Rethinking ferritin reference ranges for diagnosing iron deficiency at Denver health Free

Introduction Iron deficiency (ID) is a globally burdensome, yet highly treatable disease, affecting 2 billion total people worldwide. ID disproportionally affects children, premenopausal women, and people of low socioeconomic status (SES).¹ ID precedes iron deficiency anemia and both conditions are associated with adverse health outcomes including fatigue, decreased physical capacity, and impaired neurocognitive function. One way of diagnosing ID is by a ferritin level of <30, however reference ranges for ferritin are frequently derived from population-specific distribution curves. In healthcare systems where patients are more chronically ill at baseline, these adjusted ranges can classify ferritin levels as “normal” even when they fall below the standard diagnostic thresholds for ID. At Denver Health (DH) specifically, ferritin is only flagged as abnormal when < 5 in females and <28 in males. This misalignment in diagnostic cutoffs imposes a risk to missing diagnoses and undertreating a vulnerable population.

Methods This is an IRB approved, retrospective study looking at adults (≥18 years) who were seen at one of the seven Denver Health primary care clinics who had ferritin levels measured between January 1, 2020, and January 1, 2025. Data was obtained through the electronic medical record (EMR) with extracted data including basic demographics (age, sex, race/ethnicity, insurance status, and zip code), laboratory values (CBC, ferritin, iron/TIBC, CRP, ESR, hepatic function panel), clinical diagnoses associated with the visit, treatment history (oral or IV iron supplementation), and hospital admission data. To evaluate the potential influence of institutional diagnostic thresholds on treatment patterns, we analyzed patients with ferritin levels <5 µg/L and 5–30 µg/L. The primary endpoint was the proportion of patients with ID treated for this condition based upon the institutional threshold of ferritin <5 compared to a threshold of a ferritin <30. Secondary outcomes included subgroup analysis for treatment rates by demographic factors.

Results We identified 5,982 patients at DH primary care clinics with ferritin levels <30, meeting criteria for ID. Of these, 18.0% (n=1,077) had ferritin levels <5 µg/L, and 82.0% (n=4,905) had levels between 5–30 µg/L. Despite this, only 51% of patients received any iron supplementation. Treatment rates varied significantly by ferritin level. 85.3% of those with ferritin <5 received treatment, compared to just 43.5% of those with ferritin 5–30 (p <0.0001). Females (n=5,296) represented the majority of the study population compared to males (n=676). Females were treated 85.6% of the time if ferritin was <5, but only 41.8% if ferritin was 5-30 (p<0.0001). Males interestingly had a similar disparity despite having higher reference ranges of 84.9% for <5 and 55.7% for 5-30. Globally, patients with ferritin levels 5-30 are less likely to receive therapy at our DH primary care clinics. Racial differences in ferritin levels approached significance (p=0.0567), with Black or African American patients representing 20.1% and Asian patients 4.1% of the patient population.

Conclusions This study highlights the impact that laboratory reference ranges have on the systemic underdiagnosis and undertreatment of ID in DH safety-net clinics. This discrepancy in treating ID is likely the result of institutional reference ranges at Denver Health, as only ferritin values <5 for females are flagged as abnormal, however this does not fully explain the undertreatment in males. These thresholds, especially in women, appear to drive clinical decisions rather than current clinical guidelines. Our findings suggest that many patients at Denver Health who have clinically significant low ferritin (<30), do not receive neededtreatment. Reassessing and changing ferritin thresholds to align with evidence-practice represents a low-cost, high-impact intervention to improve recognition ID and addressing disparities and care in a vulnerable population.