Impact of optimizing the serum ferritin threshold for diagnosis of iron deficiency: A pre- and post-intervention study using EHR data

Abstract

Abstract Background Iron deficiency (ID), with or without anemia (IDA or IDWA) is one of the most prevalent and underdiagnosed conditions globally, particularly in women. Ferritin is the primary biomarker used to assess iron stores, but historically, sex-specific lower limits of normal (LLNs) have underestimated ID prevalence in females. Recent guidelines recommend a unified ferritin threshold of <30 ng/mL for diagnosing iron deficiency. In September 2023, Rochester Regional Health (RRH) laboratories adopted this evidence-based, sex-independent LLN for ferritin from 10 ng/mL in women and 22 ng/mL in men, to a unified threshold of 30 ng/mL. This study evaluates the impact of this clinical laboratory intervention on IDWA/IDA diagnosis and treatment patterns. Methods We conducted a retrospective cohort study using Epic's SlicerDicer tool to analyze RRH patient data across two 1.5-year periods: before (March 21, 2022–September 20, 2023) and after (September 21, 2023–March 20, 2025) the ferritin LLN update. Patients with ferritin <30 ng/mL were included. The primary outcome was the diagnosis rate of IDWA or IDA. Secondary outcomes included diagnosis rates by sex, iron prescriptions (oral and IV), hematology referrals, number of patients undergoing endoscopy procedures, and pRBC transfusions. Relative risks (RR) with multiplicity corrected confidence intervals (CI) were calculated for the primary and secondary outcomes. Subgroup analyses were performed by age, sex, ferritin level, and ordering department. Changes in ferritin and hematocrit levels at 3–6 months post-test were also assessed. Results Among patients with ferritin levels <30 ng/mL, 22,478 were identified in the pre- intervention period and 27,699 in the post-intervention. The proportion of ferritin tests flagged as low increased from 13% to 32% following the LLN change (p<0.001). The rate of IDWA or IDA diagnosis increased from 51.0% to 58.5% (RR 1.15; CI 1.12–1.17; p<0.001), with the largest increase among females (49.4% to 58.0%; RR 1.17) compared with males (57.0% to 60.2%; RR 1.06). Diagnosis of IDWA rose by 47% (RR 1.47; CI 1.37–1.57; p<0.001) and IDA diagnosis increased by 7% (RR 1.07; CI 1.03–1.10). Oral iron prescriptions rose by 15%, and IV iron administration increased by 17%. Hematology referrals for IDWA or IDA rose by 14% (CI 1.00–1.31), while endoscopy procedures increased by 23% (CI 1.12–1.35). There was no change in rate of packed red blood cell transfusions (RR 0.93; CI 0.79–1.11). Subgroup analyses showed the most pronounced diagnostic increase occurred among patients with ferritin levels 10-29 ng/ml in women and 22-29 ng/ml in men (newly reclassified as low), with a 33% rise in diagnoses (RR 1.33; CI 1.29–1.36). Surgical specialties had the greatest relative increase in diagnosis of IDWA/IDA —from 40.5% to 54.3% (RR 1.34; CI 1.21–1.49). Trends in iron prescriptions mirrored diagnostic patterns across departments and demographic groups. Ferritin levels increased more post-intervention vs. pre-intervention at 3–6 months follow-up (mean difference: 6.9 ng/mL; 95% CI: 5.7–8.1). Females had substantial greater increase (mean difference: 8.2 ng/mL; 95% CI: 6.9–9.5) compared with males (mean difference: 4.0 ng/mL; 95% CI: 1.4–6.6). Hematocrit also increased in both periods, the difference between the two periods was not clinically significant. Conclusion Adopting a unified, evidence-based lower limit of normal for ferritin improved the recognition and management of iron deficiency, especially in women and individuals without anemia. These results underscore the clinical value of updating ferritin thresholds and support the widespread implementation of standardized, sex-independent reference ranges to promote earlier diagnosis and treatment of iron deficiency in routine practice.