Critical Care Nephrology · Two-Week Intensive · Lecture 9 of 9

Kidney Replacement Therapy II

Dose, anticoagulation, complications, drug dosing, blood purification and what happens after the machine is switched off

40 minutes Nephrology Fellows Week 2

Based on Koyner, Handbook of Critical Care Nephrology (2021) · NTUH Yunlin Branch

Learning objectives

By the end of this session you will be able to…

Write a complete CRRT prescription that delivers 20–25 mL/kg/h — and audit what was actually delivered.
Run regional citrate anticoagulation, recognise citrate accumulation early and troubleshoot the circuit.
Name the circuit and patient complications of CRRT, including the ones the therapy itself creates.
Dose antimicrobials and key ICU drugs on CRRT without repeating the commonest error in the ICU.
Judge blood purification and poisoning indications honestly, and plan recovery and post-AKI follow-up.

Koyner Ch 15, 26, 28, 30, 33–36, 51 · Ronco Ch 143, 154, 168 · KDIGO AKI 2012

Dose

Dose is a clearance the patient receives, not a number typed into a machine — and the two are never the same.

Koyner Ch 30 · KDIGO 2012 Section 5

What "dose" means — and what it does not

In CRRT, blood flow greatly exceeds effluent flow, so dialysate and ultrafiltrate leave the filter essentially saturated with small solutes. Effluent flow therefore equals small-solute clearance.
That is a surrogate, chosen because urea is measurable — not because urea is what kills the patient. It says nothing about middle molecules, volume control or acid–base control.
For intermittent HD the metric is different: Kt/V_urea >1.2 per treatment, or urea reduction ratio >0.67, thrice weekly.
Prescribing dose without prescribing volume, potassium, phosphate and buffer targets is prescribing half a therapy.

The four dimensions of dose

Small-solute clearance (effluent)
Middle-molecule clearance (membrane, convection)
Volume: net ultrafiltration rate
Duration: hours actually on therapy

Chen H, Palevsky PM. Koyner Ch 30 · Clark WR et al. Blood Purif. 2017;44:140–155 · KDIGO AKI 2012

Prescribed is not delivered

Delivered dose = Q_effluent ÷ weight × (hours on therapy ÷ 24) × pre-dilution factor
pre-dilution factor = Q_{plasma water} ÷ ( Q_{plasma water} + Q_pre-dilution )
Q_{plasma water} = blood flow × (1 − haematocrit). Every litre given before the filter dilutes the blood you are about to clear.

~16 haverage therapy actually delivered per 24 h in a large US centre — a third of the dose lost to downtime

1.1mean Kt/V achieved on the first treatment in ATN despite a strict 1.2–1.4 protocol

3things that steal dose: circuit clotting, procedures and imaging, and access dysfunction

Venkataraman R, Kellum JA, Palevsky P. J Crit Care. 2002;17:246–250 · Palevsky PM et al. N Engl J Med. 2008;359:7–20 · Koyner Ch 30, 33

The two trials that settled the dose question

ATN (VA/NIH Acute Renal Failure Trial Network) · NEJM 2008

1,124 critically ill patients with severe AKI. Intensive strategy = IHD 6 days/week plus CVVHDF at 35 mL/kg/h when unstable; less-intensive = IHD 3 days/week plus CVVHDF at 20 mL/kg/h. 60-day mortality 53.6% vs 51.5% (p = 0.47). No difference in recovery, and none in the subgroup that stayed haemodynamically stable throughout.

RENAL · NEJM 2009

1,508 patients in Australia and New Zealand randomised to CVVHDF at 40 vs 25 mL/kg/h. 90-day all-cause mortality 44.7% in both arms (p = 0.99). No difference in dialysis dependence among survivors.

Individual-patient-data meta-analysis · NDT 2018

Pooling the randomised dose trials confirmed no survival benefit from higher-intensity CRRT — and raised the concern that higher doses may be associated with impaired recovery of kidney function.

Palevsky PM et al. N Engl J Med. 2008;359:7–20 · Bellomo R et al. N Engl J Med. 2009;361:1627–1638 · Wang Y et al. Nephrol Dial Transplant. 2018;33:1017–1024 · Koyner Ch 30

From trial result to a number you write

The standard

KDIGO: deliver an effluent volume of 20–25 mL/kg/h for CRRT in AKI
KDIGO notes this "will usually require a higher prescription" — most units prescribe 25–30 to land at 20–25
The alternative, argued by the ATN investigators: prescribe 25 and defend the uptime instead
Intermittent: Kt/V_urea >1.2 per session, three times a week

When more dose is genuinely needed

Marked hypercatabolism: rhabdomyolysis, burns, severe trauma with uncontrolled acidosis or hyperkalaemia
Tumour lysis with rising potassium and phosphate between treatments
Poisonings where clearance is the therapy (Section 05)
Any patient in whom electrolyte, acid–base or volume targets are not being met — the non-urea dimensions

Common pitfall

Escalating the effluent rate because the patient has sepsis. It has never improved survival, it accelerates antibiotic, phosphate, magnesium and micronutrient losses, and the meta-analysis signal points the wrong way for recovery. If the patient is deteriorating, the problem is almost never the dose.

KDIGO AKI 2012, Section 5 · Koyner Ch 30 · Jörres A et al. Nephrol Dial Transplant. 2013;28:2940–2945 (ERBP)

A CRRT prescription you could write tonight

70 kg man, septic shock, KDIGO stage 3 AKI, anuric, haematocrit 30%. CVVHDF with regional citrate.

