The effect on women's health of extending parental leave: a quasi-experimental registry-based cohort study

Abstract

Background: Parental leave policies have been hypothesized to benefit mothers' mental health. We assessed the impact of a 6-week extension of parental leave in Denmark on maternal mental health.

Methods: We linked individual-level data from Danish national registries on maternal sociodemographic characteristics and psychiatric diagnoses. A regression discontinuity design was applied to study the increase in parental leave duration after 26 March 1984. We included women who had given birth between 1 January 1981 and 31 December 1987. Our outcome was a first psychiatric diagnosis following the child's birth, ascertained as the first day of inpatient hospital admission for any psychiatric disorder. We presented cumulative incidences for the 30-year follow-up period and reported absolute risk differences between women eligible for the reform vs not, in 5-year intervals.

Results: In all, 291 152 women were followed up until 2017, death, emigration or date of first psychiatric diagnosis. The median follow-up time was 29.99 years, corresponding to 10 277 547 person-years at risk. The cumulative incidence of psychiatric diagnoses at 30 years of follow-up was 59.5 (95% CI: 57.4 to 61.6) per 1000 women in the ineligible group and 57.5 (95% CI: 55.6 to 59.4) in the eligible group. Eligible women took on average 32.85 additional days of parental leave (95% CI: 29.20 to 36.49) and had a lower probability of having a psychiatric diagnosis within 5 years [risk difference (RD): 2.4 fewer diagnoses per 1000 women, 95% CI: 1.5 to 3.2] and up to 20 years after the birth (RD: 2.3, 95% CI: 0.4 to 4.2). In subgroup analyses, the risk reduction was concentrated among low-educated, low-income and single women.

Conclusions: Longer parental leave may confer mental health benefits to women, in particular to those from disadvantaged backgrounds.

Keywords: Parental leave; mental disorders; quasi-experiment; social determinants of health; women’s health.

Figure 1

Study population. Women might have given birth in either or both control and treated groups during our time frame. We include robust standard errors clustered at the mother’s level to account for this issue in all models, and show that our findings are robust to a sensitivity analysis focusing on the firstborn child (see Supplementary Appendix 3, available as Supplementary data at IJE online for additional details about our model)

Figure 2

Average days of maternity leave taken by mothers before and after the reform. Each dot represents the mean number of days of birth-related leave taken for all children born on a specific day. The vertical line indicates 26 March 1984, the cut-off for eligibility for the reform

Figure 3

Cumulative incidence of psychiatric disorder diagnoses. RD refers to risk difference. Cumulative incidence curves are presented for the probability of psychiatric diagnostic during the 30-year follow-up. Risk is expressed per 1000 women

Figure 4

Cumulative incidence of psychiatric disorder diagnoses by subgroups of interest. RD refers to risk difference. Cumulative incidence curves are presented for the probability of psychiatric diagnostic during the 30-year follow-up. Risk is expressed per 1000 women