💉

Chapter 21 — ECPR Cannulation

🧭

Chapter question: Compressions are running, the criteria are met, and the clock is costing about 2% mortality a minute. How does a needle get into the right vessel when every cue that normally tells you which vessel it is has stopped working?

📅

Evidence search date: 12 September 2026.

Primary sources: ECPR and Resuscitative ECMO (Shinar and Badulak), Chapter 4 — Running the ECPR Code (personnel, room setup, code choreography and the staged model), Chapter 5 — ECMO Cannulation for ECPR (the procedure in full, cannula selection, guidewires, distal perfusion cannulas) and Chapter 6 — Hybrid Cutdown Technique (Hutin and Lamhaut, the Paris prehospital method); ELSO Red Book 6th edition Chapter 32 for the ACLS modifications, cannula sizes and connection sequence, and Chapter 4 for ultrasound technique, micropuncture, wire confirmation and difficult access; and Taha Chapter 7 for the cutdown, hybrid and distal perfusion technique.

External evidence: the ELSO Registry percutaneous-versus-surgical analysis (3,575 femoro-femoral ECPR patients); SAVE-J II's real-time ultrasound analysis (443 propensity-matched pairs); two ECPR fluoroscopy studies; the randomised femoral-access literature including the UNIVERSAL trial; and four intensivist-cannulation safety series. All from structured abstracts — nothing was retrieved in full, and every row seeded from this chapter is marked Verified = No.

A note on what this chapter can and cannot be. There is no randomised evidence comparing any ECPR cannulation technique with any other. The randomised evidence in this chapter is about femoral arterial access in elective and semi-elective cardiology, in patients with a pulse. Every number that comes from an arresting patient is observational, and much of it is single-centre. The technique described here is therefore a synthesis of textbook practice and registry association, and it is labelled as such throughout.

What this chapter covers — and what it does not

This chapter owns

Deferred to

Why cannulating an arresting patient is a different procedure; the two sub-teams and the room; the staged model and why the first stage earns its keep even if ECPR never happens; the two permitted modifications to ACLS; ultrasound anatomy and the two traps; ultrasound versus fluoroscopy; the procedure step by step, including wires, dilators and cannula sizes; percutaneous versus cutdown versus hybrid; the distal perfusion cannula under compressions; going on pump and the moment compressions stop; what to do when cannulation is failing; and training and the learning curve

Who to cannulate → Chapter 20 Why time is the indication → Chapter 19 Everything after flow is established — coronary angiography, temperature, ventilation, reperfusion → Chapter 22 General VA cannulation, arterial site selection and sizing in the non-arrest patient → Chapter 12 Limb ischaemia mechanisms, NIRS thresholds and the six-step ladder → Chapter 17 The 4-hour distal perfusion deadline in context → Chapter 14 Circuit priming, air and catastrophes → Chapters 27 and 32 Decannulation → Chapter 18 Prehospital and interhospital transport logistics → Chapter 60 Simulation programme design → Chapters 88–90

21.1 Five things that make this a different procedure

Chapter 12 taught femoral cannulation on a patient with a circulation. Almost all of it still applies. What follows is what changes when the patient has no cardiac output and somebody is compressing the chest.

🧠

Physiology — why the two cues you rely on both fail at the same moment

Outside an arrest, two independent signals tell you which vessel a needle is in: the blood pulsates, and it is red. In ECPR both fail simultaneously, and the ECPR textbook says so plainly:

"Vessel pulsatility and blood oxygenation are often unreliable for identification of arterial and venous access, as both vessels may be pulsatile during CPR and the blood from both vessels may be poorly oxygenated."

Chest compressions generate a pressure wave that is transmitted into the venous system as well as the arterial, so a femoral venous puncture can return pulsatile blood. And after twenty minutes of low flow with a systemic oxygen debt, arterial blood may be as dark as venous blood.

The consequence is structural, not a matter of care: in ECPR, vessel identification is an imaging problem, not an observational one. Everything in §21.5 and §21.6 follows from this single sentence.

What is different	Why it matters
Both confirmatory cues fail at once	Pulsatility and colour are uninterpretable. Imaging is not an adjunct here; it is the only means of vessel identification
The target moves	The Red Book notes that in patients under cardiac massage, anterior vessel entry is not always achieved and the tissue to be dilated can be substantial, leading to cannula kinking, bleeding and vascular trauma — especially on the arterial side. This is the reason superstiff wires exist in this chapter (§21.7)
Every second is a measured quantity	Beyond 30 minutes of CPR, each additional minute without extracorporeal flow adds roughly 2–2.5% mortality (Chapter 19 §19.3). A manoeuvre that costs two minutes must be worth about five percentage points of survival
Failure is not a setback, it is the outcome	The Red Book states it directly: "As unsuccessful cannulation is potentially fatal, cannulators must be meticulous in their approach." There is no plan B that preserves the patient
Complications are the normal case	In 1,644 out-of-hospital ECPR patients, complications occurred in 32.7%, the commonest being bleeding — cannulation-site bleeding 16.4%, other haemorrhage 8.5%. A recent review of intensivist cannulation concludes that ECPR carries substantially higher complication rates than other ECMO cannulation, driven by technical and environmental challenges rather than by operator specialty

⚠️

Danger — the two failure modes that kill, and neither is the one people rehearse

Teams rehearse the needle. The two events that actually end the resuscitation are elsewhere.

1. The wire is in the wrong place and nobody knows. A guidewire that has taken an unintended course and is then dilated to 25 Fr produces an injury that cannot be repaired in a patient with no circulation. The Red Book's warning is specific: confirm the wires "to avoid liver lacerations or perforations." §21.5 and §21.6 exist for this.

2. Nobody calls the switch. Percutaneous access fails in roughly 8–15% of ECPR attempts (§21.11). The harm is not the failure — it is the twelve further minutes spent on a fourth attempt because no one had decided in advance who says stop. §21.11 is about making that decision before the arrest.

21.2 Two teams, one room, and the person everything revolves around

💡

Clinical pearl — the single most important organisational sentence in ECPR

The ECPR textbook's rule for the resuscitation room is a rule about attention:

"The ECPR resuscitation team should be divided into two separate and independently functioning sub-teams... Having members of either sub-team turn attention to the actions of the other team is strongly discouraged and may result in decreased effectiveness of both sub-teams."

And the corollary, which determines the geometry of the room: "All other aspects of the ECPR resuscitation revolve around the person performing the cannulation."

The commonest organisational failure in ECPR is a single team trying to do two jobs. Compressions degrade while everyone watches the wire; or the cannulator looks up at a rhythm check and loses the field. Two teams, two leaders, and neither watches the other.

ACLS sub-team	Responsibilities
Nurse Code Team Leader	Runs the code. Ensures high-quality CPR, keeps the ACLS clock, documentation
Physician Code Team Leader	Advanced airway management, supervision of ACLS, and determining ECPR eligibility (Chapter 20)
Respiratory therapist	Assists with the airway; ventilator management
Chest compressor	Manual compressions; places and manages the mechanical compression device
Medication/electricity nurse	Intravenous access, ACLS drugs, defibrillation
Pharmacist	Prepares medications

ECPR sub-team	Responsibilities
Cannulator (and optionally a second)	Vascular access, wires, imaging, cannula insertion, connection to the circuit
Wire assistant — the sterile assistant	Wire control, dilator selection and loading, and racking the wire to prevent kinking. Prepares needles, wires, dilators and cannulas; keeps wires off non-sterile surfaces; pins the wire to form the rail along which dilators and cannulas are advanced; loads the cannula onto the wire; assists with the circuit connection
ECMO specialist — perfusionist, nurse, respiratory therapist or physician	Priming and de-airing the circuit, pump and circuit maintenance, troubleshooting
Assistant / runner — the non-sterile assistant	Hands equipment to the sterile assistant; manages the machine; passes the circuit tubing at the right moment

Three practical points from the same source. Some centres use laminated role cards giving each member their position, role and responsibilities — the purpose being as much to stop people doing other people's jobs as to remind them of their own. Because most practitioners are right-handed, the default orientation is cannulation of the right femoral vessels, and the room is built around that. And centres without the staff for two full sub-teams should recognise the limitation explicitly rather than pretend it away.

