Intravenous iron for treatment of iron deficiency anemia during pregnancy and associated maternal outcomes

Abstract Background More than 40% of pregnant patients worldwide are anemic, with at least half resulting from iron deficiency anemia (IDA). Anemia in pregnancy is linked with adverse maternal and neonatal outcomes. Treatment for IDA is iron supplementation; however, the optimal route of administration remains unclear. We sought to investigate whether patients with IDA who received intravenous iron (IVI) had decreased odds of maternal morbidity compared to patients who did not. Methods This is a retrospective cohort study of pregnant patients with presumed IDA with term deliveries at a tertiary hospital from 2013–2021. Data were extracted from the hospital's electronic medical record using standardized definitions and billing codes. Patients who received antepartum IVI were compared to patients who did not. The primary outcome was a maternal morbidity composite inclusive of receipt of blood transfusion, hysterectomy, admission to the intensive care unit or death. Bivariate analyses and multivariable logistic regression modelling were performed adjusting for potential confounders. Results Of 45,345 pregnancies, 5054 (11.1%) met eligibility criteria. Of these, 944 (18.7%) patients received IVI while 4110 (81.3%) did not. Patients who received IVI had higher risk baseline characteristics. They experienced a greater increase in hematocrit from pregnancy nadir to delivery admission (4.5% vs. 3.3%, p < .01). Despite this, patients who received IVI had higher odds of the maternal morbidity composite (OR 1.47, 95%CI 1.11–1.95). This finding persisted after adjusting for potential confounders, although the strength of the association became attenuated (aOR 1.37, 95%CI 1.02–1.85). Odds of the morbidity composite were not elevated among patients who received a full IVI treatment course (OR 1.2, 95% CI 0.83–1.90). Discussion Odds of the maternal morbidity composite were increased among patients who received IVI despite greater increases in hematocrit. The effect was attenuated after adjusting for potential confounders and was not significant among patients who completed a full treatment course.

Oral iron supplementation is the typical first-line approach as it is safe, inexpensive, and readily available. However, limitations include reliance on patient adherence and gastrointestinal side effects such as metallic taste, gastric irritation, nausea, and constipation [16,17]. Intravenous iron (IVI) is an alternative that helps to mitigate these limitations. Previous meta-analyses comparing oral and IVI found that while both are effective at increasing hematologic indices, IVI was associated with achieving higher hemoglobin levels, fewer adverse effects, and lower rates of discontinuation [18][19][20][21]. Despite this, no significant improvements in rates of maternal transfusion, cesarean delivery, or neonatal hemoglobin levels were reported [19]. However, maternal outcome data is overall sparse, owing to small sample sizes and infrequent reporting of clinical outcomes.
Given the aforementioned benefits of IVI, we hypothesized that patients who received antenatal IVI would be less likely to experience anemia-related morbidity. We sought to test this hypothesis by examining a cohort of patients with IDA and comparing maternal outcomes between those who received IVI and those who did not.

Methods
We performed a retrospective cohort study of patients who delivered at 37 weeks of gestation at a single, urban, tertiary hospital with a presumed diagnosis of IDA from 1 October 2013, when our electronic medical record (EMR) was introduced, to 31 May 2021. IDA diagnosis was based on diagnosis codes and anemia was confirmed by either pregnancy hemoglobin <11 g/dL or hematocrit <33%. We included only fullterm deliveries to ensure patients would have had sufficient opportunity for IVI if indicated. Patients at high risk for postpartum hemorrhage due to placental pathology (placenta previa and placenta accreta spectrum) were excluded based on diagnosis codes. Patients with other hematologic diagnoses (other etiologies of anemia, clotting and bleeding disorders) were similarly identified and excluded.
Typical institutional practice is to initiate oral iron supplementation promptly after an IDA diagnosis is made by using trimester-specific thresholds delineated by the American College of Obstetricians and Gynecologists (first <33%, second <32%, third <33%) and confirmatory iron studies [22]. A hospital-specific protocol then recommends a transition to IVI if hematocrit does not increase by 3% within 4 weeks. This protocol recommends five doses of iron sucrose 200 mg, which can be given as frequently as every three days.
