Cost‐Effectiveness of Ferritin Screening Thresholds for Iron Deficiency in Reproductive‐Age Women

ABSTRACT

Iron deficiency (ID) is a top five leading cause of disability‐adjusted life‐years in women of reproductive age around the world. Despite its enormous health burden, no screening guidelines exist for the detection and treatment of ID in women of reproductive age. We sought to determine the cost‐effectiveness of screening versus no screening for ID in women of reproductive age in the United States. A lifetime simulation of women of reproductive age was conducted using a Markov cohort model under three strategies: (1) no screening, (2) screening at a ferritin threshold of 15 μg/L, and (3) screening at a ferritin threshold of 25 μg/L, from the US health system perspective, and at a willingness‐to‐pay threshold of $100 000/quality‐adjusted life year (QALY). Epidemiologically informed ID prevalence estimates sourced from the National Health and Nutrition Examination Survey were employed for model parameterization. The primary outcome was the incremental cost‐effectiveness ratio (ICER, in $/QALY). Base‐case results for the three strategies accrued $209 700, $210 200, and $210 200 discounted lifetime costs and 23.6, 24.0, and 24.4 discounted lifetime QALYs, respectively. Screening at a ferritin threshold of 25 μg/L was the cost‐effective intervention with an ICER of $680/QALY (95% credible interval $350–$750/QALY). In dual base‐case analyses examining intravenous rather than oral iron repletion for treatment, screening at a ferritin threshold of 25 μg/L remained the cost‐effective intervention with an ICER of $2300/QALY (95% CI $1800–$3800/QALY). In probabilistic sensitivity analyses, screening at a ferritin threshold of 25 μg/L was the cost‐effective intervention in 100% of 10 000 second order Monte Carlo iterations.

1. Introduction

Iron deficiency (ID) is the most common and widespread micronutrient deficiency in the world and is independently associated with both all‐cause mortality and reduced health‐related quality of life [1, 2]. It is among the top five global causes of disability‐adjusted life‐years (DALYs) in women of reproductive age [1, 3, 4, 5]. ID is a causal risk factor for progression to iron deficiency anemia (IDA), which accounts for half of the 1.92 billion anemia cases globally and is the leading cause of years lived with disability (YLD) in low‐ and middle‐income countries (LMIC's) [3, 6, 7]. Despite clear recognition as a global and public health concern, the exact prevalence of ID globally is unclear, in part due to underreporting in LMIC's. However, recent findings from the United States and Canada indicate that ID may affect nearly 40% of nonpregnant females of reproductive age, most of whom have ID without anemia [8, 9, 10, 11]. Even in high‐income countries, structural barriers, such as across socioeconomic strata and self‐identified ethnicity and/or race, continue to complicate the proper identification and management of ID for women [9, 10, 12, 13].

In the United States, the burden of ID is exacerbated by underdiagnosis and undertreatment due, in part, to inappropriately low reference intervals for serum ferritin—a biomarker of total body iron stores used to diagnose ID [11, 14]. These intervals, which place the lower limit of normal ferritin at 15 μg/L or less (in many cases, in the single‐digits), are built upon biased historic patient samples and flawed assumptions of a Gaussian distribution of disease that do not reflect the true prevalence of ID in the general population [14, 15]. This includes the World Health Organization (WHO) guidelines, which recommend a ferritin threshold of 15 μg/L for ID diagnosis based upon expert opinion [5, 16, 17, 18]. Several prospective and cross‐sectional studies have clearly established that the true physiologic threshold for ID lies between serum ferritin values of 25–30 μg/L, or even as high as 50 μg/L. [14, 16, 17, 19, 20] Clinically, the discrepancy between laboratory and WHO thresholds versus physiologically based ferritin cut‐offs leads to systematic underrecognition of ID and perpetuation of its untreated sequelae. In fact, recent work has shown that the numbers needed to screen (NNTS) for ID without anemia are 7 and 3, and for ID with anemia are 16 and 36 at ferritin values of < 15 and < 25 μg/L, respectively [8]. Currently, no guidelines exist in the United States for the regular screening of ID in reproductive‐age women or standardization of ferritin reference ranges in accordance with best available evidence [14, 21]. We sought to fill this gap by conducting, to our knowledge, the first cost‐effectiveness analysis of screening ferritin thresholds for the treatment of ID in this patient population.

