Stillbirths and Newborn Deaths Are Falling in 17 Countries, but Not Evenly

Perinatal mortality fell across 17 low- and middle-income countries by a pooled 1.08 deaths per 1,000 births each year (95% CI −1.40 to −0.76) over three decades, according to a study published Aug. 24 in PLOS Medicine that re-analyzed household survey data after correcting it for known reporting errors.

The authors used 93 Demographic and Health Surveys carried out between 1990 and 2023 in 17 countries, covering 1,019,253 pregnancies reported by 733,181 women. Only pregnancies of at least seven months' gestation were included, to align with the standard definition of stillbirth. The paper is by Mohamed M. Ali of WHO's reproductive health department in Geneva, Luca Cegolon of the University of Trieste and Saverio Bellizzi of the United Nations University.

Where death registration is incomplete, household surveys are a primary source of data on stillbirths and early neonatal deaths, but the paper notes that these deaths are often underreported. The team applied a statistical model that adjusts for three specific faults: under-reporting, confusion between a stillbirth and a death in the first two days of life, and the habit of recording a newborn's age at death as exactly seven days.

Broken out, the pooled annual decline was 0.60 deaths per 1,000 births for stillbirths (95% CI −0.78 to −0.41) and 0.50 for deaths in the first week of life (95% CI −0.66 to −0.35). Country trajectories differ widely: Bangladesh's perinatal rate fell 58%, from 97.0 to 40.8 per 1,000 births between its first and last surveys, and Colombia's fell 70%, from 39.9 to 12.2.

Lower perinatal mortality was associated with urban residence, more maternal education, wealthier households and a higher national Human Development Index, while a mother's age of 30 or over was associated with higher risk. Those are associations measured in survey data, not causes the study tested. The authors say residual under-reporting is still likely and that the surveys record no causes of death.

The DHS data are public, the analysis code is posted on GitHub and Zenodo, and the authors report no funding for the work.