Two jobs, two circuits
VV ECMO replaces the lung. Blood leaves a vein, crosses the membrane, and returns to a vein. The native heart still provides all cardiac output. You are buying gas exchange while the lungs rest.
VA ECMO replaces the heart — and usually the lung with it. Blood is drained from a vein and returned to an artery, so the pump contributes to MAP and systemic flow. That is why VA creates problems VV does not: North–South (differential hypoxia), LV distension, and limb ischemia distal to a femoral arterial cannula.
ELSO names the circuit drainage then return. Common adult setups:
- VV fem–IJ — femoral drainage, right internal jugular return (or the reverse). Two single-lumen cannulas.
- VV Avalon (DLC) — one right-IJ shaft: drains SVC and IVC, returns a jet aimed at the tricuspid valve.
- VA fem–fem — femoral vein drain, femoral artery return. Retrograde aortic flow meets native LV ejection.
- V-AV — one venous drain, two returns (artery + vein). Used when VA is already in and the recovering heart plus bad lungs produce a blue upper body.
- V-VA — VV that adds an arterial limb when the right (or both) ventricles fail. Not the same as VV-A (an extra venous drain on VA).
On every hybrid the pump sets total flow. A Hoffman / gate clamp on the offshoot sets the split. The console has no “percent to the artery” menu.
What the Cardiohelp actually shows
This lab is modeled on Getinge’s Cardiohelp with an HLS Set Advanced — the compact centrifugal console used in most of these cases. The 5.7″ screen and the HLS module give you speed, measured flow, three pressures, a calculated ΔP, venous optical sats, and temperatures. They do not print recirculation percent, Qeff, or a diagnosis.
HLS 5.0 is rated about 0.5–5 L/min; HLS 7.0 about 0.5–7 L/min. A rotary knob sets RPM (speed). Flow is what the head achieves against venous return and resistance. If you spin and flow does not rise, you are drainage-limited — not under-pumped.
| Readout | What it is | How you use it |
|---|---|---|
| RPM | Pump speed you set | The only speed control. Not flow. |
| Flow (Q) | Ultrasonic measured L/min | Escalate a drop > ~0.5 L at the same RPM. |
| Pven | Drainage-line pressure | Modestly negative is normal. Very negative = suction / empty well. |
| Pint | Pre-oxygenator (internal) pressure | Rises with flow, afterload, or a clogging membrane. |
| Part | Post-oxygenator pressure | Return-side pressure. Track with Pint. |
| ΔP | Pint − Part (calculated) | A climbing ΔP with falling flow is a membrane problem. |
| Pre-oxy SvO₂ / Hb | Optical venous probe on the HLS | High SvO₂ + low patient sat is recirculation until proven otherwise. |
| Tven / Tart | Blood temperature | Fever raises VO₂; cooling is a real lever. |
Alfred-style alarm hygiene (useful even in the lab): flow ±0.5 L from the current value; Pven about 20 mmHg more negative than the working number; Pint/Part about 50 mmHg above; ΔP about 5 mmHg above the current gradient. The Cardiohelp also has an emergency drive (hand crank), a bubble detector, and ≥90 minutes of battery — know the crank before the head goes silent.
What each control does
- RPM
- Raises or lowers intended flow. On VV, titrate toward roughly 50–80% of estimated cardiac output (often ~60% when the native lung is contributing almost nothing). On VA, match metabolic demand and MAP — not a vanity flow. If Pven crashes or the circuit chatters, ease RPM and fix preload or the cannula.
- FdO₂
- Blender oxygen to the membrane — the oxygenator’s FiO₂, not the ventilator’s. It sets post-membrane PO₂. Turning it to 1.0 does nothing if the blood is recirculating, the membrane is dead, or effective flow is a fraction of cardiac output.
- Sweep (fresh gas flow)
- Gas across the membrane. This is minute ventilation for the circuit: more sweep, lower PaCO₂. It does not raise saturation. A common start is sweep ≈ 50–100% of blood flow (often 2 L/min and up). Do not dump PaCO₂ in minutes — cerebral perfusion cares. A rising PaCO₂ with a stable sat is a sweep problem.
- Ventilator / PEEP
- Still the patient’s lungs. Ultra-protective settings (often 4–6 mL/kg, Pplat ≤25, PEEP around 10) rest the parenchyma. On VA, recruitment and PEEP change the mixing point that produces North–South. Resting the lung is not abandoning it.
- Fluids / diuresis / transfusion
- Preload fills the drainage well. A dry patient chatters. Volume overload wrecks the RV and the lung. Hb is a DO₂ term: most adults tolerate ≥7 g/dL; ischemia or a DO₂ crisis may justify a higher target.
