I had the tremendous privilege last week to attend the REANIMATE course in San Diego. Run by the ED ECMO gang — Zack Shinar, Joe Bellezzo, and Scott Weingart — for the past several years REANIMATE has been teaching “endovascular resuscitation” to a global audience of ED providers, intensivists, and overall resuscitationists.
It’s 90% about ECMO. We talked about patient selection, cannulation, initiating heart-lung bypass, and critical care management afterwards. Although areas like VV ECMO came up, the main focus was on “ECPR,” or crash induction onto VA ECMO for refractory cardiac arrest. We also addressed REBOA, the overall management and flow of a cardiac arrest, and some other associated topics.
The faculty was insanely talented, universally friendly, and appropriately irreverent. Among others, the EM and CCM crowd was represented by Zack and Joe from Sharp Memorial, their colleague Chris Ho, Scott from EmCrit, Zaf Qasim to show us REBOA, transport master Leon Eydelman, and a contingent of great Alfred folks including Chris Nickson of Life in the Fast Lane. The surgical perspective was assisted by ECMO godfather Bob Bartlett, Amy Hackmann from USC, and Joe Dubose from Shock Trauma. Jim Manning showed us his SAAP catheter, and vendors were on hand to demonstrate specialized hardware. It was nuts.
The use of ECMO for ECPR is on the bleeding edge of resuscitation; while the underlying technology is decades old, the refinement and logistics needed to make it a success are only recently bearing fruit. Like all novel areas in medicine, this makes it an exciting playground, and something of an artisinal one; there are no well-established best practices, and the techniques continue to evolve every year as advances are introduced at the handful of busy centers around the world (Paris, Minnesota, the Alfred, etc) and then spread by diffusion. One wonders whether in twenty years, ECMO will be as routine as mechanical ventilation, and we’ll wonder how we ever practiced critical care without it.
I’d like to share my notes from the course, supplemented with material from the pre-conference reading and the ED ECMO podcast. As I’m no expert myself, all of this should be considered unsourced and unreliable, but perhaps it will start your wheels turning. Some of my own thoughts and editorials are woven through in boldface. It’ll be broken up into three posts due to length.
Which cardiac arrest is a candidate for ECMO?
“ECMO is for the neurologically and cardiovascularly salvageable.” Otherwise criteria are evolving.
- Witnessed arrest
- Bystander CPR (Minnesota does not care about this, but does do some prognostic stuff with lactate, etc). <10 minutes no-flow time.
- Total downtime <75 minutes
- Initial rhythm not asystole. Zack Shinar would like to move past this; although outcomes are worse in this cohort overall, presumably all asystole is not heterogeneous; some asystole patients were in VF just a moment ago, and the preponderance of asystole is so high that if we can select a good subgroup, there could be quite a few saves.
- No terminal conditions, not DNR. In ICU patients, not clear to me how we would differentiate between critically ill patients who arrest but can be saved, and those in whom arrest portends a terminal state.
- Age criteria can be flexible, but probably <75-ish
- AAA or dissection are a relative contraindication
Paris: signs of life (movement, tearing, etc) trumps all else.
The Alfred: excludes ETCO2 <10
Consider ECMO a sort of “bridge to neuro evaluation” when used for ECLS. We won’t know how the brain is doing for a few days, so give the patient that time (including hypothermia) — then you can take them off if function seems poor. Paris has 10% survival from OOHCA with field ECMO — and zero survival with poor neuro outcome, because they just take those people off support. [If we continue to expand this use, though, one can certainly imagine getting “stuck” with patients with no ability to survive off the pump and a family who won’t allow withdrawal. Is transfer to an ECMO LTAC in the future?]
ECMO in trauma? Animal models from the Swedes suggest improved markers of shock when used after a fixed dose of shock and surgical hemostasis. [I am dubious of the value here; these patients are generally dying of hemorrhagic shock more than anything else, and ECMO is more harm than help there. If they stop bleeding they do okay.]
Probably generally okay to run circuit without anticoagulation when there is coagulopathy. When asked this, vendors predictably said they don’t recommend it, suggest correcting medical coagulopathies and then anticoagulating. Good luck with that. Of note, Centrimag circuit is apparently not heparin-coated.
Evidence for ECPR
SAVE-J: 4x good neuro survival when OOHCA brought to ECPR centers.
Brussels data: Propensity matched arrests showed superior survival and neuro outcomes with ECPR.
Big 5 centers currently seeing about 35% survival with good neuro outcome with their ECPR cases. Average 59 minutes CPR.
While optimal patient selection obviously helps the numbers, the duration of arrest goes the other way — any other non-ECMO patient with 59 minutes of CPR would be declared dead by fiat, so these are basically pure saves. [I would have expected such prolonged arrests to universally have poor neuro outcomes. But it seems that we previously confounded “too much low-flow time to get ROSC” and “too much low-flow time for preserved neuro function.” In reality there appears to be an in-between space, wherein ROSC will not occur spontaneously, but if we can extrinsically restore perfusion, the brain may still recover.]
