Article Text
PDF
Abstract
Background This study evaluates the influence of socioeconomic position (SEP) over the life course on change in health-related quality of life (HRQoL) in older adults.
Methods A prospective cohort of 2117 individuals aged 60 years and over. In 2001, SEP was measured over the life course as social class in childhood (approximated by father's occupation), as educational level completed and as adult social class (occupation of household head). HRQoL was measured with the SF-36 health questionnaire. Changes from 2001 to 2003 in the scores for the physical component summary (PCS) and the mental component summary (MCS) of the SF-36 were calculated, and individuals were classified into three categories: decline (decrease of >5 points), no change (change of −5 to +5 points) and improvement (increase of >5 points) in HRQoL.
Results After adjustment for baseline HRQoL, lifestyle, chronic illness, educational level and adult social class, low childhood social class was associated with a higher risk of both a decline and an improvement in the SF-36 PCS and MCS. The risk of decline in PCS and MCS and of improvement in MCS increased with the cumulative number of adverse SEP over the life course. Subjects who rose in social class from childhood to adulthood showed the greatest improvement on the SF-36 PCS and MCS.
Conclusion These results on the relation between SEP and changes in HRQoL in older adults support the three models proposed to explain health inequalities over the life course: the existence of critical periods, the accumulation of adverse SEP and social mobility.
- Health-related quality of life
- life-course epidemiology
- old age
- older adults
- quality of life ME
- SF-36
- social differences
- socioeconomic position
- Spain
Statistics from Altmetric.com
- Health-related quality of life
- life-course epidemiology
- old age
- older adults
- quality of life ME
- SF-36
- social differences
- socioeconomic position
- Spain
Health-related quality of life (HRQoL) represents the individual's perception of health status and its impact on different spheres of life. HRQoL is a strong predictor of morbidity, mortality and health services use.1–3 Moreover, change in HRQoL provides prognostic information in addition to a single baseline measure of HRQoL.4–6
Socioeconomic position (SEP) is associated with HRQoL and with changes in HRQoL. Several studies have observed that persons with more adverse SEP in childhood or adulthood have worse HRQoL than those with more advantageous SEP, and that the number of adverse SEP exposures over the life course shows an inverse linear relation with the physical and psychosocial dimensions of HRQoL.7–15 Also, one study has shown the contribution of social mobility to functional limitation, as individuals with low childhood SEP who then experienced upward mobility in adulthood had a lower frequency of functional limitations than those with stable low SEP in childhood and in adulthood.9
These investigations support the three models proposed within the conceptual framework of ‘life-course epidemiology’ to explain socioeconomic differences in health.16–18 The existence of an independent relation of SEP in childhood and adulthood with HRQoL suggests that there are specific periods in which SEP exerts its effect on HRQoL (the ‘critical period model’). The gradual relation between the accumulation of adverse SEP over the life course and HRQoL supports the ‘cumulative risk model’. Finally, the relation between change in SEP over the life course and HRQoL supports the ‘social mobility model’.
Two investigations within the Whitehall II Study found that persons in a low adult social class had the greatest decline in physical function and mental function after 3 years of follow-up.19 20 To date, however, the influence of SEP over the life course on change in HRQoL has not been evaluated. Accordingly, we assess this relation in older adults using the three models mentioned: critical periods, accumulation of exposure and social mobility. In particular, we evaluate whether change in HRQoL after a 2-year follow-up is related to SEP in different stages of the life course, the number of adverse SEP exposures over the life course and the change in SEP from childhood to adulthood.
Methods
Study design and participants
The study methods have been reported elsewhere.21 22 Briefly, this was a population-based cohort established in 2001 and followed up for 2 years. In 2001 the cohort comprised 4008 persons, representative of the non-institutionalised population aged 60 years and over in Spain. Information was collected in the household by interview and physical examination conducted by trained and certified personnel. Subjects were contacted again in 2003, and 3235 were reached successfully. The individuals contacted did not differ significantly from those lost to follow-up in any sociodemographic characteristic or lifestyle, except for the number of chronic diseases diagnosed and reported in 2001, which was 1.4 among those contacted and 1.2 among those lost to follow-up.21
Study variables
Main variables
Three indicators of SEP in different stages of the life course were collected in 2001 for each study participant: father's occupation, educational level attained and occupation of the household head. As the occupation at labour market entry was not available, we used education as an indicator of SEP at mid-life. Moreover, we used the occupation of the household head because 75% of women in this cohort had never had paid employment.