We compared outcomes for individuals who received any antenatal IVI to individuals who did not receive any IVI. At our institution, IVI is administered in the hospital, and therefore these data were extracted directly from the medication administration record. The primary outcome was a maternal morbidity composite which included receipt of packed red blood cells during the delivery hospitalization, hysterectomy during the delivery hospitalization, intensive care unit admission within 30 days of delivery, or maternal death within one year. These outcomes were selected due to each being associated with maternal anemia [5][6][7][8][9][10][11][12][13][14][15]. Secondary outcomes included fetal or neonatal death. Outcome data were extracted directly from the EMR.
Maternal characteristics including race, ethnicity and marital status were obtained through patient selfreport within the EMR. All other maternal and neonatal characteristics were extracted directly from the EMR. Pre-pregnancy BMI was calculated using the first height and weight during the pregnancy. Nulliparity was defined as no previous births 20 weeks of gestation. Delivery EBL was based on provider documentation in the delivery record. Postpartum hemorrhage was defined as EBL 1000 ml with missing values assumed to be <1000 ml. Anticoagulation and aspirin use were based on enoxaparin, fondaparinux, heparin, warfarin or aspirin within the patient's medication list.
Hematologic indices were compared based on hematocrit as institutionally this variable drives clinical decisions. The lowest antepartum hematocrit was recorded as well as the first hematocrit value from the delivery hospitalization encounter. Median values and interquartile ranges were calculated.
Bivariate analyses were performed using chi-square and Fisher's exact tests for categorical variables, as appropriate, and Wilcoxon rank-sum tests for continuous variables based on non-parametric distributions. Both unadjusted logistic regression and multivariable logistic regression modelling of the maternal morbidity composite were performed, generating crude and adjusted odds ratios (ORs) and 95% confidence intervals (CIs). Variables significant at p < 0.05 in bivariate analyses were included for model adjustment. Delivery EBL was additionally used due to its strong independent effect on the outcome.
A subgroup analysis was performed among patients who received any IVI. Individuals receiving <5 doses were compared to individuals receiving 5 doses. Five doses were selected as the dichotomous cut point as 1000 mg of elemental iron was considered a complete treatment course. ORs for the maternal morbidity composite were determined for each group, using the non-IVI group as the reference group.
Analyses were performed using SAS 9.4 (SAS Institute, Cary, NC). This study was approved by the Yale Institutional Review Board with a waiver of informed consent.

Results
Of the 45,345 deliveries in the study period, 5054 (11.1%) had a diagnosis of IDA and met eligibility criteria. Of this cohort, 944 (18.7%) received antenatal IVI while 4110 (81.3%) did not. The median gestational age at the time of IVI initiation was 33.2 weeks (IQR 31.4-36.7).
Among patients with presumed IDA, patients who received IVI were more likely to be multiparous, be prescribed aspirin and had a different racial and ethnic distribution compared to patients who did not (Table 1).
Furthermore, patients who received IVI were more likely to have worse hematologic indices at pregnancy nadir and at delivery admission (Table 1). Specifically, those who received IVI had a significantly lower median hematocrit nadir (28.8% vs. 30.6%, p < .01) and slightly lower median admission hematocrit (33.3% vs. 33.6%, p ¼ .01) compared to patients who did not receive IVI. Additionally, patients who received IVI were more likely to have severe anemia, defined as nadir hematocrit <25%, (9.2% vs. 2.9%, p < .01) than those who did not receive IVI. Notably, patients who received IVI had a higher median increase in hematocrit from pregnancy nadir to delivery admission compared to patients who did not receive IVI (4.5% vs. 3.3%, p < .01).
Overall, 274 (5.4%) patients were identified as experiencing the maternal morbidity composite with 68 (7.2%) in the IVI group and 206 (5.0%) in the non-IVI group. In the unadjusted analysis, patients who received IVI had significantly increased odds of the morbidity composite (OR 1.47, 95%CI 1.11-1.95) ( Table 1). After adjusting for potential confounders including race and ethnicity, parity, aspirin use, delivery EBL, and hematocrit on admission, the increased risk of the maternal morbidity composite with IVI use persisted, but the strength of the association was attenuated (aOR 1.37, 95%CI 1.02-1.85). The most frequent morbidity was the receipt of maternal blood transfusion. Within the IVI group, of the 35 patients who received blood, the median was 2 units (IQR 1-2) while within the non-IVI group, of the 114 patients who received blood, the median was 2 units (IQR 1-3)[p ¼ .05]. For the secondary outcome, there was no significant difference in odds of fetal or neonatal mortality between patients who received IVI and patients who did not (0.2% vs 0.1%, p ¼ .24).
