Abstract
Objectives. To investigate the effect of social mobility and to assess the use of socioeconomic indicators in monitoring health inequities over time, we examined the association of self-rated health with socioeconomic position over the life course.
Methods. Data came from a cross-sectional telephone survey (n = 2999) that included life-course socioeconomic indicators and from a chronic disease and risk factor surveillance system (n = 26 400). Social mobility variables, each with 4 possible intergenerational trajectories, were constructed from family financial situation and housing tenure during childhood and adulthood.
Results. Low socioeconomic position during both childhood and adulthood and improved financial situation in adulthood were associated with a reduced prevalence of excellent or very good health. Trends over time indicated that socioeconomic disadvantage in adulthood was associated with poorer self-rated health.
Conclusions. Our results support policies aiming to improve family financial situation during childhood and housing tenure across the life course. Inclusion of life-course socioeconomic measures in surveillance systems would enable monitoring of health inequities trends among socially mobile groups.
Monitoring and surveillance form an important part of the international health inequities agenda.1 It is necessary to monitor health inequities in terms of socioeconomic position (SEP), gender, ethnicity, and other indicators to determine whether they are widening or decreasing over time2 and to design and evaluate policies aimed at reducing these inequities. Researchers have noted the paucity of data for monitoring inequities over time and called for the focus of such monitoring to be broadened from mortality to include health, morbidity, and risk indicators.2,3
Routine data from population health surveys and surveillance systems can be used to monitor epidemiological changes in health inequities. The US Behavioral Risk Factor Surveillance System is known as one of the most extensive population survey systems for monitoring morbidity and risk factors among different population groups over time.4–6 Other systems for monitoring health include the Demographic Surveillance Systems in 19 countries across Africa, Asia, Oceania, and Central America7; the Multiple Indicator Cluster Surveys across 64 countries8; the Demographic and Health Surveys typically conducted every 5 years in more than 75 countries9; and the Netherlands Permanent Research on Living Conditions annual surveys.2 In Australia, the South Australian Monitoring and Surveillance System (SAMSS)10 is an example of a statewide system that regularly and frequently collects, analyses, interprets, and disseminates data.11,12
Many surveillance systems use measures of current SEP to monitor inequities in health. It is widely acknowledged that SEP across the life course influences health13 and that observational studies of socially patterned exposures and outcomes should adjust for measures of SEP across the life course,14 but indicators of early-life SEP have not yet been included in population survey monitoring systems. Longitudinal cohort studies have been the preeminent design in life-course epidemiology; the indicators of life-course SEP used in these studies have not generally been applied in a surveillance context. Measuring SEP at more than 1 point over the life course allows investigation into how social mobility, both upward and downward, is associated with health outcomes. The success of policies aiming to redress the health effects of disadvantage in early life and to protect against challenges in later life can be monitored and evaluated if SEP over the life course is measured in surveillance systems.
A commonly used indicator to examine health inequities is self-rated health. Studies of the 1958 National Child Development Study British birth cohort demonstrated that lifetime socioeconomic circumstances accounted for inequities in self-rated health.15,16 Four European studies found stronger associations for SEP in adulthood with self-rated health than with childhood SEP, although this varied by country and gender.17 Trends in Finland showed improvements in self-rated health between 1972 and 1992, and inequities in self-rated health by education and income decreased over this period.18 No studies have examined trends in self-rated health among groups who have experienced socioeconomic disadvantage over the life course—whether they have experienced cumulative disadvantage throughout childhood and adulthood or upward or downward social mobility.
We analyzed data from a cross-sectional representative population survey to examine the prevalence of self-rated excellent or very good health among people who were socially mobile (upward or downward) between childhood and adulthood, with retrospectively recalled information about childhood SEP. We also analyzed surveillance data collected continuously between 2002 and 2007 to measure the prevalence of self-rated excellent or very good health among different socioeconomic groups over time and to evaluate the potential of life-course SEP indicators in surveillance systems to help monitor the direction of inequities in health among socially mobile and stable groups.
Two data sources were used for the purposes of this study: a cross-sectional survey (Health Monitor), to examine the association between self-rated health and social mobility, and a surveillance system (SAMSS), to determine trends in self-rated health over time. Because SAMSS did not collect information about early-life SEP, Health Monitor data were superimposed on SAMSS data to demonstrate how health inequities could be monitored more comprehensively if indicators of early-life SEP were included in such a surveillance system.
