Article Text
PDF
Abstract
Background The purpose of this work was to evaluate the association between socioeconomic circumstances in early life and poor self-rated health in adulthood, and to determine whether this association varies depending on birth cohort.
Methods Using data from the 2005 Living Conditions Survey carried out in a Spanish population sample, separate analyses were made for persons born between 1941 and 1959 who lived their childhood and adolescence in a period marked by absence of economic growth in Spain, and those born between 1960 and 1980. The authors estimated of the relation of financial difficulties in adolescence, occupation of father, education of father and education of mother with poor self-reported health, adjusted for the measures of socioeconomic position in adulthood. Also evaluated was the possibility of interaction between early and adult socioeconomic position and the cumulative effect of low socioeconomic position across the life course.
Results In the two cohorts, the presence of financial difficulties in adolescence showed an association with an increased risk of poor self-rated health; in contrast, the relation of occupation and education of parents with poor self-rated health varied depending on the birth cohort and gender of the study subjects. The effect of family financial situation was shown to accumulate over the life course, whereas the effect of socioeconomic position across the life course was heterogeneous when occupation and education of parents were used.
Conclusion The importance of financial difficulties in the household has probably not varied over time; in contrast, the heterogeneity of the findings regarding occupation and education of parents suggests that the importance of these indicators and, consequently, their aetiological pathways may differ depending on the gender and birth cohort of the study subjects.
- Father's education
- father's occupation
- financial difficulties in early life
- mother's education
- self-rated health
- social epidemiology
- social factors
- social mobility
- socioeconomic
Statistics from Altmetric.com
- Father's education
- father's occupation
- financial difficulties in early life
- mother's education
- self-rated health
- social epidemiology
- social factors
- social mobility
- socioeconomic
Introduction
Many studies in developed countries have evaluated the relation between socioeconomic circumstances in early life and health within the framework of life course epidemiology, which focuses mainly on the long-term effect of childhood and adolescent social factors on later disease.1 Almost all these studies have shown that, independently of socioeconomic position (SEP) in adulthood, low SEP in early life is associated with a higher risk of mortality2 3 and greater frequency of poor physical function4–6 and poor self-perceived health in adulthood.5–13
Gilman14 and Blane15 have criticised the lack of theoretical conceptualisation of the measures of SEP in childhood in these studies. As Blane himself16 observed with the measures of SEP in adulthood, the two authors have observed the interchangeable use of SEP measures in the absence of hypotheses about which aspect of each measure of SEP involves a risk to health in adulthood. This use limits the possibility of identifying possible aetiological pathways.14 15 17 For example, household income and financial difficulties during childhood are the most appropriate measures of material conditions during childhood, whereas occupation of parents represents an indirect indicator of material conditions. Education of parents is also an intellectual resource of the household related to education of the children, either directly through its influence on cognitive ability or indirectly by access to quality education, since parents with higher educational levels are likely to have better jobs, higher incomes, and greater ability to finance higher levels of education for their children.18
Although the significance of social indicators changes over time, there has been little suggestion in these investigations that the impact of SEP in childhood may differ depending on birth cohort.18 19 Birth cohorts may be differentially affected by environmental change during childhood20 (eg, improvement or deterioration of living standards), or by greater or lesser opportunities for professional training and education of parents. Consequently, the relation of SEP in early life with adult health may vary from one birth cohort to another.
Children in the lowest socioeconomic groups born in more recent decades have experienced better standards of living than those born in the first half of the 20th century. For this reason, Lawlor et al have noted that if the associations observed in previous studies largely represent the effects of extreme adverse circumstances, then one might expect a weaker, or no, association between occupation of parents and health outcomes in studies of populations born more recently.21 And the opposite could occur with education of parents. Unlike parents of those born in the latter half of the 20th century, most parents in earlier generations did not have access to formal education. Completion of high school was not then necessary for many trade and professional positions; therefore, parental education in those generations would be expected to have a weaker, or no, association with health outcomes given that education was not a strong predictor of income.
