Gac Sanit. 2022;36(5):452–458 Original article Socioeconomic inequalities in self-assessed health and mental health in Barcelona, 2001-2016 Xavier Bartoll-Rocaa,b, , Laia Palènciaa,b,c , Mercè Gotsensa,b , Carme Borrella,b,c a Agència de Salut Pública de Barcelona, Barcelona, Spain b Institut d’Investigació Biomèdica Sant Pau, Barcelona, Spain c CIBER de Epidemiología y Salud Pública (CIBERESP), Spain a r t i c l e i n f o a b s t r a c t Article history: Objective: Previous research has found persistent socioeconomic inequalities in health outcomes at the Received 17 October 2020 national level, with different patterns after the economic crisis. However, inequalities in urban areas are Accepted 15 February 2021 also important. This study analyses socioeconomic inequalities in self-assessed health and mental health in the city of Barcelona. Keywords: Method: Repeated cross-sectional design using quinquennial data from the Barcelona Health Surveys car- Socioeconomic factors ried out in 2001, 2006, 2011 and 2016 for the population older than 22 years. Robust Poisson regressions Economic recession models were used to compute socioeconomic gradients and relative (RII) and slope indexes of inequality Health status disparities Mental health (SII) by occupational social class, with stratification by sex. RII and SII were also obtained with further adjustment by employment situation. Results: A consistent socioeconomic gradient was found for all years except for 2011. Relative and absolute inequalities followed a V-shape, showing a drop during the economic crisis but widening thereafter to recover pre-crisis figures for self-assessed health and widening for mental health, in both relative and absolute terms in 2016. Adjustment for employment situation reduces inequalities but a large part of these inequalities remains, with variability across years. Conclusions: The lasting effects of the 2008 economic crisis and the austerity programmes imposed since then may have contributed to the persistence of socioeconomic inequalities in self-assessed health and the widening of those for mental health. © 2021 SESPAS. Published by Elsevier España, S.L.U. Desigualdades socioeconómicas en salud percibida y salud mental en Barcelona, 2001-2016 r e s u m e n Palabras clave: Objetivo: La investigación previa ha reportado desigualdades socioeconómicas persistentes en salud en Factores socioeconómicos el territorio nacional, con diferentes patrones después de la crisis económica. Sin embargo, las desigual- Recesión económica dades en las áreas urbanas son también importantes. Este estudio analiza las desigualdades en salud Diferencias en el estado de salud autopercibida y salud mental en la ciudad de Barcelona. Salud mental Método: Diseño de corte transversal con datos quinquenales de la Encuesta de Salud de Barcelona llevada a cabo en 2001, 2006, 2011 y 2016 para la población mayor de 22 años. Se utilizan modelos de regresión robusta de Poisson para calcular el gradiente social y los riesgos relativo (RII) y absoluto de desigualdad (SII) por clase social ocupacional, estratificando por sexo. El RII y el SII se obtienen también ajustando adicionalmente por situación laboral. Resultados: Se encuentra un gradiente social para todos los años excepto para 2011. Se observan desigual- dades relativas y absolutas en forma de V para la salud autopercibida, con una reducción durante la crisis económica, pero recuperándose a partir de esta para volver a los valores anteriores a la crisis para salud autopercibida y ampliándose para la salud mental, tanto en términos relativos como absolutos, en 2016. Ajustar por la situación laboral reduce las desigualdades, pero una gran parte de estas desigualdades permanece, con variabilidad, en los años. Conclusiones: Los efectos duraderos de la crisis económica de 2008 y los programas de austeridad impuestos desde entonces pueden haber contribuido a la persistencia de las desigualdades socioeconómi- cas en salud autopercibida y a su ampliación en la salud mental. © 2021 SESPAS. Publicado por Elsevier España, S.L.U. Corresponding author. E-mail address: xbartoll@aspb.cat (X. Bartoll-Roca). https://doi.org/10.1016/j.gaceta.2021.02.009 0213-9111/© 2021 SESPAS. Published by Elsevier España, S.L.U. X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 453 Introduction of psychosocial problems has been found across European countries among unemployed persons, those with a lower educational level Socioeconomic inequalities in health have been widely analysed and those living in urban areas.16 However, these recent increases at the national and cross-national level but there is less evidence in mental health problems can only partially be explained by trends at the urban level, even though inequalities in cities may be as in unemployment and wages. It has been alleged that the auster- important as those in countries.1 Large urban areas