Perinatal deaths are falling across 17 poorer countries, but unevenly

Stillbirths and deaths in the first days of life are among the hardest events in global health to count, and the ones easiest to lose. A new modelling study that deliberately corrects for that undercounting finds that stillbirth, early neonatal and perinatal mortality have all declined across 17 low- and middle-income countries since 1990 — by a pooled 0.60, 0.50 and 1.08 deaths per 1,000 births per year respectively — but that the gains are spread unevenly, tracking wealth, maternal education and where a woman lives [s1]. The declines are real; the inequality inside them is the story.

The context for why this is difficult: worldwide, an estimated 2.0 million babies were stillborn in 2019, and stillbirths have been falling more slowly than deaths among young children [s2]. Measuring that progress honestly means confronting how badly surveys record it.

The measurement problem the study is built around

In countries without complete civil registration, household surveys are the main source of data on births and deaths — and they systematically distort exactly the events this study measures. The authors analysed 93 Demographic and Health Surveys with reproductive-calendar data from 17 LMICs, conducted between 1990 and 2023 and covering 1,019,253 pregnancies reported by 733,181 women [s1].

Three biases recur in that kind of data, and the study models each one. Stillbirths are underreported outright. They are misclassified — a stillbirth and a death on the first day or two of life are easy to swap in a mother's retrospective account. And ages at death "heap": deaths get rounded to round-number days, often reported at seven days at the expense of the days around it [s1]. To adjust, the authors restricted the analysis to pregnancies of at least seven months' gestation, in line with the standard stillbirth definition, and applied a modified Gompertz-Makeham model built to absorb underreporting, day 0-1 misclassification and age-at-death heaping [s1].

That is the point of the paper. A raw survey number understates stillbirths and blurs the line between late fetal and early neonatal death; the correction is what lets the trend be trusted.

What the trends show

Across the 17 countries, stillbirth, early neonatal and perinatal mortality generally declined over time, though the pace varied substantially between settings [s1]. The pooled annual declines were 0.60 stillbirths per 1,000 births (95% CI 0.41-0.78), 0.50 early neonatal deaths per 1,000 (95% CI 0.35-0.66), and 1.08 perinatal deaths per 1,000 (95% CI 0.76-1.40) [s1]. (These are reported in the source as negative annual changes; the chart shows their magnitudes, with the direction — decline — in the title.)

The inequalities were consistent. Lower perinatal mortality was associated with urban residence, higher maternal education, wealthier households and higher national Human Development Index levels, while maternal age of 30 years or older was associated with higher risk [s1]. These are the familiar axes of health inequality, and finding them intact after the statistical correction suggests they are features of the underlying reality, not artefacts of who gets counted.

How it fits the global picture

The single-study declines sit inside a slower global trend. The UN Inter-agency Group for Child Mortality Estimation put the 2019 global stillbirth count at 2.0 million (90% uncertainty interval 1.9-2.2), a rate of 13.9 per 1,000 total births (90% UI 13.5-15.4), with enormous regional spread — from 22.8 per 1,000 in west and central Africa to 2.9 in western Europe [s2]. Crucially, that assessment found stillbirths falling more slowly than child deaths: a global annual rate of reduction of 2.3% from 2000 to 2019, against 2.9% for neonatal mortality and 4.3% for mortality in children aged 1-59 months [s2].

That gap is the durable finding across both papers. Stillbirths respond to the quality of care around pregnancy and birth — care that is harder to deliver, and slower to improve, than the vaccination and child-survival interventions that have driven down deaths after the first month. The distribution of progress mirrors the distribution of that care, which is the same inequality tracked in our report on maternal mortality trends and, in early childhood, in the gap between nutrition gains and stunting outcomes.

The limits

The authors are explicit that their corrections reduce but do not remove the problem: residual underreporting of pregnancy losses is possible, and the surveys carry no information on causes of death, so the study can show that perinatal mortality fell without saying which causes drove the decline [s1]. The global estimates, for their part, rest on modelling for many country-years with little direct data [s2]. Both are honest about resting on imperfect inputs — which is precisely why the modelling matters, rather than pretending the raw counts speak for themselves.

What to watch

Whether investment shifts toward the quality of intrapartum and immediate newborn care, where the slow-moving stillbirth curve is decided. The measurement fix in this study is also a warning: any country claiming rapid progress on stillbirths from unadjusted survey data should be read with the same scepticism the authors applied to their own.

Sources

• Trends of stillbirth, early neonatal and perinatal mortality rates and sociodemographic differentials of perinatal mortality rates in 17 low- and middle-income countries: A modeling study — PLOS Medicine, 2026-08-24
• Global, regional, and national estimates and trends in stillbirths from 2000 to 2019: a systematic assessment — The Lancet, 2021-08-26