Adjudicating Heart Failure as an End Point in Dialysis Trials: More Signal, Less Noise Authors: Emma O’Lone, Patrick B. Mark Published: 2026 The definition of heart failure (HF) has evolved significantly over the past decade. The diagnosis is now far more nuanced. The ACC/AHA guidelines [1] define HF as a clinical syndrome caused by structural or functional cardiac abnormalities, leading to reduced cardiac output or elevated intracardiac pressures. Diagnosis requires symptoms and/or signs of HF (e.g., dyspnea, fatigue, edema), supported by objective evidence of cardiac dysfunction (imaging, biomarkers such as natriuretic peptides, or hemodynamic data). Echocardiography is the key diagnostic tool to confirm ventricular function and guide classification into categories of HF: HF with reduced ejection fraction (EF) (HFrEF), with EF ≤40%, HF with mildly reduced EF (HFmrEF), EF 41–49%, and HF with preserved EF, EF ≥50%. HF events have been a key component of a number of important randomized controlled trials (RCTs) in nephrology including the major sodium glucose co-transporter 2 inhibitor trials in chronic kidney disease (CKD) [2] and/or HF and trials of proactive iron use [3], treatment of hyperparathyroidism or anemia [4, 5], and mineralocorticoid receptor antagonist in people with kidney failure treated with dialysis [6, 7]. Table 1 presents an overview of the diagnostic criteria for HF events in some high profile RCTs, with these trials being representative of a wider expansion of cardiorenal outcome trials being performed in CKD. Table 1.Hospitalization for HF definitions across major trials​Trial (year)Intervention and populationPrimary/secondary/compositeDefinition of hospitalizationSymptoms describedLaboratory/radiologic/hemodynamic criteriaendpointNon-dialysisDAPA-HF (2019) [2]Dapagliflozin vs. placebo in general HF population (HFrEF)Primary composite endpoint (CV death and HF)Admission ≥24h with primary HF diagnosis and new/worsening HF symptoms plus objective evidenceDyspnea, orthopnea, PND, decreased exercise tolerance, fatigue, edema, rales, ascitesBNP/NT-proBNP >500/2,000 pg/mL, radiological pulmonary congestion, echocardiographic criteria (E/e’ >15, plethoric IVC), invasive filling pressures (PCWP ≥18 mm Hg)Secondary endpointASCEND-ND (2020) [5]Roxadustat vs. darbepoetin alpha in non-dialysis CKD populationSecondary endpointAdmission ≥24h with primary HF diagnosis, new/worsening symptoms, and objective evidenceDyspnea, fatigue, orthopnea, nocturnal cough, edema, abdominal fullness, reduced urine output, dizzinessBNP/NT-proBNP elevation, pulmonary congestion on imaging, echocardiography/MRI, or invasive HF evidenceDialysisACHIEVE (2025) [7]Spironolactone vs. placebo in dialysis populationPart of composite primary outcome (CV mortality and HF)Unplanned admission with signs/symptoms of HF and unplanned mechanical fluid removal (e.g., dialysis or ultrafiltration)Dyspnea (without other cause), bilateral basilar ralesInterstitial edema on chest X-ray, apical vascular redistribution, elevated LVEDP/PCWP, or lung ultrasound evidence of interstitial edemaSecondary outcomePIVOTAL (2018) [3]IV iron therapy standard vs. increased in dialysis patientsSecondary endpointUnplanned hospital admission ≥1 night with new/worsening HF and need for HF therapyDyspnea (exertional/rest), fatigue, orthopnea, PND, edema, rales, elevated JVP, S3Radiological pulmonary edema/cardiomegaly, abnormal echo or MRI, elevated BNP/NT-proBNP, invasive catheter dataEVOLVE (2012) [4]Cinacalcet vs. placebo in dialysis patientsSecondary composite endpoint (CV mortality or HF)Admission or acute presentation with volume overload requiring mechanical fluid removal (ultrafiltration or dialysis)Dyspnea plus ≥2 of: bibasilar rales, raised JVP, interstitial edema, increased pulmonary vessel diameterCXR interstitial edema, elevated LVEDP or PCWP (Swan-Ganz)HF, hospitalization for HF, HF = HF; BNP, B-type natriuretic peptide; NT-proBNP, N-terminal pro-BNP; PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-diastolic pressure; PND, paroxysmal nocturnal dyspnea; CKD, chronic kidney disease; IVC, inferior vena cava; CV, cardiovascular. Hospitalization for HF definitions across major trials HF, hospitalization for HF, HF = HF; BNP, B-type natriuretic peptide; NT-proBNP, N-terminal pro-BNP; PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-diastolic pressure; PND, paroxysmal nocturnal dyspnea; CKD, chronic kidney disease; IVC, inferior vena cava; CV, cardiovascular. The three cornerstones of HF diagnosis and classification all present significant challenges with advancing CKD and are particularly problematic in people with kidney failure requiring dialysis. Symptoms and/or signs of HF are indistinguishable to those described with volume overload, which is common as kidney function declines and is ubiquitous in dialysis patients. In HF, reduced cardiac output and impaired forward flow lead to reflex activation of neurohormonal systems that promote sodium and water retention which, combined with increased capillary hydrostatic pressures and venous congestion, cause peripheral and pulmonary edema [8]. While both HFpEF and HFrEF cause signs of volume overload, HFrEF more typically presents with overt systolic dysfunction, and signs of low forward output (fatigue, hypotension). In contrast, HFpEF is