Setting	Value	Why this value
Blood flow (Q_b)	150 mL/min	Fixes the citrate dose; keeps filtration fraction low
Citrate ACD-A, pre-blood-pump	239 mL/h	Delivers ~3 mmol citrate per litre of blood at Q_b 150
Dialysate, calcium-free	1,100 mL/h	Diffusive workhorse; buffer reduced for the citrate load
Replacement, post-filter	300 mL/h	Adds convection without raising filtration fraction much
Net ultrafiltration	100 mL/h	Set from the fluid-balance goal, reviewed every 6 h
Calcium chloride, systemic	titrated	To systemic ionised calcium 1.0–1.2 mmol/L
Filter / set change	every 72 h	Or sooner for rising transmembrane pressure

The arithmetic

Total effluent = 239 + 1,100 + 300 + 100 = 1,739 mL/h
Prescribed = 1,739 ÷ 70 = 24.8 mL/kg/h
Plasma water = 150 × 60 × 0.70 = 6,300 mL/h
Pre-dilution factor = 6,300 ÷ 6,539 = 0.96
Corrected = 23.9 mL/kg/h
With 4 h downtime: × 20/24 = 19.9 mL/kg/h

Citrate rates from Tolwani AJ, Barker AB. Koyner Ch 33, Table 33.3 · Dose target KDIGO AKI 2012 · Koyner Ch 30

The daily dose audit — six numbers, five minutes

Hours on therapy in the last 24 h — the single biggest determinant of delivered dose. Chart it, do not estimate it.

Number of filter changes and why — clotting, scheduled change, transport, or an access alarm. Each has a different fix.

Actual effluent volume from the machine log, divided by weight and by 24 — not the prescribed rate.

Pre-dilution fraction — correct the effluent for any fluid delivered before the filter, including pre-blood-pump citrate.

Metabolic readout — urea and creatinine trend, potassium, phosphate, magnesium, bicarbonate and anion gap. This is the real adequacy test.

Cumulative and 24-hour fluid balance against the goal. Dose without volume control is an incomplete prescription.

Koyner Ch 30 · Vijayan A, Palevsky PM. Am J Kidney Dis. 2012;59:569–576

Anticoagulation

The circuit and the patient have opposite interests; regional citrate is the only strategy that serves both.

Koyner Ch 33 · Ronco Ch 142, 168 · KDIGO 2012

Choosing a strategy before choosing a drug

Clinical situation	No liver failure	Severe liver failure	Note
Low bleeding risk	Regional citrate, or unfractionated heparin	Unfractionated heparin, or none	Citrate is still first choice where it is available and monitored
High bleeding risk	Regional citrate	No anticoagulation	Regional means regional: no systemic effect if it is run properly
Heparin-induced thrombocytopenia	Regional citrate, or argatroban	Bivalirudin	Argatroban is hepatically cleared — halve it in liver disease
Active haemorrhage or post-neurosurgery	None, or regional citrate	None	Optimise flow and pre-dilution instead; accept shorter filter life

Key point

KDIGO recommends regional citrate anticoagulation as first line for CRRT in patients without a contraindication — including many patients who would previously have been given no anticoagulation at all because of bleeding risk.

Barker AB, Tolwani AJ. Koyner Ch 33, Table 33.1 · KDIGO AKI 2012, Section 5.5

Circuit patency before you reach for a drug

Filtration fraction = Q_{ultrafiltrate} ÷ ( Q_{plasma water} + Q_pre-dilution )
Q_{plasma water} = Q_b × (1 − haematocrit)
Keep it below 20–25%. Above that, the blood leaving the filter is haemoconcentrated and will clot whatever you infuse.

Why post-dilution CVVH clots more than CVVHD at the same dose: at a fixed blood flow, haematocrit and effluent rate, a purely convective mode always has the higher filtration fraction. Diffusion removes solute without removing plasma water.
Fixes, in order: correct the access → raise blood flow → move fluid from post- to pre-dilution → shift part of the dose from replacement to dialysate (i.e. CVVHDF or CVVHD).

Worked contrast

Same patient, same 25 mL/kg/h.

Post-dilution CVVH: 1,750 mL/h removed from 6,300 mL/h plasma water → FF 28% → repeated clotting.

CVVHDF as prescribed earlier: only 400 mL/h is post-filter convection → FF 6%.

Koyner Ch 33 · Joannidis M, Oudemans-van Straaten HM. Crit Care. 2007;11:218 · Ronco Ch 168

Systemic heparin: cheap, familiar, and paid for by the patient

How it is given

Bolus 25–50 U/kg (2,000–5,000 IU), then 5–10 IU/kg/h into the arterial limb of the circuit
Target circuit aPTT 45–60 s, or anti-Xa 0.3–0.6 IU/mL
Half-life ~90 min, up to 3 h in kidney failure; reversible with protamine
Heparin resistance is usually low antithrombin, not a dosing error

What it costs

Filter life tracks the aPTT, not the heparin dose: +10 s of aPTT reduced clotting but came with a 50% rise in intracranial and retroperitoneal bleeding
Bleeding complications in 10–50% of series; mortality attributable to bleeding up to 15%
HIT: falling platelets days 5–10, thrombosis not bleeding. Score it (4Ts), stop all heparin, switch to argatroban — 100 µg/kg bolus then 1 µg/kg/min, aPTT 1.5–2× baseline, 0.5 µg/kg/min in severe liver disease

Koyner Ch 33, Table 33.2 · van de Wetering J et al. J Am Soc Nephrol. 1996;7:145–150 · Link A et al. Crit Care Med. 2009;37:105–110

Regional citrate: how it actually works

ChelateCitrate infused at the start of the circuit binds ionised calcium. At 3–4 mmol citrate per litre of blood the post-filter iCa falls below 0.35 mmol/L and coagulation stops inside the circuit.

→

RemoveCalcium–citrate complexes are small and are largely cleared across the filter into the effluent — which is also why calcium is continuously lost.

