks and dampness (Table 1). penditure and household income were excluded due to the large As a result of the entire process described above, using poste- number of missing data, 33.7% and 48.9%, respectively. Thus, a rior means, point estimates of the proportions of each of the 6 EP total of 6 indicators were included (Table 1). Of note, all indicators indicators and the EP index were obtained for the 73 neighbor- are recommended and monitored by the EU Energy Poverty Ob- hoods. Henceforth, we will refer to these 6 indicators as ‘‘small- servatory, 2 of which are primary indicators and the remaining 4 area indicators’’ (SAI). Moreover, to quantify the statistical evi- are secondary indicators (EU Energy Poverty Observatory, 2021). dence of these parameters in each neighborhood, we estimated Step 2: Using the conceptual framework of the determinants the probability of each EP SAI and the index to be higher than the of health inequalities related to EP previously developed by our overall value for the city. This means that for all EP estimates (the group (Marí-Dell’Olmo et al., 2017), we identified those sociode- 6 SAIs and the index) and for each neighborhood, we calculated a mographic variables included in the BHS that could be related to probability of excess in comparison to the city average. These ex- EP (indicators selected in step 1) and which, in turn, were avail- cess probabilities were categorized with the following cut points: able at the neighborhood level from population-based records [0, 0.025), [0.025, 0.05), [0.05, 0.95), [0.95, 0.975), and [0.975, 1]. (not based on a sample). A total of 7 variables were obtained (Table 1). 3. Results Step 3: To make estimates at the small-area level (neighbor- hood) of each of the EP indicators, we used the M-model, a mul- Table 2 shows the distribution of each SAI and of the EP index tivariate model for estimation in small areas (Botella-Rocamora in the Barcelona neighborhoods, as well as their correlation ma- et al., 2015; Corpas-Burgos et al., 2019). This allowed, on the one trix. There was wide variability in the EP SAIs by neighborhood. hand, analysis of all EP indicators in the same model, taking into For example, the SAI of inadequate temperature in cold months account their possible dependence and, on the other hand, to take had a median of 8.39%, the 25th and 75th percentile values were into account the possible spatial dependence of data between ad- 6.62% and 13.28%, respectively, and their values ranged from jacent neighborhoods by means of the Leroux distribution (Leroux 2.10% to 35.38%. Therefore, all the dwellings in neighborhoods et al., 2000). above the 3rd quartile showed this problem, at least twice as In summary, a Bayesian hierarchical model was estimated often as those below the 1st quartile. The EP SAIs were highly where the dependent variables were the 6 EP indicators (obtained correlated with each other, with the SAIs referring to inadequate in step 1). As covariates of the model, all variables obtained temperatures in the cold and hot months and the availability of in step 2 were explored. Covariate effects were included into air conditioning being the most closely correlated, with correla- the small area estimates as introduced in Vergara-Hernández tions greater than 0.95. In contrast, the presence of conditions et al. (2021). Not all of the covariates mentioned were included such as leaks or dampness showed the lowest correlations with in the final model, as we did not find a relationship with EP the remaining SAIs. This is possibly the reason why this SAI indicators for all of them. To select the covariates for the final showed the lowest weight, and therefore the lowest contribution, multivariate model, we used the following criteria: (1) each of to the EP index (Table 1). the covariates had to be significantly associated with at least The rows in Table 3 show the 73 Barcelona neighborhoods, 2 dependent variables and, (2) a lower value of the deviance ordered from highest to lowest value of the EP index, while the information criterion (DIC) of multivariate models. The covariates columns show the 6 SAIs and the EP index. Each cell shows the of the final model were age group, educational level, country SAI percentages and the index values with their 95% credible of birth, housing tenure and employment status (Table 1). A intervals (95%CI). The cells are colored according to the categories random effects matrix combining the spatial and multivariate of the probability of EP excess with respect to the overall city dependence of the EP indicators was included in the model. Cases specified above. Thus, red cells indicate a high probability of the with missing values in any of the covariates were excluded in the EP SAIs and of the index being higher than that for the rest of estimation of the final model and therefore this model included the city and green cells a low probability. Fifteen neighborhoods a sample of 3378 individuals instead of 3519 (4% loss). were notable for their high values of the EP index and a significant All models were fitted using a full Bayesian approach. Posterior excess of EP in 3 or more EP SAIs. Thus, for example, the neighbor- distributions were obtained using Markov Chain-based Monte hood with the highest EP according to the index obtained (3.50) Carlo methods via WinBUGS and R (Lunn et al., 2012; R Core was Ciutat Meridiana. Examination of the different SAIs of this Team, 2021). neighborhood showed that 31.41% of households could not afford Step 4: The above-mentioned model allowed us to estimate to keep their homes at an adequate temperature during the cold the relationship between