2. Methods

2.1. Model Overview

We built a Markov cohort model of reproductive‐age women in the United States to examine the cost‐effectiveness of screening for ID with a ferritin threshold of (1) 25 μg/L versus (2) 15 μg/L (the current WHO diagnostic threshold) versus (3) no screening (i.e., the status‐quo). A 25 μg/L threshold was chosen as more physiologically appropriate than the WHO diagnostic standard of 15 μg/L, while remaining specific for ID, based on previously published work [16, 17]. Patients in our base model begin by undergoing ferritin testing for ID per screening strategy, and may progress to health states that include (1) being healthy (without ID), (2) properly identified with ID and treated, (3) living with undiagnosed ID, or (4) death; all living patients return to annual screening throughout their reproductive lifetime ending at 51 years, the average age of menopause in the United States (Figure 1) [11, 22, 23, 24, 25]. Patients identified as iron deficient are treated with iron supplementation at a rate commensurate with average menstrual losses—in line with previously proposed screening algorithms [11]. In dual base‐case analyses, patients are treated with (1) alternate‐day oral ferrous sulfate per best available evidence assuming optimal gastrointestinal absorption [26, 27], or (2) intravenous (IV) iron supplementation (see Supporting Information: Model Overview). All patients who are treated with iron supplementation may experience iron‐related adverse events including gastrointestinal effects (e.g., constipation, nausea, vomiting) for oral supplementation and infusion‐related anaphylactic reactions in the case of IV supplementation, which may lead to death.

FIGURE 1.

State‐transition schematic illustrating modeled health‐states for ferritin‐based screening versus no screening for iron deficiency (ID). In the screening strategy, individuals undergo initial ferritin screening at the age of 18 years, followed by annual testing until the age of 51. Based on ferritin screening thresholds (15 or 25 μg/L), individuals transition into one of three initial health states: Healthy (without ID), ID treatment, or undiagnosed ID. Individuals receiving iron supplementation (ID treatment) may transition to untreated ID due to permanent treatment discontinuation from either adverse events (such as gastrointestinal distress with oral iron supplementation or anaphylaxis with intravenous supplementation) or personal choice. Individuals in all health states transition to the absorbing state of death at age‐ and sex‐specific background mortality. In the comparator strategy (no ID screening), individuals remain either in the healthy (without ID) state or undiagnosed ID until death occurs. ID refers to women with iron deficiency without anemia in the base‐case, and to all women with ID (with and without anemia) in scenario analysis. [Color figure can be viewed at wileyonlinelibrary.com]

We assumed a starting age of 18 years to reflect the best available age‐specific data in all input parameters, recognizing that the median age at menarche in the United States is estimated to be 12 years and may be decreasing over time [28, 29, 30]. Model transition‐state cycles were 6 months in duration, and analysis was conducted over a lifetime time‐horizon. The primary outcome of interest was the incremental cost‐effectiveness ratio (ICER) and its reformulation, the incremental net‐monetary benefit (iNMB). We performed this analysis from the US healthcare system perspective, doing so across a range of conventionally accepted willingness‐to‐pay (WTP) thresholds in the United States ($50 000–$150 000 per quality‐adjusted life year [QALY]) and discounting cost and effectiveness by 3% annually, as recommended in the US context [31, 32]. We constructed our model using TreeAge Pro Healthcare 2024 (TreeAge Software, Williamstown, MA). Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines were implemented where applicable.

2.2. Transition Probabilities

Base‐case estimates and ranges for all input parameters used in our model are summarized in Table 1. The prevalence of ID, as defined at each ferritin cut‐off value, was sourced from epidemiologically informed estimates of US females aged 12–21 years, itself sourced from National Health and Nutrition Examination Survey (NHANES) data spanning 2003–2010 and 2015–2020 [8]. Transition probabilities for gastrointestinal or anaphylaxis adverse events related to oral or IV iron, respectively, were sourced from previously published clinical trials of iron supplementation as well as hemovigilance data from the WHO Vigibase 2008–2017 and Food & Drug Administration Adverse Event Reporting System 2014–2019 [33, 34, 35]. US Life Tables were used for age‐ and sex‐specific background mortality rates [36].

TABLE 1.

