Statistics from Altmetric.com
Occupational epidemiology studies in the 1980s and 1990s demonstrated a beneficial effect of physically active employment on risks of premature death,1 with HR of ~1.50 for sedentary employees (see the online supplement for a detailed discussion). However, that accepted wisdom was disputed in a recent meta-analysis which claimed that male employees faced ‘detrimental health consequences associated with high-level occupational activity, even after adjusting for relevant factors’.2 Such a view challenges existing physical activity guidelines and poses the intriguing problem why occupational activity might be bad for men but not women.
Here the author scrutinises the choice of articles for this meta-analysis, the characteristics of apparently hazardous activity and potential confounding by covariates, focusing on the male data. Possible reasons why women do not show an occupational paradox are discussed in the online supplemental material.
Choice of articles: search strategy
Articles for the meta-analysis were drawn from 13 countries, but 5 of 16 were written by the study authors, leading to some similarities in the determination of active individuals and choice of co-variates. Moreover, participants in 5 of 16 studies lived around Copenhagen, with a potential for subject overlap; the author does not believe that this was evaluated. Studies provided ~30 to 5668 male deaths over 3.3–50 years of follow-up, with many studies beginning when industry was already largely mechanised.
Classifying ‘active’ tasks
In contrast with the objective determinations of early occupational studies, active employment in recent articles was typically assessed by a simple questionnaire (table 1). Two reports discussed lifting, but otherwise, the effort did not exceed what might be anticipated in active leisure. Few employees reported heavy work, and thus, some studies included all except sedentary workers in their ‘active’ category. Five reports showed benefit from the occupational activity, five an adverse effect and six no effect. The earliest study cited,3 when heavy work was more widely prevalent, showed substantial protection against premature mortality, and negative effects were marked in studies covarying for leisure activity. Overall, the meta-analysis yielded a modest HR of 1.18 (95% CI 1.05 to 1.34), likely of limited clinical importance. Several studies reported similar adverse effects from ‘moderate’ and ‘heavy’ occupational activity, and risks were no greater for jobs that involved lifting. With two exceptions, HRs for heavy workers were lower for cardiac than for all-cause deaths.
Covariates: with particular attention to smoking
Most studies analysed included some covariates (table 2), but in terms of perhaps the most important, many only distinguished continuing and former smokers from never-smokers. The relative risk was 2.4 for those smoking 25 cigarettes/day, and several studies noted that smoking was associated with heavy employment (see the online supplement).
The (recent) ‘occupational paradox’
Has the nature of ‘heavy’ work changed since classical epidemiological enquiries or has the recent ‘occupational paradox’ arisen from the inadequate allowance for covariates? Holtermann and associates4 suggested six possible explanations of the phenomenon, all physical in nature: lack of worker control over physical demands, heavy lifting, inadequate recovery periods and elevation of 24-hour heart rates, persistent inflammation or too low an intensity of effort to promote health.
High-intensity effort can certainly precipitate a cardiovascular incident5; those affected are more likely than their peers to report involvement in strenuous physical activity over the previous 24 hours. Two studies noted occasional ‘heavy’ lifting,6 although the loads (>10 kg, >20 kg) were only moderate, and the affected employees had the same or lower HRs than other manual workers. Hazards were 10% greater in those with pre-existing heart disease, but ‘heavy’ workers generally showed no increase of cardiac deaths relative to their sedentary peers.
Where energy expenditures were ascertained, the added daily energy expenditure (~1000 kcal) was distributed over the day, with demands amounting to ~30% of maximal aerobic power,7 a level unlikely to cause a sustained elevation of blood pressure or cardiac death. Moreover, the ‘occupational paradox’ was associated with ‘moderate’ as well as ‘heavy’ work.
Possibly, ‘heavy’ workers think that they gain sufficient physical activity at work. However, such effort is insufficient to induce a training effect; they engage in less active leisure than their ‘sedentary’ peers and also have a somewhat smaller maximal aerobic power, although differences seem too small to explain an ‘occupational paradox’.
An inadequate allowance for covariates remains a potential explanation for the ‘occupational paradox.’ Cigarette consumption and socioeconomic status are major determinants of premature mortality, and both are associated with ‘heavy’ work. Further, it is notoriously difficult to make adequate measurements to eliminate their influence by covariance analysis. Two studies showed little change of risk when smoking was introduced into the predictive model, but in other reports, cardiac and all-cause deaths were reduced.
Socioeconomic status was evaluated in 13 of the 16 studies, usually as a 3–5 level categorisation that is unlikely to have captured all of the associated variances. Early classical studies showing benefit from occupational activity had the important advantage of making comparisons within a single work force, thus, minimising the impact of socioeconomic status and related lifestyle characteristics.
The epidemiologists of the 1980s and earlier pointed out the favourable association between heavy occupational work on cardiac health, but a third of recent studies report a negative association. The intensity of physical effort in modern industry seems unlikely to have caused harm. The author argue that the probable explanation of this ‘recent occupational paradox’ lies in the inadequate classification of occupational demands by simple questionnaires and incomplete adjustments for covariate effects, particularly from cigarette smoking and low socioeconomic status.
Contributors RJS is the sole contributor.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no additional unpublished data relevant to this paper.
Patient consent for publication Not required.
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.