dominated by diastolic dysfunction (impaired ventricular relaxation and filling), so patients may show hallmark signs of congestion. Patients with kidney failure, dependent on dialysis to maintain sodium and water homeostasis, are at high risk of developing volume overload and the corresponding congestive signs and symptoms. Furthermore, volume overload clinically indistinguishable from the syndrome of HF may be the consequence of a missed dialysis session, inadequate ultrafiltration or excessive fluid consumption, rather than primary myocardial dysfunction. The ACC/AHA guidelines use the New York Heart Association (NYHA) classification to characterize symptoms and functional capacity of patients. The classification system is not practicable in patients on dialysis because dyspnea in these patients is not solely attributable to HF and its severity changes in relation to the timing of volume removal. In 2014, the Acute Dialysis Quality initiative did propose a classification system for use in dialysis patients which is based on patient-reported dyspnea, assessed both pre- and post-ultrafiltration, in conjunction with echocardiography [9]. This has not found itself into mainstream clinical practice let alone trial outcome definitions. The principal biomarkers are the natriuretic peptides. ProBNP, released from cardiomyocytes in response to myocardial stretch, is rapidly cleaved into two circulating fragments: the biologically active 32-amino acid B-type natriuretic peptide (BNP) and the inactive 76-amino acid N-terminal proBNP (NT-proBNP) [10]. These biomarkers have been used in the diagnosis of HF as well as to monitor interventions and to predict outcomes. Both BNP/NT-proBNP are partially renally excreted and therefore increase as eGFR decreases. In dialysis patients who are euvolemic these levels can still be up to 20 times higher than normal. Both BNP/NT-proBNP levels increase further as volume overload increases. There remains strong association in the dialysis population between increasing BNP/proBNP levels and both cardiovascular mortality and all-cause mortality, independent of volume status [11]. This suggests that BNP levels can help identify those at higher risk but are not sensitive enough to diagnose HF in the dialysis population. Finally, the critical use of echocardiogram in the diagnosis of HF is not always practical in the multitude of patients that arrive at the dialysis unit with symptoms consistent with HF or fluid overload. Echocardiogram is performed to aid in the categorization of EF in people with HF. This categorization has been shown to be associated with outcomes in the general population at 1 year with HFrEF increasing the risk of all-cause mortality and all-cause hospitalization though interestingly not cardiovascular death [12]. The timing of an echocardiogram around dialysis has a significant impact on many of the reported parameters. During interdialytic intervals volume overload, acid-base, and electrolyte shifts, as well as arterial and myocardial wall changes, result in dilatation of right cardiac chambers and pulmonary circulation overload. Before dialysis, preload is higher so e’ (early diastolic myocardial velocity) tends to be higher; E and E/e’ are likely elevated due to greater filling pressures. After dialysis, preload is reduced leading to a reduction in chamber volumes, pressures; thus e′ may drop and E/e’ may change. Timing matters, measurement taken at different points (pre, post, during dialysis) can yield different classification of diastolic dysfunction if relying on e’ or E/e’. Recent studies suggest that these alterations are more extended during the long interdialytic interval or the first dialysis session of the week and are associated with excess volume overload or removal [13]. The Standardised Outcomes in Nephrology Group (SONG) have determined core outcomes for trials conducted in hemodialysis patients. Cardiovascular disease has been deemed to be one of the core outcomes and HF was consistently identified as an important outcome to both patients and healthcare professionals [14]. Contemporary trials in CKD and kidney failure are prioritizing cardiovascular outcomes; however, there remains heterogeneity in the definition and adjudication of HF events, both between dialysis and non-dialysis populations and within dialysis trials themselves (Table 1). Non-dialysis trials generally define HF as a hospital admission of ≥24 h with a primary diagnosis of HF or an urgent HF visit requiring escalation of diuretic therapy, with adjudication supported by symptoms, physical signs, imaging, and biomarkers such as BNP or NT-proBNP. Dialysis trials adapt these definitions to account for dialysis-specific interventions. The recent report by Kang et al. [15] demonstrates the challenges of determining what represents a HF event in a clinical trial in dialysis patients, using data from the EVOLVE trial [4]. The main message from this important report is that approximately half of HF events reported by site investigators were adjudicated to not represent HF events by the clinical events classification committee, whilst by contrast, additional events were adjudicated to be HF endpoints which had not been reported as such by local site investigators. This issue has been explored in depth by a recent meta-epidemiologic study