Intravenous iron supplementation in heart failure and iron deficiency: an umbrella review of systematic reviews and meta-analyses

Introduction

Iron deficiency is found in roughly one-third to one-half of patients with heart failure. Even when anemia is not present, it remains clinically important and has been linked to prognosis (1–3). Beyond its classic role in hemoglobin synthesis, iron is a critical cofactor in mitochondrial respiration, oxidative pathways, and cellular energy generation. Depleted cellular iron in heart failure is therefore directly linked to compromised skeletal-muscle function, impaired myocardial bioenergetics, and a subsequent decline in functional capacity (4–7). These changes are reflected clinically in worse health status, heavier symptom burden, repeated hospitalization, and a greater risk of adverse outcomes (2, 8–10). For this reason, recent heart failure guidelines recommend assessment of iron status and intravenous iron replacement in selected patients with confirmed iron deficiency, but these recommendations are phenotype-specific rather than applicable to the entire heart failure spectrum (11–14). The 2022 AHA/ACC/HFSA guideline gives intravenous iron replacement a Class IIa recommendation in patients with heart failure with reduced ejection fraction (HFrEF) and iron deficiency, with or without anemia, to potentially improve functional status and quality of life. The 2023 ESC focused update recommends intravenous iron supplementation in symptomatic patients with HFrEF or heart failure with mildly reduced ejection fraction (HFmrEF) and iron deficiency to alleviate heart failure symptoms and improve quality of life (Class I, Level A), while ferric carboxymaltose or ferric derisomaltose should be considered in these patients to reduce the risk of heart failure hospitalization (Class IIa, Level A). Accordingly, interpretation of intravenous iron evidence requires explicit consideration of left ventricular ejection fraction (LVEF), heart failure phenotype, and clinical setting, rather than simple generalization to all patients with heart failure.

Over the past two decades, evidence for intravenous iron therapy in heart failure has gradually expanded. Early work, including FERRIC-HF and studies of iron sucrose, suggested that patients with heart failure and iron deficiency could have better exercise tolerance, fewer symptoms, and improvement in selected biomarkers after treatment (15–17). Ferric carboxymaltose then became the formulation studied most often. In FAIR-HF, CONFIRM-HF, and EFFECT-HF, improvements were reported in patient-reported outcomes and functional capacity, with an additional signal toward fewer heart failure hospitalizations (18–20).

While recent outcome-driven trials have partially resolved the therapeutic role of intravenous iron, they have also introduced nuance to its clinical interpretation. Specifically, trials such as AFFIRM-AHF, IRONMAN, HEART-FID, and FAIR-HF2 have contributed evidence on the potential for iron supplementation to reduce heart failure-related hospitalizations; however, they have not established a definitive survival benefit (21–24). Such inconsistencies are largely attributable to variations in study populations, specific iron formulations, endpoint criteria, and observation periods across trials. Notably, although ferric carboxymaltose remains the most heavily investigated agent, the accumulation of clinical data on ferric derisomaltose has broadened the therapeutic armamentarium while compounding the complexity of meta-analytic pooling (47). Concomitant with the expanding trial literature, numerous systematic reviews have emerged to synthesize the therapeutic effects of iron therapy in heart failure. Although these reviews span diverse clinical and patient-centered endpoints, their pooled estimates are often confounded by substantial heterogeneity in study selection criteria, statistical modeling, and active treatment definitions (25–29). Several reviews focused exclusively on ferric carboxymaltose, whereas others combined multiple intravenous formulations or compared intravenous and oral iron strategies (30–34).

Despite the proliferation of secondary literature, synthesizing this evidence base remains challenging, as published reviews are frequently confounded by overlapping primary trials and discordant definitions of outcomes, treatment categories, and subgroup cohorts. Consequently, while a superficial consensus across meta-analyses may appear reassuring, it often reflects the redundant compounding of identical randomized trials rather than a truly robust, independent body of evidence. To address this evidentiary redundancy and clarify the true therapeutic signal, we conducted an umbrella review of systematic reviews and meta-analyses investigating intravenous iron therapy in patients with heart failure and concomitant iron deficiency. Beyond summarizing clinical efficacy across key domains—specifically hospitalization, mortality, functional capacity, quality of life, cardiac performance, and safety—we rigorously appraised review-level methodological quality via AMSTAR2, quantified trial redundancy using CCA, and graded the overall certainty of evidence using the GRADE framework.

Methods

Design and reporting framework

This umbrella review synthesized evidence from systematic reviews and meta-analyses that evaluated iron supplementation in adults with heart failure and iron deficiency. Study identification, screening, eligibility assessment, and reporting followed the PRISMA 2020 statement (35). The unit of inclusion was the systematic review or meta-analysis, rather than the individual trial. Primary randomized trials were extracted to assess overlap between reviews and to understand repeated contributions of the same evidence, but no patient-level reanalysis was performed.

