Chapter 8 — Weaning & Renal Recovery

🎨 Cover art prompt: An anime-style illustration of a samurai kneeling to sheathe his katana at dawn beside a river that has just begun to run clear again on its own. The hollow-fibre cartridge that had been damming the flow now sits unlit and set aside on the mossy bank, cord coiled neatly beside it. A single green shoot pushes up through the cracked riverbed stone in the foreground, and the first true current of the season moves past unaided, catching pink dawn light. Mist lifts off the water. His expression is watchful rather than triumphant — the sword is sheathed but the hand still rests on the saya. Warm rose-gold and soft jade palette, ink-wash mountains. 16:9, cinematic, negative space upper-left.

Starting CRRT is a well-studied decision. Stopping it is not. This chapter is deliberately the most explicit in the guide about what the literature does not tell us.

8.1 Guideline Anchoring

KDIGO 2012 AKI, Section 5.9 — Discontinuation:

• Discontinue RRT when it is no longer required, either because intrinsic kidney function has recovered to the point that it is adequate to meet patient needs, or because RRT is no longer consistent with the goals of care. (Not Graded)

• We suggest not using diuretics to enhance kidney function recovery, or to reduce the duration or frequency of RRT. (2B)

⚠️ Read the grading carefully. KDIGO's discontinuation recommendation is ungraded — a statement of principle, not evidence. The only graded recommendation in this section is a negative one: do not use diuretics to try to hasten recovery. There is no KDIGO-endorsed numeric weaning threshold, because the evidence to define one does not exist.

8.2 Why Creatinine Fails You Here

Serum creatinine is the marker clinicians instinctively reach for, and it is close to useless during CRRT.

• CRRT clears creatinine continuously, so the serum value reflects the machine's work, not the kidney's.
• Creatinine generation is markedly reduced in critical illness — muscle wasting, immobility, reduced hepatic synthesis. A patient may have a "reassuring" creatinine with negligible glomerular filtration.
• Volume of distribution is expanded by fluid accumulation, diluting the concentration.
• Creatinine kinetics lag — the value takes days to equilibrate after CRRT stops, which is precisely the window in which the decision is being tested.

💡 A falling creatinine on CRRT tells you the circuit is working. A stable creatinine off CRRT for 24–48 hours tells you the kidney is working. Only the second is evidence of recovery.

8.3 Urine Output — The Best Available Predictor

The most-cited evidence comes from a post-hoc analysis of the BEST Kidney observational study (Uchino et al., Critical Care Medicine 2009), examining patients in whom CRRT was discontinued.

• Urine output in the 24 hours before discontinuation was the strongest predictor of successful liberation — outperforming creatinine, urea, severity scores and demographic factors.
• In patients not receiving diuretics, a threshold around 400–500 mL/day (the reported cut-off was approximately 436 mL/day) discriminated reasonably well.
• In patients receiving diuretics, a considerably higher urine output was required for the same predictive value, and discrimination was worse — diuretic-induced urine output is not equivalent evidence of recovered clearance.

⚠️ Interpret this correctly. BEST Kidney was a large observational cohort, and this was a post-hoc analysis of a non-randomised decision made by treating clinicians. It describes what predicted success in patients whom clinicians had already chosen to stop — it is not a validated rule for deciding whom to stop. No randomised trial has prospectively validated any urine output threshold for CRRT liberation.

💡 The practical corollary is nonetheless useful: urine output produced without diuretic support is a more meaningful recovery signal than the same volume produced with a furosemide infusion. This aligns with KDIGO's 2B suggestion against using diuretics to shorten RRT.

8.4 Other Candidate Markers

⚠️ The comprehensive declared gap for this chapter: there is no randomised controlled trial comparing weaning strategies for CRRT. There is no validated protocol, no validated threshold, and no validated biomarker. Every recommendation in circulation — including this one — rests on observational data, physiology and consensus. Anyone presenting a numeric weaning rule as evidence-based is overstating the literature.

