Download:
-
In response to the global push for health equity following the World Health Organization (WHO) report on the Commission for Social Determinants of Health, reducing health disparities has become a primary goal of public health policies worldwide. In China, inequalities in health, especially among older populations, are widespread (1). Healthy aging (HA) is crucial for achieving Sustainable Development Goals (1–2). However, limited research assesses the inequality of HA, specifically concerning childhood circumstances and its impact on HA inequality, termed inequality of opportunity (IOp) (3). Most studies use a mean-based approach to identify sources of IOp in health (3). Thus, understanding the role of childhood circumstances in HA inequality across the entire HA distribution in China, especially among individuals with poorer health, is critical but currently unknown. To bridge this gap, we used data from the China Health and Retirement Longitudinal Study (CHARLS) to develop a composite HA index for assessing changes in HA inequality. We explored how childhood circumstances affect IOp in HA across the entire HA distribution. Our findings reveal an increase in both total HA inequality and inequality explained by childhood circumstances from 2011 to 2020. Factors such as household socioeconomic status in childhood and regional/urban-rural status at birth were identified as key drivers of IOp in HA. Notably, the influence of early-life access to healthcare and parental health behaviors decreased as the HA distribution moved towards individuals with better health. These results highlight the significant role of childhood circumstances in HA determination and advocate for policies enhancing childhood nutrition and health, especially within disadvantaged families.
The CHARLS is a nationally representative survey conducted by the National School of Development at Peking University, focused on individuals aged 45 and above. Since its initiation in 2011, the survey has been followed up in 2013, 2015, 2018, and 2020. For this study, data from the 2014 Life History Survey was utilized, concentrating on individuals aged 60 and older. The analysis involved 23,409 participants, with varying numbers from the CHARLS surveys conducted in 2011 (3,317), 2013 (4,038), 2015 (4,919), 2018 (4,781), and 2020 (6,354), respectively. Sample weights were applied to adjust for design effects and survey nonresponses (
Supplementary Figure S1 ) (4).The Healthy Ageing Index (HAI) consists of five domains: physical capabilities, cognitive function, physiological health, psychological well-being, and social well-being (5). A total of 26 indicators were used, each categorized into quintiles with a code from 0 to 100. The sum of all indicator scores divided by the total number of indicators yielded the HAI score, which ranges from 0 to 100 (
Supplementary Table S1 ). A higher HAI score indicates better aging status.The variables listed in
Supplementary Table S2 were categorized into two groups (6). The first group comprised demographic variables like age and sex. The second group included six domains: war exposure; childhood household socioeconomic status (SES) encompassing self-reported family financial status, parental educational levels, political status, and housing conditions; geographical location (east, central, west) and urban/rural status at birth; parental health status and behaviors during childhood (e.g., parental bedridden condition, alcohol consumption, smoking); childhood health and nutritional status (e.g., self-reported health compared to peers before age 15, childhood experiences of hunger before age 17); and childhood access to healthcare (e.g., vaccination status and type of initial doctor visit).We conducted descriptive statistical analyses to compare the circumstances and HAI scores across the years 2011, 2013, 2015, 2018, and 2020 within our sample cohort. Our sample was selected without biases related to observables, and detailed results are available upon request. To measure the IOp in HA, we employed the direct parametric method using ordinary least squares (OLS) regressions, where HAI scores served as the dependent variable and the circumstances as the independent variables. We measured inequality utilizing the mean logarithmic deviation (MLD) index (7). The proportion of absolute IOp to overall HA inequality was calculated to determine relative IOp values. We also applied unconditional quantile regression (UQR) to explore disparities in HAI scores due to circumstances at specific points in the HAI score distribution. This involved regressing the re-centered influence function (RIF) of the HAI percentile’s ranks against the circumstance variables. To ascertain the relative impact of each contextual element, we used the Shapley value decomposition technique in our regression analyses. All statistical procedures were performed using STATA (version 17.0; StataCorp, College Station, TX, US).