Eligible populations comprised adults with heart failure and iron deficiency. Heart failure was accepted according to the definitions used in the original systematic reviews, including clinically diagnosed heart failure based on established guideline criteria, documented LVEF phenotypes, including reduced, mildly reduced, or preserved ejection fraction when reported, or other validated clinical classifications. Iron deficiency was accepted according to the criteria reported in the included reviews, most commonly defined as serum ferritin <100 μg/L or ferritin 100–299 μg/L with transferrin saturation <20%. Eligible interventions primarily included intravenous iron formulations, including ferric carboxymaltose, ferric derisomaltose, iron sucrose, and other intravenous iron-carbohydrate complexes. Reviews evaluating mixed iron strategies were considered eligible when intravenous iron represented the primary intervention of interest or when intravenous iron-specific outcomes were reported separately. Oral iron studies were not considered as primary evidence for intravenous iron efficacy but were retained when included within broader iron supplementation reviews to provide contextual information regarding differences between administration routes. Comparators included placebo, usual care, or standard care. Outcomes of interest were heart failure hospitalization, total or recurrent heart failure hospitalization, first heart failure hospitalization, composite endpoints combining heart failure hospitalization with cardiovascular death, cardiovascular mortality, all-cause mortality, 6MWD, KCCQ, NYHA class, LVEF, peak oxygen uptake, serious adverse events, and any adverse events. Studies evaluating oral iron alone without intravenous iron comparison were not considered eligible for the primary effectiveness assessment.

Outcomes were categorized into clinical, functional, patient-reported, cardiac-function, and safety domains. Clinical outcomes included heart failure hospitalization, recurrent or total hospitalization, composite hospitalization–cardiovascular death endpoints, cardiovascular mortality, and all-cause mortality. Functional and patient-reported outcomes included 6MWD, New York Heart Association functional class, peak oxygen uptake, and Kansas City Cardiomyopathy Questionnaire scores. Cardiac-function outcomes included LVEF, whereas safety outcomes included serious adverse events and any adverse events.

We did not restrict this review to mortality outcomes. In heart failure, iron deficiency and iron therapy may be relevant to several areas of care, including hospitalization, symptom burden, exercise capacity, health status, and safety (12, 13, 36). Focusing on a single endpoint would not fully reflect the range of outcomes assessed in recent randomized trials.

Another issue was overlap among studies. Many published reviews were based on a relatively limited set of landmark trials. We therefore included an overlap assessment in the review design, so that agreement across reviews could be interpreted in relation to the amount of shared trial evidence (37, 38).

Protocol registration

This umbrella review was conducted according to the PRISMA 2020 statement. No protocol was prospectively registered in PROSPERO or another publicly accessible registry. The methodology was developed according to established umbrella review principles, including predefined eligibility criteria, methodological appraisal, overlap assessment, and outcome-level certainty evaluation (48).

Search strategy and eligibility criteria

We searched PubMed, Embase, Web of Science, the Cochrane Library, CNKI, and Wanfang. The search covered publications from January 2016 to 31 March 2026 to capture contemporary systematic reviews and meta-analyses developed after the expansion of large randomized trials of intravenous iron therapy in heart failure. The final search was conducted on 31 March 2026.

Search strategies combined controlled vocabulary and free-text terms related to heart failure, iron deficiency, intravenous iron, ferric carboxymaltose, ferric derisomaltose, iron sucrose, iron supplementation, systematic review, and meta-analysis. Search syntax was adapted according to the indexing requirements of each database. The complete search strategies for all databases are provided in Supplementary Table 4.

No language restrictions were applied. Duplicate records were removed using reference-management software before title and abstract screening. Detailed search strategies, database-specific terms, and screening procedures are presented in the Supplementary Material.

Eligible reviews were systematic reviews with quantitative synthesis, including conventional meta-analyses, network meta-analyses, and subgroup meta-analyses. They had to evaluate iron supplementation in adults with heart failure and iron deficiency. To be included, a review also had to report at least one outcome relevant to this umbrella review, such as clinical outcomes, functional outcomes, patient-reported outcomes, cardiac-function measures, biomarkers, or safety outcomes.

Reviews were excluded if they focused on populations without heart failure, evaluated interventions unrelated to iron supplementation, examined anemia in the absence of both heart failure and iron deficiency, or discussed only mechanistic or biomarker associations without treatment outcomes. Non-systematic evidence sources were also excluded, including protocols, narrative reviews, scoping reviews, guidelines, editorials, letters, commentaries, conference materials, case reports, animal studies, and expert opinions. In addition, we excluded individual patient data meta-analyses, pooled analyses that were not conducted as systematic reviews, and trial-level pooled analyses, because the unit of inclusion was the systematic review or meta-analysis.

When more than one version of a review was available, we retained the most complete report. This decision was based on the completeness of methodological reporting, availability of outcome data, description of the included trials, and clarity of population and intervention details. Conference abstracts were not eligible because they did not provide enough information for methodological quality assessment, overlap mapping, or certainty-of-evidence assessment.

