J Stroke Search

CLOSE


J Stroke > Volume 27(1); 2025 > Article
Fadel, Sembajwe, Li, Leclerc, Pico, Schnitzler, Fadel, Roquelaure, and Descatha: Direct and Indirect Effects of Prolonged Exposure to Long Working Hours on Risk Stroke Subtypes in the CONSTANCES Cohort
Dear Sir:
Long working hours (LWH) have been identified as a leading cause of work-related mortality and morbidity [1]. Although their associations with health effects are low to moderate, the high prevalence of LWH makes them a significant occupational health risk [2]. Earlier evidence has suggested elevated risk of stroke among workers with prolonged exposure to LWH, but differentiating stroke subtypes and causal pathways have become a key focus of research [2]. Particularly, a couple of recent studies have highlighted a potential difference in association regarding ischemic and hemorrhagic stroke [3,4]. The aim of this study was to estimate the direct and indirect effects of exposure to LWH on the risk of ischemic and hemorrhagic stroke through cardiovascular risk factors using structural equation models.
Baseline data from CONSTANCES (CONSulTANts des Centres d'Examens de Santé), a French population-based cohort were used. Details on the cohort are available elsewhere [5]. All participants who had completed the baseline questionnaires and who were identified in the National Health Data System (Système National des Données de Santé, SNDS) were included. Participants with a personal history of stroke before their inclusion were excluded. All study participants provided written informed consent prior to enrollment with appropriate institutional review (Angers University Hospital Ethics Committee, approval #2021-186). For CONSTANCES, all procedures have been approved by the Institutional Review Board (IRB) of the French Institute of Health (Inserm) (Opinion n°01-011, then n°21-842), and authorized by the by the French Data Protection Authority (“Commission Nationale de l’Informatique et des Libertés,” CNIL) (Authorization #910486).
Age, sex, smoking habits, family and personal history of cardiovascular diseases, socio-professional status according to the French classification, and exposure to LWH were collected from the questionnaire at baseline. LWH were defined as self-reported working time ≥10 h/day for at least 50 days/year. Duration of exposure to LWH was categorized by a period of every 5 years. The initial health examinations by medical professionals and the linkage to SNDS were used to identify diabetes, dyslipidemia (hypercholesterolemia or hypertriglyceridemia), and high blood pressure status according to validated algorithms (Supplementary Table 1). The SNDS is a national healthcare database and contains data from the hospital discharge database and healthcare related reimbursement in France. Cases of stroke were retrieved during the follow-up period (until 2022) using the SNDS database and were categorized as hemorrhagic (I60-61-62 except for I604/I605/I621) or ischemic (I63-64 except for I63.6) using the International Classification of Diseases 10 (ICD-10). Only the first new case of stroke during the follow-up was considered.
Associations between LWH and stroke were examined using crude and adjusted multinomial logistic regression models, adjusting for cardiovascular variables. All analyses were stratified on sex. Missing data were imputed in a sensitivity analysis using multivariate imputation by chained equations. A conceptual structural equation model was created based on current knowledge between LWH and cardiovascular risk factors, which were considered as mediators in this analysis (Supplementary Figure 1). This method allows for studying the relation between variables, especially identifying mediating factors, testing hypothesized structures, and assessing complex relationships between variables through their correlations [6]. The aim is to construct, based on a hypothesized graph of measured and latent variables as well as their casual links, the linear model that most closely yields the observed matrix of cross-correlations between variables. The quality of the model is roughly reflected in how much correlation is explained by the model. Variables were standardized or binarized for categorical data. Adequation of models was estimated using usual indicators (root mean square error of approximation, standardized root mean square residual, comparative fit index, and Tucker-Lewis Index). Direct and indirect effects were estimated from the models and 95% confidence interval with bootstrapping. All analyses were carried out using R software (version 4.3.0; R Foundation for Statistical Computing, Vienna, Austria) with the following packages: compareGroups, mice, lavaan, and manymome.
From 2012 to 2022, 174,925 participants were free of stroke at baseline, while 1,507 were excluded for their own medical history of stroke. A flow diagram is available in Supplementary Figure 2. There were 463 ischemic stroke cases and 183 hemorrhagic stroke cases during the follow-up period. A full description of the population and the results of the models is provided in Supplementary Table 2. In the male subgroup, only hemorrhagic stroke was associated with LWH for durations of exposure between 5-10 years and 10-15 years, with adjusted odds ratios of 2.75 (1.38-5.49) and 2.36 (1.11-5.02), respectively (Table 1). No significant association with LWH was found in the female subgroup. Age and high blood pressure were associated with both stroke subtypes and both sex subgroups. Results were similar in the multiple imputation models (Supplementary Table 3). All structural equation models were well or very well identified. The total effect of LWH on the risk of ischemic stroke and hemorrhagic stroke was significant in the male subgroup (Figure 1), 70% of the total effect being indirect for ischemic stroke, whereas 60% of the total effect was direct for hemorrhagic stroke, though this was close to significance. The total effect was not significant in the female subgroup, in contrast to the male subgroup for both ischemic and hemorrhagic stroke.
This study explores the pathways of the modest but robust effect of LWH on the risk of ischemic and hemorrhagic stroke, confirming a previously raised hypothesis of differential pathways [3]. In the male subgroup, there was a predominance of the indirect effect through cardiovascular risk factors for ischemic stroke contrary to hemorrhagic stroke.
This result is consistent with other studies in the literature, which have found that exposure to LWH increases harmful health behaviors [7-9], known risk factors for stroke and cardiovascular diseases in general. However, it also shows that direct effects, possibly through stress-related, pro-inflammatory, or hypercoagulability mechanisms, may play a key factor [10], though its role as a potential trigger still needs to be explored.
The differences between sexes could be related to the lower level of exposure in the female subgroup and related to hormonal status, which was not considered, in addition to possible difference in tasks outside work. Future studies will have to test these hypotheses by focusing on highly exposed women. Furthermore, the weaker effect of LWH on stroke for the ≥15 years of LWH exposure group may be caused by their higher cardiovascular risk factors frequency (Supplementary Tables 4 and 5).
Several limitations should be discussed. All cardiovascular risk factors, as well as exposure to LWH, were assessed at the inclusion, even though working conditions and health behaviors may change over time, and not all cardiovascular factors could be explored. However, we looked at lifetime exposure to LWH, any potential changes in exposures might lead to underestimating associations. Moreover, though there was an update of the cohort’s data, stroke remains a rare event, leading to a low number of events and statistical power in some subgroups, especially women.
There are important implications to these findings. Multilevel prevention seems relevant, and interventions focusing on both working conditions at the organizational level and cardiovascular risk factors at the individual level are warranted. Further research with longitudinal assessment of mediators, better inclusion of other cardiovascular and work-related factors, and focus on women is needed.

