Socioeconomic pathways to inequalities in mental and functional health: a comparative study of three birth cohorts

Socioeconomic inequalities in health (SEIH) have been remarkably resistant to policies and interventions in most industrialised societies. Recent time trend studies show stability or even an increase in absolute and/or relative socioeconomic inequalities in several health outcomes [1-4]. At the same time, significant changes in social stratification have been observed [1]. Throughout the twentieth century, access to education has improved substantially, resulting in an expansion of the number of higher educated individuals. Education is strongly related to other socioeconomic resources, such as higher skilled occupations and higher income, and both factors also have effects on health [2]. In fact, educational inequalities in health can be partly explained by the associations with income and occupation [3]. The fact that relatively many people achieved a high educational level in recent birth cohorts may have changed the way in which lower and higher educated individuals compete for jobs and gain access to higher wages [4]. This raises the question whether these changes have implications for health inequalities; specifically, whether the direct effects and indirect effects of education (via occupation and income) on health have changed over time. Therefore, this study examines whether the strength of indirect effects from education via occupational level and income on physical and mental health, have changed between three subsequent 10-year birth cohorts. We also examine to what extent potential changes in these indirect effects are due to changes in the direct effects of education on occupation and income and to changes in the direct effects of the socioeconomic indicators on health.

Fundamental cause theory (FCT) states that socioeconomic inequalities persist across time and place because social conditions are ‘fundamental causes’ of health and disease [5]. This means that the precise mechanisms and proximal risk factors that link socioeconomic inequalities to health may change over time, but those with a higher socioeconomic position consistently have better health outcomes because they have access to ‘flexible resources’ that promote health, such as money, knowledge, prestige and power [6].

Education is regarded as one of the most ‘fundamental’ among social causes of disease, because it is usually the first individual socioeconomic characteristic obtained in life, which subsequently provides access to most of the ‘flexible resources’ stipulated by FCT [7]. For example, higher educated individuals have access to less physically demanding and higher status jobs, with better financial resources, higher autonomy and more opportunities for further development [2]. Furthermore, the relationship between education and health is substantially mediated by occupation and income [3]. In addition, jobs with a higher occupational level typically provide more income [4]. As such, the relationship between education and income can also be assumed to be mediated through occupation. Therefore, education can be seen as a ‘gateway’ to further socioeconomic rewards (i.e. occupation and income) and – for a large part via these pathways – to better health [8]. Based on FCT, we expect that educational inequalities are persistent across time because a high level of education continues to provide access to flexible resources that ultimately improve health. However, several societal changes across the past decades suggest that the ways in which education operates as a gatekeeper to further socioeconomic resources and better health may have changed.

Exactly how they have changed remains uncertain and different theories seem to suggest different changes. Modernization theory argues that education grants access to human capital [9]. Individuals invest in schooling because education enhances their productivity and skills, which are rewarded by employers in the labour market [10]. It is assumed that each person with the same amount of skills is rewarded equally and that educational returns automatically increase when there is more demand than supply. Modernization theory thus states that educational attainment has become more important in modern economies as the demand for higher educated individuals has increased. Many jobs are more technologically advanced and industries demand more productivity and skills [11]. At least in western industrialized countries, the percentage of high skilled jobs has expanded over time, and these jobs are associated with higher income. If this theory is true, the effects of education on occupational skill level and income may have increased, leading to stronger indirect effects of education on health, via occupation and income.

In contrast, the positional model of education presumes that there has been an inflation of educational credentials that decreases the occupational [12] and economic returns [13] from education. The model argues that the employment structure has failed to keep up with the educational expansion [14]. Therefore, there are not enough high-skilled jobs available for the growing number of highly educated individuals. As a consequence, highly educated individuals are increasingly forced to compete for low skilled jobs. Moreover, as education is seen as a positional good, employers generally favour higher educated individuals over lower educated individuals, even if their specific skills for the job are identical [15]. Furthermore, because more high skilled jobs are available, the financial rewards offered for these jobs may decrease. Following this model, education can no longer guarantee high occupational and economic returns, and this development may neutralize some of the assumed advantage of higher educational qualifications. So, the positional model would suggest that the indirect effects of education on health via occupation and income are weaker in recent generations.

