{"paper_id":"50d43eb7-8b99-477c-a5ca-50d43d1b78de","body_text":"1 \nEvaluating treatment strategies for managing anaemia with erythropoiesis stimulating agent therapy in \nhaemodialysis patients: findings from a target trial emulation using electronic health record data \n \nAuthor list: Kate Birnie1, Fergus J Caskey1,2, Yoav Ben-Shlomo1,3, Dorothea Nitsch4,5,6, Anna Casula6, Eleanor J \nMurray7, Jonathan AC Sterne1 \n \nAffiliations: \n1. Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK \n2. Renal Unit, Southmead Hospital, Bristol, UK \n3. The National Institute for Health and Care Research Applied Research Collaboration West (NIHR ARC \nWest) at University Hospitals Bristol and Weston NHS Foundation Trust, UK \n4. Department of Noncommunicable Disease Epidemiology, London School of Hygiene & Tropical Medicine, \nLondon, UK \n5. Department of Nephrology, Royal Free London NHS Foundation Trust, London \n6. UK Renal Registry, Bristol, UK \n7. Boston University, School of Public Health, Boston, USA \n \nCorrespondence: kate.birnie@bristol.ac.uk \nKey words: Anaemia, Erythropoiesis stimulating agents, Haemodialysis, Target trial, Electronic health records \n \nWord counts \nAbstract: 291 \nMain text: 3,442 \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 \nAbstract \nBackground:  \nAnaemia is a common complication in chronic kidney disease (CKD), often managed with erythropoiesis-\nstimulating agents (ESAs) and intravenous iron. However, the optimal haemoglobin (Hb) target for treatment \nremains uncertain. We used electronic health records to emulate a target trial comparing different Hb \ntargets in haemodialysis patients, including ranges used in practice and those tested in earlier trials, to \nestimate their effects on mortality. \nMethods: We used electronic health records from selected UK kidney centres reporting to the Renal \nRegistry. Eligible patients were ≥18 years, on haemodialysis for ≥3 months, and without marked ESA \nresistance. Using the clone-censor-weight method, we emulated a trial comparing Hb targets of (1) 105–125 \ng/L versus 95–115 g/L, and (2) 125–145 g/L versus 95–115 g/L. Darbepoetin dosing followed a standardized \nprotocol with predefined dose adjustment rules and a maximum of 150 µg/week. The outcome was all-cause \nmortality over 8 months. \nResults: Among 8,628 patients from 10 kidney centres followed between 2004 and 2016; 62% were male, \nand the median age was 66 years (IQR 52–76). At baseline, mean Hb was 97.7 g/L (SD 15.1) and 79% were \nreceiving darbepoetin (median dose 30 µg/week). The estimated hazard ratios for mortality were 0.92 (95% \nCI 0.75, 1.12) comparing the 105–125 g/L and 95–115 g/L Hb target ranges, and HR 1.20 (95% CI 0.94, 1.54) \ncomparing the 125–145 g/L and 95–115 g/L ranges. Fewer patients remained adherent to the 125–145 g/L \ntarget range, limiting precision. \nConclusions: Compared with targeting a Hb range of 95-115 g/L Hb, targeting a slightly higher range of 105–\n125 g/L under controlled ESA dosing was not associated with increased mortality in haemodialysis patients. \nHowever, targeting Hb levels above 125 g/L may increase mortality risk, consistent with previous trial \nfindings. \n \n \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n3 \nLay Summary \nAnaemia, or low blood haemoglobin levels, is common in people receiving dialysis for kidney disease. It is \nusually treated with medicines that help the body make red blood cells, along with iron therapy. But the \noptimal haemoglobin target is unknown. Past clinical trials did not show clear benefits from targeting higher \nlevels and sometimes found increased risks of heart problems, so new trials are unlikely to take place. We \ndesigned a “target trial” to mimic a randomized trial using existing health records. We used data from over \n8,000 UK dialysis patients between 2004 and 2016 to emulate a trial. Aiming for a moderate haemoglobin \nrange (105–125 g/L) did not increase the risk of death compared to a lower range (95–115 g/L), whereas \naiming above 125 g/L may increase risk. These findings suggest that modestly higher haemoglobin targets, if \ncarefully managed, could improve symptoms and quality of life without increasing harm.  