Health Monitor, conducted in September 2004, used simple random sampling to select households in South Australia with a connected telephone and a number listed in the electronic white pages. The purpose of the survey was described in a letter sent to each selected household prior to interviewing. The household member who was 18 years or older and most recently had a birthday was selected for a computer-assisted telephone interview. Up to 6 callbacks were made in an attempt to interview the selected person. Nonrespondents were not replaceable. From the eligible sample (n = 4342), 714 refused to participate in the interview, 233 were not contacted after 6 attempts, 160 were incapacitated, 135 were unavailable for interviewing, 90 were excluded because of language barriers (interviews were conducted in English, Greek, Italian, and Vietnamese), and 8 terminated the interview. Three respondents were excluded because of missing data required for weighting purposes. A total of 2999 completed interviews were analyzed, a response rate of 69.1%.
Self-rated health status was assessed by the question, “In general, would you say your health was excellent, very good, good, fair, or poor?”19 In accordance with the World Health Organization definition of health as a state of complete physical, mental, and social well-being, we dichotomized responses to examine the prevalence of excellent or very good health. Respondents were asked whether a doctor had ever told them that they had diabetes and to report their height without shoes and their weight (undressed in the morning). Body mass index (weight in kilograms divided by height in meters squared) was calculated from these responses, with obesity classified as a BMI of 30 kg/m2 or higher.20 Respondents who smoked daily or occasionally were classified as current smokers, those who smoked in the past but not currently were classified as ex-smokers, and those who had never smoked or had tried it a few times but not smoked regularly were classified as nonsmokers.
Respondents' main occupation, employment status, gross annual household income, family financial situation, housing tenure, and highest level of education were included as measures of current SEP. Respondents were asked what kind of work they had done for most of their lives. Occupations were coded according to the Australian Standard Classification of Occupations, which groups occupations according to level of education, knowledge, responsibility, and on-the-job training and experience required.21 Additional demographic factors collected included age, gender, area of residence (metropolitan or rural), marital status, and country of birth.
Indicators of SEP during early life were the respondent's recall of the family structure, housing tenure, and financial situation when the respondent was aged 10 years, and the main occupation of the respondent's father and mother. These variables are commonly used indicators in the literature examining early-life SEP and health.22 The age of 10 years was used in questions about childhood experiences because it was considered that adult respondents might be able to remember, or have been more aware of, their socioeconomic circumstances during this middle-childhood period than at earlier ages. This specific age is viewed not as a critical period when developmental paths are determined but as a time when exposure to low SEP may have a particularly strong effect.23
Two social mobility variables were created from current and early-life indicators of family financial situation and housing tenure. These indicators were chosen because questions about current family financial situation and housing tenure are routinely included in the SAMSS questionnaire. For family financial situation, high SEP was indicated by “being able to save a bit or a lot,” and low SEP was indicated by “not being able to save any money at all.” For housing tenure, living in a dwelling that a respondent or other family members owned or were purchasing indicated high SEP, and low SEP was indicated by living in a dwelling that was rented privately or from the government housing authority or having other housing circumstances. Each social mobility variable comprised 4 possible intergenerational trajectories: stable high, low in childhood and high in adulthood (upward mobility), high in childhood and low in adulthood (downward mobility), and stable low.
SAMSS is a continuous, monthly telephone monitoring system of a random representative sample of South Australians of all ages (approximately 600 persons are interviewed per month). All households with a telephone number listed in the electronic white pages are eligible for selection via simple random sampling. A letter describing the purpose of the survey is mailed to selected households prior to interviewing. The person who most recently had a birthday in the selected household is interviewed via computer-assisted telephone interview by trained interviewers. SAMSS differs from Health Monitor in an individual's probability of selection because the SAMSS protocol selects respondents regardless of their age, whereas Health Monitor samples only adults. Up to 10 callbacks are made to interview the selected person. Nonrespondents are not replaced. The average response rate during the period we studied was 69.6%, ranging from 68.6% in 2003 to 2004 to 71.0% in 2005 to 2006. We analyzed responses from respondents 18 years and older who were interviewed between July 2002 and June 2007 (n = 26 400).
Self-rated health, diabetes, obesity, smoking status, current employment status, gross annual household income, family financial situation, housing tenure, highest level of education, age, gender, area of residence, marital status, and country of birth were measured as in the Health Monitor survey. Occupation and indicators of SEP during early life were not included in the SAMSS questionnaire. Social mobility variables are therefore not currently available in this surveillance system.
Data were weighted by age group, gender, geographical area, and household size to the most recent estimated residential population (a figure the Australian Bureau of Statistics derives from census data)24 to account for different probabilities of selection and response rates among demographic groups, thus ensuring that the sample accurately reflected the South Australian adult population.
