Preterm birth (PTB) and low birth weight (LBW) are common adverse perinatal outcomes, exerting both short-term and long-term effects for infants, such as infection, intellectual and mental disabilities and CVD(Reference Saigal and Doyle1,Reference Blencowe, Krasevec and de Onis2) . In 2020, the global prevalence of PTB and LBW was 9·9 % and 14·7 %, respectively, with no measurable change over the past decade(Reference Ohuma, Moller and Bradley3,Reference Okwaraji, Krasevec and Bradley4) . In China, although the prevalence of PTB and LBW is lower than the global average, the number of infants born with PTB and LBW remains substantial due to its large population size, contributing significantly to the global burden(Reference Tang, Mu and Li5). Therefore, identifying the potential modifiable factors of PTB and LBW is critical to improving the quality of birth population and preventing both short-term and long-term diseases in offspring.
Hb concentration, a key indicator of maternal nutritional status, holds a crucial role throughout pregnancy(Reference Means6). Among healthy pregnant women, plasma volume has little increase in the first trimester, followed by a sharp rise during the second trimester and peaks late in the third trimester(Reference Aguree and Gernand7). Hb concentrations exhibit fluctuations during pregnancy due to physiological hemodilution, which enhances placental nutrient delivery and gas exchange to support fetal development(Reference Sun, Gu and Wu8). Anaemia is a common nutritional deficiency among pregnant women, and its risk with adverse perinatal outcomes has been widely recognised(Reference Daru, Zamora and Fernández-Félix9). Compared with anaemia, elevated maternal Hb concentrations status are usually overlooked and perceived as a symbol of good nutritional status, which needs critical attention(Reference Wang, Lin and Su10). Nevertheless, this perception is likely to be misleading, as elevated Hb concentrations may reflect insufficient plasma volume expansion during pregnancy, which could impair placental perfusion and increase blood viscosity, thereby potentially contributing to adverse birth outcomes(Reference Liu, Li and Zhang11).
In recent years, although some small-scale studies have suggested an association between elevated Hb concentrations and increased risks of PTB and LBW(Reference Liu, Li and Zhang11–Reference Ohuma, Jabin and Young13), investigations into this relationship across different trimesters are relatively limited and controversial(Reference Scanlon, Yip and Schieve14,Reference Peng, Si and Cheng15) . Several studies have indicated that elevated Hb concentrations, particularly at ≥ 130 g/l, are associated with increased risks of PTB and LBW, while others have not identified a consistent or significant connection with adverse pregnancy outcomes(Reference Liu, Li and Zhang11,Reference Wu, Sun and Liu12,Reference Scanlon, Yip and Schieve14) . Furthermore, most existing studies have rarely accounted for the physiological fluctuations in Hb concentrations throughout pregnancy. Assessing Hb concentrations at a single time point may lead to heterogeneity in results and fail to capture the comprehensive profile of maternal Hb status(Reference Xu, Wang and Li16). In comparison, quantifying Hb changes across different trimesters and longitudinal trajectories over the entire pregnancy may better elucidate the association with PTB and LBW(Reference Peng, Si and Cheng15,Reference Randall, Patterson and Gallimore17) , thereby enhancing clinical approaches to risk stratification and targeted intervention during pregnancy.
Therefore, we conducted a retrospective cohort study to examine the link between maternal Hb concentrations during the first, second and third trimesters with the risks of PTB and LBW. Moreover, the associations of Hb concentration changes and trajectories during pregnancy with adverse outcomes were also evaluated.
Materials and methods
Study design and participants
This retrospective cohort study was conducted among 19 370 participants with complete medical information delivered at Haidian District Maternal and Child Health Care Hospital in Beijing, China, between January 2018 and December 2021. A total of 18 980 women with at least one Hb measurement were included in the two analyses, after excluding women with multiple births and those with pre-pregnancy diabetes and hypertension. In one analysis, we further divided the Hb concentration records contributed by 18 980 participants into the first, second and third trimesters. Among them, 8229 participants had Hb measurements across the first, second and third trimesters. In a secondary trajectory analysis, we excluded women who have less than three Hb measurements during pregnancy, with the remaining 14 681 participants provided 72 117 records for trajectory analysis (Figure 1). Ethical approval for this study was obtained from the Institutional Research Review Board of the National Research Institute for Family Planning, Beijing, China (IRB-2024001).
