Clinical reference

Referral and pathways

If you read one page in this section and you do not work in a renal unit, read this one. Almost everything that goes wrong at the start of dialysis was decided months earlier, by someone who was not a nephrologist and who did not know there was a decision to make.

When to refer to nephrology

Refer to specialist kidney care when the 5-year risk of kidney failure is about 3 to 5% on a validated risk equation (in practice the Kidney Failure Risk Equation), or the eGFR is below 30 mL/min/1.73 m2, or there has been a sustained fall of 20 to 30% or more in GFR.

KDIGO CKD guideline, 2024

Use an externally validated risk equation to estimate absolute risk of kidney failure over 2 to 5 years in people with CKD G3 to G5. This is a genuine shift: risk-based rather than stage-based planning.

KDIGO CKD guideline, 2024

Plan for kidney replacement therapy, meaning pre-emptive transplant work-up and/or dialysis access, when the GFR is below 15 to 20 mL/min/1.73 m2, or the 2-year risk of KRT exceeds 40%.

KDIGO CKD guideline, 2024

Start assessment for renal replacement therapy or conservative management at least 1 year before therapy is likely to be needed.

NICE NG107, 2018

Plus the classic triggers, at any eGFR

A3 albuminuria (ACR above 300 mg/g, or above 30 mg/mmol).
Rapidly progressive CKD, or a sustained fall of 20 to 30% or more in GFR.
Refractory hypertension.
Recurrent or extensive nephrolithiasis.
Hereditary kidney disease, including ADPKD, which carries particular weight in Seychelles.
Persistent haematuria.
Persistent abnormalities of serum potassium.

One point that changes referral behaviour more than any threshold: staging by eGFR alone is incomplete, and KDIGO classifies CKD by Cause, GFR category and Albuminuria category together (the CGA system).

The practical consequence is that two people can both have "stage 3" kidney disease and sit in completely different risk categories depending on their albuminuria. A patient with an eGFR of 50 and an ACR above 300 is at higher risk than a patient with an eGFR of 40 and no albuminuria, and referring on the eGFR alone will get that pair the wrong way round. Measure the ACR.

Late referral is a preventable harm

This is worth stating as bluntly as the evidence allows, because the harm is real, it is common, and it is inflicted upstream by clinicians who never see the consequence. Late referral is associated with all of the following.

Catheter starts rather than fistula or peritoneal dialysis starts, with all the infection risk that follows.
Unplanned crash-landing dialysis, usually as an inpatient, usually as an emergency.
No time to discuss conservative care, so a frail patient is dialysed by default rather than by decision.
No chance of a pre-emptive transplant, which is the best outcome available and requires months of work-up.
No time to find and work up a living donor.
Worse outcomes, and a patient who has never had a real conversation about what they wanted.

Initiation: symptoms, not numbers

Start dialysis on symptoms, not on a number. This is the most important nuance in the whole of initiation, and it is the one most often got wrong by clinicians who are not nephrologists.

The IDEAL trial randomised 828 adults with progressive advanced CKD to an early start (eGFR 10 to 15) or a late start (eGFR 5 to 7). The groups converged in practice: the early group started at a mean eGFR of about 12.0 and the late group at about 9.8, a difference of only around 2.2, and the late-start group began dialysis a median of about 6 months later. There was no significant difference in all-cause mortality or in any clinically important secondary outcome, including cardiovascular events, infection, hospitalisation and quality of life (Cooper et al., New England Journal of Medicine, 2010).

The detail that should settle the argument: about three quarters of the "late" group had to start before reaching the eGFR 5 to 7 window, because they became symptomatic. eGFR alone is a poor trigger, and the trial demonstrated it by failing to keep its own protocol.

