Gac Sanit. 2020;34(3):253–260 Original Effect of the recent economic crisis on socioeconomic inequalities in mortality in nine urban areas in Europe Laia Palènciaa,b,c,∗ , Mercè Gotsensa,b,c , Marc Marí-Dell’Olmoa,b,c , Lucia Bosakovad,e , Bo Burströmf , Cláudia Costag , Patrick Deboosereh , Dagmar Dzurovai , Michala Lustigovai , Joana Morrisonj , Paula Santanag,k , Carme Borrella,b,c a Agència de Salut Pública de Barcelona, Barcelona, Spain b CIBER de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain c Institut d’Investigació Biomèdica Sant Pau, Barcelona, Spain d Department of Health Psychology, Medical Faculty, P.J. Safarik University in Kosice, Kosice, Slovak Republic e Olomouc University Social Health Institute, Palacky University in Olomouc, Olomouc, Czech Republic f Department of Public Health Sciences, Equity and Health Policy Research Group, Karolinska Institutet, Stockholm, Sweden g Centre of Studies in Geography and Spatial Planning, University of Coimbra, Coimbra, Portugal h Interface Demography, Section Social Research, Vrije Universiteit Brussels, Brussels, Belgium i Department of Social Geography and Regional Development, Faculty of Science, Charles University, Prague, Czech Republic j Research Department of Epidemiology and Public Health, University College London, London, United Kingdom k Department of Geography and Tourism, Humanities Faculty, University of Coimbra, Coimbra, Portugal a r t i c l e i n f o a b s t r a c t Article history: Objective: To analyse socioeconomic inequalities in all-cause mortality among men and women in nine Received 22 July 2019 European urban areas during the recent economic crisis, and to compare the results to those from two Accepted 5 November 2019 periods before the crisis. Available online 23 January 2020 Method: This is an ecological study of trends based on three time periods (2000-2003, 2004-2008 and 2009-2014). The units of analysis were the small areas of nine European urban areas. We used a com- Keywords: posite deprivation index as a socioeconomic indicator, along with other single indicators. As a mortality Socioeconomic inequalities indicator, we used the smoothed standardized mortality ratio, calculated using the hierarchical Bayesian All-cause mortality Small areas model proposed by Besag, York and Mollié. To analyse the evolution of socioeconomic inequalities, we Europe fitted an ecological regression model that included the socioeconomic indicator, the period of time, and Economic crisis the interaction between these terms. Results: We observed significant inequalities in mortality among men for almost all the socioeconomic indicators, periods, and urban areas studied. However, no significant changes occurred during the period of the economic crisis. While inequalities among women were less common, there was a statistically significant increase in inequality during the crisis period in terms of unemployment and the deprivation index in Prague and Stockholm, respectively. Conclusions: Future analyses should also consider time-lag in the effect of crises on mortality and specific causes of death, and differential effects between genders. © 2019 SESPAS. Published by Elsevier España, S.L.U. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Efecto de la reciente crisis económica en las desigualdades socioeconómicas en la mortalidad en nueve áreas urbanas europeas r e s u m e n Palabras clave: Objetivo: Analizar las desigualdades socioeconómicas en la mortalidad por todas las causas en hombres y Desigualdades socioeconómicas mujeres de nueve áreas urbanas europeas durante la reciente crisis económica, y comparar los resultados Mortalidad por todas las causas con dos periodos previos a la crisis. Áreas pequeñas Método: Estudio ecológico de tendencias basado en tres periodos (2000-2003, 2004-2008 y 2009-2014). Europa Las unidades de análisis fueron las áreas pequeñas de nueve zonas urbanas europeas. Se utilizaron un Crisis económica índice compuesto de privación socioeconómica como indicador socioeconómico y otros indicadores sim- ples. Como indicador de mortalidad se usó la razón de mortalidad estandarizada suavizada, calculada utilizando el modelo jerárquico bayesiano propuesto por Besag, York y Mollié. Para analizar la evolución de las desigualdades socioeconómicas se utilizó un modelo de regresión ecológico que incluía el indicador socioeconómico, el periodo y la interacción de ambos. ∗ Corresponding author. E-mail address: lpalenci@aspb.cat (L. Palència). https://doi.org/10.1016/j.gaceta.2019.11.001 0213-9111/© 2019 SESPAS. Published by Elsevier España, S.L.U. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc- nd/4.0/). 254 L. Palència et al. / Gac Sanit. 