across a number of high impact clinical trials in dialysis [16]. ACHIEVE, for example, includes unplanned mechanical fluid removal (ultrafiltration or extra dialysis) as both diagnostic and therapeutic [7], whereas PIVOTAL focuses on hospital admissions with primary HF diagnosis but still incorporates fluid management considerations [3]. PIVOTAL's initial treatment effect was determined based on unadjudicated events, this was later changed and similar to Kang et al. [15], the events number was significantly reduced but the direction and size of treatment effect was similar [16]. EVOLVE adjudicated HF as a component of CV outcomes but did not include mechanical fluid removal explicitly [4, 16]. Furthermore, it is often hard to find definitions, e.g., to the best of our knowledge, recent large trials with key cardiovascular outcomes FLOW (in CKD) and ALCHEMIST (in dialysis) have not published their definitions used [6, 17]. The lack of transparency and differences highlight the complexity and heterogeneity of HF definitions, even within the dialysis population, emphasizing the need for context-specific adjudication criteria when comparing trial outcomes. Considering the challenges in diagnosis, it is clear to see that even with the benefit and considerable cost of adjudication in these trials, it is still hard to ensure accurate diagnosis and harder still to ensure comparisons across trials without a standardized definition for HF in trials in the dialysis population. Further research into more specific biomarkers in a dialysis population may be of benefit. For example, soluble ST2 (sST2), a member of the interleukin-1 receptor family has shown some promise in prediction of cardiovascular events. sST2 is released from the vascular endothelium and may serve as a marker of venous congestion suggesting further utility in assessment of HF [18]. sST2 has been widely mooted as a useful biomarker in management of HF mainly for risk stratification However, sST2 has only been assessed in a limited number of studies patients with kidney failure and has been shown to associate with prognosis rather than having diagnostic utility for HF [19]. Specific microRNAs and gene hubs have also been found to be associated with HF in maintenance dialysis patients which may present a new avenue in diagnosis [20]. The use of symptoms and clinical signs may be augmented by the addition of an echocardiogram as proposed by ADQI workgroup [9], although this may not be practical to implement on each admission in dialysis patients, who are frequently hospitalized with symptoms consistent with HF. Standardizing timing of echocardiograms around dialysis treatments and at a point of euvolemia should be the next step in achieving a standardized definition and would have multiple benefits. Ultimately, HF remains an important clinical outcome for people with both CKD and those with kidney failure requiring dialysis. Interventions have demonstrated the ability to reduce HF events over the course of clinical trials [2, 3]. However, HF has not been defined consistently across trials partly due to inherent challenges in defining HF in dialysis compared to CKD not requiring dialysis. A standardized definition of HF in dialysis trials would lead to improved comparability both inter and intra participants and trials. It may also avoid misclassification of diastolic dysfunction in HF and potentially reclassify the category of HF which has important treatment implications. The treatment effect of interventions on a HF outcome would be more reliably determined, improving interpretation of trial results. Without the confounding of dialysis related volume shifts in fluid status statistical power may be improved potentially meaning reduced sample sizes required for trials. A push to undertake pragmatic RCTs using routine healthcare data to ascertain clinical outcomes is attractive for reducing cost and administrative burden for both participants and trial sites. However, as Kang et al. [15] demonstrate, there is merit in using endpoint adjudication specifically for HF outcomes in patients requiring dialysis. This permits most precise determination of HF events allowing robust estimation of the effect of the interventional treatment on HF outcomes. We advocate for ongoing research into optimal assessment of HF outcomes in RCTs in kidney failure, which for now still require endpoint adjudication by an independent clinician or committee, but in future may be optimized by algorithmic approaches informed by intradialytic fluid gains, biomarkers, cardiac imaging, and capturing need for unplanned dialysis sessions. The aspiration should be to both continue to develop improved therapies to prevent and treat HF, but also to improve cardiovascular trial design and efficiency in people with CKD and/or kidney failure requiring dialysis. Conflict of Interest Statement P.M. received grants from AstraZeneca and Boehringer Ingelheim, consulting fees from AstraZeneca, Boehringer Ingelheim, Pharmacosmos, and Vifor, honoraria for lectures from AstraZeneca, Boehringer Ingelheim, and Vifor and participates in advisory boards for Vertex and NovoNordisk. E.O. has no conflict of interests. Funding Sources This commentary was not supported by any sponsor or funder. Author Contributions P.M. and E.O. contributed equally to the article.