Where the guidelines disagree: the eGFR range at which asymptomatic initiation is considered

KDIGO (2012, framing carried into 2024): Start when symptoms or signs attributable to kidney failure appear. This often, but not invariably, occurs in the GFR range between 5 and 10 mL/min/1.73 m2. KDIGO 2024 frames the decision as a composite of symptoms, signs, quality of life, preferences, GFR level and laboratory abnormalities.
NICE NG107 (2018): Consider starting dialysis when indicated by the impact of uraemic symptoms on daily living, or by biochemical measures, or by uncontrollable fluid overload, or at an eGFR of around 5 to 7 mL/min/1.73 m2 if there are no symptoms. The decision must be made jointly by the person, their family or carers where appropriate, and the healthcare team.
KDOQI (2015): Base the decision primarily on signs and symptoms of uraemia, evidence of protein-energy wasting, and the ability to safely manage metabolic abnormalities and volume, rather than on a specific level of kidney function. The 2015 update removed the previous eGFR-based cut-offs entirely.
ERBP (2011) and CSN (2014): ERBP: consider dialysis in the presence of uraemic symptoms or signs, with an explicit caution against basing the decision solely on estimated GFR. CSN: an intent-to-defer strategy, with an absolute safety threshold of about eGFR 6 as the point at which to start regardless.

The NICE range of around 5 to 7 for asymptomatic patients is more conservative than the KDIGO range of 5 to 10. Both are safe, and both are guidance rather than a rule. What all of them agree on is that the number is not the trigger.

Symptoms and signs that do trigger initiation

Uraemic symptoms: nausea, vomiting, anorexia, weight loss, fatigue, pruritus, restless legs, poor sleep, impaired concentration, metallic taste.
Uraemic serositis: pericarditis or pleuritis. An urgent indication.
Fluid overload not controllable with diuretics: breathlessness, pulmonary oedema, refractory hypertension.
Refractory hyperkalaemia or metabolic acidosis.
Protein-energy wasting or nutritional deterioration despite dietetic input.
Cognitive impairment or uraemic encephalopathy.

Conservative care conversations

Conservative care, also called comprehensive conservative management, is a legitimate, active treatment choice, and NICE requires that it be offered as a genuine option alongside kidney replacement therapy to everyone likely to need it (NICE NG107, 2018).

It is not "no treatment". It is active management of anaemia, acidosis, fluid overload, pruritus, nausea, breathlessness and pain, plus blood pressure control, medication review and deprescribing, dietetic support, advance care planning, psychological and spiritual support, and palliative care input. It is everything except dialysis.

A clinician who presents it as "or we could do nothing" has not offered it. They have foreclosed it.

What the survival evidence actually says

A systematic review and meta-analysis of 22 cohort studies including 21,344 patients (there are no randomised trials) reported a pooled adjusted hazard ratio for death, dialysis versus conservative care, of 0.47 (95% CI 0.39 to 0.57), with high heterogeneity (I2 = 55%). Unadjusted median survival ranged from 20 to 67 months on dialysis versus 6 to 31 months on conservative care across studies (Nephrology Dialysis Transplantation, 2022).

That hazard ratio must never be quoted bare, and the authors say so themselves. The comparison is badly confounded. Patients who chose conservative care were a median of 7.0 years older (range 1.0 to 21.6 years), had more comorbidity in half the studies, and had lower functional and cognitive status. Only 3 of the 22 studies used the treatment decision itself as the survival starting point, so lead-time bias is likely. Adjusting for age and comorbidity changed the estimate only modestly, which the authors read as evidence of substantial residual confounding.

In patients aged 80 or over, the mortality advantage of dialysis persisted but was substantially reduced, and most individual studies showed no statistically significant difference. A companion review found no distinct advantage for either option in health-related quality of life or symptom burden.

The authors own conclusion is the one to carry into the consultation: high risk of bias and heterogeneous reporting preclude definitive conclusions, the results cannot be translated to an individual level, and they should be interpreted cautiously.

The defensible clinical summary is this. For a younger, fitter patient, dialysis very probably buys meaningful extra years. For a patient who is very old, frail, or heavily comorbid, dialysis may buy little or no extra time, and it will certainly cost time: hospital visits, procedures, and recovery from sessions.

Some patients in that group live longer on dialysis. Some live the same length of time but spend more of it in a hospital chair. Which of those matters more is the patient decision, not the clinician, and it should be revisited rather than decided once.

Transplant work-up triggers

Transplantation is the best treatment for kidney failure for those who are suitable, and it is a treatment rather than a cure. It replaces one set of problems with another: lifelong immunosuppression, infection and cancer risk, rejection, and the fact that grafts fail.