Supplementary Table S3 presents the descriptive statistics of the study sample. The average scores for HAI were similar in 2011 (79.2) and 2013 (79.8), with a significant decrease observed in 2020 (75.9). Throughout this period, the overall inequality in HAI ranged from 0.010 to 0.016, with childhood circumstances accounting for 10.9%–14.8% of this inequality. Furthermore, the total inequality in HAI and the proportion explained by childhood circumstances increased from 2011 to 2020. Among the various childhood circumstances, household socioeconomic status (21.9%–27.6%) and regional and urban/rural status at birth (11.8%–22.1%) were the major contributing factors (Figure 1).
Figure 1.Contributions of circumstances to IOp in HAI: Mean-based Shapley decomposition.
Abbreviation: IOp=inequality of opportunity; HAI=healthy aging index; SES=socioeconomic status.Figure 2 shows a significant IOp across various quantiles, demonstrating a consistent decrease towards the higher end of the HA distribution, irrespective of utilizing the complete dataset or different time points. Furthermore, the impact of parental health status, health-related behaviors, and access to healthcare during childhood on IOp escalates towards the lower end of the HA spectrum, where individuals with the most severe health conditions are predominant (Figure 3).
Figure 2.IOp in HAI at different quantiles (MLD index).
Abbreviation: IOp=inequality of opportunity; HAI=healthy aging index; MLD=mean logarithmic deviation.
Figure 3.Contributions of circumstances to IOp in HAI: RIF-based Shapley decomposition.
Abbreviation: IOp=inequality of opportunity; HAI=healthy aging index; RIF=re-centered influence function; SES=socioeconomic status. -
The present study has demonstrated a significant and progressive increase in IOp in the HA from 2011 to 2020. Among all variables analyzed, early-life household SES emerges as the most substantial factor contributing to this disparity. This underscores the pivotal influence of household SES in shaping the health status of elderly individuals. A plausible explanation is that a higher level of parental education, typically associated with improved household SES, correlates with increased health literacy and adoption of healthier behaviors among both parents and their offspring, encompassing aspects like dietary habits and medication adherence (8). Moreover, enhanced household SES, reflected in superior financial stability, facilitates access to better healthcare services and improved nutrition, thereby limiting children’s exposure to detrimental substances and environmental hazards. These elements collectively play a role in promoting better health outcomes in later life (8). Hence, enhancing household SES through measures such as elevating parents’ educational attainment, improving financial resources within families, and creating healthier living environments during childhood emerges as a viable strategy to mitigate healthcare disparities among the elderly population.
The findings underscore the importance of improving access to healthcare for children and enhancing the health and behaviors of parents in socioeconomically disadvantaged families to reduce health disparities in the Chinese population. Disparities among older adults in China are significantly influenced by geographic location and urban/rural distinctions, possibly due to unequal allocation of healthcare resources, primary healthcare services, and welfare support (9). Additionally, our study reveals that health inequality is influenced not only by age and gender but also by differences in childhood experiences.
Moreover, the study revealed that IOp had a greater impact on individuals with poor HA, emphasizing the influence of early-life conditions on disparities, especially in the context of limited household assets. Thus, utilizing distributional decompositions indicated that concentrating solely on mean-based decomposition would neglect crucial aspects associated with early-life conditions, notably when investigating disparities in household assets at the lower spectrum. This aspect is significant as individuals in this bracket are generally perceived as less healthy (10).
This study is subject to some limitations. First, data on childhood circumstances relied on retrospective self-reporting, which may lead to recall bias. Second, the mechanisms by which childhood circumstances influence HA are not yet understood.
Our study utilized data from the 2011–2020 CHARLS dataset as our analytical sample. It is crucial in life course research to incorporate thorough assessments of early-life determinants instead of solely focusing on current adult health status, especially when examining older populations. Therefore, our study contributes significant insights for life course research targeting the reduction of health disparities. Furthermore, our distribution analysis presents valuable findings that can inform the development of targeted public health strategies to tackle health disparities.
-
No conflicts of interest.
HTML
| Citation: |
|