are a setting ity measures have contributed to the widening in mental health for global/local dynamics, from climate change to job delocalisa- inequalities.17 For instance, in England, relative inequalities in tion, migration and housing bubbles potentially affecting health mental health reduced following the recession, between 2008 and inequalities. The aim of this study was to provide evidence of trends 2010, but increased significantly more between the recovery years in socioeconomic inequalities in self-assessed health and mental 2009 and 2013 and seemed to widen thereafter.18 In Spain, since the health during the last 16 years in the urban area of Barcelona 2008 economic crisis, mental health has also clearly deteriorated.19 (Spain). The increase in inequalities is largely due to a worse trend among A common feature of socioeconomic inequalities in self- those in the lowest socioeconomic position. At national level, using assessed health, at both the cross-national and national level, is two waves of the national health survey for years 2006 and 2012, the persistence of inequalities across European countries or small increasing inequalities have been found by social class among men oscillations without trend.2 Recent trends in socioeconomic health and long-term unemployed.13,20,21 At regional level, it has been inequalities have inevitably dealt with the effects of the 2008 eco- found larger adverse effects among those with lower education nomic crisis, and inequalities since then have seemed unchanged. level in Andalusia for years 2007 and 2012,22 and using four waves For instance, one study using national surveys reported that abso- of the Basque Health Survey from 1997 to 2013, an increase in lute inequalities by education and occupation in self-assessed relative inequalities by social class among men.23 health were mostly constant across 17 European countries in 1990 This study contributes to the existing literature by extending and 2010, and while relative inequalities increased, almost no the period of analysis and showing its evolution at local level. The country consistently experienced a significant decline in either objective was to analyse socioeconomic inequalities by occupa- absolute or relative inequalities.3 Another study analysing health tional social class in the self-assessed health and mental health of inequalities by education level across 27 European countries the Barcelona population by using data from a quinquennial survey observed no discernible short-term impact of the 2008 crisis at the from 2001 to 2016. population level.4 Similarly, one study using the European Union of Income and Living Conditions (EU-SILC) panel from 2005 to 2011 Method did no find differences before and after the economic crisis among low educated and unemployed.5 Several national experiences also Design and study population show the persistence of socioeconomic health inequalities. For instance, in Japan, relative and absolute socioeconomic inequalities Data were drawn from four cross-sectional repeated waves of by household income in self-assessed health narrowed or remained the Barcelona Health Survey (BHS) for the years 2001 and 2006 stable between 1986 and 2007, despite the economic stagnation pre-crisis, 2011 crisis, and 2016 post-crisis. The BHS aims to peri- during the 90s and adverse social changes.6 Similarly, inequalities odically (i.e., every 5 years) study the health status, health-related by education level and income also appear to persist in Germany behaviours, and use of health services among the non-institutional during the period 2003-2012 after the recession of 2008.7 In Eng- population resident in the city of Barcelona. The sample unit land, self-perceived health (and mental health) remained stable consisted of randomly selected individuals from the city. All from 1996 to 2009, with no evidence of widening socioeconomic non-responders were substituted by randomly previously-selected inequalities.8 However, another study using the EU-SILC panel data persons with the same characteristics (age, sex, and area) until the but for the recovery years from 2011 to 2013 found steeper declines target sample size was achieved. The information was collected in self-assessed health among unemployed and single mothers and by trained personnel through face-to-face interviews. For the pur- no different by education level.9 These results suggest that the poses of this study, we excluded individuals younger than 23 years effects of the economic crisis may disproportionally affect specific to ensure potential compliance of the educational/training period. social groups and with larger effects with elapsed time since the The final sample size was 7740 for 2001, 2983 for 2006, 3249 for start of the economic crisis. In Spain, the evidence for trends in 2011 and 3252 for 2016. self-assessed health inequalities is not clear cut.10 