the demographic covariates and the months and that 35.95% could not do so during the warm months, EP indicators, as well as the spatial effects that could not be 41.25% of households had at least one delay in utility bills in explained by the covariates. Once these components had been es- the last 12 months, 21.88% reported leaks, dampness in walls, timated, the exhaustive demographic information for each neigh- floors, ceilings or foundations, and/or rot in floors, window frames borhood was used, together with the estimated spatial effects, to or doors, 32.29% had no means of heating the home or did not 1251 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Table 1 Characteristics of each of the 6 energy poverty indicators (step 1 of the method) and the 7 sociodemographic variables related to energy poverty available in the Barcelona Health Survey and at the neighborhood level in population-based registries (step 2 of the method). Barcelona, 2016. Name Description Data source Year Overall Number of Weight in the value for missing values 1st principal the city (%) component analysis dimension Energy poverty indicators Temp_cold Identifies households that cannot afford to Barcelona Health Survey 2016 9.36 39 0.47 maintain the dwelling at an adequate temperature during the cold months. Temp_hot Identifies households that cannot afford to Barcelona Health Survey 2016 11.41 43 0.45 maintain the dwelling at an adequate temperature during the hot months. Arrears Identifies households that have one or more Barcelona Health Survey 2016 13.90 738 0.45 arrears in utility bills in the last 12 months Conditions Identifies dwellings with leaks, dampness in Barcelona Health Survey 2016 9.66 0 0.25 walls, floors, ceilings or foundations, and/or rot in floors, window frames or doors Heating Identifies dwellings without means of heating Barcelona Health Survey 2016 9.96 20 0.40 or with central heating or room-heating appliances, but not used when necessary Air_cond Identifies dwellings without an air conditioner Barcelona Health Survey 2016 51.43 23 0.39 or with an air conditioner, but not used when necessary Sociodemographic variables related to energy poverty available in the Barcelona Health Survey and at the neighborhood level in population-based registries Sex Describes sex classified into female and male Barcelona Health Survey 2016 0 Municipal Register 2016 – Age groupa Describes age according to the groups: 0–19, Barcelona Health Survey 2016 0 20–39, 40–59, 60–79,>79 years Municipal Register 2016 – Education Describes education level grouped into 4 Barcelona Health Survey 2016 44 levela categories: no education, primary education, Municipal Register 2016 – secondary education and higher education Country of Describes country of birth classified into Barcelona Health Survey 2016 9 birtha high-income and low-and middle-income Municipal Register 2016 – countries according to the 2018 World Bank classificationb Housing Describes the tenancy regime distinguishing Barcelona Health Survey 2016 66 tenurea between paid property and others (including Census 2011 – property paying mortgage, rent at market price, social and other situations) Employment Describes employment status classified into Barcelona Health Survey 2016 43 statusa unemployment and other Municipal Register 2016 – Households Describes household composition classified Barcelona Health Survey 2016 0 composition into single-parent households and other Municipal Register 2016 – a Covariates included in the final statistical model. b The World Bank World Bank Country and Lending Groups – World Bank Data Help Desk Available online: https://datahelpdesk.worldbank.org/knowledgebase/articles/ 906519 (accessed on Jul 25, 2021). Table 2 Summaries and Spearman correlations between the 6 energy poverty small-area indicators and index. Barcelona, 2016. Energy poverty Minimum P25 P50 P75 Maximum Correlation matrix indicators and index Temp_cold Temp_hot Arrears Conditions Heating Air_cond Index Temp_cold 2.10 6.62 8.39 13.28 35.38 1.00 0.99 0.87 0.77 0.89 0.96 0.98 Temp_hot 3.12 8.50 10.75 16.49 42.72 0.99 1.00 0.84 0.78 0.84 0.95 0.97 Arrears 3.24 11.00 14.50 19.94 41.25 0.87 0.84 1.00 0.56 0.87 0.82 0.91 Conditions 4.85 7.57 9.13 11.31 22.29 0.77 0.78 0.56 1.00 0.75 0.88 0.79 Heating 2.58 7.19 9.10 11.54 32.29 0.89 0.84 0.87 0.75 1.00 0.93 0.94 Air_cond 25.60 45.21 52.29 60.88 83.19 0.96 0.95 0.82 0.88 0.93 1.00 0.98 Index −3.33 −0.68 −0.12 0.81 3.50 0.98 0.97 0.91 0.79 0.94 0.98 1.00 P: Percentile. use heating when necessary, and 83.19% had no air conditioning the respective average percentages for the city (9.36%; 11.41%; or did not use it when necessary. These values are higher than 13.90%; 9.66%; 9.96% and 51.43%) with a probability greater than 1252 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Fig. 1. Geographical distribution of the energy poverty index at the neighborhoods level (A) and probability of the index being higher than the overall value for the city (B). Barcelona, 2016. 