Among patients who received any IVI (n ¼ 944), the median number of doses was 4.5. 472 (50%) received <5 doses while 472 (50%) received 5 doses. Patients who received 5 doses were more likely to be of advanced maternal age, have commercial insurance, and be married compared to those who received <5 doses (Table 2). Patients who received 5 doses had a different racial and ethnic and pre-pregnancy BMI distribution than patients who received <5 doses. Patients who received 5 doses were earlier in gestation at the time of IVI initiation (median 32.5 weeks, IQR 30.9-25.6 vs median 34.8 weeks, IQR 33.7-37.6). Furthermore, patients who received 5 doses had a similar median hematocrit nadir (29.0% vs 28.6%, p ¼ . 19), yet, had a higher median delivery hematocrit (34.5% vs 32.1%, p <.01) compared to those who received <5 doses. Patients who received 5 doses had a higher median increase in hematocrit from pregnancy nadir to delivery admission compared to patients who received <5 doses (5.4% vs. 3.6%, p < .01) ( Table 2). Finally, patients who received <5 doses of IVI had increased odds of the maternal morbidity composite (OR 1.7, 95%CI 1.2-2.4) compared to patients who received no IVI. On the contrary, the odds of the maternal morbidity composite were not increased among patients who received 5 doses (OR1.2, 95%CI 0.8-1.9).

Discussion
We found that patients with presumed IDA who received antenatal IVI, despite having a more robust increase in hematocrit, did not have maternal morbidity benefits when compared to patients who did not receive IVI. These findings were unexpected and differed from our study hypothesis.
With regard to hematologic indices, our findings were in line with existing literature demonstrating a more robust increase in indices with the use of IVI [18][19][20]. Furthermore, in the sub-analysis, we found that patients who received 5 doses of IVI had a greater increase in median hematocrit from pregnancy nadir to delivery admission compared to patients who received <5 doses, suggesting there may be a benefit of prolonged IVI therapy. While many factors may contribute to these findings, we speculate that this more robust response to IVI is related to improved systemic absorption as well as improved treatment adherence given the decreased side effects with IVI.
Despite this more robust increase in median hematocrit, the unadjusted analysis showed increased odds of the maternal morbidity composite in the IVI group compared to the control group (OR 1.47, 95%CI 1.11-1.95). After adjusting for potential confounders, Data are presented as n (%) or median (interquartile range). BMI: body mass index; IVI: intravenous iron; OR: odds ratio; CI: confidence interval; aOR: adjusted odds ratio. a Blood loss as estimated by the provider and recorded in the nursing electronic delivery record. b As documented in electronic health record medication list as enoxaparin, fondaparinux, heparin or warfarin or aspirin.
increased odds persisted, although the effect was attenuated (aOR 1.37, 95%CI 1.02-1.85). It should be noted, though, that rates of maternal transfusion (3.7% vs. 2.8%), hysterectomy (0.9% vs. 0.6%), and ICU admission (3.2% vs. 2.0%) were low in both groups. We suspect that this increased morbidity is not related to dangers of IVI, but rather that patients who were prescribed IVI were considered by their providers to be at increased risk for anemia-related morbidity. While we attempted to adjust for some of these potential confounders in our adjusted analysis, we likely were unable to account for all differences. The degree of baseline anemia was more pronounced in the IVI group compared to the non-IVI group. Increased incidence of severe anemia (9.2% vs. 2.9%, p < .01) and lower median hematocrit nadir (28.8% vs. 30.6%, p <.01) likely contributed to the difference in the primary outcome. Despite this clinically significant difference in antepartum hematocrit nadir, the hematocrit at the time of delivery admission was clinically quite similar, although still statistically significant (33.3% vs. 33.6%, p ¼ .01). While the degree of anemia was adjusted for in the analysis, it may not have fully accounted for these inherently different populations.
Additionally, other high-risk characteristics were more often exhibited by patients who received IVI. We sought to account for several potential confounders through study methodology by excluding high-risk placental pathologies and hematologic diagnoses. In our statistical analysis, we further adjusted for significant baseline differences. When adjusting for these potential confounders, the strength of the association was attenuated suggesting that these characteristics played an important role in the development of the primary outcome. Despite our best efforts, it is likely that we were unable to account for all potential confounders and suspect that the IVI group had additional high-risk baseline characteristics making them more susceptible to anaemia-related morbidity.
