Houston, TX (November 8, 2025) — Advanced chronic kidney disease (CKD) presents a formidable clinical challenge: determining which patients will derive greater benefit from conservative, non-dialytic management versus initiating dialysis. This therapeutic decision significantly impacts patient survival outcomes, quality of life, and healthcare resource allocation. Addressing this critical gap, a team of researchers from the University of California Los Angeles and the Veterans Affairs Greater Los Angeles Healthcare System has developed and rigorously validated an innovative mortality prediction model that estimates individualized survival probabilities under conservative management compared to dialysis. This novel tool draws upon extensive, real-world data from two national cohorts — the Veterans Affairs and OptumLabs® DataWarehouse databases — showcasing a significant advance in personalized nephrology care. The findings were unveiled at ASN Kidney Week 2025, held November 5–9 in Houston, Texas.
The traditional clinical paradigm has often leaned toward preemptive dialysis initiation for patients diagnosed with advanced CKD, frequently before robust evidence can confirm that this approach enhances survival or quality of life for every individual. With heterogeneous patient profiles and varied progression trajectories, a one-size-fits-all strategy fails to capture nuanced risk-benefit considerations. The researchers meticulously analyzed a multitude of clinical and laboratory variables associated with mortality risk. Principal among these were age, baseline and trajectory of estimated glomerular filtration rate (eGFR), albuminuria levels, frailty indices, nutritional status gauged by serum albumin and body mass index (BMI), recent hospitalization history, underlying comorbidities such as heart disease and sepsis, tobacco use, and whether the patient transitioned onto dialysis.
By integrating these multidimensional risk factors, the prediction model offers moderate discrimination — effectively stratifying patients based on risk — alongside acceptable calibration, ensuring predicted mortality probabilities closely align with actual observed outcomes. Such statistical performance metrics underscore the model’s utility in clinical contexts where granular survival estimation can directly inform patient-centered decision-making. Unlike existing tools that may assess risk in isolation, this model contextualizes survival probabilities by contrasting conservative management against dialysis initiation. Thus, it empowers nephrologists, patients, and their care partners with evidence-based insights tailored to individual health profiles.
Dr. Connie Rhee, the study’s corresponding author, elaborated on the model’s development as part of the NIH-funded OPTIMAL study — “Defining Optimal Transitions of Care in Advanced Kidney Disease: Conservative Management vs. Dialysis Approaches.” The study’s overarching mission is to establish a rigorous evidence base surrounding conservative management strategies in advanced CKD patients, a cohort traditionally underrepresented in clinical trials. Dr. Rhee emphasized that such tools are pivotal for enhancing shared decision-making conversations, enabling clinicians and patients to weigh the risks and benefits of dialysis initiation versus conservative care in light of personalized survival forecasts.
The methodology underpinning the model’s construction involved comprehensive retrospective analyses of national cohort datasets. These datasets included demographic variables, laboratory values, and clinical event histories, enabling the researchers to capture longitudinal health trajectories among veterans and broader patient populations. Sophisticated statistical modeling techniques, likely including survival analysis frameworks such as Cox proportional hazards models and machine learning algorithms, contributed to the identification of salient predictors and the formulation of the risk score. Validation procedures were performed across independent patient cohorts, confirming the model’s reproducibility and robustness.
This research emerges amidst growing recognition that conservative management — encompassing symptom control, management of comorbidities, and supportive care without dialysis — can be a viable and sometimes preferable option for selected patients with advanced CKD. Particularly among older, frail individuals or those with substantial comorbidity burdens, refraining from dialysis may align better with patient values and yield comparable survival with enhanced quality of life. Yet, clinical tools to precisely identify such candidates have been scarce, underscoring the significance of this prognostic innovation.
The practical implications extend beyond survival predictions. This model facilitates nuanced risk communication, bridging complex medical data and patient comprehension. By providing individualized mortality estimates, patients and families can better grasp prognostic realities and engage in deliberations reflecting their goals, preferences, and lifestyle considerations. This patient-centric approach aligns with evolving paradigms in nephrology that prioritize autonomy, individualized care planning, and holistic treatment strategies.
Moreover, the model’s development using large, heterogeneous datasets such as the Veterans Affairs and OptumLabs® DataWarehouse enhances its external validity and applicability across diverse CKD populations. The inclusion of veterans — typically an older, multimorbid cohort — and other insurance-based patient data underscores the model’s relevance for broad clinical practice settings. Importantly, the study also examined the impact of recent hospitalizations and acute illnesses such as sepsis on mortality risk, highlighting the dynamic interplay between chronic kidney function decline and concurrent health events.
The broader nephrology community eagerly anticipates further dissemination of these findings at Kidney Week 2025, the world’s premier nephrology conference drawing over 12,000 professionals globally. The platform provides an unparalleled venue for sharing advances, debating clinical practices, and fostering collaborations to elevate kidney disease care. The presentation of this mortality prediction model aligns with ongoing efforts to innovate management paradigms and personalize care in advanced CKD.
Future directions envisioned by the research group include prospective validation studies, integration of the model into electronic health record systems for real-time clinical decision support, and exploration of patient-reported outcomes alongside survival metrics. The synergy between data-driven prognostication and patient engagement tools promises to transform nephrology practice by facilitating evidence-based, tailored treatment pathways that respect patient priorities.
In summary, the development and validation of this mortality prediction model mark a transformative step toward optimizing care transitions in advanced CKD. By quantifying individualized survival likelihoods for conservative management versus dialysis, it equips clinicians and patients with actionable data to guide complex therapeutic decisions. As conservative management gains recognition as a legitimate alternative to dialysis for selected patients, such tools become indispensable in advancing nephrology toward precision medicine and shared decision-making excellence.
Subject of Research: Development and validation of a mortality prediction model comparing conservative management versus dialysis in advanced chronic kidney disease patients.
Article Title: Development and Validation of a Mortality Prediction Model for Conservative Management vs. Dialysis in Advanced CKD: Analysis of Two National Cohorts
News Publication Date: November 8, 2025
Web References:
– American Society of Nephrology: www.asn-online.org
– ASN Kidney Week 2025 (#KidneyWk)
Keywords: Advanced Chronic Kidney Disease, Conservative Management, Dialysis, Mortality Prediction Model, Personalized Medicine, Shared Decision-Making, Kidney Week 2025, Veterans Affairs Dataset, OptumLabs DataWarehouse, Survival Analysis, Nephrology, Risk Stratification
