Maternal anemia is a well‐recognized global health problem; however, there remain questions on specific hemoglobin (Hb) thresholds that predict health risk or protection for mother and child. We conducted a systematic review and meta‐analysis to examine the associations of maternal Hb concentrations with a range of maternal and infant health outcomes, accounting for the timing of measurement (preconception, and first, second, and third trimesters), etiology of anemia, and cutoff category. The systematic review included 272 studies and the meta‐analysis included 95 studies. Low maternal Hb (130 g/L) was associated with increased odds of SGA, stillbirth, preeclampsia, and gestational diabetes. Relationships varied by the timing of measurement and cutoff category (stronger associations with lower cutoffs); limited data were available on anemia etiology. There were insufficient data for other maternal outcomes and long‐term child health outcomes. Current data are insufficient for determining if revisions to current Hb cutoffs are required. Pooled high‐quality individual‐level data analyses, as well as prospective cohort studies, would be valuable to inform the reevaluation of Hb cutoffs.

Background: Growing evidence suggests low and high maternal hemoglobin (Hb) concentrations may have adverse consequences for maternal and child health. There remain questions on specific Hb thresholds to define anemia and high Hb as well as how cutoffs may vary by anemia etiology and timing of assessment.Methods: We conducted an updated systematic review (using PubMed and Cochrane Review) on low (< 110 g/L) and high (≥ 130 g/L) maternal Hb concentrations and associations with a range of maternal and infant health outcomes. We examined associations by timing of Hb assessment (preconception; first, second, and third trimesters, as well as at any time point in pregnancy), varying cutoffs used for defining low and high hemoglobin concentrations and performed stratified analyses by iron-deficiency anemia. We conducted meta-analyses to obtain odds ratios (OR) and 95% confidence intervals.Results: The updated systematic review included 148 studies. Low maternal Hb at any time point in pregnancy was associated with: low birthweight, LBW (OR (95% CI) 1.28 (1.22–1.35)), very low birthweight, VLBW (2.15 (1.47–3.13)), preterm birth, PTB (1.35 (1.29–1.42)), small-for-gestational age, SGA (1.11 (1.02–1.19)), stillbirth 1.43 (1.24–1.65)), perinatal mortality (1.75 (1.28–2.39)), neonatal mortality (1.25 (1.16–1.34), postpartum hemorrhage (1.69 (1.45–1.97)), transfusion (3.68 (2.58–5.26)), pre-eclampsia (1.57 (1.23–2.01)), and prenatal depression (1.44 (1.24–1.68)). For maternal mortality, the OR was higher for Hb < 90 (4.83 (2.17–10.74)) than for Hb < 100 (2.87 (1.08–7.67)). High maternal Hb was associated with: VLBW (1.35 (1.16–1.57)), PTB (1.12 (1.00-1.25)), SGA (1.17 (1.09–1.25)), stillbirth (1.32 (1.09–1.60)), maternal mortality (2.01 (1.12–3.61)), gestational diabetes (1.71 (1.19–2.46)), and pre-eclampsia (1.34 (1.16–1.56)). Stronger associations were noted earlier in pregnancy for low Hb and adverse birth outcomes while the role of timing of high Hb was inconsistent. Lower Hb cutoffs were associated with greater odds of poor outcomes; for high Hb, data were too limited to identify patterns. Information on anemia etiology was limited; relationships did not vary by iron-deficiency anemia.Conclusion: Both low and high maternal Hb concentrations during pregnancy are strong predictors of adverse maternal and infant health outcomes. Additional research is needed to establish healthy reference ranges and design effective interventions to optimize maternal Hb during pregnancy.