There is also evidence of the influence of SEP of parents on educational opportunity and attainment of their children; in turn, educational attainment is a powerful predictor of adult income and occupation. The strength of these relationships is context specific and varies in men and women.1 These relationships may also be different depending on the birth cohort; therefore, the model that explains the relation between SEP in early life and adult health will also vary from one birth cohort to another. Different findings show that social accumulation (continuation of early life social conditions) and social mobility (movement up or down the social hierarchy) across the life course affect adult health, and that both processes can operate simultaneously.22 23 However, the importance of social mobility could be a characteristic of younger cohorts, since they have enjoyed greater educational opportunities; therefore, downward social mobility due to poor health is more likely in younger than in older cohorts.
The purpose of this work was to evaluate the association between socioeconomic circumstances in early life and poor self-reported health in adulthood, and to determine if this association varies depending on the birth cohort. For this purpose, we used various measures of SEP in adolescence in two cohorts of Spanish citizens born during the 20th century.
Methods
Study population
Participants were drawn from the 2005 wave of the Living Conditions Survey. In this survey respondents were selected by stratified multistage sampling of the non-institutionalised population residing in Spain. The first stage units—census sections—were grouped into strata by size of the municipal population and were then selected with a probability proportional to the size of the population of the stratum. The second stage units—households—were selected within each census section with the same probability by systematic sampling with random start. Within each household, all persons aged 16 years or older were selected to complete the questionnaire. In the 2005 wave, 16 383 subjects were selected.
The Living Conditions Survey includes information on different socioeconomic characteristics of respondents. In 2005 information was also collected on the socioeconomic situation in adolescence of respondents who were born between 1941 and 1980. The reference period used to define adolescence was the time when the respondent was between 12 and 16 years of age.
Spain suffered a period of economic stagnation between the end of the Civil War in 1936–39 and the late 1950s; no economic growth was observed until well into the 1960s.24–26 As a result of different economic policies to liberalise the economy in the early 1960s, Spain experienced an unprecedented period of industrialisation, to the extent that economic growth rates between 1960 and 1975 were among the highest in the world.27 For this reason the analyses were made in two birth cohorts: on the one hand, the 4547 participants born between 1941 and 1959, whose childhood and adolescence were marked by low standards of living, and on the other hand, the 6403 participants born between 1960 and 1980.
The analyses were made separately by sex, given that some studies have shown that the effect of SEP in early life on poor self-rated health in adulthood may differ in men and women.11 12
Measures
The measures of SEP in adolescence were financial difficulties as an indicator of material household resources, occupation of father as an indicator of earnings and economic conditions, and education of father and mother as an indicator of intellectual household resources. Occupation of mother was not used because two-thirds of the mothers were housewives. The non-response rates for financial difficulties, occupation of father, education of father and education of mother were 2%, 7%, 6% and 4%, respectively. These measures of SEP during adolescence and the measures of SEP at the time of interview are shown in table 1. Spearman's correlation coefficient between the measures of SEP during adolescence and the measures of SEP in adulthood ranged between 0.25 and 0.30.
Definition of the indicators of socioeconomic position used in this study
The indicator of health status was poor self-rated health, which was measured with the following question: ‘Would you say your overall health is very good, good, fair, poor or very poor?’ A respondent was considered to have poor health when the response was fair, poor or very poor.
Statistical analysis
We first calculated the percentage of respondents who reported poor self-rated health according to each of the SEP measures in adolescence. We then calculated the association between the measures of SEP in adolescence and poor self-rated health using the percentage ratio estimated by binomial regression. We used the percentage ratio rather than the OR because the OR overestimates the association when the frequency of the dependent variable is higher than 0.20.28 29 Moreover, this overestimate would be greater in participants born between 1941 and 1959, since they have the highest percentage of poor self-rated health.