One study using the National Health Survey (NHS) for year 2006 and 2011 reported increasing inequalities in self-assessed health by education level.11 Variables A worse trend in self-assessed health among unemployed was also reported using the Living Conditions Survey for years 2007 and 1) Dependent variables 2011.12 A finding that was not confirmed applying matching tech- Self-assessed health. Participants were asked about their per- nics to data from NHS for the years 2006 and 2011.13 Interestingly, ceived general health using the question: “How is your health in a study including four waves 2005, 2008, 2011, 2014 of the Bank of general? There were five possible answers that were dichotomized Spain’s Survey of Household Finances found worse self-assessed in good health status (categories “excellent”, “very good” and health trends among family types characterized with a low job “good”) and poor health (“fair” and “bad”). intensity.14 Again, the results point that the economic crisis may Mental health was measured with the 12-item version of have exacerbated inequalities through weakening the labour mar- the General Health Questionnaire (GHQ-12). This questionnaire ket position of specific social groups for which economic recovery focuses on temporary problems, occurring in the last 30 days, and fades away. is a widely used screening instrument to detect risk of current, Unlike the persistent inequalities reported in self-assessed diagnosable psychiatric disorders. Questions were answered on a health, mental health can be more sensitive to situations of short- 4-point Likert scale (from 1 to 4). Following recommendations, we term employment and material deprivation. There is an extensive used a 2-point scoring method, rating a problem as absent (0) or literature on mental health trends since the economic recession. present (1). Participants scoring 3 or more were classified as having Systematic reviews have concluded a worsening of the mental poor mental health.24 health trends since the economic crisis.15 As expected, a higher risk 2) Independent variables 454 X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 Occupational social class is a measure of socioeconomic posi- cumulative frequency of occupational class instead of its categori- tion based on a graded hierarchy of occupations ranked according cal original form. The robust Poisson regression for binary data has to skill. The exact coding is based on the 5-category classification been proved to yield reliable RII estimates and has the advantage proposed by the Spanish Society of Epidemiology:25 occupational over the binomial model of minimizing omitted variable bias.26 social class (OSC) I, managers of firms with ≥10 employees and pro- Absolute differences are measured by the SII corresponding to the fessionals with a university degree (the reference category); OSC II, difference between the estimated prevalences at the two extremes managers of firms with < 10 employees and professionals without of occupational class of the fitted model. The baseline analysis was a university degree; OSC III, intermediary occupations and support; adjusted for the sociodemographic variables of age, marital status OSC IV, supervisors and semi-qualified workers; OSC V, unskilled and foreign birth. A second model was adjusted for employment sit- workers; where OSC I represents the highest and OSC V the lowest uation to assess its potential mediating role in the RII and SII and to socioeconomic position. Hereafter, non-manual class-term is used account for the changing composition of the population and work to refer OSCs I and II and manual class as OSC from III to V. Respon- force during the study period. All computations were performed dents who never had an occupation were classified according to with the statistical software Stata v15. the social class of the reference person of the household. The syntax used and the detailed regression results are provided 3) Control variables as supplementary material in online Appendix. Control variables include age, marital status in three categories: being married as the base category, single, and the third being sep- arated, widowed or divorced, being born in a foreign country (yes Results or no), and employment status in six categories (employed as the reference category, homemaker, unemployed, student, retired and Demographic and socioeconomic changes during the study disabled). All models were stratified by sex. period are shown in Table 1. Comparing the first and final year, there was an upward trend in the share of the population older than 75 years, unmarried people, those in the highest occupational class, Data analysis unemployed persons and a large increase in the share of immigrant population. The trends for health outcomes were favourable for We conducted a descriptive statistical analysis and calculated poor self-assessed health (decreasing by 2.6% for men and 4.5% for age-standardised prevalences with confident intervals in poor self- women), but were unfavourable for poor mental health (increasing assessed health and mental health. To test for associations between by 6.3% for men and 2.5% for women). health outcomes and the occupational class for each sex and sur- Age-standardised prevalences with confidence intervals in poor vey year, the relative risk was obtained by running a robust Poisson self-assessed health and mental health by occupational class, sex regression with the occupational class as the categorical variable. and years are shown in Table 2. The broad picture is that in the As a summary measure of socioeconomic inequalities, the rela- post-crisis period (2016) inequality levels recovered from the pre- tive index of inequality (RII) and the slope index of inequality (SII) crisis periods, with 2011 —during the economic crisis— being the were computed. The RII was also obtained by running a robust year when inequalities reduced due to a simultaneous increase in Poisson regression model but fitted against the midpoints of the adverse health status in non-manual classes and a reduction of poor Table 1 Sociodemographic and health characteristics, Barcelona 2001-2016. Men Women 2001 2006 2011 2016 2001 2006 2011 2016 N = 3288 N = 1073 N = 1409 N = 1413 N = 3595 N = 1164 N = 1436 N = 1569 Age group 23-35 years 27.0 30.5 26.5 23.0 23.7 26.7 25.1 22.6 36-45 years 19.0 21.3 21.3 23.0 20.3 19.3 19.9 19.9 46-55 years 17.3 15.9 17.4 19.0 16.6 17.5 17.8 17.6 56-65 years 16.5 14.0 14.6 13.8 16.4 14.8 15.9 15.2 66-75 years 13.5 10.7 11.0 12.1 14.6 12.5 10.6 13.4 76 years or more 6.7 7.6 9.2 9.1 8.4 9.2 10.7 11.8 Marital status Married 68.9 59.7 58.2 54.3 62.8 54.7 51.7 52.6 Single 25.5 33.7 35.0 35.4 20.4 27.6 31.1 29.4 Widowed, separated, divorced 5.6 6.6 6.8 10.2 16.9 17.7 17.2 17.9 Foreign born 3.1 16.2 14.9 22.6 3.7 12.4 16.5 24.3 Occupational class OSC I 17.4 20.6 20.4 22.6 9.8 12.5 16.0 20.7 OSC II 9.9 13.6 10.7 8.7 9.2 11.1 13.5 10.6 OSC III 31.3 21.2 23.5 23.3 31.2 28.0 29.0 25.8 OSC IV 35.8 37.4 33.6 35.7 35.5 32.9 26.8 26.6 OSC V 5.7 7.2 11.9 9.9 14.3 15.5 14.6 16.4 Employment status Employed 64.5 70.9 59.5 60.8 46.2 57.4 54.7 54.8 Homemaker 0.1 0.2 0.0 0.0 28.8 17.2 8.7 10.2 Unemployed 5.2 4.4 11.0 10.5 4.8 5.3 10.6 9.1 Student 24.4 21.8 25.1 22.8 16.3 17.0 21.2 21.9 Retired 2.6 1.5 1.7 1.9 1.9 1.2 2.1 1.3 Disabled 3.2 1.2 2.7 4.0 2.0 1.9 2.7 2.7 Health outcomes Poor self-assessed health 20.0 18.1 18.0 17.4 27.9 28.6 23.2 23.4 Poor mental health 10.2 9.5 11.6 16.5 17.5 20.5 18.1 20.0 OSC: occupational social class, with OSC I the more advantaged and OSC V the less advantaged social class. X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 455 Table 2 Age-adjusted prevalences and 95% confidence intervals of poor self-assessed health and mental health by occupational social class and sex, Barcelona, 2001-2016. 2001 2006 2011 2016 Men Poor self-assessed health OSC I 9.3 (7.0-11.7) 10.2 (6.3-14.2) 14.1 (10.1-18.1) 12.0 (8.6-15.5) OSC II 9.6 (7.3-11.9) 15.5 (11.3-19.7) 23.6 (18.9-28.3) 16.0 (11.4-20.6) OSC III 20.1 (17.3-22.9) 17.9 (12.2-23.6) 19.3 (14.4-24.3) 17.2 (12.5-21.9) OSC IV 27.9 (24.6-31.3) 22.9 (17.5-28.4) 18.8 (14.1-23.5) 19.8 (15.3-24.4) OSC V 29.4 (25.4-33.5) 30.9 (24.4-37.5) 20.1 (16.3-23.8) 26.3 (21.7-30.8) Poor mental health OSC I 6.8 (4.8-8.8) 7.4 (4.0-10.9) 10.6 (7.1-14.1) 8.4 (5.4-11.3) OSC II 10.8 (8.3-13.2) 10.0 (6.5-13.5) 12.0 (8.4-15.5) 9.2 (5.7-12.9) OSC III 9.6 (7.5-11.6) 11.1 (6.4-15.8) 8.3 (4.9-11.8) 14.7 (10.4-19.2) OSC IV 11.7 (9.3-14.1) 9.3 (5.6-13.1) 11.9 (7.9-15.7) 20.1 (15.6-24.8) OSC V 14.8 (11.7-17.9) 13.6 (8.7-18.4) 18.4 (14.7-22.0) 28.8 (24.1-33.5) Women Poor self-assessed health OSC I 12.8 (9.8-15.7) 14.4 (9.5-19.4) 20.3 (15.9-24.8) 11.3 (8.5-14.2) OSC II 16.2 (13.9-19.9) 18.1 (13.1-23.1) 16.0 (12.2-19.9) 18.9 (14.5-23.2) OSC III 19.8 (16.9-22.8) 23.8 (18.0-29.6) 23.5 (19.5-28.5) 22.2 (17.2-27.2) OSC IV 32.5 (29.2-35.8) 32.6 (26.5-38.7) 24.8 (19.2-30.4) 26.7 (21.2-32.3) OSC V 42.0 (38.2-45.8) 43.0 (36.6-49.5) 27.0 (21.8-32.3) 34.7 (29.6-39.8) Poor mental health OSC I 11.2 (8.4-14.0) 11.4 (7.0-15.9) 15.9 (11.9-19.9) 12.4 (9.4-15.4) OSC II 16.5 (13.5-19.5) 14.4 (9.8-19.0) 13.5 (9.9-17.1) 10.4 (7.0-13.9) OSC III 15.5 (12.8-18.1) 19.7 (14.3-25.2) 20.9 (16.1-25.8) 15.8 (11.4-20.2) OSC IV 18.3 (15.5-21.0) 22.8 (17.4-28.2) 19.5 (14.4-24.7) 27.2 (21.6-32.8) OSC V 23.1 (19.8-26.4) 30.5 (24.5-36.5) 17.5 (13.0-22.1) 29.0 (24.1-33.8) OSC: occupational social class, with OSC I the more advantaged and OSC V the less advantaged social class. health status in the manual classes. For men, the gap in poor self- Discussion assessed health by class squeezes in 2011 (6.0%) but widens again in 2016 (14.3%) below the pre-crisis gap (20.1% for 2001 and 20.7% The use of data from the Barcelona city quinquennial health for 2006). The