2020;34(3):253–260 Resultados: Se observaron desigualdades significativas en la mortalidad en los hombres para casi todos los indicadores socioeconómicos, periodos y áreas urbanas. Sin embargo, no hubo cambios significativos en las desigualdades en el periodo de crisis. Aunque las desigualdades entre las mujeres fueron menos comunes, hubo un incremento significativo en las desigualdades en mortalidad en el periodo de crisis en términos de desempleo y del índice de privación en Praga y Estocolmo, respectivamente. Conclusiones: Futuros análisis deberán tener en cuenta el tiempo transcurrido entre la crisis y su efecto en la mortalidad, así como diferentes causas de mortalidad y el efecto diferencial entre géneros. © 2019 SESPAS. Publicado por Elsevier España, S.L.U. Este es un artı́culo Open Access bajo la licencia CC BY-NC-ND (http://creativecommons.org/licenses/by-nc-nd/4.0/). Introduction knowledge and resources related to policies promoting health and health equity across European regions, with a focus on metropoli- Multiple studies have evaluated the relationship between eco- tan areas. We performed an ecological study of trends based on nomic recessions and changes in the health of the population. Some three periods, two before the economic crisis (2000-2003 and studies have shown improvements in health and health-related 2004-2008), and one during the crisis (2009-2014). The units of behaviours (procyclical effect), while others have shown worsening analysis were small areas in nine European urban areas: Athens health (countercyclical effect) during periods of economic crises.1 metropolitan area, Barcelona city, Berlin-Brandenburg Metropol- Concerning the recent financial crisis in Europe that started at the region, Brussels-Capital Region, Lisbon Metropolitan Area, Greater end of 2008, some studies have reported a decrease in the con- London, Prague city, Turin city, and Stockholm metropolitan area. sumption of tobacco and alcohol, reduced sedentary behaviour, These areas were selected because of their availability within the and reduced incidence of traffic injuries. Conversely, there has EURO-HEALTHY project, which covers a broad range of geographic been an increase in suicides, homicides, drug consumption, AIDS, areas and socioeconomic conditions20 . The study population con- and unmet healthcare needs overall.2–4 Consistent with this, it sisted of individuals residing in these areas during the three time has been suggested that the number of suicides increased and periods. that mental health deteriorated during the crisis, while there has been mixed evidence for self-rated health and other indicators.5 Information sources The effect of a recession on population health depends on various factors such as the extent, nature, and duration of the economic For most of the urban areas, mortality and population data were recession, co-existing economic and social protection policies, pre- available for the three time periods, and socioeconomic indica- dominant socio-cultural values, and the general well-being of the tors for 2001 (Tables 1 to 3). Mortality data were mainly obtained population.6 These factors affect intermediate determinants of from mortality registers. Population data stratified by age (5-year health inequalities such as unemployment, job insecurity, and lack groups) and sex were obtained from census or population registers. of income and housing, all of which generally deteriorate during Socioeconomic data were mainly obtained from the census records. economic crises.7 There have been contradictory results regarding the impact of Variables and indicators economic recessions on all-cause mortality. Some studies have observed an increase and decrease in mortality during economic In this study we analysed all-cause mortality in each small area expansions and recessions, respectively,8,9 while others have of residence during the three time periods. The mortality indicator shown the opposite.10,11 Various possibilities have been proposed used for this analysis was the Standardized Mortality Ratio (SMR). to explain these contradictory results: the use of individual versus For descriptive purposes we used the Indirectly Standardised Rate aggregated data; the wealth of the countries analysed (high income (ISR), which we calculated by multiplying the SMR for a study pop- versus low-middle income); analysis of levels of mortality versus ulation by the crude rate in the standard population (defined as the inequalities in mortality; the time lag between the crisis and the European Union [EU]-28 in the year 2007). effects studied; and the type of welfare state policies implemented We used a composite deprivation index in 2001 as a socioe- in the countries studied.12 conomic indicator for each small area. This index was the first Economic crises have a worse impact on people from disadvan- component of a principal components analysis performed within taged social classes, ethnic minorities, the unemployed, and those each urban area.21 It includes unemployment in people aged ≥16 with lower educational levels.13 Therefore, the negative effects on years; the percentage of manual workers in people aged ≥16 years; health are also likely to be greater among these groups. However, the percentage of people aged 25-64 years with primary education there has been a general lack of studies that analyse the impact of as their highest education (ISCED 0 and 1, except London where it economic recessions on inequalities in mortality,14,15 especially in was ISCED 0, 1 and 2); and the percentage of