0.95. Finally, 20 of the 73 neighborhoods had a significantly In these areas, there was also a high percentage of dwellings higher EP index (probability greater than 0.95) than the average without heating or which, even if heating was available, did not use it when necessary (figures S5.A and S5.B), e.g., 32.29% value for the city. in Ciutat Meridiana, 25.68% in La Barceloneta and 23.66% in El Fig. 1 shows the spatial distribution of the EP index and its Raval (Table 3). Finally, in the city, 51.43% of dwellings reported probability of being higher, for each neighborhood, than for the not having air conditioning or having it but not using it when necessary and, according to our study, this percentage increased city overall. The geographical pattern of the EP index was uneven, in the city center and southeast, and in various northern neigh- with the northernmost neighborhoods, some in the north-west, borhoods, especially those located in the north and northeast those in the historic city center (around the port area) and those (Figures S6.A and S6.B). In some of these neighborhoods, this in the south-east showing the highest values of the EP index SAI could even reach values above 75%, as is the case of Ciutat Meridiana (83.19%), Baró de Viver (81.83%), Vallbona (76.19%) and (Fig. 1A). In addition, there were 3 spatial clusters of neighbor- El Raval (76.00%). hoods with significantly higher EP indexes (probability greater than 0.95) than the average value for the city (Fig. 1B). 4. Discussion However, examination of the different geographical patterns 4.1. Main findings of the EP SAIs (see Appendix B) revealed some differences with respect to the index. The 2 EP SAIs related to inadequate house- This study identified substantial geographical inequalities in hold temperatures in the cold and warm months followed a the distribution of EP in the city of Barcelona. The highest EP levels were found in the northernmost neighborhoods, some in similar geographical pattern to each other and to the EP index the north-west, those in the historic city center and those in (Figures S1.A, S1.B, S2.A and S2.B). However, the EP SAI referring the south-east. Within these areas, there were 3 well-defined to the warm months showed fewer significant percentages (than spatial clusters of neighborhoods with a significant excess of EP compared with the city as a whole. the EP SAI considering the cold months) in 2 neighborhoods close to the coast (El Gòtic and Barceloneta). The SAI referring 4.2. Interpretation of results to late payment of energy bills showed a significant cluster of high percentages in the historic center and followed a more EP is a complex, multidimensional and dynamic construct that affects different people depending on the combination of multi- spatially heterogeneous pattern, with numerous neighborhoods ple social and economic determinants (Bouzarovski and Petrova, with high percentages in the north and northwest (figures S3.A 2015; Marí-Dell’Olmo et al., 2017). Important intermediate de- and S3.B). The SAI of dwellings with physical deficiencies (figures terminants are low household income, low energy efficiency of housing and high energy prices (Hills, 2011). These intermediate S4.A and S4.B) showed significantly higher values than the city determinants are in turn the result of structural determinants in the historic city center, with 2 neighborhoods in the north related to the labor, housing and energy markets and their poli- of the city (Ciutat Meridiana and Baró de Viver) standing out. cies (Recalde et al., 2019). All these complexities make measuring 1253 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Table 3 Energy poverty index and small-area indicators with their 95% credible intervals (95%CI) and their probability of being higher than the overall value for the city, for each of the 73 neighborhoods. Barcelona, 2016. (continued on next page) 1254 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Table 3 (continued). EP a difficult task that requires methods that take into account there are older, hastily constructed buildings with low construc- the multiple characteristics of EP and allow its optimal quan- tion standards (Ajuntament de Barcelona, 2016b; Donat et al., tification. In recent years, various proposals have been made 2015). On the other hand, the less significant percentages of to try to measure EP more comprehensively through composite the SAI of inadequate temperatures in the warm months in 2 indicators (or indices). These range from simple combinations of neighborhoods close to the coast could be due to the effect of EP indicators to more complex methods, vary in the geographical sea breezes on daytime temperatures in these neighborhoods level of analysis, and have been proposed for different contexts (Yamamoto and Ishikawa, 2020). (e.g. high- and low-income countries) (Siksnelyte-butkiene et al., In turn, the SAI referring to arrears in utility bills showed 2021). The index presented in this study is based on 6 underlying a similar geographical pattern to that of the income of its in- indicators, which are also estimated at the neighborhood level, a habitants. Except for some neighborhoods on the north-eastern more detailed disaggregation level than that used at the sampling border of the city, such as La Verneda i la Pau and El Besos stage. i el Maresme, which have very low incomes (Ajuntament de In Barcelona, most of the key intermediate determinants of Barcelona, 2017), their values for this EP SAI were not signifi- EP show a geographical pattern in the city. Additionally, these cantly higher than the Barcelona average. In contrast, the income geographical patterns may differ across these determinants. For level of Barri Gòtic (located in the city center) is medium-high, example, the household disposable income per capita index cal- but the percentage of households with late payment of bills is culated for each neighborhood shows large contrasts between 24.24% (1.73 times higher than that of the city as a whole). This the north and west of the city (Ajuntament de Barcelona, 2017). could be because the disposable household income per capita Regarding proxy indicators of the energy efficiency of dwellings, indicator calculated by the City Council may be affected by an such as the age of buildings, there