gap in poor mental health remains fairly stable from surveys from 2001 to 2016 revealed persistent absolute and rela- 2001 to 2011 (between 6.2% and 8.0%) but reaches a maximum in tive inequalities by occupational social class in self-assessed health 2016 (20.4%). Among women, the gap in poor self-assessed health and mental health outcomes for men and women across years by class follows a similar pattern of that of men, squeezing in 2011 except during the economic crisis in 2011, which showed a drop. (6.7%) but widening again in 2016 (23.4%) below to the pre-crisis Absolute inequalities in mental health widened in the post-crisis gap (29.2% for 2001 and 28.6% for 2006). The gap in mental health period in 2016 in both men and women to values higher than pre- also reaches a minimum in 2011 (1.6%) but widens again in 2016 crisis values. Adjustment for employment status reduced absolute (16.6%) below to pre-crisis levels in 2006 (19.1%). For men and inequalities in both sexes but a large share of inequality remains. women, the increases in 2016, specially for poor mental health, Our results suggest that the recurrent inequality was due to are driven by the worsening among manual classes. faster increases in poor self-assessed health and mental health Tables 3 and 4 show the relative risk by occupational class, the among persons with lower socioeconomic position after the relative index of inequality and the absolute difference of poor economic crisis compared to those with a more advantaged socioe- health outcomes and year for men and women. The first shows conomic position, for whom it declined. This result contrasts with a socioeconomic gradient for most years, with the population in recent findings of a slowdown in declines in poor self-assessed the lower occupational classes having a higher risk of experienc- health among people with both low and high socioeconomic posi- ing poor self-perceived and mental health than the higher ones. tion in the countries worst hit by the economic crisis, leaving Both the RII and the SII for self-assessed health followed this V- inequalities unchanged.4 An examination of the Spanish case con- shape trend of vanishing inequalities during the economic crisis firmed previous studies that the economic crisis could have affected and for mental health even widened again in 2016. For poor self- more intensely those in lower socioeconomic classes.10,14 The assessed health, the RII (95% confidence interval) decreased among increase in health inequalities recorded in Barcelona in 2016 may men, from 3.42 (2.61-4.47) in 2001 to a non-significant value in be related with labour reforms enacted after the economic crisis. 2011, but increased to 2.83 (1.81-4.42) in 2016. The same occurred The worsening of the work environment has been pointed as a for poor mental health, from 2.04 (1.37-3.03) in 2001 to a non- potential mechanism to the increasing inequalities in health.12,23 significant value in 2011, and rose again to 4.02 (2.43-6.64) in 2016. The Spanish labour reform of 2012 made layoff cheaper, weakened For women, the RII for poor self-assessed health decreased from coverage of collective agreements and increased job insecurity.27 3.16 (2.55-3.91) in 2001 to 1.66 (1.14-2.40) in 2011 and rose again to Our analysis that includes a longer term, probably captures bet- 3.92 (2.67-5.76) in 2016. For poor mental health, the RII decreased ter the lasting effect of the relentless crisis than previous research, from 1.83 (1.39-2.41) in 2001 to a non-significant value in 2011 and especially for mental health. Absolute and relative inequalities in rose again to 3.55 (2.31-5.47) in 2016. For men and women, abso- mental health for men and women in 2016 widened compared lute inequalities in self-assessed health also followed a V-shape and to figures for previous years. These findings are close to the pat- widened for mental health in 2016. When we adjusted for employ- tern followed by England from 1991 to 2014, when inequalities ment situation, inequalities for self-assessed health and mental in mental widened after the recession and during the auster- health were reduced for both men and women but a large share ity measures.18 The adverse working conditions, job insecurity, of inequality remains with variability across the years. higher unemployment rates and poor prospects for younger peo- ple disproportionately affected persons with lower socioeconomic 456 X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 Table 3 Risk of poor self-assessed health and mental health by occupational class, relative index of inequality (RII) and absolute difference (SII) with 95% confident intervals for men, Barcelona, 2001-2016. 