Descriptive analyses were conducted with SPSS version 14 (SPSS Inc, Chicago, IL). For comparison, the profile of respondents in Health Monitor, conducted during September 2004, and the profile of respondents in SAMSS during 2004 are shown in Table 1. We used the χ2 test to compare men and women on the social mobility variables. To examine the association between self-rated health and social mobility, we calculated relative risks and adjusted for age by log binomial regression with Stata version 9 (StataCorp LP, College Station, TX).
TABLE 1 Demographic Profile of Respondents: Health Monitor, Australia, 2004, and South Australian Monitoring and Surveillance System (SAMSS), 2002–2007
| Health Monitor, % (95% CI) | SAMSS, % (95% CI) | |
| Gender | ||
| Men | 49.0 (47.2, 50.8) | 48.6 (47.3, 50.0) |
| Women | 51.0 (49.2, 52.8) | 51.4 (50.0, 52.7) |
| Age, y | ||
| 18–24 | 12.1 (11.0, 13.3) | 11.1 (10.3, 12.0) |
| 25–34 | 17.2 (15.9, 18.6) | 17.4 (16.4, 18.4) |
| 35–44 | 19.3 (17.9, 20.8) | 19.5 (18.5, 20.6) |
| 45–54 | 18.2 (16.8, 19.6) | 18.4 (17.4, 19.4) |
| 55–64 | 13.9 (12.7, 15.2) | 14.1 (13.2, 15.0) |
| 65–74 | 9.8 (8.8, 10.9) | 9.9 (9.1, 10.7) |
| ≥ 75 | 9.5 (8.5, 10.6) | 9.6 (8.9, 10.4) |
| Country of birth | ||
| Australia | 77.9 (76.4, 79.3) | 78.4 (77.3, 79.4) |
| United Kingdom/Ireland | 11.1 (10.1, 12.3) | 11.5 (10.7, 12.4) |
| Other | 11.0 (9.9, 12.1) | 10.1 (9.4, 11.0) |
| Marital status | ||
| Married/living with partner | 67.7 (66.0, 69.3) | 68.7 (67.5, 69.9) |
| Separated/divorced | 7.0 (6.2, 8.0) | 6.6 (6.0, 7.3) |
| Widowed | 6.4 (5.6, 7.3) | 6.2 (5.6, 6.9) |
| Never married | 18.9 (17.5, 20.3) | 18.4 (17.4, 19.5) |
| Residencea | ||
| Metropolitan | 74.0 (72.4, 75.6) | 72.8 (71.6, 74.0) |
| Rural | 26.0 (24.4, 27.6) | 27.2 (26.0, 28.4) |
| Educational attainment | ||
| Secondary/high school | 51.7 (50.0, 53.5) | 55.6*** (54.3, 56.9) |
| Trade/apprenticeship/certificate | 30.4 (28.8, 32.1) | 24.9*** (23.8, 26.1) |
| Undergraduate degree or more | 17.8 (16.5, 19.2) | 19.5 (18.5, 20.6) |
| Gross annual household income, AU $ | ||
| ≤ 20 000 | 15.1 (13.9, 16.5) | 15.9 (15.0, 16.9) |
| 20 001–40 000 | 18.4 (17.0, 19.8) | 19.4 (18.4, 20.5) |
| 40 001–60 000 | 18.3 (16.9, 19.7) | 16.7 (15.8, 17.7) |
| > 60 000 | 37.5 (35.7, 39.2) | 35.0* (33.7, 36.2) |
| No response | 10.8 (9.7, 11.9) | 13.0** (12.1, 13.9) |
| Employment status | ||
| Full time | 41.8 (40.1, 43.6) | 44.1* (42.8, 45.4) |
| Part time/casual | 18.6 (17.3, 20.1) | 18.1 (17.1, 19.2) |
| Unemployed/unable to work | 5.6 (4.8, 6.4) | 5.2 (4.7, 5.9) |
| Home duties | 9.1 (8.1, 10.2) | 8.1 (7.4, 8.8) |
| Retired | 19.4 (18.0, 20.9) | 20.8 (19.7, 21.9) |
| Student | 4.5 (3.8, 5.3) | 3.5* (3.1, 4.0) |
| No response | 1.0 (0.7, 1.4) | 0.1*** (0.1, 0.3) |
| Occupationb | ||
| Manager/administrator/professional/ associate professional | 31.3 (29.6, 32.9) | … |
| Tradesperson/advanced clerical or service/ intermediate clerical/service/sales/production/transport | 40.0 (38.3, 41.8) | … |
| Elementary clerical, sales, service/laborer | 20.2 (18.8, 21.7) | … |
| No response/home duties/student/never worked | 8.5 (7.6, 9.6) | … |
| Current family financial situation | ||
| Can't save at all | 30.0 (28.4, 31.7) | 29.9 (28.7, 31.2) |
| Save a bit | 50.3 (48.5–52.1) | 50.6 (49.0, 51.9) |
| Save a lot | 18.1 (16.8, 19.5) | 16.9 (16.0, 17.9) |
