Blood donor populations reveal a clear association between ferritin and change in haemoglobin levels

Summary

Many blood establishments worldwide monitor serum ferritin alongside mandatory haemoglobin (Hb) screening to better protect donors from iron deficiency and anaemia. However, the relationship between ferritin and Hb, and the ferritin level that indicates iron deficiency, remains unclear. Whole blood donation results in significant iron loss, and repeated donations can deplete iron stores. This study analysed over 1 million whole‐blood donations from four countries to explore the association between Hb change and ferritin levels. Hb change was defined relative to a donor's initial Hb level. A consistent two‐phase relationship emerged: At low ferritin levels, Hb change is linearly associated with log ferritin; above a certain threshold, this association disappears as donors recover their reference Hb. The transition point and slope of this association differ by population. These results suggest that ferritin thresholds for identifying limited Hb recovery are not universal but population‐specific, influenced by biological and procedural differences, including ferritin assay variability. While the overall pattern is consistent, the absence of standardized procedures and assays limits the ability to define global ferritin thresholds for donor care. This underscores the importance of localized approaches to ferritin‐based donor management and the need for harmonized methodologies across blood services.

Many blood establishments around the world monitor both ferritin and haemoglobin (Hb) levels to better safeguard donors against iron deficiency and anaemia. However, the exact relationship between ferritin and Hb, and the ferritin threshold indicative of iron deficiency, remains uncertain. This study analysed ferritin and changes in Hb across more than 1 million whole blood donations from the Netherlands, South Africa, the United States and England. Hb change was determined relative to each donor's initial Hb level. The analysis revealed a consistent two‐phase pattern: At low ferritin levels, Hb change showed a linear relationship with log‐transformed ferritin; above a certain ferritin threshold, this association disappears as donors recover their initial Hb. Notably, both the slope and ‘changepoint’ of this relationship varied across populations.

INTRODUCTION

Maintaining the health of blood donors is a primary responsibility of blood establishments. Because of the loss of haemoglobin (Hb)‐bound iron during blood donation, regular whole blood donors are vulnerable to developing iron deficiency (ID). ¹ Premenopausal women are particularly affected, experiencing an average loss of 60% of their total iron stores with each blood donation, in contrast to the 30% loss observed in men. ² However, donation‐induced iron loss in men also deserves attention. While higher allowed donation frequency is a key factor causing both young male and female donors to eventually reach similarly low ferritin levels in some settings, iron deficiency in young male donors can occur independent of frequent donations. ³ , ⁴ To mitigate the risk of anaemia and ID, blood establishments are required to monitor blood donors' Hb levels. While Hb monitoring ensures adequate blood iron content for oxygen transport, it does not provide information on iron stores. ¹ Thus, relying solely on Hb monitoring may not be sufficient to prevent depletion of donor iron reserves.

Circulating ferritin level is the most widely used marker of iron stores and low ferritin is a specific indicator of depleted iron stores. ⁵ However, the relationship between ferritin level and Hb recovery remains insufficiently explored. While ferritin testing is increasingly introduced in blood establishments, there is no consensus on the ferritin threshold that best reflects sufficient iron stores. ⁶ , ⁷ , ⁸ Current thresholds for low ferritin vary widely between expert recommendations, with suggestions ranging from 12 to 50 ng/mL, depending on the population and context. ⁹ , ¹⁰ , ¹¹ , ¹² Despite these differences, the association between ferritin and Hb has important implications for donor care and the development of evidence‐based management protocols.

Blood donors represent a unique population for studying this relationship because they experience repeated, controlled iron loss and undergo regular monitoring of Hb and ferritin levels. Additionally, since donors are generally healthy and inflammation—which can elevate ferritin—is rare, there is minimal risk of disease‐related confounding. ¹³ , ¹⁴ This study examines the association between ferritin and change in Hb levels in blood donors from multiple countries. By analysing data from diverse donor populations, we identify a consistent pattern in the relationship between ferritin and Hb. These findings provide valuable insights into how ferritin and Hb levels can be used conjointly to optimize donor management and improve the understanding of post‐donation recovery dynamics.