💡

Clinical pearl — the prehospital team is three people

The Paris prehospital ECPR team consists of exactly three: a physician trained in ECPR implementation, an anaesthetic nurse trained in circuit priming and intensive care, and a specially trained paramedic who drives, supplies all equipment to the two gowned operators, and organises patient extraction. They join a mobile intensive care team already on scene.

The same three-person configuration is used inside and outside the hospital. A team that can run with three in a stairwell can certainly run with three in a resuscitation room, and the discipline of designing for the smaller number is worth importing.

One detail from that chapter is easy to dismiss and is not: lighting. The authors call adequate lighting an underappreciated key to successful cutdown cannulation and state that a headlamp for the cannulators is crucial. A resuscitation-room overhead light illuminates the resuscitation, not the groin.

21.3 The staged model, and why the first stage earns its keep

The ECPR textbook describes ECPR as a staged procedure. The two stages set out below are taken from its account; the third — circuit connection and initiation — is covered in §21.10.

flowchart TD
    A["ECPR activation<br/>Chapter 20 gate 2"] --> B["STAGE 1<br/>Ultrasound-guided femoral<br/>ARTERIAL and VENOUS<br/>angiocatheters"]
    B --> C{"ROSC?"}
    C -- "Yes" --> D["Stop here.<br/>You have gained a femoral<br/>arterial line and central access<br/>and lost nothing"]
    C -- "No, and criteria met" --> E["STAGE 2<br/>Exchange catheters for<br/>ECMO cannulas over wires.<br/>NO further defibrillation"]
    E --> F["STAGE 3<br/>Connect, de-air, increase RPM,<br/>unclamp — 21.10"]
    F --> G["Flow established.<br/>Compressions stop.<br/>Chapter 22"]
    style B fill:#fff3cd
    style D fill:#d1e7dd
    style E fill:#fff3cd
    style G fill:#d1e7dd

Stage 1 — the catheters that are useful whether or not ECPR happens

Stage 1 is percutaneous placement of femoral arterial and venous angiocatheters — typically a small-bore micropuncture needle and Seldinger sheath in the artery and a larger-bore sheath in the vein, both ideally under ultrasound guidance. They are the conduits through which the ECMO wires will pass in Stage 2.

They are also, immediately, two of the most useful monitors available during a cardiac arrest, and this is the argument that makes Stage 1 worth doing early rather than late.

📚

Evidence — what an intra-arrest femoral arterial line tells you

The ECPR textbook lists four uses, and each is a different question the trace answers.

1. It measures the quality of the compressions. "Transduction allows for a good metric of CPR quality with higher mean arterial pressure indicating higher quality chest compressions." Chapter 19 §19.2 gave the physiology: with a central venous pressure of 5–10 mmHg, a diastolic pressure of 25–30 mmHg approximates a coronary perfusion pressure of 20 mmHg, and human data record no ROSC where coronary perfusion pressure failed to rise above 15 mmHg.

2. It tracks a target that is moving. "As the arrest proceeds, compliance of the chest changes making the ideal vector of compression dynamic. Broken ribs can alter the ideal compression direction and the arterial line can give clues to these changes." The compression that was optimal at minute five is not necessarily optimal at minute twenty-five, and the arterial trace is the only bedside signal that says so.

3. It diagnoses pseudo-pulseless electrical activity — organised electrical activity with mechanical contraction too weak to produce a palpable pulse — which may benefit from vasopressors rather than from compressions.

4. It predicts re-arrest in a patient who achieved ROSC, as the short half-life of adrenaline elapses.

Certainty: low — this is expert practice from a textbook, not trial evidence. But none of the four claims requires a trial to be useful, and all four are free once the catheter is in.

💡

Clinical pearl — Stage 1 has no downside case

The staged model's real contribution is that it decouples the decision to get access from the decision to cannulate.

If ROSC occurs after Stage 1, the patient has a femoral arterial line and femoral venous access — both things a post-arrest patient needs anyway — and nothing has been lost. If ROSC does not occur, the hardest and slowest part of Stage 2 has already been done, and the ECMO wires pass through conduits already in the vessels.

This is the one place in Part IV where doing something early costs nothing if you turn out not to need it. Everything else in ECPR is a commitment.

Stage 2 — the exchange, and the moment defibrillation stops

Stage 2 exchanges the Stage 1 catheters for the ECMO cannulas: a guidewire through each conduit catheter, serial dilation, cannulas over the wires. If the circuit is not pre-primed, priming happens in parallel with dilation. Anticoagulation is given (§21.7).

And at this point defibrillation stops. The textbook's reasoning is twofold: by the time a patient reaches Stage 2 they have not responded to earlier shocks, so further attempts are assumed futile; and defibrillation attempts interrupt the time-critical process of cannulation.

21.4 The only two permitted modifications to ACLS

The Red Book is unusually precise here. CPR continues as directed by the code leader under standard ACLS protocols, and there are exactly two modifications.

Modification	As stated	Why
1. Defibrillation is suspended during guidewire insertion	"In the event of refractory VT/VF, defibrillation attempts are temporarily suspended during guidewire insertion until established on VA ECMO support"	Stated reason: shocks have already failed and they interrupt a time-critical process. This book adds a mechanical reason that is at least as important: a defibrillated patient moves, and a wire that is half-way up an artery when the body jerks is a dissection. That addition is reasoning, not a published finding
2. Adrenaline boluses stop when the circuit is being connected	"Discontinuing adrenaline boluses when the circuit is being connected to the cannulas (to avoid inadvertent hypertension)"	The pump is about to deliver full cardiac output into a maximally vasoconstricted circulation. The Red Book adds that the code leader should anticipate that vasopressor and inotrope infusions may need to be rapidly weaned (§21.10)

⚠️

Danger — "only two modifications" is the point, not a caveat

Everything else in the resuscitation continues unchanged: compression rate, depth, fraction, airway management, the rest of the drug protocol, the search for reversible causes.

The commonest drift in an ECPR code is that the ACLS degrades because attention has moved to the groin. Chapter 19 §19.2 gives the cost: coronary perfusion pressure builds over three or four compressions and dissipates within seconds of stopping, so a pause costs the pause plus the rebuild — and the patient is being perfused entirely by those compressions until the pump takes over.

The ACLS sub-team exists precisely so that the quality of CPR is somebody's only job while the cannula is going in.

21.5 The ultrasound — anatomy, two views, and two traps

Because vessel identification cannot be done by pulsatility or colour (§21.1), the ultrasound is not an aid to the procedure. It is the procedure's sensory apparatus.

The 2025 American Society of Echocardiography guideline on ultrasound-guided vascular cannulation sets out three distinct roles, and it is worth keeping them separate: pre-cannulation vessel assessment, dynamic guidance during cannulation, and identification of local complications. The same document is candid that although the literature increasingly indicates ultrasound improves success and reduces complications, the quality of that evidence remains weak.

The target

The common femoral artery, at the level of the femoral head, just proximal to the bifurcation into the superficial femoral and deep femoral (profunda) arteries. Pre-cannulation assessment at that level also shows calcification and peripheral arterial disease, which may change the site or the equipment before anything is punctured.

⚠️

Danger — the two anatomical traps, one at each end of the target zone

TRAP 1 — too low. Proximal to the bifurcation, the femoral vein lies medial to the common femoral artery. Distal to the bifurcation, the femoral vein lies POSTERIOR to the superficial femoral artery. The Red Book spells out the consequence: vessel entry distal to the bifurcation risks inadvertent entry through the superficial femoral artery into the femoral vein, which can be complicated by inappropriate venous cannulation and arteriovenous fistula formation.

TRAP 2 — too high. Entry above the inguinal ligament puts the puncture where compression cannot control it, and retroperitoneal haemorrhage in a fully heparinised patient on a pump is among the worst outcomes of this procedure.

The longitudinal view identifies the femoral head and shows where the common femoral artery dives into the pelvis cephalad to it. That is the landmark that keeps you out of both traps.