Base‐case input parameters and probability distributions.

	Base case estimate	Probabilistic sensitivity analysis distribution and range	References
Clinical parameters
Cohort age at start	18	Fixed	—
Age of menopause	51	Fixed	McKinlay et al.
Discount rate	0.03	Fixed	Sanders et al.
Prevalence of ID at ferritin 15 μg/L	0.17	ß‐PERT (0.154, 0.192)	Weyand et al.
Prevalence of ID at ferritin 25 μg/L	0.386	ß‐PERT (0.358, 0.409)	Weyand et al.
Prevalence of iron deficiency without anemia ID patients	0.836	ß‐PERT (0.808, 0.864)	Weyand et al.
Probability of oral ferrous sulfate related gastrointestinal AE	0.190145335	ß‐PERT (0.152116, 0.228174)	von Siebenthal et al.
Probability of IV iron dextran related anaphylaxis	7.90E‐06	ß‐PERT (6.32E‐0.6, 9.48E‐0.6)	Durup et al.
Utilities
Baseline utility of ID health state	Age‐dependent	ß‐PERT	Peuranpaa et al.
Baseline utility of perfect health	Age‐dependent	ß‐PERT	Jiang et al.
Utility decrement of oral iron related gastrointestinal AE	0.0878	ß‐PERT (0.07024, 0.10536)	Sullivan et al.; Ito et al.
Utility decrement of IV iron anaphylaxis	0.09	ß‐PERT (0.072, 0.108)	Shaker et al.
Costs (USD)
Average annual healthcare costs, US women	Age‐dependent	Fixed	US Medical Expenditures Panel Survey
Cost of ferritin lab test	13.63	Gamma (305, 22.403)	CMS Clinical Laboratory Fee Schedule
Cost of complete blood count with differential	7.77	Gamma (225, 0.035)	CMS Clinical Laboratory Fee Schedule
Cost of 6 months of oral ferrous sulfate supplementation, alternate day dosing	66.9	Gamma (306, 4.585)	GoodRx
Cost of 1000 mg IV iron dextran	353.38	Gamma (225, 0.637)	CMS Average Sales Price Data
Cost of IV iron dextran administration, per nursing time	252.75	Gamma (301, 1.194)	US Bureau of Labor Statistics

2.3. Costs

All costs were adjusted to 2024 US dollars using the medical component of the consumer price index [37]. The costs of ferritin and complete blood count laboratory testing and oral ferrous sulfate were sourced from the Centers for Medicaid & Medicare Services (CMS) 2024 Clinical Laboratory Fee Schedule and over‐the‐counter sales price data (e.g., GoodRx), respectively [38, 39]. The cost of IV iron dextran was sourced from CMS 2024 Average Sales Price data [40]. The cost of IV iron administration was also taken into account using median hourly nursing wages per the US Bureau of Labor Statistics [41]. Age‐ and sex‐specific average annual healthcare costs for women were incorporated and sourced from the US Medical Expenditures Panel Survey [42].

2.4. Quality‐Adjusted Life‐Years

Health outcomes estimated by our model were expressed in QALYs, a measure that accounts for both health‐related quality of life and length of life. Health‐related quality of life was quantified by utility values, which range from 0 (death) to 1 (perfect health). Disease‐ and sex‐specific QALYs for patients with ID were informed by EQ‐5D (EuroQol‐5D) index values drawn from the largest (n = 236) prospective study of the effects of ID and anemia on health‐related quality of life in women of reproductive age [43]. These utilities were then adjusted to account for potential geographic and temporal differences to the modern‐day United States, derived using concurrent country‐specific general population EQ‐5D reference norms [44, 45]. We opted for this approach because it conservatively produced a lower quality‐of‐life benefit for starting with a lower serum ferritin than what was reported in this prospective cohort after adjusting for menstrual blood loss and the degree of anemia. The baseline utility of perfect health for US women was multiplicatively incorporated in an age‐dependent manner based on the EQ‐5D‐5L for the general population of the United States [46]. The utility decrement of adverse events related to oral or IV iron supplementation was also taken into account [47, 48, 49].

See Supporting Information: Scenario, Threshold, and Sensitivity Analyses.

