Pregnancy has long been monitored with a simple blood test and a simple rule: if haemoglobin falls below a fixed threshold, the mother is diagnosed with anaemia and treated accordingly. A new analysis published in Nature Health suggests that this one-size-fits-all approach may be doing a disservice to millions of women and their babies. Using a rigorous meta-analytic framework designed to quantify both risk and the certainty of the evidence behind it, an international team of researchers has mapped the relationship between maternal haemoglobin concentration and a broad spectrum of pregnancy outcomes, and the resulting picture is anything but a straight line. Instead of a single cliff-edge threshold separating health from harm, the study reveals continuous, nonlinear curves in which risk climbs steadily as haemoglobin drifts away from an optimum, whether upward or downward, with deviations on the low end carrying particularly steep consequences.

The work belongs to the Burden of Proof family of studies, an approach that has gained prominence in recent years for the way it handles a chronic problem in medical research: overconfident conclusions drawn from heterogeneous evidence. Traditional meta-analyses typically pool risk estimates and report a single number, often accompanied by confidence intervals that can dramatically understate the true uncertainty when the underlying studies differ in design, population and quality. The Burden of Proof framework, developed to bring a level of transparency more familiar from grading systems in clinical guidelines, generates risk–response curves across the full range of exposure, applies corrections for bias and imprecision, and assigns star ratings that signal how much confidence readers should place in each finding. In this application, the exposure was maternal haemoglobin concentration measured during pregnancy, and the outcomes spanned the most consequential endpoints in obstetric medicine.

The list of outcomes examined reads like a catalogue of the leading causes of maternal and neonatal suffering worldwide. On the maternal side, the researchers modelled the relationship between haemoglobin and all-cause maternal mortality, postpartum haemorrhage and maternal sepsis, three conditions that together account for a substantial share of pregnancy-related deaths, particularly in low- and middle-income countries. On the neonatal side, they assessed all-cause neonatal mortality, preterm birth, low birth weight and large for gestational age, an outcome that has received far less attention in the haemoglobin literature but which the analysis suggests is also sensitive to maternal oxygen-carrying capacity. For each outcome, the team constructed dose–response curves that estimate how risk changes with every incremental shift in haemoglobin, rather than comparing broad categories such as “anaemic” versus “non-anaemic”.

What emerged from this modelling is a consistent pattern of nonlinearity. For outcomes where the evidence reached at least moderate strength, denoted by star ratings above two, the haemoglobin concentration associated with the lowest predicted risk ranged from approximately 109 to 135 grams per litre. That range is notable for how much it overlaps with levels that many clinical guidelines would still classify as mild anaemia. The World Health Organization defines anaemia in pregnancy as a haemoglobin concentration below 110 grams per litre, a cut-off that has remained essentially unchanged for decades and is applied uniformly regardless of trimester, geography or the specific outcome a clinician is trying to prevent. The new findings suggest that the risk landscape around that threshold is far more gradual and more individualized than the binary language of “anaemic” and “not anaemic” implies.

Perhaps the most striking feature of the results is the symmetry of harm. High haemoglobin, often dismissed as a benign finding or even a sign of good health, was also associated with elevated risk for several outcomes. This is biologically plausible: elevated haemoglobin in pregnancy can reflect haemoconcentration, reduced plasma volume expansion, chronic hypoxia, smoking, or underlying conditions such as pre-eclampsia, all of which may compromise placental function. At the same time, the analysis found that deviations on the low end of the distribution, the classical territory of anaemia, carried especially steep increases in risk for mortality and haemorrhage-related outcomes. The curve is therefore not a simple U with a broad flat bottom; it is an asymmetric landscape in which small departures in either direction matter, and departures downward matter more.

The implications of this asymmetry become vivid when one considers population-level mathematics. Because pregnancy haemoglobin distributions in many countries are centered close to the threshold region identified by the study, even modest shifts in the population distribution, the kind that could be produced by changes in iron supplementation policy, food fortification, or the prevalence of parasitic infections such as malaria and hookworm, could meaningfully alter the total burden of postpartum haemorrhage, preterm birth and neonatal death. A woman whose haemoglobin sits at 105 grams per litre rather than 115 is not merely crossing an administrative line; she is moving along a risk curve whose slope steepens as concentrations fall. Conversely, the findings caution against indiscriminate iron supplementation aimed at pushing haemoglobin ever higher, since the upper end of the curve carries its own hazards.

The star ratings attached to each outcome give clinicians and policymakers a way to calibrate their confidence. Findings supported by stronger evidence, indicated by ratings above two stars, included several of the mortality and birth-weight outcomes, meaning that the nonlinear associations survived the framework’s corrections for bias and heterogeneity. Outcomes with weaker evidence remain suggestive rather than definitive, and the authors are careful to frame their curves as the best current synthesis of available data rather than as immutable biological law. This honesty about uncertainty is itself a feature of the Burden of Proof approach, which was designed precisely to prevent the field from mistaking noisy pooled estimates for settled truth. In an area of medicine where interventions such as blood transfusion and aggressive iron therapy carry real risks of their own, knowing how much to trust a risk estimate is not an academic luxury.

