Both low and high haemoglobin in pregnancy tracked with preterm birth, by gestational week

Haemoglobin is one of the most routinely measured numbers in pregnancy. A finger-prick or blood draw at a prenatal visit gives it, and low values are treated as anaemia to be corrected. But the thresholds that count as abnormal are inconsistent, and they rarely account for two things that plainly matter: how far along the pregnancy is, and how high above sea level the mother lives. A large cohort study published on 4 November in JAMA Network Open tried to pin those thresholds down against a single outcome — preterm birth [s1].

What was studied

The researchers built a multisource database from five medical centres in Yunnan, a Chinese province with a steep altitudinal gradient, linking electronic medical records, preconception examinations and geospatial data [s1]. They included singleton pregnancies delivered between 1 January 2017 and 31 July 2025, and operationally defined preterm birth as a live birth from 28 weeks 0 days through 36 weeks 6 days of gestation [s1].

Rather than treat haemoglobin as a single number, they grouped prenatal measurements into two-week windows across pregnancy and asked, at each window, whether low or high values were associated with later preterm birth [s1]. That design is the point of the paper: it allows the "abnormal" line to move as pregnancy progresses, instead of applying one cut-off throughout.

What they found

Of 141,943 singleton live births, among women with a mean age of 29.7 years, 12,607 — 8.88% — were preterm [s1]. The relationship between haemoglobin and preterm birth was U-shaped and depended on gestational age: both ends of the distribution carried risk, and the thresholds shifted across pregnancy [s1].

At the low end, haemoglobin below 12.00 g/dL measured around 13 weeks was associated with preterm birth, with an adjusted odds ratio of 1.46 (95% CI, 1.17-1.80) compared with the intermediate group [s1]. The association persisted later in pregnancy at a lower cut-off: from 27 weeks onward, haemoglobin below 11.00 g/dL carried an adjusted odds ratio of 1.48 (95% CI, 1.25-1.74) [s1]. At the high end, elevated haemoglobin above 13.50 g/dL from 19 weeks onward was associated with preterm birth, with an adjusted odds ratio of 1.28 (95% CI, 1.04-1.55) [s1].

Altitude changed the map. At high altitude, the thresholds were generally higher, and a low haemoglobin concentration from 21 weeks was persistently associated with preterm birth [s1]. Subtype analyses suggested low-haemoglobin associations appeared earlier for medically indicated preterm birth, while high-haemoglobin associations in the middle-to-late pregnancy window showed up for both spontaneous and medically indicated subtypes [s1].

How to read it

The intuitive half of this is familiar: anaemia in pregnancy has long been linked to adverse outcomes, and low haemoglobin tracking with preterm birth fits that picture [s1]. The less familiar half is that high haemoglobin also carried risk. That is biologically plausible — elevated haemoglobin can reflect poor plasma volume expansion or impaired placental perfusion rather than robust health — but it cuts against the instinct that more is better, and it is the finding most likely to surprise clinicians.

The effect sizes are modest. Adjusted odds ratios between about 1.3 and 1.5 describe a real but moderate shift in risk, not a doorway between safe and doomed pregnancies [s1]. These are associations drawn from records, so they cannot establish that correcting haemoglobin to any particular target would change the outcome. Confounding is the standing hazard of this kind of study: the conditions that drive haemoglobin up or down — iron status, inflammation, blood-volume changes, underlying disease — are themselves tied to preterm birth, and adjustment can only go so far.

The altitude result is where the practical value lies. If the haemoglobin number that signals trouble genuinely differs by elevation and by gestational week, then a single national cut-off will misclassify women in both directions — flagging some as anaemic who are not, and reassuring others who are at risk [s1]. The authors frame their thresholds as candidates to refine how a routine measurement is interpreted, not as rules to act on, and they are explicit that the findings await external validation [s1].

This is also a study rooted in one province and one population, which limits how far the specific numbers travel [s1]. The transportable idea is methodological: that the meaning of a haemoglobin value depends on when in pregnancy and where it was measured. Whether gestational-age- and altitude-adapted thresholds actually improve prediction — or outcomes — is the question a future, prospective study would need to answer before anything changes at the bedside.