The Longitudinal Aging Study Amsterdam (LASA) is an ongoing longitudinal, multidisciplinary study, that focuses on physical, cognitive, social, and emotional functioning of Dutch older adults [19, 20]. Respondents were randomly selected from the population registers of 11 urban and rural municipalities in the west, northeast and south of the Netherlands. Trained interviewers visited respondents at home and data were collected by means of personal interviews that lasted approximately 2 hours. We selected the baseline measurement and two refresher measurements for inclusion in the study. The baseline measurement was collected in 1992 and included 3107 respondents aged 55–84 years old (cooperation rate was 62%). The first refresher cohort was collected in 2002–2003 and included 1002 adults aged between 55 and 64 years (cooperation rate was 62%). The third refresher cohort was collected in 2012–2013 and consisted of 1023 adults also aged between 55 and 64 years (cooperation rate was 63%). Given our focus on comparing independent groups of adults of the same age but observed at different moments in time (cross-sectional), we selected only those individuals aged between 55 and 64 years from the baseline measurement, which yielded 988 respondents. We, henceforward, refer to the cohorts by birth year which are 1928/37, 1938/47 and 1948/57 respectively. Included were 472 men and 516 women in the 1928/37-cohort, 475 men and 527 women in the 1938/47-cohort, and 496 men and 526 women in the 1948/57-cohort.

This study investigated differences between three birth cohorts of young-old adults born between 1928–1937, 1938–1947 and 1948–1957 in the mediating role of occupation and income in the education–health relationship. Results suggest that socioeconomic health inequalities have persisted but that the associations between the socioeconomic indicators producing them has changed between the cohorts.

Absolute and relative educational inequalities in functional limitations and depressive symptoms persisted or increased in later birth cohorts. However, changes in absolute inequalities were more pronounced than in relative inequalities. These findings are largely in line with prior studies [27‐30]. Following FCT, we argue that education, in our study, indeed functions as a fundamental cause of disease as it remained stable across cohorts. We might infer that flexible resources, such as power, prestige and financial rewards, are, as such, continuously accessed through educational level.

We observed changes in two socioeconomic pathways to health. The indirect effect of education on functional limitations via occupation increased from the 1928/37-cohort to the 1948/57-cohort. The indirect effect of education on functional limitations via income decreased in later cohorts. In the introduction we argued that modernization theory and the positional model lead to opposing expectations; while modernization theory led us to expect that the effects of education on occupation and income increased, the positional model suggests that these effects would have remained stable or decreased. Our findings were partly consistent with both theories. Those with higher education became more likely to obtain a higher skilled job in more recent generations, supporting the modernization theory; it suggests that high-level occupational opportunities kept up with the educational expansion that occurred in the Netherlands around the 1950s [17]. As a consequence, education has become a more important determinant of people’s job skill level. However, the fact that education’s effect on income, after accounting for mediation, decreased across cohorts suggests that highly educated individuals only gain economic advantage over their lower educated peers when they also obtain a high level job. In earlier cohorts, the larger direct effect of education on income suggests that educational qualifications could, independently from occupational position, result in higher incomes, for example via on-the-job training or gradually increasing wages through the accumulation of experience in the same type of job. In more recent cohorts, it appears that such options may have become less accessible. The positional model suggests that there is credential inflation in the chances to obtain a job and in economic rewards [15]. While we found that the occupational structure seemingly kept up with the educational expansion, the economic returns from education in the absence of obtaining a higher skilled job decreased [13]. In addition, this decrease occurred more strongly for men than for women. As such, credential inflation may have occurred for economic returns from educational level but not through occupation.

In general, we found larger and more consistently increasing inequalities in functional limitations compared to depressive symptoms. This underscores the importance of examining multiple types of health outcomes in research on socioeconomic inequalities. One explanation for the differences may lie in differences in risk factors for these health outcomes. It has been shown that functional limitations are more strongly affected by health behaviours and depressive symptoms more strongly by factors such as stress exposure and social support [31]. Furthermore inequalities in health behaviours such as smoking [32] and alcohol use [33] increased in the past decades, whereas this is less so for psychosocial factors [34].

We found that the magnitude of relative SEIH persisted and even increased in recent generations of 55–64 year old men and women. While the role of education as gatekeeper in health inequalities remains, its pathway through socioeconomic resources changed across cohorts. The pathway in which education determines health through income has partly been replaced by a pathway in which education determines health through occupational skill level. Apparently, it has become more difficult for lower educated individuals to obtain a higher skilled job, which increasingly and negatively affects their health in recent generations. An important takeaway for policy and intervention is to strengthen the occupational position of lower educated individuals through offering more opportunities for on-the-job training.