The impact of response to iron therapy on maternal and neonatal outcomes among pregnant women with anemia.

This study was designed to evaluate whether successful response to iron therapy (“successful treatment”) affected maternal and neonatal outcomes in women with anemia with adequate prenatal care in a large population-based cohort. Successfully treated patients had a considerable reduction in adjusted odds of PE and PTB. Similar to untreated women with anemia, refractory women had higher odds of most maternal outcomes, including PTB, CMM, PE, blood transfusion, endometritis, and cesarean delivery. All groups with anemia had a higher odds of PPH. Interestingly, there was a considerable reduction in the odds of all women with anemia delivering an SGA neonate. This study confirmed that pregnant persons with iron deficiency anemia have higher odds of perinatal morbidity after adjusting for potential confounders. Moreover, it uniquely demonstrated that risk is dependent on (1) receipt of iron supplementation and (2) “successful treatment” under a considerable and clinically relevant increase in hemoglobin. Specifically, we showed that successful treatment of anemia with iron therapy can reduce the odds of maternal and neonatal complications. This study describes a comprehensive evaluation of outcomes associated with anemia and response to treatment and supplementation. A search for the terms “iron therapy,” “iron supplementation,” “iron deficiency anemia,” “maternal morbidity,” “neonatal morbidity,” and “maternal outcomes” in PubMed ranging from 1980 revealed no similar study inclusive of the approach and data with findings herein. Approximately one-third of our population was composed of women who were pregnant and with anemia; although this was higher than the typical prevalence in North America, it was not surprising given that most of our patients had other risk factors for anemia, such as concurrent medical comorbidities. 17 , 18 Anemia has been associated with PTB in several previous and smaller case-control studies. 2 , 6 – 8 The underlying true causes of this association are poorly understood, but we speculate that it could be attributable to associated placental hypoxia and/or increased oxidative stress that is hypothesized to lead to hypertensive disorders and PTL and PPROM. 19 We noted that our patients with anemia had higher odds of PTB as a group, but this effect differed in each subgroup. Our patients with anemia who were successfully treated had a considerably reduced odds of PTB, which seemed to be related to a reduction in PTL, PPROM, and PE. Our refractory and untreated women had an increased odds of PTB, which may potentially be related to persistent increased oxidative stress and placental hypoxia. However, this study did not test that hypothesis. A single retrospective study in Hungary reported that iron therapy mitigated the increased risk of PTB in their population with anemia. 19 Conversely, a large meta-analysis of 48 randomized controlled trials found that iron therapy was not considerably associated with a reduction in PTB among women who were pregnant with anemia. 20 However, unlike our study, neither of these other studies stratified by treatment response nor success of treatment (which may be a reflection of the duration of therapy), which may explain our differing interpretations. It is possible that iron therapy would reduce placental and neonatal stress by supporting hematopoiesis and oxygen-carrying capacity. However, this benefit was not realized among women who were pregnant and had anemia that was refractory to therapy, and their PTB rate was increased. Therefore, iron alone did not seem to provide this benefit unless occurring in the context of correcting anemia. Although one might have expected to see an improvement in a CNM with a reduction in the risk of PTB, this was not observed here as there was no difference in CNM for our successfully treated group. When assessing all patients with anemia, there was a considerable increase in the adjusted odds of PE, which has been demonstrated in previous studies. 2 , 21 Moreover, 1 Cochrane review noted no significant difference in the rate of PE among women receiving daily iron therapy vs controls. 22 However, this was applied to all pregnant women regardless of anemia status, and many of the studies they included were from low-income countries, which may not apply to our population. These studies did not evaluate the effect of iron therapy on outcomes. We noted that there was a 25% reduction in the adjusted odds of PE with successful iron therapy compared with the reference group. This benefit was not seen among refractory patients. Our findings suggested that the association between anemia and PE may be reversed among women with adequate treatment of iron deficiency anemia. It is possible that this association is related to the resolution of anemia with subsequent improved oxygen-carrying capacity of maternal blood and, therefore, reduced oxidative stress at the placental-maternal interface. 23 , 24 In addition, there were several interesting findings from our secondary outcomes. The rate of PPH was considerably increased in our cohort with anemia. Interestingly, even successful treatment was associated with a 40% increase in the odds of intrapartum and PPH, even after adjusting for confounders. The cause for this association was likely multifactorial and more likely to represent those at risk of PPH being repeatedly advised and counseled regarding benefits and efficacy of iron therapy (given their at-risk status). However, we could not decipher from our database whether this had occurred and remains speculative but informed by our general approach to maternal care. Moreover, patients who were either untreated or refractory had an increased odds of cesarean delivery. Although there was an increased odds of cesarean delivery among our successfully treated patients, this was not substantial after controlling for confounders. We speculate that the association between anemia and cesarean delivery could also arise from other factors or comorbidities, including decreased placental reserve from recent or current anemia and increased maternal fatigue during labor. 