→

MetaboliseCitrate returning to the patient is metabolised by liver, kidney and skeletal muscle; each citrate yields three bicarbonate. This is an alkali load, not a neutral one.

→

ReplaceA systemic calcium infusion returns the lost calcium and restores normal coagulation in the patient. Anticoagulation stays regional.

Key point

Citrate is simultaneously an anticoagulant, a buffer, a sodium load and a calorie source (~200 kcal/day). Every one of those four properties can become a complication.

Koyner Ch 33, Figure 33.1 · Oudemans-van Straaten HM, Ostermann M. Crit Care. 2012;16:249 · Ronco Ch 143 · Koyner Ch 15

The citrate monitoring set

Measurement	Target	Frequency	What it tells you
Post-filter ionised calcium	0.25–0.35 mmol/L	q6h if citrate is titrated; not needed if citrate is fixed to a constant blood flow	Whether the circuit is actually anticoagulated
Systemic ionised calcium	1.0–1.2 mmol/L	q6h, then q12h once stable at 48–72 h	Whether the calcium infusion is keeping up
Total calcium ÷ ionised calcium	<2.5	at least twice daily	The accumulation alarm — the single most important derived number
Bicarbonate, pH, anion gap	trend, not a value	q6h	Too much citrate (alkalosis), too little (acidosis), or accumulation (acidosis with a wide gap)
Sodium	stable	q6h	Hypertonic citrate solutions carry a large sodium load
Magnesium	replace to normal	daily	Citrate chelates magnesium too, and it is lost in the effluent

Koyner Ch 33 · Morabito S et al. Clin J Am Soc Nephrol. 2014;9:2173–2188 · Schneider AG et al. Crit Care. 2017;21:281

Citrate accumulation — recognise it in four lines

Pitfall — the picture that is missed on nights

Systemic ionised calcium falls despite the nurse repeatedly increasing the calcium infusion; total calcium rises; a metabolic acidosis worsens with a widening anion gap. The total-to-ionised calcium ratio crosses 2.5. Nothing else on the chart looks new.

Mechanism

Citrate is not being metabolised, so it accumulates and binds calcium systemically. Extra calcium is simply chelated — total calcium climbs, ionised calcium does not.

Who

Acute or acute-on-chronic liver failure, cardiogenic or septic shock with a high lactate, profound hepatic hypoperfusion. Low muscle mass reduces the reserve.

First moves

Lower the target citrate concentration or the blood flow; raise the effluent rate to clear more citrate; increase monitoring to q4h.

If it persists

Switch anticoagulation — no anticoagulation with optimised flow, or argatroban if the bleeding risk allows.

Liver failure is not an absolute contraindication

The L-CAT observational study and a subsequent meta-analysis found citrate can be used safely in liver failure and after liver transplantation — with a reduced citrate target, a higher effluent rate and tighter monitoring. Fear of citrate should not push a coagulopathic cirrhotic onto systemic heparin.

Koyner Ch 33 · Meier-Kriesche HU et al. Crit Care Med. 2001;29:748–752 · Slowinski T et al. Crit Care. 2015;19:349 (L-CAT) · Zhang W et al. Crit Care. 2019;23:22

Regional citrate troubleshooting at the bedside

What you see	Mechanism	First move
Circuit clots, post-filter iCa >0.4	Too little citrate for the blood flow being run, or blood flow drifting up	Increase citrate dose (or reduce Q_b if citrate is fixed); recheck post-filter iCa in 1 h
Circuit clots, post-filter iCa on target	Access dysfunction, recirculation, kinking, or filtration fraction too high	Reposition or replace the catheter; move to pre-dilution or add dialysate to drop the filtration fraction
Metabolic alkalosis	Citrate-derived bicarbonate exceeds the buffer requirement	Reduce blood flow (less citrate delivered) or increase the effluent rate; reduce buffer in the dialysate
Metabolic acidosis, ratio <2.5	Citrate delivery insufficient to buffer the patient's acid load	Increase blood flow or reduce the effluent rate; add bicarbonate
Metabolic acidosis, ratio >2.5, wide anion gap	Citrate accumulation — impaired metabolism	Reduce citrate target, increase effluent rate; if unresolved change anticoagulant
Hypernatraemia	Hypertonic citrate: 4% trisodium citrate contains 420 mmol/L sodium	Use a hypotonic dialysate or replacement fluid, or switch to ACD-A; recalculate the sodium balance
Systemic iCa low, ratio normal	Calcium replacement not matching effluent calcium losses	Increase the calcium infusion; confirm the line is running and not co-infused with phosphate
Systemic iCa high	Over-replacement, or a calcium-containing fluid reversing the circuit effect	Reduce the calcium infusion; check the fluid bag actually hung
Falling magnesium	Chelation by citrate plus effluent loss	Replace magnesium daily; do not wait for arrhythmia

Koyner Ch 33 · Morabito S et al. Clin J Am Soc Nephrol. 2014;9:2173–2188 · Schneider AG et al. Crit Care. 2017;21:281 · Ronco Ch 168

Citrate vs heparin: what the randomised evidence shows

Bai M et al. · Intensive Care Med 2015 — 11 RCTs, 992 patients

Risk of circuit loss lower with citrate than regional heparin (HR 0.52, 95% CI 0.35–0.77) and than systemic heparin (HR 0.76, 95% CI 0.59–0.98). Risk of bleeding lower than systemic heparin (RR 0.36, 95% CI 0.21–0.60). More HIT with heparin, more hypocalcaemia with citrate. No survival difference.