Study selection

Records were de-duplicated before screening. Titles and abstracts were screened against predefined eligibility criteria, and potentially relevant reports underwent full-text assessment. When full-text PDFs were unavailable, eligibility was evaluated using alternative sources, including abstracts, PubMed Central records, bibliographic databases, publisher webpages, and other accessible sources. Reports were not excluded solely because a downloadable PDF could not be obtained. Two authors independently participated in the study selection and data extraction process. QL conducted the primary screening, extraction, methodological assessment, and evidence synthesis, while YZ independently cross-checked eligibility decisions, extracted information, reviewed methodological assessments, and critically evaluated the certainty assessment and interpretation of findings. Any disagreements were resolved through discussion and consensus.

Prioritization of representative reviews

Although 45 systematic reviews and meta-analyses fulfilled the predefined eligibility criteria, many reviews included the same randomized controlled trials and addressed highly overlapping clinical questions. Directly synthesizing all eligible reviews in equal detail would therefore risk duplicating evidence and artificially inflating the apparent evidence base. To reduce redundant evidence contribution, 18 representative reviews were prioritized for detailed methodological appraisal, overlap assessment, GRADE evaluation, and evidence synthesis. The remaining 27 eligible reviews were retained to describe the broader review-level evidence context and to support concordance assessment.

Prioritization was based on predefined hierarchical criteria. First, reviews most closely matching the predefined population, intervention, comparator, and outcome framework were prioritized. Second, priority was given to reviews providing clear, extractable, outcome-specific quantitative estimates. Third, when overlap existed, reviews incorporating contemporary randomized controlled trial evidence, including newer landmark trials, were prioritized. Fourth, among otherwise comparable reviews, preference was given to reviews providing broader and less duplicated evidence coverage.

When multiple reviews fulfilled similar criteria, the review providing the most comprehensive and interpretable evidence synthesis was selected. Methodological reporting completeness was considered only to ensure transparency and reproducibility of assessment, rather than as a measure of review quality or treatment effect. AMSTAR2 methodological confidence was not used as a selection criterion, and reviews were not prioritized according to favorable or unfavorable findings. The prioritization process and rationale for individual non-prioritized reviews are provided in Supplementary Table 5. Reviews were classified according to their contribution to the umbrella review as quantitative synthesis, outcome-specific synthesis, or contextual evidence.

Data extraction

For each included review, data were extracted on publication characteristics, review methodology, study populations, interventions, comparators, outcomes, and principal findings. Extracted variables included first author, publication year, journal, review type, number of included studies and randomized controlled trials, total participants, intervention and comparator characteristics, follow-up duration when available, outcome measures, effect estimates, heterogeneity statistics, risk-of-bias assessment methods, publication-bias assessment, funding, and conflicts of interest. Heart failure phenotype, LVEF eligibility criteria, and the acute vs. chronic or stable clinical setting were additionally summarized from the included reviews and underlying landmark randomized trials where clearly reported, to inform interpretation of evidence generalizability. Data extraction was based primarily on the main text, tables, Supplementary Materials, and forest plots. Abstract information was used only when full-text information was unavailable or incomplete. Extracted data were cross-checked against the source reviews by the second author, and discrepancies were resolved through discussion and consensus.

To improve consistency across reviews, participant counts were recorded as total participants, while subgroup-specific sample sizes were documented separately when relevant. Review types were categorized as meta-analysis, systematic review and meta-analysis, network meta-analysis, or subgroup meta-analysis. Interventions were classified as intravenous iron, ferric carboxymaltose, ferric derisomaltose, oral iron, or mixed iron formulations.

For outcomes reported in multiple reviews, preference was given to estimates that most closely matched the predefined outcome definitions. Recurrent-event outcomes were extracted separately from first-event outcomes whenever this distinction was reported. When multiple estimates for the same outcome were available within a review, the primary result identified by the review authors was selected, and supplementary analyses were recorded separately. This approach aimed to minimize misclassification of distinct outcomes such as first hospitalization, recurrent hospitalization, total hospitalization, and composite hospitalization–mortality endpoints. Representative estimates were selected according to predefined criteria, including alignment with prespecified outcome definitions, clinical interpretability, inclusion of contemporary randomized evidence, and avoidance of duplicate representation of overlapping evidence.

Methodological quality assessment

The methodological quality of included reviews was evaluated using AMSTAR2 (39). Each review was assessed across the 16 AMSTAR2 items, with emphasis on the critical domains identified by the instrument. Overall confidence was classified according to AMSTAR2 recommendations as high, moderate, low, or critically low. AMSTAR2 ratings were used to characterize the quality of the review evidence and were not converted into numerical scores.

Overlap assessment

To assess overlap among reviews, all primary randomized trials included in the representative reviews were identified and mapped into a citation matrix. Trial names were harmonized when different reviews used alternative naming conventions for the same study. The resulting matrix was used to determine the extent to which reviews relied on shared primary evidence.

Overlap was quantified using CCA, calculated as (Nr−Ns)/[(Nc × Ns)−Ns], where Nc denotes the number of reviews, Ns the number of unique primary studies, and Nr the total number of study occurrences within the matrix (37, 38). CCA values were interpreted using established thresholds for slight, moderate, high, and very high overlap.

