Mechanical Circulatory Support
A map of the devices, what each one actually does to the circulation, and how to decide which failing patient needs which machine.
Why support a circulation at all
Inotropes buy cardiac output by raising myocardial oxygen consumption in a heart that is already ischemic or exhausted. At some point that exchange stops being worth it. Mechanical support takes over part of the work instead, so the ventricle can do less and still deliver more. That is the whole idea. Everything else is engineering.
Support is not treatment. None of these devices fixes anything. They hold a circulation together while something else happens, and if nothing else is going to happen, you have bought suffering rather than time. Before you pick a device, answer three questions out loud, in front of the team.
Left, right, or both. Left-sided devices unload the left ventricle and, in doing so, deliver more preload to a right ventricle that may already be drowning. Right-sided devices push blood into a pulmonary circulation and a left heart that has to accept it. Put a device on the wrong side and you will make the patient worse within the hour. Get the answer from echo and from filling pressures, not from the blood pressure.
A patient with a P/F ratio of 60 and a normal ejection fraction needs gas exchange and nothing else. A patient in cold, wet shock with clear lungs needs flow. A post-arrest patient with pulmonary edema and a stunned ventricle needs a circuit with an oxygenator in it. Only ECMO oxygenates. Every other device here moves blood the patient's own lungs have already had to deal with.
Recovery, a durable ventricular assist device, transplant, or withdrawal. Those are the four. Bridge to decision is a legitimate answer for about seventy two hours and then it is not an answer any more. If the patient is not a transplant or device candidate and the myocardium will not recover, the honest plan is comfort, and that conversation belongs to the family before the cannulas go in. Have it early and write it down properly, as covered in goals of care and code status.
A patient supported for weeks with no destination, awake enough to know it, and a family who were never told this was possible, is the worst outcome in this field. It is not created by bad intentions. It is created by an absent plan and a series of individually reasonable escalations.
The comparison table
Read the afterload column first. It is the one people get wrong, and it explains most of the complications you will see.
| Device | Supports | Flow added | Effect on LV afterload | Access | Anticoagulation | Typical duration |
|---|---|---|---|---|---|---|
| IABP | LV, partially. Needs a beating, ejecting heart | 0.5 to 1.0 L/min | Reduces it, by deflating just before systole | Percutaneous femoral artery, 7 to 8 Fr | None to low, unit-dependent | Days |
| Impella CP | LV, substantially | Up to 3.7 L/min | Reduces it, and directly unloads the ventricle | Percutaneous femoral artery, 14 Fr pump | Moderate systemic, plus a heparinized purge | Days, up to about a week |
| Impella 5.5 | LV, full support | Up to about 6 L/min | Reduces it, fully unloads the ventricle | Surgical cutdown, axillary artery or direct aortic | Moderate systemic, plus purge | Weeks. Patient can be mobilized |
| Impella RP | RV only. IVC to pulmonary artery | Up to 4 L/min | None directly. Raises LV preload | Percutaneous femoral vein, 22 Fr | Moderate systemic, plus purge | Days to about two weeks |
| VA ECMO | Both ventricles and the lungs. Full support | 3 to 6 L/min | Increases it. Retrograde aortic flow | Femoral vein and femoral artery, or central | Full systemic | Days to weeks |
| VV ECMO | Lungs only. Gas exchange, no circulatory support | None. Flow is neutral | None | Femoral and jugular veins, or one dual-lumen cannula | Full systemic, often a lower target | Weeks, occasionally months |
| TandemHeart | LV. Drains left atrium, returns to femoral artery | Up to 4 to 5 L/min | Increases it, while unloading LV preload | Femoral vein with transseptal puncture, plus femoral artery | Full systemic | Days to about two weeks |
| Durable LVAD | LV, indefinitely. The destination, not a rescue | 4 to 10 L/min | Reduces it. Continuous unloading | Surgical: inflow at the LV apex, outflow to the ascending aorta | Warfarin plus aspirin, lifelong | Years |
VV ECMO does not support the circulation. It takes blood from a vein, oxygenates it, and returns it to a vein. The patient's own heart still has to pump it. A patient in cardiogenic shock does not get VV ECMO.
VA ECMO increases left ventricular afterload. It returns blood retrogradely into the aorta, and the failing ventricle has to eject against it. Alone among the pumps here, it loads the left ventricle rather than unloading it, which is why every VA patient needs a deliberate answer to how the LV is being vented.
The escalation ladder
The textbook sequence runs: optimize volume and rhythm, start an inotrope, add a vasopressor if the pressure will not hold, place a balloon pump, escalate to an Impella, go to VA ECMO if that is not enough. Each rung adds flow and adds risk, and you move up when the patient tells you to: a lactate that will not clear, a mixed venous saturation that stays low, a cardiac index under about 2.2, a urine output that has stopped.
Now the honest part. In real cardiogenic shock the ladder is frequently skipped, and skipping it is often correct. A patient in SCAI stage E with a lactate of 12 does not need a device that adds 0.8 L/min, and marching them up rung by rung wastes the only thing they do not have. Every hour of hypoperfusion costs kidneys, liver and brain you will spend three weeks trying to get back. Match the device to the deficit, not to the protocol. The counterweight: the biggest device is not automatically right either, and VA ECMO in a ventricle that is still ejecting can precipitate pulmonary edema on full mechanical support.
The most useful development in this field over the last decade is not a pump. It is the shock team: a defined protocol, a single number to call, and a small group with interventional cardiology, cardiac surgery, heart failure and critical care on it who assemble within minutes and make one decision together. The mechanism is not mysterious. It replaces a chain of sequential referrals with one conversation, and it means somebody is accountable for the exit strategy from hour zero.
