Job strain and ischemic heart disease: the balance of methodological bias and implications for prevention. Response to: Bonde JP et al. The demands-control-support work stress model and risk of ischemic heart disease: causal inference based on observational epidemiology.
Mathilde Lavigne-Robichaud, Paul Landsbergis, et al. • Scandinavian Journal of Work, Environment and Health • 2026
The balance of methodological bias in the job strain and ischemic heart disease literature is more plausibly downward than unpredictable, suggesting the pooled relative risk estimate of 1.14 is most likely underestimated.
Key Findings
Results
The pooled relative risk estimate (RRE) of 1.14 for job strain and ischemic heart disease is most likely underestimated due to common limitations in the available literature that tend to bias results toward the null.
Acknowledged sources of underestimation include nondifferential exposure misclassification, overadjustment for cardiometabolic risk factors, and healthy-worker survivor selection.
Dichotomizing exposure by combining active and passive exposures into a single 'non-high-strain' reference category may attenuate risk estimates by increasing heterogeneity in the reference group.
Workers with passive job exposure may also be at increased IHD risk, further undermining the validity of this reference category.
Sex-based attenuation may occur because women develop IHD at older ages partly due to pre-menopausal estrogen cardioprotection, meaning working-age follow-up captures fewer events, reducing pooled estimates and precision.
Results
Job-exposure matrix (JEM) studies do not fully capture individual-level variability in exposure within occupational categories, meaning lower JEM estimates should not be interpreted as evidence of upward bias in self-reported studies.
Even when exposure values are imputed within subgroups defined by sex and age, JEMs do not fully capture individual-level variability in exposure within occupational categories.
The resulting non-differential misclassification in JEM studies likely attenuates estimates.
Bonde et al. acknowledge this limitation but do not take it into account in their conclusion.
Results
Negative affectivity as a proposed source of overestimation of the job strain–IHD association is not supported in prospective studies with control for anger, hostility, and cynicism.
Negative affectivity is the mechanism by which adverse health perceptions could inflate individual-level exposure reports.
Prospective studies controlling for anger, hostility, and cynicism do not support this mechanism as a source of overestimation.
Two referenced studies support this conclusion (Lavigne-Robichaud et al. 2023; Tiwa Diffo et al. 2024).
Results
Health-reporting bias (reverse causation), whereby workers with prodromal IHD symptoms may over-report perceived job strain, is not supported as a source of overestimation in prospective studies.
In prospective studies excluding early incident IHD events, associations were not attenuated and, if anything, marginally strengthened.
This finding provides no support for reverse causation as a source of overestimation.
Two referenced studies support this conclusion (Lavigne-Robichaud et al. 2023; Kivimäki et al. 2012).
Results
A prospective cohort study designed to address multiple sources of underestimation estimated that 18.2% of incident IHD cases were attributable to job strain exposure.
This estimate contrasts sharply with the 3.4% population attributable fraction (PAF) reported by the IPD-Work Consortium.
The study was specifically designed to address sources of underestimation including exposure misclassification, alternative referent group definitions, and other attenuation sources identified in the literature.
The referenced study is Lavigne-Robichaud et al. 2025, published in JACC Advances.
There has been considerable debate about whether the IPD-Work Consortium PAF of 3.4% was itself an underestimate.
Discussion
Methodological characteristics in the reviewed literature are presented as isolated binary indicators rather than as interdependent dimensions, which hinders overall appraisal of study quality.
This critique is directed at the Bonde et al. review's approach to evaluating study methodology.
The authors argue this framing prevents a valid appraisal of the available evidence.
The supporting reference for overestimation used by Bonde et al. relies on a 4-item measure of perceived stress, limiting its relevance to job strain specifically.
Conclusions
Debates about the precise magnitude of the job strain–IHD association should not delay the development and evaluation of workplace interventions to reduce job strain and improve cardiovascular health.
The authors argue this conclusion follows from the substantial burden of IHD.
The balance of methodological bias is characterized as more plausibly downward than unpredictable.
This recommendation is framed in the context of the methodological debate having implications for burden estimation and prevention policy.
What This Means
This research is a formal scientific response to a review paper that concluded work stress (specifically 'job strain' — having high demands but low control at work) has at most a small effect on the risk of heart disease. The authors of this response argue that the evidence has been systematically underestimated due to a number of well-documented methodological problems. These include how job stress is measured and categorized, the fact that women develop heart disease later in life (meaning studies often miss many of their cases), and the use of crude comparison groups that mix together workers with very different actual stress levels. When these biases are accounted for, the true effect of job strain on heart disease risk is likely meaningfully larger than the 14% increased risk reported in the pooled analysis.
The authors also push back on the claim that previous studies may have overestimated the job strain–heart disease link. Two commonly raised concerns — that stressed people are more likely to exaggerate their stress levels (negative affectivity), and that people who are already developing heart disease symptoms might report more stress (reverse causation) — are not supported by the best available prospective research. When studies have specifically accounted for these factors or excluded people who developed heart disease early in follow-up, the associations either stay the same or get slightly stronger. A study designed to correct for multiple sources of underestimation found that as many as 18.2% of heart disease cases could be attributable to job strain, compared to the 3.4% figure from an earlier large consortium analysis.
This research suggests that the scientific debate over exactly how large the job strain–heart disease connection is should not be used as a reason to delay action. Given that heart disease remains a leading cause of death and disability, and given that the weight of methodological evidence suggests the true effect is likely being underestimated rather than overestimated, this paper argues there is sufficient basis to develop and test workplace programs aimed at reducing job strain to protect workers' cardiovascular health.
Mathilde Lavigne-Robichaud, Paul Landsbergis, Chantal Brisson, Grace Sembajwe, M. Gilbert-Ouimet, Jian Li, et al.. (2026). Job strain and ischemic heart disease: the balance of methodological bias and implications for prevention. Response to: Bonde JP et al. The demands-control-support work stress model and risk of ischemic heart disease: causal inference based on observational epidemiology.. Scandinavian Journal of Work, Environment and Health. https://doi.org/10.5271/sjweh.4315