CCA was calculated among the 18 representative reviews included in the quantitative synthesis. Therefore, this estimate reflects redundancy within the prioritized evidence set rather than overlap across all 45 eligible reviews. The overlap analysis was not used as a criterion to select representative reviews; instead, it was performed after prioritization to characterize the degree of redundancy among the reviews contributing to the primary synthesis.

Certainty of evidence

GRADE was used to assess certainty for each outcome (40). Evidence derived from randomized controlled trials was initially considered high certainty and was downgraded according to risk of bias, inconsistency, indirectness, imprecision, and publication bias.

In this umbrella review, AMSTAR2 methodological confidence and GRADE certainty were considered complementary but distinct assessments. AMSTAR2 evaluates the methodological rigor of systematic reviews, whereas GRADE evaluates the certainty of evidence supporting individual clinical outcomes. Therefore, low AMSTAR2 confidence did not automatically result in downgrading of all outcomes to low or very low certainty. Instead, AMSTAR2 limitations informed interpretation of the overall evidence context.

For outcomes with consistent effects across reviews, support from larger randomized trials, acceptable precision, and clinically coherent findings, certainty was maintained when the GRADE domains did not justify further downgrading. Conversely, certainty was downgraded when estimates were affected by substantial heterogeneity, indirect comparisons, imprecision, limited trial numbers, or concerns regarding publication bias.

Final certainty categories were classified as high, moderate, low, or very low. Certainty judgments were performed for heart failure hospitalization, total heart failure hospitalization, first heart failure hospitalization, composite heart failure hospitalization plus cardiovascular death, cardiovascular mortality, all-cause mortality, 6MWD, KCCQ, LVEF, and serious adverse events.

Evidence map

We also constructed an evidence map to present the findings more clearly across outcomes. The map summarized two aspects at the same time: the direction of effect and the certainty of evidence. Outcomes were grouped as beneficial, no clear effect, uncertain, or potentially harmful. Certainty was classified using GRADE. We also included information on the amount of available evidence, because a favorable direction of effect is easier to interpret when it is supported by a less sparse evidence base.

When participant numbers were not reported for a specific outcome, we used the closest available review-level participant count as a proxy and recorded this decision.

Results

Search results and selection of reviews

The database search identified 515 records: 72 from PubMed, 190 from Embase, 117 from Web of Science, 5 from the Cochrane Library, 83 from CNKI, and 48 from Wanfang. After removal of 109 duplicate records, 406 records underwent title and abstract screening. Of these, 316 were excluded, and 90 reports were assessed for eligibility. Forty-five systematic reviews and meta-analyses fulfilled the predefined eligibility criteria.

Because multiple eligible reviews addressed similar clinical questions with substantial overlap in randomized trial populations, a predefined prioritization strategy was applied to identify representative reviews for detailed methodological appraisal, overlap assessment, GRADE evaluation, and evidence synthesis. Eighteen representative reviews were selected based on methodological reporting transparency, availability of extractable outcome data, relevance to the PICO framework, recency of evidence, and minimization of redundant synthesis. The remaining eligible reviews were retained as part of the overall evidence base but were not included in the detailed synthesis because they provided limited additional independent information beyond the prioritized reviews. The selection process is summarized in the PRISMA 2020 flow diagram (Figure 1).

Characteristics of included reviews

The 18 representative reviews were published between 2016 and 2026. Most were systematic reviews with meta-analysis; one network meta-analysis and one subgroup meta-analysis were included. The number of included studies ranged from 4 to 14, and most reviews primarily synthesized randomized controlled trials. Table 1 summarizes the characteristics of included reviews, including population, intervention, comparator, outcomes, and AMSTAR2 confidence rating.

Table 1. Characteristics of representative systematic reviews included in the umbrella review.

Total heart failure hospitalization, composite heart failure hospitalization plus cardiovascular death, all-cause mortality, and serious adverse events were rated as low certainty. First heart failure hospitalization, cardiovascular mortality, 6MWD, KCCQ, and LVEF were rated as very low certainty because of greater inconsistency, imprecision, indirectness, or limited available evidence. No outcome achieved high certainty.

Most representative reviews evaluated intravenous iron therapy, particularly ferric carboxymaltose. Some reviews incorporated mixed iron formulations or included oral iron comparisons; these were interpreted primarily as contextual evidence because the primary clinical question of this umbrella review concerned intravenous iron therapy. Reviews evaluating mixed strategies were retained when intravenous iron represented the predominant intervention or when intravenous iron-specific outcomes were reported separately. Later reviews incorporated contemporary trials including AFFIRM-AHF, IRONMAN, HEART-FID, and FAIR-HF2, resulting in greater emphasis on hospitalization outcomes, composite clinical endpoints, and mortality.

AMSTAR2 methodological confidence

The methodological confidence of included reviews was generally limited. Seven reviews were rated as low confidence and eleven as critically low confidence according to AMSTAR2. No review achieved moderate or high confidence.