What to expect from a well-run circuit

Most citrate circuits should reach the scheduled 72-hour set change
A filter life consistently under 24 h is a system problem: access, filtration fraction, or citrate dose — in that order
Log why every filter was changed; the pattern names the fault

What citrate does not do

It does not improve survival — it buys uptime, and uptime buys delivered dose
It does not remove the need for a functioning catheter
It does not run itself: it needs a protocol, trained nurses and 6-hourly numbers

Bai M et al. Intensive Care Med. 2015;41:2098–2110 · Wu MY et al. Am J Kidney Dis. 2012;59:810–818 · Zhang Z, Hongying N. Intensive Care Med. 2012;38:20–28 · Koyner Ch 33

Complications

Half of them belong to the circuit, half to the patient — and some of the worst are created by the therapy working exactly as designed.

Koyner Ch 33–34 · Ronco Ch 143, 154

Circuit complications

Mechanical

Clotting — rising transmembrane and pre-filter pressures, dark striations in the fibres. Costs dose and costs blood — a clotted adult circuit is roughly 150–200 mL not returned
Blood leak — fibre rupture at high transmembrane pressure; effluent turns pink and the detector alarms. Stop, do not return the blood
Air embolism — entrained through the pre-pump segment where pressure is subatmospheric: loose connectors, cracks, unclamped ports. Air detectors have made it rare, not impossible

Access and membrane

Recirculation at fixed blood flow: ~4% internal jugular, 5% subclavian, 10% femoral — and 20–30% with the lines reversed, which happens in up to half of treatments
Access dysfunction — pressure alarms, positional flow, fibrin sheath. Catheter design, tip position and length matter (Lecture 8)
Bioincompatibility — complement and leukocyte activation; the classic bradykinin reaction with an AN69 membrane in a patient on an ACE inhibitor: flushing, hypotension and bronchospasm within minutes of connection

Koyner Ch 33 · Ronco Ch 154 · Leblanc M, Fedak S, Moskis G, et al. Blood recirculation in temporary central catheters for acute hemodialysis. Clin Nephrol. 1996;45:315–319

Patient complications

Haemodynamic

Hypotension at connection — extracorporeal volume, vasodilation from cool fluid, bradykinin
Excessive net ultrafiltration outrunning plasma refill
Arrhythmia from rapid potassium, magnesium or calcium shifts

Metabolic

Hypophosphataemia — near-universal after 24–48 h; weakens the diaphragm and delays weaning. Use a phosphate-containing solution or supplement systematically
Hypokalaemia and hypomagnesaemia
Citrate, lactate or glucose gain from the solutions — or glucose loss into glucose-free effluent

Nutritional and infective

Amino acid loss ~0.2 g per litre of effluent; albumin and protein loss up to 20 g/day with some membranes
Water-soluble vitamin, selenium and trace element depletion
Catheter-related bloodstream infection — the risk that scales with every extra day of therapy

Druml W, Kalantar-Zadeh K. Koyner Ch 15, Table 15.3 · Ronco Ch 143 · Yang Y, Zhang P, Cui Y, et al. Crit Care. 2013;17:R205

The therapy as the problem: temperature and dialytrauma

Pitfall — the patient who cannot mount a fever

Room-temperature fluid running at 2 L/h through an extracorporeal circuit removes substantial thermal energy. The patient becomes normothermic while septic. Never use temperature to exclude a new infection in a patient on CRRT — use the white count trend, the vasopressor requirement, cultures and the catheter site.

Cooling is also a tool: it reduces vasodilation and can lower vasopressor requirement, and a CRRT circuit can be used to deliver targeted temperature management where the equipment allows.
The cost is shivering (which raises oxygen consumption), impaired coagulation and masked infection. Warm deliberately or cool deliberately — do not drift.

Dialytrauma

The collective harm from the therapy itself: heat loss, loss of electrolytes and divalent ions, water-soluble vitamins and trace elements, enhanced antibiotic clearance, membrane bioincompatibility and repeated hypotension.

Its most important consequence is the possibility that CRRT delays renal recovery — through recurrent intradialytic hypotension and through the higher-dose signal in the individual-patient-data meta-analysis. This is the argument for the lowest effective dose and the earliest reasonable stop.

Maynar Moliner J, Honoré PM, Sánchez-Izquierdo Riera JA, et al. The dialytrauma concept. Blood Purif. 2012;34:177 · Wang Y et al. Nephrol Dial Transplant. 2018;33:1017–1024 · Ronco Ch 143, 172 · Koyner Ch 34

Drug dosing & nutrition on CRRT

A patient on 25 mL/kg/h is not anuric to a drug — they have a creatinine clearance of about 30 mL/min, and almost nobody prescribes as if that were true.

Koyner Ch 15, 26 · Ronco Ch 073, 078

What governs drug removal

CL_total = CL_non-renal + CL_{residual kidney} + CL_CRRT
CL_CRRT ≈ Q_effluent × sieving coefficient (sieving ≈ unbound fraction)
25 mL/kg/h in a 70 kg patient = 1,750 mL/h = 29 mL/min. That is your patient's drug-clearing "GFR" — and it does not stop overnight.

Cleared well by CRRT

Molecular weight <1,000 Da
Protein binding low — the free fraction is what sieves
Volume of distribution <0.8 L/kg (water-soluble)
High effluent rate, long uninterrupted therapy

Barely touched

Large volume of distribution — digoxin, tricyclics, amiodarone
Heavily protein bound — echinocandins, ceftriaxone, most antipsychotics
Large molecules — insulin (>5,000 Da)
But note: hypoalbuminaemia raises the free fraction and therefore raises removal of protein-bound drugs

Jang SM, Mueller BA. Koyner Ch 26 · Roberts DM et al. Crit Care Med. 2012;40:1523–1528

The commonest error in the ICU

Pitfall — the renal-failure dose written on admission and never revisited

"Underdosing is overprevalent." With ceftazidime 2 g every 12 h only 53% of CRRT patients reached the pharmacodynamic target; with cefepime 2 g every 12 h, none did. In RENAL, trough concentrations varied 6.7-fold for meropenem and 10.5-fold for tazobactam between patients on identical orders — 15% never reached the MIC target and 40% missed the higher target, while 10% were frankly excessive.