Transverse or longitudinal — and why the answer is both

View	What it does well	What it does badly
Transverse	Rapidly identifies the bifurcation and the vein–artery relationship; easier to perform	Less likely to achieve the anterior vessel entry that large-bore cannulation requires — and anterior entry is exactly what a 25 Fr cannula needs
Longitudinal	Shows anterior puncture, confirms intraluminal guidewire placement, identifies the femoral head, keeps you below the inguinal ligament, and approximates the trajectory of the vessel so the cannula runs along its axis	Harder to perform, particularly under time pressure and with a moving field

The Red Book's sequence is to begin transverse to find the bifurcation, then use longitudinal for the puncture itself. The transverse view answers where; the longitudinal view answers how.

Micropuncture

Standard	Micropuncture
18-gauge needle, 0.035 inch guidewire	21-gauge needle, 0.018 inch guidewire, small sheath
A wrong entry is a large hole	The small needle can be removed if vessel entry is incorrect; the small sheath allows angiographic evaluation of the entry site; if the site is unsatisfactory the sheath can be removed with easier haemostasis

🚩

Pitfall — in suspected pulmonary embolism, ultrasound is not optional

The Red Book carries a specific instruction that is easy to miss and has no substitute: in the event of pulmonary embolism, ultrasound visualisation is mandatory to avoid inadvertent cannulation of a thrombosed vein.

Chapter 20 §20.5.8 identified pulmonary embolism as one of the highest-yield ECPR aetiologies — 19.3% favourable neurological outcome in the SAVE-J II pulseless-electrical-activity analysis, the best of any cause in that subgroup. It would be a particular waste to lose one of those patients to a drainage cannula placed into an occluded iliofemoral segment.

If the working diagnosis is pulmonary embolism, look at the vein before you puncture it, and look at the contralateral one too.

21.6 Controversy 1 — ultrasound, fluoroscopy, or both?

⚖️

Controversy 1 — Which imaging should guide ECPR cannulation, and is it worth moving the patient to get it?

The question. Ultrasound is portable, free of radiation and available at the bedside. Fluoroscopy shows where the wire has actually gone. The ECPR literature contains studies favouring each, and one of them requires transporting an arresting patient to another room.

What the randomised evidence says — and all of it is in patients with a pulse.

— A 2020 meta-analysis of 7 randomised trials (3,180 patients) comparing ultrasound-guided with standard femoral arterial access: first-attempt success 82.0% versus 58.7% (RR 1.36, 1.17–1.57); vascular complications 1.3% versus 3.0% (RR 0.48, 0.25–0.91); access-site haematoma RR 0.41 (0.20–0.83); venepuncture 3.6% versus 12.1% (RR 0.32, 0.20–0.52); major bleeding RR 0.57 (0.24–1.32), not significant.

— A 2015 meta-analysis of 4 trials (1,422 patients): 49% reduction in overall complications (RR 0.51, 0.28–0.91) and 42% improvement in first-attempt success (RR 1.42, 1.01–2.00).

— But the largest single randomised trial, UNIVERSAL (621 patients, 2022), was null on its primary endpoint. Against a background of fluoroscopic landmarking, the composite of major bleeding or major vascular complications at 30 days was 12.9% with ultrasound versus 16.1% without (OR 0.77, 0.49–1.20, P = 0.25). Yet the mechanical findings held: first-pass success 86.6% versus 70.0% (OR 2.76, 1.85–4.12, P less than 0.001), fewer attempts (1.2 versus 1.4), venipuncture 3.1% versus 11.7% (OR 0.24, 0.12–0.50) — with similar times to access (114 versus 129 seconds, P = 0.34).

— A randomised trial of ultrasound versus fluoroscopy for femoral arterial access (635 patients): successful cannulation 93% versus 86% (P = 0.002), first-attempt 74% versus 42%, venipunctures 2% versus 10%, median time 80 versus 100 seconds — with no difference in complications at 24 hours or at 30–90 days. Among trainees the gap widened: with ultrasound their time matched consultants' (85 versus 77 seconds); with fluoroscopy they were far slower (137 versus 86 seconds).

What the ECPR-specific evidence says — all observational.

— SAVE-J II, 443 propensity-matched pairs: real-time ultrasound guidance shortened cannulation by 2.5 minutes (95% CI −3.7 to −1.3, P less than 0.001), with no difference in catheter-related complications (OR 1.51, 0.64–3.74) or in poor neurological outcome (OR 1.08, 0.83–1.59).

— Adding fluoroscopy to ultrasound, 73 ECPR patients: complications 8.7% versus 36.0% (P = 0.022), adjusted OR 0.14 (P = 0.024) — and the time from hospital arrival to extracorporeal circulation was identical, a median of 17.0 minutes in both groups.

— Moving the patient to the angiography room, 59 ECPR patients, before-and-after: vascular complications 40.6% (14/32) in the emergency department without fluoroscopy versus 10% (2/20) after transfer to the angiography room (P = 0.014), with time from emergency department arrival to ECMO initiation unchanged (median 23.0 versus 25.5 minutes, P = 0.71).

— Fluoroscopy-guided percutaneous implantation in the catheterisation laboratory, 56 patients including 12 in refractory arrest under mechanical compression: cannulas successfully implanted and the desired flow obtained in every patient, with no vessel perforation or dissection.

Resolution — they are not competitors, and the literature reads as a contradiction only because it is being asked one question instead of two.

Ultrasound answers: am I in the right vessel, at the right place, on the first pass? Every study agrees it does this, randomised and observational alike. The mechanical benefits — more first-pass success, fewer attempts, fewer venipunctures — are the most consistent finding in the whole of this literature.

Fluoroscopy answers a different question: where has the wire actually gone? That is the question behind the catastrophic complications — the liver laceration, the wrong vessel, the kinked wire dilated to 25 Fr — and it is a question ultrasound answers poorly once the wire has left the groin.

And in ECPR, the mechanical benefit is the outcome benefit, because Chapter 19 established that time is the indication. A 2.5-minute saving is worth roughly five to six percentage points of mortality on the Red Book's 2–2.5% per minute estimate. That is a larger effect than most of the complication differences being argued about.

Practical position. Use ultrasound always — it is the only vessel-identification tool that works in an arresting patient. Add fluoroscopy wherever it exists, because both ECPR studies that added it found large complication reductions at no time cost. Where the choice is between cannulating in the emergency department and moving to an angiography suite, the single before-and-after study supports moving; it is one small study, and the decision should rest on your own door-to-flow audit rather than on its odds ratio.

Certainty: moderate for the mechanical benefits of ultrasound; low for adding fluoroscopy in ECPR; very low for moving the patient.

🚩

Pitfall — one odds ratio in this section is too wide to use

The before-and-after study of transfer to the angiography room reports an adjusted odds ratio for vascular complications of 9.92, with a 95% confidence interval of 2.04 to 81.2.

An interval spanning a forty-fold range is compatible with a small effect and with an enormous one. The direction is used in this chapter; the point estimate is not. The underlying proportions — 40.6% versus 10% — are reported because they are what was observed, with the caveat that this is a before-and-after comparison in 52 analysed patients at one centre, where practice will have changed in other ways across the study period.

This is the fifth published numerical problem flagged in Part IV.

21.7 The procedure

What follows is the composite percutaneous technique as described by the ECPR textbook and the Red Book. Certainty: expert practice throughout. No step below has randomised support.