The most frequent methodological limitations included absence of prospectively registered protocols, incomplete reporting of excluded full-text studies with reasons, inadequate duplicate processes for study selection or data extraction, limited incorporation of risk-of-bias assessments into interpretation, and incomplete assessment of publication bias.

Overlap among primary randomized trials

The 18 representative reviews included 23 unique randomized controlled trials. These primary trials appeared 163 times across the citation matrix, resulting in a CCA of 35.81%, indicating substantial overlap among representative reviews included in the quantitative synthesis.

Frequently reused trials mainly included intravenous iron studies, such as FAIR-HF, CONFIRM-HF, EFFECT-HF, AFFIRM-AHF, IRONMAN, HEART-FID, FAIR-HF2, FERRIC-HF, Myocardial-IRON (52), and IRON-CRT (50), and earlier iron sucrose studies. IRONOUT-HF was identified in some mixed iron supplementation reviews and was retained only for overlap characterization; it was not considered as evidence for intravenous iron efficacy interpretation. Trial names and aliases were harmonized before calculation. The final overlap parameters were Nc = 18, Ns = 23, and Nr = 163. The citation overlap matrix is shown in Figure 2.

Heart failure phenotypes represented in the randomized evidence base

The randomized evidence underlying the included reviews was derived predominantly from patients with reduced LVEF, with several contemporary trials extending eligibility into the mildly reduced ejection fraction range. Stable chronic heart failure trials such as FAIR-HF, CONFIRM-HF, EFFECT-HF, FERRIC-HF, IRONMAN, HEART-FID, FAIR-HF2, Myocardial-IRON, and IRON-CRT predominantly enrolled patients with reduced LVEF, although several used eligibility thresholds that extended into the contemporary mildly reduced EF range. AFFIRM-AHF evaluated patients stabilized after an acute heart failure hospitalization and used a broader LVEF eligibility threshold below 50%, thereby representing a post-acute population extending beyond strictly reduced ejection fraction. Smaller earlier iron sucrose studies also contributed to the evidence base but were less consistently reported across reviews. Evidence specifically derived from heart failure with preserved ejection fraction (HFpEF) populations was limited. Accordingly, the review-level evidence should not be interpreted as representing the full spectrum of heart failure phenotypes.

Because the remaining 27 eligible reviews were not included in the quantitative synthesis due to limited additional independent information, CCA was interpreted as a measure of overlap within the synthesized evidence base rather than the entire eligible review landscape.

Sensitivity analysis of evidence direction across all eligible reviews

A qualitative concordance assessment was performed by comparing conclusions from the 18 prioritized reviews with those reported in the remaining 27 eligible reviews. The overall direction of evidence was consistent: intravenous iron was generally associated with reduced heart failure hospitalization, while effects on mortality and functional outcomes remained uncertain. No additional eligible review identified findings that would materially change the direction or interpretation of the main conclusions.

Filled cells indicate inclusion of the corresponding randomized controlled trial in each systematic review. The corrected covered area (CCA) was used to quantify overlap among reviews. The matrix includes only the 18 representative reviews selected for detailed synthesis.

Heart failure hospitalization

Review-level findings for heart failure hospitalization were generally consistent with a favorable association of intravenous iron therapy, though the magnitude of the clinical benefit varied depending on the specific endpoint definition and statistical approach utilized by the authors. To maintain clinical and methodological rigor, we analyzed these findings according to three distinct estimands: first-event hospitalization (reflecting time-to-first clinical worsening), total or recurrent hospitalization (reflecting overall disease burden), and composite clinical endpoints combining hospitalization with cardiovascular mortality. For first-event heart failure hospitalization, the clinical signal was generally favorable but showed some statistical variability across the reviews. Graham et al. reported a reduction in first heart failure hospitalization with a risk ratio of 0.83 (95% CI, 0.71–0.97). Martens et al. reported a favorable effect for total heart failure hospitalization, with a rate ratio of 0.81 (95% CI, 0.74–0.89). Other reviews, including Vukadinović et al., Rangwala et al., Shalabi et al., and Ahmed et al., also reported directionally favorable effects for hospitalization-related outcomes.

Total or recurrent heart failure hospitalization outcomes showed favorable estimates, although interpretation was limited by heterogeneity and low certainty evidence. Martens et al. reported a rate ratio of 0.739 (95% CI, 0.661–0.827), while Padda et al. reported a pooled estimate of 0.59 (95% CI, 0.40–0.88). However, heterogeneity was substantial in some analyses, with I² values approaching 75%–78%. Evidence concerning first heart failure hospitalization was less consistent. While Ahmed et al. (29) demonstrated a significant reduction with an odds ratio of 0.855 (95% CI, 0.744–0.983), alternative analyses—such as the review by Tanashat et al. (41)—yielded less certain estimates with confidence intervals crossing the null. Conversely, composite endpoints combining heart failure hospitalization with cardiovascular death consistently favored intravenous iron. Reviews by Taha et al. (32), Mhanna et al. (42), and Mohamed et al. (43) reported pooled effect estimates ranging from 0.79 to 0.89, though differences in outcome definitions and follow-up durations precluded direct statistical comparison. Within the consolidated evidence map, overall heart failure hospitalization was categorized as beneficial, whereas first-event hospitalization remained uncertain because of limited and inconsistent evidence. According to the GRADE assessment, the certainty of evidence was moderate for overall heart failure hospitalization, low for both recurrent/total hospitalization and composite endpoints, and very low for first heart failure hospitalization.