The flow diagram of study participants.
Definition of exposure
Hb concentration was quantified in a unit of g/l using an automatic haematology analyser for routine blood tests of venous blood samples from participants. Meanwhile, Hb concentrations were categorised into four categories using intervals of 10 g/l: < 110 g/l, 110–119 g/l, 120–129 g/l and ≥ 130 g/l. A similar classification has been documented in a prior study(Reference Xu, Wang and Li16). Due to the rapid expansion of plasma volume during the mid-to-late stages of pregnancy, we reclassified Hb concentrations into three categories (< 120 g/l, 120–129 g/l and ≥ 130 g/l) to better characterise the joint effects of elevated Hb concentrations in the second and third trimesters.
Gestational age was calculated based on the difference between the detection time of Hb measurement and the last menstrual period. The period of pregnancy was then stratified into three trimesters according to gestational age: first trimester (0–12 weeks), second trimester (13–27 weeks) and third trimester (28 weeks until delivery). We also evaluated changes in Hb concentrations during pregnancy using Hbfs (second trimester in Hb minus first trimester in Hb), Hbft (third trimester in Hb minus first trimester in Hb) and Hbst (third trimester in Hb minus second trimester in Hb) and categorised them into four categories based on quartiles. Additionally, given the Hb concentration fluctuations during pregnancy, the gestational age time point windows were further stratified according to the Chinese Guideline for Preconception and Prenatal Care (2018)(18). These time points were defined as 1–13+6, 14–19+6, 20–24+6, 25–28+6, 29–32+6, 33–36+6 and ≥ 37 weeks of gestation.
In this study, for women with more than one Hb record in a certain trimester or time point window, the record closest to the highest frequency detection weeks was selected(Reference Xu, Cheng and Dong19). Any remaining measurements in this trimester and time point window were discarded.
Definition of outcome
The primary outcomes of this study included PTB and LBW. PTB was identified by delivery before 37 completed weeks of gestation. LBW was defined as a neonatal birth weight below 2500 g.
Covariate assessment
Covariates in all multivariable-adjusted models, including physical examinations and diagnostic information, were extracted from the electronic medical record system. The adjusted covariates were as follows: age (< 35, ≥ 35 years), preconception BMI (< 18·5, 18·5–23·9, 24·0–27·9, ≥ 28·0 kg/m2), education (junior high school and below, senior high school or above), residence (rural, urban), nationality (han, other), parity (primipara, multipara), gestational diabetes mellitus (yes, no), hypertensive disorders of pregnancy (yes, no), thyroid dysfunction (yes, no), hepatitis B virus infection (yes, no), history of adverse pregnancy outcomes (yes, no) and history of cesarean section (yes, no).
Statistical analysis
Continuous variables were expressed as means with standard deviations or medians with interquartile ranges, and categorical variables were described as counts and percentages. Differences in characteristics across trajectory groups were analysed using the χ 2 test or the Kruskal–Wallis test.
To identify trajectories of Hb concentrations during pregnancy, a latent class growth mixed model was employed. The optimal latent class growth mixed model was established according to the following criteria(Reference Xu, Cheng and Dong19): (1) the Bayesian information criterion, with lower values indicating a superior model fit; (2) requiring more than 5 % of the participants in every trajectory class and (3) ensuring that each trajectory group has high average posterior probabilities (> 0·7).
