You are here

Article Text

Article menu
Global health
Original research
Educational inequalities in adult mortality: a systematic review and meta-analysis of the Asia Pacific region
  1. Kathryn Christine Beck1,2,
  2. Mirza Balaj2,
  3. Lorena Donadello2,
  4. Talal Mohammad2,
  5. Hanne Dahl Vonen2,
  6. Claire Degail2,
  7. Kristoffer Eikemo2,
  8. Anna Giouleka2,
  9. Indrit Gradeci2,
  10. Celine Westby2,
  11. Kam Sripada2,3,
  12. Magnus Rom Jensen4,
  13. Solvor Solhaug4,
  14. Emmanuela Gakidou5,6,
  15. Terje Andreas Eikemo2
  1. 1Department of Public Health and Nursing, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norway
  2. 2Centre for Global Health Inequalities Research, Department of Sociology and Political Science, Norwegian University of Science and Technology, Trondheim, Norway
  3. 3Centre for Digital Life Norway, Norwegian University of Science and Technology, Trondheim, Norway
  4. 4Library Section for Humanities, Education and Social Sciences, University Library, Norwegian University of Science and Technology, Trondheim, Norway
  5. 5Institute for Health Metrics and Evaluation, University of Washington, Seattle, Washington, USA
  6. 6Department of Health Metrics Sciences, School of Medicine, University of Washington Seattle Campus, Seattle, Washington, USA
  1. Correspondence to Kathryn Christine Beck; kathryb{at}stud.ntnu.no

Abstract

Objectives In this study, we aim to analyse the relationship between educational attainment and all-cause mortality of adults in the high-income Asia Pacific region.

Design This study is a comprehensive systematic review and meta-analysis with no language restrictions on searches. Included articles were assessed for study quality and risk of bias using the Joanna Briggs Institute critical appraisal checklists. A random-effects meta-analysis was conducted to evaluate the overall effect of individual level educational attainment on all-cause mortality.

Setting The high-income Asia Pacific Region consisting of Japan, South Korea, Singapore and Brunei Darussalam.

Participants Articles reporting adult all-cause mortality by individual-level education were obtained through searches conducted from 25 November 2019 to 6 December 2019 of the following databases: PubMed, Web of Science, Scopus, EMBASE, Global Health (CAB), EconLit and Sociology Source Ultimate.

Primary and secondary outcome measures Adult all-cause mortality was the primary outcome of interest.

Results Literature searches resulted in 15 345 sources screened for inclusion. A total of 30 articles meeting inclusion criteria with data from the region were included for this review. Individual-level data from 7 studies covering 222 241 individuals were included in the meta-analyses. Results from the meta-analyses showed an overall risk ratio of 2.40 (95% CI 1.74 to 3.31) for primary education and an estimate of 1.29 (95% CI 1.08 to 1.54) for secondary education compared with tertiary education.

Conclusion The results indicate that lower educational attainment is associated with an increase in the risk of all-cause mortality for adults in the high-income Asia Pacific region. This study offers empirical support for the development of policies to reduce health disparities across the educational gradient and universal access to all levels of education.

PROSPERO registration number CRD42020183923.

  • PUBLIC HEALTH
  • EPIDEMIOLOGY
  • SOCIAL MEDICINE

Data availability statement

Data are available on reasonable request. Researchers interested in using the extracted data from systematic review articles may contact TAE, Department of Sociology and Political Science, NTNU with this request at terje.eikemo@ntnu.no. These data include information extracted from published journal articles found through the systematic review literature searches.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/bmjopen-2021-059042

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Strengths and limitations of this study

    • The present study provides the most up-to-date collection of evidence regarding educational inequalities in adult all-cause mortality for the high-income Asia Pacific Region.

    • The present systematic review used a global, language-unrestricted search strategy to gather all available data.

    • The present systematic review and meta-analysis is the first to our knowledge to use harmonised education groups for this region, allowing more comparable effect sizes.

    • This study employed a random-effects meta-analysis model, giving a more robust estimate given the high levels of heterogeneity.

    • High levels of heterogeneity in the meta-analysis likely are due to the differences between the populations within the studies and differences between methodologies and measurements used.

    Introduction

    Research in public health often fails to examine how the societal conditions we live in shape our health and life chances. These societal conditions in which we are born and grow include our education, income, employment status, housing and work conditions, and have great and complex influences on our subsequent health and disease status.1 While it may seem intuitive that those making a larger income would experience better health than someone living in poverty, the social determinants also act on a gradient, as that with every additional year of education health status improves. As a result, research to explain and measure inequities in health is needed to guide policy changes aimed at not only closing the gap between the most and least advantaged groups, but ultimately by reducing the inequalities between groups across the social gradient in health.

    Education is often used as a proxy measure for socioeconomic status (SES) in health inequalities research, as it is relatively constant after young adulthood while having a significant impact on later measures of SES. It is also influenced through parental characteristics and therefore, to some degree, can act as an indicator of early-life socioeconomic conditions.2 Though having a higher level of education may help one by increasing prospects of a good job with a decent income and thus improving access to a healthy lifestyle, the exact pathways in which education influences health are complex and intertwining. Egerter et al3 described the pathways in which educational attainment affects one’s health through health knowledge and behaviours, healthcare access, working conditions, income, social network and social standing.