Mortality outcomes

Cardiovascular mortality

Evidence for cardiovascular mortality remained inconclusive. Graham et al., Martens et al., Shalabi et al., and Ahmed et al. reported estimates generally in the range of risk ratios or odds ratios of approximately 0.87–0.90. However, several analyses had confidence intervals crossing the null, and statistical significance was inconsistent across reviews.

Cardiovascular mortality was therefore classified as uncertain in the evidence map. GRADE certainty was rated as very low because of inconsistency, imprecision, and methodological limitations of the available review-level evidence.

All-cause mortality

Evidence for all-cause mortality did not demonstrate a reliable treatment effect. Qian et al., Vukadinović et al., Wang et al., and Mhanna et al. reported estimates generally close to the null, with risk ratios or odds ratios approximately between 0.90 and 0.95. Earlier reviews were limited by smaller sample sizes and wider confidence intervals.

The direction of effect was classified as no clear effect in the evidence map. GRADE certainty was rated as low because of imprecision and limitations of the contributing reviews.

Functional capacity, health status, and cardiac function

Six-minute walk test

Effects on exercise capacity were heterogeneous. Taha et al. and Pintaningrum and Yunandari reported improvements in 6MWD of approximately 29–31 m, whereas Padda et al. reported a smaller mean difference of 14.03 m with confidence intervals crossing the null.

The direction of effect for 6MWD remained uncertain despite favorable estimates reported in some reviews, because of substantial heterogeneity and very low certainty evidence.

KCCQ

KCCQ outcomes were inconsistent across reviews. Taha et al. and Padda et al. reported variable effects, with some analyses suggesting improvement and others showing no statistically significant difference. High heterogeneity and differences in patient populations, follow-up duration, and treatment protocols limited interpretation.

KCCQ effects remained uncertain, with some analyses suggesting improvement whereas others showed no statistically significant difference. Interpretation was limited by inconsistent estimates, heterogeneity, and very low certainty evidence.

LVEF

Evidence for LVEF improvement was also uncertain. Mhanna et al. and Padda et al. reported mean differences of approximately 2.7% and 4.54%, respectively; however, confidence intervals were imprecise and clinical significance remained uncertain.

LVEF was classified as uncertain with very low GRADE certainty.

Safety outcomes

No clear increase in serious adverse events was identified across available reviews. Rangwala et al. reported a risk ratio of 0.98 (95% CI, 0.93–1.03), while Padda et al. reported an estimate of 0.73 (95% CI, 0.49–1.10).

However, definitions and reporting methods for serious adverse events varied across reviews, and heterogeneity was moderate to high in some analyses. Serious adverse events were classified as showing no clear effect, with low GRADE certainty.

Any adverse events were reported less consistently. Several reviews summarized adverse events narratively or used trial-specific definitions, limiting direct comparison. Infusion reactions, infection-related outcomes, hypotension, and gastrointestinal events were not uniformly reported across iron formulations. Ferric carboxymaltose-associated hypophosphatemia and hypersensitivity reactions were also not consistently captured across reviews; therefore, uncommon or formulation-specific harms could not be fully excluded.

GRADE summary and evidence map

GRADE certainty varied across outcomes. Heart failure hospitalization was rated as moderate certainty because the direction of effect was consistent across multiple reviews, supported by contemporary randomized trial evidence, and generally associated with clinically meaningful estimates with acceptable precision. However, certainty was not rated as high because methodological limitations of the contributing reviews, substantial overlap among primary trials, and variability in hospitalization definitions remained important sources of uncertainty. This moderate certainty reflects consistency and coherence at the outcome level rather than high methodological confidence of the contributing systematic reviews.

The evidence map summarizes the direction of effect and certainty of evidence across major outcomes. Panel A summarizes the direction of effect, certainty of evidence, and number of contributing reviews. Panel B presents outcome-level classification according to effect direction and GRADE certainty. Bubble size and numbers adjacent to bubbles represent the number of representative reviews/meta-analyses contributing to each outcome. Interpretation should consider that evidence certainty reflects outcome-level GRADE assessment rather than methodological confidence of individual reviews. The outcome-level evidence map is shown in Figure 3.

Source review(s) indicates the representative systematic reviews/meta-analyses from which the corresponding effect estimates and evidence interpretations were derived. When multiple reviews reported similar outcomes, representative estimates were selected according to predefined criteria, including outcome definition alignment, clinical interpretability, contemporary evidence inclusion, and avoidance of duplicate representation of overlapping evidence. The outcome-specific evidence and corresponding GRADE certainty are summarized in Table 2.

Table 2. Summary of evidence by outcome.