Join the ELSO registry to contribute to the data!
Running the ECPR arrest
EM, critical care, cardiology, cardiac surgery could all be your cannulators. Nurses or perfusionists or other can run your pumps.
Sharp suggests for the code:
- 2 docs: Code leader (thinks big thoughts, intubates, echos) and cannulator
- 2 RNs: Nurse code leader (runs ACLS) and med-electric nurse (gives drugs and shocks)
- 2 techs: Compressors
- Wire assistant to help cannulator
- Pump manager (perfusionist or trained RN)
- Misc: runner, pharmacist, etc.
Stages of arrest:
- Stage I: Patient moved to bed. Cannulator immediately uses ultrasound to place a 5 Fr arterial and 9 Fr venous sheath in right femoral. These are used for access and BP transduction. Nurse leader runs ACLS code. Doc leader intubates and then starts putting together history.
- Hard stop once access obtained: “Is this an ECMO candidate?” If code leader says yes based on the big picture, proceed.
- Stage II: Heparin given. Cannulator dilates access up and places ECMO cannulae. ECMO person gets pump ready. Stop attempts at defibrillation — risky to cannulator, and unlikely to work if it hasn’t yet.
- Hard stop once cannulas in: “Is this an ECMO candidate?” If nothing has changed, proceed.
- Stage III: Confirm heparin was given. Connect cannulas using underwater seal. Start bypass and stop compressions.
First goal is to save without ECMO. Run a good, coordinated arrest with high-quality CPR and you’ll get saves before you ever get to bypass. Don’t miss an easily correctable cause such as hyperkalemia, pneumothorax, etc.
If mechanical CPR device desired, apply early during Stage I, as you’ll never get another chance once cannulation starts. Not mandatory, but it does help quiet down the scene, and reduces movement of the patient during cannulation.
There is no commitment to ECMO until you begin dilating the artery (at that point, probably more risk to stop than to go ahead and insert a cannula). So start all codes with Stage I, and you’ll be ready to do ECMO, but never required.
The ECMO machine
The components, in order:
- Venous cannula from patient
- On some circuits: Rapid prime connector for introducing IV fluid
- Flow probe (some need contact gel, some don’t)
- Pre-oxygenator sampling port and/or continuous SvO2 monitor
- Oxygenator. Connected by water hoses to heater/cooler. Also connected to gas supply (pure oxygen or blender).
- Post-oxygenator sampling port
- Stopcock or Y leading to perfusion catheter
- Arterial cannula to patient
Some circuits have a shortcut (recirculation bridge) between the two patient limbs, which (when opened and patient limbs clamped) turns the circuit into a closed loop. This allows retained flow during priming. On other circuits, there is no bridge, but the two patient limbs begin as a single closed loop (i.e. a single piece of tubing), which you cut open after priming to create the arterial and venous limbs.
Pump is generally centrifugal (“whirls” the blood outward); roller pumps caused a lot of hemolysis and are rarely used for adults now. These work best when upright. They are controlled by setting an RPM, and the resulting flow is a measured figure, depending on preload/afterload. Most devices have a hand crank backup in case the pump fails.
Default gas flow is 100% oxygen from a regular flowmeter. Optimally, you can add a blender to titrate the gas mix (FdO2).
- Maquet Cardiohelp: Small and convenient.
- Centrimag: Totally contactless pump (rotor levitates on magnets), supposedly causes less hemolysis. However, cannot use a hand crank, so must have a backup pump (at the bedside at all times, not elsewhere).
- Tandemheart: Can be used for ECMO. Simple.
- Bolus 5000 units of heparin (to patient, or may be added to priming solution)
- Clamp arterial/venous tubing. Close recirculation bridge if present. Mount oxygenator and pump. Put gel on flow sensor if needed. Turn on heater and connect water lines if using heater/cooler (okay to skip initially, especially if hypothermia desired).
- Remove yellow deairing port from top of oxygenator
- Attach priming fluid (Isolyte, LR, whatever). Can deair bags by inverting and squeezing air into circuit.
- Hang fluid above circuit and open clamps to gravity prime until circuit full. Lower patient limbs to prime those too.
- Start the pump. Run at ~4 L for at least 10-15 seconds. Stop and check for bubbles; if present, keep running. Replace cap on deairing port.
- Connect gas to oxygenator inlet.
- Change any vented caps for regular caps.
Big summary points
- We don’t know exactly the right patients for ECPR, but there’s clearly benefit in the right ones. Not sure yet how to manage referral (hub-and-spoke) models, since time is of the essence.
- An ECPR service can be run by different folks, but a system needs to be in place before hand. You probably need cardiac surgery, vascular surgery, and IR all available, but not necessarily part of the core team or in-house.
- Equipment is all a little different and continually evolving.
Continue to Part II on Cannulation and Part III on pump management and critical care.