The importance of both workplace and private life factors in psychological distress: a large cross-sectional survey of French railway company employees
- First Online:
- Cite this article as:
- Evans, D., Mallet, L., Flahault, A. et al. Soc Psychiatry Psychiatr Epidemiol (2013) 48: 1211. doi:10.1007/s00127-012-0605-7
- 248 Views
The psychological well-being of employees is a priority in occupational health. This study aimed to estimate the prevalence of psychological distress among employees of a large French company, to calculate the associations between distress and stressors in the workplace and private life domains, and to explore confounding across stressor domains.
8,058 employees of the French national railways company completed a nation-wide survey in 2006 (94.3 % participation). Psychological distress was measured by the 12-item General Health Questionnaire and 21 potential stressors and socio-demographic factors by a self-administered questionnaire. Stressors were summarized in scores for work pressure, workplace conflict, and personal life domains. Risk ratios (RRs) between psychological distress and stressors were calculated using robust-variance Poisson regression.
The prevalence of psychological distress was 32.8 % (95 % CI 31.8–33.9 %), higher among women (48.9 %, 95 % CI 46.5–51.7 %) than men (30.1 %, 95 % CI 29.0–31.2 %). Each stressor domain was associated with distress in the final model containing likely confounders and all three domains (RR highest vs. lowest level—work pressure: men 1.55, 95 % CI 1.42–1.70, women 1.42, 95 % CI 1.23–1.63; work conflict: men 2.63, 95 % CI 2.38–2.91, women 1.98, 95 % CI 1.70–2.30; life concerns: men 2.04, 95 % CI 1.86–2.23, women 1.53, 95 % CI 1.32–1.78). The mutually adjusted RRs for the stressor domains were smaller than the unadjusted RRs.
Almost one-third of all employees and one-half of female employees experienced psychological distress. All three stressor domains were associated with psychological distress and adjustment reduced the association size, suggesting possible over-estimation if one or more domains are omitted from the survey.