| Don't know/no response | 1.7 (1.3, 2.2) | 2.5* (2.2, 3.0) |
| Current housing tenure | ||
| Own or buying home | 82.4 (81.0, 83.7) | 83.1 (82.1, 84.0) |
| Rental, retirement home, other | 17.5 (16.1, 18.8) | 16.9 (16.0, 17.9) |
| Self-rated health | ||
| Excellent/very good | 53.5 (51.8, 55.3) | 54.7 (53.4, 56.0) |
| Good/fair/poor | 46.5 (44.7, 48.2) | 45.3 (44.0, 46.6) |
| Smoking status | ||
| Nonsmoker/ex-smoker | 80.9 (79.5, 82.3) | 80.5 (79.5, 81.5) |
| Current smoker | 19.1 (17.7, 20.5) | 19.5 (18.5, 20.5) |
| Body mass index, kg/m2 | ||
| Not obese | 83.8 (82.5, 85.1) | 82.2 (81.2, 83.2) |
| Obese ( ≥ 30) | 16.2 (14.9, 17.5) | 17.8 (16.8, 18.8) |
| Diabetes | ||
| No | 93.4 (92.4, 94.2) | 92.5 (91.7, 93.1) |
| Yes | 6.6 (5.8, 7.6) | 7.5 (6.9, 8.3) |
Note. CI = confidence interval. For the Health Monitor survey, n = 2999; for SAMSS, n = 5541. Ellipses indicate questions not included in SAMSS.
a Residence was coded as metropolitan or rural on the basis of postcode.
b Occupations were coded according to the Australian Standard Classification of Occupations,21 which groups occupations according to level of education, knowledge, responsibility, and the on-the-job training and experience required.
*P < .05; **P < .01; ***P < .001, for comparisons with the Health Monitor survey participants.
Respondents with high SEP in both childhood and adulthood were the reference category. Variables that were associated with self-rated health in age-adjusted models (P < .25)25 were entered into a multivariate model. The variance inflation factor was used to check for multicollinearity among the variables included in the model. Multicollinearity was not found to be a concern (variance inflation factor < 15). Nonsignificant (P ≥ .05) covariates were removed until a final model was obtained. Relative risks for the age-adjusted and multivariate models were estimated through log binomial regression from the generalized linear modeling command in Stata. We chose this procedure instead of calculating odds ratios with logistic regression because odds ratios can overstate the association when the outcome is common. We used Stata to conduct the generalized linear modeling because SPSS does not take account of noninteger weights in this procedure. In cases in which convergence was not achieved with this method, we used a Poisson generalized estimated equation.26 Differences were reported as significant when P < .05, P < .01, and P < .001. We used the χ2 for trend test to examine trends in self-rated health among different groups over time.
The prevalence of excellent or very good health status among respondents to the Health Monitor survey was 53.5% (95% confidence interval = 51.8, 55.3) and was not significantly different between men and women. Self-rated health prevalence was not significantly different between Health Monitor and SAMSS samples.
The distribution of social mobility variables and early-life SEP variables is shown in Table 2. Approximately one quarter of respondents experienced upward mobility in family financial situation, and one fifth experienced upward mobility in housing tenure. Social mobility was not significantly different between men and women. Overall, 8.2% of respondents experienced upward mobility, and 2.4% experienced downward mobility in both family financial situation and housing tenure.