people aged 25-64 urban areas. Further, studies that have analysed the effects of the years with a university education (ISCED 5 and 6).22 There were economic crisis on socioeconomic inequalities in mortality have no data for the percentage of manual workers in Stockholm, so the yielded mixed results.16–19 index was constructed using the other three available indicators. The aim of this study is to analyse socioeconomic inequalities in all-cause mortality among men and women living in nine European Data analysis urban areas during the recent economic crisis, and to compare the results to those from two periods before the crisis. The SMR depends on population size because its variance is inversely proportional to the expected values. Therefore, areas with Method low populations tend to present very variable estimates. To smooth the SMR, we used the hierarchical Bayesian model proposed by Design, units of analyses and study population Besag, York and Mollié,23 which takes two types of random effects into account, spatial and heterogeneous. The former considers This study is part of a wider European project called EURO- the spatial structure of the data, while the latter deals with non- HEALTHY (http://www.euro-healthy.eu/), which aims to increase structural (non-spatial) variability. We used the following model L. Palència et al. / Gac Sanit. 2020;34(3):253–260 255 to estimate smoothed SMR (sSMR) for each sex, cause of death, and Results period: The cities and metropolitan areas included in the study are Oi ∼ Poisson (Ei i ) described in Table 1. The least populated area is Turin city with nearly 900.000 people and the most populated is Greater Lon- don with more than 7 million people. Berlin-Brandenburg has the log(i ) = ˛ + Si + Hi (model 1) smallest number of areas (30 areas) and Turin has the highest where, for each area i, Oi is the number of observed cases, Ei is the (2678 census tracts). The population sizes of the small areas varied expected number of cases, ␪i is the expected sSMR with respect to inversely with their number in a given region. the European population, Si is the spatial effect, and Hi is the het- The periods of analyses, deaths and crude and indirectly stan- erogeneous effect. The expected number of cases were calculated dardised mortality ratios (ISMR) are described in Table 2. Indirectly by indirect standardization, taking the mortality rates for 5-year standardised mortality rates tend to diminish over time in all urban age groups of the EU-28 in 2007 (approximately the year in the areas (second with respect to the first period, and third with respect middle of the study period) as reference. to the second period). The geographical distribution of the sSMR, for each time period, Table 3 shows the socioeconomic indicators and their cor- was represented using septile maps. The socioeconomic indicators relation with deprivation index. Prague is the area with the and the composite deprivation indicator were also represented as lowest median percentages of unemployment, Brussels-Capital septile maps. Region has the lowest median percentages of manual workers, To analyse the evolution of socioeconomic inequalities, we fit- Berlin-Brandenburg has the lowest median percentages of people ted an ecological regression model that takes into account the with primary education at most, and Greater London the highest composite deprivation index (D), the period (through two dummy median percentages of university education. Conversely, Berlin- variables P2 and P3) and their interaction: Brandenburg is the area with the highest median percentage of unemployment, Turin has the highest percentage of manual work- Oit ∼ Poisson (Eit it ) ers, Greater London has the highest percentages of people with primary education at most (although in this case the variable was constructed in a different way to the other cities), followed by Lis- log(it ) = ˛ + ˇ1 Di + ˇ2 P2t + ˇ3 P3t + ˇ4 P2t Di + ˇ5 P3t Di + Sit bon, which also has the lowest percentages of university education. + H it (model 2) The deprivation index shows a high negative correlation with the percentage of university education (all above 0.5), and a generally positive correlation with the other variables. In particular, a mod- where, for each area i and period t (t = 1 for the first period, t = 2 erate positive correlation is shown with the percentage of manual for the second period and t = 3 for the third period), Oit is the num- workers and individuals with primary education only, and a lower ber of observed cases, Eit is the expected number of cases, ␪it is correlation with the percentage of unemployment, in some cases the sSMR with respect to the European population, Sit is the spa- around 0. tial effect, and Hit is the heterogeneous effect. Finally, P2t and P3t Figure 1 shows the distribution of the deprivation index and the take the following values: Pjt = 1 if j = t, and 0 otherwise, where j = 2 sSMR for the three periods among men living in London and women or 3. The expected number of cases was calculated as in the previ- living in Barcelona (chosen for descriptive purposes). Men living ous