    Despite the growing focus on health inequalities, this field has been dominated by research describing trends in North American and European countries, often leaving other contexts under-researched. In the Asia Pacific region, specifically Japan, South Korea, Singapore and Brunei Darussalam, research into health inequalities gained increasing attention in the 1990s and early 2000s, with most research conducted in Japan and South Korea. While comparative studies are rare, one study comparing educational inequalities in mortality between Japan and the USA found that the magnitude of the inequalities was similar.4 This suggests that the intensity of such inequalities may be similar to those in Europe and the USA. Traditionally, this region has had the longest life expectancies (LE), relatively little inequality and tight supportive social networks.5 When comparing LE by welfare regime, the East Asian regime had higher average LE than the traditionally high LE of Scandinavian welfare regimes, suggesting that other social and cultural factors may be of importance in this region.6 However, as rapid economic growth driven by drastic technological innovation and growing globalisation contribute to the high LE and improved population health, inequalities also rise.7 If we want to reduce inequalities, prevent the widening of the health gap and diminish disparities across the social gradient, we first need to quantify the level of inequality so we can monitor these trends over time and enact policies that reduce these unjust inequities.

    Therefore, the aim of this study is to analyse the relationship between educational attainment and all-cause mortality of adults in the high-income Asia Pacific region, consisting of Japan, South Korea, Singapore and Brunei Darussalam. This region has been classified by the Global Burden of Diseases, Injuries and Risk Factors Study led by the Institute for Health Metrics and Evaluation by epidemiological similarity and geographic closeness. This analysis focuses on relative inequalities in mortality, which may paint a dire picture of widening inequalities despite targeted public health interventions. However, in reality, the absolute rates of mortality often have drastically improved, and inequalities in absolute terms are often seen to have actually reduced over time.8 Therefore, it is important to consider that both reductions in overall mortality rates, and compositional changes in education groups may contribute to rising relative inequalities in mortality, despite improvements in absolute inequalities.8

    Methods

    Search strategy and selection criteria

    A literature search was conducted using the databases PubMed (NIH/NLM), Web of Science (Clarivate), Scopus (Elsevier), EMBASE (Elsevier), Global Health (CAB-EBSCO), EconLit (EBSCO) and Sociology Source Ultimate (EBSCO) from 25 November 2019 to 6 December 2019 by experienced librarians. The search was limited to records published since 1980 without any language limitations. The search was constructed using key terms: education, socioeconomic status, health inequalities, adult, mortality, all-cause mortality and death. The specific search string contains commonly related terms and synonyms to the key terms above. Rather than using the Boolean term “AND” for the search string, we used proximity searching. Setting the proximity to 10 ensures the two words or blocks must occur within ten words of each other, yielding a more relevant and manageable result through a higher bar for inclusion in the search results. The literature search string used is provided in online supplemental appendix A. A handsearch for relevant articles not found through database searching was conducted by searching the reference lists of obtained articles such as literature reviews and narrative reviews.

    Abstracts of articles found through the literature search were screened by two independent reviewers. In the case of disagreement, a third independent reviewer determined the final decision. During the abstract screening, all articles that mentioned social group analysis were included. This was done as many studies may not explicitly state educational attainment as a measure examined, but rather simply state ‘socioeconomic variables’. During full article reading, two reviewers independently screened each article for inclusion, with a third review used in case of disagreement. The inclusion and exclusion criteria are summarised in table 1 according to the SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research type) framework. Articles were assessed for risk of bias and quality by one reviewer independently using the Joanna Briggs Institute (JBI) Critical Appraisal checklists for the study’s design (cohort or cross-sectional).9 10 Studies received a score of 1 if the criterion was met and 0 if not, or if it was unclear. The scores were added together and classified into equal quality categories based on the maximum score possible. For cohort studies, the fair category included scores from 0 to 3, good quality was scores from 4 to 7 and excellent was scores from 8 to 11. For cross-sectional studies, the categories for fair, good and excellent were determined from scores ranging from 0 to 2, 3 to 5 and 6 to 8, respectively.

    View this table:
    Table 1

    SPIDER (Sample, Phenomenon of Interest, Design, Evaluation, Research type) framework for inclusion and exclusion criteria

    All relevant data were extracted using a standardised extraction template. The template includes information such as location (country or region), date(s), sample sizes, method of measuring exposure (education) and outcome (mortality), confounders adjusted for in the mutivariable analysis, effect measure estimates, educational groups definition according to study, corresponding years of education and others. Extractors used ISCED mappings11 to determine corresponding numerical years of education for articles that reported only education categories. Illiterate was considered 0 years of education, while literate was considered ≥1 year. Articles were extracted independently by reviewers due to the large number of records. A quality control random sample of 10% of the extractions from each reviewer were extracted in duplicate by experienced reviewers. Following data extraction, articles which contained data for the High-Income Asia Pacific region were selected for inclusion into this study.