To determine whether the measures of SEP in adolescence were associated with poor self-rated health independently of SEP in adulthood, we first estimated the percentage ratio adjusted for age, and then the percentage ratio adjusted for age and the measures of SEP in adulthood.
To test the social mobility hypothesis, we also evaluated whether the association between SEP in adolescence and poor self-rated health differed depending on SEP in adulthood. For this purpose we used the change of deviance to assess the possibility of interaction between SEP in adolescence and SEP in adulthood, by comparing the model without interaction with a model containing the appropriate interaction term. For cases in which no interaction was observed, the social mobility hypothesis was ruled out, and we then evaluated the cumulative effect of having low SEP across the life course. The evaluation of interaction was made using the same SEP measure across the life course and adjusting for the rest of the SEP measures in adulthood. The analyses were made using SAS 9.1 for Windows (SAS, Cary, North Carolina, USA).
Results
Table 2 shows the number of subjects interviewed and the percentage of poor self-reported health by the measures of SEP in adolescence. The percentage of poor self-reported health was higher in the cohort born between 1941 and 1959 than in the cohort born between 1960 and 1980. For all measures of SEP this percentage showed an increasing gradient from the category representing the least adverse position to the one representing the most adverse position.
Percentage of poor self-rated health according to measures of socioeconomic position in adolescence
The relation between the measures of SEP in adolescence and poor self-reported health can be seen in table 3. The presence of financial problems in the household was the only measure that showed a relation with poor self-reported health in all the population groups analysed; specifically, an increased risk of poor self-reported health. Occupation of father was independently related with poor self-reported health in men born before between 1941 and 1959: those whose fathers were manual workers had a 1.21 (95% CI 1.03 to 1.43) greater risk than those whose fathers were non-manual workers. After adjusting for the measures of SEP in adulthood, education of father was not independently related with poor self-rated health in men, but it was in women, although the direction of the relation was opposite in the two cohorts of women analysed. Low educational level of the father was associated with a decreased risk of poor self-rated health in women born before 1960 (0.84, 95% CI 0.72 to 0.99), but with an increased risk of poor self-rated health in women born afterwards (1.35, 95% CI 1.08 to 1.71). Education of mother showed a stronger association with poor self-rated health in women than in men in both cohorts, but the relation disappeared after adjusting for the measures of SEP in adulthood.
Association between measures of socioeconomic position and poor self-rated health
A sensitivity analysis was performed to test whether a different cut-off point for self-reported health would change the results. This was done by excluding the category of ‘fair’ from poor self-rated health. In the 1941–1959 birth cohort, the age-adjusted percentage ratios were higher, but after adjusting for the measures of SEP in adulthood, the magnitude of the association was similar to that observed in table 3. In the 1960–1980 birth cohort, the models did not converge, since only 3% of subjects responded that their health was ‘poor’ or ‘very poor’.
In men and women born between 1941 and 1959, no evidence of interaction was found between the measures of SEP in adolescence and in adulthood (table 4). The effect of family financial situation was seen to be cumulative across childhood and adulthood. The effect of having a manual versus non-manual occupation in men in this cohort was also found to be cumulative across both stages of the life course, and the effect of low educational level in women was cumulative across the life course when considering the education of the mother and respondent.
Associations between measures of socioeconomic position throughout the life course and poor self-rated health adjusted for age and other measures of socioeconomic position in adulthood (birth cohort 1941–1959)
As in the cohort born between 1941 and 1959, the effect of family financial situation was cumulative across childhood and adulthood in men and women born between 1960 and 1980 (table 5). Men who belonged to a non-manual class in adolescence but who moved down to a manual class in adulthood had a higher risk of poor self-rated health than men who had a manual occupation at both stages, although no interaction was seen between the two measures of occupation. Interaction was detected, however, between educational level of parents and respondent: subjects who had reached only a low educational level but whose parents had a high educational level showed a greater risk of poor self-rated health than subjects with low educational level whose parents also had low educational level.
