Having decided to dialyse, the work becomes a prescription — a set of numbers and choices that determine how well and how safely the circuit runs. CRRT is forgiving of the unstable patient but unforgiving of a careless prescription: get the filtration fraction or the citrate wrong and the filter clots or the calcium drifts. This chapter is the operating manual — how to write the order, read the running circuit, and fix it when it misbehaves.
Diffusion and convection: how solute leaves
Two physical processes clear solute, and the modes are just their combinations. Diffusion moves solute down a concentration gradient across the membrane into dialysate run countercurrent to the blood; it clears small solutes such as urea and potassium efficiently. Convection drags solute along with plasma water that is ultrafiltered across the membrane, replaced by a substitution fluid; it clears larger middle molecules somewhat better. CVVHD uses diffusion, CVVH uses convection, CVVHDF uses both, and SCUF removes fluid alone with negligible solute clearance. For the small solutes that matter most in AKI, convective and diffusive modes perform comparably, so the choice between them is a matter of what your unit runs well rather than a clinical advantage.
The dose, and the downtime gap
The CRRT dose is the effluent rate — the sum of dialysate, replacement fluid, and net ultrafiltration — normalised to body weight. The evidence from Chapter 13 sets the target: deliver 20 to 25 mL/kg/h, because intensive dosing above this does not improve survival and adds harm. The practical subtlety is the gap between prescribed and delivered dose. Circuits clot, filters are changed, the patient goes to imaging, alarms interrupt — and this downtime can cost 10 to 25% of the prescribed dose. So you prescribe somewhat higher, around 25 to 30 mL/kg/h, precisely so that the delivered dose lands in the target range despite the interruptions. Prescribing exactly 20 to 25 guarantees under-delivery.
Blood flow, filtration fraction, and clotting
Filter clotting is the commonest practical problem, and three settings govern it. Blood flow of roughly 100 to 200 mL/min carries the blood through the circuit fast enough to support the effluent and the citrate, and faster flow lowers the filtration fraction. That fraction — the proportion of plasma water removed as ultrafiltrate — should stay below about 20 to 25%, because beyond that the blood leaving the filter is so haemoconcentrated that it clots. Pre-dilution replacement, given before the filter, lowers the filtration fraction and prolongs circuit life, at the cost of diluting the solute reaching the membrane and so reducing efficiency, which is compensated by prescribing a higher effluent. Post-dilution is more efficient but clots sooner. Balancing flow, fraction, and dilution against the anticoagulation is the craft of keeping a filter alive.
Regional citrate: the calcium dance
Regional citrate anticoagulation, preferred for its lower bleeding and longer circuit life, works by chemistry. Citrate infused into the blood entering the circuit chelates ionised calcium, and because the clotting cascade needs calcium, the circuit is anticoagulated while the patient is not — the calcium is replaced after the filter or returned systemically. Running it means watching two calcium measurements: the circuit ionised calcium, kept deliberately low to anticoagulate, and the patient's systemic ionised calcium, kept normal. The hazard is citrate accumulation, which happens when the liver and muscle cannot metabolise the citrate load — in liver failure or shock with poor perfusion. Accumulating citrate keeps chelating calcium, so the total calcium rises while the ionised calcium falls and the total-to-ionised ratio climbs above about 2.5, accompanied by a high-anion-gap metabolic acidosis from the unmetabolised citrate anions. Confusingly, the opposite metabolic picture — a metabolic alkalosis — signals too much citrate buffer being normally metabolised to bicarbonate, not accumulation. Reading the calcium ratio and the acid-base together tells you which problem you have.
Fluid balance: where CRRT shines
The single greatest advantage of CRRT is precise, gentle fluid control. Because removal is continuous, you can prescribe an hourly net ultrafiltration to a daily goal and adjust it in real time, which is exactly what the haemodynamically fragile patient needs — the slow removal that intermittent dialysis cannot match. The discipline is to set a net removal that meets the decongestion goal without outrunning the patient's ability to refill the intravascular space, since over-aggressive ultrafiltration causes the hypotension CRRT was chosen to avoid. Net ultrafiltration is a prescription to be reviewed, not a fixed setting.
Fluids, electrolytes, and the metabolic price
Continuous clearance has a metabolic cost. CRRT efficiently removes phosphate and potassium, so hypophosphataemia and hypokalaemia are among its commonest complications — best pre-empted by using phosphate-containing solutions and adding potassium to the fluids or supplementing directly, rather than chasing the levels after they fall. The buffer in the replacement and dialysate fluids is normally bicarbonate; lactate-buffered fluids are a poor choice in liver failure or lactic acidosis, where lactate metabolism is impaired. And because the circuit cools the blood continuously, hypothermia is common and can mask fever and shivering increase metabolic demand — a blood warmer is standard. None of these is dramatic, but each is a predictable, preventable harm of running the circuit.
SLED: the pragmatic middle
Prolonged intermittent renal replacement therapy — SLED — occupies a sensible middle ground. Run over 6 to 12 hours at blood and dialysate flows lower than conventional intermittent dialysis, it achieves much of CRRT's haemodynamic tolerance through slower solute and fluid shifts, while needing less anticoagulation and freeing equipment, staff, and the patient for parts of the day. It can be scheduled overnight to allow daytime mobility and procedures. For many units, especially where CRRT resources are stretched, SLED delivers equivalent clearance with practical advantages, and the survival evidence shows no disadvantage. It is increasingly the default for the patient who is too unstable for standard intermittent dialysis but does not need the round-the-clock control of CRRT.
Drugs, and where the evidence still argues
Finally, the circuit clears drugs, and the clinically dangerous version of this is antibiotic under-dosing in the septic patient — a recurring theme — so doses must be set for the prevailing CRRT clearance rather than for anuria. As for the evidence: the dose target and the survival equivalence of modalities are firm, and the superiority of citrate over heparin for circuit life and bleeding is well supported. What remains genuinely uncertain is whether convective or diffusive clearance is better for middle molecules in any way that changes outcomes, whether specialised high-cut-off membranes add anything, and the precise net-ultrafiltration strategy that best balances decongestion against perfusion. These are the open questions worth holding lightly while delivering the well-supported basics reliably.