Why has the field relied on fixed thresholds for so long? The answer is partly historical and partly practical. The original anaemia cut-offs were derived from studies conducted decades ago, often in populations and settings very different from those where they are applied today. A single threshold is also administratively convenient: it can be printed on a lab report, encoded in a clinical algorithm, and taught in a single sentence. But convenience comes at a cost. The new analysis demonstrates that the optimal haemoglobin concentration likely differs depending on which outcome one is trying to prevent, with the low-risk regions for preterm birth, low birth weight and maternal haemorrhage not perfectly overlapping, and it raises the prospect of thresholds that vary by trimester, reflecting the physiological dilution of haemoglobin that normally occurs as plasma volume expands through the second and third trimesters.

The study’s authors argue that their results support a move toward more refined, trimester-specific and outcome-specific guidance, a shift that would better identify individuals genuinely at elevated risk and spare others unnecessary intervention. Such a transition would not be trivial. It would require updating diagnostic criteria, retraining health workers, and in many settings expanding access to the repeated haemoglobin measurements that individualized care demands. But the potential payoff is substantial in a world where anaemia in pregnancy remains one of the most common medical conditions encountered in antenatal care, affecting an estimated large fraction of the roughly 140 million births that occur each year, and where the outcomes studied here remain leading causes of death and disability for mothers and newborns alike.

For researchers, the study also opens a methodological door. The Burden of Proof framework, with its explicit star ratings and bias corrections, could be applied to other continuous biomarkers in obstetrics, from ferritin and transferrin saturation to blood pressure and gestational weight gain, replacing categorical definitions with curves that respect the continuous biology underlying them. For the moment, the message for clinical practice is more modest but no less important: haemoglobin in pregnancy should be read not as a pass-or-fail number against a single threshold, but as a position on a risk gradient, with the safest ground lying roughly between 109 and 135 grams per litre and with danger increasing, quietly and nonlinearly, the further a patient strays from that range in either direction.

Subject of Research: Dose–response relationships between maternal haemoglobin concentration during pregnancy and maternal and neonatal outcomes, assessed with the Burden of Proof meta-analytic framework

Subject of Research: Medicine

Article Title: Maternal and neonatal outcomes associated with haemoglobin concentration in pregnancy: a Burden of Proof study

Article References: Nanditha, N. G. A., Rhee, D. K., Teply, C. J., Tandiono, H. A., Okorie, I., Taylor, H. J., Stollfus, B. L., Chu, H. T., Noyes, T. S., Humphrey, K., Markt-Maloney, J., Memetova, A., Chung, E., Harris, A. A., Fuller, K. C., Dirac, M. A., Bhutta, Z. A., Gardner, W. M., Ershova, K. I., … Kassebaum, N. J. (2026). Maternal and neonatal outcomes associated with haemoglobin concentration in pregnancy: a Burden of Proof study. Nature Health. https://doi.org/10.1038/s44360-026-00129-7

Image Credits: AI Generated

DOI: 10.1038/s44360-026-00129-7

Keywords: pregnancy, haemoglobin, anaemia, maternal mortality, neonatal outcomes, postpartum haemorrhage, preterm birth, low birth weight, Burden of Proof, dose–response, nonlinear risk, trimester-specific thresholds

Cite Scienmag News
APA MLA Chicago

Harold Sullivan. (September 5, 2026). Pregnancy haemoglobin levels linked to maternal and newborn health outcomes. Scienmag. https://scienmag.com/pregnancy-haemoglobin-levels-linked-to-maternal-and-newborn-health-outcomes/

Copy citation Download RIS

Tags: continuous risk assessment in pregnancyeffects of low and high haemoglobin during pregnancyevidence-based analysis of pregnancy blood thresholdsevidence-based approach to pregnancy blood testsimpact of anemia diagnosis thresholds on maternal and newborn healthimpact of maternal anaemia on newborn healthimplications for clinical guidelines on pregnancy anaemiaimplications of haemoglobin variability on fetal developmentimportance of nuanced risk evaluation in obstetric careimproving pregnancy health monitoring andindividualized thresholds for haemoglobin during pregnancylimitations of fixed haemoglobin thresholds in pregnancymaternal and fetal health risk factorsmaternal blood health and pregnancy complication riskmeta-analytic approach to pregnancy blood healthmeta-analytic methods in pregnancy researchnonlinear dose-response curves in pregnancy studiesnonlinear relationship between haemoglobin and pregnancy risksnonlinear risk curves in pregnancy health researchPregnancy haemoglobin levels and maternal health outcomesreevaluating anemia treatment guidelines