3. Results

3.1. Dual Base‐Case Analyses

The estimated total cost, QALYs, and net monetary benefit associated with each screening strategy at a life‐time horizon are reported in Table 2. In the base case, screening for ID at a threshold of 25 μg/L of ferritin is the cost‐effective strategy compared to both no screening and a threshold of 15 μg/L. Compared to the status quo (no screening), a 25 μg/L ferritin cut‐off costs an additional $540 while generating an incremental QALY gain of 0.8, yielding an ICER of $680/QALY (95% credible interval $350–$750/QALY). In the second base case analysis in which all patients found to have ID are treated with IV iron dextran, as opposed to oral supplementation, screening with a ferritin threshold of 25 μg/L is again the preferred strategy, accruing an additional $1800 in costs and 0.8 QALYs compared to the status quo, for an ICER of $2300/QALY (95% credible interval $1800–$3800/QALY).

TABLE 2.

Dual base case (oral and intravenous iron treatment) analysis results and net monetary benefits for ferritin screening at 25 μg/L, 15 μg/L, and no screening from a US healthcare system perspective. Screening at 25 μg/L ferritin is the cost‐effective strategy. Reference for all accepted willingness‐to‐pay thresholds in the United States ranges from $50 000/QALY to $150 000/QALY.

Strategy	Cost (USD)	Incremental cost (USD)	Effectiveness (QALYs)	Incremental effectiveness (QALYs)	Net monetary benefit (USD)
Base‐case: oral iron treatment
No screening	209 700	—	23.6	—	2,150 000
Ferritin 15 μg/L	210 200	490	24.0	0.3	2 184 000
Ferritin 25 μg/L	210 200	540	24.4	0.8	2 227 000
ICER = $680/QALY [95% CI $350–$750/QALY]
Base‐case: intravenous iron treatment
No screening	209 700	—	23.6	—	2,150 000
Ferritin 15 μg/L	210 700	1100	24.0	0.4	2 184 000
Ferritin 25 μg/L	211 500	1800	24.4	0.8	2 228 000
ICER = $2300/QALY [95% CI $1800–$3800/QALY]

Abbreviations: QALY = quality‐adjusted life‐years; USD = United States dollar.

3.2. Deterministic, Probabilistic, and Threshold Sensitivity Analyses

The parameters to which the model was most sensitive were, in descending order: the utility of living without ID, the utility of living with ID, and the prevalence of ID defined at a ferritin threshold of 25 μg/L. However, no change in these or any other parameters caused the model to favor screening at 15 μg/L or no screening (Figure 2). Two‐way sensitivity analysis demonstrated that screening at 25 μg/L continued to be the favored strategy when the utility of the ID and perfect health states are pushed to the extreme maximum and minimum of their ranges, respectively (i.e., under the assumption of near parity in health‐related quality‐of‐life between patients with ID and those who are healthy) (Figure S1). Further, threshold analysis revealed that, across commonly accepted WTP thresholds from $50 000–$150 000/QALY, screening at 25 μg/L ferritin remains the cost‐effective strategy with a difference in utility between living with ID and without ID equal to < 0.004–0.005 QALYs in the dual base‐case (Table S1). An additional threshold analysis indicated that at a WTP of $100 000/QALY, the cost of an individual work‐up to establish the underlying cause of ID in diagnosed patients may reach $34 000 before screening at 25 μg/L becomes cost‐ineffective; at a WTP of $50 000/QALY and $150 000/QALY these thresholds are $17 000 and $52 000, respectively (Figure S3). In probabilistic sensitivity analysis, screening at 25 μg/L was favored in 100% of 10 000 Monte Carlo iterations in the dual base‐case.

FIGURE 2.

Tornado diagram of one‐way sensitivity analyses demonstrating net monetary benefits with ferritin screening (25 μg/L) versus no screening. Each row illustrates analysis results when one model parameter is varied across its range. Parameters that produced > 10% change in the incremental net monetary benefit (iNMB) when varied are shown. Ranges utilized in analyses are detailed in Table 1. Blue denotes iNMB changes associated with lower values, while red denotes those associated with higher values. iNMB = incremental net monetary benefit; QALY = quality‐adjusted life‐year; USD = United States dollar; WTP = willingness‐to‐pay. [Color figure can be viewed at wileyonlinelibrary.com]

See Supporting Information: Scenario Analyses.