Supplementary materials

Supplementary materials related to this article can be found online at https://doi.org/10.5853/jos.2024.02586.
Supplementary Table 1.
Algorithms used to detect diabetes, high blood pressure, and dyslipidemia status using data from the CONSTANCES cohort and the National Health Data System (Système National des Données de Santé, SNDS)
jos-2024-02586-Supplementary-Table-1.pdf
Supplementary Table 2.
Description and univariate tests of the population included according to stroke subtypes
jos-2024-02586-Supplementary-Table-2.pdf
Supplementary Table 3.
Multivariable analysis according to stroke subtypes with imputation for missing data by multivariate imputation by chained equations
jos-2024-02586-Supplementary-Table-3.pdf
Supplementary Table 4.
Sensitivity analysis of cardiovascular risk factors in the LWH exposure subgroups
jos-2024-02586-Supplementary-Table-4,5.pdf
Supplementary Table 5.
Sensitivity analysis of cardiovascular risk factors in the LWH exposure subgroups
jos-2024-02586-Supplementary-Table-4,5.pdf
Supplementary Figure 1.
Conceptual models used for the structural equation model analyses. Latent variables are in ellipses and observed variables in rectangles.
jos-2024-02586-Supplementary-Fig-1,2.pdf
Supplementary Figure 2.
Flow diagram of the studied population.
jos-2024-02586-Supplementary-Fig-1,2.pdf

