iPhoneIcon_BigDespite all the thousands of man-hours and gazillions of money going into researching stuff like therapeutic hypothermia and adrenaline in cardiac arrest we really haven’t got the breakthrough we deserve. With the exception of pockets of excellence, ROSC rates have been parked in the 5-10% range since the 80s or so.

Then a couple of days ago I had the privilege of picking the brains of a remarkable senior colleague of mine.  He blew me away with an absolutely encyclopaedic knowledge of EM and cardiac arrest literature. He also shared some fascinating insights into the possible future of cardiac arrest management. Actually, these are pretty exciting times in cardiac arrest research.

One of the things he discussed was resuscitative thoracotomies and open heart massage in non-traumatic cardiac arrest. He suggested the largely forgotten art of open chest compressions is due for a reassessment and a revival. He suggested we should consider opening the chest, outside the operating theatre, of our cardiac arrest patients in order to directly compress the hearts instead of performing the old gold standard closed chest compressions. He went as far as to suggest this should be done in the EDs by ED docs or even in the field by the prehospital teams.

I spent some time looking into it.

History lesson
Open heart massage in cardiac arrest is not a new concept. It’s efficacy in laboratory animal experimental cardiac arrest was demonstrated already in 1874 by one professor Schiff. Then in 1903 and 1904 the first human open chest CPR attempts were reported (D.G. Zesas 1903 and W.W. Keen 1904). Both of them reported patients suffering anaesthetic chloroform collapse and arrest.

After that two massive case series have been published. The vast majority of these patient suffered operating theatre cardiac arrest. In 1953 Stephenson et al published a case series of 1200 theatre arrests that had open chest CPR. The recovery rate was 28%. Briggs et al reported theatre cardiac arrest from the Massachusets General during a 30 year period. In the patients where open chest CPR was initiated within 4 minutes, 58% recovered and were neurologically intact. In contrast, the Beth Israel study published in NEJM 1983 reported reported only 14% survival rates in hospital cardiac arrests who had closed chest CPR.

So, there were some promising findings. Then in the 1960s Kouwenhoven et al and many others popularised the closed chest compressions and blew the open chest CPR out of the water. To be fair, sternal compressions was described already  in 1786 by an Enfield surgeon named John Sherwin but never entered the mainstream until the 1960s. Since then closed chest has been the gold standard for the layman on the street as well as for the hospital resus super-specialists. Outside trauma, I think it is fair to say that open chest compressions in cardiac arrest nowadays is exclusively performed in theatre. Actually, the only times I have ever seen open chest compressions being done is during cardiac surgery or in trauma cases who had their chests cracked anyway.

The physiology of closed and open chest compressions
ACLS3_3-1We all know that early defibrillation and quality compressions are what matters in cardiac arrest. The end-point of compressions is generate cardiac output and coronary perfusion pressure.

Coronary perfusion pressure (CPP) is the difference between aortic diastolic pressure and the left ventricular end-diastolic pressure. In the working heart the coronary perfusion pressure is cycling between 0 and 70-80 mmHg. At least 60 mmHg or so is required for normal coronary blood flow.

When performing CPR we need to achieve a CPP of at least 15 mmHg in order to achieve ROSC. 15 mmHg certainly doesn’t guarantee ROSC but that’s how good our compressions need to be. There is an important paper in JAMA 1990 where Paradis et al demonstrated how a CPP of less than 15 mmHg resulted in 100%  failure to resuscitate. With closed chest CPR we rarely achieve that outside the lab. Several studies report abysmal CPPs with averages ranging from 1-9 mmHg (1,2).

Here is where open chest direct heart compressions might refind it’s place in cardiac arrest management. Open chest compressions are so much more efficient. The pressure gradient generated is significantly higher.

There is an interesting paper from 1995 by Michael Boczar et al. The authors study 10 out-of-hospital cardiac arrest patients admitted to their ED where the ROSC wasn’t achieved within 30 minutes. At an early stage during the resuscitation they inserted aortic and central venous pressure catheters in order to calculate the mean coronary perfusion pressure. Baseline measurements were measured during five minutes of standard closed chest compressions delivered by a chest compression device. Then, after five minutes, they performed a left lateral thoracotomy and obtained new measurements during open chest compressions.

The results are striking. Mean coronary perfusion pressure in closed chest CPR was 7.3 +/- 5.7. As is stated above, that well below the 15 mmHg we want to achieve. Now, compare that to the mean value for open chest compressions. During open chest compressions the coronary perfusion pressures averaged  32.6 +/- 17.8.

Another way to look at this is cardiac output. In 1965  Del Guercio et al demonstrated how  open-chest compressions resulted in a mean cardiac index of 1.31 L/min/m2 compared to only 0.61 L/min/m2 during closed-chest cardiac massage. The Boczar group only measured cardiac output in one of their 10 patients: the cardiac index increased from a mean of 0.56 to 2.23 L/min/m2

Open chest compressions are vastly superior to our standard closed chest compressions.

Take home message
Open chest compressions in non-traumatic cardiac arrest is not news, but since the popularisation of the closed chest compressions it is now rarely performed outside operating theatres.  Perhaps it is time to revisit the concept as a way forward in cardiac arrest management? It makes a lot of sense.

If our goal with compressions is to reperfuse the heart in order to achieve ROSC and provide cardiac output then the theoretical framework and evidence is in place. There is ample evidence demonstrating that open chest compressions are superior to our normal closed chest compressions as far as perfusing the heart goes.