METHODS

Setting

We obtained data on Hb levels from regular blood donations from donor databases from blood establishments in four countries: The Netherlands (Sanquin), South Africa (South African National Blood Service, SANBS), England (NHS Blood and Transplant, NHSBT, data from the INTERVAL trial ¹⁵ ), and the United States of America (USA, Vitalant). Across these blood establishments, the minimum donation intervals range from 56 (males) to 122 (females) days (Table S1). Hb screening is mostly conducted through capillary measurements, with deferral thresholds of either 12.5 and 13.5 g/dL or 12 and 13 g/dL, for males and females, respectively, below which donors are deferred for periods ranging from 1 day to 1 year (Table S1). Ferritin screening is conducted in three of four blood establishments and iron supplements are offered in South Africa and advised in the USA (Table S1). In the INTERVAL trial, the study population was subject to a policy and measurement regime that differed from usual practice at the NHSBT. ¹⁵ In the 2‐year trial, males were randomized to a donation interval of 8, 10 or 12 weeks and females to 12, 14 or 16 weeks. To evaluate the longer term risks and benefits of varying inter‐donation intervals, 20 757 participants were followed for two additional years in an extension study. ¹⁶

Measurement methods

In the Netherlands, ferritin was measured on the Architect Ci8200 (Abbott Laboratories, Illinois). In the INTERVAL trial, ferritin was measured on the Cobas e801 (Roche Diagnostics, Basel). ¹⁵ These measurements are traceable to the first WHO Human Liver Ferritin International Standard (IS) (80/602). In the United States, ferritin was measured on the AU680 (Beckman‐Coulter, Inc., Brea, CA). In South Africa, ferritin was measured on the UniCel Dxl 800 Access Immunoassay (Beckman‐Coulter, Inc., Brea, CA). Both are traceable to the third WHO IS (94/572). In all settings, ferritin was measured in serum samples. In the Netherlands, South Africa and the United States, Hb was measured using a capillary skin‐prick measurement, using the HemoCue 201 in the Netherlands and the HemoCue 301 (Angelholm, Sweden) in South Africa and the United States. In the INTERVAL trial, Hb was measured in venous blood samples on the Sysmex XN‐2000 (Sysmex, Kobe). ¹⁵

Datasets and variables

We defined reference Hb values for each donor according to the criteria outlined in Table 1. In short, the reference Hb is the Hb level that is not affected by donation (yet). Subsequently, following the criteria in Table 1, we selected follow‐up visits with Hb and ferritin measurements for each donor. The data analysed consisted of follow‐up data where the change in Hb was calculated as the change at follow‐up relative to the reference Hb level. We allow multiple donations from one donor to be included.

TABLE 1.

Data selection and labelling.

Data selection	Reference Hb	Follow‐up Hb or ferritin
Description	Hb measured at visit that is assumed to be unaffected by blood donation. This Hb level is used to calculate changes in Hb at follow‐up donations	Hb and ferritin levels measured at a donation that is preceded by a (recent) donation
Definition	This can be: R1: Hb level measured in a donor who has not donated in a prior 2‐year period. This could be a donor's first‐ever visit or the first visit after an inactive phase OR R2: Average of Hb values measured at the first two ever blood bank visits, only if no blood donation was made at the first visit a R3: If multiple potential reference visits precede a follow‐up visit, the most proximal Hb level is used	This can be any Hb and ferritin level measured during a visit that is: F1: Preceded by a donation that provided a reference Hb level AND F2: At which both ferritin and Hb were measured AND F3: Where at least one successful erythrocyte‐containing donation occurred in the prior 2‐year period (this could include a donation acquired at the reference donation visit and/or donations at other visits occurring after the reference donation visit but before this visit)
		Follow‐up visits are valid for any donation product (e.g. whole blood or plasma donation) or donation outcome (whether the donor was allowed to donate during that visit or not)

^a For centres with an ‘intake’ visit without blood collection, averaging over the first two visits reduces measurement uncertainty.

In all analyses, we used the log10‐transformed ferritin level (ng/mL). Hb change (g/dL) was defined as follows: For each donor, a reference Hb level (g/dL) was assigned, being the Hb level at the first ever visit or the visit after reactivation because it is assumed this is the setpoint Hb of each donor. ¹⁷ In the Netherlands, the mean Hb level of the first two pre‐donation visits was taken as the reference Hb level, as both measurements are conducted prior to any donation (Table 1). At each follow‐up visit, the change relative to the reference Hb level was calculated by subtracting the reference Hb level from the Hb level at donation.

We additionally analysed data from the FinDonor study conducted at the Finnish Red Cross Blood Service. ¹⁴ As the inclusion and exclusion criteria could not be applied to these study data, further details and results are displayed in Supplementary Materials (Tables S4–S6, Figure S6).