Fullloading dose regardless of kidney function — often larger, because the volume of distribution is expanded

Intervaladjust the interval and the maintenance dose, not the loading dose

TDMtherapeutic drug monitoring beats every table, including the next one

Lewis SJ, Mueller BA. Semin Dial. 2014;27:441–445 · Roberts DM et al. Crit Care Med. 2012;40:1523–1528 · Koyner Ch 26

Cheat sheet — assuming effluent 20–25 mL/kg/h and no residual function

Agent	Loading dose	On CRRT	The point to remember
Vancomycin	20–25 mg/kg	Then 7.5–10 mg/kg q12h or 15–20 mg/kg q24h	Dose to AUC₂₄/MIC 400–600 by monitoring — never to a table
Piperacillin–tazobactam	4.5 g	4.5 g q8h, extended infusion over 4 h	Time above MIC is what kills; extending the infusion is free
Meropenem	1 g	1 g q8h; 2 g q8h for MIC ≥2 mg/L or CNS infection	The q12h "renal" dose is an end-stage dose, not a CRRT dose
Cefepime	2 g	2 g q12h; 2 g q8h for Pseudomonas	Watch neurotoxicity — non-convulsive status is under-recognised
Aminoglycosides	Higher than usual (expanded V_d)	Extended interval, redose on measured level	Concentration-dependent: chase the peak, then wait
Fluconazole	800 mg	400–800 mg q24h	Small, water-soluble, minimally bound — it sieves freely and is easily underdosed
Levetiracetam	Standard	500–1,000 mg q12h	Renally cleared and dialysable; underdosing causes breakthrough seizures
Acyclovir	Standard mg/kg	5–10 mg/kg q24h	Also a nephrotoxin — keep the patient volume-replete
Low-molecular-weight heparin	—	Avoid	Unpredictable in kidney failure; use unfractionated heparin instead
Argatroban	100 µg/kg	1 µg/kg/min; 0.5 in severe liver disease	Hepatically cleared — kidney function and CRRT do not change the dose
Echinocandins	Standard	No adjustment	Highly protein bound and large — CRRT does not remove them

Koyner Ch 26, Tables 26.1–26.2 · Trotman RL et al. Clin Infect Dis. 2005;41:1159–1166 · Heintz BH, Matzke GR, Dager WE. Pharmacotherapy. 2009;29:562–577

Nutrition on CRRT

Target	Value
Energy	20–25 kcal/kg/day, maximum 30; ~20 in patients over 60
Protein / amino acids on KRT	1.2–1.5 g/kg/day; up to 1.7 g/kg/day if hypercatabolic
Glucose	2–3 g/kg/day; target blood glucose <180 mg/dL
Water-soluble vitamins	Twice the recommended daily allowance; extra thiamine early
Selenium	Supplement — it is cleared by CRRT

The hidden ledger

Out: ~0.2 g of amino acids per litre of effluent — roughly 10–15 g/day at standard dose; protein loss up to 20 g/day across some membranes; phosphate, magnesium, selenium, thiamine, vitamin C
In: citrate ~200 kcal/day; ACD-A adds dextrose calories; lactate-buffered solutions add several hundred more. Count them, or you will overfeed
Triglycerides above ~800 mg/dL interfere with the therapy and clot the filter — remember propofol is a lipid

Koyner Ch 15, Table 15.4 (Druml W) · Ronco Ch 073, 143 · European nutrition-society targets as summarised in Koyner Ch 15

Blood purification & advanced extracorporeal therapy

A beautiful hypothesis, twenty years of trials, and a short list of indications that actually survived.

Koyner Ch 28, 34–36 · Ronco Ch 100, 160, 193

The sepsis blood-purification hypothesis

Circulating PAMPs — endotoxin above all — trigger an overwhelming cytokine response; the anti-inflammatory counter-wave then leaves the host immunosuppressed.
Higher plasma endotoxin and cytokine levels track with more severe AKI and higher mortality. Removing them ought to help.
Four device families: endotoxin adsorption (polymyxin B), cytokine adsorption (CytoSorb), combined membranes (oXiris, which can also provide KRT), and plasma exchange.
The theoretical hole: cytokines are produced by tissue macrophages, and plasma levels may not fall even when the device is clearing them — the circulation refills from the interstitium.

High-volume haemofiltration

IVOIRE randomised septic shock with AKI to 70 vs 35 mL/kg/h and found no survival difference. In the ATN and RENAL dose trials, higher intensity likewise gave no benefit in the sepsis subgroups.

Higher volumes also mean more antibiotic, electrolyte and micronutrient loss — dialytrauma scaled up.

Jansen A, Pickkers P. Koyner Ch 34 · Joannes-Boyau O et al. Intensive Care Med. 2013;39:1535–1546 (IVOIRE) · Koyner Ch 36

What the adsorption trials actually showed

EUPHAS · JAMA 2009 — polymyxin B haemoperfusion

64 patients with abdominal septic shock, stopped early for a survival signal. Hazard ratio 0.36 (0.16–0.80) — but absolute 28-day mortality 32% vs 53%, p = 0.09, and mortality was never the powered endpoint.

ABDO-MIX · Intensive Care Med 2015

243 patients with peritonitis and septic shock. No benefit on survival, organ dysfunction or inflammatory markers; 38% of sessions were incomplete.

EUPHRATES · JAMA 2018

Septic shock with confirmed endotoxaemia (endotoxin activity assay ≥0.60) and a genuine sham control. No difference in 28-day mortality, and the assay did not fall more with treatment than with sham. The post-hoc "addressable endotoxaemia" subgroup is hypothesis at best.