are notable geographical dif- ecological fallacy. According to the National Institute of Statis- ferences that reflect the historical process of urbanization of the tics (Instituto Nacional de Estadística, 2020), many areas of this city. Thus, the city center has a high percentage of old buildings neighborhood have a much lower average income level (per (pre-1940), while in other areas most of the housing stock was person) than that of the rest of the city, while in other areas built in the decades of greatest growth of the city, because the there has been a major gentrification where many residents of the city had to absorb a large wave of migration (between 1961 and neighborhood have been replaced by inhabitants from northern 1980), and finally, in the eastern part of the city, there have Europe, with higher purchasing power (Ajuntament de Barcelona, been large urban development operations in the last 30 years 2017; Sánchez-Ledesma et al., 2020). This artifact could modify (Ajuntament de Barcelona, 2021, 2016b; Donat et al., 2015). The the average income level of the neighborhood but have little state of preservation of buildings is also unevenly distributed influence on the percentage of people with difficulties in paying across the city. The highest percentages of buildings considered bills. to be in a dilapidated or poor state of repair are found in the With regard to the spatial distribution of the SAI that identified city center and in neighborhoods with greater socioeconomic dwellings with physical deficiencies, such as the presence of leaks deprivation, where the percentage can exceed 10% (Ajuntament or damp, the highest values were concentrated in neighborhoods de Barcelona, 2021, 2016b). Therefore, the uneven spatial distri- with older buildings and poor construction quality, such as those bution observed in this study in the EP SAIs and index was similar in the historic city center. In addition, the Ciutat Meridiana neigh- to that expected and is consistent with the different patterns of borhood stands out, where 94.85% of dwellings were built hastily some of its determinants in the city (Robinson et al., 2018). between 1961 and 1980 and many of them show construction EP SAIs related to inadequate home temperatures in the cold deficiencies, and Baró de Viver, with 14.95% of dwellings in a poor and warm months follow a very similar geographical pattern, state of repair (Ajuntament de Barcelona, 2021). probably due to poor building isolation. In addition, neighbor- An important finding was the continued existence of neigh- hoods with higher values for these SAIs tend to be those where borhoods in the city with a high percentage of the population that 1255 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 did not have heating or did not use heating when necessary. This as it is the people themselves who assess whether, for example, factor is relevant because, although Barcelona enjoys a temperate the thermal comfort is adequate or not(Thomson et al., 2017a). climate, cold temperatures currently have a greater impact on Finally, it should be noted that official statistics sometimes make mortality in the city than warm temperatures, and it is estimated the most affected groups invisible, such as, in the case of the study that in the period 1992–2015 they were responsible for more in question, people who do not have a rental or property contract than 5% of mortality in Barcelona (Marí-Dell’Olmo et al., 2019b). (squatters) and who therefore cannot access even regular energy Barcelona also has periods of the year with high or extremely supplies. high temperatures. In the last 34 years, there have been 8 heat waves and, due to climate change, it is estimated that this fre- 5. Policy implications, future work and conclusion quency will increase significantly, with between 1 and 4 a year by the end of the century (Ajuntament de Barcelona, 2018a). This study has different EP policy implications in Barcelona. Moreover, these high temperatures and heat waves in Barcelona First, local governments could implement policies to reduce geo- pose a demonstrated risk and have a major impact on people’s graphical inequalities by focusing their EP interventions on the health (Borrell et al., 2006; Ingole et al., 2020; Marí-Dell’Olmo neighborhoods with the highest levels and specifically on the et al., 2019b). However, a high percentage of the households in populations most vulnerable to EP within these neighborhoods. the city report not having air conditioning or not using it when In this sense, the Barcelona City Council has implemented a necessary. In this study, this percentage rose to more than 75% in network of Energy Advice Points (EAP), which offer personalized some of the low and very low-income neighborhoods. attention on energy rights to the affected population (Ajuntament de Barcelona, 2018b). Currently, there is an EAP in each of the 10 4.3. Limitations and strengths districts of Barcelona, so they are evenly distributed throughout the city. Our index could be useful to improve this service with Obtaining reliable EP estimates at the small-area level is by proportional universality, taking into account the areas where no means a trivial task, since most information sources (usually there is a greater need to reinforce the EAPs and place new surveys) used to obtain EP indicators are not usually represen- resources in those neighborhoods with higher levels of EP. In tative at this geographical disaggregation. Several approaches to the same vein, interventions to improve the energy efficiency of estimating EP at the small-area level have been used (Baker buildings could be prioritized in neighborhoods with the worst et al., 2003; Castaño-Rosa et al., 2019; Centre for Sustainable EP SAIs related to inadequate temperatures, physical deficiencies Energy, 2001; DECC, 2009; Fahmy et al., 2011b; Gouveia et al., and lack of heating or air conditioning such as, for example, 2019; HM Government, 2013; März, 2018; Morrison and Shortt, funding programs for the energy rehabilitation of buildings. 