Statistical analysis

We investigated the relationship between Hb change and ferritin levels at a population level. We performed separate analyses in every country using the same publicly available analysis code. We further stratified the analyses by sex. We did not account for the correlation between multiple follow‐up donations from one donor or the time between the reference and follow‐up visit, but we did conduct sensitivity analyses in the Netherlands, South Africa and the United States using only the first follow‐up donation and visits falling within the first quartile of the dataset's time between reference and follow‐up visits (Figures S1 and S2).

After observing a clear changepoint in the rolling mean of Hb change plotted as a function of log ferritin in each study, we fitted a two‐segmented line to the data. This approach allowed accounting for the presence of two distinct segments with different slopes in the association between log ferritin and Hb change. We determined the changepoint of the association, defined as the ferritin level at which the slope of the line transitions from one segment to the other, using maximum likelihood estimation (Supplementary Materials, p. 2). We performed the fitting procedure twice: once assuming the second segment of the line was flat (i.e. fixed the slope at 0) and once where both segments' slopes were estimated from data. We generated confidence intervals from 1000 bootstrap samples. Analyses were conducted using R version 4.1.1. The analysis code is available through GitHub via https://github.com/Sanquin/Ferritin‐threshold‐for‐iron‐deficiency‐assessment.

This secondary analysis of de‐identified data was granted expedited approval by the McGill University Institutional Review Board.

RESULTS

Using the eligibility criteria in Table 1, a total of 1 021 356 blood donations were included in the analysis across the four study populations. The distribution of Hb levels was comparable across populations, but ferritin levels showed more variation (Table 2). The USA dataset skewed younger because of a policy to test all donors of age 16–18 for ferritin. In the Netherlands, donors included in the study were more experienced blood donors compared to those in other populations, likely because they are included based on routine ferritin measurements at every fifth donation; those with more ferritin measurements are by default more experienced. By including blood establishments with different donation interval policies and even one study cohort, the time and number of donations between the reference and follow‐up visits differed widely (Table 2).

TABLE 2.

Dataset characteristics per country for male and female blood donors (N = donations).

	The Netherlands	South Africa	England	USA
Date range	Sep 2017 to Dec 2022	Feb 2019 to Dec 2022	Mar 2014 to Aug 2016	Mar 2017 to Oct 2022
Males	N = 196 595	N = 205 195	N = 20 975	N = 65 052
Age (years), median [IQR]	52.0 [36.0, 61.0]	40.9 [29.9, 52.6]	51.7 [40.4, 60.8]	18.0 [17.0, 18.0]
Total donations per donor, median [IQR]	26 [13, 40]	10 [6, 15]	12 [10–14]	3 [2, 5]
Donation frequency (donations/year), median [IQR]	3.0 [2.3, 3.7]	3.9 [2.7, 5.0]	3.0 [2.5, 3.5]	2.8 [1.8, 4.0]
Ferritin (ng/mL), median [IQR]	38.0 [24.0, 56.0]	28.5 [13.3, 58.7]	32.0 [17.0, 56.0]	41.0 [25.0, 63.0]
Reference Hb (g/dL), median [IQR]	15.15 [14.50, 15.79]	15.79 [14.66, 16.60]	14.82 [14.34, 15.63]	15.79 [14.82, 16.60]
Follow‐up Hb (g/dL), median [IQR]	14.99 [14.34, 15.79]	15.63 [14.66, 16.60]	14.34 [13.54, 15.15]	15.63 [14.66, 16.60]
Time between reference and follow‐up visits (days), median [IQR]	2548 [732, 3898]	763 [496, 1087]	721 [692, 1043]	319 [162, 542]
Donations between reference and follow‐up visits, median [IQR]	18 [6, 33]	6 [3, 10]	9 [7, 11]	1 [1, 3]
Females	N = 217 144	N = 196 082	N = 19 697	N = 100 616
Age (years), median [IQR]	43.0 [28.0, 56.0]	37.6 [27.2, 49.5]	48.4 [35.8, 58.3]	18.0 [17.0, 49.0]
Total donations per donor (N), median [IQR]	13 [7, 23]	7 [5, 12]	8 [6, 10]	4 [2, 8]
Donation frequency (donations/year), median [IQR]	2.0 [1.5, 2.4]	3.2 [2.1, 4.4]	2.0 [1.5, 2.5]	2.8 [1.8, 4.2]
Ferritin (ng/mL), median [IQR]	28.0 [18.0, 42.0]	17.7 [9.60, 32.8]	26.0 [14.0, 45.0]	22.0 [13.0, 36.0]
Reference Hb (g/dL), median [IQR]	13.62 [13.13, 14.18]	14.02 [13.37, 14.82]	13.37 [12.73, 14.02]	13.86 [13.21, 14.50]
Follow‐up Hb (g/dL), median [IQR]	13.70 [13.05, 14.34]	13.86 [13.05, 14.66]	13.21 [12.57, 13.86]	13.54 [13.05, 14.34]
Time between reference and follow‐up visits (days), median [IQR]	1416 [611, 3409]	698 [420, 1062]	714 [680, 975]	375 [189, 734]
Donations between reference and follow‐up visits, median [IQR]	7 [4, 18]	4 [2, 8]	6 [5, 8]	2 [1, 4]