Where this leaves us

CytoSorb removes IL-6 across the filter but has not changed plasma cytokine levels or outcomes in randomised trials; oXiris has retrospective data only. Routine blood purification for sepsis is not supported. Standard CRRT in sepsis-associated AKI is organ support, not mediator therapy.

Cruz DN et al. JAMA. 2009;301:2445–2452 · Payen DM et al. Intensive Care Med. 2015;41:975–984 · Dellinger RP et al. JAMA. 2018;320:1455–1463 · Schädler D et al. PLoS One. 2017;12:e0187015 · Koyner Ch 34

Therapeutic plasma exchange: where it does earn its place

Established ICU indications

TTP — daily exchange is the emergency; start on clinical suspicion, with steroids and caplacizumab/rituximab per protocol
Anti-GBM disease — exchange plus immunosuppression, urgently in diffuse alveolar haemorrhage or dialysis-requiring disease
ANCA vasculitis — reserve for diffuse alveolar haemorrhage or the most severe kidney disease; PEXIVAS reduced the enthusiasm considerably
Myasthenic crisis, Guillain–Barré, and selected antibody-mediated transplant rejection

Where it does not

Sepsis — a meta-analysis of four small trials in 194 patients showed no overall mortality benefit; the adult subgroup signal (RR 0.63) is not practice-changing
Myeloma cast nephropathy — routine exchange is not supported; any benefit appears limited to biopsy-confirmed cast nephropathy with light-chain–guided dosing
Remember the cost: citrate load and hypocalcaemia, coagulation factor and immunoglobulin depletion, and allergic or transfusion reactions with donor plasma

Koyner Ch 34 · Rimmer E et al. Crit Care. 2014;18:699 · Walsh M et al. N Engl J Med. 2020;382:622–631 (PEXIVAS) · Ronco Ch 160

CRRT on ECMO: three ways to connect, three sets of problems

Configuration	Advantage	Problem
Separate circuits	Independent flow and ultrafiltration control	Another large catheter under systemic anticoagulation; air entrainment risk during venous cannulation
In-line haemofilter	Cheap, small priming volume	Relies on external infusion pumps — fluid-balance errors over 800 mL have been reported
CRRT machine spliced into the ECMO circuit	Pre- and post-oxygenator ports are easy to use; the oxygenator acts as a bubble and clot trap	Blood loss at high-pressure connections, air entrapment at low-pressure ones

Numbers that explain the alarms

About half of ECMO patients need KRT; AKI on ECMO has been associated with a four-fold increase in mortality
CRRT machines expect venous pressures of 0 to 20 mmHg. The ECMO circuit runs from below −100 mmHg pre-pump to above +300 mmHg post-pump
Systemic ECMO anticoagulation usually protects the KRT circuit too; if ECMO is run without it, anticoagulate the KRT circuit with citrate
Early on, capillary leak means poor tolerance of fluid removal; cannula "chattering" means you are ultrafiltrating too fast

Koyner Ch 35, Figures 35.1–35.2 · Tymowski CD, Augustin P, Houissa H, et al. CRRT connected to ECMO: managing high pressures. ASAIO J. 2017;63:48–52

Poisoning: which toxins, and why

Dialysable = low molecular weight + volume of distribution <1 L/kg + protein binding <80% + low endogenous clearance
Extracorporeal clearance should add at least ~30% to total clearance to be worth doing. Intermittent HD is preferred over CRRT — CRRT clears roughly 80% less — with CRRT added afterwards where rebound is expected.

Poison	Dialyse when	Stop when
Methanol / ethylene glycol	Coma, seizures, new visual deficit, pH ≤7.15, persistent acidosis despite antidote, anion gap >24; or methanol >50 mg/dL with no alcohol dehydrogenase blocker (>60 on ethanol, >70 on fomepizole)	Methanol <20 mg/dL with clinical improvement; continue fomepizole and folate throughout
Salicylate	Altered mental status, hypoxaemia needing oxygen, level >100 mg/dL (>90 with kidney impairment); consider at pH ≤7.20	At least 6 h, or level <19 mg/dL; keep the bicarbonate infusion running between sessions
Lithium	Reduced consciousness, seizures or dysrhythmia at any level; or impaired kidney function with level >4.0 mmol/L; consider if >5.0 or if time to <1.0 exceeds 36 h	Level <1.0 mmol/L — then recheck for 12 h; add CRRT to blunt intracellular rebound
Valproate	Cerebral oedema or shock; level >1,300 mg/L. Consider if coma needing ventilation, level >900 mg/L, hyperammonaemia, or pH <7.10	Clinical improvement or level 50–100 mg/L; protein binding saturates in overdose, which is why it becomes dialysable
Metformin	Shock or reduced consciousness; pH <7.0; lactate >20 mmol/L	Acidosis resolved — the acidosis, not the drug, is the target; expect rebound from the erythrocyte compartment
Do not dialyse	Tricyclic antidepressants and digoxin (large volume of distribution, heavily bound), cocaine (endogenous clearance already exceeds anything you can add)

EXTRIP Workgroup systematic reviews, summarised in Auguste BL, Juurlink DN. Koyner Ch 28, Tables 28.1–28.10 · Roberts DM et al. Crit Care Med. 2015;43:461–472 (methanol) · Decker BS et al. Clin J Am Soc Nephrol. 2015;10:875–887 (lithium)

Recovery & life after AKI

The consult does not end when the machine stops — the highest-yield intervention in this whole lecture happens after discharge.