19 Interestingly, the rate of blood transfusion and hysterectomy were, like cesarean delivery, only considerably increased among refractory and untreated patients with anemia after controlling for confounders. This may have been, in part, related to lower average starting hemoglobin in addition to the increased risk of PPH and subsequent need for surgical treatment of hemorrhage. Women with untreated anemia were not only 3 times more likely to undergo hysterectomy but also 3 times more likely to receive ICU care after adjusting for confounders. This suggested that anemia in pregnancy is associated with major maternal morbidity and increased healthcare costs, although the causal relationship cannot be determined in the current study. Multiple studies have noted an association between both occurrence and severity of maternal anemia in association with risk of delivery of an SGA neonate. 21 , 25 , 26 This may be because of the effect of considerably lower hemoglobin on placental perfusion with oxygenated blood, leading to decreased fetal growth. 25 A recent study found that although the odds of SGA considerably increased among gravidae with anemia compared with the reference population, the odds considerably decreased for women who were pregnant and only manifested mild anemia (hemoglobin level between 9 and 10.9 g/dL). 2 The adjusted odds of delivering an SGA neonate were considerably reduced among our untreated and refractory populations, both of which had a median hemoglobin in this range. Interestingly, the odds of SGA also decreased among our gravidae with successfully treated anemia, which may indicate these patients, like our other groups with anemia, had a mild anemia in the third trimester of pregnancy. Our results and previous studies suggested that fetal growth and placental nutrient exchange may be contributory factors in mild maternal iron depletion and anemia. 27 We are unsure why our refractory population did not respond to iron therapy. This could be related to unmet challenges to compliance, an undiagnosed secondary micronutrient deficiency, or coexisting morbidities that are associated with reduced absorption (ie, celiac disease, Helicobacter pylori infection, gastritis, or inflammatory bowel disease). 28 , 29 The findings from this study emphasized the need to adequately treat and monitor response to iron therapy among women with anemia in pregnancy. This included appropriate counseling on when to take the supplement to optimize absorption and reduce side effect profile. Given the findings associated with refractory and untreated anemia compared with successfully treated women with anemia, anemia unresponsive to oral iron supplementation warrants consideration for evaluation of coexistent comorbidities and potential responsiveness to IV iron therapy. Moreover, we cannot explain why women with sufficient prenatal care as included in our study did not receive treatment of iron deficiency anemia in the untreated group. This may be related to unmet challenges with compliance with recommended care, but that is speculative. Our lack of detailed knowledge emphasized the need to directly query women in pregnancy for important social determinants of health that not only impact their care but are also associated with perinatal morbidity. 30 Although this study demonstrated an association between several maternal and neonatal outcomes with responsiveness to iron therapy for the treatment of anemia in pregnancies, cause and effect could not be determined, and several questions remain. Further research is necessary to identify the mechanisms by which iron deficiency anemia contributes to these pathologies. In addition, future studies need to evaluate the impact of iron deficiency without anemia on similar outcomes, which was not assessed in this study. Moreover, this study reiterated some urgency in determining the optimal iron regimen for iron deficiency anemia in pregnancy and evaluating why some compliant women fail to respond to adequate therapy. Lastly, our results provided further rationalization for future clinical research into how protocols for the management of iron deficiency anemia in pregnancy could be of benefit. Our study benefited from a large sample size and a population-based approach, which created an ethnically and socioeconomically diverse cohort. The database had extensive demographic and comorbidity information, which could be considered and accounted for in our analysis. However, 1 limitation was that our uniform working definition of anemia was based on nadir hemoglobin on admission to labor and delivery and/or the use of ferrous sulfate. It was assumed that women treated with iron had iron deficiency as iron studies were not abstracted on every patient within this database. In our experience, nearly all women with anemia in our cohort have abnormal iron studies suggestive of iron deficiency. However, this may have incorrectly classified some women within the cohort. In addition, there was no compliance or dosing data, and some providers likely failed to address challenges, which limited our patient’s ability to comply with recommended therapy. However, these limitations were reflective of the “real-world” clinical setting, and cohort stratification and cohort size minimized these limitations.