Abbreviations: IQR, interquartile range; NA, not available.

We assessed the association between Hb change and log ferritin levels and observed that the association is similar across countries and sexes. However, the slopes of the first segment, the changepoint of the association and the slopes of the second segment differ between countries (Table 3). Figure 1A provides a visual example of this Hb‐ferritin relation along with the rolling mean (window of n = 1000). In Figure 1B, the fitted association and its changepoint in Dutch male donors are displayed. For the other study populations, these graphs can be found in Figures S1 and S2. Interestingly, the graphs show that, for most donor populations, the second segment of the association, that is, after the changepoint, flattens at a Hb change of 0 g/dL. However, for females in The Netherlands and South Africa and males in South Africa, the association flattens at a mean increase in Hb of approximately 0.25 g/dL (Table S2).

TABLE 3.

Changepoint statistics and 95% confidence interval retrieved from 1000 bootstrap samples.

	The Netherlands (Sanquin)	South Africa (SANBS)	England (INTERVAL)	United States (Vitalant)
Males	N = 196 595	N = 205 389	N = 20 975	N = 65 052
Slope first segment (g/dL)	1.66 (1.60–1.74)	2.01 (1.93–2.09)	2.08 (1.98–2.16)	2.74 (2.55–2.92)
Changepoint (ng/mL)	30.3 (28.6–31.5)	21.6 (20.7–22.7)	43.6 (31.2–46.4)	24.2 (23.2–25.5)
Slope second segment (g/dL)	0	−0.19 (−0.23 to −0.16)	0	0
Females	N = 217 144	N = 196 479	N = 19 697	N = 100 616
Slope first segment (g/dL)	0.97 (0.92–1.02)	1.74 (1.71–1.80)	1.60 (1.51–1.69)	1.22 (1.16–1.34)
Changepoint (ng/mL)	25.4 (23.7–27.3)	19.6 (18.4–20.2)	32.4 (30.2–34.6)	22.2 (20.2–24.4)
Slope second segment (g/dL)	0.06 (0.03–0.11)	0.03 (0.02–0.10)	0	0.18 (0.13–0.24)

FIGURE 1.

Random sample of 10 000 Dutch males with the rolling mean of log ferritin (A) and the two‐segmented association between Hb change and log ferritin and its changepoint estimated by maximizing likelihood (B). The changepoint is the optimal ferritin level at which slope of the line representing this association changes to create a two‐segmented line. The analysis was conducted using log₁₀‐transformed ferritin values, the x‐axis displays actual ferritin concentrations (ng/mL) on a logarithmic (log₁₀) scale to aid interpretability. Green: rolling mean line (window = 1000); red: fitted association; yellow: changepoint ferritin level.

The slope of the first segment varied between countries and sexes, from 0.97 to 2.74 g/dL (Table 3). The changepoint in the association ranges from 18.6 to 43.6 ng/mL in the four study populations and varied more between populations than between sexes within populations (Table 3). In all populations, the ferritin level of the changepoint for males was higher than for females, but confidence intervals overlap between males and females in England and the United States. Confidence intervals for the changepoint levels between countries (within sexes) mostly do not overlap, suggesting significant differences between countries (Table 3). In four out of eight sex and country combinations analysed, the second segment slope was non‐zero, but the magnitude of the second segment's slope was less than 10% of the magnitude of the first segment's slope (Table 3). In sensitivity analyses using only a donor's first follow‐up visit to rule out effects of repeated measurements (Table S2) or shorter times between reference visits and return to rule out effects of time between donations (Table S3), the changepoints did not differ significantly from the main dataset analysed and did not change by more than 12%. Slopes for the first segment differed significantly in most cases and changed by up to 40% (Tables S2 and S3).