2001 2006 2011 2016 Poor self-assessed healtha OSC I 1 (ref.) 1 (ref.) 1 (ref.) 1 (ref.) OSC II 0.97 (0.58-1.61) 1.37 (0.72-2.60) 1.41 (0.91-2.20) 1.38 (0.75-2.52) OSC III 2.05d (1.51-2.79) 1.82b (1.11-2.97) 1.47b (1.02-2.11) 1.55b (1.03-2.32) OSC IV 2.78d (2.07-3.73) 2.29d (1.48-3.55) 1.39 (0.99-1.97) 1.84c (1.28-2.64) OSC V 2.78d (1.94-4.00) 3.13d (1.82-5.37) 1.54b (1.02-2.34) 2.79c (1.81-4.29) RII base 3.42d (2.61-4.47) 3.34b (2.04-5.49) 1.45 (0.98-2.14) 2.83d (1.81-4.42) SII base 19.63d (14.84-24.43) 19.90d (10.70-29.11) 4.62 (−1.83-11.06) 15.52d (7.98-23.07) RII adjusted 2.75d (2.1-3.59) 3.39d (2.04-5.62) 1.22 (0.82-1.8) 2.08c (1.34-3.24) SII adjusted 12.25d (8.29-16.20) 17.20d (8.63-25.77) 1.81 (−3.37-6.99) 7.54c (2.62-12.46) Poor mental healtha OSC I 1 (ref.) 1 (ref.) 1 (ref.) 1 (ref.) OSC II 1.46 (0.91-2.33) 1.23 (0.57-2.68) 1.11 (0.63-1.97) 1.28 (0.64-2.55) OSC III 1.44 (0.98-2.12) 1.42 (0.74-2.73) 0.80 (0.47-1.37) 1.78b (1.09-2.93) OSC IV 1.75c (1.22-2.53) 1.25 (0.67-2.32) 1.03 (0.66-1.62) 2.40d (1.55-3.73) OSC V 2.22c (1.36-3.62) 1.84 (0.83-4.05) 1.75b (1.07-2.87) 3.43d (2.09-5.63) RII base 2.04d (1.37-3.03) 1.48 (0.71-3.08) 1.64 (0.90-2.98) 4.02d (2.43-6.64) SII base 6.12c (2.47-9.73) 3.72 (−3.04-10.48) 4.72 (−1.16-10.61) 20.10d (11.14-29.06) RII adjusted 1.67b (1.12-2.48) 1.41 (0.68-2.91) 1.28 (0.69-2.35) 3.17d (1.89-5.31) SII adjusted 4.08b (0.69-7.46) 2.76 (−3.09-8.61) 2.42 (−2.75-7.60) 12.54d (5.45-19.63) a Risk by class corresponds to the base model. b p < 0.05. c p < 0.01. d p < 0.001. OSC: occupational social class, with OSC I the more advantaged and OSC V the less advantaged social class; Baseline model included age, marital status and foreign born; adjusted model added base adjustment for employment situation. Table 4 Risk of poor self-assessed health and mental health by occupational class, relative index of inequality (RII) and absolute difference (SII) with 95% confident intervals for women, Barcelona 2001-2016. 2001 2006 2011 2016 Poor self-assessed healtha OSC I 1 (ref.) 1 (ref.) 1 (ref.) 1 (ref.) OSC II 1.36 (0.89-0.06) 1.22 (0.64-2.33) 0.85 (0.54-1.35) 1.50 (0.91-2.48) OSC III 1.62c (1.15-2.30) 1.76b (1.04-3.00) 1.26 (0.89-1.78) 1.80c (1.20-2.68) OSC IV 2.53d (1.80-3.54) 2.32c (1.38-3.90) 1.27 (0.90-1.80) 2.38d (1.60-3.52) OSC V 3.04d (2.16-4.28) 2.83d (1.66-4.82) 1.48b (1.02-2.14) 3.35d (2.25-5.00) RII base 3.16d (2.55-3.91) 2.93d (2.02-4.25) 1.66b (1.14-2.40) 3.92c (2.67-5.76) SII base 27.38c (22.06-32.69) 29.34c (18.50-40.18) 10.70b (2.24-19.16) 23.65d (15.65-31.64) RII adjusted 3.06d (2.47-3.8) 2.66d (1.83-3.88) 1.44b (1.01-2.06) 3.40d (2.31-5.00) SII adjusted 21.37d (16.24-26.50) 22.46d (12.08-32.84) 5.16b (0.02-10.31) 18.47d (10.92-26.02) Poor mental healtha OSC I 1 (ref.) 1 (ref.) 1 (ref.) 1 (ref.) OSC II 1.47 (0.99-2.19) 1.20 (0.61-2.37) 0.90 (0.54-1.50) 0.80 (0.46-1.39) OSC III 1.40 (0.99-1.97 1.69 (0.99-2.86) 1.36 (0.91-2.02) 1.23 (0.83-1.83) OSC IV 1.67c (1.20-2.33) 1.88b (1.13-3.15) 1.27 (0.85-1.91) 2.13d (1.48-3.07) OSC V 2.03d (1.43-2.87) 2.21c (1.29-3.77) 1.01 (0.63-1.63) 2.32d (1.58-3.41) RII base 1.83d (1.39-2.41) 2.17d (1.38-3.41) 1.16 (0.76-1.79) 3.55d (2.31-5.47) SII base 9.44d (5.03-13.84) 15.98c (6.29-25.68) 2.39 (−4.46-9.25) 23.78d (14.51-33.06) RII adjusted 1.77d (1.34-2.35) 1.97c (1.24-3.12) 1.03 (0.67-1.59) 2.94d (1.91-4.52) SII adjusted 8.13d (3.86-12.39) 13.77c (3.84-23.70) 0.73 (−4.67-6.12) 17.83d (9.35-26.32) a Risk by class corresponds to the base model. b p < 0.05. c p < 0.01. d p < 0.001. OSC: occupational social class with OSC I the more advantaged and OSC V the less advantaged social class; Baseline model included age, marital status and foreign born; adjusted model added base adjustment for employment situation. position.28 Indeed, this hypothesis is partially confirmed by the recent increases in mental health problems.22 Even though it has reduction in absolute inequalities in mental health after adjustment been alleged that the health system in Spain has managed to for employment situation. Our results differ from some previous avoid irreversible deterioration,29 the magnitude of the social cri- research that finds increasing inequalities for men only.20,21,23 sis depends on the welfare response to the economic crisis.30 In In Barcelona, temporary work, part-time work and other pre- the case of Barcelona, the trend towards a reduction in health carious working arrangements spread across the work force but inequalities seems to have been offset by the lasting effects of the affected women more than men.28 However, absolute inequalities economic crisis and the austerity response, reversing any reduction in 2016, after adjustment for employment, still remained above in inequalities and making them persistent or even widening. mean trends, and consequently a large share in inequalities remains An unexpected finding was that, during the economic crisis in unexplained, with employment situation only partly explaining 2011, this population showed the largest reduction in inequali- X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 457 ties. It has been suggested that