    Effect Size Computation

    Primary education was classified as middle school education or less, equal to 0–9 years of schooling. Secondary education was classified as high school, equal to 10–12 years of schooling and tertiary education was classified as a college education or higher, equal to 13 years of schooling or more. Risk ratios (RR) and 95% CIs were calculated for an overall effect size for each study, using the raw population numbers for each educational group with the tertiary educational group used as the reference category. The primary factors for effect size calculation and therefore inclusion into the meta-analysis were: (1) the design of the study and (2) availability of data needed for effect size calculation and (3) the study’s educational attainment grouping. Three articles met the requirements to be included in the meta-analysis but used the same underlying dataset. Due to this, we excluded two12 13 of the articles which used only one wave of the NHANES survey and therefore had a smaller sample size and less follow-up time than the article from Khang and Kim which remained in the analysis.14 A random-effects model with inverse variance weighting was used to complete the meta-analysis. To assess for the presence of publication bias, where smaller studies with only highly significant effect sizes are more likely to be published, funnel plots are usually created and examined for asymmetry. However, funnel plots are only appropriate for detecting publication bias when studies included in the meta-analysis come from one underlying population.15 As the number of studies included in our meta-analysis is below the threshold for statistical tests for funnel plot asymmetry to be reliable, and our population does not come from one single underlying population and rather is a heterogeneous population, a funnel plot would not be appropriate for this meta-analysis.15 Statistical analyses were completed using RStudio with R V.4.0.3 and the ‘meta’ R package.16 17

    Assessing heterogeneity and sensitivity analyses

    As heterogeneity is expected between included studies due to the variation between participants' characteristics and settings, a test for heterogeneity was conducted. This was first done by using Cochrane’s Q test to assess heterogeneity, and the heterogeneity was then quantified by using an I2 statistic. As recommended by Cochrane, a p value of less than 0.1 was considered statistically significant heterogeneity.18 The degree of heterogeneity in the I2 statistic was determined as being low (0%–25%), medium (25%–75%) or high (75%–100%).18 A random-effects model was chosen a priori for the meta-analysis to account for the expected high levels of heterogeneity and give a more robust estimate. To assess whether the choice of tertiary education as the reference group in the meta-analysis influenced the results, we conducted a sensitivity analysis with secondary education as the reference category. This did not alter the main findings and the results from the sensitivity analysis are available in online supplemental appendix B, online supplemental figures B1 and B2.

    Patient and public involvement

    Due to the global nature of the systematic review strategy, it was not feasible to involve the public in the design or conduct of this study.

    Results

    The literature search yielded 15 017 records after the first-degree removal of duplicates. The handsearching of reference lists of relevant records yielded an additional 384 records and after duplicate removal, 15 345 records continued to the abstract screening phase. Of these, 1799 articles were assessed for eligibility in the full-text screening, and 579 articles met the criteria for data extraction for the global review. In total, 30 articles were conducted in one or more countries within the high-income Asia Pacific region as shown in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram in figure 1. Of the 30 studies included in this synthesis, 24 are longitudinal cohort studies, 5 are cross-sectional studies using unlinked death certificate data, and 1 is a prospective pooled cohort analysis.

    Figure 1
    Figure 1

    PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)flow diagram. *Twelve extractions from one book.

    Study characteristics

    Table 2 gives an overview of the descriptive characteristics of the included studies. One study is a prospective pooled cohort study that uses data from both Japan, South Korea and Singapore.19 Aside from the pooled cohort study, 14 studies each (48.3%) were conducted in Japan and South Korea, 1 study (3.45%) was conducted in Singapore and no studies were conducted in Brunei Darussalam. A majority (19) of the studies were representative of the entire population (63.3%), while others were representative of specific cities, municipalities or population groups. Leaving 11 studies that were not representative of the entire population (36.6%). In total, two studies used data from as early as 1963 and 1970, while four studies included earliest data from the late 1980s, and 26 studies included data from the 1990s and onwards. Among the cohort studies, five studies (20%) were classified as ‘good’ quality through the JBI risk of bias and quality critical appraisal checklists, while 20 (80%) were of ‘excellent’ quality and 0 studies were of ‘fair’ quality. For cross-sectional studies, all five (100%) scored in the ‘good’ quality category. The risk of bias and study quality scores and categories are presented for each included study in table 2 and individual scores for each quality assessment criterion are presented in online supplemental appendix C, online supplemental table C1 and C2.

    View this table:
    Table 2

    Descriptive characteristics of included studies

    Analysis of the association between educational attainment and mortality

    Seven studies covering 222 241 individuals and 17 551 deaths allowed for analysis of all-cause mortality by harmonised educational categories. Figure 2 shows the risk of mortality by primary educational status with the tertiary educational category used as the reference category. The individual studies with the calculated (unadjusted) RR and 95% CIs, and the overall random-effects estimate are listed along with a forest plot visualisation of the studies’ estimates. For the studies included in the meta-analysis, six studies (86%) were categorised as ‘excellent’ quality14 20–24 through the risk of bias and quality assessment, while one study (14%) was of ‘good’ quality,25 and no studies were of ‘fair’ quality.

    Figure 2
    Figure 2

    Forest plot for primary education versus tertiary education. Primary education equates to middle school or less (0–9 years), while tertiary education equates to college or higher (≥13 years). Tertiary education is the reference category. RR, risk ratio.

    The overall estimate indicated that individuals with primary education had 2.40 times the risk for all-cause mortality compared with tertiary educated individuals (95% CI 1.74 to 3.31, z=5.33, p<0.01). In this analysis, RRs ranged from 1.21 to 6.13, with one estimate including the null value.17 The Q statistic was statistically significant (Q=236.98, p˂0.01), and the I2 value estimates that about 97% of the variation across studies is due to heterogeneity, rather than chance, both of which suggest significant heterogeneity.

    Figure 3 illustrates the meta-analysis and forest plot using secondary education compared with tertiary education (reference) in mortality risk. In this analysis, the overall effect estimate shows a statistically significant increase in the risk of mortality by secondary education, with a 29% increase in risk (z=2.87, p˂0.01). Meaning those with 10–12 years of schooling had 1.29 times the risk of mortality compared with those with 13 or more years of schooling. However, multiple studies did not show as clear of an increase in risk as in previous analyses. Three of the seven studies’ effect sizes or 95% CI included the null value of 1, indicating a non-statistically significant or null effect of secondary education on the risk of mortality in these estimates. The Q statistic was statistically significant (Q=57.50, p˂0.01), and the I2 value estimates that about 90% of the variation across studies is due to heterogeneity, both of which suggest significant heterogeneity.