Step	Detail	Why it is done this way
Prepare the field	Clean both groins with antiseptic, with or without shaving; apply a large femoral drape. The non-sterile assistant confirms the circuit is primed, activated and free of bubbles	Both groins because the second leg may be needed (§21.11), and because arterial and venous cannulation may be contralateral
Choose the sites	See the flag below — the Red Book gives two different answers in two chapters	—
Access the vein first, then the artery	18-gauge needle with a syringe under continuous negative pressure, under ultrasound; or a micropuncture set (§21.5). A stiff J-tipped guidewire is advanced to the chest in each vessel	The vein is larger and more forgiving; and the venous cannula can be connected early (below)
Confirm both wires before anything is dilated	Fluoroscopy: wire to the left of the spine confirms aortic position, to the right of the spine confirms inferior vena cava — not reliable in congenital heart disease. Transoesophageal echocardiography: the venous wire is seen crossing the right atrium into the superior vena cava, the arterial wire in the aorta. Transthoracic echocardiography is very challenging with continuous compressions. Verify freedom of the wire tip	This is the step that prevents the unsurvivable complication. §21.1's second danger
Anticoagulate	50–100 U/kg of heparin, after wire position is confirmed	Confirmed first, because a heparinised patient with a misplaced wire is a worse problem
Nick the skin	Scalpel incision under 1 cm, over the needle — not over the wire. Dissect to the fascia with a small Kelly clamp	Damaging the wire makes advancement of dilators and cannulas difficult, and a damaged wire cannot be relied on as a rail
Dilate	Typically only 1–2 dilators are necessary. When exchanging a dilator on one wire, leave the dilator in the vessel on the other wire to limit blood loss. Hold pressure over the access site. Ensure dilators move independently of the guidewire at all times and that the wire does not kink	Every exchange is a bleeding opportunity in a heparinised patient with no clotting reserve
Rack the wire	The wire assistant alternately advances and retracts the wire slightly during dilation, and pins the wire to form the rail	Prevents kinking — a kinked wire is the commonest reason a cannula will not advance
Insert the cannulas	After the dilator equal to or slightly smaller than the cannula, the assistant loads the cannula onto the wire and the cannulator advances it. Venous tip in the mid right atrium or at the right-atrial/superior-vena-caval junction	—
Manage the first cannula while the second goes in	Either connect the venous cannula to the circuit at this point — which allows transfusion through it, displaces the blood standing in the lumen and reduces clot, and is useful if there is blood loss during a complicated arterial cannulation — or flush it with heparinised saline, or backflush it intermittently	A large-bore cannula full of static blood in a low-flow patient is a clot waiting to embolise
Secure	Multiple sutures to skin. Secure the circuit to the leg as well as the cannula	Accidental decannulation during patient movement is a named complication, and an ECPR patient is about to be moved

🚩

Pitfall — the Red Book contradicts itself on which leg

Chapter 4 states: "If possible, femoral arterial and venous cannulation should be done contralaterally to avoid placing the cannulas in the same leg, with concomitant compromise of both arterial inflow and venous return."

Chapter 32 states: "Unilateral cannulation can help localization of the second vessel to be found and punctured for guidewire insertion."

Both are printed here and neither is asserted. They are not quite answering the same question: the first is about the limb's circulation afterwards, the second about finding the second vessel quickly when visualisation is poor. This book's reading — labelled as reasoning — is that the contralateral argument is the stronger one whenever the vessels can be seen, because Chapter 17 established that the venous and arterial halves of limb ischaemia are separate problems and putting both cannulas in one leg creates both at once. The unilateral argument applies when finding the second vessel is the rate-limiting step, which is a statement about your imaging rather than about your patient.

This is the second internal contradiction found inside a single source in this book.

Cannula selection

Flow through a cannula follows Poiseuille's relation, and the term that dominates is the radius to the fourth power.

$Q = \Delta P \times \frac{\pi}{8} \times \frac{1}{\eta} \times \frac{r^4}{L}$

Blood is not Newtonian and flow in a large cannula is often turbulent, so this is a guide to relative behaviour rather than a calculation — but doubling the radius increases flow sixteen-fold, while doubling the length halves it. Chapter 3 established the principle and Chapter 12 sharpened it for the artery; Chapter 17's completion of the rule is the one that matters most here: the operative variable is the cannula-to-vessel ratio, not the absolute French size.

Cannula	Red Book Chapter 32 (ECPR)	Red Book Chapter 4 (general)
Arterial (return)	15–17 Fr, and 17–19 Fr in larger males	—
Venous (drainage)	25–27 Fr multistage; a single-stage drainage cannula is acceptable	A 25 Fr multistage cannula should suffice in most adults

One caution from the intensivist-cannulation literature: smaller arterial cannulas may reduce vascular complications, though survival is adversely affected when such complications occur — so the trade is between a complication you can often manage and a flow deficit you cannot. Chapter 17 §17.3 established that a limb complication is a systemic marker, not a local one.

💡

Clinical pearl — the superstiff wire is the answer to the moving target

The Red Book identifies exactly the two situations this chapter is about — "patients that are obese or in patients under cardiac massage" — and gives the same solution for both. In these patients anterior vessel entry is not always achieved and the tissue to be dilated is substantial, which produces cannula kinking, bleeding and vascular trauma, especially on the arterial side.

In these situations extra support wires are essential, and the described technique is:

place a small sheath over the guidewire → advance a 6 Fr JR4 or multipurpose coronary catheter over an 0.038 inch wire into the proximal descending thoracic aorta under fluoroscopy → exchange the standard wire for the extra-support wire → remove the catheter and sheath → dissect, dilate and cannulate. A similar process can be used on the venous side.

If your unit cannulates obese patients or does ECPR, a superstiff wire belongs in the pack, not in the store cupboard.

21.8 Controversy 2 — percutaneous, cutdown, or hybrid?

⚖️

Controversy 2 — Which access technique should an ECPR programme use?

The question. Three techniques are in current use, they differ by a factor of three in the time they take, and the largest dataset comparing two of them found a benefit in a place nobody predicted.

The three techniques.

— Percutaneous Seldinger. Needle, wire, sequential dilators, cannula over wire, under imaging. Reported cannulation in 6–8 minutes in some studies. Now dominant: in the ELSO Registry, 2,749 of 3,575 (77%) femoro-femoral ECPR cannulations were percutaneous, and the proportion rose from 18% to 89% between 2008 and 2019. In a systematic review of 24 ECPR protocols, percutaneous Seldinger was the preferred method in 44%.

— Surgical cutdown. Open exposure, ligatures, arteriotomy or venotomy, cannula, ligation or an end-to-side graft. Frequently requires 20–30 minutes. Similar limb ischaemia rates to percutaneous, but greater vascular control in severely diseased vessels and the ability to size the cannula to the vessel under direct vision.

— Hybrid cutdown. An incision below the inguinal ligament exposes the vessels; the needle is then inserted percutaneously distal to the incision and enters the vessel under direct visualisation; Seldinger technique proceeds from there and the cutdown is closed. About 20 minutes.

What the registry comparison found, and it is not what you would predict. Among 3,575 ELSO Registry femoro-femoral ECPR patients, percutaneous cannulation was associated with fewer severe neurological complications — 13% versus 19%, adjusted OR 0.62 (0.46–0.83), P = 0.002 — with no difference in in-hospital mortality (OR 0.93, 0.73–1.17), limb ischaemia (OR 0.84, 0.58–1.20) or cannulation-site bleeding (OR 0.90, 0.66–1.22). The comparison held across levels of centre percutaneous experience and centre ECPR volume.

Why would access technique affect the brain and not the leg? This book's reading, offered as reasoning rather than as a finding: it is a time effect wearing a neurological mask. Cutdown takes 20–30 minutes against 6–8. Chapter 19 established that the organ that pays for delay is the brain, and that the leg is injured by the cannula rather than by the clock. A technique that adds fifteen minutes of low-flow time should produce exactly this pattern — more hypoxic brain injury, unchanged vascular complications. If that reading is right, the finding is not an argument for percutaneous access as such; it is another measurement of the low-flow gradient.

The case for the hybrid technique, from the team that uses it exclusively. The Paris prehospital programme adopted it in 2011 after several failed attempts to implement ECPR percutaneously, and reports mean implementation time 21.3 minutes with a 7.6% failure rate. For comparison, the same chapter cites a median 19 minutes and a 14.5% failure rate for ultrasound-guided percutaneous access, and a study in which 8% of patients cannulated percutaneously under ultrasound required a switch to a surgical approach.

Its four stated advantages are specific and none of them is speed:

1. It needs no imaging at all — decisive in a prehospital setting, and the reason a Paris team can cannulate in a stairwell.

2. Direct visual identification of vein from artery — which matters because anatomic variations are frequent and the femoral vessels often lie directly on top of one another, and because §21.1 established that pulsatility and colour have both failed.

3. It increases the likelihood of a single needle puncture per vessel, reducing bleeding — and any bleeding that does occur is immediately visible and manageable.