At the time of this analysis, a total of 43,580 pregnancies had been enrolled in the perinatal database since its inception ( Figure 1 ). After excluding individuals with missing data from key variables, a final study cohort of 20,690 patients was retained for further analysis. Information regarding prenatal care is listed in Table 1 . The median hemoglobin was 10.2 (9.5–10.6), 12.0 (11.5–13.2), 10.3 (9.7–10.7), and 12.3 (11.7–13.0) for refractory, successful, untreated, and reference groups, respectively. Maternal demographics and comorbidities are represented in Tables 2 and 3 , respectively. Successfully treated patients were more likely to be White non-Hispanic, older, nulliparous, and married with a high school education and have private insurance ( Table 2 ). Untreated and refractory patients had similar demographics, and specifically, they were less likely to be nulliparous and married or have a high school education and more likely to be a minority, make <$35,000 per year, and have government insurance. Total number and percent and cORs and aORs for each outcome are listed in Tables 4 to 7 . The odds of PTB were significantly increased among refractory (aOR, 1.44; 95% CI, 1.16–1.76) and untreated (aOR, 1.45; 95% CI, 1.26–1.67) patients but was significantly decreased for successfully treated patients even after controlling for confounders (aOR, 0.59; 95% CI, 0.47–0.72) ( Figure 2 ; Tables 4 and 5 ). Moreover, 85.1% of preterm deliveries among all groups occurred at >32 weeks of gestation. There was a significantly decreased odds of very PTB (<32 weeks of gestation) among successfully treated patients (aOR, 0.32; 95% CI, 0.14–0.62). There was no difference in very PTB for refractory (aOR, 1.11; 95% CI, 0.62–1.88) or untreated (aOR, 1.33; 95% CI, 0.92–1.91) patients compared with the population without anemia. The causes of PTB varied by subcategory ( Figure 3 ). There was a significant reduction in the odds of occurrence PTB because of preterm labor (PTL) (aOR, 0.49; 95% CI, 0.33–0.70), preterm premature rupture of membranes (PPROM) (aOR, 0.60; 95% CI, 0.38–0.88), and PE (aOR, 0.47; 95% CI, 0.28–0.74) among successfully treated patients. The odds of PTB because of PE significantly increased among refractory (aOR, 1.75; 95% CI, 1.17–2.53) and untreated (aOR, 1.60; 95% CI, 1.23–2.08) patients. These findings were similar when evaluating the odds of developing PE at all gestational ages. The odds of PE significantly decreased among successfully treated patients (aOR, 0.75; 95% CI, 0.61–0.91). There was a significant increase in the odds of PE among refractory (aOR, 1.54; 95% CI, 1.24–1.89) and untreated (aOR, 1.44; 95% CI, 1.25–1.67) patients. CMM was significantly increased among refractory (aOR, 1.70; 95% CI, 1.44–2.01) and untreated (aOR, 1.71; 95% CI, 1.52–1.92) patients ( Tables 4 and 5 ; Figure 2 ). The odds of CMM was not significantly different between successfully treated patients and patients with anemia after controlling for confounders (aOR, 0.89; 95% CI, 0.77–1.03). Refractory and untreated patients had considerably increased odds of hysterectomy and blood transfusion. Moreover, untreated patients had considerably increased odds of chorioamnionitis, endometritis, and need for ICU care. There was no maternal death in the cohort. All anemic categories had an increased odds of PPH after controlling for confounders. There was a significant increase in the odds of cesarean delivery for refractory (aOR, 1.53; 95% CI, 1.33–1.75) and untreated (aOR, 1.38; 95% CI, 1.25–1.51) patients, but there was no difference for successfully treated patients (aOR, 1.09; 95% CI, 0.98–1.21) compared with patients without anemia. The evaluation of neonatal outcomes revealed that refractory and untreated patients with anemia were considerably more likely to have a baby with TTN after controlling for preterm status, but this was not substantial for successfully treated patients ( Tables 6 and 7 ; Figure 2 ). All anemic categories had a considerable reduction in the odds of delivering an SGA neonate even after controlling for confounders. There was a considerable increase in the odds of delivering an LGA neonate for successfully treated and untreated patients. There was no difference in the odds of CNM.