The survival evidence, from Wolfe et al. (New England Journal of Medicine, 1999): the annual death rate was 16.1 per 100 patient-years for all dialysis patients, 6.3 for dialysis patients on the transplant waiting list, and 3.8 for transplant recipients. The gap between the first two is selection, and it is why the honest comparison is transplant versus waiting list, not transplant versus all dialysis patients.

Transplantation carries early risk. The relative risk of death in the first 2 weeks after transplantation was 2.8 times that of matched waiting-list dialysis patients, survival probability became equal within 5 to 673 days across the subgroups examined, and long-term mortality was 48 to 82% lower among recipients, with a relative risk at 18 months of 0.32 (95% CI 0.30 to 0.35) (Wolfe et al., 1999).

The framing for a patient: transplantation is a bet that pays off, but the payment comes first. You are more likely to die in the weeks after a transplant than you were on dialysis, and then much less likely to die for every year afterwards. The break-even point is measured in months.

What should trigger a work-up

Consider planning for pre-emptive transplantation and/or dialysis access when the GFR is below 15 to 20 mL/min/1.73 m2, or when the 2-year risk of KRT exceeds 40% (KDIGO, 2024).
Offer a pre-emptive living donor transplant, where there is a suitable living donor, or pre-emptive listing for deceased donor transplantation, to people considered eligible after full assessment (NICE NG107, 2018).
Time on dialysis before transplant is itself an adverse prognostic factor for graft and patient survival in registry data. Every month of delay accrues cardiovascular damage.
Finding a living donor takes time: conversations, screening, and sometimes a paired or pooled exchange. That clock starts when someone mentions it, and nobody mentions it late.
Adopt a restrictive transfusion policy in transplant candidates, to minimise allosensitisation. HLA antibodies from previous transplants, pregnancies and transfusions make finding a compatible kidney much harder.
Eligibility is decided by a transplant centre after assessment, not by a threshold and not by a website. Cardiac fitness, respiratory reserve and frailty matter more than chronological age, and many programmes transplant people in their seventies and beyond.

A local caveat on transplantation

A hard local caveat, and we would rather state it than let a clinician assume otherwise.

No kidney transplant programme, and no overseas transplant referral pathway for Seychellois patients, is documented in any source we could trace. That is not the same as saying none exists: it means it is not published, and we will not invent it. If you need to know, ask the renal unit or the Ministry of Health directly.

The regional pattern is at least consistent with caution. Mauritius has no transplant service. The Maldives has no domestic transplant programme and funds transplantation abroad, in India or Sri Lanka. Across the Indian Ocean islands, haemodialysis dominates, peritoneal dialysis is rare to non-existent, and domestic transplantation is largely absent, so for many patients transplant means flying out.

The practical consequence for referral is that the transplant conversation may need to start earlier here than a clinician trained in a high-income system would expect, because the pathway itself may need to be constructed rather than merely entered.

Sources

KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of CKD (KDIGO, 2024)
KDIGO 2012 Clinical Practice Guideline for the Evaluation and Management of CKD (dialysis initiation framing) (KDIGO, 2012)
KDOQI Clinical Practice Guideline for Hemodialysis Adequacy: 2015 Update (KDOQI, American Journal of Kidney Diseases, 2015)
NG107: Renal replacement therapy and conservative management (NICE, 2018)
ERBP position statement on the timing of dialysis initiation (European Renal Best Practice, 2011)
A randomized, controlled trial of early versus late initiation of dialysis (the IDEAL trial) (Cooper BA et al., New England Journal of Medicine, 2010)
Comparison of mortality in all patients on dialysis, patients on dialysis awaiting transplantation, and recipients of a first cadaveric transplant (Wolfe RA et al., New England Journal of Medicine, 1999)
Survival of patients who opt for dialysis versus conservative care: a systematic review and meta-analysis (22 cohort studies, 21,344 patients) (Nephrology Dialysis Transplantation, 2022)
Canadian Society of Nephrology: intent-to-defer strategy for dialysis initiation (Canadian Society of Nephrology, 2014)
Kidney Transplant: eligibility, work-up and what to expect (NIDDK)
Global Dialysis Perspective: Mauritius (Kidney360, 2021)
Global Dialysis Perspective: Maldives (Kidney360, 2023)