2008; Robinson et al., 2018; Sanchez-Guevara et al., 2019; Simoes Finally, in the context of a climate emergency, reliable esti- et al., 2016; Wilson et al., 2012), ranging from simple procedures mates at the neighborhood level, such as the percentage of people such as looking for reliable proxy indicators of EP available at with inadequate summer temperatures in their homes or the the small-area level, to employing procedures with several steps percentage of people who do not have air conditioning or who (including several information sources with different scales and do not use it when necessary, could be useful from a climate statistical methods). justice point of view. These estimates allow both to prioritize the The main strength of this study is that, for the first time in targeting of interventions to reduce the health impacts of heat Barcelona, it obtained reliable estimates of EP at the neighbor- on the most vulnerable people and to strengthen existing ones, hood level, identifying significant geographical differences that such as increasing the quantity and quality of vegetation and could be highly useful for local EP policies. To obtain these esti- shaded areas in the city, improving the heatwave early warning mates, we designed a sophisticated new method, employing sta- systems, or through the creation of climate shelters. For example, tistical models that draw on the information provided by all the schools playgrounds of most affected neighborhoods could be EP indicators used, taking into account the information available transformed into climate shelters (introducing vegetation, shaded in adjacent neighborhoods, which is a key aspect in small-area areas and aquatic components) and be opened to the wider public estimation models. Moreover, our multivariate approach consid- in non-school period (Ajuntament de Barcelona, 2019; UIA - ers the EP indicators as potentially dependent. This assumption Urban Innovative Actions, 2019) However, in all these cases, once preserves the potential relationship, in contrast to models assum- the areas have been prioritized, interventions should focus on ing independence, which is the cornerstone for the subsequent people who are particularly vulnerable to EP and its effects on PCA used to construct the EP index. However, as a drawback, health, such as children, the elderly, women from disadvantaged less accurate estimates of EP SAIs may occur in some neighbor- social classes, dependent people and people with illnesses (Oliv- hoods at the outskirts of the city, due to the lack of information eras et al., 2020; Thomson and Snell, 2013). Therefore, a good from adjacent neighborhoods outside the city (frontier effect) practice could be, implementing care or social superblocks, ori- (Martinez-Beneito and Botella-Rocamora, 2021). ented to the integral attention of vulnerable people by domiciliary Another strength of the study is the use of the last BHS as teams (Barcelona city council, 2020). an information source, in which questions on EP were included The estimation of EP indicators and the EP index at small-area for the first time, providing a unique opportunity to explore this level as well as the proposed methodology may lead to interesting phenomenon with information from more than 3500 individuals. future research. The EP index could be used to study hypotheses Nonetheless, we were unable to include EP indicators based on in groups of neighborhoods according to their EP or to study its expenditure and income, which are the ones traditionally most association or interaction with other indicators as, for example, used in this type of studies. The variables needed to construct health indicators or other situations of socioeconomic hardship, them had a high percentage of missing values in the BHS, which like residential or food insecurity. On the other hand, from the invalidated their use. However, as previously stated, the indica- next editions of the city health survey, it will be possible to tors used are part of the main indicators recommended by the perform a spatio-temporal small-area analysis that will allow to EU Energy Poverty Observatory and cover different expressions study the evolution of the geographical pattern of EP, to quantify of EP (EU Energy Poverty Observatory, 2021). Furthermore, the the evolution of EP in each neighborhood in a reliable manner subjective indicators used take into account the different energy and to detect those neighborhoods where EP is increasing signifi- needs of households and are considered bottom-up information, cantly. For example, in Barcelona, this would allow assessment of 1256 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Fig. A.1. Geographical distribution, identification (ID) number and name of the neighborhoods of Barcelona city. neighborhood compliance with the climate plan target of elimi- CRediT authorship contribution statement nating EP by 2030 (Ajuntament de Barcelona, 2018a). Moreover, the methodology developed in this study can be replicated in Marc Marí-Dell’Olmo: Conceptualization, Methodology, For- representative surveys of specific population groups with greater mal analysis, Writing – original draft, Writing – review & edit- vulnerability to EP, such as children