We used an institutional review board (IRB)-approved, subject-consented perinatal database, which is collected and maintained by trained full-time research personnel at Baylor College of Medicine in Houston, Texas (PeriBank; H-26364). All gravid patients who delivered at our 2 institutional hospitals (Ben Taub Hospital and Texas Children’s Pavilion for Women) were approached and consented to participate in our perinatal database. In addition to detailed data abstraction from the electronic medical record into the database, additional information was obtained by direct maternal interview in their native language. In the current study, more than 100 variables relevant to the hypothesis and study aims were used. 10 – 12 The PeriBank database and biorepository involved active consent for participation from all participants and did not rely on volunteers to self-identify or enter data. 10 The rate of enrollment did not change considerably throughout the study period; consent rates ranged from 86% in 2012 to 90% in 2019. Regular quarterly audits were done to ensure data accuracy, and any adjudication of cases was performed by maternal-fetal medicine physicians. The database specifically captured medication prescriptions by maternal interview and review of prescriptions or medication dispensed within the electronic medical record. 10 – 12 Women were specifically asked about prenatal vitamin and iron therapy as examples of medications used in pregnancy during interviews with study personnel during enrollment to labor and delivery. The use of the PeriBank database for this study was approved by the Baylor College of Medicine IRB (H-45877). All women with singleton pregnancies and sufficient prenatal care from August 2011 to November 2019 were included. For women with ≥2 pregnancies within this time frame, the first delivery in the PeriBank database was chosen for inclusion for the overall analysis. Sufficient prenatal care was defined by the Kotelchuck Adequacy of Prenatal Care Index. 11 , 13 , 14 This model has been previously validated and took into account the number of prenatal visits and gestational age at initiation of prenatal care and delivery. 11 , 15 Sufficient prenatal care was defined as beginning prenatal care before 20 weeks of gestation and attending 50% to 100% of recommended prenatal visits based on gestational age at delivery. Patients enrolled in PeriBank receive prenatal care at several clinics before delivering at our 2 enrollment and collection sites, where they were recruited at the time of delivery. Their prenatal care followed the ACOG recommendations and institutional guidelines for prenatal care. At our institution, patients were initially treated with supplemental iron therapy if their hemoglobin is below the ACOG cutoffs for anemia. If iron studies were performed at a later time and indicated iron deficiency, they were continued on iron supplementation for the remainder of the pregnancy. Moreover, most clinics performed hemoglobin electrophoresis to rule out hemoglobinopathies in all patients. Patients with another known cause of anemia, such as vitamin B12 deficiency, thalassemia, or sickle cell disease, and those who were actively bleeding on admission to labor and delivery were excluded from the study. Furthermore, patients with an antenatal diagnosis of known or suspected placenta accreta spectrum disorder (including previa, accreta, increta, and percreta) were excluded as they often received iron prophylactically at our institution. Maternal demographics, diagnoses, medications, and outcomes were uniformly abstracted for patients. Those patients who received iron therapy other than that included in a prenatal vitamin were considered to have a diagnosis of iron deficiency. Supplements included the brand names and generic forms of ferrous sulfate, ferrous gluconate, ferrous fumarate, and ferrous glycinate. The administration of intravenous (IV) iron was not reported consistently and so was not included in this analysis. The sample size for this analysis was exhaustive for the number of participants meeting the criteria in the database. Patients were categorized into 1 of 2 groups: anemic and nonanemic. Patients without anemia (eg, women who were pregnant without anemia) were used as the reference group. Patients with anemia were considered as those who were treated with an iron supplement outside of prenatal vitamin or presented to labor and delivery with anemia as defined by the ACOG criteria. 5 This included a hemoglobin level of <11 g/dL in the third trimester of pregnancy or 10.5 g/dL if delivered in the second trimester of pregnancy. Patients with anemia were further sub-categorized by whether (1) they received iron therapy and (2) whether they had normal hemoglobin at the time of admission for delivery. Categories were labeled as “untreated and anemic” for women who were anemic on admission to labor and delivery and did not receive iron supplementation, “successfully treated” for women who arrived with normal hemoglobin and reported taking iron supplementation, and “refractory anemic” for those who were anemic on admission to labor and delivery despite taking an iron supplement. Recorded hemoglobin for this analysis was determined at the time of admission to labor and delivery unit on all patients in both institutions. The primary maternal outcomes designated a priori as of interest were PTB before 37 weeks of gestation and PE. Other maternal outcomes