the economic crisis offered some Transparency declaration social groups the possibility of acquiring healthier habits: more hours of sleep, more physical exercise, less demanding work, more The corresponding author on behalf of the other authors guar- home-made cooking, and more time for leisure.31 Some scholars antee the accuracy, transparency and honesty of the data and have stressed the differential effects of the economic crisis on spe- information contained in the study, that no relevant information cific social groups, depending on what the dominant mechanism has been omitted and that all discrepancies between authors have is, a coping response or a stressful one.32 Nonetheless, our results been adequately resolved and described. also show an increase in adverse health outcomes during the eco- nomic crisis for people with high socioeconomic position. It also Authorship contributions has been pointed that the populations with better objective health indicators report worse perceived health, due to an adverse social X. Bartoll-Roca designed the study and carried it out. L. Palència context, known as the “status inconsistency effect”, and on the con- and M. Gotsens provided methodological support. C. Borrell con- trary “survivor employees” in an adverse work environment may tributed in writing and discussing the results. All the authors have overestimate their living conditions, known as the “other’s unem- contributed corrections and improvements in the writing and final ployment effect”12,32 . The lack of a worsening of an adverse health presentation of the study. effect amidst the economic crisis may also be due to a delayed effect. It has been emphasised that inequalities have stronger effects on Funding self-assessed health in follow-up analyses with the incorporation of lag effects.33 None. A strength of our study is that it used robust at the city level and covered a wide time span. It is recommended to report both RII and SII, both measures have the advantage of including the information Conflicts of interest available from all socioeconomic position. However, changes in the composition of those groups can lead to bias in the interpretation of None. disparities.34 Our inequality estimates can be conservative because, with increasing precariousness, occupational class may not fully Appendix A. Supplementary data capture emerging sources of inequality such as those engaged in highly temporary jobs, or other adverse working conditions. In Supplementary data associated with this article can be found, in this regard, the GHQ is also known to have lower sensitivity to the online version, at doi:10.1016/j.gaceta.2021.02.009. more disadvantaged groups so that underlying inequalities can be underestimated.35 References In summary, the lasting effects of the 2008 economic cri- sis and the austerity programmes imposed since then may have 1. Glaeser EL, Resseger M, Tobio K. Inequality in cities. J Reg Sci. 2009;49:617–46. contributed to the persistence of socioeconomic inequalities in self- 2. Kunst AE, Bos V, Lahelma E, et al. Trends in socioeconomic inequalities in self- assessed health in 10 European countries. Int J Epidemiol. 2005;34:295–305. assessed health and the widening of those for mental health. 3. Hu Y, van Lenthe FJ, Borsboom GJ, et al. Trends in socioeconomic inequalities in self-assessed health in 17 European countries between 1990 and 2010. J Epidemiol Community Health. 2016;70:644–52. 4. Mackenbach JP, Valverde JR, Artnik B, et al. Trends in health inequalities in 27 European countries. Proc Natl Acad Sci USA. 2018;115:6440–5. 5. Abebe DS, Tøge AG, Dahl E. Individual-level changes in self-rated health before What is known about the topic? and during the economic crisis in Europe Economic crisis in Europe. Int J Equity Health. 2016;15:1–8. 6. Hiyoshi A, Fukuda Y, Shipley MJ, et al. Inequalities in self-rated health in Several studies show the persistence of economic inequal- Japan 1986-2007 according to household income and a novel occupational ities in health outcomes. With some fluctuations during the classification: national sampling survey series. J Epidemiol Community Health. economic crisis of 2008, socioeconomic inequalities in poor 2013;67:960–5. self-assessed health persist or even widens for poor mental 7. Wachtler B, Hoebel J, Lampert T. Trends in socioeconomic inequalities in self- rated health in Germany: a time-trend analysis of repeated cross-sectional health. health surveys between 2003 and 2012. BMJ Open. 2019;9:e030216. 8. Maheswaran H, Kupek E, Petrou S. Self-reported health and socio-economic What does this study add to the literature? inequalities in England, 1996-2009: repeated national cross-sectional study. Soc Sci Med. 2015:135–46, 136-137. Potential downward trends in poor self-assessed and men- 9. Nelson K, Tøge AG. Health trends in the wake of the financial crisis — increasing tal health in pre-crisis period are reverted in post-crisis period. inequalities? Scand J Public Health. 