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n4 \nIntroduction \nAnaemia, defined by the World Health Organisation as haemoglobin (Hb) levels <130 g/L in men and <120 \ng/L in non-pregnant women1, leads to reduced quality of life2, fatigue3, decreased exercise capacity4 and \nshortness of breath5 in individuals with chronic kidney disease (CKD) and end-stage kidney disease who \nrequire dialysis.6 Erythropoiesis stimulating agents (ESAs), along with intravenous iron supplementation, are \nused to increase Hb levels, but the optimal level of Hb that should be targeted remains unknown. UK clinical \nguidelines recommend a target Hb of 100-120 g/L7 and Kidney Disease Outcomes Quality Initiative (KDOQI) \nguidelines recommend Hb should generally be in a target range of range of 110 to 120 g/L and the target \nshould not be greater than 130 g/L.8 A 2025 update to the Kidney Disease: Improving Global Outcomes \n(KDIGO) guideline (currently in draft and under public review) recommends targeting Hb levels below 115 \ng/L.9  \nRandomized trials in patients with CKD10-12 and end-stage kidney disease13 found either no change or an \nincrease in the risk of cardiovascular events with higher Hb targets. However, reaching a target Hb of 120 g/L \nrequires high doses of ESA in some individuals and high ESA doses are associated with a higher risk of death \nin observational studies and post hoc analyses of randomized trials.14-16 It has been proposed that high ESA \ndoses can cause thrombosis17 in the presence of relative iron deficiency, as well as arterial hypertension, \nendothelial activation, increased platelet reactivity, increased blood coagulability, and accelerated tumour \ngrowth, and inflammation.18-21 As a result, commonly used Hb targets have an upper limit of 120 g/L. \nIt is also possible that the higher risks associated with high ESA doses simply reflect that ESA resistance is a \nmarker of other co-morbidities (inflammation and cardiovascular disease).  If so, individuals without ESA \nresistance may benefit from a higher Hb target than the one typically used. Because it is unlikely that new \nrandomized trials will be conducted, this question needs to be addressed using observational data. We used \nUK electronic health record data to emulate a target trial comparing the effects of ESA treatment strategies \non mortality in haemodialysis patients. The target ranges compared were (1) 105–125 g/L versus 95–115 g/L \nand (2) 125–145 g/L versus 95–115 g/L. \n \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n5 \nMethods \nData \nThe UK Renal Registry (UKRR) collects clinical and biochemical data for all UK patients aged ≥18 years \nreceiving kidney replacement therapy (KRT).22 A bespoke data extraction was obtained for selected kidney \ncentres reporting to the UKRR. In each quarter of a calendar year, centres reporting <60% of haemodialysis \npatients being treated with erythropoiesis-stimulating agents (ESAs) were considered to have incomplete \ndata so were excluded. Data were extracted from 2004 to 2016, though the contribution period varied by \ncentre based on when prescribing data were sufficiently complete. Analyses were restricted to patients \nprescribed darbepoetin, as dose equivalency with other ESAs (e.g., epoetin) was insufficiently similar to \nallow combined analyses. \n \nResearch Ethics and Informed Consent  \nThe processing of UKRR data for research has been approved by the NRES Committee North East Newcastle \nand North Tyneside 1 Research Ethics Committee, reference 21/NE/0045. The UKRR has section 251 \npermissions to use data for research without individual patient consent. \n \nEmulating a target trial \nThe target trial emulation and analysis methodology has been described in detail previously.23 Patients were \neligible if they were ≥18 years old and had been on haemodialysis for ≥3 months. We excluded patients with \nevidence of marked ESA resistance (operationalised as receiving a high darbepoetin dose [≥120 µg/week]) \nand with low Hb [<80 g/L]) at baseline. The outcome was all-cause mortality over 8 months. \nWe first compared Hb target range 105–125 g/L with range 95–115 g/L. These ranges were chosen to fall \nslightly above and below the UK guideline range of 100–120 g/L7 while reflecting the spectrum of Hb targets \nrecommended or observed in international guidelines. We also compared target range 125–145 g/L with \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n6 \nrange 95–115 g/L. The 125–145 g/L Hb range was chosen to align with near-complete correction of anaemia, \nconsistent with historical randomized trials in which Hb targets approached or exceeded 130 g/L.10-13 \n \nDarbepoetin dose adjustments followed a standardized protocol designed to reflect typical clinical practice. \nDecisions to modify doses were based on current and recent Hb levels, whether the dose had been changed \nin the previous month, and observed Hb response to prior dosing adjustments (Figure 1). Dose changes were \nallowed only within a predefined acceptable range, with a maximum dose of 150 µg/week permitted \n(Supplementary Figure A1). Patients showing signs of ESA resistance requiring dose escalations were \ncensored unless the dose remained within the allowed limits. Additionally, a grace period (within which \nchanges should be made) of up to one month was incorporated to reflect delays in dose implementation. \nThis grace period helped accommodate treatment variability and minimized censoring caused by minor \ndeviations from the dosing rules. \nWe used the three-step ‘clone, censor, inverse-probability weight’ framework for comparing sustained \ntreatment strategies24. 