Notes

Funding statement
The CONSTANCES Cohort Study was supported and funded by the Caisse nationale d’assurance maladie (CNAM); it is an “Infrastructure Nationale en Biologie et Santé” and benefits from ANR (ANR-11-INBS-0002) grant funding. CONSTANCES is also partly funded by Merck Sharp & Dohme, AstraZeneca, Lundbeck and L’Oréal through Inserm-Transfert.
Conflicts of interest
The authors have no financial conflicts of interest.
Author contribution
Conceptualization: MF, AD, YR. Study design: MF, AD. Methodology: MF, AD. Statistical analysis: MF, ERF. Writing—original draft: MF. Writing—review & editing: GS, JL, AL, FP, AS, YR. Approval of final manuscript: all authors.

Acknowledgments

The data of the CONSTANCES cohort are protected by our national regulatory agency (“Commission nationale de l’informatique et des libertés“, n°910486). The CONSTANCES cohort is “an open epidemiological laboratory,” access to and data is available on request (http://www.constances.fr/index_EN.php#propose). Access to some confidential data, on which is based this work, has been made possible within a secure environment offered by CASD - Centre d’accès sécurisé aux données (Ref. 10.34724/CASD).

Figure 1.
Direct and indirect effect of long working hours on (A) ischemic stroke and (B) hemorrhagic stroke through cardiovascular factors for males.
jos-2024-02586f1.jpg
Table 1.
Multivariable multinomial regression analysis according to stroke subtypes
Male
Female
Ischemic stroke
Hemorrhagic stroke
Ischemic stroke
Hemorrhagic stroke
Adjusted OR (95% CI) Adjusted OR (95% CI) Adjusted OR (95% CI) Adjusted OR (95% CI)
Age (yr) 1.08 (1.06-1.09) 1.08 (1.05-1.10) 1.06 (1.04-1.08) 1.04 (1.01-1.07)
Body mass index
 25-30 kg/m2 (vs. <25 kg/m2) 0.81 (0.62-1.07) 0.90 (0.55-1.46) 1.14 (0.73-1.76) 0.86 (0.48-1.55)
 ≥30 kg/m2 (vs. <25 kg/m2) 0.72 (0.48-1.06) 0.61 (0.29-1.31) 1.71 (1.05-2.79) 0.89 (0.43-1.85)
High blood pressure status
 Yes (vs. No) 1.41 (1.06-1.86) 1.73 (1.04-2.89) 2.01 (1.33-3.03) 1.73 (1.00-3.03)
Diabetes status
 Yes (vs. No) 1.06 (0.69-1.63) 1.14 (0.53-2.46) 0.60 (0.22-1.66) 1.33 (0.46-3.84)
Dyslipidemia status
 Yes (vs. No) 1.06 (0.82-1.38) 1.13 (0.71-1.80) 1.07 (0.72-1.59) 1.46 (0.85-2.51)
Familial history of cardiovascular diseases
 Yes (vs. No) 1.29 (0.83-2.00) 0.73 (0.26-1.99) 1.11 (0.60-2.08) 1.53 (0.73-3.22)
Smoking
 <30 pack-years (vs. 0 pack-years) 1.09 (0.83-1.43) 0.85 (0.54-1.36) 1.04 (0.72-1.52) 1.13 (0.67-1.89)
 ≥30 pack-years (vs. 0 pack-years) 1.64 (1.10-2.46) 0.67 (0.27-1.63) 1.73 (0.79-3.83) 3.52 (1.52-8.14)
Occupation
 Self-employed/chief executive officer/professional (vs. High-skilled white-collar) 1.11 (0.82-1.52) 1.16 (0.67-2.03) 0.71 (0.43-1.19) 0.97 (0.48-1.94)
 Low-skilled white-collar (vs. High-skilled white-collar) 1.34 (0.88-2.05) 1.24 (0.56-2.74) 0.97 (0.64-1.47) 1.47 (0.81-2.66)
 Blue-collar (vs. High-skilled white-collar) 1.45 (1.00-2.11) 1.43 (0.72-2.85) 0.71 (0.25-1.99) 2.35 (0.87-6.36)
Long working hours exposure
 5-10 years (vs. <5 years) 0.87 (0.48-1.57) 2.75 (1.38-5.49) 1.38 (0.64-2.99) 2.25 (0.95-5.31)
 10-15 years (vs. <5 years) 1.18 (0.69-2.00) 2.36 (1.11-5.02) 0.29 (0.04-2.09) 1.16 (0.28-4.80)
 ≥15 years (vs. <5 years) 1.16 (0.85-1.58) 1.08 (0.58-1.99) 0.60 (0.24-1.48) 1.29 (0.51-3.31)
OR, odds ratio; CI, confidence interval.
* Multivariable models are adjusted for age, body mass index, high blood pressure, diabetes, dyslipidemia, smoking, and occupation.