8.5 The Cost of Getting It Wrong in Either Direction

Stopping too early — recurrent uraemia, fluid re-accumulation, electrolyte derangement, and the need for a second catheter and a second circuit, with the associated infection and vascular risks.

Continuing too long — ongoing anticoagulation exposure, catheter-related bloodstream infection risk, hypothermia, micronutrient and amino acid losses (Ch. 6), drug clearance complexity (Ch. 7), immobility, and cost.

✅ A finding from Chapter 3 belongs here too. In STARRT-AKI, among patients alive at 90 days, RRT dependence was higher in the accelerated-initiation arm (10.4% vs 6.0%). This is direct randomised evidence that extracorporeal support is not a neutral intervention with respect to kidney recovery — more RRT was associated with more long-term RRT dependence, not less. It is a reason to take liberation seriously as an active goal rather than a passive endpoint.

⚠️ A note on causality: STARRT-AKI's RRT-dependence finding is a secondary outcome, and the mechanism — whether earlier RRT impairs recovery, or reflects differences in ascertainment and in who received RRT at all — is not established. Treat it as an important signal, not a proven mechanism.

8.6 A Defensible Practical Framework

Offered explicitly as physiology and consensus, not as evidence-based protocol:

1. Ask the question daily. Liberation should be an active daily assessment, exactly as ventilator weaning is. The commonest reason CRRT continues is that nobody asked whether it should.
2. Establish that the original indication has resolved. No ongoing severe acidosis, hyperkalaemia, refractory overload or uraemic complication.
3. Look for spontaneous urine output — rising, and preferably without diuretic support. KDIGO suggests against using diuretics to shorten RRT (2B); diuretic-driven output is weaker evidence of recovery.
4. Confirm haemodynamic and volume stability — vasopressors weaning, no ongoing large-volume resuscitation that will immediately reaccumulate.
5. Consider a measured creatinine clearance during a planned pause if the decision is genuinely uncertain and the collection is feasible. Recognise the thresholds are not validated.
6. Stop and observe, rather than taper. There is no evidence base for gradual weaning of continuous therapy. A trial off is an assessment, and it is reversible.
7. Monitor closely for 24–72 hours after stopping — urine output, potassium, acid–base, fluid balance, and creatinine trend. Remember creatinine will rise initially as it re-equilibrates; a rise in the first 24 hours is expected and is not by itself failure.
8. Leave the catheter in until you are confident, balancing this against infection risk. Removing it prematurely converts a straightforward restart into a new procedure.
9. If restarting is required, it is not a failure. It is information: the kidney was not ready. Reassess and repeat.

8.7 Renal Recovery and What Follows

• Recovery may be incomplete. A meaningful proportion of survivors of dialysis-requiring AKI do not return to baseline renal function, and AKI is an established risk factor for subsequent chronic kidney disease.
• Follow-up matters. Survivors of AKI requiring RRT warrant nephrology follow-up and monitoring of renal function, blood pressure and proteinuria after discharge — a step frequently omitted in the transition out of intensive care.
• Avoid the second hit. Nephrotoxin exposure, contrast, and hypotension in the recovery window are modifiable and consequential.

💡 The final clinical pearl of this guide: the moment CRRT stops is not the end of the AKI episode. It is the beginning of the recovery phase, and the recovery phase is where the long-term renal outcome is largely determined.

8.8 Chapter Summary

1. KDIGO's discontinuation guidance is ungraded; its only graded recommendation is against using diuretics to shorten RRT (2B).
2. Creatinine is unreliable during CRRT — it reflects the circuit, not the kidney.
3. Urine output without diuretics is the best available predictor — from a post-hoc observational analysis, not a trial.
4. No RCT has ever compared CRRT weaning strategies. No threshold, protocol or biomarker is validated.
5. STARRT-AKI found higher 90-day RRT dependence with accelerated initiation — take liberation seriously as an active goal.
6. Ask the liberation question every day.