Brazil's childbirth haemorrhage deaths have not fallen in a decade

Obstetric haemorrhage killed 2,985 women in Brazil between 2015 and 2024, accounting for 17.2% of all maternal deaths at a rate of 10.86 per 100,000 live births — and a nationwide analysis published on 6 August found no significant change in that rate across the decade [s1].

The data

The study is a nationwide population-based cross-sectional analysis drawing on two Brazilian national systems: the Mortality Information System (SIM) and the Live Birth Information System (SINASC) [s1]. Deaths were classified into three groups by cause: first-half-of-pregnancy haemorrhage, placental or intrapartum haemorrhage, and postpartum haemorrhage [s1].

The distribution was relatively even. Postpartum haemorrhage accounted for 38.96% of deaths, placental or intrapartum haemorrhage for 33.74%, and first-half-of-pregnancy haemorrhage for 27.30% [s1]. That spread is worth noting on its own: haemorrhage prevention in maternal health is often discussed as a postpartum problem, and in this dataset postpartum causes account for well under half the deaths.

The disparities

The gradients across population groups are the study's central finding, and they are steep.

Indigenous women had a 2.87-fold higher prevalence of haemorrhagic maternal death than White women (95% CI 2.23–3.70), and Black women a 1.66-fold higher prevalence (95% CI 1.44–1.92) [s1]. Women aged 40 and over faced nearly four times the risk of women aged 20 to 29 (prevalence ratio 3.90; 95% CI 3.38–4.51) [s1].

The steepest gradient was educational. Women with no formal education had almost ten times the risk of women with 12 or more years of education (PR 9.68; 95% CI 7.46–12.56) [s1]. Geographically, Brazil's North region showed the highest excess risk (PR 1.84; 95% CI 1.56–2.15) [s1].

These variables overlap heavily with one another, and the available reporting describes them as prevalence ratios by group rather than as an adjusted model isolating each one. Education, region, and race are not independent in Brazil, so a tenfold educational gradient and a 2.87-fold Indigenous gradient are almost certainly describing overlapping populations rather than additive risks.

What "no significant temporal change" means

The flat trend is arguably the more consequential result. The study opens by describing obstetric haemorrhage as a leading preventable cause of maternal mortality worldwide [s1] — and a preventable cause whose national rate does not move across ten years is a statement about a health system, not about biology. Whatever changed in Brazilian obstetric practice over that decade did not change this outcome at national scale [s1].

The study does not attempt to explain why. Its conclusion frames the requirement in two parts — strengthening evidence-based haemorrhage care and addressing persistent racial, educational and regional inequalities [s1] — which reads as a judgement that the clinical toolkit alone has not been the binding constraint.

Limits of the design

This is a cross-sectional analysis of national administrative records, and it inherits their limitations. Maternal deaths are identified through death certificate coding, so both the count and the cause assignment depend on how consistently certificates are completed and coded — variation the study's design cannot correct for.

Race in SINASC and SIM is recorded categorically, and the Indigenous category in particular covers a small absolute number of deaths, which is reflected in the wide confidence interval around the 2.87 ratio (2.23–3.70) [s1]. The North region result likewise cannot separate geography from the population composition and health-service density within it.

Most importantly, this is a study of deaths, not of care. It cannot say at what point in the pathway — antenatal detection, delivery setting, availability of blood products, time to referral — the failures occurred, because those variables are not in the datasets used.

What the pattern points at

The combination the study documents — a flat national rate alongside very large between-group differences [s1] — is the signature of variation in care rather than of a uniform clinical limitation. If haemorrhage deaths were being driven mainly by the intrinsic difficulty of treating catastrophic bleeding, the risk would not scale so sharply with a mother's education or the region she lives in. The authors' framing of the remedy in two parts, clinical and structural [s1], follows from that.

What to watch

Whether Brazil's maternal mortality surveillance produces facility-level or process-level analyses that can locate the failures this study can only bound, and whether any intervention targeting the North region or Indigenous maternity care shifts the ratios reported here.

Sources

• Maternal mortality from obstetric hemorrhage in Brazil: a 10-year analysis (2015 to 2024) — The Journal of Maternal-Fetal & Neonatal Medicine, 6 August 2026