1. Cloning: All patients met both strategies’ eligibility criteria at baseline, defined as \nthe latest of the dates when their dialysis centre became eligible and when they first met the eligibility \ncriteria. At this point (time zero of the target trial), their follow-up was duplicated, with one copy (‘clone’) \nassigned to each strategy being compared. 2. Censoring: Follow-up ended when the patient’s data was no \nlonger compatible with the strategy assigned to the clone because they received a darbepoetin dose outside \nthe acceptable range for the assigned strategy. 3. Inverse probability (IP) weighting, to account for the \nselection bias introduced by the informative censoring in the previous step.  \nData were organized into discrete 28-day intervals25 for each patient. Time zero was set at baseline. Each \nmonth, darbepoetin dose adjustments were recorded along with the corresponding Hb level that informed \nthe treatment decision. Darbepoetin dose was recorded as zero if the patient was not receiving darbepoetin \nor as a numerical value if prescribed. Non-zero doses were log-transformed for normality. Follow-up ended \neight months after baseline, death, kidney transplantation or a change to peritoneal dialysis, or loss to \nfollow-up, whichever happened first. \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n7 \nBaseline covariates included: patient age (years), sex, kidney centre (the largest centre was chosen as the \nreference category), primary kidney disease (diabetes, glomerulonephritis, pyelonephritis, polycystic kidneys \nand ‘other’ [hypertension, kidney vascular disease, other, uncertain]) and co-morbidities present at the start \nof KRT (angina, angioplasty, claudication, chronic obstructive pulmonary disease, diabetes not causing kidney \nfailure, ischaemic / neuropathic ulcers, liver disease, malignancy, previous myocardial infarction within last 3 \nmonths prior to starting KRT, previous myocardial infarction >3 months prior to start of KRT, previous \ncoronary artery bypass graft or coronary angioplasty, amputation for peripheral vascular disease, \nsymptomatic cerebrovascular disease, heart failure, and whether the patient was a smoker). Time-updated \ncovariates included: Hb (g/L), darbepoetin dose (µg/week), white blood cell count (109/L), albumin (g/L), \nferritin (µg/L), adjusted calcium (mg/dL), C-reactive protein (mg/L), urea reduction ratio (dialysis adequacy \n%) number of blood tests in the previous 28 days, time since eligibility into the study (months).  \nStatistical analysis \nDeriving weights \nBecause follow-up for a clone was censored when patients received a darbepoetin dose outside the \nacceptable range for the strategy assigned to that clone, models for treatment (darbepoetin dose levels) \nwere used to derive the probability of being censored, based on covariates up to and including the previous \nmonth. The treatment models (described in detail previously)23 included (1) logistic regression to estimate \nthe probability of receiving zero darbepoetin, with separate models for patients receiving and not receiving \ndarbepoetin in the previous month, to model cessation of darbepoetin and remaining off darbepoetin \nrespectively; (2) a multivariable heteroskedastic linear regression model for log darbepoetin dose among \nthose with non-zero dose; and (3) multinomial logistic regression for extreme doses (e.g. 2.5 or 150 \nµg/week) in that month. The multinomial models were for all possible categories of dose given the previous \nmonth. For example, from a dose of 2.5 µg/week, the possible categories are: zero dose, the same dose, an \nincrease consistent with both strategies, an increase consistent with the higher of the Hb target strategies \nonly, or an increase inconsistent with both strategies. Due to small sample sizes in some categories, only Hb \nand month were included as covariates in the multinomial models. \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n8 \nThe probability of adhering to the assigned treatment strategy in each month was derived by combining \nprobabilities from all relevant models. The cumulative probability of remaining uncensored was derived by \nmultiplying the probabilities of remaining uncensored during each month since baseline, with inverse \nprobability of censoring weights (IPCW) derived as the reciprocal of these probabilities. Patients could also \nbe censored for other reasons: separate logistic regression models, based on covariates up to and including \nthe previous month, were used to calculate censoring weights for patients who changed from haemodialysis \nto peritoneal dialysis, received a kidney transplant or were lost to follow-up and IPCW defined in the same \nway. IPCWs for each type of censoring were multiplied together to give the final consolidated weight for \neach strategy, patient and month.  \nPooled logistic regression models, weighted using the IPCW for each strategy, was used to estimate the \nmortality hazard ratio (HR) for each comparison. These models included cubic splines with 2 knots for month \n(to allow for non-linear relationships) and used robust standard errors for clustering by patient. In sensitivity \nanalyses the IPCW were truncated at the 99th, 95th and 90th percentiles (chosen a priori) to mitigate the \nimpact of large weights. We estimated weighted survival curves under each strategy using inverse \nprobability of treatment and censoring weights, truncated at the 99th percentile to improve stability. \nStatistical analyses were conducted in Stata 17. Confidence intervals were obtained using a nonparametric \nbootstrap procedure with 1,000 replicates. Figures of flow diagrams and dose change visualisations were \ngenerated in R (version 4.5.1) using the DiagrammeR, ggplot2, dplyr, and tibble packages. \n \nResults \nCharacteristics of the study cohort \nA total of 8,628 patients from 10 kidney centres were eligible for the study (Supplementary Figure A2). The \nmedian age was 66 years (IQR 52–76) and 5,351 patients (62%) were male (Table 1). Diabetes and \nglomerulonephritis were the primary kidney diseases in 1,783 (20.7%) and 1,228 (14.2%) patients \nrespectively. At baseline 6,773 (78.5%) patients were being treated with darbepoetin, with a median dose of \n30 µg/week (IQR 20, 50) and a mean Hb of 97.7 g/L (SD 15.1).  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n9 \n \nPredictors of darbepoetin treatment and dosing \nPredictors of darbepoetin dosing were consistent with clinical expectations. Prior dosing history was strongly \nassociated with current dose, and recent rises in haemoglobin were associated with dose reductions or \nstopping darbepoetin. Among those already receiving darbepoetin, patients with heart failure or with \nfrequent blood tests in the prior month were more likely to have darbepoetin stopped. Among those not \nreceiving darbepoetin, higher ferritin and CRP levels were associated with a lower likelihood of remaining on \nzero dose, consistent with anaemia management decisions in the context of inflammation. \n \nComparison of haemoglobin target ranges: 105–125 g/L vs 95–115 g/L \nFor the comparison of haemoglobin target ranges 105–125 g/L versus 95–115 g/L, Hb levels started to \ndiverge by the second month, with the 105–125 g/L target group maintaining higher Hb levels for the \nremainder of follow-up (Figure 2, panel A). By month 5, mean Hb in the 105–125 g/L strategy was 114.9 g/L \n(95% confidence interval [CI]: 114.5, 115.4 g/L), while in the 95–115 g/L strategy, it was 111.5 g/L (111.1, \n112.0 g/L). By month 5, the geometric mean darbepoetin dose (for those being treated with darbepoetin) \nwas 39.3 µg/week (95% CI 38.4, 40.2 µg/week) in the 105–125 g/L strategy and 34.9 µg/week (95% CI 34.1, \n35.7) in the 95–115 g/L strategy (Figure 2, panel C). As expected, the number of patients in the analysis \nreduced over time as patients stopped adhering to the assigned strategies, died, were censored or were lost \nto follow-up. At four months 56% patients remained in the 105–125 g/L strategy and 60% in the 95–115 g/L \nstrategy. By eight months, 37% remained in the 105–125 g/L strategy and 35% the 95–115 g/L strategy.  \nThere were 373 deaths during 39,155 patient-months follow-up in the 105–125 g/L strategy group, \ncompared to 434 deaths during 40,387 patient-months in the 95–115 g/L strategy group. In the weighted \nanalysis, which accounted for baseline and time-updated covariates, the estimated hazard ratio (HR) \ncomparing the 105–125 g/L versus the 95–115 g/L strategy was 0.92 (95% CI: 0.75, 1.12). The median weight \nwas 1.5, with the 90th, 95th, and 99th percentiles at 5.2, 8.8, and 25.5, respectively. Truncating weights at \nthe 99th, 95th, and 90th percentiles yielded similar effect estimates with narrower CIs: HRs of 0.91 (95% CI: \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n10 \n0.80, 1.03), 0.90 (95% CI: 0.82, 0.99), and 0.89 (95% CI: 0.82, 0.97), respectively. The unweighted estimate, \nwhich does not account for confounding, was 0.89 (95% CI 0.83, 0.94).  \n \nComparison of haemoglobin target ranges: 125–145 g/L vs 95–115 g/L \nFor the 125–145 g/L Hb strategy, mean Hb increased across follow-up, reaching 119.2 g/L (95% CI: 117.7, \n120.6 g/L) by month 8 (Figure 2, panel B), when the geometric mean of darbepoetin dose (for those being \ntreated with darbepoetin) was 49.4 µg/week (95% CI 46.6, 52.3 µg/week) (Figure 2, (Figure 2, panel D). \nThere were 187 deaths during 21,663 patient-months follow-up in the 125–145 g/L strategy group. Only 667 \npatients (8%) in the 125–145 g/L strategy group remained uncensored by month eight, indicating that this \nstrategy was not often followed in clinical practice. The median weight was 1.5, with the 90th, 95th, and \n99th percentiles at 6.2, 11.2, and 43.8, respectively.  \nThe estimated HR comparing the 125–145 g/L with the 95–115 g/L target was 1.20 (95% CI: 0.94, 1.54) after \ntruncating the weights at the 99th percentile. Truncating weights at the 95th, and 90th percentiles \nprogressively attenuated effect estimates and narrowed CIs: HR 1.11 (95% CI: 0.93, 1.32), and 1.03 (95% CI: \n0.88, 1.21), respectively. The fully weighted model produced unstable estimates due to a small number of \nvery large weights and is therefore not presented. The unweighted estimate was 0.95 (95% CI 0.85, 1.07). \nThe estimated survival curves for all three strategies are shown in Figure 3. The curve for the 125–145 g/L \ntarget strategy lies below those for the other two target ranges from around month 4 onward, although the \nconfidence intervals are wide. Confidence intervals widened over time, reflecting greater uncertainty due to \nthe increased proportion of patients whose follow-up was censored, reducing the amount of available data \nat later time points.  \n \nDiscussion \nUsing observational electronic health record data from the UKRR, we emulated target trials evaluating the \neffects of different Hb target strategies on mortality among haemodialysis patients. Because these strategies \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n11 \ncannot be distinguished by patients’ characteristics at the start of follow-up, we employed the clone-censor-\nweight method to emulate the target trial. We found no evidence of increased all-cause mortality comparing \nthe 105–125 g/L Hb with the 95–115 g/L target range over 8 months of follow-up. The 105–125 g/L target is \nslightly higher than the UK Kidney Association Clinical Practice Guidelines recommendation of 100–120 g/L7 \nand also exceeds the 115 g/L upper target limit suggested by KDIGO.9 The comparison of the 125–145 g/L \nversus 95–115 g/L target ranges suggested a potential increased mortality with the 125–145 g/L target \nrange. \n \nA randomized trial of 1,233 haemodialysis patients with cardiovascular comorbidity, published in 1998,13 was \nterminated early when it became clear that benefit from full correction of anaemia (compared to partial \ncorrection) was highly improbable. Randomized trials examining different Hb targets in patients with CKD \ninclude the CHOIR trial (published 2006),12 which randomized 1,432 patients with CKD not on KRT to Hb \ntargets of 135 g/L or 113 g/L. Patients assigned to the higher Hb target had an increased risk (HR 1.34; 95% \nCI: 1.03, 1.74) of the primary composite outcome (death, myocardial infarction, hospitalization for heart \nfailure, or stroke) compared to the lower Hb target. In contrast, the CREATE trial (published 2006),10 \nrandomized 603 patients with CKD not on KRT to a target Hb of 130–150 g/L or 105–115 g/L: the HR for \ncardiovascular events was 0.78 (95% CI: 0.53, 1.14). The TREAT trial, published in 200911, included 4,038 \npatients with diabetes, CKD, and anaemia. Participants were randomly assigned to achieve a haemoglobin \nlevel of approximately 130 g/L using darbepoetin alfa or placebo with rescue darbepoetin alfa if Hb <90 g/L). \nThe HR for the composite outcome of death or cardiovascular events was 1.05 (95% CI: 0.94, 1.17), but \nstroke incidence was higher in the darbepoetin alfa arm (HR 1.92; 95% CI: 1.38, 2.68). \nIn published trials, patients assigned to higher Hb targets typically received median ESA doses two to three \ntimes greater than those assigned to lower targets.21 Furthermore, a secondary analysis of CHOIR study data \nfound that failure to reach the target Hb and high ESA doses were associated with a higher risk of primary \noutcomes, including death, myocardial infarction, congestive heart failure, or stroke.16 An observational \nstudy using the target trial approach in 22,474 dialysis patients aged ≥65 years with diabetes and \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n12 \ncardiovascular disease from the United States Renal Data System (USRDS) evaluated clinical strategies for \nmanaging anaemia with epoetin therapy. 