model. Changes between periods in the relationship between in areas in London with lower deprivation (green, mainly in the the socioeconomic deprivation index and mortality were evaluated inner part of London) also have lower mortality (green). Conversely, by the interaction terms in model 2. areas with higher deprivation (brown) tend to show higher mortal- In the two models, the spatial effect was assigned an intrinsic ity. However, we do not observe similar patterns in socioeconomic conditional autoregressive prior distribution, which assumes that deprivation and mortality among women in Barcelona. the expected value of each area coincides with the mean of the Figure 2 illustrates, for each of the three periods, the relative spatial effect of the adjacent areas. Its variance ␴s 2 depends on risks of mortality for the 5% of small areas with the highest depriva- the number of adjacent areas, since those areas that are more con- tion compared to the 5% of small areas with the lowest deprivation, nected will show lower variability.24 The heterogeneous effect was as well as the levels of statistical significance of their difference. Also represented using independent normal distributions with mean shown is the relationship of unemployment and primary education 0 and variance ␴h 2 . A uniform distribution U(0,∞) was assigned with mortality. Regarding the deprivation index, we see significant to the standard deviations ␴s and ␴h. A normal vague prior dis- inequalities (credible interval not containing 1) among men in all tribution, with mean 0 and precision 0.001, was assigned to the periods and areas, except for Prague where there were no signif- parameters ␣, ␤1, ␤2, ␤3, ␤4 and ␤5. icant inequalities in any of the periods. The greatest inequalities Since the composite deprivation index scale is dimensionless were found in Stockholm, where the areas with highest depriva- and arbitrarily fixed, we calculated the increase in risk that would tion have around 80% higher mortality than those with the lowest correspond to a change in the composite deprivation index from its deprivation. The results for primary education are similar with 5th percentile value (low deprivation) to its 95th percentile value minor differences. For example, there were significant inequalities (severe deprivation). Relative risk estimates were obtained based in Prague during the first period, but not in Berlin-Brandenburg. In on the mean of their posterior distribution, along with correspond- addition, there was a statistically significant increase in inequal- ing 95% credible intervals. ities between the first and second periods in Turin. Results were Analyses were also performed for the single socioeconomic also similar for unemployment. indicators: unemployment, percentage of manual workers, the per- The number of significant inequalities was much lower among centage of those with primary education only, and the percentage women (Fig. 2). According to the deprivation index, there were of those with university education (results are shown for unem- inequalities during all periods in Brussels, Lisbon, London, and ployment and for primary education while the others can be found Turin. Inequalities became significant during the third period in in online Appendix). Athens and Berlin (although this change was not statistically sig- All analyses were performed using R 3.5.0 25 and the R-INLA nificant). Inequalities were not significant in any of the periods package 26 . in Prague or Barcelona, although in Barcelona they were almost 256 Table 1 Description of the nine European urban areas: number and type of small areas, total population and first, second and third quartiles of the population by small area in the first year available, for men and women. Urban area Short name Number of Type of small areas Population small areas (first year available) Men Women Total p25 p50 p75 Total p25 p50 p75 Athens metropolitan Athens 40 Municipalities 1.577.172 18.565 29.745 35.489 1.710.446 20.136 32.163 39.965 area Barcelona city Barcelona 1491 Census tracts 697.563 365 457 577 796.497 418 517 648 Berlin-Brandenburg Berlin 30 Parishes for Berlin and 2.927.616 66.326 96.176 129.157 3.047.188 68.041 97.454 130.560 Metropolregion municipalities for Brandenburg Brussels-capital region Brussels 145 Neighbourhoods 464.364 2.727 4.004 5.707 505.673 3.058 4.288 6.172 Lisbon metropolitan Lisbon 188 Parishes 1.275.813 2.694 5.437 8.962 1.386.314 2.938 5.835 9.904 area Greater London London 983 Census tracts 3.597.120 3.442 3.810 4.284 3.725.283 3.526 3.960 4.382 Prague city Prague 57 Districts 549.652 1.010 2.206 15.001 610.466 1.024 2.100 14.838 L. Palència et al. / Gac Sanit. 