4. It works in patients who are not amenable to percutaneous access — obesity, atherosclerosis.

Resolution — the technique is chosen by the environment and by what the operator does often, not by a general ranking.

Percutaneous is fastest, is now the default, and the registry association favours it. It should be the primary technique of almost every in-hospital programme.

But a programme should be competent in two of the three. The textbook's own formulation is the right one: "Having a second technique that the provider is skilled at performing is critical in these cases." The two circumstances that defeat percutaneous access — no imaging, and anatomy that will not take a blind wire — are precisely the circumstances in which the alternative has to be available immediately, not summoned.

Certainty: low. One large registry comparison with an unexplained pattern of benefit; single-programme series for the hybrid technique; no randomised comparison of any technique against any other in ECPR, and none is registered.

🚩

Pitfall — cutdown bleeding can relapse the moment the pump starts

The hybrid-technique chapter records something that applies to every open or hybrid approach and is easy to be caught by:

"Applying compression often suffices to stem the bleed. If not, vascular clamping can be necessary. In this case, cannulation of the other leg might become necessary. Although the bleeding might have stopped, relapse can occur once the pump is activated."

A groin that is dry at a mean arterial pressure of 20 mmHg generated by compressions is not necessarily dry at 65 mmHg generated by a centrifugal pump. Look again after flow is established, before the patient is moved.

The same source notes that transfusion is often required regardless, from heparinisation of the circuit and dilution of coagulation factors.

21.9 The distal perfusion cannula — Chapter 12's rule under compressions

Chapter 12 established the rule and Chapter 17 explained why it matters: the distal perfusion wire goes in before the arterial cannula, because distal flow is much lower once the arterial cannula is in place. Chapter 21 is where that rule meets a patient who has no circulation and a team with no spare minutes.

📚

Evidence — the ECPR exception is stated explicitly, and it is a deferral rather than an abandonment

Taha's chapter gives the rule and then the exception in consecutive sentences:

"It is favorable to place a wire distally prior to placement of the actual arterial cannula, since distal flow is usually considerably lower after placement of the arterial cannula. However, during an emergency situation or in case of patients with ECPR, this can be done later after the femoral artery cannulation."

That is permission to defer, not permission to omit, and the deadline is Chapter 14's: the Red Book's ECPR chapter states that where femoral cannulation has been used and a distal perfusion cannula has not already been positioned, this should be undertaken ideally within 4 hours to reduce the risk of limb ischaemia.

The technique (Taha): locate the common femoral bifurcation with ultrasound; access the superficial femoral artery antegrade, just past the bifurcation; ensure the wire passes freely down the leg; place the sheath; connect it by high-pressure tubing to the Luer lock port on the arterial cannula. Retrograde placement via the posterior tibial or dorsalis pedis artery is the alternative.

Certainty: expert practice. The 4-hour figure has no trial behind it.

Finding	Certainty
In a 56-patient percutaneous VA ECMO series including 12 refractory arrests, ipsilateral limb ischaemia occurred in 1 of 8 (13%) with an upfront perfusion sheath versus 2 of 3 (75%) without it (P = 0.15)	Very low — denominators of 8 and 3. The direction is used; the percentages are not a usable estimate
In 427 screened VA ECMO patients, arterial complications 37% (mainly ischaemia, then bleeding, dissection and compartment syndrome) and venous complications 27% (thrombosis 21%, pulmonary embolism 7%). Vascular surgery was needed in 19%; major amputation in 1%. A distal perfusion cannula was placed at cannulation in 24% and secondarily in 16%	Low — single centre, ten years, abstract only
In the same cohort, risk factors for leg ischaemia at the time of cannulation were elevated D-dimers, lower near-infrared spectroscopy on the cannulated leg, and lack of a distal perfusion cannula. The best discriminative parameter was the DIFFERENCE in regional oximetry between the non-cannulated and the cannulated leg. During ongoing support, only the lack of a distal perfusion cannula was associated with ischaemia	Low — but it independently confirms Chapter 17's bilateral-probe rule
In a 42-patient ultrasound-guided ECPR programme, distal limb ischaemia occurred in 38.1% and cannulation-site bleeding in 31.0%, all managed conservatively	Low — and a reminder that the problem is not solved by technique alone
Complications associated with decannulation were similar between percutaneous and surgical approaches (18% versus 17%, P = 0.295)	Low. Relevant to Chapter 18

💡

Clinical pearl — how to hold the rule in an ECPR code

Put the distal wire in first if it costs you nothing. In a straightforward cannulation with two operators and good imaging, the wire goes down the superficial femoral artery while the arterial access is being prepared and it costs seconds.

If it would cost a minute, defer it — and hand it to a named person with a time. Chapter 19 §19.5 made the same argument about the sweep gas: the decisions that get lost in ECPR are the ones that belong to nobody in particular. "Distal perfusion cannula, Dr X, within four hours" is a handover; "we'll do it later" is not.

And put the regional oximetry probes on both legs before you leave the resuscitation room (Chapter 17). The discriminating signal is the difference between the two, and you cannot compute a difference retrospectively from one probe.

21.10 Going on pump, and the moment compressions stop

Step	Detail
Connect	Despite the time pressure, all air is purged during connection and final checks are undertaken to avoid embolisation. The Red Book is explicit that this step is not shortened
Increase revolutions	RPM are increased gradually to generate adequate positive pressure in the return limb of the circuit before the clamps come off
Unclamp	Clamps are removed to establish antegrade ECMO flow
Stop compressions	"At which stage mechanical chest compressions can be discontinued." One practical formulation from the ECPR textbook is to continue compressions during the initial seconds of extracorporeal support until total blood flow approaches 2.5–3 L/min
Anticipate the hypertension	The code leader should anticipate that vasopressor and inotrope infusions may need to be rapidly weaned, and adrenaline boluses have already stopped (§21.4)
Take the gas	An arterial blood gas from the arterial cannula insertion, or from the right radial line. Chapter 19 §19.5 owns what happens next, and the one instruction that matters is that somebody is named and a time is set

⚠️

Danger — the two minutes after flow starts are the least supervised in the whole pathway

Everyone in the room has been working towards a single event, and the event has just happened. What follows immediately is a cluster of decisions that Chapter 19 §19.5 identified as ownerless: the oxygen fraction is 1.0, the sweep is high, the circuit is at ambient temperature, and the vasopressors are still running into a circulation that now has a pump in it.

And a second, mechanical hazard: the groin that stopped bleeding under compressions may start again at pump pressures (§21.8), and the patient is about to be moved.

Build the handover into the protocol rather than leaving it to relief. Flow established is the start of Chapter 22, not the end of Chapter 21.

21.11 When cannulation is failing

This section exists because the decision to change technique is the decision nobody makes in time.

Reported failure	Setting
7.6% failure rate	Hybrid cutdown, Paris, in-hospital and prehospital
14.5% failure rate	Ultrasound-guided percutaneous, as cited in the same chapter
8% of patients cannulated percutaneously under ultrasound required a switch to a surgical approach — a rate that fell when fluoroscopy was added	ECPR, single centre
Failure to achieve flows occurred exclusively with peripheral attempts (8 of 52); difficult cannulation in 17% of peripheral and 13% of central attempts	92 paediatric in-hospital ECPR events. Paediatric data, and Chapter 62 owns them — recorded here because the pattern is instructive
75% (15 of 20) of two-person teams succeeded after a two-hour training course, in a swine arrest model, mean time 22 minutes 10 seconds (95% CI 17:25–26:54). "The most frequent reason for unsuccessful ECPR was the inability to obtain appropriate vascular access"	Emergency medicine residents, animal model

💡

Clinical pearl — the two-attempt rule, and it does not come from the ECPR literature

A prospective five-year study of 1,794 landmark-guided central venous access procedures by experienced operators found that more than one cannulation attempt was a risk factor for failed catheterisation and for other mechanical complications, and concluded:

"After two unsuccessful cannulation attempts failure and associated complications are very likely."

In that cohort, where cannulation failed at the attempted site, 35.3% of those procedures were accompanied by further complications — the failure and the injury travel together.