26  The study found similar 6-month survival targeting a medium \nhaematocrit range (34.5%–39.0%) and a lower haematocrit range (30.0%–34.5%), HR 0.98 (95% CI 0.78–\n1.24). A further analysis of USRDS data using the target trial approach with the parametric g-formula found \nthat, compared to a low-haematocrit strategy (30–33%), the estimated risk of death was 4.6% higher (95% \nCI: 4.4–4.9) under a high-haematocrit strategy (36–39%) and 1.8% higher (95% CI: 1.7–1.9) under a mid-\nhaematocrit strategy (33–36%).27 This study used an 18-month follow-up and included patients with a \nhistory of congestive heart failure or ischaemic heart disease in the two years prior to study entry. \n \nPrevious analyses of UKRR data have shown that anaemia management patterns for haemodialysis patients \nin the UK changed considerably between 2005 and 2013.28 There was a decrease in ESA use, the average \ndose administered, and the achieved Hb levels in UK haemodialysis patients over time. These trends suggest \nthat the results from randomized trials like CHOIR, CREATE, and TREAT, along with updated clinical \nrecommendations, have influenced clinical decision-making, prompting a more conservative approach to \nanaemia management in haemodialysis patients. The Proactive IV Iron Therapy in Haemodialysis Patients \n(PIVOTAL) trial, published after our data extraction period, is likely to have further influenced clinical practice \nby demonstrating that a proactive high-dose intravenous iron regimen reduced ESA requirements and \nimproved clinical outcomes compared to a reactive low-dose approach.29 As a result, current ESA prescribing \nis often combined with diverse iron strategies, which can complicate the interpretation of ESA effects alone. \nBy analysing data that precede the PIVOTAL trial, our study offers a more isolated assessment of ESA \ntreatment strategies, minimising influence from evolving iron management practices. Our results suggest \nthat a modestly higher Hb target may not be detrimental to patient survival, providing that excessive ESA \ndosing is avoided.  \n \nA strength of our analysis was the availability of detailed observational data, which allowed us to emulate a \ntarget trial and assess dynamic treatment strategies reflecting clinical practice. The UKRR is a large and \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n13 \nrepresentative database allowing trends in clinical practice patterns to be captured.30 By structuring the data \ninto monthly time periods we were able to incorporate information on Hb levels that informed darbepoetin \ntreatment decisions. The study design utilised variability in practitioners' usual darbepoetin dosing practices \nover time, assuming this variation was not systematically linked to other, unmeasured, factors influencing \nmortality. However, our study was limited by incomplete data returns, as some kidney centres do not \nroutinely record computerised data on ESA dose or drug type. We therefore restricted analyses to centres \nwhere at least 60% of haemodialysis patients were reported as receiving ESA treatment. A key limitation of \nthis study is its observational nature: we cannot exclude unmeasured confounding. However, we controlled \nfor the key variables that predict treatment decisions and/or are associated with all-cause mortality.  \n \nIn conclusion, our target trial emulation using EHR data found no evidence that targeting a Hb range of 105–\n125 g/L, compared with a 95–115 g/L range, with a maximum darbepoetin dose of 150 µg/week, increases \nall-cause mortality at 8 months in haemodialysis patients. However, targeting a range of 125–145 g/L may \nincrease mortality compared with a 95–115 g/L range. Haemodialysis patients may benefit from modestly \nhigher Hb, in terms of symptoms and quality of life, under a dosing strategy that limits changes in dose and \nmaximum permissible dose of ESAs.  \n \nDisclosures / Conflict of interest statement \nNone to declare.  \n \nData sharing statement \nThe dataset is not available due to privacy restrictions. Researchers may apply for access to UK Renal \nRegistry data through the UK Kidney Association. Information and application procedures is provided at \nhttps://www.ukkidney.org/audit-research/how-access-data. \n \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n14 \nAcknowledgements \nWe thank all the UK kidney centres for providing data to the UK Renal Registry. We thank Miguel Hernán for \ncomments on the analyses. We thank Charles Thompson for his input into the original study idea and the \ndevelopment of the trial design.  \n \nFunding \nKB was supported by a Medical Research Council (MRC) UK fellowship (R137881-101). YBS is partly funded \nby National Institute for Health and Care Research Applied Research Collaboration West (NIHR ARC West) \nand University of Bristol. \n \nAuthor contributions \nJACS, KB, FJC, YBS and DN designed the study. FJC and AC advised on the data. KB performed the analyses. \nJACS and EJM advised on the analyses. All authors contributed to interpreting results. KB wrote the initial \npaper draft and all other authors edited and provided feedback on drafts. All authors read and approved the \nfinal manuscript. \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n15 \nReferences \n1. WHO. Haemoglobin concentrations for the diagnosis of anaemia and assessment of severity.  