2020;34(3):253–260 Stockholm Stockholm 1299 Census tracts 897.487 218 560 1.050 936.977 232 599 1.104 metropolitan area Turin city Turin 2678 Census tracts 425.782 88 129 196 465.987 96 142 215 Table 2 Description of mortality in each urban area: years, number of deaths, crude mortality rate and indirectly standardized mortality rate by 100,000 inhabitants, for men and women in each study period. Men First period Second period Third period Urban area Years Deaths Crude MR ISMR Years Deaths Crude MR ISMR Years Deaths Crude MR ISMR Athens 2000-2003 57289 913.94 1329.29 2004-2008 72605 957.29 1266.20 2009-2013 75057 1015.67 1210.48 Barcelona 2000-2003 29273 1016.28 1172.40 2004-2008 37353 983.81 1118.13 2009-2013 35957 934.24 980.28 Berlin 2002 26148 893.15 1459.56 2006 26477 875.65 1268.69 2011 28294 965.51 1159.72 Brussels 2001-2003 10491 - - 2004-2008 21915 - - 2009-2011 12932 - - Lisbon 2000-2003 53581 1049.29 1584.01 2004-2008 65655 1016.61 1397.99 2009-2012 52774 994.68 1242.07 London 2000-2003 112182 775.33 1361.42 2004-2008 125657 670.99 1180.11 2009-2014 140361 574.13 993.37 Prague 2001-2003 18714 1129.37 1579.29 2004-2008 29515 1024.37 1369.61 2009-2014 35262 968.24 1230.16 Stockholm 2001-2003 22074 814.13 1185.68 2004-2008 36184 768.33 1091.54 2009-2011 21178 730.04 1041.12 Turin 2000-2003 18366 1075.12 1225.83 2004-2008 22530 1054.51 1102.93 2009-2013 22871 1059.69 983.65 Women First period Second period Third period Urban area Years Deaths Crude MR ISMR Years Deaths Crude MR ISMR Years Deaths Crude MR ISMR Athens 2000-2003 56116 825.04 982.35 2004-2008 72172 873.49 918.19 2009-2013 75402 930.85 853.27 Barcelona 2000-2003 30374 933.19 721.60 2004-2008 39335 933.80 710.23 2009-2013 39347 924.69 667.14 Berlin 2002 32571 1068.89 1067.22 2006 30731 986.41 935.49 2011 30257 990.74 847.86 Brussels 2001-2003 12793 - - 2004-2008 26302 - - 2009-2011 15292 - - Lisbon 2000-2003 50900 915.51 1002.16 2004-2008 62484 878.87 886.45 2009-2012 51575 873.87 812.25 London 2000-2003 120110 803.55 1018.40 2004-2008 131077 678.16 903.58 2009-2014 143304 572.66 787.47 Prague 2001-2003 21317 1164.47 1060.48 2004-2008 32758 1055.24 929.31 2009-2014 38139 989.32 855.36 Stockholm 2001-2003 24154 854.90 880.23 2004-2008 39558 812.04 839.40 2009-2011 23654 795.13 854.07 Turin 2000-2003 19803 1059.50 826.11 2004-2008 24085 1027.39 765.82 2009-2013 25257 1063.57 716.15 ISMR: indirectly standardized mortality rate; MR: mortality rate. Note: Crude MR and ISMR are not displayed for Brussels as population data were interpolated. L. Palència et al. / Gac Sanit. 2020;34(3):253–260 257 statistically significant during the last period. There were inequal- ities in Stockholm during the first period, but these disappeared in the second and appeared again in the last period; this change was University education statistically significant. In this case, women in areas with highest -0.61 -0.64 -0.95 -0.56 -0.57 -0.86 -0.97 -0.50 -0.70 Correlation indicator with deprivation index deprivation have 31% higher mortality than those from areas with the lowest deprivation. Overall, the results were similar for pri- mary education and for unemployment. In Prague, unemployment inequalities appeared in the third period, where the change from education the second period is statistically significant. In the third period, women from areas with the highest unemployment have 16% Primary higher mortality than those from areas with the lowest unemploy- 0.48 0.48 0.49 0.62 0.56 0.36 0.17 0.43 0.09 ment. Description of the socioeconomic indicators: year, first, second and third quartiles of the indicators by small area. Correlation between the indicators and the composite deprivation index. Discussion Manual 0.48 0.49 0.21 0.49 0.56 0.35 0.45 0.60 - In this study, we found statistically significant inequalities in mortality among men for almost all of the socioeconomic indica- Unemployment tors, periods, and urban areas studied. However, for all urban areas, no statistically significant changes occurred in the period of eco- nomic crisis (third period) when compared to the previous period. In women, statistically significant inequalities in mortality were -0.03 -0.04 0.42 0.34 0.45 0.16 0.34 0.20 0.20 less common. However, there was a statistically significant increase in deprivation and unemployment inequalities during the crisis period in Stockholm and Prague, respectively. For some urban areas 32.4 38.2 32.9 45.6 25.6 24.7 34.0 17.0 34.0 p75 University education and indicators, there were no significant inequalities in the two first periods, although there were statistically significant inequalities in the third period. 22.8 26.1 17.8 32.1 22.1 24.6 11.5 31.0 10.8 p50 Changes in mortality due to the economic crisis 16.9 16.9 23.8 23.2 17.4 17.8 10.8 p25 6.3 6.3 There has been much debate about whether mortality rates increase or decrease during economic crises.12,27-33 There has been little consensus, except for the general observation that the number 27.7 23.3 45.7 62.8 21.6 30.5 10.0 p75 2.0 7.0 Primary education of suicides increases and that of traffic injuries decreases, at least during recent economic crises.5,34-36 We did not find any previous studies that focused on urban 22.4 19.2 14.9 37.6 15.7 54.0 p50 Note: Primary education stands for ISCED 0 and 1 except for London where it stands for ISCED 0, 1 and 2. 7.8 0.9 4.0 areas, so it is difficult to compare our results with others. Never- theless, we observe that mortality during the crisis seems to follow 14.4 12.1 41.2 30.3 the downward trend observed before the crisis. However, it is not p25 8.7 6.7 2.2 9.2 0.6 possible to determine whether the rate was different to that in the periods prior to the crisis. 