included cesarean delivery, placenta abruption, intrapartum hemorrhage, postpartum hemorrhage (PPH), and composite maternal morbidity (CMM). CMM included any of the following: hypertensive disorders of pregnancy, chorioamnionitis, endometritis, placental abruption, blood transfusion, maternal ICU admission, hysterectomy, pulmonary edema, or maternal death. The primary neonatal outcome was small for gestational age (SGA). Other neonatal outcomes evaluated included transient tachypnea of the newborn (TTN), retinopathy of prematurity, and composite neonatal morbidity (CNM). CNM included any of the following: 5-minute Apgar score of ≤3, respiratory distress syndrome, suspected or proven newborn sepsis, seizure, stillbirth, or neonatal death. SGA neonates were defined as below the 10th percentile using national birthweight reference data, which is stratified by race and ethnicity and gestational age at delivery. 16 Large for gestational age (LGA) neonates were defined as above the 95th percentile using the same reference. Baseline maternal characteristics, comorbidities, demographics, and adverse perinatal outcomes were analyzed for all eligible participants. These included maternal age, parity, marital status, body mass index at time of delivery, educational achievement, race and ethnicity, income, and insurance type. Furthermore, maternal comorbidities were evaluated, which included chronic hypertension (CHTN), gestational diabetes mellitus (GDM), type 1 diabetes mellitus, type 2 diabetes mellitus, deep venous thrombosis and pulmonary embolus, hypothyroidism, hyperthyroidism, cardiac disease, seizure disorder, asthma, endometriosis, cancer, psychiatric disease, and substance use. Descriptive statistics were used to report all variables of interest. Continuous variables were presented as median (interquartile range) using the Kruskal-Wallis test. Categorical variables were presented as number (percentage) and were evaluated using the chi-square test of association. All groups were initially compared as a whole to evaluate for significance. Odds ratios (ORs) were calculated by comparing all patients with anemia to the reference group with anemia. Crude ORs (cORs) and adjusted ORs (aORs) were reported. aORs and 95% confidence intervals (CIs) were calculated using logistic regression. Variables included in the regression analyses were age, nulliparity, education, race and ethnicity, composite medical comorbidity, and tobacco use. A numeric composite of medical comorbidities was created, which included CHTN, pregestational diabetes mellitus and GDM, asthma, thyroid disease, seizure disorder, thromboembolism, cardiac disease, and psychiatric disease. For example, if a patient had 3 of these conditions, then they would receive a value of 3 for this composite. A P value of <.05 or CI excluding 1.0 was considered significant. All statistical analyses were performed, and graphs were created using the SAS software (version 9.4; SAS Institute Inc, Cary, NC).

Using a large, diverse, population-based cohort, we observed a considerable association between unsuccessful treatment of iron deficiency anemia during pregnancy and adverse perinatal outcomes, inclusive of PTB. Accurate diagnosis of the underlying cause of anemia during pregnancy, and enabling correct choice and duration of treatment, may play a key role in reducing maternal and neonatal morbidities and mortalities.

Anemia during pregnancy is one of the most common diagnoses among gravidae, given that it is estimated to affect 40% of women worldwide. 1 The prevalence of anemia among pregnant women in the United States has been estimated to be approximately 10% to 12%. 2 – 4 The American College of Obstetricians and Gynecologists (ACOG) has defined pathologic anemia in pregnancy as a hemoglobin level of <11.0 g/dL in the first and third trimesters of pregnancy and a hemoglobin level of <10.5 g/dL in the second trimester of pregnancy. 5 The ACOG recommends screening all pregnant women for anemia and treating women with suspected iron deficiency with iron supplementation. Of note, 1 study estimated that only 50% of pregnant women with anemia respond to iron therapy. 1 Studies have shown that beyond physiological dilutional anemia of pregnancy, iron deficiency is the most common cause of anemia in pregnancy and is associated with an increased risk of adverse pregnancy outcomes, such as preterm delivery, preeclampsia (PE), cesarean delivery, and perinatal death. 2 , 6 – 8 In addition, women with anemia have an increased risk of perinatal blood transfusions, intensive care unit (ICU) admissions, and postpartum depression. 3 A 2011 Cochrane review evaluated different treatments of iron deficiency anemia in pregnancy for the prevention of perinatal morbidity and concluded that there is insufficient evidence to suggest an efficacious reduction in the incidence of perinatal complications. 9 They concluded that further studies are needed to evaluate whether treatment of iron deficiency anemia can improve maternal and neonatal outcomes. Given these knowledge gaps, we undertook a large population-based cohort study to evaluate whether treatment of anemia during pregnancy with oral iron therapy was associated with a change in the occurrence of associated perinatal outcomes, such as preterm birth (PTB) and PE. We further evaluated whether this effect was modified by treatment and response, as measured by correction of the maternal hemoglobin at delivery.