Whereas the father's occupation does not correspond to a specific period of the subjects' childhood, the occupation of the household head refers to the current or most recent occupation. The classification of occupations comprised 16 groups, which were classified into four categories of social class: I (professionals, managers, proprietors and clerical workers); II (self-employed farm workers); III (skilled and unskilled manual workers) and IV (paid farm workers). We lumped the four categories into two groups: categories I and II to represent high social class and categories III and IV to indicate low social class. This decision aimed to facilitate data analysis and to increase statistical power to test the study hypothesis. Finally, in the preindustrial society, self-employed farmers were landowners who worked long hours of heavy manual labour. In this study, we assigned self-employed farmers to the high social class because there is evidence that, during the first third of the 20th century, they had better living conditions than paid farm workers.23
Education is not a direct measure of position within the social class structure, but it is a mechanism to improve such a position.24 During the first half of the 20th century a large proportion of children left school without completing primary education, because of the scarcity of public educational resources or because of the need to get a paid job to contribute to the family income. In our cohort, education and family income are correlated, so that education is an indirect indicator of SEP in middle age.25
HRQoL was measured with the SF-36 health questionnaire, the generic instrument most widely used to assess HRQoL.26 The Spanish version of the SF-36 has previously been used to measure HRQoL in the elderly,27 and has shown good reproducibility and validity.28 It is made up of 36 items, which assess the following eight components or scales of HRQoL: physical functioning; role-physical and bodily pain, which reflect the physical component of health; social functioning, role-emotional and mental health, which cover the psychosocial aspects; and vitality and general health, which give an overall idea of subjective health, and are thus associated with both the physical and mental aspects. From these eight scales, we calculated the scores for the physical component summary (PCS) and the mental component summary (MCS) of the SF-36. Higher values in both scores indicate better health.29 For instance, in 2001 mean scores of PCS and MCS among individuals with no chronic diseases were 49.8 and 52.6, respectively. Among individuals with one, two or three diseases, mean PCS was 45.4, 40.4 and 36.8, respectively; corresponding values for MCS were 50.8, 47.1 and 42.4. There is evidence that a 5-point change in the scores of the SF-36 has clinical relevance.30
The outcome variable was change in HRQoL from 2001 to 2003, calculated as the difference in the PCS and MCS scores between the 2 years. We classified the individuals into three categories: decline (decrease of >5 points), no change (a decrease of 5 to an increase of 5 points) and improvement (increase of >5 points) in HRQoL.
Potential confounders
We collected data on sociodemographic variables, lifestyle, use of health services and chronic diseases that may act as confounders of the study relation because in our cohort,21 22 31–33 and in the literature,34 they have been shown to be associated with SEP and HRQoL. In particular, in 2001 we collected data on sex, age, social network (marital status and cohabitation), smoking, alcohol consumption, physical activity and central obesity, defined as waist circumference greater than 88 cm in women and greater than 102 cm in men. Also, we registered the frequency of consultations with the primary care physician and the following diseases diagnosed by a physician and reported by participants: ischaemic heart disease; stroke; chronic obstructive pulmonary disease; cancer at any site; diabetes mellitus; osteoarthritis and depression requiring treatment. We also obtained data on the incidence of the same diseases from 2001 to 2003.
Statistical analysis
Of the 3235 persons followed up until 2003, the following were excluded from the analysis: 45 for lacking data on the SF-36 in 2001; 229 who died between 2001 and 2003; 581 without the SF-36 data in 2003 and 263 for not reporting any SEP indicator. The analyses were thus conducted with 2117 persons. Compared with the whole cohort at baseline, individuals included in this analysis were younger, with higher education and were less frequently sedentary. Also, they more frequently had moderate and excessive alcohol consumption and abdominal obesity. We did not observe differences in the rest of the study variables.
The association between life-course SEP and improvement or decline in HRQoL was summarised by OR, and their corresponding 95% CI, obtained from polytomous logistic regression. We first tested the critical period hypothesis by examining the relation between each of the three SEP indicators and the change in HRQoL. We then evaluated the relation between the accumulation of adverse SEP over the life course and change in HRQoL. For this purpose we developed an index that sums the number of low SEP exposures over life, including low social class in childhood, low educational level and low social class in adulthood. The score on this index ranges from 0 to 3. Finally, we tested the social mobility hypothesis by creating a variable that combines childhood and adult social class. Educational level was not considered in this case because it reflects a socioeconomic dimension, which is different from social class. Although education is an indirect indicator of social class, it is also linked to cognitive function and other instrinsic characteristics of individuals. As such, education might contribute to explain the possible influence of social mobility on HRQoL change. However, to examine this issue was not an objective of this study.