- Pressors / inotropes
- On VA they compete with the pump. More inotrope can shove the mixing cloud distally (worse North–South) and load the LV. Treat the physiology, not the MAP in isolation.
- Hoffman / gate clamp
- Hybrid only — a manual restrictor on the offshoot, not a pump menu. Closing the venous return pushes flow to the artery; opening it steals arterial flow. Keep each limb ≥ about 1 L/min (ultrasonic flow probe on the real circuit) to avoid stasis.
Oxygenation is a ratio. CO₂ is sweep.
On VV, the patient’s arterial sat is a mixture of ECMO-oxygenated blood and whatever the native lung (and recirculation) leave behind. If the lungs add nothing, sat tracks effective ECMO flow / cardiac output. That is why 2.8 L at FdO₂ 0.6 on a 5 L heart still looks like ARDS — the settings are simply too low (case: freshly cannulated).
DO₂ also falls when Hb is low or VO₂ is high (fever, agitation, shivering). Night desaturation with unchanged flow is often demand, not a new cannula problem. Transfuse, cool, sedate — then spin.
CO₂ clearance is almost all sweep. Pink sats and a pH of 7.15 with PaCO₂ 80 is not an FdO₂ failure. Raise sweep; do not exchange a working membrane.
Recirculation is a pattern, not a percent
Some returned blood is always re-drained on VV. Pumps and HLS modules do not calculate the fraction. The textbook formula
Recirc % = (Spre − SvO₂) / (Spost − SvO₂) × 100
needs a true mixed-venous sat from before the drainage cannula — which you do not have on the console. Spre (pre-oxy) approaching Spost with a still-blue patient is the bedside equivalent of “the oxygenator is seeing its own exhaust.”
Treat it as recirculation when:
- Patient SpO₂ is low and pre-oxy SvO₂ is high (often above 80–85%).
- Raising RPM makes sats worse or unchanged — more short-circuit, not more DO₂.
- Post-membrane gas is fine. The membrane works; the jet is in the wrong place.
Fix distance and direction: pull the return off the drainage jet, rotate an Avalon so the jet faces the tricuspid, ease RPM, add a second drainage cannula if anatomy will not cooperate. Do not max FdO₂ and call it treated.
Drainage insufficiency, chatter, and Pven
The centrifugal head can only pump what the venous well provides. Chatter is the tubing flutter of an empty atrium or a cannula against a wall. Pven dives (often past −50 to −80). Flow falls at the same RPM.
First move is usually down on RPM, then volume, then look for abdominal hypertension, kinks, or malposition. Spinning faster on a chattering circuit shears red cells (rising plasma-free Hb, dark urine) and still does not raise effective flow.
When the membrane is the problem
A working HLS oxygenator adds O₂ and removes CO₂ with a stable, modest ΔP. Suspect thrombosis or failure when ΔP climbs as flow slips, or when a post-membrane ABG shows the blood leaving the oxygenator is still hypoxic / hypercapnic.
Pre/post gases separate “patient problem” from “circuit problem.” Sweep tricks do not fix a clotted membrane. Plan the exchange before the gradient becomes a crisis. A stable ΔP with bad patient sats is usually recirculation, low flow, or native-lung shunt — not an oxygenator swap.
Air in the circuit, and a silent pump
Air is a clamp-and-clear emergency. On VA, clamp the return first so arterial air does not reach the brain; Trendelenburg; de-air or exchange the set; then find the entrainment site (loose connection, empty venous well, open stopcock). The Cardiohelp bubble detector is a backup, not a plan.
If the pump head stops (console failure, power, decoupling), you have minutes. Hand-crank / emergency drive to keep forward flow. On VA, a stopped pump can allow retrograde aortic flow down the arterial cannula — clamp strategy matters during the switch. Practice this before night call.
Accidental decannulation is hemorrhage plus loss of support. Hold pressure, clamp what is left, call for blood and a surgeon, and do not lose the airway while you stare at the floor. Prevention is two-person turns and secured lines.
VA-specific failure modes
Inadequate flow
Cardiogenic shock on 2 L of VA is still shock. Raise RPM/flow, support volume, and watch the LV — more afterload can hide in a “better” MAP.
North–South (Harlequin)
Femoral arterial return is retrograde. Recovering LV ejects poorly oxygenated blood into the coronaries and arch; ECMO blood supplies the lower body. Detect it with a right-hand (or right-ear) sat and a right-radial ABG — not a femoral sat. Legs can be 98% while the brain is 80%.