and adolescents, thus also ing, Supervision. Laura Oliveras: Conceptualization, Methodol- favoring the visibility of geographical inequalities in these groups. ogy, Formal analysis, Writing – review & editing. Carlos Vergara- Hernández: Conceptualization, Methodology, Formal analysis, This study can also be replicated in regions with surveys that Writing – review & editing. Lucia Artazcoz: Conceptualization, include questions on EP and with representative registers in the Writing – review & editing. Carme Borrell: Conceptualization, small-area of study that have some socioeconomic indicators Writing – review & editing. Mercè Gotsens: Conceptualization, shared with the surveys and, therefore, a detailed document (Ap- Writing – review & editing. Laia Palència: Conceptualization, pendix B) is provided to replicate the method used in this study. Writing – review & editing. María José López: Conceptualization, In addition, calculation of these indicators before and after the Writing – review & editing. Miguel A. Martinez-Beneito: Con- implementation of a local policy or intervention, could also allow ceptualization, Methodology, Formal analysis, Writing – review & evaluation of its effectiveness (Gouveia et al., 2019). Finally, the editing, Supervision. EP index or indicators obtained can be used for the construction of other indexes at the small area level, for example, to create Declaration of competing interest heat wave vulnerability indexes. In conclusion, this study has obtained several indicators and The authors declare that they have no known competing finan- a composite indicator (index) of EP at the small-area level in cial interests or personal relationships that could have appeared the city of Barcelona using a novel method. These SAIs have to influence the work reported in this paper. allowed analysis of the multidimensionality of EP in the city and its diverse geographical patterns, and identification of the most Data availability statement affected neighborhoods and, therefore, those most susceptible to interventions to alleviate EP. Moreover, the calculation of these Data available upon request. SAIs in future surveys will allow improved surveillance of EP and Acknowledgment monitoring and evaluating its trends over time at the small-area level. In short, these indicators are a very useful tool for policy All authors approved the version of the manuscript to be makers and local authorities. published. Funding Institutional review board statement The study is part of the PENSA Project (‘‘Pobreza energética y This study was approved by the Clinical Research Ethics Com- salud desde múltiples perspectivas: análisis individual, geográfico mittee of Hospital del Mar (2015/6155/l). y temporal’’), which received a research grant from the Carlos 1257 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 III Institute of Health, Ministry of Economy and Competitive- Carrere, J., Peralta, A., Oliveras, L., López, M.J., Marí-Dell’Olmo, M., Benach, J., ness (Spain), awarded on the 2015 call under the Health Strat- Novoa, A.M., 2020. Energy poverty, its intensity and health in vulnerable populations in a Southern European city. Gac. Sanit. http://dx.doi.org/10. egy Action 2013–2016, within the National Research Program 1016/j.gaceta.2020.07.007. aimed at Societal Challenges, within the Technical, Scientific and Castaño-Rosa, R., Sherriff, G., Thomson, H., Guzmán, J.S., Marrero, M., 2019. Innovation Research National Plan 2013–2016, with reference Transferring the index of vulnerable homes: Application at the local-scale PI15/02006, co-funded with European Union ERDF funds (Euro- in England to assess fuel poverty vulnerability. Energy Build. 203, http: pean Regional Development Fund). This study was also partially //dx.doi.org/10.1016/j.enbuild.2019.109458. Centre for Sustainable Energy, 2001. Fuel poverty indicator predicting fuel funded by ‘‘CIBER Epidemiología y Salud Pública’’ (CIBERESP) sub- poverty at the local level the. program ‘‘Energy Poverty and Health’’. L.O. was supported by Corpas-Burgos, F., Botella-Rocamora, P., Martinez-Beneito, M.A., 2019. On the a Rio Hortega grant (CM19/00184—Contratos Rio Hortega/Fondo convenience of heteroscedasticity in highly multivariate disease mapping. Social Europeo) from the Carlos III Institute of Health. The authors Test 28, 1229–1250. http://dx.doi.org/10.1007/s11749-019-00628-8. DECC, 2009. 2009 Sub-regional fuel poverty methodology and documentation. also acknowledge the support of the research Grant PI19/2009 pp. 1–32. (co-funded with ERDF grants) of Carlos III Institute of Health. Donat, C., Cónsola, A., Sánchez, A., 2015. Anàlisi i diagnosi de la situació de l’habitatge a Barcelona. Pla d’Habitatge de Barcelona. Appendix A EU Energy Poverty Observatory, 2021. Indicators & data [WWW Document]. URL https://www.energypoverty.eu/indicators-data. (Accessed 18 May 2021). Fahmy, E., Gordon, D., 2007. Updating the Fuel Poverty Indicator for England. See Fig. A.1. University of Bristol. Fahmy, E., Gordon, D., Patsios, D., 2011a. Predicting fuel poverty at a small-area Appendix B. Supplementary data level in England. Energy Policy 39, 4370–4377. http://dx.doi.org/10.1016/j. enpol.2011.04.057. Fahmy, E., Gordon, D., Patsios, D., 2011b. Predicting fuel poverty at a small-area Supplementary material related to this article can be found level in England. Energy Policy http://dx.doi.org/10.1016/j.enpol.2011.04.057. online at https://doi.org/10.1016/j.egyr.2021.12.025. Gómez-Navarro, T., Calero-Pastor, M., Pellicer-Sifres, V., Lillo-Rodrigo, P., Alfonso- Solar, D., Pérez-Navarro, A., 2021. Fuel poverty map of Valencia (Spain): Results of a direct survey to citizens and recommendations for policy making. References http://dx.doi.org/10.1016/j.enpol.2021.112162. Gouveia, J.P., Palma, P., Simoes, S.G., 2019. Energy poverty vulnerability index: Ajuntament de Barcelona, 2016a. El Pla dels Barris de Barcelona [WWW A multidimensional tool to identify hotspots for local action. Energy Rep. 5, Document]. URL https://pladebarris.barcelona/. (Accessed 6 July 2021). 