Differences between occupations and their cumulative effects during working life are thought to contribute to variations in the incidence and mortality associated with several chronic diseases of aging (1). The Whitehall study of British civil servants was the first to demonstrate marked differences between mortality among individuals from different positions in the social hierarchy of occupations (2, 3). Higher mortality from cardiovascular diseases and a range of other causes of death occurred among those with the lowest-status occupations compared to those with higher managerial jobs although the extent to which this could be attributed to occupational, environmental, genetic, individual socio-economic, or lifestyle factors was unclear. Since then, many other studies have reported a relationship between various occupational characteristics and exposures and increased ill health and mortality later in life (4–6).

Most studies of the impact of occupations on future health have focused on mortality or on the incidence of specific diseases (5, 7–9). However, with many societies experiencing rapid aging of their populations, another social and economic imperative is to maintain an individual’s independence as long as possible, thereby reducing the need for institutional care. Disability-free survival, defined as survival free of severe physical disability or dementia has been proposed as a surrogate measure of independence, of particular relevance to studies of older people (10). To date, there is little information available about the impact of occupational factors on this important measure of population health.

The Aspirin in Reducing Events in the Elderly (ASPREE) study was a large, randomized clinical trial of low-dose aspirin and placebo in which disability-free survival was the primary outcome. Australian participants were healthy adults aged ≥70 years when recruited in 2010–2014 (11). Within a sub-study, an abbreviated occupational history, which included jobs held between approximately 1950 and 2009, was obtained. This analysis aims to determine the relationship between occupational characteristics and disability-free survival and mortality in healthy older adults. Understanding this relationship may help improve health outcomes and quality of life for older adults after retirement and inform public health policies and intervention strategies to promote healthy aging.

A total of 12,215 participants who provided work histories were included in this analysis (Figure 1). Table 1 shows their baseline characteristics. The predominant occupations for men were craft and trade workers (n = 1,349), followed by professionals (n = 1,072) and machine operators (n = 791). Among women, the predominant ISCO-88 occupations were clerks (n = 2,197), professionals (n = 1,429), and service workers or retail salespersons (n = 1,140; Figure 2).

More than half of the men worked in ‘blue-collar’ occupations (56%), and men were more likely than women to have a job involving vigorous physical activity (39% vs. 6%). Women predominantly worked indoors (92%), and just over half had roles that involved sitting and/or standing without physical activity (51%). Among men, 3% were current and 54% were former smokers, while among women, 2% were current and 31% were former smokers.

This large prospective survivor’s cohort study is the first to examine the association between occupational characteristics and disability-free survival in an older adult population in the post-retirement period. Disability-free survival is a surrogate measure of independent living which is particularly relevant in an aging population. The principal finding was a reduction in disability-free survival associated with the least skilled or ‘elementary’ occupations. These include cleaners, garbage collectors, and laborers in agriculture, fisheries, mining, construction, mining, and manufacturing.

Among the specific job characteristics, those with the major adverse influence on disability-free survival were working in an adverse social environment and work that involved a high accident risk. All of these are likely to be characteristics of ‘elementary’ occupations and to provide a potential explanation for the lower likelihood of healthy survival among these workers.

Of the ASPREE participants who died, 45% developed either dementia or persistent physical disability prior to death (17). However, all-cause mortality was not substantially different among the four-digit ISCO-88 occupation groups, including elementary occupations. This indicates that the significant reduction in disability-free survival among the elementary occupational groups is likely to have resulted from an increase in physical disability or dementia related to occupational history (18).

The relatively similar mortality among participants, including those working in elementary occupational groups contrasts with previous reports including the United States National Longitudinal Mortality Study (1) and the Whitehall study of United Kingdom civil servants (2). A report using Korean insurance claim data also found higher mortality among lower occupational categories and a study from Japan found a high mortality in elementary occupations with most of the mortality due to cancer (19, 20). As all participants in ASPREE were recruited as healthy, the analyses of mortality may be biased due to the selective exclusion of those most likely to die prematurely from occupational factors. Alternatively, it might reflect a declining social gradient within the Australian workforce or the impact of occupational health and safety standards introduced progressively over the working life of the ASPREE participants.