GRADE, grading of recommendations assessment, development and evaluation; HF, heart failure; KCCQ, Kansas City Cardiomyopathy Questionnaire; LVEF, left ventricular ejection fraction; 6MWD, 6 min walk distance; IV, intravenous; FCM, ferric carboxymaltose.

Discussion

Synthesizing these data reveals a relatively localized primary finding: review-level evidence most consistently supports a potential reduction in heart failure hospitalizations with intravenous iron therapy, whereas its therapeutic efficacy regarding cardiovascular or all-cause mortality, functional capacity, health status, and LVEF remains far less conclusive. This disparity underscores that the clinical benefits of intravenous iron are not uniformly established across all clinical and surrogate endpoints. The review-level evidence requires cautious interpretation. All 18 representative reviews were rated as low or critically low confidence by AMSTAR2 (39), and many conclusions were derived from primary trials with substantial overlap. Only 23 unique randomized trials contributed to the core evidence base, while these trials appeared 163 times across reviews, resulting in a CCA of 35.81%. Therefore, the apparent abundance of published reviews likely overestimates the amount of independent evidence. The prioritization of 18 representative reviews from the 45 eligible reviews was performed to minimize redundant synthesis of highly overlapping evidence while preserving the broader eligible evidence base.

Crucially, the methodological quality of the synthesized reviews and the certainty of specific clinical outcomes represent distinct dimensions of evidence credibility. Although all prioritized reviews were characterized by low or critically low confidence under AMSTAR 2, the heart failure hospitalization outcome was consistently supported across reviews and backed by large-scale randomized trials. Consequently, the moderate GRADE certainty assigned to this endpoint reflects the consistency and precision of the outcome-level data itself, rather than implying an absence of methodological flaws in the secondary literature. The rating was capped at moderate because review-level deficiencies, extensive trial overlap, and discordant hospitalization definitions persist as key sources of residual uncertainty. The hospitalization signal is clinically plausible. Iron is involved in mitochondrial respiration, oxidative metabolism, skeletal-muscle energetics, and myocardial energy production, in addition to its role in hemoglobin synthesis (4, 7, 44). In patients with heart failure, iron deficiency may impair peripheral oxygen utilization and reduce exercise tolerance even in the absence of overt anemia. Restoration of iron availability through intravenous administration may improve functional reserve and reduce susceptibility to clinical deterioration, providing a potential explanation for the more consistent hospitalization findings.

Crucially, mitigating hospitalization does not guarantee a concomitant reduction in mortality. Survival in heart failure patients is governed by a multitude of competing risks (11, 13)—such as age, renal insufficiency, ischemic heart disease, arrhythmic burden, comorbidities, and baseline guideline-directed medical therapy—which collectively dilute the observable impact of iron repletion on hard survival endpoints. This complex interplay may help explain why the evidence for hospitalization reduction has been more consistent than that for a mortality benefit. Moreover, the lack of a consistent signal for LVEF improvement leaves effects on resting left ventricular systolic recovery uncertain; peripheral and metabolic mechanisms may contribute to the observed hospitalization findings. These findings should also be interpreted in the context of the evolving clinical trial evidence. Earlier studies, including FAIR-HF, CONFIRM-HF, and EFFECT-HF, primarily focused on symptoms, functional capacity, and quality of life (18–20), whereas subsequent trials increasingly evaluated clinical events. AFFIRM-AHF demonstrated fewer heart failure hospitalization events after acute heart failure stabilization without a clear cardiovascular mortality benefit (21). IRONMAN expanded the evidence base for ferric derisomaltose and recurrent-event outcomes (22). HEART-FID and FAIR-HF2 provided larger contemporary datasets but did not establish a clear mortality advantage (23, 24). Collectively, the trial trajectory supports hospitalization reduction as the most reproducible signal, while mortality and remodeling outcomes remain unresolved.

The choice of iron formulation is important for clinical interpretation. Most available evidence is derived from intravenous ferric carboxymaltose, including FAIR-HF, CONFIRM-HF, EFFECT-HF, AFFIRM-AHF, HEART-FID, and FAIR-HF2. Evidence for ferric derisomaltose is more limited and is mainly derived from IRONMAN and subsequent reviews incorporating this trial. Earlier iron sucrose studies were generally smaller and contributed less contemporary outcome evidence. Although some eligible reviews included mixed iron strategies or oral iron comparisons, these findings should not be interpreted as equivalent evidence for intravenous iron efficacy. Oral iron differs from intravenous formulations in bioavailability and iron delivery kinetics, and IRONOUT-HF did not demonstrate improvement in exercise capacity with oral iron (45). Therefore, the conclusions of this umbrella review primarily reflect intravenous iron therapy rather than iron supplementation as a broad therapeutic category.

Because most available evidence was derived from ferric carboxymaltose and direct comparative evidence between intravenous iron formulations remains limited, findings were interpreted primarily as class-level effects rather than evidence of superiority of any specific formulation.

The most consistent review-level signal concerns heart failure hospitalization. Evidence supporting reductions in mortality, cardiac remodeling, and durable functional improvement remains less certain. This distinction is clinically relevant because a therapy may reasonably reduce hospitalization risk without demonstrating a comparable effect on survival or structural cardiac outcomes.