187–201. http://dx.doi.org/10.1016/j.egyr.2018.12.004. Ajuntament de Barcelona, 2016b. Pla pel dret a l’habitatge 2016 - 2025: anàlisi Hills, J., 2011. Fuel poverty: The problem and its measurement - interim report i diagnosi. pp. 1–219. of the fuel poverty review. Ann. Phys. Ajuntament de Barcelona, 2017. Distribució territorial de la renda familiar HM Government, 2013. Fuel poverty report –updated August 2013. 2012 58. disponible per càpita a Barcelona (2016). Ingole, V., Marí-Dell’Olmo, M., Deluca, A., Quijal, M., Borrell, C., Rodríguez- Ajuntament de Barcelona, 2018a. Pla Clima 2018–2030. Sanz, M., Achebak, H., Lauwaet, D., Gilabert, J., Murage, P., Hajat, S., Ajuntament de Barcelona, 2018b. Energy advising points [WWW Document]. Basagaña, X., Ballester, J., 2020. Spatial variability of heat-related mortality URL https://habitatge.barcelona/en/housing-services/problems-paying-your- in Barcelona from 1992–2015: A case crossover study design. Int. J. Environ. home/energy-rights/energy-advising-points. (Accessed 18 May 2021). Res. Public Health 17, http://dx.doi.org/10.3390/ijerph17072553. Ajuntament de Barcelona, 2019. Refugis climàtics a les escoles [WWW Instituto Nacional de Estadística, 2020. Atlas de distribución de renta de Document]. URL https://www.barcelona.cat/barcelona-pel-clima/ca/escoles- los hogares [WWW Document]. URL https://www.ine.es/experimental/atlas/ refugi-climatic. (Accessed 17 May 2021). experimental_atlas.htm. (Accessed 8 July 2021). Ajuntament de Barcelona, 2021. Oficina municipal de dades [WWW Document]. Jessel, S., Sawyer, S., Hernández, D., 2019. Energy, poverty, and health in climate URL https://ajuntament.barcelona.cat/estadistica/angles/index.htm. (Accessed change: A comprehensive review of an emerging literature. Front. Public 8 July 2021). Health 7, 357. http://dx.doi.org/10.3389/FPUBH.2019.00357. Baker, W., Starling, G., Gordon, D., 2003. Predicting Fuel Poverty at the Local Leroux, B.G., Lei, X., Breslow, N., 2000. Estimation of Disease Rates in Small Level: Final Report on the Development of the Fuel Poverty Indicator. Cent. Areas: A New Mixed Model for Spatial Dependence. Springer, New York, NY, Sustain. Energy, Bristol. pp. 179–191. http://dx.doi.org/10.1007/978-1-4612-1284-3_4. Barcelona city council, 2020. Care superblocks recognised for their Lunn, D., Jackson, C., Best, N., Thomas, A., Spiegelhalter, D., 2012. The BUGS Book comprehensive assistance [WWW Document]. URL https://www.barcelona. A Practical Introduction to Bayesian Analysis. cat/infobarcelona/en/tema/social-services/care-superblocks-recognised-for- Marí-Dell’Olmo, M., Novoa, A.M., Camprubí, L., Peralta, A., Vásquez-Vera, H., their-comprehensive-assistance_1000018.html. (Accessed 23 November Bosch, J., Amat, J., Díaz, F., Palència, L., Mehdipanah, R., Rodríguez-Sanz, M., 2021). Malmusi, D., Borrell, C., 2017. Housing policies and health inequalities. Int. Bartoll, X., Barranda, L., González, J., Pérez, K., Pasarín, M., Rodríguez-Sanz, M., J. Health. Serv. 47, 207–232. http://dx.doi.org/10.1177/0020731416684292. Borrell, C., 2018. Manual metodològic de l’Enquesta de Salut de Barcelona Marí-Dell’Olmo, M., Sheehan, M.C., Villalbí, .J.R., 2019. Planning for climate 2016/17. Barcelona. change workshop: final report. Barcelona. Borrell, C., Marí-Dell’Olmo, M., Rodríguez-Sanz, M., Garcia-Olalla, P., Caylà, J.A., Marí-Dell’Olmo, M., Tobías, A., Gómez-Gutiérrez, A., Rodríguez-Sanz, M., García Benach, J., Muntaner, C., 2006. Socioeconomic position and excess mortality de Olalla, P., Camprubí, E., Gasparrini, A., Borrell, C., 2019b. Social inequalities during the heat wave of 2003 in Barcelona. Eur. J. Epidemiol. 21, 633–640. in the association between temperature and mortality in a South European http://dx.doi.org/10.1007/s10654-006-9047-4. context. Int. J. Public Health 64, 27–37. http://dx.doi.org/10.1007/s00038- Bosch, J., Palència, L., Malmusi, D., Marí-Dell’Olmo, M., Borrell, C., 2019. The 018-1094-6. impact of fuel poverty upon self-reported health status among the low- Marmot Review Team, 2011. The Health Impacts of Cold Homes and Fuel income population in Europe. Hous. Stud. 1–27. http://dx.doi.org/10.1080/ Poverty. Friends of the Earth & The Marmot Review Team, http://dx.doi.org/ 02673037.2019.1577954. 10.1136/bmj.d2807. Botella-Rocamora, P., Martinez-Beneito, M.A., Banerjee, S., 2015. A unifying Martín-Consuegra, F., Hernández-Aja, A., Oteiza, I., Alonso, C., 2019. Distribución modeling framework for highly multivariate disease mapping. Stat. Med. 34, de la pobreza energética en la ciudad de Madrid (España). Eure 45, 133–152. 1548–1559. http://dx.doi.org/10.1002/sim.6423. http://dx.doi.org/10.4067/S0250-71612019000200133. Bouzarovski, S., 2013. Energy poverty in the European Union: landscapes of Martinez-Beneito, M.A., Botella-Rocamora, P., 2021. Disease Mapping from vulnerability. Wiley Interdiscip. Rev. Energy Environ. http://dx.doi.org/10. Foundations to Multidimensional Modeling. Chapman and Hall/CRC. 1002/wene.89. März, S., 2018. Assessing the fuel poverty vulnerability of urban neighbourhoods Bouzarovski, S., Petrova, S., 2015. A global perspective on domestic energy using a spatial multi-criteria decision analysis for the German city of deprivation: Overcoming the energy poverty-fuel poverty binary. Energy Res. Oberhausen. Renew. Sustain. Energy Rev. 82, 1701–1711. http://dx.doi.org/ Soc. Sci. 10, 31–40. http://dx.doi.org/10.1016/j.erss.2015.06.007. 