Koyner Ch 51 · Ronco Ch 023, 028 · KDIGO 2012

Defining, predicting and permitting recovery

What recovery means

Independence from KRT is the pragmatic endpoint — it is not the same as return to baseline GFR
Acute kidney disease is the 7–90 day window in which most of the trajectory is decided; beyond 90 days it is CKD
Reassess weekly: a patient who is still anuric at 4 weeks has a very different conversation ahead

What predicts it

Rising urine output off diuretics — the strongest daily signal; our unit thresholds and the weaning ladder are in Lecture 8
Measured creatinine clearance on a timed collection confirms what urine output suggests
Against recovery: older age, pre-existing CKD, proteinuria, prolonged oliguria, ongoing sepsis, multiple insults
The furosemide stress test predicts progression; it does not cause recovery

Key point

Recovery is something you permit, not something you wait for. Avoid recurrent intradialytic hypotension, keep the dose at the low end of the target, remove nephrotoxins, and stop early — resuming KRT is far easier than undoing a delayed recovery.

Koyner Ch 51 · Ronco Ch 023, 028 · Wang Y et al. Nephrol Dial Transplant. 2018;33:1017–1024

Post-AKI syndrome: the risk you inherit

~30%of AKI survivors are readmitted with a recurrent episode

80%of survivors did not know they had had AKI

<10%referred for nephrology follow-up in one cohort; 24% among those meeting agreed criteria

1.53hazard ratio for kidney disease progression with higher post-AKI albuminuria (ASSESS-AKI)

Risk of CKD and ESKD rises with AKI stage but is present even after mild, fully "recovered" AKI
Cardiovascular consequences are real: new hypertension — plausibly salt-sensitive — and a strong association with incident heart failure
Early outpatient nephrology follow-up after dialysis-requiring AKI has been associated with lower mortality

Heung M. Koyner Ch 51, Table 51.1 · Hsu CY, Chinchilli VM, Coca S, et al. JAMA Intern Med. 2020;180:402–410 (ASSESS-AKI) · Harel Z, Wald R, Bargman JM, et al. Kidney Int. 2013;83:901–908 · Siew ED et al. Am J Nephrol. 2019;49:449–459 · KDIGO AKI 2012

The handoff — seven lines that change a life

Name the diagnosis — "KDIGO stage 3 AKI, sepsis-associated, dialysis-requiring", written in the discharge summary and said to the patient. Most survivors are never told.

Peak stage, cause and duration of KRT, plus the date it stopped — the next clinician cannot risk-stratify without these.

Nephrotoxin reconciliation — stop NSAIDs, review contrast plans, adjust or restart renally cleared drugs as function recovers. Under-dosing after recovery is as common as over-dosing during AKI.

Renin–angiotensin blockade — restart or initiate once kidney function is stable, not on the day of discharge. The data support safety with monitoring.

Creatinine and a urine albumin-to-creatinine ratio within 3 months — both, as KDIGO asks. Proteinuria is the strongest post-AKI predictor and the one everyone forgets to measure.

Sick-day rules — hold diuretics, ACE inhibitors/ARBs and metformin during vomiting, diarrhoea or febrile illness. Teach it before discharge.

A named follow-up with a date and a clinician — nephrology after severe or dialysis-requiring AKI, primary care with a clear plan after milder episodes. Blood pressure control and rehabilitation belong in the same letter.

Heung M. Koyner Ch 51, Table 51.2 · KDIGO AKI 2012, Section 3.5 · ADQI 2018 quality-of-care-in-AKI consensus, summarised in Koyner Ch 51

Cases, pitfalls & closing the arc

Three consults that use everything in this lecture — and one that uses everything in the course.

Koyner Ch 26, 30, 33, 51

Case 1 · The circuit that will not survive the shift

62 M, septic shock, day 3. Post-dilution CVVH at 25 mL/kg/h on systemic heparin through a 20 cm right femoral catheter. This is the third filter in 18 hours.

Weight 70 kgBlood flow 120 mL/minHct 34%Post-dilution 1,750 mL/hTMP 280 → 340 mmHgaPTT 52 sPlt 96 → 41 ×10⁹/L

Two separate things are wrong here. Name both, and fix them in the right order.

Think 60 seconds · answer on the next slide

Case 1 — fix the physics before the pharmacology

Drop the filtration fractionMove fluid to pre-dilution and shift most of the dose to dialysate — CVVHDF instead of post-dilution CVVH. Raise blood flow if the access tolerates it.

→

Fix the accessA 20 cm femoral catheter recirculates ~10%, more if the lines are reversed. Reposition, replace, or move to a longer catheter.

→

Score the 4TsPlatelets down >50% at day 5 on heparin. Stop all heparin, send HIT assays, do not give platelets.

→

Change anticoagulationRegional citrate is first line here and covers both problems. Argatroban if citrate is contraindicated.

Key point

A circuit that clots on adequate anticoagulation is almost never an anticoagulation problem. Work outwards: access, then flows, then filtration fraction, then the drug.

Koyner Ch 33 · Ronco Ch 154, 168 · KDIGO AKI 2012, Section 5.5

Case 2 · Ventilator-associated pneumonia that is not responding

68 M, 80 kg, CVVHDF running since admission. Pseudomonas aeruginosa from a tracheal aspirate, meropenem MIC 2 mg/L. The admitting team wrote "renal failure" doses on day 0 and nothing has changed since. Day 4, still febrile.

Effluent 2,000 mL/hMeropenem 500 mg q12hVancomycin 1 g q24h, no loadVanc trough 8.4 mg/LUO 60 mL/dayTemp 38.9 °CWBC 22

Convert the effluent rate into a clearance. Now judge these two orders.

Hands up — how many would have changed the antibiotic instead?

Case 3 · Hour 20 of citrate in acute-on-chronic liver failure

54 F, alcohol-related cirrhosis with acute decompensation, noradrenaline 0.35 µg/kg/min, lactate 7.8 mmol/L. CVVHD with regional citrate. The nurse has increased the calcium infusion twice overnight because the ionised calcium keeps drifting down.