We performed logistic regression models to calculate OR and 95 % CI for assessing the relationship of Hb concentrations in different trimesters and Hb changes across pregnancy with outcomes. The maternal Hb concentration in 110–119 g/l served as the reference group across pregnancy, and the reference group for Hb change was defined as the first quartile (Q1). Logistic regression models were also used to examine the association of Hb trajectories with PTB and LBW. The dose–response relationships of maternal Hb concentrations and Hb changes with odds of PTB and LBW were estimated through restricted cubic spline with three knots. Covariates in the above analysis including age, preconception BMI, parity, education, residence, nationality, gestational diabetes mellitus, hypertensive disorders of pregnancy, thyroid dysfunction, hepatitis B virus infection, history of adverse pregnancy outcomes and history of cesarean section.
All statistical analyses were carried out using R, version 4.2.1 (R Project for Statistical Computing). A two-sided P < 0·05 was set as the threshold for statistical significance.
Results
Maternal baseline characteristics
Among 18 980 women included in the cohort study, the median age of participants was 31·00 (interquartile range, 29·00–34·00) years. Higher proportions of participants were younger than 35 years (77·87 %), with normal preconception BMI (70·40 %), senior high school or above (89·10 %), urban areas (73·44 %) and primipara (61·69 %) were the majority. Details of the participants’ characteristics were shown in Table 1, and the characteristics of participants in trajectory analysis were presented in the online Supplementary Table S1.
Demographic and clinical characteristics of pregnant women enrolled in this study
IQR, interquartile range; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; HBV, hepatitis B virus.
Association of Hb concentrations with low birth weight and preterm birth in different trimesters
As shown in Table 2, maternal Hb concentrations in the first trimester were not associated with the odds of PTB or LBW. In the second trimester, women with Hb 120–129 g/l showed no statistically significant increase in the odds of PTB but had a 44 % (OR: 1·44, 95 % CI 1·16, 1·80) increased odds of LBW compared with the reference group (110–119 g/l). While those with Hb ≥ 130 g/l carried higher odds of PTB and LBW, with an adjusted OR of 1·74 (95 % CI 1·29, 2·31) and 2·31 (95 % CI 1·65, 3·20). In the third trimester, the OR of exposure to Hb 120–129 g/l for PTB was 1·26 (95 % CI 1·04, 1·51) and for LBW was 1·72 (95 % CI 1·36, 2·17) compared with Hb of 110–119 g/l. Women with Hb ≥ 130 g/l exhibited a higher odds of PTB (OR: 1·45, 95 % CI 1·12, 1·86) and LBW (OR: 2·60, 95 % CI 1·96, 3·43). No associations between anaemia and an increased odds of PTB or LBW were found in different trimesters.
Associations between maternal Hb concentrations in pregnancy with odds of PTB and LBW
PTB, preterm birth; LBW, low birth weight; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; HBV, hepatitis B virus.
Model Ⅰ was adjusted for age, preconception BMI, parity, education, residence and nationality.
Model Ⅱ was additionally adjusted for GDM, HDP, thyroid dysfunction, HBV infection, history of adverse pregnancy outcomes and history of cesarean section.
The dose–response relationships between Hb concentrations across pregnancy and odds of PTB and LBW were displayed in online Supplementary Figure S1. In the first trimester, Hb concentrations were associated with PTB and LBW odds in a J-shaped (both non-linear P > 0·05). In the second and third trimesters, the same trends were observed, and the Hb threshold was determined to be 116 g/l (both non-linear P < 0·05).
Association of Hb changes with low birth weight and preterm birth across the pregnancy period
Table 3 revealed the associations between changes in Hb concentrations during pregnancy and the odds of LBW and PTB. Hb changes from the first to the second trimester were not notably related to PTB odds. However, women in the highest quartile of Hb changes during this period exhibited a 62 % (OR: 1·62, 95 % CI 1·15, 2·31) increased odds of LBW compared with those in the lowest quartile. From the first to the third trimester, individuals in the third and fourth quartiles showed elevated odds of both PTB and LBW. A similar pattern was observed for Hb changes from the second to the third trimester. Moreover, a generally linear dose–response relationship was identified between Hb changes and the odds of PTB and LBW (Figure 2).