4. Discussion

We assessed the cost‐effectiveness of both an annual ferritin‐based screening strategy for the diagnosis and treatment of ID (versus no screening) and that of a more physiologically relevant higher ferritin threshold from a US healthcare system perspective [15, 16, 18]. Our results across dual base‐case, sensitivity, and scenario analyses indicate that ID screening with a ferritin cut‐off of 25 μg/L is a cost‐effective intervention that improves quality‐adjusted life expectancy in reproductive‐age women compared to screening at 15 μg/L (WHO diagnostic standard) or no screening, regardless of the route of iron supplementation. We posit that the low ICERs found across our analyses are driven primarily by the relatively low cost of both ID screening and treatment, the specificity of ferritin‐based testing for ID, alongside the benefit to health‐related quality‐of‐life of correction of ID for reproductive‐age women. This benefit is mediated by a greater proportion of women accurately diagnosed and treated for ID at a higher ferritin threshold, and remains robust with a marginal increase in utility from ID treatment and with a maximum cost of additional diagnostic work‐up for ID that is far below what reproductive age women experience.

Our model results extend the existing literature in several ways. First, we add to the large body of evidence and calls from clinicians across specialties, despite lack of such guidance from either the US Preventative Service Task Force (USPSTF) or Centers for Disease Control, for regular ferritin‐based screening of ID [14, 15, 21, 43]. This urgent need is underscored by growing recognition of the enormous healthcare burden of ID (likely underestimated due to inconsistent definitions and incomplete reporting), and its impairing effects on energy, mood, and cognition when untreated [8, 13, 50, 51, 52]. Though ID is estimated to underlie over half of the nearly 2 billion cases of anemia worldwide, evaluation for anemia itself is insufficient to adequately identify individuals with ID as patients may persist with long periods of negative iron balance and iron‐deficient states before any effect on hemoglobin production can be observed [8, 53, 54, 55]. Proper testing not only identifies ID before further development of iron‐restricted erythropoiesis and subsequent anemia, but proactive iron supplementation may also protect against progression of symptoms in latent or early‐symptomatic patients [11]. Even when screening does occur, studies from high‐income jurisdictions suggest that ID often persists for years before resolution, particularly among females of reproductive age, again highlighting the need for regular, repeat screening and prompt action once ID is diagnosed [13, 56].

Second, our results demonstrate the cost‐effectiveness of standardizing and raising the lower limit of normal ferritin above the WHO recommended 15 μg/L based on rigorous clinical and translational studies of true physiological levels of ID [15, 16, 17]. Importantly, this model does not assume a ferritin threshold of 25 μg/L as most sensitive and specific for the proper diagnosis of ID. Rather, it builds upon a preponderance of evidence indicating that a ferritin of at least 25–30 μg/L (and even as high as 50 μg/L) is a more appropriate threshold than previously established cut points for ID [14, 57, 58]. This proof‐of‐concept analysis for population‐level implementation of increased ferritin thresholds has many implications. Inappropriately low ferritin reference ranges that do not accurately reflect physiologic iron stores in women inadvertently contribute to suboptimal care through the systematic underdiagnosis and subsequent undertreatment of ID [14]. This underrecognition not only leads to underestimation of the true healthcare burden of ID but also acts as a barrier for proper insurance coverage of parenteral iron, which is essential for women who do not tolerate or adequately respond to oral formulations [14, 57].

Our model has several strengths. First, we apply the best available epidemiological data from NHANES for our model input parameters, a nationally representative sample of n = 3490 patients [8]. Our model comprehensively compares the current practice of no screening with two screening strategies, one using an increased physiologically grounded ferritin threshold and the other using the WHO diagnostic threshold [14]. We also conducted an analysis of treatment with IV iron dextran (as opposed to oral ferrous sulfate), accounting for both the cost of the product and nursing‐driven infusion time, with our findings remaining robust throughout. Third, we performed key scenario, sensitivity, and threshold analyses to further quantify areas of uncertainty in our model towards better informing clinical decision making. Together, these nonnormatively denote thresholds beyond the bounds of clinical reality for which screening would become cost ineffective. For example, the maximum threshold cost of ID workup, although most women of reproductive age with ID do not require such testing, far exceeds even the most expensive invasive tests including the work‐up for celiac disease (i.e., upper and small bowel endoscopy with biopsy in addition to serologic tests) and colonoscopy [59].