52.3 43.1 26.6 57.7 51.0 50.5 20.2 48.0 p75 - Changes in socioeconomic inequalities in mortality Manual 44.2 46.3 12.6 35.4 34.8 23.9 46.5 40.2 Even if mortality does not increase for the entire population p50 - during periods of economic crisis, it may increase in certain socioe- conomic groups, e.g. in socioeconomically deprived groups, who 31.5 39.5 24.4 24.9 21.6 31.9 29.0 are more negatively affected by economic crises14 . Some studies p25 8.0 - have shown an increase in inequalities following previous crises, such as during the Southeast Asian, Japanese, and Soviet Union 12.6 19.7 13.8 20.9 p75 crises, whereas these patterns were not so clear for the Nordic 9.9 8.5 8.9 5.3 8.9 countries.14 Following the recent crisis, two studies reported an Unemployment increase in socioeconomic inequalities in total mortality in two 16.1 14.9 10.5 10.2 p50 areas of Spain (Barcelona city and Andalucía autonomous commu- 7.2 6.5 5.9 9.0 5.0 nity), one of which found increasing inequalities in men only.16,17 Two other studies found a decrease in inequalities in premature 14.4 11.8 p25 mortality in Spain18 and in the Lisbon Metropolitan area.19 The last 7.8 8.8 6.1 4.3 4.2 3.8 7.0 authors suggest that the decrease in inequalities is due to stability or deterioration of health in the middle and upper-middle social 2001 2001 2002 2001 2001 2001 2001 2001 2001 Year groups, and improvements in the most disadvantaged groups. They argue that this is probably an unexpected effect of austerity poli- cies, which have promoted the emergence of newly-deprived and Urban area vulnerable groups.19 Stockholm Barcelona Brussels London Athens In our study we did not find statistically significant changes Prague Lisbon Berlin Table 3 Turin between periods in socioeconomic inequalities in men. Two sta- tistically significant changes occurred among women in the crisis 258 L. Palència et al. / Gac Sanit. 2020;34(3):253–260 Figure 1. Deprivation index and smoothed standardized mortality ratio (sSMR), in septiles, in each of the periods in men in London and women in Barcelona. Figure 2. Association between socioeconomic indicators and mortality, relative risk (RR) and 95% credible intervals (CI) for men and women in nine urban areas (Note 1: RRi corresponds to the increase in risk that would correspond to a change in the composite deprivation index from its 5th percentile value (low deprivation) to its 95th percentile value (severe deprivation) in the period i; Note 2: * indicates that RR2 is statistically significantly different than RR1 and + means that RR3 is statistically significantly different than RR2 ). L. Palència et al. / Gac Sanit. 2020;34(3):253–260 259 period: deprivation inequalities increased in Stockholm, and unem- ployment inequalities increased in Prague. These countries were not affected especially hard by the economic crisis compared to other countries, and indeed to other crises.37,38 In addition, they What is known about the topic? have quite different state welfare policies.39 It is well known that Socioeconomic inequalities in health tend to be greater in Sweden has one of the most generous welfare policies, so the strong urban areas. inequalities found are quite counterintuitive. There have been contradictory results regarding the impact of economic recessions on all-cause mortality. Differences in inequality due to gender What does this study add to the literature? In our study we found more socioeconomic inequalities among Significant inequalities in mortality among men were found men than among women, which has already been reported in pre- for almost all the socioeconomic indicators, periods and urban vious studies.38 Surprisingly, however, we found some increases areas studied. However, no significant changes occurred in the in inequality among women only. Allowing for spurious relation- period of economic crisis. In women, inequalities were less frequent. However, in two urban areas there were statistically ships and factors other than the crisis that could provoke such an significant increases in inequalities in the crisis period. increase in inequalities, other studies have indeed found similar results. The economic crisis of 1994 in Sweden was associated with greater inequalities in suicide among women, which was partially related to greater job insecurity and poorer working conditions in Editor in charge the female-dominated public sector after the recession.40 In the late 1990s economic crisis in South Korea, following neoliberal restruc- María-Victoria Zunzunegui. turing imposed by the IMF, females of working age were more affected than males in terms of life expectancy and mortality.41 Transparency declaration These authors argued that, since the economic crisis, women from disadvantaged classes had to join the labour force to compen- The corresponding author on behalf of the other authors guar- sate for their husbands’ unemployment and wage cuts, and thus antee the accuracy, transparency and honesty of the data and suffered both household