2011. \nGeneva, World Health Organization. \n2. van Haalen H, Jackson J, Spinowitz B, et al. Impact of chronic kidney disease and anemia on health-\nrelated quality of life and work productivity: analysis of multinational real-world data. BMC Nephrol 2020; \n21: 88. 20200307. DOI: 10.1186/s12882-020-01746-4. \n3. Fishbane S and Spinowitz B. Update on Anemia in ESRD and Earlier Stages of CKD: Core Curriculum \n2018. Am J Kidney Dis 2018; 71: 423-435. 20180111. DOI: 10.1053/j.ajkd.2017.09.026. \n4. Odden MC, Whooley MA and Shlipak MG. Association of chronic kidney disease and anemia with \nphysical capacity: the heart and soul study. J Am Soc Nephrol 2004; 15: 2908-2915. DOI: \n10.1097/01.ASN.0000143743.78092.E3. \n5. Mathias SD, Blum SI, Sikirica V, et al. Symptoms and impacts in anemia of chronic kidney disease. J \nPatient Rep Outcomes 2020; 4: 64. 20200729. DOI: 10.1186/s41687-020-00215-8. \n6. Moura AF, Moitinho JAO, da Luz LG, et al. Anemia in Dialysis Patients. In: Fadem SZ, Moura-Neto JA \nand Golper TA (eds) Complications in Dialysis: A Clinical Guide. Cham: Springer International Publishing, \n2023, pp.157-170. \n7. Bhandari S, Oliveira B, Spencer S, et al. Clinical Practice Guidelines. Anaemia of Chronic Kidney \nDisease. 5th 2009-2012 ed. https://www.ukkidney.org/health-professionals/guidelines/guidelines-\ncommentaries: UK Renal Association, 2024. \n8. KDOQI. KDOQI Clinical Practice Guideline and Clinical Practice Recommendations for anemia in \nchronic kidney disease: 2007 update of hemoglobin target. Am J Kidney Dis 2007; 50: 471-530. DOI: \n10.1053/j.ajkd.2007.06.008. \n9. Group KDIGOKAW. KDIGO Clinical Practice Guideline for Anemia in Chronic Kidney Disease: Public \nReview Draft, https://kdigo.org/wp-content/uploads/2024/11/KDIGO-2025-Anemia-in-CKD-\nGuideline_Public-Review-Draft_Nov42024.pdf (2024, accessed 14/07/2025). \n10. Drueke TB, Locatelli F, Clyne N, et al. Normalization of hemoglobin level in patients with chronic \nkidney disease and anemia. N Engl J Med 2006; 355: 2071-2084. 355/20/2071 pii ;10.1056/NEJMoa062276 \ndoi. \n11. Pfeffer MA, Burdmann EA, Chen CY, et al. A trial of darbepoetin alfa in type 2 diabetes and chronic \nkidney disease. N Engl J Med 2009; 361: 2019-2032. NEJMoa0907845 pii ;10.1056/NEJMoa0907845 doi. \n12. Singh AK, Szczech L, Tang KL, et al. Correction of anemia with epoetin alfa in chronic kidney disease. \nN Engl J Med 2006; 355: 2085-2098. 355/20/2085 pii ;10.1056/NEJMoa065485 doi. \n13. Besarab A, Bolton WK, Browne JK, et al. The effects of normal as compared with low hematocrit \nvalues in patients with cardiac disease who are receiving hemodialysis and epoetin. N Engl J Med 1998; 339: \n584-590. 10.1056/NEJM199808273390903 doi. \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n16 \n14. Kilpatrick RD, Critchlow CW, Fishbane S, et al. Greater epoetin alfa responsiveness is associated with \nimproved survival in hemodialysis patients. Clin J Am Soc Nephrol 2008; 3: 1077-1083. CJN.04601007 pii \n;10.2215/CJN.04601007 doi. \n15. Solomon SD, Uno H, Lewis EF, et al. Erythropoietic response and outcomes in kidney disease and \ntype 2 diabetes. N Engl J Med 2010; 363: 1146-1155. 2010/09/17. DOI: 10.1056/NEJMoa1005109. \n16. Szczech LA, Barnhart HX, Inrig JK, et al. Secondary analysis of the CHOIR trial epoetin-alpha dose and \nachieved hemoglobin outcomes. Kidney Int 2008; 74: 791-798. 2008/07/04. DOI: 10.1038/ki.2008.295. \n17. Streja E, Kovesdy CP, Greenland S, et al. Erythropoietin, iron depletion, and relative thrombocytosis: \na possible explanation for hemoglobin-survival paradox in hemodialysis. Am J Kidney Dis 2008; 52: 727-736. \n2008/09/02. DOI: 10.1053/j.ajkd.2008.05.029. \n18. Bennett CL, Lai SY, Henke M, et al. Association between pharmaceutical support and basic science \nresearch on erythropoiesis-stimulating agents. Arch Intern Med 2010; 170: 1490-1498. DOI: \n10.1001/archinternmed.2010.309. \n19. Bennett CL, Lai SY, Sartor O, et al. Consensus on the Existence of Functional Erythropoietin Receptors \non Cancer Cells. JAMA Oncology 2016; 2: 134-136. DOI: 10.1001/jamaoncol.2015.3940. \n20. Singh AK. Resolved: Targeting a Higher Hemoglobin Is Associated with Greater Risk in Patients with \nCKD Anemia: Pro. Journal of the American Society of Nephrology 2009; 20: 1436. DOI: \n10.1681/ASN.2009040444. \n21. Vaziri ND and Zhou XJ. Potential mechanisms of adverse outcomes in trials of anemia correction with \nerythropoietin in chronic kidney disease. Nephrol Dial Transplant 2009; 24: 1082-1088. 20081105. DOI: \n10.1093/ndt/gfn601. \n22. Plumb LA, Hamilton AJ, Inward CD, et al. Continually improving standards of care: The UK Renal \nRegistry as a translational public health tool. Pediatr Nephrol 2018; 33: 373-380. 2017/06/24. DOI: \n10.1007/s00467-017-3688-2. \n23. Birnie K, Tomson C, Caskey FJ, et al. Comparative Effectiveness of Dynamic Treatment Strategies for \nMedication Use and Dosage: Emulating a Target Trial Using Observational Data. Epidemiology 2023; 34: 879-\n887. 