We also recognize several limitations in our model. First, we do not account for states of increased iron demand throughout women's reproductive lifespan, namely pregnancy, where more than 80% of women have ferritin levels below 30 μg/L in the third trimester [60]. We recognize that such states will bring about a concomitant increase in the risk of ID, particularly if needs are not met through either diet or additional supplementation [12, 21]. This represents an important avenue for future research, particularly given evidence of the detrimental fetal and maternal effects of ID in pregnancy despite the USPSTF finding insufficient justification for screening of asymptomatic pregnant persons [21]. Second, our model does not explore all possible formulations of iron supplementation for the treatment of ID, nor the clinical reality of heterogeneous treatment across both routes of administration with formulations differing in cost and administration. However, we chose commonly used oral and IV products in our dual base‐case analyses. Third, since no on‐treatment specific utility data have been reported for women with ID undergoing iron supplementation using generic‐preference based measures of health‐related quality of life suitable for health economic analyses, we estimated the utility of the ID and perfect health states from a prospective cohort of Finnish women with ID without anemia, standardized for time period and geography using validated EQ‐5D norms for US women [43, 46]. This was further addressed in extensive sensitivity and threshold analyses. Fourth, despite data that higher ferritin thresholds (i.e., up to 50 μg/L) may better reflect the physiologic threshold for ID, we opted to evaluate ferritin thresholds for which (1) US‐specific epidemiologic estimates have been published and (2) which did not require assumptions to substitute for absence of data on the possibility of potential iron over‐treatment in a subcohort of patients when the ferritin threshold specificity for ID diagnosis is less than ~100% (> 25–30 μg/L). Fifth, we did not account for variability among different commercially available serum ferritin assays while recognizing that this variability could serve to limit the difference in cost‐effectiveness of different close‐together ferritin thresholds, such as 25 versus 30 μg/L [15, 61, 62]. Sixth, blood transfusion as treatment for ID was not considered in this work; it is the subject of future work in jurisdictions beyond the US where red blood cell transfusion may appear to be a less expensive approach (without accounting for transfusion complications) to treating ID than iron infusion. Finally, although the findings from our analysis of ID screening may not be directly generalizable outside the United States, they hold important implications for other countries. The identified ICERs are well below WTP thresholds in most countries around the world; a thorough evaluation of the burden of ID globally, particularly in LMICs, where ID contributes to significant morbidity and mortality, would enable the adaptation of our model beyond the high‐income country context.

In conclusion, to the best of our knowledge, we conducted the first cost‐effectiveness analysis of ferritin‐based screening for ID in reproductive‐age women in the United States. We found that, regardless of WTP or route of iron supplementation, annual screening for ID with a 25 μg/L ferritin threshold is a cost‐effective intervention that improves quality‐adjusted life expectancy compared to both screening with the WHO diagnostic standard of 15 μg/L and no screening. These results are driven by the additional proportion of women accurately identified and treated for ID (as opposed to the status quo of underdiagnosis and undertreatment) when routine screening is implemented and ferritin‐based screening thresholds are established. The relevance and importance of implementing ID screening can be better understood by comparing its ICER with those of currently recommended screening programs in the United States. The estimated ICERs for breast cancer, cervical cancer, and colorectal cancer screening are $12 710, $10 260, and $13 700 per QALY, respectively [63]. The ICER for ID screening is also lower than that of hypertension screening ($33 800/QALY) and cholesterol screening ($48 500/QALY) [64]. In the US context, our findings provide a strong economic rationale for the establishment of ID screening for reproductive‐age women, of whom ~800 000 adult women in every US birth cohort are expected to have ID [8, 65].

Author Contributions

D.W., S.I., A.C., and G.G. conceived the study design. All authors wrote and edited the manuscript.

Ethics Statement

The authors have nothing to report.

Conflicts of Interest

A.C. has served as a consultant for MingSight, New York Blood Center, Pfizer, Sanofi, and Synergy and has received authorship royalties from UpToDate. The other authors declare no conflicts of interest.

Supporting information

Data S1.Supporting Information.

Data Availability Statement

Data used in this study are from publicly sourced research, outlined within the article, and are available on email request from the corresponding author.