2017;45:22–9. 10. Ministerio de Sanidad, Consumo y Bienestar Social. Crisis económica y salud en Health inequalities persist even after accounting by employ- España. Madrid. Informes, estudios e investigación; 2018. ment situation. 11. Barroso C, Abásolo I, Cáceres JJ. Health inequalities by socioeconomic character- istics in Spain: the economic crisis effect. Int J Equity Health. 2016;45:22–9. What are the implications of the results? 12. Calzón S, Fernández A, Del Amo González MDPL, et al. Diferencias por sexo de la salud percibida antes y durante la crisis económica (2007 y 2011). Rev Esp Salud Publica. 2017;91:e1–12. Local governments face global threats with limited local 13. Urbanos-Garrido RM, Lopez-Valcarcel BG. The influence of the economic crisis competences and financing. The reversion of the austerity on the association between unemployment and health: an empirical analysis response would help to reduce socioeconomic inequalities in for Spain. Eur J Health Econ. 2015;16:175–84. health. 14. Saez M, Vidiella-Martin J, Casasnovas GL. Impact of the great recession on self- perceived health in Spain: a longitudinal study with individual data. BMJ Open. 2019;9:1–9. 15. Frasquilho D, Matos MG, Salonna F, et al. Mental health outcomes in times of eco- nomic recession: a systematic literature review. BMC Public Health. 2016;16:1. 16. Pinto-Meza A, Moneta MV, Alonso J, et al. Social inequalities in mental health: results from the EU contribution to the World Mental Health Surveys Initiative. Editor in charge Soc Psychiatry Psychiatr Epidemiol. 2013;48:173–81. 17. Barr B, Kinderman P, Whitehead M. Trends in mental health inequalities in Eng- land during a period of recession, austerity and welfare reform 2004 to 2013. Miguel ÿngel Negrín Hernández. Soc Sci Med. 2015;147:324–31. 458 X. Bartoll-Roca et al. / Gac Sanit. 2022;36(5):452–458 18. Thomson RM, Niedzwiedz CL, Katikireddi SV. Trends in gender and socioe- 26. Zou G. A modified Poisson regression approach to prospective studies with conomic inequalities in mental health following the Great Recession and binary data. Am J Epidemiol. 2004;159:702–6. subsequent austerity policies: a repeat cross-sectional analysis of the Health 27. Eichhorst W, Wehner C, Eichhorst W, et al. Labor market reforms in Europe: Surveys for England. BMJ Open. 2018;8:e0222924. towards more flexicure labor markets? IZA. 2016. No. 9863. 19. Moreno-Küstner B, Masedo Gutierrez AI. Economic crisis and mental health- 28. Departament d’Estudis i Programació. El treball assalariat a Barcelona 2015. findings from Spain. Die Psychiatrie. 2017;14:95–102. Temporalitat, intensitat laboral i durada dels contractes. Barcelona: Ajuntament 20. Bartoll X, Palència L, Malmusi D, et al. The evolution of mental health in Spain de Barcelona; 2017. during the economic crisis. Eur J Public Health. 2014;24:415–8. 29. Lopez-Valcarcel BG, Barber P. Economic crisis, austerity policies, health and fair- 21. Moncho J, Pereyra-Zamora P, Tamayo-Fonseca N, et al. Is recession bad for your ness: lessons learned in Spain. Appl Health Econ Health Policy. 2017;15:13–21. mental health? The answer could be complex: evidence from the 2008 crisis in 30. Bacigalupe A, Shahidi FV, Muntaner C, et al. Why is there so much controversy Spain. BMC Med Res Methodol. 2018;18:78. regarding the population health impact of the great recession? Reflections on 22. Córdoba-Doña JA, Escolar-Pujolar A, San Sebastián M, et al. How are the three case studies. Int J Health Serv. 2016;46:5–35. employed and unemployed affected by the economic crisis in Spain? Educational 31. Ruhm CJ. Are recessions good for your health? Q J Econ. 2000;115:617–50. inequalities, life conditions and mental health in a context of high unemploy- 32. Catalano R, Goldman-Mellor S, Saxton K, et al. The health effects of economic ment. BMC Public Health. 2016;16:1–11. decline. Annu Rev Public Health. 2011;32:431–50. 23. Bacigalupe A, Esnaola S, Martín U. The impact of the Great Recession on mental 33. Kondo N, Sembajwe G, Kawachi I, et al. Income inequality, mortality, and self health and its inequalities: the case of a Southern European region, 1997-2013. rated health: meta-analysis of multilevel studies. BMJ. 2009;339:b4471. Int J Equity Health. 2016;15:17. 34. Regidor E. Measures of health inequalities: part 2. J Epidemiol Community 24. Goldberg D. Manual of General Health Questionnaire. UK: Windsor; 1978. Health. 2004;58:900–3. 25. Domingo-Salvany A, Bacigalupe A, Carrasco JM, et al. Propuestas de clase social 35. Bell T, Watson M, Sharp D, et al. Factors associated with being a false posi- neoweberiana y neomarxista a partir de la Clasificación Nacional de Ocupaciones tive on the General Health Questionnaire. Soc Psychiatry Psychiatr Epidemiol. 2011. Gac Sanit. 2013;27:263–72. 2005;40:402–7.