First: fix the lung (FiO₂, PEEP, recruitment, diuresis, bronchoscopy). Increasing VA flow to shove the mixing point up is often disappointing and loads the LV. If the heart is recovering but the lungs are not, convert toward V-AV (RIJ venous return + Hoffman clamp) rather than just spinning.
LV distension
VA afterload plus a closed aortic valve = a still, thrombosing LV and pulmonary edema. Pulse pressure shrinking is the bedside clue. Unload (inotrope, Impella / atrial septostomy / vent — in this lab, “LV unload”), and do not chase a high MAP with more pressor.
Limb ischemia
The femoral arterial cannula occupies the vessel. A cool, mottled foot needs a distal perfusion catheter now, not after the CK peaks. NIRS / Doppler if you have them; compare both feet every time you walk in.
Tamponade on VA
Equalization can be masked by circuit flow. Post-cannulation or post-arrest with a vanishing pulse pressure and rising CVP — think echo, not just more RPM.
RV failure on VV
VV does not unload the right ventricle. Hypoxic vasoconstriction, high PEEP, and acidosis still tax the RV. Protect it with the ventilator and vasoactives; if the heart is the new problem, that is the indication to add an arterial limb (V-VA).
Avalon Elite dual-lumen cannula
Getinge’s Avalon Elite is a single right-IJ, kink-protected bicaval cannula. Two drainage lumens pull from the SVC and IVC; the return jet should aim at the tricuspid valve so oxygenated blood crosses into the RV instead of re-entering the IVC port. Depth and rotation both matter (CFD and clinical series). There is no femoral pair — both lines on the diagram belong at the right neck.
- Jet off the TV (malrotation) looks exactly like fem–fem recirculation: high pre-oxy, low patient sat. Echo, then rotate; do not spin.
- Migration after sitting — chatter, ugly Pven, tense abdomen (hepatic-vein / deep IVC seat), climbing pre-oxy. Re-measure insertion depth from the skin.
- Over-spin — even a well-aimed jet recirculates and shears as flow climbs past what one shaft can separate. Ease RPM; a vanity sat at 4300 is how you get chatter and a rising pfHb.
Hybrids and the Hoffman clamp
The offshoot is almost always the low-resistance path. Unclamped V-AV: most of the pump flow takes the venous return, arterial Q collapses, MAP falls — “the Y stole the artery.” Unclamped V-VA after conversion: the new arterial limb is high resistance, so it under-fills until you restrict the venous side.
Clamp the venous return enough that Qart supports MAP and Qven still oxygenates the upper body. Alfred and similar programs keep ≥1 L/min in each limb (often watched with a clamp-side flow probe). Over-clamp the venous limb and you recreate Harlequin plus stasis in that cannula. The lab’s slider is that clamp: open % is venous-return share, not pump speed.
Hemolysis, bleeding, infection, demand
- Hemolysis — rising plasma-free Hb, dark urine, falling Hb. Cause is shear: suction, kink, thrombosis, or ridiculous RPM. Fix the circuit; the number is a symptom.
- Bleeding / high ACT — hold or drop anticoagulation, local control, products. Short periods of lower AC at higher flow are sometimes safer than a dripping cannula site. ECMO is not a reason to ignore a surgical bleeder.
- Sepsis — the circuit does not treat infection. Fever and vasodilation raise demand; you may need more flow and a real source-control plan.
When to come off
VV wean readiness is a lung that can do the job: improving compliance and CXR, tolerable gases on modest vent settings, hemodynamics that do not require the circuit as a crutch. A wean trial turns sweep down (often to zero) while blood flow continues so the membrane does not clot, then you watch sats and CO₂. Have a recannulation plan. Do not declare victory on one blood gas.
How to work a case in this lab
Same order every time — it is the order that keeps people alive at 03:00.
- Look at the patient: sat (right hand on VA), MAP, pulse pressure, the foot, the urine.
- Look at the console: flow vs RPM, Pven, ΔP, pre-oxy SvO₂. Compare pre-oxy to patient sat before you touch FdO₂.
- Name the problem in one sentence (settings too low, recirc, sweep, suction, membrane, North–South, steal…).
- Do the matching move. Harmful clicks are the ones that treat the wrong organ.
- If you neglect hypoxemia, shock, or a dead circuit, the patient will declare — pre-arrest, then a code. ROSC needs the cause fixed plus perfusion.
Guided mode forces three decisions. Free play is the same physiology without the multiple choice. The debrief is the attending in the doorway.