    Figure 3
    Figure 3

    Forest plot for secondary education versus tertiary education. Secondary education equates to high school (10–12 years), while tertiary education equates to college or higher (≥13 years). Tertiary education is the reference category. RR, risk ratio.

    Discussion

    This study provides a comprehensive systematic review with meta-analysis describing the influence of educational attainment on adult all-cause mortality in the high-income Asia Pacific region. The results reveal that lower educational attainment is associated with a significant increase in the risk of mortality for adults in the high-income Asia Pacific region. Despite impressive improvements in health and living conditions in the region in the past 50 years,5 26 the inequalities in educational attainment and mortality have persisted. Though overall mortality rates in the population have declined, trends in relative inequalities show a persistence and widening in the region.27 28 A study by Kasajima and Hashimoto28 examined absolute and relative educational disparities in mortality in Japan and found relative educational inequalities in all-cause mortality persisted despite improvement in average mortality rates. Interestingly, the researchers found that absolute inequalities also widened for causes of mortality linked to lifestyle and behavioural factors, as well as an overall worsening of average mortality for vulnerable populations such as youth and women.28 In South Korea, Bahk, Lynch and Khang27 found compositional changes in educational groups as a likely cause of the increase in absolute inequalities in mortality and mortality decline, which may have contributed to the increase of relative inequalities. When using measures more robust to such compositional changes, researchers found stable trends for both relative and absolute inequalities in recent years.27 The pathways in which these trends in all-cause mortality have developed may also be different when examining relative or absolute inequalities. Khang et al29 found that when looking at relative educational inequalities in mortality, material factors explained 29% of the excess mortality risk compared with 78.6% of the absolute risk.29 This shows how relative and absolute inequalities may develop differently, and while we may see worsening inequalities when measured on a relative scale, absolute inequalities may have remained stable or even improved.

    Similar multinational studies of high-income countries in Europe have found large relative educational inequalities in all-cause mortality, as found in our study. Our estimates suggest a great relative disparity in total mortality for the lowest educated individuals compared with those with a college education or higher. Similarly, when estimating relative inequalities in Western European countries, Mackenbach et al30 find an increase in risk of all-cause mortality between 57% and 115% for the lowest educated men compared with the highest educated men, and between 37% and 105% for the lowest educated women, in the years 2005–2009.30 The researchers also examined the trends in absolute and relative inequalities for these high-income countries between 1990–1994 and 2005–2009, finding that while many countries have seen an increase in the relative educational inequalities for both men and women, many countries have also seen an impressive reduction in absolute educational inequalities in mortality.30 When looking over a longer time period (1979–2014), Mackenbach et al31 again find an almost universal widening of relative educational inequalities in mortality in Western Europe, likely resulting from decreasing mortality rates across the population.31 The trend of declining mortality and stable or declining absolute inequalities remained despite periods of economic crisis and increases in unemployment and poverty.31 Due to data restrictions, we were not able to examine long-term trends in relative inequalities in this analysis. If we assume the trends in the high-income Asia Pacific region follow the same pattern seen in other high-income countries such as in Western Europe, we could expect to see rising relative educational inequalities with declining overall mortality and perhaps stable or improving absolute inequalities, as seen in studies by Kasajima and Hashimoto28 and Bahk, Lynch and Khang.27 There is still a need however for further analysis into educational inequalities in this region and analyses into the long-term trends in both relative and absolute terms.

    To the best of our knowledge, one other systematic review and meta-analysis of educational inequalities in all-cause mortality for the Asian context has been completed.32 This review included, among others, Japan, South Korea and Singapore, with 11 studies from these countries in the meta-analyses. Researchers compared the highest available education level to the lowest level in each study, finding an overall RR of 1.29 (95% CI 1.17 to 1.43) for those with ‘low education’ compared with those with ‘high education’ in their subgroup consisting of Japan, South Korea, and Singapore.32 Overall, the results from Vathesatogkit, Batty and Woodward32 support the results found in our analyses, however, their overall estimate includes the highest adjusted estimate from each study, while ours is based on unadjusted estimates, likely contributing to the lower estimate compared with ours. All included studies in the meta-analysis from Vathesatogkit, Batty and Woodward32 except one, which did not meet our inclusion criteria, were included in our review—suggesting that the review process was thorough, and all relevant literature was included. Therefore, this study provides both an updating of the existing literature review and provides new knowledge on the relationship between education and health in this region, through harmonised educational groups.