Associations between changes of maternal Hb concentrations in different trimester with odds of PTB and LBW
PTB, preterm birth; LBW, low birth weight; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; HBV, hepatitis B virus; Hbfs, second trimester in Hb minus first trimester in Hb; Hbft, third trimester in Hb minus first trimester in Hb; Hbst, third trimester in Hb minus second trimester in Hb.
Model Ⅰ was adjusted for age, preconception BMI, parity, education, residence and nationality.
Model Ⅱ was additionally adjusted for GDM, HDP, thyroid dysfunction, HBV infection, history of adverse pregnancy outcomes and history of cesarean section.
Dose–response associations of maternal Hb change during pregnancy with odds of adverse pregnancy outcomes. The dashed vertical line indicates the estimated threshold value from the RCS model. Models were adjusted for age, preconception BMI, parity, education, residence, nationality, GDM, HDP, thyroid dysfunction, HBV infection, history of adverse pregnancy outcomes and history of caesarean section. PTB, preterm birth; LBW, low birth weight; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; HBV, hepatitis B virus; Hbfs, second trimester in Hb minus first trimester in Hb; Hbft, third trimester in Hb minus first trimester in Hb; Hbst, third trimester in Hb minus second trimester in Hb; RCS, restricted cubic spline.
Joint effects of Hb concentrations during the second and third trimesters on low birth weight and preterm birth odds
Compared with women with Hb < 120 g/l in both the second and third trimesters, women with Hb ≥ 130 g/l during the second and third trimesters exhibited a 2·53-fold (OR: 2·53, 95 % CI 1·52, 4·05) increase in the odds of PTB. Among women with Hb 120–129 g/l in the second trimester and 130 g/l in the third trimester, the odds of PTB were elevated by 71 % (OR: 1·71, 95 % CI 1·06, 2·64). A comparable pattern of association was observed in the analysis of LBW (online Supplementary Figure S2).
Association of Hb trajectories with low birth weight and preterm birth
As presented in Figure 3, two distinct trajectories of Hb concentration during pregnancy were identified using latent class growth mixed model, 61·79 % of the women demonstrated a decline-gradual-rise trajectory, while the remaining 38·21 % exhibited a decline-sharp-rise trajectory. Compared with those in the decline-gradual-rise group, women with the decline-sharp-rise group had a 48 % (OR: 1·48, 95 % CI 1·26, 1·75) higher odds of PTB but showed no increased odds of LBW (Table 4).
Maternal Hb trajectories during pregnancy.
Associations between trajectories of maternal Hb and the odds of adverse birth outcomes
PTB, preterm birth; LBW, low birth weight; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; HBV, hepatitis B virus.
Model Ⅰ was adjusted for age, preconception BMI, parity, education, residence and nationality.
Model Ⅱ was additionally adjusted for GDM, HDP, thyroid dysfunction, HBV infection, history of adverse pregnancy outcomes and history of cesarean section.
Discussion
In this retrospective cohort study, we found that elevated Hb concentrations, particularly ≥ 120 g/l in mid-to-late gestation, were associated with a J-shaped increase in the odds of PTB and LBW. Moreover, insufficient reductions or abnormal trajectories of maternal Hb concentrations throughout pregnancy have been implicated in an increased odds of adverse pregnancy outcomes.