Growing evidence supports the role of preconception maternal nutritional status (PMNS) on birth outcomes; however, evidence of relationships with child growth are limited. We examined associations between PMNS (height, weight and body mass index- BMI) and offspring growth during the first 1000 days. We used prospective cohort data from a randomized-controlled trial of preconception micronutrient supplementation in Vietnam, PRECONCEPT (n = 1409). Poisson regression models were used to examine associations between PMNS and risk of offspring stunting (<-2 HAZ) at 2 years. We used path analytic models to examine associations with PMNS on fetal growth (ultrasound measurements) and offspring HAZ at birth and 2 years. All models were adjusted for child age, sex, gestational weight gain, education, socioeconomic status and treatment group. A third of women had a preconception height < 150cm or weight < 43 kg. Women with preconception height < 150 cm or a weight < 43 kg were at increased risk of having a stunted child at 2 years (incident risk ratio IRR: 1.85, 95% CI 1.51–2.28; IRR 1.35, 95% CI 1.10–1.65, respectively). While the traditional low BMI cut-off (< 18.5 kg/m²) was not significant, lower BMI cut-offs (< 17.5 kg/m² or < 18.0 kg/m²) were significantly associated with 1.3 times increased risk of child stunting. In path models, PMNS were positively associated with fetal growth (ultrasound measurements) and offspring HAZ at birth and 2 years. For each 1 standard deviation (SD) increase in maternal height and weight, offspring HAZ at 2 years increased by 0.30 SD and 0.23 SD, respectively. In conclusion, PMNS influences both offspring linear growth and risk of stunting across the first 1000 days. These findings underscore the importance of expanding the scope of current policies and strategies to include the preconception period in order to reduce child stunting.

Anemia remains an important global health problem. Inexpensive, accurate, and noninvasive solutions are needed to monitor and evaluate anemia in resource-limited settings. We evaluated the performance of multiple point-of-care hemoglobin devices, including a novel noninvasive smartphone application tested on Apple® and Android® cell phones, Masimo Pronto®, and HemoCue® Hb-301 and Hb-801, against a gold-standard hematology analyzer (reference hemoglobin) using venous blood. We examined correlations between hemoglobin devices and reference hemoglobin, device accuracy (average bias, Bland-Altman plots, clinical performance) and classification bias (sensitivity, specificity) among 299 refugees (10mo-65y) in Atlanta, GA. Semi-structured interviews (n = 19) with participants and staff assessed usability and acceptability. Mean reference hemoglobin was 13.7 g/dL (SD:1.8) with 12.5% anemia. Noninvasive hemoglobin devices were not well correlated with reference hemoglobin (Apple® R²= 0.08, Android® R²= 0.11, Masimo Pronto® R²= 0.29), but stronger correlations were reported with HemoCue® Hb-301 (R²= 0.87) and Hb-801 (R²= 0.88). Bias (SD) varied across each device: Apple®: -1.6 g/dL (2.0), Android®: -0.7 g/dL (2.0), Masimo Pronto®: -0.4 g/dL (1.6), HemoCue® Hb-301: +0.4 g/dL (0.7) and HemoCue® Hb-801: +0.2 g/dL (0.6). Clinically acceptable performance (within ± 1 g/dL of reference hemoglobin) was higher for the invasive devices (HemoCue® Hb-301: 90.3%; HemoCue® Hb-801: 93.4%) compared to noninvasive devices (Apple®: 31.5%; Android®: 34.6%; Masimo Pronto®: 49.5%). Sensitivity and specificity were 63.9% and 48.2% for Apple®, 36.1% and 67.6% for Android®, 45.7% and 85.3% for Masimo Pronto®, 54.3% and 97.6% for HemoCue® Hb-301, and 66.7% and 97.6% for HemoCue® Hb-801. Noninvasive devices were considered easy to use and were the preferred method by participants. Among the only studies to compare multiple point-of-care approaches to hemoglobin testing, the diagnostic ability of HemoCue® was comparable to reference hemoglobin, while noninvasive devices had high user acceptability but considerable biases. Improvements in noninvasive device performance and further testing in anemic populations are recommended before broader use.