Associations between measures of socioeconomic position throughout the life course and poor self-rated health adjusted for age and other measures of socioeconomic position in adulthood (birth cohort 1960–1980)
Discussion
Financial difficulties in adolescence are related with poor self-rated health independently of SEP in adulthood; in contrast, the relation of occupation of father and education of parents with poor self-rated health varies depending on the birth cohort and gender of the study subjects. Furthermore, the effect of family financial situation is cumulative over the life course, whereas the effect of SEP across the life course is heterogeneous when other measures of SEP are analysed.
Study strengths and limitations
Although this was a cross-sectional study, the possibility of reverse causality as an explanation for the association between the indicators of SEP in adolescence and poor self-rated health is implausible since these indicators refer to the parents. A strength of the study is the availability of measures of SEP in adolescence and in adulthood, as these made it possible to evaluate the effect of social accumulation and social mobility on adult health. The percentage of non response to the measures of SEP was low; because respondents with low SEP are much more likely not to respond, a low non-response rate is less likely to affect the estimates.
Given that the data are cross-sectional, it may be impossible to disentangle age effects from birth cohort effects. The age-adjusted percentage ratios for poor self-reported health by SEP in adolescence are in some measure a reflection of SEP in adulthood; this may explain why they are higher in the 1960–1980 birth cohort since these subjects are younger and, as is well known, the relation between one's own SEP and health decreases with age. Nevertheless, when we adjusted for the measures of SEP in adulthood, an independent effect of the measures of SEP in adolescence was observed and, in many cases, this effect was similar in the two cohorts or even greater in the cohort born between 1941 and 1959.
Another limitation could be a possible bias in the classification of occupation of father, education of mother and father, and financial difficulties in childhood, since these variables are based on individual recall. Any misclassification would most probably be non-differential, given that it is unlikely that recall inaccuracy of SEP in adolescence would be affected by self-rated health status, and thus would underestimate the association. However, the probability of this bias is small since these variables were grouped into broad categories.
An explanation for part of the heterogeneity of the findings could be that respondents incorporate different health dimensions in evaluating their own health and assign greater or lesser weight to one or another dimension depending on SEP, gender and age.30 31 However, the similarity of the findings obtained with regard to financial difficulties in all the population groups studied suggests that the heterogeneity of the findings for the rest of the measures of SEP in adolescence is probably due to the different significance of these measures in the different population groups.
Comparison with previous studies
Several studies conducted in other countries based on information collected around 2000 have also found a relation between financial difficulties in childhood and poor self-rated health in adulthood in different age groups,5 11 although the relation has sometimes been observed in men only12 and other times in women only.11 A study in an Australian population sample of persons aged 16 years and over also found a cumulative effect of family financial situation across childhood and adulthood.13
On the other hand, studies in populations in Great Britain8 and Holland7 9 have found a relation between occupational class of father and poor self-rated health in adulthood. Likewise, Power et al8 observed a cumulative effect on poor self-rated health of lifetime socioeconomic conditions based on occupation. However, the results of other investigations in which various populations were analysed in a single study have not been consistent.6 10
The findings of studies on the influence of education of parents on poor self-rated health in adulthood have also been inconsistent. Two studies that used education of the parent with the highest educational level found opposite results: one, in a sample of US adults aged 50 years or over found an independent relation between low parental education and poor self-rated health,5 while another study in a sample of Finnish adults of between 40 and 60 years of age did not find this relation.11 Another study in Finland, in subjects aged 18–39 years, found that education of the father was not related with poor self-rated health; conversely, education of the mother was related with poor self-rated health in women but not in men.12
Interpretation of findings
The cohort of subjects born between 1941 and 1959 lived their adolescence during a period of economic stagnation in Spain lasting from the end of the Civil War in 1939 to the late 1950s: 53% of subjects in this cohort reported that financial problems in the home were experienced frequently or almost always during their adolescence, whereas this percentage was 36% in the cohort born between 1960 and 1980. In any event, despite improved material living standards in subjects in the younger cohort, our findings on the relation between financial difficulties and poor self-rated health support the theory that material living standards are important for health throughout the life course.