Within the composite, receipt of a blood transfusion was the most commonly experienced morbidity. The median number of units transfused in the IVI group was 2 (IQR 1-3) versus 2 (IQR 1-2) in the non-IVI group(p ¼ .05). This finding narrowly failed to achieve statistical significance and given the low frequency of transfusion, our ability to draw conclusions is somewhat limited. Further studies with larger cohorts are warranted to determine if receipt of IVI allows for the receipt of fewer units when transfusion is indicated.
The sub-analysis additionally showed that patients who received 5 doses had a greater increase in median hematocrit compared to those who received <5 doses (5.4% vs. 3.6%, p < .01). We found that half of the patients received <5 doses and thus likely did not have the full effect that could have been offered by the full treatment protocol. As suspected, in a subanalysis comparing only patients who received 5 doses to patients who did not receive IVI, the odds of experiencing the maternal morbidity composite were no longer significant (OR 1.2, 95%CI 0. 8-1.9). This observation is interesting and seems to suggest that despite higher risk baseline characteristics and more significant anemia, with a complete treatment course, these patients can have similar odds of the maternal morbidity composite compared to controls.
Most patients started IVI in the third trimester. Patients were likely identified by routine third-trimester labs, underwent a 4-week trial of oral iron and then initiated IVI following a suboptimal response. Given late identification with a subsequent trial of oral iron, many patients simply did not have the opportunity for a full treatment course. Opportunities to improve outcomes may involve earlier recognition. Earlier screening for iron deficiency may be one strategy which is supported by data showing 42% of patients are iron deficient in the first trimester with or without concomitant anemia [23]. Further study is needed to evaluate whether early supplementation in non-anemic iron deficient patients provides benefit. Additionally, patients with advanced gestation at the time of diagnosis may benefit from forgoing a trial of oral iron. These areas for potential intervention warrant further study.
The strength of this study is that it aims to target a gap in the literature. Existing obstetric data studying IVI use for the treatment of IDA focuses primarily on safety and improving hematologic indices. There is a paucity of literature specifically examining its effect on maternal outcomes. This study aims to target this gap in the literature using real-world practice settings. To our knowledge, this is one of the largest studies with the primary aim of studying maternal morbidity outcomes related to the receipt of antenatal IVI [18,19].
This study, however, is not without its limitations. First, the retrospective study lends itself to the potential for selection bias. We attempted to account for this in our study design and adjusted analysis, however, we were likely unable to account for all confounders. Additionally, the study methodology relied on accurate and complete documentation within the EMR. Whenever able, we performed quality control of diagnosis codes, however, the potential for coding errors remains. One such limitation is that the diagnosis of IDA relied on diagnosis codes and hematologic parameters. We did not have access to iron studies to confirm the diagnosis of iron deficiency. Despite this, we believe our cohort truly reflects an iron-deficient population due to IDA being the primary aetiology of antenatal anaemia as well as the exclusion of patients with confounding hematologic diagnoses. Additionally, the prevalence IDA in our population was 11%, which is conservatively in line with national estimates [24].
Another potential limitation was that while IVI protocols were standardized, oral iron dosing and formulations varied among providers. We were unable to collect specific data on practice patterns due to the frequent use of over-the-counter iron which could not be abstracted accurately from the EMR. For this reason, our study can only comment on IVI use and not oral iron use.
This study identified additional areas warranting further investigation. Preterm birth was intentionally excluded to allow for sufficient opportunity for IVI exposure, however, future investigations should include all deliveries as preterm birth and small for gestational age have been associated with IDA. Finally, cost data need to be evaluated as the non-obstetric literature has advocated for the expansion of IVI use due to the cost-benefit of this practice [25].
All of this taken together demonstrates that while patients who received IVI had higher odds of the maternal morbidity composite, they only narrowly fared worse despite having higher risk baseline characteristics compared to the control group. Through a more substantial increase in hematocrit with receipt of IVI, patients with lower pregnancy hematocrit nadirs were able to achieve a clinically similar hematocrit by the time of delivery admission. A randomized control trial is necessary to conclusively study the effect of IVI on maternal morbidity in patients with IDA.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
The author(s) reported there is no funding associated with the work featured in this article.