Socioeconomic status (SES)1 and gender play a key role in the health of individuals and populations, the most disadvantaged groups having poorer outcomes in both physical2 and mental health.3 Social determinants have a greater impact in these early stages of life due to the special vulnerability of childhood, adolescence and youth to environmental stress which includes the moment of transition from educational settings to the workplace.4–6 In a systematic review, Reiss found that children and adolescents with poorer SES are two or three times more likely to develop mental health disorders.7 She also concluded that the theory of social causality plays a more relevant role than the theory of selection by health to understand the cycle of deprivation between generations that creates inequalities in mental health in children and adolescents.7 Most studies have evidenced consistent gender differences, internalizing disorders (anxiety, depression, obsessive–compulsive disorder and eating disorders) being more frequent in girls and externalizing ones (psychosis, attention deficit hyperactivity disorder [ADHD], autism and conduct disorders) in boys.8,9 The organization and characteristics of health systems can reduce inequalities generated by other social determinants or, on the contrary, amplify them.10,11 In line with this, it has been suggested that early intervention is critical to tackling adverse conditions experienced during childhood and adolescence that contribute to poorer levels of health in youth and adulthood.12,13

Studies analysing inequalities in mental health are generally based on surveys with questionnaires referring to the presence of psychiatric symptoms.7,11 On the other hand, the implementation of the electronic health record (EHR) in primary care and the coding of episodes in outpatient clinics and acute hospitals have created new sources of information about childhood, adolescence and youth for mental health epidemiology.8,9,14,15 New designs based on the analysis of complete population registries (real-world data) such as those available in Scandinavia8,9 and other European regions, including the Basque Country (Spain),14 offer a great opportunity to assess inequalities in mental health.16,17 The Basque Country has a population of 2,178,000 people and the funding of healthcare provision is based on the Beveridge model, the Basque Health Service being devolved to the regional government which has implemented an organization-wide database that has been validated for research.14,15 The association of premature mortality with the presence of mental illness has been previously evidenced in Danish and Swedish registries.18,19 It is not clear, however, whether that association may be mediated by SES, underlining the need to ascertain the separate weights of mental disorders and SES in the excess risk of death.20

The objective of this study was to identify and analyze inequalities by gender and SES in mortality and the prevalence of diagnosed mental health disorders in children, adolescents and young people up to 30 years of age in the Basque Country (Spain).

By December 31, 2018, the Basque Health Service database contained 609,381 individuals in the age range between 1 and 30 years, of which 15.9% (96,671) had been diagnosed with a mental disorder (Table 1). The low-SES group represented 7.8% of the population. From this more disadvantaged group, 22.1% of the individuals experienced some type of mental health disorder, compared to 11.4% in the higher income group. Half of the deaths in this age group during 2018 had a mental disorder diagnosis. Disaggregated cases for psychiatric hospital admissions and deaths are shown in Tables SM2, SM3 and SM4.

Table 2 shows the prevalence of the groups of diagnosed mental disorders as of 31 December 2018 according to age group, gender and SES. The low-SES group had a higher lifetime prevalence of mental disorders across five of the six categories; the exception being drug use disorders which were more prevalent in medium-SES individuals. The percentage of cases was higher in males for drug use (3.6% vs 2.8%), psychosis (1.4% vs 0.6%), conduct disorders (5.7% vs 3.7%) and ADHD (4.0% vs 1.5%), while females had more diagnoses of anxiety (9.0% vs 5.5%) and depression (1.0% vs 0.6%).