To test each of the three hypotheses, we built two models. The first one adjusted only for sex, age and baseline HRQoL to control for regression to the mean in the change in HRQoL. The second model additionally adjusted for the remaining confounders. The models used to test the critical period hypothesis included all three SEP indicators simultaneously to examine whether the association of each one with change in HRQoL was independent of the other SEP indicators. When the social mobility hypothesis was tested, we also examined the interaction between social class in childhood and in adulthood (eg, whether the study association differed between subjects with low social class in childhood and high social class in adulthood and subjects with low social class in both life periods). For this purpose we used likelihood ratio tests, which compare models with and without interaction terms (products of the SEP categories in childhood and adulthood).
We used likelihood ratio χ2 tests and score χ2 tests to assess goodness of fit in all models. Also, we computed the squared correlation between the predicted and observed values to determine the predictive ability of the models.35 The results of all tests support the contention that the models in this manuscript represent a reasonably valid description of study data.
The analyses were performed with SAS, version 9.1 for Windows.36
Results
Tables 1 and 2 show the characteristics of study participants according to life-course SEP indicators. In general, baseline HRQoL was lower, lifestyles were less healthy and the use of health services was more frequent in individuals with lower SEP and less favourable socioeconomic trajectories. The exceptions were smoking and physical activity, which did not show a statistically significant association with the two indicators of social class or with social mobility. Most of the baseline diseases, as well as their incidence in 2001–3, also did not show an association with the SEP indicators and their life trajectories. All potential confounders were associated with the change in the SF-36 PCS or MCS, with the exception of age, cohabitation, ischaemic heart disease and cancer in 2001, and cancer incidence from 2001 to 2003 (see supplementary table available online only).
Characteristics of study subjects in 2001 according to three indicators of SEP over the life course
Characteristics of study subjects according to accumulation of adverse SEP over the life course and social class mobility from childhood to adulthood
Table 3 shows the association of childhood social class, educational level and social class in adulthood with change in HRQoL. After adjustment for all confounders, low social class in childhood was associated with a higher risk of decline in both PCS and MCS, as well as with an improvement in MCS. In addition, low educational level was associated with a greater risk of decline in MCS. Adult social class was not associated with change in HRQoL.
Association between social class in childhood, educational level and social class in adulthood and 2-year change in HRQoL, as assessed by the SF-36 physical summary and mental summary in older adults
In fully adjusted analyses, the risk of decline in SF-36 PCS and MCS increased linearly with the cumulative number of adverse SEP over the life course (table 4). In particular, compared with subjects who had never been in low SEP, those who were always in low SEP had an OR of 1.47 (95% CI 1.02 to 2.12) for a decline in PCS and an OR of 2.07 (95% CI 1.45 to 2.97) for a decline in MCS. On the other hand, even though cumulative adverse SEP did not show a clear dose–response relation with the likelihood of improved MCS, the probability was higher in those who were always in a low SEP (OR 1.64; 95% CI 1.08 to 2.48) (table 4).
Association between cumulative adverse SEP over the life course and 2-year change in HRQoL, as assessed by the SF-36 physical summary and mental summary in older adults
Table 5 shows that, compared to subjects who were always in high social class, those who remained in low social class, and those who rose in social class, showed a greater risk of decline in SF-36 PCS and MCS. In fully adjusted analyses, the association with improved PCS showed evidence of interaction between social class in childhood and adulthood (p=0.035). Those who rose in social class were more likely to improve on PCS (OR 1.85; 95% CI 1.18 to 2.90) and MCS (OR 1.65; 95% CI 1.11 to 2.44).
Association between social class mobility from childhood to adulthood and 2-year change in HRQoL, as assessed by the SF-36 physical summary and mental summary in older adults
Discussion
In our study, childhood social class was associated with a decline in the physical and mental components of HRQoL and with an improvement in the mental component. Also, the number of adverse SEP exposures over the life course was associated with a reduction of the physical component of HRQoL, and with both a reduction and a gain in the mental component. Finally, individuals who rose in social class from childhood to adulthood showed an improvement in both components of HRQoL.