10.1016/j.rser.2017.07.006. Bouzarovski, S., Tirado Herrero, S., 2017. The energy divide: Integrating Mercuriali, L., Oliveras, L., Marí-Dell’Olmo, M., Gómez-Gutiérrez, A., Montalvo, T., energy transitions, regional inequalities and poverty trends in the Euro- García-Rodriguez, A., Pérez, G., Villalbí, J.R., 2021. Hacia un sistema de pean Union. Eur. Urban Reg. Stud 24, 69–86. http://dx.doi.org/10.1177/ vigilancia de salud pública para el cambio climático en ciudades. Gac. Sanit. 0969776415596449. 1258 M. Marí-Dell’Olmo, L. Oliveras, C. Vergara-Hernández et al. Energy Reports 8 (2022) 1249–1259 Morrison, C., Shortt, N., 2008. Fuel poverty in Scotland: refining spatial resolution Siksnelyte-butkiene, I., Streimikiene, D., Lekavicius, V., Balezentis, T., 2021. En- in the scottish fuel poverty indicator using a GIS-based multiple risk index. ergy poverty indicators : A systematic literature review and comprehensive Health Place 14, 702–717. http://dx.doi.org/10.1016/j.healthplace.2007.11. analysis of integrity. Sustain. Cities Soc. 67, 102756. http://dx.doi.org/10. 003. 1016/j.scs.2021.102756. Oliveras, L., Artazcoz, L., Borrell, C., Palència, L., López, M.J., Gotsens, M., Simoes, S.G., Gregório, V., Seixas, J., 2016. Mapping fuel poverty in Portugal. Peralta, A., Marí-Dell’Olmo, M., 2020. The association of energy poverty with health, health care utilisation and medication use in southern Europe. SSM Energy Procedia 106, 155–165. http://dx.doi.org/10.1016/j.egypro.2016.12. - Popul. Health 12, 100665. http://dx.doi.org/10.1016/j.ssmph.2020.100665. 112. Oliveras, L., Borrell, C., González-Pijuan, I., Gotsens, M., López, M.J., Palència, L., Sun, Y., Sundell, J., 2011. Life style and home environment are associated with Artazcoz, L., Marí-Dell’Olmo, M., 2021a. The association of energy poverty racial disparities of asthma and allergy in Northeast Texas children. Sci. Total with health and wellbeing in children in a Mediterranean city. Int. J. Environ. Environ. 409, 4229–4234. http://dx.doi.org/10.1016/j.scitotenv.2011.07.011. Res. Public Health 18, 5961. http://dx.doi.org/10.3390/ijerph18115961b. Thomson, H., Bouzarovski, S., Snell, C., 2017a. Rethinking the measurement of en- Oliveras, L., Peralta, A., Palència, L., Gotsens, M., López, M.J., Artazcoz, L., Bor- ergy poverty in Europe: A critical analysis of indicators and data. Indoor Built rell, C., Marí-Dell’Olmo, M., 2021b. Energy poverty and health: Trends in the European Union before and during the economic crisis, 2007–2016. Health Environ. http://dx.doi.org/10.1177/1420326X17699260, 1420326X1769926. Place 67, 102294. http://dx.doi.org/10.1016/j.healthplace.2020.102294. Thomson, H., Snell, C., 2013. Quantifying the prevalence of fuel poverty across Openshaw, S., 2016. Optimal zoning systems for spatial interaction models: 9. the European Union. Energy Policy 52, 563–572. http://dx.doi.org/10.1016/j. pp. 169–184. http://dx.doi.org/10.1068/A090169. enpol.2012.10.009. R Core Team, 2021. R: A Language and Environment for Statistical Computing. Thomson, H., Snell, C., Bouzarovski, S., 2017b. Health, well-being and energy Recalde, M., Peralta, A., Oliveras, L., Tirado-Herrero, S., Borrell, C., Palència, L., poverty in Europe: A comparative study of 32 European countries. Int. J. Gotsens, M., Artazcoz, L., Marí-Dell’Olmo, M., 2019. Structural energy poverty Environ. Res. Public Health 14, http://dx.doi.org/10.3390/ijerph14060584. vulnerability and excess winter mortality in the European Union: Exploring the association between structural determinants and health. Energy Policy Tirado Herrero, S., 2018. Indicadors municipals de pobresa energètica a la ciutat 133, 110869. http://dx.doi.org/10.1016/j.enpol.2019.07.005. de Barcelona. Barcelona. Robinson, C., Bouzarovski, S., Lindley, S., 2018. Getting the measure of fuel UIA - Urban Innovative Actions, 2019. GBG_AS2C - blue, green & grey. Adapting poverty: The geography of fuel poverty indicators in England. Energy Res. schools to climate change. [WWW Document]. URL https://uia-initiative.eu/ Soc. Sci. 36, 79–93. http://dx.doi.org/10.1016/j.erss.2017.09.035. en/uia-cities/barcelona-call3. (Accessed 23 November 2021). Sanchez-Guevara, C., Núñez Peiró, M., Taylor, J., Mavrogianni, A., Vergara-Hernández, C., Marí-Dell’Olmo, M., Oliveras, L., Martínez-Beneito, M.A., Neila González, J., ÚÑez Peiró, M.N., Peiró, P., Taylor, J., Mavrogianni, A., 2021. Taking Advantage of Sampling Designs in Bayesian Spatial Small Area González, J.N., González, G., Núñez Peiró, M., Taylor, J., Mavrogianni, A., Neila González, J., 2019. Assessing population vulnerability towards summer Survey Studies. https://arxiv.org/abs/2112.05468. energy poverty: Case studies of Madrid and London. Energy Build. 190, Wilson, T., Robertson, J., Hawkins, L., 2012. Poverty in Scotland fuel poverty ev- 132–143. http://dx.doi.org/10.1016/j.enbuild.2019.02.024. idence review : Defining, measuring and analysing fuel poverty in Scotland. Sánchez-Ledesma, E., Vásquez-Vera, H., Sagarra, N., Peralta, A., Porthé, V., Díez, È., p. 165. 2020. Perceived pathways between tourism gentrification and health: A Yamamoto, Y., Ishikawa, H., 2020. Influence of urban spatial configuration and participatory photovoice study in the Gòtic neighborhood in Barcelona. Soc. sea breeze on land surface temperature on summer clear-sky days. Urban Sci. Med. 258, 113095. http://dx.doi.org/10.1016/j.socscimed.2020.113095. Clim. 31, 100578. http://dx.doi.org/10.1016/j.uclim.2019.100578. 1259