    Previous research has hypothesised the good health of Japan as being a result of high levels of collectivism, social cohesion and job security.33 Before the introduction of neo-liberal market reforms, employers in Japan and South Korea relied heavily on long-term workers with seniority-based wage systems. Through this framework, employees enjoyed relatively high levels of job security and low levels of income inequality between workers. However, due to the introduction of these reforms sparked by economic crisis, employers began shifting to non-regular workers and pay-per-performance schemes, resulting in rising insecurity and inequality with fewer social insurance protections for workers and weakening social cohesion.34 This may have increased inequalities in South Korea as the relative risk of mortality for men of low education in 1980 was 3.44, increasing to 6.41 in 2000 after the introduction of these market reforms, when compared with highly educated men.27

    Perhaps because of this transformation, the educational systems have seen skyrocketing competitiveness and high-income and middle-income families are drastically outspending low-income families on education.35–37 The phenomenon of ‘shadow education’ or private supplemental educational lessons referred to as juku in Japan and hagwon in South Korea is by no means new or isolated to East Asian countries. The percentage of average monthly household educational expenditure on private supplementary education shows an unequal and rising trend across income quintiles from 2002 to 2013 in Singapore.38 The wealthiest 20% increased their educational expenditures by 1.49% during this 10-year timespan, while the poorest 20% increased their expenditure by 0.62%. These expenditure trends and skyrocketing competition may result in the educational system being a reproducer of class status instead of a mechanism for social mobility and may lead to an increase in future inequalities.34 On the one hand, education can be used as a powerful tool to eliminate inequities and promote social mobility, while on the other hand, when middle-income and high-income families drastically outspend low-income families on supplemental schooling, education is used to reproduce and exacerbate inequities in society.

    Implications and future research

    By demonstrating a gradient effect of education on mortality, we offer empirical support for policies that aim to improve morbidity and mortality across all socioeconomic groups rather than focus on closing the gap between the most and least advantaged groups. We also demonstrate strong support for policy aimed at improving access to education for all, from primary through higher education, as we see a significant impact of education on the health and life of all individuals.

    Although this review offers valuable empirical evidence, further research is needed on this topic and region. Additional longitudinal data are also needed, especially in South Korea, Singapore and Brunei Darussalam. Further research is also needed to examine the mechanisms in which education influences health and mortality, although this requires a rich, comprehensive evidence base, and so far, is not feasible in this region. This review makes an essential step in the identification and monitoring of educational inequalities in mortality in the high-income Asia Pacific region, and future research is needed to monitor these trends across time and eventually to reduce disparities in health, allowing all individuals the opportunity to live a long healthy life.

    The results and interpretations of this review should be taken with consideration to its limitations. We cannot be sure that our conclusions apply to countries with data restrictions, namely Singapore and Brunei Darussalam. As all the studies relied on self-reported education levels, we cannot rule out the potential for response bias in some individuals if they could not accurately remember their total years of education or their highest level of educational attainment. Additionally, the high levels of heterogeneity in the analysis should not be ignored. This is most likely due to the differences between the populations within the studies and differences between methodologies and measurements used. While it is not surprising that there are high levels of heterogeneity in the meta-analysis, due to the between country nature of the study and the inability to control for all confounders within the primary studies, the interpretation of this meta-analysis should be taken with caution as we cannot be sure the effect seen from these studies would be the same for the entire population either within or outside of the region studied.

    To our knowledge, this is the most up-to-date review of educational attainment and adult mortality conducted in this region. This study benefits from using education as a measure of SES, as this is a consistent and early indicator of an individual’s SES.39 This review also used a thorough global search without language restriction allowing for near-complete coverage of all relevant articles. Additionally, the meta-analysis used harmonised educational categories in estimating the adult all-cause mortality. Lastly, this study employed a random-effects model, giving a more robust estimate given the high levels of heterogeneity.

    Conclusion

    This study provides empirical evidence to support the association between educational attainment and adult all-cause mortality in the high-income Asia Pacific region. We see a gradient effect of education on mortality as with every step up the educational ladder, individuals experience a reduced risk of mortality. Overall, these results offer a basis for evidence-based policy decisions to reduce health disparities across the educational gradient and improve access to education from primary to higher education. Further research is needed to expand the limited research base in this region, to allow for consistent monitoring of these trends, and to support further policy changes aimed at reducing health disparities.

    Data availability statement

    Data are available on reasonable request. Researchers interested in using the extracted data from systematic review articles may contact TAE, Department of Sociology and Political Science, NTNU with this request at terje.eikemo@ntnu.no. These data include information extracted from published journal articles found through the systematic review literature searches.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Not applicable.