The impact of elevated Hb on PTB and LBW was particularly pronounced during the second and third trimesters. Several studies analysing Hb concentrations across different trimesters in relation to adverse outcomes have similarly demonstrated that elevated Hb concentrations during the second and third trimesters are associated with increased risks of PTB and LBW(Reference Liu, Li and Zhang11–Reference Ohuma, Jabin and Young13). Divergently, two small-scale studies conducted in China identified that the risk of LBW was only elevated when Hb concentrations reached ≥ 130 g/l in the third trimester(Reference Liu, Li and Zhang11,Reference Wu, Sun and Liu12) . Another multinational study reported that significant increases in PTB risk occurred at substantially higher Hb concentrations of up to 165 g/l during mid-to-late pregnancy(Reference Ohuma, Jabin and Young13). With a relatively larger sample size, our study revealed an association of Hb concentrations ≥ 120 g/l during the second and third trimesters with increased odds of PTB and LBW, indicating that the current thresholds for defining high Hb concentrations during pregnancy may require reassessment. Nevertheless, limited studies have also found that Hb concentrations ≥ 130 g/l during the first trimester increase the risk of PTB and LBW(Reference Wang, Lin and Su10,Reference Tang, Luo and Sheng20) . Given that plasma volume expansion was minimal in early pregnancy(Reference de Haas, Ghossein-Doha and van Kuijk21), whether a high Hb concentration in the first trimester contributing to adverse outcomes required further investigation through high-quality studies.
Notably, beyond trimester-specific Hb effects, our study also found that women with insufficient Hb concentrations decline during pregnancy faced significantly higher odds of PTB and LBW. With advancing research into haematological changes during pregnancy, several studies have examined the implications between Hb changes and adverse outcomes. A study involving 1986 Japanese participants reported a mean Hb concentration of 12·5 g/dl in early pregnancy and found that each 1 g/dl increase in Hb from early to late pregnancy was linked to a 76·1 g reduction in birth weight(Reference Jwa, Fujiwara and Yamanobe22). Similar to our results, Peng et al. have shown a negative correlation of both Hb decline from the first to second trimester and from the first to third trimester with the risk of LBW(Reference Peng, Si and Cheng15). Since the inverse relationship between Hb concentrations and plasma volume expansion during pregnancy, insufficient reductions in Hb may indicate either suboptimal plasma volume adaptation or an underlying pathological condition, both of which are recognised contributors to impaired fetal growth(Reference Aguree and Gernand7).
Considering the substantial plasma volume expansion during mid- to late pregnancy(Reference de Haas, Ghossein-Doha and van Kuijk21), considerable physiological fluctuations in maternal Hb concentrations were expected. Although the Hb changes between the second and third trimesters reflected hemodilution dynamics, it failed to represent the actual Hb status at each corresponding time point. Therefore, we specifically analysed the joint effect of Hb concentrations in the second and third trimesters on adverse outcomes. Our findings demonstrated that women with Hb ≥ 130 g/l during both the second and third trimesters had the highest odds of PTB and LBW. To the best of our knowledge, such findings have not been previously documented. The results in our study highlight the importance of identifying and managing high-risk populations to mitigate adverse pregnancy outcomes and improve the quality of the birth population.
With the variability of Hb concentrations throughout pregnancy, a more specific and sensitive approach can enhance the identification of women at elevated risk for PTB and LBW. Based on findings from an Australian study, Randall et al. indicated that maintaining a high Hb trajectory was associated with a higher risk of PTB(Reference Randall, Patterson and Gallimore17). While our study did not observe a significant association between abnormal Hb trajectory and LBW, this discrepancy may imply different critical windows for the development of PTB and LBW. Our findings, combined with limited prior studies, suggest the necessity for expanded research to uncover additional abnormal Hb trajectories and support targeted clinical risk stratification.