Our findings are broadly aligned with contemporary international guideline recommendations, while also helping to define the boundaries of the underlying evidence. The 2022 AHA/ACC/HFSA guideline assigns intravenous iron replacement a Class IIa recommendation in patients with HFrEF and iron deficiency, with or without anemia, to improve functional status and quality of life. For symptomatic patients with HFrEF or HFmrEF and iron deficiency, the 2023 ESC focused update recommends intravenous iron for symptom and quality-of-life improvement (Class I, Level A), and supports consideration of ferric carboxymaltose or ferric derisomaltose for reducing heart failure hospitalization risk (Class IIa, Level A) (12, 13). In the present umbrella review, the most consistent review-level signal concerned heart failure hospitalization, whereas mortality benefit remained uncertain and functional or health-status outcomes were heterogeneous and often supported by low or very low certainty evidence. This pattern is consistent with guideline emphasis on selected clinical benefits, but it should not be interpreted as class-wide support across all heart failure phenotypes.

Importantly, the randomized evidence contributing to both current guideline recommendations and the present umbrella review was concentrated predominantly in patients with reduced ejection fraction, with some contemporary trials extending eligibility into the mildly reduced ejection fraction range. Evidence specifically derived from HFpEF remains limited. Because randomized evidence was derived predominantly from patients with reduced or mildly reduced LVEF, the findings should not be directly generalized to patients with preserved ejection fraction. Treatment decisions should therefore remain individualized and should consider heart failure phenotype, ferritin and transferrin saturation levels, renal function, recent hospitalization history, background guideline-directed medical therapy, comorbidities, clinical setting, and availability of specific iron formulations. Based on the current review-level evidence, the most consistent evidence of a potential clinical benefit of intravenous iron concerns overall heart failure hospitalization, whereas a mortality-modifying effect has not been established.

This umbrella review has several strengths. We searched major international and Chinese databases and integrated multiple methodological approaches, including AMSTAR2, GRADE, CCA analysis, and evidence mapping. This allowed assessment not only of treatment effects but also of methodological confidence, certainty of evidence, direction of effect, and redundancy across reviews. The prioritization of representative reviews was designed to reduce repeated synthesis of overlapping evidence while maintaining transparency regarding the broader eligible literature (55).

Although CCA demonstrated substantial overlap among the prioritized reviews, this estimate should not be generalized to all eligible reviews because it was calculated within the evidence set included in the quantitative synthesis. The prioritization strategy may have influenced the quantitative synthesis; however, the qualitative concordance assessment across all eligible reviews suggested that inclusion of the broader review landscape would unlikely alter the principal conclusions.

Another limitation is that heart failure phenotype was not uniformly defined or reported across the contributing reviews and underlying trials. LVEF eligibility thresholds varied, and several older studies predated contemporary HFrEF, HFmrEF, and HFpEF classifications. Therefore, phenotype-specific comparative conclusions could not be established, and extrapolation to HFpEF should be made cautiously. Rather than viewing intravenous iron as a universally established class-wide therapy for all heart failure settings, clinical application should be framed around the specific clinical phenotypes, iron formulations, and care settings validated in large-scale trials. Current evidence suggests a potential reduction in overall heart failure hospitalization in selected patients with heart failure, predominantly those with reduced LVEF and with some trials extending eligibility into the mildly reduced EF range; interpretation is constrained by high trial overlap, review-level methodological deficiencies, and varying hospitalization definitions.

Future evidence synthesis should move beyond repeated pooling of the same trial base. More informative approaches would include formulation-specific analyses separating ferric carboxymaltose from ferric derisomaltose, setting-specific analyses distinguishing acute decompensated from stable chronic heart failure, and consistent reporting of first vs. recurrent hospitalization outcomes. Future trials and reviews should also incorporate post-treatment iron status, renal-function strata, sex-specific effects, heart failure phenotype, and background guideline-directed medical therapy to improve clinical interpretation and guideline implementation.

Conclusion

To conclude, cumulative review-level evidence, derived predominantly from patients with reduced LVEF and with some trials extending into the mildly reduced EF range, probably supports a reduction in heart failure hospitalization with intravenous iron in patients with iron deficiency. This pattern is broadly consistent with contemporary AHA/ACC/HFSA and ESC guideline recommendations. However, the available evidence should not be generalized across the entire heart failure spectrum, particularly to HFpEF, for which dedicated randomized evidence remains limited. Class-wide generalizations across intravenous iron formulations and clinical contexts are further constrained by extensive trial overlap, methodological limitations of the contributing reviews, and inconsistent outcome definitions. The effects on survival, long-term functional status, and cardiac remodeling remain uncertain.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcvm.2026.1902879/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: David Hesketh Roberts, University of Liverpool, United Kingdom

Reviewed by: Bambang Dwiputra, National Cardiovascular Center Harapan Kita, Indonesia

Xu Zhang, Henan Academy of Sciences, China