DISCUSSION

This study analysed the relationship between change in Hb levels and ferritin levels in blood donors across four populations. The findings reveal a consistent association across countries, characterized by a two‐phase pattern where lower ferritin levels correspond to a more pronounced decrease in Hb (a sign of iron depletion‐limited erythropoiesis), followed by a plateau indicating sufficient iron availability for Hb recovery to the level at the start of donorship. Despite this consistency, the slope of the association and the changepoint varied across populations and sexes, reflecting the potential influence of biological, environmental and procedural factors.

The association we observed between Hb change and log ferritin levels tends to flatten after the changepoint at a Hb change level of 0 g/dL, indicating stable Hb levels in case of sufficient iron reserves. However, an exception arises in females from The Netherlands and South Africa, where the association flattens at a higher Hb change of approximately 0.25 g/dL. This may suggest an underestimation of reference Hb levels in postmenopausal female donors whose reference Hb levels were established during their premenopausal phase. As a result of menstrual blood loss in the premenopausal phase and hormonal changes in menopause, the premenopausal reference Hb may be lower than the reference Hb levels in the postmenopausal phase. ³ , ¹⁷ Indeed, in postmenopausal females with a shorter donation history, who are more likely to have a reference Hb measured after menopause, the association stabilizes at approximately 0 g/dL change in Hb (Figures S3–S5). This suggests a distinct Hb reference level for the premenopausal and postmenopausal phase, which is in line with recent literature on individual haematological setpoints and the potential effect of hormonal changes. ¹⁷ In settings other than the Netherlands and South Africa, this bias may be less likely to occur due to generally shorter donor careers or study‐based data with more recent reference Hb measurements.

The slope of the first segment of the association and the position of the changepoint varied notably between populations. For instance, the slopes in The Netherlands were less steep than those in the other countries. This disparity could reflect differences in genetic makeup and composition of the population, lifestyle characteristics such as dietary habits and iron supplementation and blood bank policies, but it also reflects differences in the assessment of ferritin. Although the changepoints observed in our study are generally in line with ferritin thresholds of approximately 25 ng/mL described previously, ¹⁸ comparison of our results with prior studies, and between populations within our study, is complicated by the lack of standardization in ferritin assays. To address this challenge, the WHO has developed four reference materials for ferritin assay calibration since the 1980s. ¹⁹ , ²⁰ , ²¹ , ²² However, continuity in value assignment between these preparations has been lacking, and more recent studies suggest that these materials may not be optimally commutable—behaving differently across assays when applied to native samples. ²⁰ , ²¹ , ²³ , ²⁴ , ²⁵ While the direction of differences observed between blood establishments in our study aligns with differences observed between assays in recent harmonization efforts, the lack of assay standardization continues to complicate comparability across blood establishments. ²³ , ²⁵ The lack of standardization in ferritin assays across countries may have contributed to variability in deferral thresholds as well, with thresholds as low as 5 ng/mL in South Africa compared to 15 ng/mL in other centres, where national policies have adapted to best match local assay performance and implications thereof. Assay harmonization would be a crucial step towards improving ferritin‐based donor assessments.

Despite the insights provided, several limitations to our study warrant consideration. We used capillary Hb measurements to calculate Hb change in all settings except England, where venous Hb measurements were taken. Hb measurement results in capillary samples are known to be somewhat lower and less precise compared to those in venous samples, potentially affecting the precision of our findings. However, this limitation should be partly mitigated by our focus on population‐level associations in these large cohorts. While we identified a population‐level association, further research is needed to explore individual‐level variation in the ferritin level needed for Hb recovery. This would help in tailoring thresholds to individual donors and designing personalized donation strategies, taking into account personal factors such as baseline iron levels, dietary habits and genetic differences. Moreover, although blood donor populations provide a valuable resource to study the mechanism between blood loss and iron parameters due to their regular blood donations and low CRP levels, the populations used are not representative of the general population due to the healthy donor effect. ¹⁴ , ²⁶ Additionally, in the case of study‐based data, specifically as from the INTERVAL study, the study population may not be representative of the country's donor population. ¹⁵ , ¹⁶ Moreover, it remains unclear to what extent donors experience symptoms related to anaemia or ID with declining Hb levels from an individual homeostatic setpoint. ¹⁷ It is also important to note that the reference Hb value used here does not necessarily reflect a true homeostatic balance, particularly in premenopausal and postmenopausal women.