economic collapse and unstable working information contained in the study, that no relevant information conditions.41 has been omitted and that all discrepancies between authors have been adequately resolved and described. Limitations and strengths Authorship contributions We only had socioeconomic data for 1 year and thus considered L. Palència, M. Gotsens, M. Marí-dell’Olmo and C. Borrell them constant during all of the periods. Although socioeconomic designed the paper, and analysed and interpreted the data. L. indicators may have changed, the ranking of the areas in terms of Bosakova, B. Burström, C. Costa, P. Deboosere, D. Dzurova, M. deprivation did not change much and we were not using the value Lustigova, J. Morrison and P. Santana provided and interpreted itself but only the distribution. the data. L. Palència and C. Borrell wrote the first version of The sizes of the small areas differ between the urban areas stud- the manuscript. M. Gotsens, M. Marí-dell’Olmo, L. Bosakova, B. ied, in fact some of the areas may not be considered as small areas Burström, C. Costa, P. Deboosere, D. Dzurova, M. Lustigova, J. Mor- given their size, which may partly explain why the results are not rison and P. Santana revised it critically. All authors read and consistent. Smaller areas are more homogeneous and the possi- approved the final manuscript. bility of observing higher effects is higher, which may be the case in Stockholm, London and Turin. So the comparison of RRs among Acknowledgements urban areas is not very suitable in this case. Finally, we did not consider any lag between the cause (eco- We thank Sani Dimitroulopoulou and Christina Mitsakou for nomic recession) and its effects (mortality). It has been reported providing the data from London, Klea Katsouyanni for the data that the impact of unemployment and other adverse circumstances from Athens, Jürgen Schweikart and Conrad Franke for the on health in times of crisis only becomes evident after many years, data from Berlin, and Giuseppe Costa and Nicolás Zangarini for especially in disadvantaged populations.14 the data from Turin. However, this study has important strengths. It fills the gap in research related to the effect of the crisis on socioeconomic inequal- Funding ities in mortality in urban areas. Also, it reports mortality and socioeconomic data collected in a comparable way for a number This study is a part of the EURO-HEALTHY project (Shaping of urban areas from different parts of Europe. EUROpean policies to promote HEALTH equity) and has received funding from the European Union’s Horizon 2020 research and innovation programme (Grant Agreement No 643398). Dagmar Conclusions Dzúrová and Michala Lustigova were also supported by Charles University (UNCE/HUM 018). We observe persistent socioeconomic inequalities in mortality Cláudia Costa was supported by an Individual Doctoral Fellow- among men, but these did not change during the periods studied. ship funded by the Foundation for Science and Technology, FCT While inequalities among women were less common, there was (SFRH/BD/132218/2017). a statistically significant increase in inequality for certain indica- tors and areas studied. Future analyses should consider a time lag Conflicts of interests between the crisis and mortality and specific causes of death, and should pay special attention to differential effects between genders. None. 260 L. Palència et al. / Gac Sanit. 2020;34(3):253–260 Appendix A. Supplementary data 20. Costa C, Santana P, Dimitroulopoulou S, et al. Population Health Inequali- ties Across and Within European Metropolitan Areas through the Lens of the EURO-HEALTHY Population Health Index. Int J Environ Res Public Health. Supplementary data associated with this article can be found, in 2019;16. the online version, at doi:10.1016/j.gaceta.2019.11.001. 21. Domínguez-Berjón MF, Borrell C, Cano-Serral G, et al. Constructing a deprivation index based on census data in large Spanish cities (the MEDEA project). Gac Sanit. 2008;22:179–87. References 22. International Standard Classification of Education (ISCED) 1997. United Nations Educational, Scientific and Cultural Organization (UNESCO). 1. Catalano R, Goldman-Mellor S, Saxton K, et al. The health effects of economic 23. Besag J, York J, Mollié A. Bayesian image restoration, with two applications in decline. Annu Rev Public Health. 2011;32:431–50. spatial statistics. Ann Inst Statist Math. 1991;43:1–59. 2. Kondilis E, Giannakopoulos S, Gavana M, et al. Economic crisis, restrictive poli- 24. Martínez-Beneito MA, Botella-Rocamora P, Botella-Rocamora P. Disease map- cies, and the population’s health and health care: the Greek case. Am J Public ping. Boca Raton: Taylor & Francis; 2019. Health. 2013;103:973–9. 25. R core Team. R: a language and environment for statistical computing. Vienna, 3. Filippidis FT, Gerovasili V, Millett C, et al. Medium-term impact of the economic Austria; 2018. Available at: http://www.r-project.org/. crisis on mortality, health-related behaviours and access to healthcare in Greece. 