Our results on the relation between SEP and decline in HRQoL supports the models of critical periods and of an accumulation of adverse socioeconomic circumstances. Furthermore, the improvement in the physical component of HRQoL is compatible with the social mobility model, and the improvement in the mental component is compatible with all three models proposed to explain health disparities over the life course.
Most studies have observed that adulthood represents a critical period for the action of socioeconomic circumstances on HRQoL. In particular, adults with lower educational level or social class have poorer physical and mental function.7 8 13–15 Likewise, in some studies socioeconomic circumstances in adulthood explain the association between childhood SEP and HRQoL, because this association decreases or disappears after adjusting for SEP in adulthood.8 13 Furthermore, in the Whitehall Study adults with lower SEP showed a greater decline in physical and mental function after 3 years of follow-up.19 20
In our study, social class in childhood showed an independent relation with the decline in the physical and mental components of HRQoL. However, we did not find a relation between adult social class and a decline in SF-36 PCS and MCS, nor was educational level associated with decreased PCS. These findings suggest that in older Spaniards early life is a critical period for the decline in HRQoL associated with low SEP. In this cohort, the father's social class is strongly related to educational level and to social class in adulthood; thus, social class in adulthood possibly reflects the influence of a number of social circumstances over the life course.
Previous studies have reported that the accumulation of adverse socioeconomic exposures is associated with worse physical and mental function in adults.10–15 In our study, the risk of a decline in the physical and mental components of HRQoL increased with the number of adverse SEP exposures over the life course, both when using the three SEP indicators and when using only social class in childhood and in adulthood. This finding is relevant for two reasons: first, because lifestyles and chronic morbidity do not explain this relation; and second, because the relation is observed for indicators of SEP, such as educational level or social class in adulthood, which are not associated with a HRQoL decline. This is important within the theory of the life course influence of SEP on health, because a health indicator may exert a contribution to the effect of other indicators despite lacking an independent association with health.
We have not found any studies on the relation between SEP and improvement in HRQoL. However, some investigations have reported that low SEP may be associated with recovery from disability.37 38 In our study, low social class in childhood and cumulative adverse SEP were associated with improvements on the mental components of HRQoL, but not the physical components. This may partly be due to a certain regression to the mean, given the very low baseline SF-36 MCS score of individuals with lower SEP.
Another notable finding was the improvement in HRQoL among those who rose in social class over the life course. These persons, and those who were in high social class in both childhood and adulthood, had the highest PCS and MCS scores in 2001. It is possible that some persons who rose in social class over the life course use persons from their original social class as the reference when evaluating their health, so that the difference between the two groups would increase over time.
Finally, this study has some limitations. The data refer only to persons who completed the information on HRQoL in 2001 and 2003. Subjects who completed the SF-36 in 2003 had better HRQoL in 2001 and higher education than those who did not complete the questionnaire. This may have led to an underestimation of the observed associations. Another limitation is the possible misclassification of the father's occupation, because it is based on individual recall and may vary over time. However, it is not likely that it has produced a substantial bias in study results because the father's occupation was grouped into broad categories. Also the information on HRQoL was obtained in face-to-face interviews in 2001 and by telephone in 2003. However, results of administering the SF-36 by telephone are equivalent to and of similar validity to those obtained in face-to-face interviews.39 Finally, morbidity, which was used as an adjustment variable in the models, was also self-reported. Nevertheless, there is a high level of agreement between self-reported diseases and medical records in the elderly.40
In conclusion, our findings on the relation between SEP and change in HRQoL in older adults are compatible with all three models proposed to explain health inequalities over the life course: the existence of critical periods for the effect of SEP, the accumulation of adverse SEP and social mobility.
What is already known on this subject
Persons with more adverse SEP in childhood or adulthood have worse HRQoL than those with more advantageous SEP. The number of adverse SEP exposures over the life course shows an inverse linear relation with the physical and psychosocial dimensions of HRQoL.
What this study adds
The relation between SEP and change in HRQoL in older adults is compatible with all three models proposed to explain health inequalities over the life course: the existence of critical periods for the effect of SEP, the accumulation of adverse SEP and social mobility.
References
Footnotes
Contributors AOR and LMLM contributed equally to this paper.
Funding This study was funded by FIS grant 09/01626.
Competing interests None declared.
Ethics approval This study was conducted with the approval of the Clinical Research Ethics Committee of the ‘La Paz’ University Hospital in Madrid, Spain.
Provenance and peer review Not commissioned; externally peer reviewed.