Nothing about that finding is specific to ECPR, which is exactly why it is usable. It is the best-evidenced stopping rule available, and it is about attempts rather than minutes, which is what a cannulator can actually count while working.

Write your switch trigger before the arrest, in the protocol, with a named person who calls it. Two failed attempts at a vessel, or a stated number of minutes, whichever comes first — and then the alternative technique, the other groin, or the second operator.

Problem	Response
Cannot identify vein from artery	This is §21.1's problem. Do not resolve it by colour or pulsatility. Reposition the probe, use the longitudinal view, confirm with fluoroscopy or transoesophageal echocardiography. If imaging cannot answer it, the cutdown answers it by looking (§21.8)
Wire will not advance	Suspect a kink. Rack the wire. Do not force a dilator over a wire that is not moving independently. Consider a superstiff wire via the exchange technique in §21.7
Cannula will not advance despite a good wire	Usually non-anterior vessel entry or substantial tissue — the moving-target problem of §21.1. Superstiff wire; consider hybrid cutdown
Bleeding at the site	Compression first. If compression fails, vascular clamping — and then the other leg may be needed. Expect relapse when the pump starts (§21.8). Expect to transfuse
Two failed attempts at a vessel	Switch — technique, side, or operator. The named person calls it
Vascular injury recognised during cannulation	Taha's rule: "Attempting surgical repair and completing the cannulation immediately are imperative; otherwise, failure is likely." The injury does not mean the resuscitation stops
Inadequate venous drainage after flow is established	In one 42-patient ECPR series, 7.1% (3 of 42) required conversion to veno-veno-arterial ECMO for inadequate venous drainage. Chapter 3 owns the configuration; the point here is that it is a foreseeable step, not a rescue

21.12 Training, competence, and the learning curve

Chapter 19 §19.9 established that the unit of intervention in ECPR is the programme. Chapter 20 §20.9 showed that the criteria list is the dial between a unit's survival figure and its competence. This section is the third side of the same argument: the cannulation itself is a trainable skill with a measurable learning curve, and the published safety data are reassuring in a specific and conditional way.

Series	Findings
Intensivist-led cannulation programme, 4 years, quaternary centre	402 cannulations in 194 episodes involving 179 patients, including 69 veno-arterial initiations of which 36 were ECPR. 394 of 400 successful (98.5%). 32 complication events (7.96%): 15 low significance (3.7%), 10 medium (2.5%), 7 high clinical significance (1.7%)
Intensivist-led cannulation, tertiary academic centre, 2019–2024	213 cannulations in 106 episodes, 99 patients, including 4 ECPR. 9 complications (4.23%), of which 6 high significance (2.82%) — four of those were cannulation-related bleeding requiring at least two units of red cells
Correspondence on a systematic review of intensivist cannulation	Intensivist cannulation is generally safe and feasible when supported by structured training, credentialing, and immediate surgical backup for complications such as vascular injury. Veno-venous carries relatively low complication rates; veno-arterial carries higher risks; and ECPR is characterised by substantially higher complication rates, likely driven by technical and environmental challenges rather than operator specialty
A 720-patient single-centre experience, 159 percutaneous VA	Vascular complications or limb ischaemia leading to surgical revision in 16.9%; blood loss and cannula relocation were the other main problems; Harlequin syndrome in 8.8%; cannulation failure and malfunction were infrequent. The authors' conclusion: "As lack of experience is the trigger of many complications, adequate training of cannulation techniques is essential."

📚

Evidence — a published training pathway, with a number attached

The Paris hybrid-cutdown team describes a four-step pathway, and it is the only one in this chapter's sources that specifies a threshold for independent practice.

1. A gel pelvis model for the first steps — location and sequence.

2. Cadavers, ideally perfused — "perfused cadavers make the cannulation process even more realistic... the advantage of locating the vessels and seeing/feeling the difference between the vein and the artery."

3. Real-time simulation, which the authors call crucial, because "at this point the cannulator sees how and why cannulation can fail. For example, the trainee recognizes the importance of a ready guidewire for the speed and success of a one stick insertion."

4. Proctored practice: "a new cannulator must assist at least 5 procedures and do 5 proctored procedures before being able to cannulate alone."

Certainty: expert practice, single programme. No comparative evidence supports 5 and 5 over any other number. It is nonetheless the only published threshold, and a programme with no threshold at all is worse off than one that borrows this.

💡

Clinical pearl — the Vienna curve, and what it honestly contains

A programme that restructured itself between 2020 and 2023 reported favourable neurological outcome across 192 ECPR patients rising year on year: 15% (5/34), 19% (8/42), 23% (12/53), 37% (23/63) — and in out-of-hospital arrest specifically, 7%, 14%, 17%, 32%.

The authors do not claim this as a cannulation effect, and they should not. Over the same period the rates of witnessed arrest, bystander CPR and initial shockable rhythm all increased, and low-flow durations fell. Their own summary is that the improvement is "a summation effect of training, patient selection, and process standardisation."

That honesty is the point. It is the same loop Chapter 20 §20.9 described: a maturing programme simultaneously gets better at the procedure and more selective about who receives it, and the published survival figure moves for both reasons. A unit that watches its own curve rise should ask which of the two is happening, because only one of them is a skill.

21.13 The errors that recur

Error	Why it is wrong
Identifying the vessel by pulsatility or by the colour of the blood	Both fail simultaneously in an arresting patient. Compressions make veins pulsatile and oxygen debt makes arterial blood dark
Dilating before both wires are confirmed	This is the step that produces the unsurvivable complication. Confirm left-of-spine and right-of-spine, or confirm on transoesophageal echocardiography, and verify freedom of the wire tip
Entering distal to the femoral bifurcation	The femoral vein lies posterior to the superficial femoral artery there — the needle can pass through the artery into the vein, producing venous cannulation through an artery and an arteriovenous fistula
Entering above the inguinal ligament	Retroperitoneal haemorrhage that compression cannot reach, in a fully heparinised patient
Using the transverse view for the puncture	It is less likely to achieve the anterior entry that large-bore cannulation requires. Transverse to find, longitudinal to puncture
Making the skin nick over the wire	A damaged wire is a damaged rail. Nick over the needle
Forcing a dilator over a wire that will not move	The wire is kinked. Dilators must move independently of the wire at all times. Rack the wire
Removing the dilator from the second vessel while exchanging on the first	Two open tracts in a heparinised patient with no clotting reserve. Leave the dilator in
Leaving the first cannula full of static blood	Connect it to the circuit or flush it. A large-bore cannula of standing blood in a low-flow patient is a clot
Both cannulas in the same leg without a reason	Compromises arterial inflow and venous return in one limb simultaneously — and Chapter 17 established these are two separate problems
Continuing to defibrillate while a wire is being advanced	One of only two permitted ACLS modifications exists to prevent this. A shocked patient moves
Continuing adrenaline boluses as the circuit is connected	Full pump flow into a maximally vasoconstricted circulation. Stop the boluses and anticipate weaning the infusions
Letting CPR quality degrade because attention moved to the groin	The reason there are two sub-teams. A pause costs the pause plus the rebuild of coronary perfusion pressure
Persisting with a third and fourth percutaneous attempt	After two unsuccessful attempts, failure and complications are very likely. Write the switch trigger and name who calls it
Omitting the distal perfusion cannula because the wire was not placed first	The ECPR exception is a deferral, not an exemption, and its deadline is 4 hours
Putting a regional oximetry probe on one leg	The discriminating signal is the difference between the two legs. You cannot compute it later
Treating "flow established" as the end	The groin may bleed again at pump pressures, the patient is about to be moved, and a cluster of ownerless reperfusion decisions has just begun (Chapter 22)