Only in men born between 1949 and 1959 did we observe an independent association between occupational class of the father and poor self-rated health and a cumulative effect of being in a manual versus non-manual occupational class across both stages of the life course. Occupation is an indicator of earnings and economic circumstances, and the relation of occupation of father with health in adulthood has been attributed to the effect on health of material resources in early life.32 33 Occupation is an indirect indicator of material resources; therefore, it may be that the effect of occupation of father is shown only in the population cohort whose adolescence coincided with the more economically adverse period. However, this fails to explain why the relation was not found in women in this birth cohort. Moreover, the fact that financial difficulties in childhood, a direct indicator of material circumstances, showed a relation with poor self-rated health in all four population groups analysed raises important doubts about the mechanism by which occupation of father would be related with poor self-rated health in men born before 1960.
Regarding education of parents, the findings of our study suggest that this indicator of SEP during adolescence is important for health in women, but not in men. In both cohorts of women, education of father showed an independent relation with poor self-rated health, although the direction of the relation was different in each cohort. Likewise, low education of the mother was strongly associated with poor self-rated health in both cohorts of women. The association disappeared after adjusting for the other measures of SEP, although it almost reached statistical significance in the cohort born between 1941 and 1959 (1.32, 95% CI 0.99 to 1.77). Furthermore, in this cohort of women the effect of low education of the mother and/or the respondent on poor self-rated health was cumulative.
The importance of parental education for health of women has been observed in two studies in the Finnish population. One is the aforementioned study in young adults13; the other study, in a middle-aged population, found that low parental education was associated with physical functioning in women, but not in men, although the association disappeared after adjusting for women's own education.34 It remains an open question why the intellectual and cognitive resource represented by parental education is important for health in women.
The findings on socioeconomic life course based on education suggest the existence of health-related social mobility in a small number of men and women born between 1960 and 1980. Persons in this cohort enjoyed more opportunities for professional training and education during their adolescence than those in the cohort born before 1960, since compulsory education up to age 14 years was not established in Spain until 1970.35 Accordingly, the greater risk of poor self-rated health in subjects with low educational level whose parents had high educational level may be due to childhood health problems that prevented them from benefiting from these opportunities during their adolescence and young adulthood.
Thus, the impact of social mobility might also be a selection effect, and those who had social mobility are likely to be a selection of those who experienced health problems or other factors related to health. This may also be the reason for the high risk of poor self-rated health in men in this cohort who had a manual occupation but whose fathers had a non-manual occupation, since in this cohort a minimum level of professional education and training is essential for any career or employment.
In summary, the consistency of the findings with regard to financial difficulties suggests that the importance of this indicator has not varied over time; in contrast, the heterogeneity of the findings on occupation of father and parental education suggests that its importance and, consequently, the aetiological pathways to the health outcome may differ depending on the birth cohort and gender of the study subjects.
What is already known on this subject
Children in the lowest socioeconomic groups born in more recent decades have experienced better standards of living than those born in the first half of the 20th century. One might expect a weaker, or no, association between occupation of parents and health outcomes in populations born more recently.
Most parents of persons born in the first half of the 20th century did not have access to formal education. In those generations one might expect a weaker, or no, association between education of parents and health outcomes, given that education was not necessary for many trade and professional positions.
What this study adds
In two cohorts of Spanish people—those born between 1941 and 1959 and between 1960 and 1980— financial difficulties in adolescence are related with poor self-rated health independently of socioeconomic position (SEP) in adulthood. In contrast, the relation of occupation of father and education of parents with poor self-rated health varies depending on the birth cohort and gender of the study subjects.
The effect of family financial situation on poor self-rated health is cumulative over the life course, whereas the effect of SEP position across the life course when other measures of SEP are analysed is heterogeneous.
References
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.