Our findings have significant implications for blood donor management and clinical practice. The observed association and changepoint highlight ferritin levels that indicate insufficient iron availability for erythropoiesis. However, the lack of standardization across ferritin assays limits the establishment of a universal cut‐off for deficient iron stores for erythropoiesis. Future research should prioritize efforts to standardize ferritin measurements, as this would enable a more accurate assessment of whether changepoints are consistent across different populations. If changepoints remain variable despite assay standardization, it would suggest that underlying differences in population characteristics, lifestyle factors or blood donation policies play a significant role. Understanding these factors could help refine donor‐screening strategies and improve tailored interventions for iron deficiency management.

The concept of a personal reference for haematological parameters, as applied here to Hb, is not new. Studies on haematological setpoints have shown that these measures are stable, patient‐specific deep phenotypes, suggesting that deviations from an individual's homeostatic baseline, rather than fixed population‐wide cut‐offs, may be more relevant for clinical and donor management. ¹⁷ Our findings support this idea in the context of blood donation, demonstrating that Hb levels drop relative to an individual's own reference when ferritin is depleted. This underscores the need to move beyond one size fits all thresholds towards personalized donor management strategies that maintain homeostatic Hb levels. ²⁷ Such an approach could better address iron deficiency anaemia, account for individual variability in iron homeostasis and improve donor retention, health outcomes and overall blood donation efficiency.

In conclusion, our study reveals an association between Hb change and ferritin which demonstrates at which level and to what extent insufficient iron stores limit Hb recovery. It also underscores the complexity of ferritin assessment in blood donors, emphasizing the importance of measurement method‐specific considerations in the interpretation of ferritin levels, revisiting reference intervals and addressing assay standardization challenges.

AUTHOR CONTRIBUTIONS

MA and ET initiated the topic to be researched. MJ conceived of the presented idea. AM, MP and MJ designed the analyses. AM, ET, WL and HQ conducted the analyses supervised by MJ, KvdH, MA, WAR, EA and EDA. AM, MP, KvdH, MJ, HM and DS interpreted the results. AM wrote the manuscript. All authors critically reviewed and approved the manuscript.

FUNDING INFORMATION

This work was established as part of the Sanquin Blood Supply Foundation ‘Product and Process Development Cellular Products’ grants PPOC21‐10/L2590 and PPOC23‐09/L2753. Authors WL and WAR were supported by an Early‐Career Scientific Research Grant from the Association for the Advancement of Blood and Biotherapies (AABB) National Blood Foundation.

CONFLICT OF INTEREST STATEMENT

None of the authors have indicated a competing interest. AM, ET, MP, RS, TB, YP, KvdH, HM, DS, MA and MJ are employed by a national or regional blood bank, responsible for (part of) the blood supply in their respective countries.

Supporting information

Data S1.

Meulenbeld A, Turkulainen EV, Li W, Pothast MR, Qi H, Allara E, et al. Blood donor populations reveal a clear association between ferritin and change in haemoglobin levels. Br J Haematol. 2025;207(3):1096–1103. 10.1111/bjh.70066

DATA AVAILABILITY STATEMENT

Analysis code used to produce the results reported in this article is available online (https://github.com/Sanquin/Ferritin‐threshold‐for‐iron‐deficiency‐assessment). Access to individual participant data, after de‐identification, that underlies the results reported in this article (text, tables, figures and appendices) can be requested by researchers from General Data Protection Regulation compliant institutions, after the provision of a methodologically sound and feasible proposal to the corresponding author immediately following publication. A material transfer or research collaboration agreement has to be agreed and signed with the researcher. The INTERVAL Study Group has previously published its trial protocol, statistical analysis plan, informed consent form and other relevant study documents (https://www.donorhealth‐btru.nihr.ac.uk/studies/interval‐study/). Bona fide scientists can seek access to relevant de‐identified individual participant data (and a copy of the trial's data dictionary) by applying to the INTERVAL Data Access Committee at the following: helpdesk@intervalstudy.org.uk. The INTERVAL Data Access Committee reviews (supplemented, when required, by expertise from additional scientists external to the Committee) applications according to usual academic criteria of scientific validity and feasibility. Following approval by the INTERVAL Data Access Committee, a material transfer or research collaboration agreement will be agreed and signed with the applicants. Applicants might be requested to provide reimbursement of data management or preparation costs, as the INTERVAL trial is no longer in receipt of funding. Applicants will be required to provide updates to the INTERVAL Data Access Committee on their use of the INTERVAL trial data, including provision of copies of any publications. Applicants will be required to adhere in publications to the INTERVAL trial's policy for acknowledgement of the trial's funders, stakeholders and scientific or technical contributors.