Background: Non-anemic iron deficiency (NAID) is estimated to affect up to 42% of early pregnancies and is associated with adverse outcomes including low birth weight, preterm birth, postpartum depression, and impacts on fetal brain development. Despite this, universal screening for iron deficiency is not currently recommended by any national, foundation, or society guidelines. We conducted a cross-sectional study to assess NAID prevalence at the initial obstetric visit and identify clinical characteristics that were predictive of NAID.

Study Design: English- and Spanish-speaking adults ≥18 years with confirmed intrauterine pregnancy <13 weeks were enrolled at their first obstetric (OB) visit. Participants completed surveys regarding relevant medical history, diet, and symptoms associated with anemia and iron deficiency and underwent concurrent laboratory testing with a full iron panel (serum iron, total iron binding capacity, saturation, ferritin) and complete blood count (CBC). Iron deficiency was defined as ferritin <30 ng/mL OR ferritin <50 ng/mL and transferrin saturation <20%; anemia was defined as hemoglobin (Hgb) <11.0 g/dL. Participants were categorized by iron status: iron deficiency anemia (IDA), NAID, or non-anemic iron repletion (NAIR).

Patient demographics, clinical characteristics, and survey responses were summarized by means (sd) or frequencies (%). Summary statistics were compared between NAID/IDA and NAIR groups using independent, two-sample t-tests or Fisher's exact tests. Characteristics significantly different between groups at the 0.05 level were considered for inclusion in a multiple logistic regression model. Stepwise variable selection was used to develop a final multiple logistic regression model estimating the probability of NAID/ID. A Receiver Operating Characteristic (ROC) curve was estimated using cross-validation, and the Area Under the ROC curve (AUC) was estimated with a 95% CI. The best cutoff for the predicted probability was determined by Youden's Index. Sensitivity, specificity, PPV, and NPV of the cutoff were estimated with 95% CIs.

Results: In total, 200 participants were enrolled from May 2024 to March 2025. Of these, 176 had complete data available for analysis. Of these, 54% (n=95) had NAID, 5% (n=9) IDA, and 41% (n=72) were NAIR. There were significant differences between the IDA/NAID vs NAIR groups in all measured red cell parameters, including Hgb (12.6 [1.1] vs 13.0 [0.8], p = 0.009), mean corpuscular volume (87.9 [6.1] vs 91.3 [5], p = 0.0002), and ferritin (20.1 ng/ml [10.2] vs 71.5 [51.6], p <0.0001).

The following clinical characteristics were significantly different between the IDA/NAID vs NAIR groups on univariate analysis: Latinx ethnicity (45% vs 28%, p = 0.035), heavy menses (35% vs 17%, p = 0.021), pagophagia (31% vs 11%, p = 0.003), prior prescription for iron replacement (38% vs 19%, p = 0.012), prior diagnosis of anemia (37% vs 21%, p = 0.03), prior anemia treatment (43% vs 24%, p = 0.01), prior over the counter (OTC) iron supplement (50% vs 30%, p = 0.013), and body mass index (BMI 29.5 vs 26.5, p = 0.004).

There were no significant differences in race, age, number of prior pregnancies, postpartum hemorrage, prior diagnosis of iron deficiency, bleeding disorder, restless leg symptoms, prior blood transfusion, autoimmune disease, prior bariatric surgery, or history of bleeding. No dietary or food safety survey responses were significantly different between groups.

On multivariable logistic regression, only pagophagia (p = 0.006), prior OTC iron supplement use (p = 0.03), and BMI (p = 0.007) remained significant. The model achieved an AUC of 0.68 (95% CI: 0.60, 0.76, P<0.001). Youden index identified a cutoff of 0.551 for the model, yielding sensitivity 69%, specificity 64%, with PPV 73.5%, and NPV 59%.

Conclusion: This cross-sectional study identified that over half of all women presenting for their first prenatal OB visit were iron deficient. Analysis of our survey results found that few established risk factors for ID were significantly associated with ID in this cohort. The prediction model developed from these results was modestly accurate, highlighting the challenges of identifying individuals at risk for ID based on self-reported clinical features alone. These findings support the need for universal screening for iron deficiency in all pregnant individuals at the first prenatal visit rather than targeted assessment based on clinical risk factors alone.

This content is only available as a PDF.
2025
Sign in via your Institution