Table 3 and Table SM5 present the results of the logistic regression models estimating the adjusted odds ratios (AORs) with confidence intervals (CIs) of being diagnosed with any of the mental disorders adjusting for age group, gender and SES. A clear gradient appeared for any mental health disorder, with the lowest AOR in medium-to-high SES females (AOR: 1) followed by medium-to-high SES males (AOR: 1.13 [CI: 1.12–1.15]), low-SES females (AOR: 1.70 [CI: 1.65–1.76]) and low-SES males (AOR: 2.22 [CI: 2.15–2.29]), the association with SES being stronger than that with gender. The likelihood of psychosis was markedly higher in males of low SES than females of medium-to-high SES, with AORs across the age groups of 22.42 (CI: 19.29–26.05) (25–30 years), 7.08 (CI: 5.98–8.37) (19–24 years), 4.36 (CI: 3.61–5.27) (13–18 years) and 6.22 (CI: 5.06–7.65) (1–12 years). The AORs were also notably high in the case of conduct disorders (6.14 [CI: 5.33–7.07], 2.55 [CI: 2.32–2.80], 2.55 [CI: 2.35–2.76] and 2.91 [CI: 2.67–3.18]), drug use (3.90 [CI: 3.50–4.35], 2.02 [CI: 1.82–2.24] and 1.55 [CI: 1.26–1.91]) and ADHD (5.81 [4.50–7.52], 3.70 [CI: 3.31–4.13], 3.21 [2.92–3.54] and 3.14 [CI: 2.68–3.67] respectively). For internalizing clusters (anxiety and depression), the direction of the gender association was reversed, with a lower likelihood of these conditions in males (AORs of 0.58 [CI: 0.57–0.59] and 0.58 [CI: 0.54–0.61] respectively) (Table SM5).

Table 4 shows that the likelihood of any admission to a psychiatry ward followed a clear gradient in the older groups (19–24 and 25–30 years), with medium-to-high SES women at the lowest risk (AOR:1) and low-SES men the highest (AOR: 6.66 [CI: 5.25–8.44]); females were only more likely to be admitted in the 13–18-year-old group. The presence of mental illness was significantly associated with the likelihood of dying (AOR=2.38 [CI: 1.49–3.78]) as can be seen in Table 4 and the most striking result was that low SES was associated with a more than five-fold higher likelihood, the AOR being as high as 10.33 (CI: 4.43–24.07) in 25–30-year-olds. The association of death with gender was less clear, the only significant AOR being that of 19–24-year-old men (3.39 [CI: 1.12–10.31]).

To our knowledge, this is the first study analysing inequalities in mental health in children, adolescents and youth in a large general population registry simultaneously considering gender and SES. Besides a higher prevalence of mental disorders, low SES and male gender were associated with an excess mortality risk. Similarly, admission to a psychiatric ward showed inequalities by both SES and gender. These findings are discussed in detail below.

Our results confirm a gradient in the likelihood of diagnosed mental disorders when considering subgroups by gender and SES.7,10 The group with the poorest mental health is that of low-SES males whose likelihood of being diagnosed with substance use, conduct disorders or ADHD was three to four times higher than that of the reference group (medium-to-high SES females), and most strikingly, their likelihood of being diagnosed with psychosis was eight times higher. As expected given the literature,8,9 the prevalence of internalizing mental disorders (depression and anxiety) is higher in females, their risk of this type of diagnosis being nearly double that in males. The models are consistent in showing that the risk of all categories of mental disorder in lower SES groups is approximately double that in higher SES groups across all the ages studied.

Two mechanisms have been proposed to explain the association between SES and poorer health: social causality and health selection.7,10 The finding that inequalities appear at an early age strengthens the evidence for an association with low family income, which would support social causality.22 On the other hand, the alternative theory of selection for poor family health may also play a role, especially, in the age range covered by our study. Hoffman et al. reported that in the transition from childhood to working age, the two paths were similarly important, while in old age, causation is more important than selection.23 Although children's health does not determine family income, as Reiss pointed out, mental health problems among parents may underlie generational cycles of deprivation across generations.7 Further, child and adolescent resilience to stress is very dependent on interactions with their parents or legal guardians. As Oliver-Parra et al. highlight, not only SES but also parents’ mental disorders play a key role in the risk of mental disorders in children, enhancing the theory of selection by health.17 The complexity of psychological development makes it difficult to distribute weights between the two theories,7 the aetiology of mental problems involving a wide range of genetic and environmental determinants.24

