Does work-to-family conflict really matter for health? Cross-sectional, prospective cohort and fixed-effects analyses.

It is well known that work-to-family conflict (WFC) is negatively associated with employees' health outcomes, including mental health and health behavior. However, the associations may be overstated because of insufficient control for unobserved individual attributes. To address this possibility, we compared the associations between WFC and health observed from a cross-sectional, prospective cohort and from fixed-effects regression models. We analyzed data from a Japanese occupational cohort survey of 15,102 observations from 7551 individuals (5947 men and 1604 women), which were collected in two waves with a one-year interval. We constructed a binary variable of high WFC and considered psychological distress measured using the Kessler 6 (K6) score, job and life dissatisfaction, and five types of health behavior (current smoking, problem drinking, leisure-time physical inactivity, sickness absence, and refraining from medical care). Results showed that for men, a high WFC increased the probability of reporting psychological distress (K6 score ≥ 5); this increased by 12.4% in a fixed-effects model. The association was substantially limited, as compared to the increase of 30.9% and 23.2% observed in cross-sectional and prospective cohort models, respectively; however, the association remained significant. Similar patterns were observed for job and life dissatisfaction. In contrast, the associations of WFC with all five types of health behavior were non-significant after controlling for fixed effects. We obtained generally similar results for women and found no substantial gender difference in the fixed-effects models. We concluded that the associations of WFC with employees' mental health and subjective well-being were robust, whereas the association between WFC and health behavior was generally limited.