TABLE 2 Distribution of Early-Life Socioeconomic Position (SEP) Variables Among Men and Women: Health Monitor, South Australia, 2004
| Men, % (95% CI) | Women, % (95% CI) | |
| Family financial situation | ||
| High childhood and adulthood SEP | 38.4 (36.0, 40.9) | 38.7 (36.3, 41.2) |
| Low childhood, high adulthood SEP (upward mobility) | 25.4 (23.3, 27.7) | 22.9 (20.8, 25.0) |
| High childhood, low adulthood SEP (downward mobility) | 10.7 (9.2, 12.3) | 12.8 (11.2, 14.6) |
| Low childhood and adulthood SEP | 15.5 (13.8, 17.5) | 15.9 (14.2, 17.8) |
| Don't know/Not stated | 9.9 (8.5, 11.6) | 9.7 (8.3, 11.2) |
| Housing tenure | ||
| High childhood and adulthood SEP (downward mobility) | 61.8 (59.3, 64.2) | 59.8 (57.4, 62.3) |
| Low childhood, high adulthood SEP (upward mobility) | 21.0 (19.0, 23.1) | 19.6 (17.7, 21.6) |
| High childhood, low adulthood SEP (downward mobility) | 10.3 (8.8, 11.9) | 12.5 (10.9, 14.2) |
| Low childhood and adulthood SEP | 5.3 (4.2, 6.5) | 6.0 (5.0, 7.3) |
| Don't know/Not stated | 1.7 (1.1, 2.4) | 2.0 (1.4, 2.9) |
| Father's main occupationa | ||
| Manager/administrator/professional/associate professional | 37.4 (34.9, 39.9) | 40.2 (37.8, 42.7) |
| Tradesperson/advanced clerical or service/intermediate clerical/service/sales/ production/ transport | 40.9 (38.4, 43.4) | 41.5 (39.1, 44.1) |
| Elementary clerical/sales/service/laborer | 17.8 (15.9, 19.8) | 13.0*** (11.4, 14.7) |
| Don't know/no response/unable to code/not in labor force | 3.9 (3.0, 5.0) | 5.3 (4.3, 6.5) |
| Mother's main occupationa | ||
| Manager/administrator/professional/associate professional | 15.4 (13.7, 17.4) | 17.5 (15.6, 19.4) |
| Tradesperson/advanced clerical or service/ intermediate clerical/service/sales/ production/transport | 19.1 (17.1, 21.1) | 22.2* (20.2, 24.4) |
| Elementary clerical/sales/service/laborer | 13.4 (11.7, 15.2) | 15.3 (13.6, 17.2) |
| Home duties | 49.7 (47.1, 52.2) | 43.3*** (40.9, 45.8) |
| Don't know/no response/unable to code/not in labor force | 2.4 (1.7, 3.3) | 1.7 (1.2, 2.5) |
| Family structure when respondent was 10 years of age | ||
| Both biological/adoptive parents in residence | 89.8 (88.2, 91.3) | 87.3* (85.6, 88.9) |
| Step/blended family | 3.2 (2.4, 4.2) | 3.1 (2.4, 4.1) |
| Single parent | 5.3 (4.3, 6.6) | 7.9** (6.7, 9.4) |
| Other | 1.6 (1.1, 2.4) | 1.5 (1.0, 2.2) |
| No response | 0.1 | 0.1 |
Note. CI = confidence interval. For men, n = 1469; for women, n = 1530.
a Occupations were coded according to the Australian Standard Classification of Occupations,21 which groups occupations according to level of education, knowledge, responsibility, and the on-the-job training and experience required.
*P < .05; **P < .01; ***P < .001, for comparisons with men.
In the age-adjusted model, the proportion of respondents reporting excellent or very good health was significantly lower among respondents who experienced upward (51.0%) or downward (53.5%) social mobility in family financial situation than among those who experienced high family financial situation during both childhood and adulthood (61.7%). Respondents who experienced disadvantage during both childhood and adulthood in either family financial situation (41.8%) or housing tenure (33.8%) were also less likely to report excellent or very good health. Other factors associated with excellent or very good health are shown in Table 3.
TABLE 3 Unadjusted Numbers, Proportions, and Age-Adjusted and Multivariate Relative Risks of Variables Associated With Excellent or Very Good Health: Health Monitor, South Australia, 2004
| Excellent or Very Good Health, No./Total Responses (%) | Age-Adjusted Model, RR (95% CI) | Multivariate Model, RR (95% CI) | |
| Family financial situation | |||
| High childhood and adulthood SEP (Ref) | 714/1157 (61.7) | 1.00 | 1.00 |
| Low childhood, high adulthood SEP (upward mobility) | 369/724 (51.0) | 0.87** (0.79, 0.96) | 0.88** (0.80, 0.97) |