References

1. Pega F, Náfrádi B, Momen NC, Ujita Y, Streicher KN, Prüss-Üstün AM, et al. Global, regional, and national burdens of ischemic heart disease and stroke attributable to exposure to long working hours for 194 countries, 2000-2016: a systematic analysis from the WHO/ILO joint estimates of the work-related burden of disease and injury. Environ Int 2021;154:106595.
crossref pmid pmc
2. Härmä M, Kecklund G, Tucker P. Working hours and health - key research topics in the past and future. Scand J Work Environ Health 2024;50:233-243.
crossref pmid pmc
3. Fadel M, Sembajwe G, Li J, Leclerc A, Pico F, Schnitzler A, et al. Association between prolonged exposure to long working hours and stroke subtypes in the CONSTANCES cohort. Occup Environ Med 2023;80:196-201.
crossref pmid
4. Hannerz H, Albertsen K, Burr H, Nielsen ML, Garde AH, Larsen AD, et al. Long working hours and stroke among employees in the general workforce of Denmark. Scand J Public Health 2018;46:368-374.
crossref pmid pdf
5. Goldberg M, Carton M, Descatha A, Leclerc A, Roquelaure Y, Santin G, et al. CONSTANCES: a general prospective population-based cohort for occupational and environmental epidemiology: cohort profile. Occup Environ Med 2017;74:66-71.
crossref pmid pmc
6. Amorim LD, Fiaccone RL, Santos CA, Santos TN, Moraes LT, Oliveira NF, et al. Structural equation modeling in epidemiology. Cad Saude Publica 2010;26:2251-2262.
crossref pmid
7. Baek SU, Lim MH, Kim T, Lee YM, Won JU, Yoon JH. Relationship between long working hours and smoking behaviors: evidence from population-based cohort studies in Korea. Scand J Work Environ Health 2024;50:257-267.
crossref pmid pmc
8. Baek SU, Won JU, Lee YM, Yoon JH. Association between long working hours and diet quality and patterns: a latent profile analysis of a nationally representative sample of Korean workers. Prev Med 2024;180:107890.
crossref pmid
9. Lee DW, Jang TW, Kim HR, Kang MY. The relationship between working hours and lifestyle behaviors: evidence from a population-based panel study in Korea. J Occup Health 2021;63:e12280.
crossref pmid pmc pdf
10. Virtanen M, Kivimäki M. Long working hours and risk of cardiovascular disease. Curr Cardiol Rep 2018;20:123.
crossref pmid pmc pdf


ABOUT JoS
AUTHOR INFORMATION
ARTICLE CATEGORY

Browse all articles >

BROWSE ARTICLES
Editorial Office
Department of Neurology, Asan Medical Center,Ulsan University College of Medicine
88, Olympic-ro 43-gil, Songpa-gu, Seoul 05505, Korea
Submission, status and progress, etc ⟫ E-mail: editor@j-stroke.org
Website and system ⟫ E-mail: journal@m2community.co.kr
Publishing company ⟫ E-mail: ka72sus@smileml.com
Developed in M2PI
Copyright © 2025 by Korean Stroke Society.
Close layer
prev next