20230926. DOI: 10.1097/EDE.0000000000001649. \n24. Hernán MA. How to estimate the effect of treatment duration on survival outcomes using \nobservational data. Bmj 2018; 360: k182. 2018/02/09. DOI: 10.1136/bmj.k182. \n25. Fewell Z, Hernán MA, Wolfe F, et al. Controlling for Time-dependent Confounding using Marginal \nStructural Models. The Stata Journal 2004; 4: 402-420. DOI: 10.1177/1536867x0400400403. \n26. Zhang Y, Thamer M, Kaufman J, et al. Comparative effectiveness of two anemia management \nstrategies for complex elderly dialysis patients. Medical care 2014; 52 Suppl 3: S132-139. DOI: \n10.1097/MLR.0b013e3182a53ca8. \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n17 \n27. Zhang Y, Young JG, Thamer M, et al. Comparing the Effectiveness of Dynamic Treatment Strategies \nUsing Electronic Health Records: An Application of the Parametric g-Formula to Anemia Management \nStrategies. Health Serv Res 2018; 53: 1900-1918. 20170530. DOI: 10.1111/1475-6773.12718. \n28. Birnie K, Caskey F, Ben-Shlomo Y, et al. Erythropoiesis-stimulating agent dosing, haemoglobin and \nferritin levels in UK haemodialysis patients 2005-13. Nephrol Dial Transplant 2017; 32: 692-698. DOI: \n10.1093/ndt/gfw043. \n29. Macdougall IC, White C, Anker SD, et al. Intravenous Iron in Patients Undergoing Maintenance \nHemodialysis. N Engl J Med 2019; 380: 447-458. 20181026. DOI: 10.1056/NEJMoa1810742. \n30. Macdougall IC, Tomson CR, Steenkamp M, et al. Relative risk of death in UK haemodialysis patients \nin relation to achieved haemoglobin from 1999 to 2005: an observational study using UK Renal Registry data \nincorporating 30,040 patient-years of follow-up. Nephrol Dial Transplant 2010; 25: 914-919. gfp550 pii \n;10.1093/ndt/gfp550 doi. \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n18 \nTable 1 Characteristics of study participants at baseline \nCovariate Category/unit N (%) or summary \nAge (years) Median (IQR) 66 (52-76) \nSex Male 5351 (62.0%) \n \nFemale 3277 (38.0%) \nPrimary kidney disease Diabetes 1783 (20.7%) \n \nGlomerulonephritis 1228 (14.2%) \n \nPolycystic kidneys 534 (6.2%) \n \nPyelonephritis 717 (8.3%) \n \nMiscellaneous 4366 (50.6%) \nYear of entering study 2004 777 (9.0%) \n \n2005 737 (8.5%) \n \n2006 635 (7.4%) \n \n2007 593 (6.9%) \n \n2008 595 (6.9%) \n \n2009 570 (6.6%) \n \n2010 600 (7.0%) \n \n2011 1061 (12.3%) \n \n2012 777 (9.0%) \n \n2013 767 (8.9%) \n \n2014 640 (7.4%) \n \n2015 447 (5.2%) \n \n2016 429 (5.0%) \nOn darbepoetin treatment Yes 6773 (78.5%) \n \nNo 1855 (21.5%) \nDarbepoetin dose, for those \ntreatment (µg/week) Median (IQR) 30 (20, 50) \nHb (g/L) Mean (SD) 97.7 (15.1) \nFerritin (µg/L) <300 3018 (35.0%) \n \n300-449 1545 (17.9%) \n \n500+ 1501 (17.4%) \n \nMissing 2564 (29.7%) \nC-reactive protein (mg/L) 0 or not tested 4432 (51.4%) \n \n0.1 to 4.9 426 (4.9%) \n \n5 to 19.9 1844 (21.4%) \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n19 \nCovariate Category/unit N (%) or summary \n \n20+ 1926 (22.3%) \nBlood tests in previous month (N) 0 2072 (24.0%) \n \n1 3933 (45.6%) \n \n2 853 (9.9%) \n \n3 433 (5.0%) \n \n4+ 1337 (15.5%) \nAlbumin (g/L) <35 4066 (47.1%) \n \n35-39 2497 (28.9%) \n \n40+ 1750 (20.3%) \n \nMissing 315 (3.7%) \nWhite blood count (109/L) <6 2316 (26.8%) \n \n6-6.9 1380 (16.0%) \n \n7-7.9 1370 (15.9%) \n \n8-8.9 1113 (12.9%) \n \n9+ 2449 (28.4%) \nAdjusted calcium (mg/dL) <2.3 3667 (42.5%) \n \n2.3-2.39 1741 (20.2%) \n \n2.4-2.49 1383 (16.0%) \n \n2.5+ 1428 (16.6%) \n \nMissing 409 (4.7%) \nUrea reduction ratio (%) <60 1692 (19.6%) \n \n60 to 69 1826 (21.2%) \n \n70 to 74 994 (11.5%) \n \n75-79 633 (7.3%) \n \n80+ 320 (3.7%) \n \nMissing 3163 (36.7%) \n \n \n   \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n20 \nFigure 1 Dose decision protocol \n \n \nNotes: Hb: haemoglobin. * When the decision is to change the dose, the new dose needs to be within \nacceptable levels (see Figure A1). The figure has been adapted from Birnie et al., Epidemiology 2023;34:879–\n887.23  \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n21 \nFigure 2 Mean Hb and geometric mean darbepoetin dose, with 95% CI, over time, by treatment strategy \n \nGeometric mean darbepoetin dose (for non-zero doses) with 95% confidence (CI) intervals from a weighted \nanalysis \n  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint \n\n22 \nFigure 3 Estimated weighted survival curves for all-cause mortality for the three target Hb strategies \n \nNote: shaded areas represent 95% confidence intervals derived from bootstrap resampling  \n . CC-BY 4.0 International licenseIt is made available under a \nperpetuity. \n is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted October 19, 2025. ; https://doi.org/10.1101/2025.10.17.25338210doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}