    References

    1. ↵
      1. World Health Organization (WHO)
      . A conceptual framework for action on the social determinants of health: debates, policy & practice, case studies, 2010.
    2. ↵
      1. Galobardes B,
      2. Shaw M,
      3. Lawlor DA, et al
      . Indicators of socioeconomic position (Part 1). J Epidemiol Community Health 2006;60:7–12.doi:10.1136/jech.2004.023531pmid:http://www.ncbi.nlm.nih.gov/pubmed/16361448
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Egerter S,
      2. Braveman P,
      3. Sadegh-Nobari T,
      4. Gossman-Kahn R,
      5. Dekker M
      . Education matters for health. exploring the social determinants of health.; 2011.
    4. ↵
      1. Chiu C-T,
      2. Hayward M,
      3. Saito Y
      . A comparison of educational differences on physical health, mortality, and healthy life expectancy in Japan and the United States. J Aging Health 2016;28:1256–78.doi:10.1177/0898264316656505pmid:http://www.ncbi.nlm.nih.gov/pubmed/27590801
      OpenUrlCrossRefPubMed
    5. ↵
      1. Marmot MG,
      2. Smith GD
      . Why are the Japanese living longer? BMJ 1989;299:1547–51.doi:10.1136/bmj.299.6715.1547pmid:http://www.ncbi.nlm.nih.gov/pubmed/2514896
      OpenUrlFREE Full Text
    6. ↵
      1. Abdul Karim S,
      2. Eikemo TA,
      3. Bambra C
      . Welfare state regimes and population health: integrating the East Asian welfare states. Health Policy 2010;94:45–53.doi:10.1016/j.healthpol.2009.08.003pmid:http://www.ncbi.nlm.nih.gov/pubmed/19748149
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Chetty R,
      2. Grusky D,
      3. Hell M, et al
      . The fading American DREAM: trends in absolute income mobility since 1940. Science 2017;356:398–406.doi:10.1126/science.aal4617pmid:http://www.ncbi.nlm.nih.gov/pubmed/28438988
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Mackenbach JP
      . Re-thinking health inequalities. Eur J Public Health 2020;30:615.doi:10.1093/eurpub/ckaa001pmid:http://www.ncbi.nlm.nih.gov/pubmed/32558893
      OpenUrlCrossRefPubMed
    9. ↵
      1. Joanna Briggs Institute
      . Checklist for analytical cross sectional studies, 2017.
    10. ↵
      1. Joanna Briggs Institute
      . Checklist for cohort studies, 2017.
    11. ↵
      1. UNESCO
      . ISCED mappings, 2016.
    12. ↵
      1. Khang Y-H,
      2. Kim HR
      . Explaining socioeconomic inequality in mortality among South Koreans: an examination of multiple pathways in a nationally representative longitudinal study. Int J Epidemiol 2005;34:630–7.doi:10.1093/ije/dyi043pmid:http://www.ncbi.nlm.nih.gov/pubmed/15746204
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Khang Y-H,
      2. Kim HR
      . [Socioeconomic mortality inequality in Korea: mortality follow-up of the 1998 National Health and Nutrition Examination Survey (NHANES) data]. J Prev Med Public Health 2006;39:115–22.pmid:http://www.ncbi.nlm.nih.gov/pubmed/16615265
      OpenUrlPubMed
    14. ↵
      1. Khang Y-H,
      2. Kim H-R
      . Socioeconomic inequality in mortality using 12-year follow-up data from nationally representative surveys in South Korea. Int J Equity Health 2016;15:51.doi:10.1186/s12939-016-0341-9pmid:http://www.ncbi.nlm.nih.gov/pubmed/27001045
      OpenUrlPubMed
    15. ↵
      1. Terrin N,
      2. Schmid CH,
      3. Lau J, et al
      . Adjusting for publication bias in the presence of heterogeneity. Stat Med 2003;22:2113–26.doi:10.1002/sim.1461pmid:http://www.ncbi.nlm.nih.gov/pubmed/12820277
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. R Studio Team
      . RStudio: integrated development environment for R. MA, Boston, 2021.
    17. ↵
      1. Balduzzi S,
      2. Rücker G,
      3. Schwarzer G
      . How to perform a meta-analysis with R: a practical tutorial. Evid Based Ment Health 2019;22:153–60.doi:10.1136/ebmental-2019-300117pmid:http://www.ncbi.nlm.nih.gov/pubmed/31563865
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Deeks JJ,
      2. Higgins JPT,
      3. Altman DJ
      . Chapter 10: Analysing data and undertaking meta-analyses. In: Cochrane Handbook for systematic reviews of interventions version 61: cochrane 2020, 2020.
    19. ↵
      1. Yang K,
      2. Zhang Y,
      3. Saito E, et al
      . Association between educational level and total and cause-specific mortality: a pooled analysis of over 694 000 individuals in the Asia cohort Consortium. BMJ Open 2019;9:e026225.doi:10.1136/bmjopen-2018-026225pmid:http://www.ncbi.nlm.nih.gov/pubmed/31444178
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Nishi A,
      2. Kawachi I,
      3. Shirai K, et al
      . Sex/gender and socioeconomic differences in the predictive ability of self-rated health for mortality. PLoS One 2012;7:e30179.doi:10.1371/journal.pone.0030179pmid:http://www.ncbi.nlm.nih.gov/pubmed/22276157
      OpenUrlCrossRefPubMed
    21. ↵
      1. Ito S,
      2. Takachi R,
      3. Inoue M
      . Education in relation to incidence of and mortality from cancer and cardiovascular disease in Japan 2008;7.
    22. ↵
      1. Hirokawa K,
      2. Tsutusmi A,
      3. Kayaba K
      . Impacts of educational level and employment status on mortality for Japanese women and men: the Jichi medical school cohort study. Eur J Epidemiol 2006;21:641–51.doi:10.1007/s10654-006-9049-2pmid:http://www.ncbi.nlm.nih.gov/pubmed/17048083