| High childhood, low adulthood SEP (downward mobility) | 189/353 (53.5) | 0.87* (0.76, 0.99) | 0.94 (0.83, 1.07) |
| Low childhood and adulthood SEP | 197/472 (41.8) | 0.70*** (0.62, 0.80) | 0.81** (0.72, 0.92) |
| Don't know/no response | 137/294 (46.7) | 0.78** (0.68, 0.91) | 0.87* (0.75, 1.00) |
| Housing tenure | |||
| High childhood and adulthood SEP (Ref) | 1030/1823 (56.5) | 1.00 | 1.00 |
| Low childhood, high adulthood SEP (upward mobility) | 319/608 (52.5) | 0.97 (0.89, 1.07) | 1.03 (0.94, 1.13) |
| High childhood, low adulthood SEP (downward mobility) | 182/342 (53.1) | 0.93 (0.82, 1.05) | 1.00 (0.88, 1.13) |
| Low childhood and adulthood SEP | 57/170 (33.8) | 0.61*** (0.47, 0.78) | 0.71** (0.56, 0.91) |
| Don't know/no response | 17/56 (31.3) | 0.59* (0.40, 0.89) | 0.69 (0.47, 1.02) |
| Smoking status | |||
| Nonsmoker/ex-smoker (Ref) | 1360/2427 (56.0) | 1.00 | 1.00 |
| Current smoker | 246/572 (43.0) | 0.72*** (0.64, 0.81) | 0.77*** (0.69, 0.86) |
| Body mass index, kg/m2 | |||
| Not obese (Ref) | 1432/2514 (57.0) | 1.00 | 1.00 |
| Obese (≥ 30 kg/m2) | 174/485 (35.9) | 0.64*** (0.56, 0.73) | 0.67*** (0.58, 0.77) |
| Diabetes | |||
| No (Ref) | 1549/2801 (55.3) | 1.00 | 1.00 |
| Yes | 57/198 (28.8) | 0.55*** (0.44, 0.70) | 0.66*** (0.53, 0.82) |
| Residencea | |||
| Metropolitan (Ref) | 1166/2220 (52.5) | 1.00 | 1.00 |
| Rural | 440/779 (56.5) | 1.08* (1.00, 1.17) | 1.09* (1.01, 1.18) |
| Country of birth | |||
| Australia (Ref) | 1288/2336 (55.1) | 1.00 | 1.00 |
| United Kingdom/Ireland | 176/334 (52.7) | 1.00 (0.89, 1.13) | 1.01 (0.90, 1.13) |
| Other | 143/329 (43.3) | 0.82* (0.71, 0.96) | 0.86* (0.75, 0.99) |
| Marital status | |||
| Married/living with partner (Ref) | 1124/2030 (55.3) | 1.00 | 1.00 |
| Separated/divorced | 103/211 (48.8) | 0.90 (0.79, 1.03) | 0.98 (0.87, 1.11) |
| Widowed | 82/192 (42.7) | 0.95 (0.80, 1.12) | 0.96 (0.82, 1.12) |
| Never married | 297/566 (52.5) | 0.83** (0.72, 0.95) | 0.87* (0.77, 0.99) |
| Employment status | |||
| Full time (Ref) | 739/1254 (59.0) | 1.00 | 1.00 |
| Part time/casual | 333/559 (59.6) | 1.00 (0.91, 1.10) | 1.04 (0.94, 1.14) |
| Unemployed/unable to work | 49/167 (29.3) | 0.50*** (0.37, 0.67) | 0.62*** (0.46, 0.84) |
| Home duties | 132/273 (48.4) | 0.83* (0.72, 0.96) | 0.87* (0.76, 1.00) |
| Retired | 255/582 (43.8) | 0.81* (0.68, 0.97) | 0.90 (0.77, 1.05) |
| Student | 82/134 (61.0) | 0.99 (0.79, 1.24) | 1.05 (0.85, 1.31) |
| No response | 16/30 (51.7) | 0.91 (0.60, 1.38) | 1.02 (0.72, 1.45) |
| Gender | |||
| Men (Ref) | 777/1469 (52.9) | 1.00 | … |
| Women | 829/1530 (54.2) | 1.03 (0.96, 1.12) | … |
| Educational attainment | |||
| Secondary/high school (Ref) | 793/1552 (51.1) | 1.00 | … |
| Trade/apprenticeship/certificate/diploma | 496/913 (54.4) | 1.04 (0.96, 1.14) | … |
| Undergraduate degree or more | 316/533 (59.3) | 1.12* (1.01, 1.25) | … |
| Gross annual household income, AU $ | |||
| ≤ 20 000 (Ref) | 669/1123 (59.6) | 1.00 | … |
| 20 001–40 000 | 307/548 (55.9) | 0.94 (0.85, 1.04) | … |
| 40 001–60 000 | 299/551 (54.2) | 0.93 (0.84, 1.04) | … |
| > 60 000 | 174/454 (38.4) | 0.69*** (0.60, 0.79) | … |
| No response | 157/322 (48.8) | 0.85* (0.72, 0.99) | … |
| Occupationb | |||
| Manager/administrator/professional/ associate professional (Ref) | 546/937 (58.2) | 1.00 | … |
| Tradesperson/advanced clerical or service/ intermediate clerical/service/sales/production/transport | 630/1199 (52.6) | 0.89* (0.82, 0.97) | … |
| Elementary clerical, sales, service/laborer | 297/607 (48.9) | 0.82** (0.73, 0.92) | … |
| No response | 133/255 (52.1) | 0.90 (0.77, 1.06) | … |
| Father's main occupationb | |||
| Manager/administrator/professional/associate professional (Ref) | 654/1164 (56.2) | 1.00 | … |
| Tradesperson/advanced clerical or service/intermediate clerical/service/sales/ production/ transport | 658/1237 (53.2) | 0.95 (0.87, 1.03) | … |
| Elementary clerical/sales/service/laborer | 231/460 (50.2) | 0.90 (0.80, 1.02) | … |