Systemic iCa 0.86 mmol/LTotal Ca 2.35 mmol/LpH 7.24HCO₃ 17Anion gap 22Post-filter iCa 0.28Na 141

Calculate the total-to-ionised calcium ratio. What is happening, and what do you change first?

Think 45 seconds

Quick poll

Day 12 of CRRT. Vasopressors stopped 48 hours ago, sepsis source controlled, urine output 950 mL/day off diuretics, creatinine plateauing, potassium and bicarbonate normal.

A. Increase the effluent dose to 30 mL/kg/h to "clear him properly" · B. Stop CRRT, observe, and book creatinine plus urine albumin-to-creatinine at 3 months with a named clinician · C. Start a furosemide infusion to prove recovery · D. Convert to thrice-weekly intermittent HD for another two weeks

Hands up — then say which answer changes this patient's outcome at one year

Closing the arc: nine lectures, one way of thinking

Recognise (L1, L4)

Stage it with KDIGO, distrust creatinine's lag, and use biomarkers and imaging to separate haemodynamic from structural injury.

Perfuse and protect (L2, L3)

Read both sides of the perfusion gradient, test before you give fluid, choose the vasopressor for the phenotype, and remove the nephrotoxin.

Correct the milieu (L5, L6, L7)

Sodium, potassium, calcium, phosphate, magnesium and acid–base are the daily work — and the commonest reason a patient needs the machine at all.

Support and stop (L8, L9)

Get access, choose the modality, start when the patient — not the number — demands it. Then prescribe a dose, defend the circuit, dose the drugs, feed the patient, and stop as early as you safely can.

The through-line

Every lecture answered the same question in a different setting: what is this kidney being asked to do, and what is stopping it? Machines are the last answer to that question, never the first.

Critical Care Nephrology two-week intensive, Lectures 1–9 · Koyner, Handbook of Critical Care Nephrology, 2021

Key takeaways

Effluent flow is small-solute clearance. Target 20–25 mL/kg/h delivered; prescribe 25–30 or defend the uptime — but audit the delivered number every morning.
ATN and RENAL closed the dose question: more is not better, and higher doses may delay recovery. Escalate dose for hypercatabolism or poisoning, never for sepsis alone.
Regional citrate is the KDIGO first-line anticoagulant. Post-filter ionised calcium 0.25–0.35, systemic 1.0–1.2, and a total-to-ionised calcium ratio above 2.5 means accumulation — reduce citrate and raise effluent, do not give more calcium.
A circuit that clots on good anticoagulation has a physics problem: calculate the filtration fraction, keep it under 20–25%, and fix the access first.
Your patient's drug-clearing "GFR" is about 30 mL/min. Load fully, adjust the interval, extend the β-lactam infusion, and let therapeutic drug monitoring override any table — including ours.
Blood purification for sepsis remains unproven; plasma exchange and EXTRIP indications are narrow and specific. Know which list a device belongs on before you connect it.
Stop early, then hand off properly: named diagnosis, nephrotoxin reconciliation, sick-day rules, and creatinine plus proteinuria within three months with a named clinician.

References & further reading

Where to go deeper

Palevsky PM, Zhang JH, O'Connor TZ, et al; VA/NIH Acute Renal Failure Trial Network. Intensity of renal support in critically ill patients with acute kidney injury. N Engl J Med. 2008;359:7–20.
Bellomo R, Cass A, Cole L, et al; RENAL Replacement Therapy Study Investigators. Intensity of continuous renal-replacement therapy in critically ill patients. N Engl J Med. 2009;361:1627–1638.
Wang Y, Gallagher M, Li Q, et al. Renal replacement therapy intensity for acute kidney injury and recovery to dialysis independence: a systematic review and individual patient data meta-analysis. Nephrol Dial Transplant. 2018;33:1017–1024.
Bai M, Zhou M, He L, et al. Citrate versus heparin anticoagulation for continuous renal replacement therapy: an updated meta-analysis of RCTs. Intensive Care Med. 2015;41:2098–2110.
Morabito S, Pistolesi V, Tritapepe L, Fiaccadori E. Regional citrate anticoagulation for RRTs in critically ill patients with AKI. Clin J Am Soc Nephrol. 2014;9:2173–2188.
Roberts DM, Roberts JA, Roberts MS, et al. Variability of antibiotic concentrations in critically ill patients receiving continuous renal replacement therapy: a multicentre pharmacokinetic study. Crit Care Med. 2012;40:1523–1528.
Dellinger RP, Bagshaw SM, Antonelli M, et al. Effect of targeted polymyxin B hemoperfusion on 28-day mortality in patients with septic shock and elevated endotoxin level (EUPHRATES). JAMA. 2018;320:1455–1463.
Joannes-Boyau O, Honoré PM, Perez P, et al. High-volume versus standard-volume haemofiltration for septic shock patients with acute kidney injury (IVOIRE). Intensive Care Med. 2013;39:1535–1546.
Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl. 2012;2:1–138 (Sections 3.5 and 5). · Koyner JL, Topf JM, Lerma EV, eds. Handbook of Critical Care Nephrology. Wolters Kluwer; 2021 (Ch 15, 26, 28, 30, 33–36, 51). · EXTRIP Workgroup recommendations, extrip-workgroup.org.

Critical Care Nephrology · Two-Week Intensive · Lecture 9 of 9

Thank you

Nine lectures, two weeks, one habit: ask what the kidney is being asked to do — then be the person on the round who knows the number.

NTUH Yunlin Branch Nephrology · Based on Koyner, Handbook of Critical Care Nephrology (2021) and Ronco, Critical Care Nephrology