Implementing this in the ED or in the prehospital field is not as far-fetched as one might think. Performing resuscitative thoracotomies outside the OR is not controversial anymore. Its frequently done in EDs in trauma. Many prehospital systems have been doing this outside hospitals for years with good outcomes in trauma.

As for how this would play out in practice I don’t know. How would it fit in with ECMO, anticoagulants, thrombolysis? Surely, cannulating patients during open-chest compressions will be a lot easier than on a bouncing traditional CPR patient…


This entry was posted in CPR, Emergency Medicine, Prehospital Medicine. Bookmark the permalink.


  1. Matthieu K says:

    Great post! It is indeed striking to see how little improvement in cardiac arrest outcome we have managed over the years. I was wondering what happened to the minimally invasive direct cardiac massage (umbrella-like) devices? Thanks.

  2. James DuCanto, M.D. says:

    My understanding of the improved physiology of open chest massage involves the cross clamping of the thoracic aorta (ATLS course, late 90’s).

    Pharmacologic methods to (in effect) cross clamp the aorta interest me–I’ll have to look that literature up, i.e., the use of Norepi infusions to MAP>80 during closed chest compressions. The issue here is monitoring—there generally is not any measurement of the blood pressure during CPR unless the patient has an indwelling arterial line, and, to be honest, the moment-to-moment waveform and numbers of the arterial line are not all they are cracked up to be–they can be confusing. Think about it–what type of monitors are we using during CPR that give us information that help us to tailor the resuscitative effort? End-Tidal CO2 is a step forward, of course, but it tells you little about how to change the resuscitation–it only tells you if the global effort is effective.

    This may seem like a step “back” in technology and method, but the use of a manual blood pressure cuff on the arm with ultrasound auscultation of the brachial artery to measure the MAP would provide more useful information (potentially) than an arterial line. The goal is a MAP>80. The heart won’t restart if we don’t put fuel in the tank, i.e., optimize coronary perfusion pressure.

    Thanks again–you have a unique and informative blog.

    • Christopher says:

      We’ve begun placing SpO2 and NIBP on all cardiac arrest patients in the field and have tried monitoring the MAP during resuscitation. Sometimes we’ll get readings and other times we will not. Hard to tell if the lack of readings indicate poor CPR, the need for vasopressors, or the inability for NIBP to work in times of severe shock/cardiac arrest.

  3. Jared says:

    Im not convinced yet. I am convinced that it is worth reevaluating in a contemporary context, but until there is contemporary data I would consider open-chest massage a topic for scholarly discussion instead of a therapy to offer every arrest patient. There are 2 main reasons I feel less than enthusiastic about opening a chest in every arrest.
    1) I would be cautious about taking any operating room scenarios and trying to apply them to an ER; it is a whole different world, not only in terms of chaos, but also in terms of resources and patient population. It doesn’t seem reasonable to compare outcomes when one patient is wheeled unresponsive into your resus bay with minimal hx, and the other arrests on the OR table, with tubes and lines in place and an anesthetist, surgeon, first assist and OR nurses in place and scrubbed (as appropriate of course – I can’t imagine an anesthetist scrubbing in).
    2) I don’t think think the older studies can be compared to what we do now. I didn’t read the old studies because I can’t access them at the moment, but I expect the criterion have differed somewhat since then. Most contemporary studies are including endpoint such a modified Ranking score, or cerebral performance category; the above studies seem to consider mortality as endpoint (can’t access the old studies, haven’t read them, please correct me if Im mistaken), with the exception of Briggs et al which reads a nebulous “neurologically intact”.
    What were the standards of care then and how do they differ from now? ACLS has changed a few times since Ive started, and Id be interested in learning how it was done in the 60’s.
    Aside from algorithms, the modalities we can offer have also changed significantly. Balloon pumps aside, we have LVADS that can match or exceed a normal CO, or can ECMO to generate CO and meet oxygenation requirements. This is becoming a more popular peri-arrest, or intra-arrest intervention. See the CHEER study (linked on the homepage below). If ECMO is invasive, a thoracotomy certainly is more so. For those interested check out .

    I am certain there are subgroups other than the post-op cardiac surgery patient who would benefit from open vs closed CPR; but I think Boczar et al are correct in saying “Further outcome studies are needed to determine the timeliness and appropriate indications for open-chest cardiac massage.”

    Great topic to revisit, Im stoked to see where this conversation goes.

  4. Naeem Toosy says:

    Though when BP is measured while using a mechanical compression device…near normal BPs are often seen….sadly not necessarily translating in ROSC though!

  5. Pingback: COCHRANE + LUCAS = NO LOVE |

  6. see this Twitter response: randomized trial showed no difference.

  7. K says:

    Thanks Stephen,
    I’d love to see that paper but cant find it on pubmed or anywhere else. Could you send a link to it or even the PDF?

    Also there are rumours of a japanese paper from the 1990s?

    • It is an abstract and I put an image of the scanned abstract into the tweet. Reference is: Sterner S et al. Annals of EM 1991; 20(9):949. I was involved: when I was a resident I did a dozen thoracomtomies as part of this study. It was not published as a paper because it had been done under a concept known as “implied consent” for unconscious individuals (and was, of course, IRB approved). By the time it was done, that concept was under harsh criticism and only later did the FDA come up with new rules for emergency research. Patients were only randomized after failing 3 shocks plus Epi. All were out of hospital arrests.

  8. Pingback: The LITFL Review 130 - LITFL

Leave a Reply

Your email address will not be published. Required fields are marked *