The patterns of gender inequality described in our study are consistent with findings from both retrospective surveys11 and population registries.8,9,17 Specifically, in our study, the risk in young males was significantly greater for any mental disorder and externalizing disorders like conduct disorders and substance use, and much higher for psychosis and ADHD. In contrast, the risk of internalizing disorders like depression and anxiety in males was approximately half that in females. As in the aforementioned studies, the overall prevalence of any mental disorder in the population between 1 and 30 years of age is higher in males. This finding is explained by the fact that while the risk of anxiety and depression diagnoses in adolescents and young people is higher in females, the prevalence of other diagnoses (substance use, psychosis and conduct disorders) is much lower.8,9,11. In contrast, studies based on self-reported surveys of the general Spanish population indicate that mental health disorders are more frequent in females than in males.24,25 Women systematically have poorer perceived health in surveys, while young men are more frequently diagnosed with mental health problems by health services.8,9 The explanation lies in the use of questionnaires such as the General Health Questionnaire that are better at identifying internalizing disorders.26 Externalizing disorders are more visible because they directly affect interpersonal relations and behaviour, hampering personal relationships in the family, at school and in the community. In addition, ADHD is closely related to school performance. Moreover, in both genders, the social stigma associated with mental disorders hampers access to mental health services.27

Previously, excess mortality has been found in young people with mental disorders in Danish and Swedish registries but without taking into account SES.18,19,28 In addition to confirming the mortality excess, the novelty of our work is to have considered SES which has revealed its even greater impact, lower income being associated with a five-fold higher risk of death, making the social gradient clearly visible.10 We did not ascertain the causes of that excess in our population. An excess has been described throughout life in people with mental disorders, and 77.7% of this has been attributed to physical health problems, such as cardiovascular diseases (29.9%) and cancer (13.5%), and 13.9% to suicide.20

The existence of clear inequalities by SES suggests the need for early identification of mental health disorders and initiation of social, family and school interventions to improve social and community resilience, providing resources in society that buffer the negative effects of short-term risk factors.12,16 The health system can provide care for individuals when problems are diagnosed, but primary and secondary prevention requires other stakeholders to be active.12,16 Early treatment is crucial but does not avoid the stigma associated with mental disorder diagnoses and treatments.27 Stereotypes, prejudices and discrimination associated with stigma make the effects of mental illness go beyond its symptoms and associated disabilities.29 In this context, primary prevention targeting all adolescents would strengthen the resilience of adolescents and young people to environmental stress helping them to manage conflicts efficiently.12 Besides primary prevention, interventions for secondary prevention could be implemented for high-risk adolescents.30

Data routinely recorded from the EHR in childhood, adolescence and youth could help to stratify populations that might benefit from specific interventions.16 Given their higher absolute risk, primary prevention would especially benefit low-SES young people and adolescents and their families.31 Addressing social determinants is necessary to reduce mental health inequalities as the global burden of these disorders is unlikely to be alleviated simply by improving better access to mental health services.32 The United Nations in its 2030 Agenda for sustainable development has highlighted education as a key dimension in the goal of reducing health inequalities in Goal 4 (Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all). In addition, Objective 3 (Ensure healthy lives and promote well-being for all at all ages) makes explicit reference to the promotion of mental health.33,34

JM conceived and designed the research. IL, OI, and MMA obtained the data, interpreted the data and drafted the results section. AGP, CLH, AF and IIB reviewed and designed methods to classify mental disorders, performed the analyses and drafted the corresponding methods section. JM, IL, JA and EdM drafted the manuscript and approved the final manuscript. All authors revised the manuscript for important intellectual content and approved the final manuscript. All authors had full access to the full data in the study and accepted responsibility to submit for publication.

This study was performed in line with the principles of the Declaration of Helsinki and received approval from the Basque Ethical Committee.

This work is included within the UPRIGHT project, which is funded by the European Union Horizon 2020 Research and Innovation program under grant agreement No. 754919. This paper reflects only the views of the authors, and the European Union is not responsible for any use that may be made of the information it contains. The funding body has had no role in the study design, writing of the protocol or in the decision to submit the paper for publication.

The authors report no biomedical financial interests or potential conflicts of interest.