Highlights
– A 2025 Cochrane review of 24 non-randomised studies (26,127 people aged ≥65 with stage 5 CKD) found very low‑certainty evidence when comparing conservative kidney management (CKM) to dialysis.
– Observational synthesis suggested higher all-cause mortality with CKM (RR 1.28; 95% CI 1.17–1.41) but evidence is very low certainty due to severe confounding and heterogeneity.
– Health-related quality of life differences were inconsistent and imprecise; key outcomes such as cardiovascular events, malnutrition, sarcopenia, residual kidney function, and adverse events were not reported across studies.
Background and clinical context
Advanced chronic kidney disease (CKD) and kidney failure (often described as stage 5 CKD or end-stage kidney disease) are increasingly prevalent in older populations. As life expectancy rises and the population ages, clinicians increasingly encounter decisions about initiation of dialysis versus conservative kidney management (CKM) for frail or comorbid older adults. CKM (also termed conservative or supportive care) is a planned, patient-centred approach focused on symptom control, comorbidity management, advance care planning and palliative measures without initiating maintenance dialysis.
Choosing dialysis or CKM in older people is a complex, preference-sensitive decision. Dialysis can extend life in many patients, but it also imposes treatment burden (frequent clinic visits, vascular access procedures, dialysis-related symptoms) and may have variable effects on quality of life, particularly for older, multimorbid, or frail patients. High-quality comparative evidence to guide these decisions has been limited.
Study design and methods
The referenced work is a Cochrane systematic review (Yang et al., 2025) that aimed to evaluate the effects of CKM compared with dialysis specifically in people aged 65 years and older with kidney failure. The review searched multiple bibliographic databases, trial registries and grey literature up to 22 September 2025. Both randomised and non-randomised studies were eligible. Two reviewers performed independent risk of bias assessments using the ROBINS-I tool for non-randomised studies. Treatment effects were pooled with random-effects meta-analysis where appropriate and evidence certainty was graded with GRADE.
Key outcomes prespecified as critical were all-cause mortality, cardiovascular death, cardiovascular events and health-related quality of life (HRQoL). Important outcomes included hospitalisation, overall adverse events, malnutrition, sarcopenia and residual kidney function.
Included studies
No randomised controlled trials were identified. The review included 24 non-randomised studies enrolling a total of 26,127 people. Study designs, patient selection criteria, definitions of CKM, baseline comorbidity and frailty profiles varied substantially across included studies, and many had substantial imbalances in prognostic factors between CKM and dialysis groups.
Key findings and interpretation
All-cause mortality
Pooled analysis from 23 studies (24,628 participants) suggested a higher observed risk of death with CKM compared with dialysis: 813 deaths per 1,000 with CKM versus 630 per 1,000 with dialysis (RR 1.28, 95% CI 1.17 to 1.41). Heterogeneity was very high (I² = 89%). The authors judged the certainty of this evidence to be very low.
Interpretation: although the pooled estimate indicates an association between CKM and higher mortality, the observational nature of included data makes causal inference unreliable. Selection bias and confounding by indication are major concerns: patients directed to CKM were often older, more frail and had higher comorbidity burdens or poorer functional status at baseline than those selected or referred for dialysis. Many studies lacked adequate adjustment for key prognostic factors or used differing analytical approaches, contributing to heterogeneity.
Cardiovascular death and events
Three studies (262 participants) reported cardiovascular mortality: 114 per 1,000 with CKM versus 66 per 1,000 with dialysis (RR 1.72, 95% CI 0.68 to 4.34; I² = 0%). Evidence certainty was very low. No included study reported cardiovascular events as an outcome.
Interpretation: data are too sparse and imprecise to draw conclusions about cardiovascular outcomes. Absence of consistent reporting of cardiovascular events is a notable evidence gap given their importance in kidney failure.
Health-related quality of life
Two studies (186 participants) reported generic HRQoL using SF-36 component scores. For the Physical Component Summary (PCS), the mean difference favored dialysis by 1.46 points (MD −1.46, 95% CI −12.08 to 9.16; I² = 77%). For the Mental Component Summary (MCS), the mean difference was −2.50 points (95% CI −7.82 to 2.82; I² = 19%). Certainty was very low.
Interpretation: estimates are imprecise, come from small samples, and are unlikely to capture patient-centred outcomes that older people often prioritise (symptom burden, independence, functional status, and time spent at home). Minimal clinically important differences for these scores vary by context; the pooled estimates here do not provide robust evidence that dialysis improves HRQoL in this population.