| Don't know/no response/unable to code/not in labor force | 63/139 (45.5) | 0.81 (0.66, 1.01) | … |
| Mother's main occupationb | |||
| Manager/administrator/professional/associate professional (Ref) | 280/494 (56.7) | 1.00 | … |
| Tradesperson/advanced clerical or service/ intermediate clerical/service/sales/ production/transport | 364/620 (58.7) | 1.04 (0.91, 1.17) | … |
| Elementary clerical/sales/service/laborer | 233/430 (54.0) | 0.96 (0.83, 1.10) | … |
| Home duties | 699/1393 (50.2) | 0.94 (0.84, 1.05) | … |
| Don't know/no response/unable to code/not in labor force | 30/62 (48.5) | 0.88 (0.65, 1.19) | … |
| Family structure when respondent was 10 years of age | |||
| Both biological/adoptive parents in residence (Ref) | 1448/2655 (54.5) | 1.00 | … |
| Step/blended family | 52/94 (55.0) | 0.99 (0.80, 1.22) | … |
| Single parent | 85/199 (42.7) | 0.78** (0.65, 0.93) | … |
| Other | 20/47 (42.7) | 0.80 (0.57, 1.11) | … |
| No response | 1/3 (29.4) | … | … |
Note. RR = relative risk; CI = confidence interval. Ellipses indicate variables not included in final multivariate model.
a Residence was coded as metropolitan or rural on the basis of postcode.
b Occupations were coded according to the Australian Standard Classification of Occupations,21 which groups occupations according to level of education, knowledge, responsibility, and the on-the-job training and experience required.
*P < .05; **P < .01; ***P < .001.
After adjustment for other significant covariates in the multivariate model, upward social mobility in family financial situation and low family financial situation or housing tenure during both childhood and adulthood remained significantly associated with a lower prevalence of excellent or very good health, compared with high SEP during both childhood and adulthood. Downward mobility in family financial situation or housing tenure was not significantly associated with lower self-rated health in the multivariate model.
The trend over time in the proportion reporting excellent or very good health, assessed in SAMSS for each quarter between July 2002 and June 2007, is shown by family financial situation and housing tenure in Figure 1. We found no significant trends over time in the proportion of respondents reporting excellent or very good health among respondents who were able to save money (χ2 trend = 3.51; P = .06) or those who were not able to save money (χ2 trend = 0.37; P = .54). We observed a significant upward trend over time in the proportion reporting excellent or very good health among respondents who owned or were purchasing their dwelling (χ2 trend = 6.42; P = .01), but the trend among respondents who were renting or who had other housing circumstances was not significant (χ2 trend = 1.08; P = .30).
FIGURE 1 Proportion reporting excellent or very good health status by current socioeconomic position over time and indicators of social mobility for (a) family financial situation and (b) housing tenure: Health Monitor, Australia, 2004, and South Australian Monitoring and Surveillance System (SAMSS), 2002–2007.
We superimposed the social mobility variable from the cross-sectional Health Monitor survey on the SAMSS data (Figure 1). This illustrated the increased detail that could be obtained if such social mobility variables were available over time in the surveillance data. These data suggest that an advantaged family financial experience during childhood partially protected against the effect of low SEP in adult life. Experiencing a disadvantaged family financial situation during childhood, irrespective of family financial situation during adulthood, was associated with a lower prevalence of excellent or very good health. A relatively advantaged SEP, measured by housing tenure during either childhood or adulthood, appeared to be beneficial for self-rated health, in contrast with a relatively disadvantaged housing tenure experience across both childhood and adulthood.