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Kim JM,
      2. Son N-H,
      3. Park E-C, et al
      . The relationship between changes in employment status and mortality risk based on the Korea labor and income panel study (2003-2008). Asia Pac J Public Health 2015;27:NP993–1001.doi:10.1177/1010539513486923pmid:http://www.ncbi.nlm.nih.gov/pubmed/23674827
      OpenUrlCrossRefPubMed
    24. ↵
      1. Kim GR,
      2. Jee SH,
      3. Pikhart H
      . Role of allostatic load and health behaviours in explaining socioeconomic disparities in mortality: a structural equation modelling approach. J Epidemiol Community Health 2018;72:545–51.doi:10.1136/jech-2017-209131pmid:http://www.ncbi.nlm.nih.gov/pubmed/29459378
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Honjo K,
      2. Iso H,
      3. Fukuda Y, et al
      . Influence of municipal- and individual-level socioeconomic conditions on mortality in Japan. Int J Behav Med 2014;21:737–49.doi:10.1007/s12529-013-9337-7pmid:http://www.ncbi.nlm.nih.gov/pubmed/24057408
      OpenUrlPubMed
    26. ↵
      1. Ikeda N,
      2. Saito E,
      3. Kondo N, et al
      . What has made the population of Japan healthy? The Lancet 2011;378:1094–105.doi:10.1016/S0140-6736(11)61055-6
      OpenUrl
    27. ↵
      1. Bahk J,
      2. Lynch JW,
      3. Khang Y-H
      . Forty years of economic growth and plummeting mortality: the mortality experience of the poorly educated in South Korea. J Epidemiol Community Health 2017;71:282–8.doi:10.1136/jech-2016-207707pmid:http://www.ncbi.nlm.nih.gov/pubmed/27707841
      OpenUrlAbstract/FREE Full Text
    28. ↵
      1. Kasajima M,
      2. Hashimoto H
      . Social policies and change in education-related disparities in mortality in Japan, 2000-2010. SSM Popul Health 2020;12:100692.doi:10.1016/j.ssmph.2020.100692pmid:http://www.ncbi.nlm.nih.gov/pubmed/33241104
      OpenUrlPubMed
    29. ↵
      1. Khang Y-H,
      2. Lynch JW,
      3. Yang S, et al
      . The contribution of material, psychosocial, and behavioral factors in explaining educational and occupational mortality inequalities in a nationally representative sample of South Koreans: relative and absolute perspectives. Soc Sci Med 2009;68:858–66.doi:10.1016/j.socscimed.2008.12.003pmid:http://www.ncbi.nlm.nih.gov/pubmed/19121885
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Mackenbach JP,
      2. Kulhánová I,
      3. Artnik B, et al
      . Changes in mortality inequalities over two decades: register based study of European countries. BMJ 2016;353:i1732.doi:10.1136/bmj.i1732pmid:http://www.ncbi.nlm.nih.gov/pubmed/27067249
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Mackenbach JP,
      2. Valverde JR,
      3. Artnik B, et al
      . Trends in health inequalities in 27 European countries. Proc Natl Acad Sci U S A 2018;115:6440–5.doi:10.1073/pnas.1800028115pmid:http://www.ncbi.nlm.nih.gov/pubmed/29866829
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Vathesatogkit P,
      2. Batty GD,
      3. Woodward M
      . Socioeconomic disadvantage and disease-specific mortality in Asia: systematic review with meta-analysis of population-based cohort studies. J Epidemiol Community Health 2014;68:375–83.doi:10.1136/jech-2013-203053pmid:http://www.ncbi.nlm.nih.gov/pubmed/24407596
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Marmot M
      . Status syndrome. Signif 2004;1:150–4.doi:10.1111/j.1740-9713.2004.00058.x
      OpenUrl
    34. ↵
      1. Koo H
      . The changing faces of inequality in South Korea in the age of globalization. Korean Studies 2008;31:1–18.doi:10.1353/ks.2008.0018
      OpenUrl
    35. ↵
      1. Gan AL,
      2. Shahrill M
      . Why the need for private tuition in mathematics? investigating shadow education in Brunei Darussalam. AIP Publishing LLC, 2019.
    36. ↵
      1. Hannum E,
      2. Ishida H,
      3. Park H, et al
      . Education in East Asian societies: postwar expansion and the evolution of inequality. Annu Rev Sociol 2019;45:625–47.doi:10.1146/annurev-soc-073018-022507
      OpenUrl
    37. ↵
      1. Teo P,
      2. Koh D
      . Shadow education in Singapore: a Deweyan perspective. Educational Philosophy and Theory 2020;52:869–79.doi:10.1080/00131857.2019.1700500
      OpenUrl
    38. ↵
      1. Peng N, Lecturer of Social Sciences at Singapore Management University
      . Inequality and the social compact in Singapore macro trends versus lived realities. Southeast Asian Economies 2020;36:355–79.doi:10.1355/ae36-3e
      OpenUrl
    39. ↵
      1. Valkonen T
      . Problems in the measurement and international comparisons of socio-economic differences in mortality. Soc Sci Med 1993;36:409–18.doi:10.1016/0277-9536(93)90403-Qpmid:http://www.ncbi.nlm.nih.gov/pubmed/8434266
      OpenUrlCrossRefPubMedWeb of Science
      1. Fujino Y,
      2. Tamakoshi A,
      3. Iso H, et al
      . A nationwide cohort study of educational background and major causes of death among the elderly population in Japan. Prev Med 2005;40:444–51.doi:10.1016/j.ypmed.2004.07.002pmid:http://www.ncbi.nlm.nih.gov/pubmed/15530597
      OpenUrlCrossRefPubMedWeb of Science
      1. Minagawa Y,
      2. Saito Y
      . Active social participation and mortality risk among older people in Japan: results from a nationally representative sample. Res Aging 2015;37:481–99.doi:10.1177/0164027514545238pmid:http://www.ncbi.nlm.nih.gov/pubmed/25651580