21.14 Key points

Both cues that normally identify a vessel fail at the same moment. Compressions make veins pulsatile and systemic oxygen debt makes arterial blood dark. In ECPR, vessel identification is an imaging problem, not an observational one — and that single fact generates most of this chapter.
The target moves. Under cardiac massage, anterior vessel entry is often not achieved and the tissue to be dilated is substantial, producing kinking, bleeding and vascular trauma — especially on the arterial side. A superstiff wire is the answer, and it belongs in the pack.
Every second is priced. Beyond 30 minutes of CPR, each further minute costs roughly 2–2.5% mortality. A manoeuvre that costs two minutes must be worth about five percentage points.
Complications are the normal case, not the exception — 32.7% in 1,644 out-of-hospital ECPR patients, mostly bleeding. ECPR carries substantially higher complication rates than other ECMO cannulation, driven by the technical and environmental conditions rather than by who is holding the needle.
Two sub-teams, two leaders, and neither watches the other. The commonest organisational failure is one team trying to do both jobs, and CPR quality is what degrades. Everything in the room is arranged around the cannulator.
Stage 1 — arterial and venous angiocatheters — is the one thing in Part IV that costs nothing if you turn out not to need it. If ROSC follows, the patient has a femoral arterial line and central access. If it does not, the slowest part of the cannulation is already done.
An intra-arrest arterial line answers four questions: how good are the compressions, has the ideal compression vector changed, is this pseudo-pulseless electrical activity, and is this patient about to re-arrest.
There are exactly two permitted modifications to ACLS: defibrillation is suspended during guidewire insertion, and adrenaline boluses stop as the circuit is connected. Everything else continues unchanged, and the point of naming only two is that the rest must not drift.
Target the common femoral artery at the femoral head, just proximal to the bifurcation. Below it, the vein lies posterior to the superficial femoral artery and a needle can pass through the artery into the vein. Above the inguinal ligament, bleeding is beyond the reach of compression.
Transverse to find, longitudinal to puncture. The transverse view is less likely to achieve the anterior entry that a 25 Fr cannula needs.
Confirm both wires before anything is dilated — left of the spine is aorta, right of the spine is inferior vena cava, or transoesophageal echocardiography — and verify freedom of the wire tip. This is the step that prevents the complication no one survives.
Ultrasound and fluoroscopy answer different questions and are not competitors. Ultrasound: am I in the right vessel, first pass. Fluoroscopy: where has the wire actually gone. Both ECPR studies that added fluoroscopy found large complication reductions at no time cost.
In ECPR the mechanical benefit of ultrasound is the outcome benefit, because time is the indication. A 2.5-minute saving is worth five to six percentage points of mortality — larger than most of the complication differences being argued over.
Percutaneous is fastest (6–8 minutes), is now 89% of ELSO Registry practice, and is associated with fewer severe neurological complications (13% versus 19%, adjusted OR 0.62) with no difference in vascular complications — a pattern this book reads as a time effect wearing a neurological mask, since cutdown takes 20–30 minutes.
A programme should be competent in two techniques. The hybrid cutdown needs no imaging, identifies vein from artery by looking, favours a single puncture per vessel, and works in patients percutaneous access cannot serve. Paris adopted it after percutaneous ECPR failed to establish itself, and reports 21.3 minutes and a 7.6% failure rate.
After two unsuccessful attempts at a vessel, failure and complications are very likely. Write the switch trigger into the protocol and name who calls it, because the decision to change technique is the one nobody makes in time.
The distal perfusion rule is deferred in ECPR, not waived, and the deferral has a 4-hour deadline. Hand it to a named person with a time. Put regional oximetry probes on both legs before leaving the room — the discriminating signal is the difference.
Purge the air even under time pressure; raise the RPM before unclamping; keep compressing until flow approaches 2.5–3 L/min; then stop. Anticipate hypertension as the pump meets a vasoconstricted circulation.
A groin that is dry under compressions may bleed at pump pressures, and the patient is about to be moved.
Cannulation is a trainable skill with a published pathway — model, cadaver, simulation, then assist five and perform five proctored procedures — and safety series report 98.5% success with 1.7% high-significance complications in trained intensivist hands, conditional on structured credentialing and immediate surgical backup.

Cross-references

Backwards

Chapter 3 — cannula flow physics and configuration notation; the Poiseuille relation used in §21.7.
Chapter 12 — VA cannulation in the patient with a circulation. This chapter is that chapter under compressions, and §21.9 resolves its distal-perfusion-wire rule for ECPR.
Chapter 14 — the first hours on VA ECMO, and the 4-hour distal perfusion deadline that §21.9 inherits.
Chapter 17 — the six mechanisms of limb ischaemia, the bilateral NIRS probe, the cannula-to-vessel ratio, and why arterial and venous limb problems are separate.
Chapter 19 §19.2 — the coronary-perfusion-pressure physiology that makes compression pauses expensive, and the arterial-line targets used in §21.3.
Chapter 19 §19.3 — the 2–2.5% per minute figure that prices every step in this chapter.
Chapter 19 §19.5 — the ownerless reperfusion decisions that begin the moment §21.10 ends.
Chapter 20 §20.5.8 — pulmonary embolism as a high-yield aetiology, which is why §21.5's thrombosed-vein warning matters.
Chapter 20 §20.8 and §20.9 — gate 2 happens while the wire is going in; and the selection loop that §21.12's Vienna curve illustrates.

Forwards

Chapter 22 — everything after flow is established. §21.10 hands over mid-sentence.
Chapter 18 — decannulation, to which §21.9's finding of equal complication rates between percutaneous and surgical removal belongs.
Chapter 27 and Chapter 32 — circuit priming, de-airing, air entrainment and circuit catastrophes.
Chapter 60 — prehospital and interhospital ECPR, where the three-person team and the hybrid cutdown belong in full.
Chapter 62 — paediatric ECPR, which owns the central-versus-peripheral cannulation data referenced in §21.11.
Chapters 88–90 — simulation, which owns the training pathway in §21.12 and the evidence that an ECPR simulation programme shortens real cannulation times.

References

All rows seeded from this chapter are marked Verified = No — every external source below is from a structured abstract.

ECPR cannulation — imaging

Kashiura M, et al. Effect of ultrasonography and fluoroscopic guidance on the incidence of complications of cannulation in extracorporeal cardiopulmonary resuscitation in out-of-hospital cardiac arrest: a retrospective observational study. BMC Anesthesiology. 2017. DOI 10.1186/s12871-016-0293-z.
Nakatsutsumi K, et al. Time-saving effect of real-time ultrasound-guided cannulation for extracorporeal cardiopulmonary resuscitation: a multicenter retrospective cohort study. Resuscitation. 2023. DOI 10.1016/j.resuscitation.2023.109927. SAVE-J II; 443 propensity-matched pairs.
Tanaka S, et al. Venoarterial extracorporeal membrane oxygenation for cardiopulmonary resuscitation: a retrospective study comparing the outcomes of fluoroscopy. Heliyon. 2024. DOI 10.1016/j.heliyon.2024.e24565. Adjusted odds ratio 9.92 (2.04–81.2) — flagged in §21.6 and the point estimate is not used.
Goslar T, et al. Emergency percutaneous implantation of veno-arterial extracorporeal membrane oxygenation in the catheterisation laboratory. EuroIntervention. 2016. DOI 10.4244/EIJ-D-15-00192.
Vegas A, et al. Guidelines for performing ultrasound-guided vascular cannulation: recommendations of the American Society of Echocardiography. Journal of the American Society of Echocardiography. 2025. DOI 10.1016/j.echo.2024.12.004. Replaces the 2011 guideline. Source of the three roles of ultrasound and of the candid statement that the evidence remains weak.

Femoral access — randomised evidence, all in patients with a circulation

Sorrentino S, et al. Standard versus ultrasound-guided cannulation of the femoral artery in patients undergoing invasive procedures: a meta-analysis of randomized controlled trials. Journal of Clinical Medicine. 2020. DOI 10.3390/jcm9030677.
Sobolev M, et al. Ultrasound-guided catheterization of the femoral artery: a systematic review and meta-analysis of randomized controlled trials. Journal of Invasive Cardiology. 2015. DOI not established [VERIFICATION REQUIRED].
Jolly S, et al. Routine ultrasonography guidance for femoral vascular access for cardiac procedures: the UNIVERSAL randomized clinical trial. JAMA Cardiology. 2022. DOI 10.1001/jamacardio.2022.3399. NCT03537118. Null on its primary endpoint; positive on every mechanical endpoint.
Stone P, et al. A prospective, randomized study comparing ultrasound versus fluoroscopic guided femoral arterial access in noncardiac vascular patients. Journal of Vascular Surgery. 2020. DOI 10.1016/j.jvs.2019.09.051.