Other outcomes
None of the included studies reported data on cardiovascular events, malnutrition, sarcopenia, residual kidney function, or adverse events—outcomes highly relevant to older patients’ functional status and treatment burden.
Risk of bias and limitations of the evidence
The authors judged the body of evidence to be of very low certainty primarily because of limitations inherent to non-randomised designs: confounding by indication, selection bias, imbalance in baseline prognostic variables, variable definitions of CKM, incomplete outcome reporting and heterogeneity in follow-up duration. Meta-analytic heterogeneity (I²) was substantial for several outcomes. The absence of randomized data means we cannot reliably separate the effect of treatment choice from baseline differences in patient health and goals of care.
Clinical implications — practical guidance for clinicians
1. Shared decision-making remains paramount. Given very low-certainty comparative evidence, decisions about CKM versus dialysis should be individualized, incorporating patient values, preferences, life expectancy, comorbidity, frailty, cognitive function and anticipated treatment burden.
2. Do not use eGFR alone to dictate dialysis initiation in older adults. Consider trajectory, symptom burden, and functional status. Multidisciplinary assessment (nephrology, geriatrics, palliative care, nursing, social work) improves alignment of treatment with patient goals.
3. For older, frail patients with high comorbidity and limited life expectancy, CKM is an appropriate, legitimate, and patient-centred option that emphasizes symptom control, advance care planning and quality of life; many patients prioritise remaining at home and avoiding treatment burden.
4. If dialysis is chosen, clinicians should optimize vascular access planning, incremental schedules where appropriate, and proactive symptom management and rehabilitation to reduce treatment burden and preserve function.
5. Document goals of care and advance directives, and integrate palliative care principles early—irrespective of the treatment pathway selected.
Research priorities
1. High-quality comparative research: pragmatic randomized trials would provide the most robust evidence but may be challenging; alternatives include well-designed prospective cohort studies with standardized baseline assessment, advanced causal inference methods (e.g., instrumental variable analyses) and trial emulation approaches.
2. Standardize CKM definitions and develop a core outcome set for trials and observational studies that includes survival, symptom burden, functional status, time at home, caregiver outcomes and healthcare utilization.
3. Larger, longitudinal studies should report cardiovascular events, nutritional and sarcopenia measures, residual kidney function trajectories, adverse events and cost-effectiveness to help inform trade-offs between survival and quality of life.
4. Research should examine heterogeneity of treatment effect: which subgroups (by frailty, comorbidity, functional status, cognitive function) benefit most from dialysis versus CKM?
Expert commentary and contextualization
The Cochrane review highlights an important real-world problem: clinicians and patients make high-stakes decisions with limited high-quality comparative evidence. Observational signals of higher mortality with CKM are expected given selection of sicker patients into non-dialysis pathways, and they should not be interpreted as definitive evidence that dialysis is superior for all older adults. Clinical guidelines emphasize individualized care and shared decision-making; this review reinforces that message and underscores the need for better data.
Conclusion
Current evidence comparing conservative kidney management with dialysis for people aged 65 years and older with kidney failure is derived entirely from non-randomised studies and is judged to be of very low certainty. Observational pooled estimates suggest higher mortality with CKM, but substantial confounding and heterogeneity limit causal interpretation. Data on cardiovascular events, nutritional status, sarcopenia, residual kidney function and adverse events are largely absent. Clinicians should use shared decision-making tailored to patients’ goals and frailty profile, integrate palliative care principles, and advocate for research that addresses the substantial evidence gaps.
Funding and trial registration
The systematic review was supported by the National Evidence‑based Healthcare Collaborating Agency (NA21‑001). The protocol is available at https://doi.org/10.1002/14651858.CD015151.
Key reference
Yang JW, Natale P, Kim S, Kim M, Jeon MS, Yi D, Hong YA, Chung S, Park WY, Hyun YY, Kwon SH, Shin SJ, Park DA, Kim J, Jung JH, Strippoli GF, Lee JY; supported by Cochrane Kidney and Transplant. Conservative kidney management versus dialysis for stage 5 chronic kidney disease in older people. Cochrane Database Syst Rev. 2025 Dec 9;12(12):CD015151. doi: 10.1002/14651858.CD015151.pub2. PMID: 41363177; PMCID: PMC12687411.