Unintended Perinatal Health Consequences Associated With a Swedish Family Policy

Key Points Question What is the association of a Swedish family policy protecting parental leave payment levels in short birth intervals with perinatal health outcomes? Findings In this cross-sectional study using an interrupted time series analysis of 1 762 784 births in Sweden, stricter birth spacing requirements for protected parental leave benefits were associated with significantly increased odds of preterm births and low birth weight but not of small for gestational age at preterm or stillbirths. Subsequently relaxed birth spacing requirements were associated with decreased odds of preterm births and low birth weight. Meaning Family policies aiming to provide socioeconomic protections may motivate changes in fertility behaviors with negative consequences for perinatal health.

eMethods.Interrupted time series methods with aggregated data eTable 1. Interrupted time series results for all perinatal health outcomes eTable 2. Interrupted time series results for all perinatal health outcomes, adjusted for seasonality eTable 3. Interrupted time series results for low birthweight at term eTable 4. Interrupted time series results for preterm births, by mother's country or region of birth eTable 5. Interrupted time series results for preterm births, by parity eTable 6. Interrupted time series results for preterm births, using gestational age calculated with last menstrual period information only eTable 7. Interrupted time series results for preterm births, excluding extremely preterm births before week 29 eTable 8. Interrupted time series results for all perinatal health outcomes, using aggregated data eFigure 1. Interrupted time series model of low birthweight at term eFigure 2. Interrupted time series models of preterm births, by mother's country or region of birth eFigure 3. Interrupted time series models of preterm births, by parity eFigure 4. Descriptive monthly rate of births at age 40 or above (Jan 1974-Dec 1991)  and monthly rate of pregnancies with diabetes (Jan 1974-Dec 1989) eFigure 5. Interrupted time series model of preterm births, using gestational age calculated with last menstrual period information only eFigure 6. Interrupted time series model of preterm births, excluding extremely preterm births before week 29 This supplementary material has been provided by the authors to give readers additional information about their work.

eMethods. Interrupted time series methods with aggregated data
After aggregating data by birth month, we again employed an interrupted time series design with multiple treatment periods. 1,2Based on the two reform dates (January 1980 and January 1986), the analyses were divided into three periods spanning a total of six years (January 1974-December 1979; January 1980-December 1985; and January 1986-December 1991) covering the three phases relative to changes in the speed premium policy as described in the paper.The data used and the analyzed outcomes are the same as in the main analysis.
We modeled the monthly rate of each outcome by fitting a linear ordinary least squares regression. 3hus, the regression coefficients reflect the average percentage-point change in the outcome rate from one month to the next.We used the Cumby-Huizinga test (actest command) to detect the appropriate autocorrelation structure per outcome. 4Models were adjusted for seasonality when 12month autocorrelation was detected by fitting Fourier terms (i.e., pairs of sine and cosine functions). 2 Confidence intervals based on lag-specific Newey-West standard errors were estimated to account for residual autocorrelation.
All statistical analyses were conducted using Stata Version 16.

eFigure 4 .
Descriptive monthly rate of births at age 40 or above (Jan 1974-Dec 1991) and monthly rate of pregnancies with diabetes (Jan 1974-Dec 1989) Panel A: Births to mothers age 40 or above (Jan 1974-Dec 1991); B: Births to mothers diagnosed with maternal diabetes before or during pregnancy (Jan 1974-Dec 1989).Dots indicate the observed rates while the solid black line represents the fitted linear trend.The dashed vertical lines mark the introduction of the 24-month speed premium in January 1980 and the later change to 30 months in January 1986.eFigure 5. Interrupted time series model of preterm births, using gestational age calculated with last menstrual period information only Dots indicate the observed probability, solid lines the estimated probability, and shaded areas the 95% confidence intervals (CI) of the outcome from the individual-level interrupted time series analysis (see eTable 6).The dashed vertical lines mark the introduction of the 24-month speed premium in January 1980 and the later change to 30 months in January 1986.eFigure 6. Interrupted time series model of preterm births, excluding extremely preterm births before week 29 Dots indicate the observed probability, solid lines the estimated probability, and shaded areas the 95% confidence intervals (CI) of the outcome from the individual-level interrupted time series analysis (see eTable 7).The dashed vertical lines mark the introduction of the 24-month speed premium in January 1980 and the later change to 30 months in January 1986.

8. eTable 1. Interrupted time series results for all perinatal health outcomes Preterm Low birthweight Small-for- gestational-age at preterm Stillbirth
1. Results are reported in eTable Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of the outcome of interest (see Figure3for time series plots).*Cumulative change (and 95% CI) over the respective 6-year periods: e.g., between January 1980 and December 1985, the odds of preterm birth increased 0.3% every month, amounting to a 26.35% increase over the full 71 months (1.0033^71=1.2635).eTable 2.

Interrupted time series results for all perinatal health outcomes, adjusted for seasonality
*Cumulative change (and 95% CI) over the respective 6-year periods: e.g., between January 1980 and December 1985, the odds of preterm birth increased 0.3% every month, amounting to a 25.20% increase over the full 71 months (1.0032^71=1.2520).eTable 3.

Interrupted time series results for low birthweight at term
Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of low birthweight (see eFigure 1 for time series plot).©2024 Debiasi E et al.JAMA Pediatrics.eTable 4.

Interrupted time series results for preterm births, by mother's country or region of birth
Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of preterm birth by mothers' country or region of birth (see eFigure 2 for time series plots).eTable 5.

Interrupted time series results for preterm births, by parity
Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of preterm birth by parity (see eFigure 3 for time series plots).eTable 6.

Interrupted time series results for preterm births, using gestational age calculated with last menstrual period information only
Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of preterm birth (see eFigure 5 for time series plot).eTable 7.

Interrupted time series results for preterm births, excluding extremely preterm births before week 29
Data are odds ratios with 95% confidence intervals (CI) indicating the monthly percentage change in odds of preterm birth (see eFigure 6 for time series plot).eTable 8.

Interrupted time series model of low birthweight at term
Dots indicate the observed probability, solid lines the estimated probability, and shaded areas the 95% confidence intervals (CI) of the outcome from the individual-level interrupted time series analysis (see eTable 3).The dashed vertical lines mark the introduction of the 24-month speed premium in January 1980 and the later change to 30 months in January 1986.

Interrupted time series models of preterm births, by parity
Panel A: Preterm birth with Parity 1; B: Preterm birth with Parity>1.Dots indicate the observed probability, solid lines represent the estimated probability, and shaded areas represent the 95% confidence intervals (CI) of the outcome from the individual-level interrupted time series analysis (see eTable 5).The dashed vertical lines mark the introduction of the 24-month speed premium in January 1980 and the later change to 30 months in January 1986.