A relatively advantaged family financial situation during childhood appeared to be protective against low self-rated health in adulthood. Participants in these surveys who experienced downward mobility in family financial situation between childhood and adulthood had a prevalence of excellent or very good health similar to that of respondents who experienced high SEP during childhood and adulthood. Experiencing a disadvantaged family financial situation in childhood had enduring effects on self-rated health in adulthood: upward mobility to a relatively advantaged SEP in adulthood was still associated with a lower prevalence of excellent or very good health. Conversely, participants who experienced upward or downward mobility in housing tenure had self-rated health in adulthood similar to those who experienced relatively advantaged housing tenure throughout their lives. These data suggest that the low probability of excellent or very good health associated with living in rented or other housing during childhood may be attenuated by purchasing or owning a home in adulthood. Provision of good-quality public housing for many working-class people in South Australia during the 1960s and 1970s may have contributed to their improved health status, although the availability of public housing has declined since the 1990s.27
Our results provide support for policies that aim to improve SEP, particularly for those that aim to improve family financial situation during childhood or housing throughout life. Policies that minimize the extent to which the socioeconomic disadvantage experienced by parents is passed on to their children are important in reducing inequities that arise from intergenerational transmission of disadvantage.28 Policies focusing on early child development and the healthy environment of children support the upward mobility of children from disadvantaged backgrounds.29
Our study did not provide support for the theory of health-selective mobility, which postulates that the upwardly mobile have better health and the downwardly mobile have worse health than do their destination group.30–32 Health-selective mobility argues against policies promoting social mobility, because it implies that social mobility could increase health inequities. We found, however, that housing tenure correlated with similar self-rated health scores among the upwardly mobile and those who were advantaged in childhood and adulthood. In measures of family financial situation, by contrast, the upwardly mobile had poorer health than did their destination group. These results for housing tenure and family financial situation over the life course support the theory that SEP effects accumulate across childhood and adulthood.33
The sampling frame for Health Monitor and SAMSS was generated from the electronic white pages, so people without a listed telephone were ineligible for selection. The majority of households owned telephones, however, and these surveys have been shown to be representative, with few health or socioeconomic differences found between those with and without a listed number.34,35
SAMSS and Health Monitor necessarily rely on retrospective recall of early-life information. No information was available in this study about the accuracy of recall. Validity of self-reported housing tenure or family financial situation may decrease as length of time since childhood increases. The survey items about housing tenure and family financial situation during childhood generated a high response rate: answers were missing for housing tenure in only 1.7% of completed interviews and for family financial situation in 9.0%. Previous analyses demonstrated that missing data on these indicators, although not occurring completely at random, were associated with only a few observed variables, suggesting that recall of childhood housing tenure and family financial situation are useful indicators of early-life SEP.36
These cross-sectional data did not provide information about causal relationships between SEP and health. Surveillance systems are designed to monitor characteristics in the population over time and detect changes. For example, changes in the housing market, with ownership becoming less affordable, or changes in the demographic profile of the population may influence the proportion of homeowners or renters who experience excellent or very good health. A strength of SAMSS is that a random sample of South Australians is interviewed each month on a wide range of health-related indicators. Consecutive samples can be combined to provide sufficient statistical power for examining particular groups in the population. Associations between social mobility and many other health outcomes, including chronic conditions, risk factors, and behaviors, could be provided from SAMSS data over time if early-life SEP were included in the questionnaire.
Housing tenure and family financial situation reflect material resources and are important determinants of ill health.37 Housing tenure is also related to self-rated health through psychosocial mechanisms, because social meanings such as self-identity, pride, and a place for refuge are attached to owning a home.38 It will be important to determine if results from this study can be replicated in the South Australian population at other times and in other settings before they are recommended for routine use in surveillance systems. A limitation of housing tenure as a measure of social mobility in surveillance systems is that its meaning and importance to material circumstances may vary across temporal and geographical contexts. Home ownership is generally considered indicative of advantage in Australia, a country in which home ownership rates are relatively high. In countries in which home ownership rates are lower, or declining, not owning a home is not necessarily a marker of socioeconomic disadvantage.39 Family financial situation is closely related to income, which is considered a direct measure of material living standards.37 Family financial situation measures perceptions of how well the family lives, or lived, within their income and may allow more comparison across different income groups in various temporal, demographic, geographic, and cultural contexts. Questions about family financial situation are also associated with lower nonresponse rates than are income items, at least in South Australia.
Existing surveillance data show that socioeconomic disadvantage, indicated by not being able to save any money or not owning or purchasing a home, is negatively associated with excellent or very good self-rated health. Surveillance data can tell us whether reform policies and programs actually reduce inequities in health. Current surveys, which only measure current SEP, can reveal a narrowing or widening of the gap between the health of relatively advantaged and disadvantaged groups. Without any measures of early-life SEP, however, these surveys can offer no evidence about whether increasing or decreasing health inequities are related to early-life experiences. Including indicators of social mobility by measuring early-life SEP as well as current SEP would enable stratification of health trends by social mobility variables. This would enable us to assess not only whether we are closing the gap but also whether policies introduced across the life course and reform aimed at early life make a difference.
Acknowledgments
The South Australian Monitoring and Surveillance System was funded by the South Australian Department of Health.
The authors acknowledge the statistical advice provided by Thomas Sullivan.
Human Participant Protection
This study satisfied all criteria for the ethical treatment of human participants. Health Monitor surveys were approved by the South Australian Department of Health Human Research Ethics Committee.
