Chapter 1 — Physical Principles & Mechanisms

CRRT is applied physics before it is applied medicine. Almost every bedside decision in the chapters that follow resolves back to the four mechanisms and three variables described here.

1.1 The Four Mechanisms of Clearance

Diffusion

Solute moves down its concentration gradient across a semipermeable membrane, following Fick's law:

J ∝ (D × A × ΔC) / Δx

D = diffusion coefficient (inversely related to molecular size) · A = surface area · ΔC = concentration gradient · Δx = membrane thickness

Two consequences matter clinically:

1. Efficiency falls steeply with molecular weight. Superb for urea and potassium, mediocre for vancomycin, effectively useless for β₂-microglobulin.
2. In CRRT, diffusion is dialysate-limited. Because Qd (1–2.5 L/h) is an order of magnitude below Qb (6–15 L/h), dialysate leaving the filter is essentially fully saturated with small solutes.

💡 Clinical pearl: In CVVHD, small-solute clearance ≈ Qd. Raising blood flow from 150 to 250 mL/min will not meaningfully increase urea clearance — it improves circuit longevity and filtration fraction, not dose. This is the opposite of intermittent HD, where Qd is high and clearance is blood-flow limited.

Convection

Water is driven across the membrane by a hydrostatic gradient and solute is dragged with it — solvent drag. No concentration gradient required.

Convective clearance = Q_uf × Sc

Because the water carries any solute that fits through the pore, convective efficiency declines far more gently with molecular weight. This is the entire rationale for convective modalities when middle molecules are the target.

Adsorption

Solutes bind the membrane polymer itself via hydrophobic and electrostatic interaction. Cytokines, peptides, endotoxin fragments and several drugs are adsorbed. Higher-adsorptive polymers include AN69, surface-treated AN69, PMMA and polyethersulfone.

Two defining properties:

• Saturable. Binding sites fill, typically within the first hours of filter life. Adsorptive clearance is front-loaded and decays.
• Invisible to dose calculation. Adsorbed solute never appears in the effluent, so it is not counted in mL/kg/h.

⚠️ Declared evidence gap. Adsorption is a real kinetic phenomenon. Whether adsorption-optimised membranes or sorbent cartridges improve patient-centred outcomes in sepsis or septic AKI is not established — the trial base is small, heterogeneous, and dominated by surrogate endpoints (cytokine levels, vasopressor dose). Do not present adsorption as an outcome-proven therapy.

Ultrafiltration

Q_uf = K_UF × TMP × surface area

K_UF = ultrafiltration coefficient (mL/h/mmHg/m²)

Only net UF changes fluid balance. A circuit removing 2,000 mL/h with 1,900 mL/h replaced delivers a net UF of 100 mL/h.

1.2 Mechanism-to-Solute Mapping

💡 Clinical pearl: "Diffusion or convection?" is really "what size is my target solute?" Below ~500 Da the two are near-equivalent and diffusion is cheaper and more circuit-friendly. Above ~5,000 Da, only convection (± adsorption) is operative.

1.3 The Three Variables That Govern the Circuit

Transmembrane Pressure (TMP)

TMP = [(P_blood-in + P_blood-out) / 2] − P_effluent

Some devices apply an additional oncotic pressure correction term.

The exact formula and alarm thresholds are device-specific. Working ceilings around 250–300 mmHg are commonly quoted, but the authoritative source is the machine's own specification, not a remembered number. TMP's diagnostic value lies in its trend:

Sieving Coefficient (Sc)

Sc = C_ultrafiltrate / C_plasma, more rigorously Sc = C_uf / [(C_in + C_out) / 2]

Sc = 1 → passes as freely as water · Sc = 0 → complete rejection

Determined by:

• Molecular radius vs. pore-size distribution — the dominant factor
• Protein binding — only the unbound fraction sieves. The single most important principle in Chapter 7.
• Electrical charge — see Gibbs–Donnan
• Concentration polarisation — a secondary protein layer builds at the blood–membrane interface, functionally narrowing pores and reducing Sc for middle molecules over filter life even while TMP still looks acceptable

Filtration Fraction (FF)

FF = Q_uf / plasma flow, where plasma flow = Qb × (1 − Hct)

High FF means blood leaving the fibre is markedly haemoconcentrated — rising haematocrit, rising viscosity, protein layering, accelerated clotting.

📐 Convention: keep FF below 20–25%. This threshold is derived from membrane physiology and expert consensus. No randomised trial has established an optimal filtration fraction. Treat it as a rational engineering constraint, not a guideline recommendation.

Three levers reduce FF: increase Qb, reduce Q_uf, or shift replacement fluid to pre-dilution. Developed fully in Chapter 4.

1.4 The Gibbs–Donnan Effect

Plasma proteins — principally albumin — are non-diffusible and carry a net negative charge at physiological pH. Retained in the blood compartment, they electrostatically hold cations back and repel anions forward.

Magnitude roughly 5% — the sodium sieving coefficient is approximately 0.94–0.96, not 1.0.

💡 Why it matters:

1. Sodium is not freely sieved. Effluent sodium slightly underestimates plasma sodium — relevant when correcting dysnatraemias.

2. Divalent cations are affected more than monovalent. This contributes to the offset between pre- and post-filter ionised calcium, the measurement backbone of regional citrate anticoagulation (Ch. 5).

3. It is a small effect. It should refine your interpretation of numbers, not dominate your prescription.

1.5 Membrane Properties & Backfiltration

• K_UF defines water permeability and separates high-flux from low-flux membranes. CRRT uses high-flux membranes exclusively.
• Retention onset vs. cut-off: conventional high-flux CRRT membranes are engineered to retain albumin. High cut-off (HCO) and medium cut-off membranes shift the curve upward to clear larger middle molecules — at the explicit cost of albumin loss. Whether that trade-off benefits patients remains unresolved.
• Backfiltration: along the fibre, the local pressure gradient can reverse, allowing dialysate to move into the blood compartment. This is why CRRT fluids must be sterile and pyrogen-free — the dialysate compartment is not hermetically separated.

1.6 Guideline Anchoring

KDIGO 2012 AKI, Section 5:

• Use CRRT rather than intermittent RRT in haemodynamically unstable patients. (2B)

• Use CRRT rather than intermittent RRT in acute brain injury or other causes of raised intracranial pressure or generalised cerebral oedema. (2B)

⚠️ Declared literature gap: KDIGO and ADQI standardise nomenclature and dose definition, not membrane physics. There is no high-quality evidence linking any membrane physical property — K_UF, polymer chemistry, pore-size distribution — to survival in critically ill AKI. Chapter 1 is mechanism, not mandate.

1.7 The Five Sentences That Carry Forward

1. Diffusion clears small solutes and is dialysate-limited in CRRT.
2. Convection clears middle molecules at Q_uf × Sc.
3. Adsorption is real, saturable, uncounted in dose — and unproven for outcomes.
4. Filtration fraction is the master variable for circuit longevity.
5. Only unbound solute sieves.