Monitoring indicators at the small area level has been a growing area of research in the last few decades, allowing the detection of widening socioeconomic inequalities.1 In particular, the use of official registers has enabled identification of socioeconomic health inequalities in mortality in small areas across European and Spanish cities.2,3 Estimation of health inequalities at the small areal level is of a major interest from a policy perspective as it allows improvement and monitoring of community health interventions. Interest in this topic has increased since the economic crisis of 2008, with systematic reviews reporting worsening in a variety of health outcomes and especially in mental health trends.4,5 In Spain, since the 2008 economic crisis, mental health has clearly deteriorated especially among men in the lowest socioeconomic position and the long-term unemployed.6 Previous evidence indicates that the urban environment is associated with worse mental health, although identification of a causal link is hampered by the complex interactions between urban environments and individual characteristics.7–9 However, there is scarce evidence of trends in inequalities at the small area level using survey data. Administrative delimitation of small areas does not generally respect an even distribution of population sizes so these small areas, as neighbourhoods, can register too few cases than statistically recommended10. In this case, indirect estimates for small areas can be obtained with spatial information under the principle that “the closer, the more similar”.11 Information from contiguous areas can be borrowed to smooth the estimate, so that a higher weight is assigned to the estimate of the closest areas, decreasing the problem of small sample size. The most common practice is to use random area effects to account for between-area variation.12

When smoothing techniques are used from surveys, small area estimation has to deal with missing values, and take into account the sampling design used in the estimates. Fortunately, recent approaches based on Bayesian smoothing techniques have been developed to account for sampling design.13,14 Any model that does not take sampling weights into account will be rendered biased. In the case of binary outcomes, sampling weights can be incorporated into the model by several possible transformations of the (binary health) outcome variable.14 In this study, we used the logit-normal transformation to obtain smoothed estimates of mental health at the neighbourhood level. Our investigation extends previous work on small area estimation using survey data in Spain by incorporating for the first time a complex survey design into small area estimation.15

We restricted our analysis to Barcelona city. The city lies in the North-East coast of Spain and has 1.7 million inhabitants. To our knowledge, it is the only city in Spain with data on mental health obtained regularly every 5 years between 2001 and 2016 through the Barcelona Health Survey (BHS). The BHS collects information from a representative sample of each of the 10 administrative districts of the city but not for the 73 neighbourhoods. These neighbourhoods group populations of unequal size, the largest with a population around 60 000 and the smallest with 500, with sharp socioeconomic differences.16

The objective of this study was, first, to estimate the prevalences of poor mental health in Barcelona at the neighbourhood level from 2001 to 2016 and, second, to analyse the association of mental health with neighbourhood income and its trend during this period.

In this study, we obtained small area estimates (at the neighbourhood level) for poor mental health incorporating a complex survey design through the logit-normal transformation. Smoothed estimates successfully reduced variability and provided a reliable picture of mental health inequalities. We observed inequalities in poor mental health across neighbourhoods (except for 2011), which increased in 2016.

A profile of inequalities that decreased during the economic crisis in 2011 but widened again in 2016 was also found in England.23 The decrease in inequalities during the economic crisis depended on whether the dominant response was a coping response or a stressful one.24 For instance, a pioneering study investigating the relationship between economic conditions and mortality rates during the period 1972-1991 argued that a larger proportion of the population may acquire healthier habits, such as more hours of sleep, more physical exercise, less demanding work and more time for leisure, leading to a reduction in health inequalities.25 Conversely, the Spanish labour reform of 2012 may have helped to prolong the adverse effects of the economic downturn through higher unemployment rates, worse working conditions, and greater job insecurity, leading to poor prospects for younger people and disproportionately affecting the population with a lower socioeconomic position.26

As stressed by the recent above-mentioned literature,9,27 individual determinants of mental health probably have stronger mechanisms than neighbourhood settings. However, the relentless effects of the economic crisis of 2008 have exacerbated divergence in neighbourhood vulnerability. It has been suggested that mechanisms of divergence may be greater in countries, such as in Spain, with a higher proportion of home-ownership and with a tight rental housing market. In this situation, it is more difficult for low- to middle-income groups and first home buyers to obtain a mortgage than in the past, while the financial effort of renting increases, so that this population is forced to move to more deprived areas.28 In particular, evictions have been a major problem in Barcelona. From the first registers in 2013 to 2016, 12,333 evictions have been recorded, which may have contributed to increasing inequalities in the 2016 wave, as evicted persons report poor mental health up to 6 times more frequently than the general population.29

Our result agrees with recent cross-sectional studies reporting weak evidence in showing increasing inequalities in the risk of psychosocial problems among neighbourhoods.30 Although the Neighbourhood Law was withdrawn in 2011 during the economic crisis, Barcelona has continued to carry out the municipal programme Health in the neighbourhoods aimed at reducing health inequalities. The programme reached 13,600 people among 25 of the most disadvantaged neighbourhoods in the city. Participants in these community interventions ranked mental health problems as among the first priorities needing to be tackled.31 Our study cannot be interpreted as a direct evaluation of this programme but it does provide indicators on the geographical scale in which the interventions are carried out and allows interventions to be focused on the neighbourhoods most requiring them without neglecting the importance of individual effects. Because some authors have found only weak evidence of area-based interventions to avoid the adverse “neighbourhood effect” on mental health,32–34 a mix of individual-level targeted and area policy is likely to be needed.

As stated, the association observed cannot be interpreted as causal, due to the repeated cross-sectional design rather than longitudinal or instrumental variable techniques, and to the lack of a proper measure of the neighbourhood environment. A limitation of our estimated models to further explore is the potential confounding between fixed and random effects. Spatial confounding occurs when covariates have a spatial pattern and are collinear with spatial random effects. To overcome this problem, it is proposed to specify random effects as orthogonal to fixed effects.35 However, while adding spatially correlated random effects does not adjust the fixed effects for the missing covariates, it does smooth the fitted values. Another limitation of our study is that we were unable to stratify by sex due to the small sample size in some areas. In the context of this small sample size, the results should be interpreted with caution if the adjacent neighbourhoods differ in their socioeconomic and environmental settings. In this case, the smoothed estimates could deviate from real values.

Data available to persons on request to xbartoll@aspb.cat from the Institut d’Investigació Biomèdica Sant Pau, Barcelona (Spain), in accordance with the data policy applicable to the Barcelona health survey.

Statistical codes to the Bayesian estimates are provided in the online Appendix.

Miguel Ángel Negrín Hernández.

The corresponding author on behalf of the other authors guarantee the accuracy, transparency and honesty of the data and information contained in the study, that no relevant information has been omitted and that all discrepancies between authors have been adequately resolved and described.

X. Bartoll-Roca and M. Marí-Dell’Olmo contributed to the planning, conception, design and implementation of the study. M. Marí-Dell’Olmo contributed to the finalisation and analysis of the data. X. Bartoll-Roca drafted the manuscript. M. Marí-Dell’Olmo, M. Gotsens and Laia Palència contributed to methodological revision of the manuscript. K. Pérez and E. Díez contributed to critical discussion, and C. Borrell to comments on the neighbourhood environment and to the final version of the manuscript.

None.

None.