The mechanisms linking elevated maternal Hb concentrations to PTB and LBW are distinct. The onset of PTB may be partially attributed to the role of Hb in nitric oxide regulation and endothelial function. Nitric oxide is an effective vasodilator that relaxes vascular smooth muscle cells(Reference Jin and Loscalzo23). Free Hb can bind to nitric oxide and inactivate it, which may exacerbate vasospasm, trigger uterine contractions and increase the risk of PTB(Reference Gladwin, Crawford and Patel24–Reference Zullino, Buzzella and Simoncini26). In addition, inadequate plasma volume expansion during pregnancy may directly reduce blood flow in low-resistance microvascular beds such as the placenta, resulting in impaired placental perfusion and oxygenation(Reference Soma-Pillay, Catherine and Tolppanen27,Reference Morley, Debant and Walker28) . Placental hypoxia and ischemia can stimulate the release of corticotropin-releasing hormone, and elevated corticotropin-releasing hormone levels have been linked to PTB(Reference Thomson29,Reference Wadhwa, Garite and Porto30) . The occurrence of LBW may be related to the following mechanisms. Elevated Hb concentrations indicate a failure of plasma volume expansion during pregnancy, which has been associated with fetal growth-restricting conditions such as pregnancy-induced hypertension and pre-eclampsia(Reference Scanlon, Yip and Schieve14,Reference de Haas, Ghossein-Doha and van Kuijk21) . Additionally, insufficient plasma volume expansion during pregnancy also leads to increased blood viscosity, resulting in hindered nutrient transport across the placenta and eventual impairment of fetal growth and development(Reference Liu, Li and Zhang11). Moreover, elevated Hb concentrations may indicate higher Fe stores, which could contribute to oxidative stress, lipid peroxidation and DNA damage in placental cells, thereby compromising fetal growth(Reference Young, Oaks and Rogers31,Reference Oaks, Jorgensen and Baldiviez32) . In addition to the relevant mechanisms derived from placental adaptive response, further research is needed to reveal more biological mechanisms in the future.
The present study has several strengths. First, this is the first multi-dimensional study based on the Hb concentrations and changes at different trimesters and the trajectory of Hb concentrations during pregnancy to comprehensively elucidate their associations with PTB and LBW. Second, compared with other studies, this research provided compelling evidence on the association of high Hb concentrations during pregnancy and adverse outcomes. Additionally, the study identified critical thresholds of high Hb linked to increased odds of adverse outcomes, particularly in the mid-to-late gestation.
However, some study limitations should be acknowledged. First, this study was conducted in Beijing, where access to healthcare and nutritional conditions are relatively favourable. In addition, excluding women with fewer than three Hb measurements may have preferentially retained those receiving more consistent antenatal care, potentially introducing selection bias. These factors may influence the distribution and clinical interpretation of Hb concentrations, thereby limiting the generalisability of our findings to populations in less-resourced or differently structured healthcare settings. Thus, the observed associations in our study should be interpreted with caution, and further studies in more diverse populations are warranted. Second, the lack of detailed information on maternal Fe status, dietary intake and supplementation limited our ability to explore potential underlying mechanisms and may have introduced residual confounding. More well-designed studies are recommended to investigate deeply. Third, although our study adjusted potential confounding variables as much as possible, some unmeasured confounders could not be thoroughly addressed in our analysis.
In conclusion, this study demonstrates that Hb ≥ 120 g/l in mid-to-late pregnancy is linked to increased odds of PTB and LBW. Furthermore, insufficient decline and abnormal trajectory in Hb concentrations significantly elevate the odds of adverse outcomes, emphasising the importance of clinical surveillance of Hb concentrations and timely targeted interventions.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114526107685
Acknowledgements
The authors would like to thank all dedicated health workers in the Haidian District Maternal and Child Health Care Hospital and participants who took part in the construction of this medical database.
This work was supported by the National Key Research and Development Program of China (grant no. 2021YFC2700705).
Conceptualisation: J. L. (Jiaxin Li) and Y. Y.; methodology: J. L. (Jiaxin Li) and Y. Z.; validation: J. L. (Jiaxin Li) and Y. Y.; formal analysis: J. L. (Jiaxin Li) and Y. Z.; investigation: J. L. (Jiaxin Li), Y. Z., C. Z., T. T., Y. J., X. L., H. W., J. L. (Jueming Lei), Z. C., S. W., H. Y. and X. Z.; writing – original draft: J. L. (Jiaxin Li), Y. Z. and C. Z.; writing – review and editing: Y. Y.; supervision: X. Z. and Y. Y. All authors have read and agreed to the published version of the manuscript.
The authors report no conflict of interest.