26. Rue H, Martino S, INLA:. functions which allow to perform a full Bayesian analysis Sci Rep. 2017;7:46423. of structured additive models using integrated nested laplace approximaxion. 4. Simou E, Koutsogeorgou E. Effects of the economic crisis on health and health- 2009. care in Greece in the literature from 2009 to 2013: a systematic review. Health 27. Falagas ME, Vouloumanou EK, Mavros MN, et al. Economic crises and mortality: Policy (New York). 2014;115:111–9. a review of the literature. Int J Clin Pract. 2009;63:1128–35. 5. Parmar D, Stavropoulou C, Ioannidis JPA. Health outcomes during the 2008 28. Stuckler D, Meissner C, Fishback P, et al. Banking crises and mortality during the financial crisis in Europe: systematic literature review. BMJ. 2016;354:i4588. Great Depression: evidence from US urban populations, 1929-1937. J Epidemiol 6. Elliott E, Harrop E, Rothwell H, et al. The impact of the economic downturn on Community Health. 2012;66:410–9. Health in Wales: a review and case study [working paper]. Vol 134. Cardiff; 2010. 29. Edwards R. Who is hurt by procyclical mortality? Soc Sci Med. 2008;67: 7. Marmot MG, Bell R. How will the financial crisis affect health? BMJ. 2051–8. 2009;338:b1314. 30. Regidor E, Barrio G, Bravo MJ, et al. Has health in Spain been declining since the 8. Gerdtham U-G, Ruhm CJ. Deaths rise in good economic times: evidence from economic crisis? J Epidemiol Community Health. 2014;68:280–2. the OECD. Econ Hum Biol. 2006;4:298–316. 31. Ruhm CJ. Recessions, healthy no more? J Health Econ. 2015;42:17–28. 9. Granados JAT. Recessions and mortality in Spain, 1980-1997. Eur J Popul/Rev 32. Strumpf EC, Charters TJ, Harper S, et al. Did the Great Recession affect mortality Eur Démographie. 2005;21:393–422. rates in the metropolitan United States? Effects on mortality by age, gender and 10. Svensson M. Do not go breaking your heart: do economic upturns really increase cause of death. Soc Sci Med. 2017;189:11–6. heart attack mortality? Soc Sci Med. 2007;65:833–41. 33. Toffolutti V, Suhrcke M. Assessing the short term health impact of the Great 11. Gerdtham U-G, Johannesson M. Business cycles and mortality: results from Recession in the European Union: a cross-country panel analysis. Prev Med Swedish microdata. Soc Sci Med. 2005;60:205–18. (Balt). 2014;64:54–62. 12. Suhrcke M, Stuckler D. Will the recession be bad for our health? It depends. Soc 34. Antonakakis N, Collins A. The impact of fiscal austerity on suicide mortality: Sci Med. 2012;74:647–53. evidence across the ‘Eurozone periphery.’. Soc Sci Med. 2015;145:63–78. 13. Winters L, McAteer S, Scott-Samuel A. Assessing the impact of the economic 35. Baumbach A, Gulis G. Impact of financial crisis on selected health outcomes in downturn on health and wellbeing. Liverpool Public Health Observatory. Obser- Europe. Eur J Public Health. 2014;24:399–403. vatory Report Series No. 88.;. 2012. 36. Santana P, Costa C, Cardoso G, et al. Suicide in Portugal: spatial determinants in 14. Bacigalupe A, Escolar-Pujolar A. The impact of economic crises on social inequal- a context of economic crisis. Health Place. 2015;35:85–94. ities in health: what do we know so far? Int J Equity Health. 2014;13:52. 37. Hassler J. Sweden in past, current and future economic crises. A report 15. Borrell C, Palència L, Marí Dell’Olmo M, et al. Socioeconomic inequalities in for the OECD Economics Department. 2010. Available at: http://hassler- suicide mortality in European urban areas before and during the economic j.iies.su.se/PAPERS/OECDSwedenAug2010.pdf. recession. Eur J Public Health. 2019 Aug 13, pii: ckz125. 38. Landesmann M. Effects of the Euro crisis on Europe’s periphery (East, 16. Maynou L, Sáez M, López-Casasnovas G. Has the economic crisis widened the South and Southeast). Vienna Inst Int Econ Stud. 2012. Available at: intraurban socioeconomic inequalities in mortality? The case of Barcelona, https://www.fiw.ac.at/fileadmin/Documents/Veranstaltungen/Vorlesungen/ Spain. J Epidemiol Community Health. 2016;70:114–24. Literatur-EffectsoftheEuroCrisisonEuropesPeriphery.pdf. 17. Ruiz-Ramos M, Córdoba-Doña JA, Bacigalupe A, et al. Crisis económica al ini- 39. Ferrarini T, Sjoberg O. Social policy and health: transition countries in a com- cio del siglo XXI y mortalidad en España. Tendencia e impacto sobre las parative perspective. Int J Soc Welf. 2010;19:S60–88. desigualdades sociales. Informe SESPAS 2014. Gac Sanit. 2014;28 (Supl 1): 40. Hiyoshi A, Kondo N, Rostila M. Increasing income-based inequality in sui- 89–96. cide mortality among working-age women and men, Sweden, 1990-2007: 18. Regidor E, Vallejo F, Granados JAT, et al. Mortality decrease according to socioe- is there a point of trend change? J Epidemiol Community Health. 2018;72: conomic groups during the economic crisis in Spain: a cohort study of 36 million 1009–15. people. Lancet. 2016;388:2642–52. 41. Son M, Cho Y, Oh J, et al. Social inequalities in life expectancy and mortality 19. Nogueira H. What is happening to health in the economic downturn? A view of during the transition period of economic crisis (1993-2010) in Korea. Int J Equity the Lisbon Metropolitan Area, Portugal. Ann Hum Biol. 2016;43:164–8. Health. 2012;11:71.