ECPR cannulation — technique and outcomes

Wang L, et al. Percutaneous cannulation is associated with lower rate of severe neurological complication in femoro-femoral ECPR: results from the Extracorporeal Life Support Organization Registry. Annals of Intensive Care. 2023. DOI 10.1186/s13613-023-01174-1. 3,575 patients, 2008–2019.
Inoue A, et al. Extracorporeal cardiopulmonary resuscitation in adult patients with out-of-hospital cardiac arrest: a retrospective large cohort multicenter study in Japan. Critical Care. 2022. DOI 10.1186/s13054-022-03998-y. 1,644 patients; complications in 32.7%; the complication dataset for this chapter.
Dewolf P. A systematic review of current ECPR protocols. A step towards standardisation. Resuscitation Plus. 2020. DOI 10.1016/j.resplu.2020.100018. 24 articles, 1,723 patients; percutaneous Seldinger preferred in 44% of protocols.
Kwinta A, et al. Feasibility and outcomes of anaesthesiology- and intensive care-led ECPR in a hospital without cardiac surgery: a 2.5-year prospective registry. BMC Anesthesiology. 2026. DOI 10.1186/s12871-026-03678-2. Source of the 7.1% veno-veno-arterial conversion rate and the 38.1% distal limb ischaemia figure.
Fisser C, et al. Arterial and venous vascular complications in patients requiring peripheral venoarterial extracorporeal membrane oxygenation. Frontiers in Medicine. 2022. DOI 10.3389/fmed.2022.960716. 427 screened patients; the between-legs NIRS difference finding.
Rupprecht L, et al. Pitfalls in percutaneous ECMO cannulation. Heart, Lung and Vessels. 2015. DOI not established [VERIFICATION REQUIRED]. 720 patients, 159 percutaneous veno-arterial.
Marquez AM, et al. A single centre experience and outcomes with central and peripheral cannulations for paediatric extracorporeal cardiopulmonary resuscitation. Resuscitation. 2025. DOI 10.1016/j.resuscitation.2025.110678. Paediatric — held for Chapter 68, Paediatric Considerations.

Training, competence and the learning curve

Duffin SC, et al. An intensivist-led ECMO accreditation pathway and safety data over the first 4 years. Critical Care and Resuscitation. 2023. DOI 10.1016/j.ccrj.2023.11.006. 402 cannulations; 98.5% success; 1.7% high-significance complications.
Davis J, et al. Safety outcomes of intensivist-led extracorporeal membranous oxygenation cannulation. ASAIO Journal. 2024 (abstract 104). DOI 10.1097/01.mat.0001070012.89490.89.
Sato R. Safety of ECMO cannulation: organization and standardized training matters. Journal of Intensive Care Medicine. 2025. DOI 10.1177/08850666251386397. Correspondence reply. Source of the statement that ECPR complication rates are driven by technical and environmental challenges rather than operator specialty.
Magnet I, et al. Extracorporeal cardiopulmonary resuscitation: outcomes improve with center experience. Annals of Emergency Medicine. 2025. DOI 10.1016/j.annemergmed.2024.12.004. The Vienna curve, and its authors' own attribution to a summation of training, selection and standardisation.
Moreno AR, et al. Training of emergency medicine residents to initiate extracorporeal membrane oxygenation cardiopulmonary resuscitation (ECPR). AEM Education and Training. 2025. DOI 10.1002/aet2.70116. Swine model; the most frequent reason for failure was inability to obtain vascular access.
Schummer W, et al. Mechanical complications and malpositions of central venous cannulations by experienced operators. Intensive Care Medicine. 2007. DOI 10.1007/s00134-007-0560-z. 1,794 procedures. The source of the two-attempt rule in §21.11 — not an ECPR study, which is why it generalises.
Burns B, et al. Common femoral artery access in emergency medicine. Emergency Medicine Journal. 2025. DOI 10.1136/emermed-2025-215295. Reports that ECPR is now an American Heart Association Class 2 recommendation for cardiac arrest. [VERIFICATION REQUIRED] — the class subdivision (2a or 2b) is not stated in the retrieved abstract and is not guessed here.

Textbooks

Shinar Z, Badulak J. ECPR and Resuscitative ECMO. Chapter 4, Running the ECPR Code (Bellezzo) — personnel, the two sub-teams, role cards, room setup, the staged model and the intra-arrest arterial line; Chapter 5, ECMO Cannulation for ECPR — cannulation strategies, cannula selection, guidewires, the procedure, distal perfusion cannulas and procedural pearls; Chapter 6, Hybrid Cutdown Technique for ECPR Implementation (Hutin and Lamhaut) — the Paris method, results, benefits, complications and the training pathway; Chapter 12 — the prehospital three-person team. [VERIFICATION REQUIRED] — page numbers not confirmed.
ELSO Red Book, 6th edition, Chapter 32 — the two ACLS modifications, cannula sizes, the cannulation checklist, wire confirmation, the pulmonary-embolism ultrasound instruction, the connection sequence and the 4-hour distal perfusion window; Chapter 4, Cannulation — ultrasound views, the bifurcation trap, micropuncture, wire position by fluoroscopy, contralateral cannulation, dilator discipline, difficult access and the superstiff-wire exchange, and the open femoral approach. [VERIFICATION REQUIRED].
Taha AR, Caridi-Scheible M, Leiendecker E, et al. ECMO: A Practical Guide to Management, Chapter 7 — the distal perfusion cannula technique and the explicit ECPR exception to the wire-first rule, surgical cutdown, the modified Seldinger hybrid technique, and cannulation complications. [VERIFICATION REQUIRED].
Comprehensive Healthcare Simulation: ECMO Simulation, Chapter 23 — ultrasound-guided vessel identification, puncture and wire approval as discrete simulated skills, and the three levels of simulation fidelity. [VERIFICATION REQUIRED]. Note: Chapters 88–90 are checklist and algorithm chapters, not simulation chapters — no chapter in this book currently owns ECMO simulation, and this material is therefore unhoused.

📅

Chapter status

Drafted and audited 12 September 2026. Ten-pass quality control completed: clinical, physiology, evidence, citation, numerical, safety, contradiction, redundancy, bedside utility and literature-currency passes.

This is a technique chapter with no randomised evidence behind its technique. No trial has compared any ECPR cannulation method with any other, and none is registered. The randomised evidence quoted here is from femoral arterial access in patients with a circulation, and it is labelled as such wherever it appears. Everything drawn from an arresting patient is observational and much of it is single-centre.

Two controversies were set out rather than smoothed. §21.6 reconciles an apparently contradictory imaging literature by separating two questions — ultrasound answers am I in the right vessel, fluoroscopy answers where has the wire gone — and argues that in ECPR the mechanical benefit is the outcome benefit, because a 2.5-minute saving is worth five to six percentage points of mortality on the Red Book's own estimate. §21.8 examines the ELSO Registry finding that percutaneous cannulation reduces severe neurological complications without changing vascular ones, and reads it — as this book's reasoning, labelled — as a time effect wearing a neurological mask, since cutdown takes fifteen to twenty minutes longer.

The numerical audit flagged one problem and did not correct it: an adjusted odds ratio of 9.92 with a confidence interval from 2.04 to 81.2, a forty-fold span; the direction is used and the point estimate is not. A second figure — 1 of 8 versus 2 of 3 for upfront distal perfusion — is reported with its denominators and explicitly not used as an estimate.

The contradiction audit found a second internal contradiction inside a single source: the Red Book advises contralateral femoral arterial and venous cannulation in Chapter 4 and notes the advantage of unilateral cannulation in Chapter 32. Both are printed, neither is asserted, and this book's preference for contralateral access where the vessels can be seen is labelled as reasoning built on Chapter 17.

The redundancy audit removed general VA cannulation technique, which belongs to Chapter 12, limb-ischaemia management, which belongs to Chapter 17, and everything after flow is established, which belongs to Chapter 22. The chapter ends deliberately mid-handover.