The results of this study add to a growing literature on the effect of occupational characteristics and exposures on other aspects of health in later life. These include a recent Global Burden of Disease study, which reported an increase in disability-adjusted life years attributable to several occupational factors (21). Several population-based cohorts have also reported links between physically demanding jobs, shift work, ergonomic and psychological stress factors, and all-cause mortality (4, 5, 7–9). The results of recent community-based cohort studies investigating the relationship between occupational factors and mortality are summarized in Supplementary Table S5. In particular, the association between heavy physical work and mortality, particularly in men has also been documented in other occupational studies (4, 6, 22).

Overall, this study suggests that some specific occupational characteristics may lead to adverse outcomes later in post-retirement age. Despite this, it is impossible to exclude the likelihood that the reduction in healthy survival reflects socioeconomic disadvantage, poor health behaviors, and the likelihood that individuals with pre-existing physical or cognitive limitations select elementary occupations. Smoking, in particular, was more common among individuals working in low-skilled labor-intensive occupations, but the influence of cigarette smoking was at least partially controlled in our analysis.

The public health significance of these findings arises from the substantial proportion of the workforce who undertake elementary and laboring occupations for a significant part of their working lives. Based on this analysis, these individuals may generate substantial additional costs for supportive care and social services in their older years. This emphasizes the need to determine more precisely whether job characteristics or other (e.g., lifestyle) characteristics lead to adverse health outcomes. Meanwhile, attention is needed to mitigate the potentially harmful effects of the negative attributes identified in this study.

This study has several strengths. For the first time, we investigated the association between occupational factors and disability-free survival and all-cause mortality in a sample from a healthy general population aged 70 years and older, in both men and women. The ASPREE study was subject to rigorous quality control, and all endpoints were actively screened, and verified by access to clinical records, which allowed the identification of the proximal cause of death. We used FINJEM, which is designed for a general population-based study. The advantage of using JEM is that it excludes recall bias and minimizes differential misclassification. In addition, the FINJEM period allows us to consider exposure at the time of employment. Finally, we adjusted for smoking as a confounder in the final models and performed a range of sensitivity analyzes to overcome the effect of residual confounding in the analyses.

Limitations included the restriction of occupational histories to the three longest-held jobs rather than a complete job history. Increased risks among women are likely to be underestimated as a result of their shorter duration in the workforce in this cohort (23). Alcohol intake is a potential confounder but was not included in the models because of the likely low validity of self-reports, its close correlation with smoking, and the very small number of individuals in this population declaring a high alcohol intake. FINJEM classifies exposures based on job titles only and cannot account for variations in characteristics of jobs with the same title. Being based on Finnish occupational data, the use of FINJEM to assess occupational factors, such as adverse social climate, might not necessarily reflect Australian working conditions.

It should be noted that the current investigation considered all Australian ASPREE participants, irrespective of their treatment allocation to either low-dose aspirin or placebo in the main intervention trial. The decision to include all Australian participants was made in light of the widespread use of aspirin in the population, with 7.2% of Australian participants being regular users prior to the trial (24). Nevertheless, sensitivity analyses demonstrated that adjusting for additional confounding factors had no impact on the overall effect estimates, as presented in Supplementary Table S4. A second analysis in which the results were presented adjusted for aspirin use in the trial allocation (data not shown) also found no discernible differences between the two groups.

In summary, this study reports the impact of occupational characteristics on disability-free survival among a large older initially healthy population recruited from the Australian community. Among the various ISCO-88 job categories, only those belonging to the least skilled ‘elementary’ occupations experienced a reduction in disability-free survival. Among the specific job characteristics, work in occupations with a high accident risk or in an adverse social climate were both associated with reduced disability-free survival, and potentially contributed to health disadvantage. All-cause mortality was not substantially increased among any of the occupational groupings or with any specific occupational characteristics. This may reflect survivor bias among individuals in this age group, especially those selected for their health at baseline.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Monash University Ethics Committee (project numbers CF11/1100 and CF11/1935). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

SA, GB, and JM conceived and designed the study. AO had significant involvement in data collection. KR and CT completed the ISCO-88 coding. SA conducted the cross work to Finnish-O codes and performed data analysis, interpretation, producing the initial draft of the manuscript. SA and JM approved the final version of the manuscript for publication and jointly took responsibility for the accuracy and integrity of all aspects of the research. All authors contributed to the article and approved the submitted version.