Learn your institution's activation criteria and its number. That is the practically useful thing to take from this page, and it works like any other rapid response activation: you do not need to be sure, you need to call early.
SCAI shock stages A to E
Cardiogenic shock used to be described in adjectives, which meant two clinicians could look at the same patient and disagree about whether it was shock at all. The SCAI classification replaced that with five stages. Its main value is as a shared language: say stage C on a phone call and everyone knows what you mean.
Not in shock. A patient who could plausibly get there, such as a large anterior infarct or an acute decompensation of chronic heart failure, but currently warm, well perfused, normal lactate, lucid. The point of the stage is to make you watch.
Pre-shock. Relative hypotension or tachycardia and elevated filling pressures, but perfusion is preserved: normal lactate, warm extremities, clear mentation. This is where recognition changes outcomes and where it is most often missed, because the numbers still look survivable.
Hypoperfusion requiring an intervention beyond volume. Cold, clammy, mottled, oliguric, confused, with a rising lactate and creatinine. This is the patient who gets an inotrope, a vasopressor, or a device. Most people picture stage C when they say cardiogenic shock.
Stage C that has not responded. The initial intervention is in and the patient is no better or worse: lactate climbing despite therapy, escalating pressor requirement, deepening acidosis. This is the escalation stage, and if you are calling the shock team late, this is usually where you are.
Circulatory collapse. Refractory hypotension on multiple agents, profound acidosis, refractory arrhythmia, arrest with ongoing CPR. Survival depends on flow within minutes, which in practice means VA ECMO in a center prepared for it. Cardiac arrest is also recorded as a separate modifier at any stage, because it changes prognosis independently.
Stage is dynamic, so restage at every reassessment rather than treating the admission label as fixed. Filling pressures and cardiac output from a PA catheter make staging far less speculative than doing it from a blood pressure and a hunch.
Right ventricular failure
This is the thing that gets missed, because everybody's mental model of shock is a left ventricle. The right ventricle is a thin-walled, compliant, low-pressure chamber built to move a large volume against almost no resistance. It tolerates volume badly and pressure worse, and it fails with a normal-looking left ventricle on the echo report.
How to recognize it
- A high CVP with an unimpressive wedge. A central venous pressure that is high in absolute terms and high relative to the wedge. When right atrial pressure approaches or exceeds the wedge, think RV.
- A low pulmonary artery pulsatility index: PA systolic minus PA diastolic, divided by right atrial pressure. A low value means the RV is not generating pulse pressure against its load, and it is one of the better bedside predictors of RV failure after left-sided support.
- The clinical picture: distended neck veins, a pulsatile liver, edema, rising bilirubin and creatinine from venous congestion, and clear lungs. Congested kidneys, not dry ones.
- On echo: a dilated, poorly contracting RV, a flattened septum, a D-shaped left ventricle, significant tricuspid regurgitation, and reduced excursion of the tricuspid annulus.
The instinct is that the RV is preload dependent, so volume will help. In early RV infarction with low filling pressures, sometimes it does. In an already dilated RV it does the opposite: wall stress rises, tricuspid regurgitation worsens, the septum bows into the left ventricle, and LV filling falls through ventricular interdependence. Cardiac output drops after the bolus. Meanwhile rising RV wall tension and falling systemic pressure squeeze RV coronary perfusion, so the ventricle becomes ischemic as well as overstretched.
Give volume only if the filling pressures say the patient is genuinely underfilled, in small aliquots, and reassess. Otherwise the treatment is usually the opposite.
Medical measures first. Correct hypoxia, hypercapnia and acidosis, because all three raise pulmonary vascular resistance, and reduce mean airway pressure where you can. Add an inotrope such as dobutamine or milrinone. Maintain systemic pressure to perfuse the RV, noting that vasopressin raises systemic resistance with little effect on the pulmonary bed. Consider inhaled nitric oxide or epoprostenol. Restore sinus rhythm if you can, because a failing RV depends on the atrial contribution more than you expect.
The devices. Isolated RV failure is supported by the Impella RP, which draws from the inferior vena cava and delivers to the pulmonary artery, or by a dual-lumen cannula in the internal jugular that drains the right atrium and returns to the pulmonary artery, with an oxygenator spliced in if the lungs also need help. Biventricular failure, or RV failure with hypoxemia, is VA ECMO territory. Remember the iatrogenic version too: unloading the left ventricle increases the volume the right ventricle must deliver, so a marginal RV can decompensate hours after a left-sided device goes in. A suction alarm on a left-sided pump is a failing right ventricle until proven otherwise.
Where to go next
The intra-aortic balloon pump covers counterpulsation physiology, reading timing off the arterial waveform and the four timing errors, trigger selection in atrial fibrillation and during pacing, the contraindications including the absolute one of aortic regurgitation, weaning by ratio, and removal.
Impella covers the platform sizes and what each is for, reading the placement signal and the motor current, why a suction alarm means stop and think rather than turn it up, the purge system and what rising or falling purge pressure means, hemolysis management, and the contraindications that catch people out, particularly LV thrombus.
ECMO covers the VA and VV configurations, sweep versus flow and which dial changes what, differential hypoxia and why you sample from the right radial artery, left ventricular distension and the venting strategies, circuit inspection, bleeding, and weaning trials.
For the numbers behind every decision on this page, read the Swan-Ganz catheter for cardiac output, filling pressures and the derived indices, and arterial lines and CVP for the waveforms underneath them.
This is not medical advice. It is a teaching outline for clinicians and clinicians in training. Mechanical circulatory support is initiated and managed by specialist teams under institution-specific protocols. Nothing here substitutes for those protocols, your perfusionist, or your attending.