      OpenUrlCrossRefPubMed
      1. Tani Y,
      2. Kondo N,
      3. Nagamine Y, et al
      . Childhood socioeconomic disadvantage is associated with lower mortality in older Japanese men: the JAGES cohort study. Int J Epidemiol 2016;45:dyw146.doi:10.1093/ije/dyw146pmid:http://www.ncbi.nlm.nih.gov/pubmed/27401729
      OpenUrlPubMed
      1. Sugisawa H,
      2. Liang J,
      3. Liu X
      . Social networks, social support, and mortality among older people in Japan. J Gerontol 1994;49:S3–13.doi:10.1093/geronj/49.1.S3pmid:http://www.ncbi.nlm.nih.gov/pubmed/8282987
      OpenUrlCrossRefPubMedWeb of Science
      1. Ishizaki T,
      2. Kai I,
      3. Imanaka Y
      . Self-Rated health and social role as predictors for 6-year total mortality among a non-disabled older Japanese population. Arch Gerontol Geriatr 2006;42:91–9.doi:10.1016/j.archger.2005.05.002pmid:http://www.ncbi.nlm.nih.gov/pubmed/16046010
      OpenUrlCrossRefPubMedWeb of Science
      1. Iwasaki M,
      2. Otani T,
      3. Ohta A, et al
      . Rural-urban differences in sociodemographic, social network and lifestyle factors related to mortality of middle-aged Japanese men from the Komo-Ise cohort study. J Epidemiol 2002;12:93–104.doi:10.2188/jea.12.93pmid:http://www.ncbi.nlm.nih.gov/pubmed/12033534
      OpenUrlCrossRefPubMedWeb of Science
      1. Honjo K,
      2. Iso H,
      3. Ikeda A, et al
      . Employment situation and risk of death among middle-aged Japanese women. J Epidemiol Community Health 2015;69:1012–7.doi:10.1136/jech-2015-205499pmid:http://www.ncbi.nlm.nih.gov/pubmed/26043897
      OpenUrlAbstract/FREE Full Text
      1. Liang J,
      2. Bennett J,
      3. Krause N, et al
      . Old age mortality in Japan: does the socioeconomic gradient interact with gender and age? J Gerontol B Psychol Sci Soc Sci 2002;57:S294–307.doi:10.1093/geronb/57.5.S294pmid:http://www.ncbi.nlm.nih.gov/pubmed/12198109
      OpenUrlCrossRefPubMedWeb of Science
      1. Iwasa H,
      2. Kawaai C,
      3. Gondo Y, et al
      . Subjective well-being as a predictor of all-cause mortality among middle-aged and elderly people living in an urban Japanese community: a seven-year prospective cohort study. Geriatr Gerontol Int 2006;6:216–22.doi:10.1111/j.1447-0594.2006.00351.x
      OpenUrl
      1. Jung-Choi K,
      2. Khang YH,
      3. Cho HJ
      . Changes in contribution of causes of death to socioeconomic mortality inequalities in Korean adults. J Prev Med Public Health 2011;44:249–59.doi:10.3961/jpmph.2011.44.6.249pmid:http://www.ncbi.nlm.nih.gov/pubmed/22143175
      OpenUrlCrossRefPubMed
      1. Suh G-H
      . Predictors of mortality in an aging community-based cohort in Korea. Psychogeriatrics 2006;6:10–18.doi:10.1111/j.1479-8301.2006.00118.x
      OpenUrlCrossRef
      1. Lim D,
      2. Kong KA,
      3. Lee HA, et al
      . The population attributable fraction of low education for mortality in South Korea with improvement in educational attainment and no improvement in mortality inequalities. BMC Public Health 2015;15:313.doi:10.1186/s12889-015-1665-xpmid:http://www.ncbi.nlm.nih.gov/pubmed/25880221
      OpenUrlCrossRefPubMed
      1. Son M,
      2. Armstrong B,
      3. Choi JM
      . Relation of occupational class and education with mortality in Korea. 2, 2002.
      1. Khang Y-H,
      2. Kim HR
      . Relationship of education, occupation, and income with mortality in a representative longitudinal study of South Korea. Eur J Epidemiol 2005;20:217–20.doi:10.1007/s10654-005-0370-ypmid:http://www.ncbi.nlm.nih.gov/pubmed/15921038
      OpenUrlCrossRefPubMedWeb of Science
      1. Khang Y-H,
      2. Lynch JW,
      3. Kaplan GA
      . Health inequalities in Korea: age- and sex-specific educational differences in the 10 leading causes of death. Int J Epidemiol 2004;33:299–308.doi:10.1093/ije/dyg244pmid:http://www.ncbi.nlm.nih.gov/pubmed/15082630
      OpenUrlCrossRefPubMedWeb of Science
      1. Khang Y-H
      . Relationship between childhood socio-economic position and mortality risk in adult males of the Korea labour and income panel study (KLIPS). Public Health 2006;120:724–31.doi:10.1016/j.puhe.2006.02.008pmid:http://www.ncbi.nlm.nih.gov/pubmed/16824564
      OpenUrlCrossRefPubMedWeb of Science
      1. Ma S,
      2. Cutter J,
      3. Tan CE, et al
      . Associations of diabetes mellitus and ethnicity with mortality in a multiethnic Asian population: data from the 1992 Singapore National health survey. Am J Epidemiol 2003;158:543–52.doi:10.1093/aje/kwg199pmid:http://www.ncbi.nlm.nih.gov/pubmed/12965880
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors The study concept and strategy was developed by MB, KS, EG and TAE. The search strategy was developed and executed by MRJ and SS. Study screening and data extraction was completed by KCB, LD, TM, HDV, CD, KE, AG, IG and CW. Statistical analyses were completed by KCB. The manuscript was drafted by KCB and revised by TAE, MB and KS. KCB accepts responsibility for the finished work as the guarantor. All authors approved of final draft to be submitted.

    • Funding This study was funded by Norges Forskningsråd (288638).

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.