Wednesday 12 September 2012

Does CPR save lives?

In TV land, most people not only survive CPR (cardiopulmonary resuscitation), but they go home from hospital and function normally afterwards (yes, they actually studied this). In reality, less than 10% survive, and for many that do, it’s not a good life. The low number of people that survive and function well after CPR (as low as 2% in some studies) leads to a question: if the results with CPR are so dismal, what would the results be if we did not do CPR? In other words: does CPR itself do anything; were those that survived going to survive anyway?

In-hospital survival is a little better, but interestingly the survival rate did not change over 14 years in this study of over 400,000 patients, despite multiple changes in the way CPR is delivered over that time. I am aware of before-and-after studies where the survival improved after implementing a change in CPR, notably this study from Seattle. But this study used a more aggressive shock protocol (defibrillation) that may have been responsible for the difference (please note: this blog is about CPR, not electrical defibrillation). Also, the biases of before-and-after studies are such that they nearly universally show improvement over time (for many reasons that will be covered in a future blog post). The bias in these studies is so strong, I only sit up and take notice when one of these studies shows no improvement over time.

Other studies have shown that the technique of delivering CPR doesn’t matter. Using fancy devices doesn’t help (Cochrane review). Giving adrenaline (epinephrine) to these patients can improve the chance of restoring circulation prior to arriving at hospital, but to no avail: it makes little difference to the survival, and is associated with worse functional outcomes (here).

It doesn’t even matter if you breathe for them or not (here and here), and I would have considered that to be pretty important. In fact, breathing for the patient is now supposed to be bad, because it increases the pressure in the chest, decreasing the venous flow into the heart. So what does the chest compression do if it doesn’t increase the pressure in the chest? I get confused reading about the mechanics of this stuff (cardiac pump theory vs. thoracic pump theory etc.) but as many of you know by now, I am a pragmatist: I am less concerned with the biological mechanism, and more concerned with whether or not it saves lives.

Is there a downside to CPR?
This study showed a 21% complication rate at autopsy, and an 89% rate of missed diagnoses, many of which would have made the CPR futile. Common complications are rib fractures, lung damage, vomiting and aspiration (gastric contents into the lungs).

While the response to CPR doesn’t depend much on the technique, it might depend on what is wrong with the patient, which could vary from a simple faint to a massive lung clot. In both these extreme cases, it wouldn’t matter whether or not you had CPR. In the first case (fainting) CPR would just add harm until the patient woke up anyway, and in the second case (massive clot) you would be wasting your time because the event is not survivable. In between though, there are other conditions for which it is possible that CPR may be of benefit, and this makes it hard to convincingly state that CPR never works. There may be people wishing to comment, with personal experience of patients surviving after CPR. My question to them will be: When we look at that case objectively, how can we be certain that the patient survived because of the CPR?

The bottom line.
The results of CPR are a lot worse than most people think, and in most cases CPR will not make a difference to the outcome. CPR fits the pattern of treatments discussed in this blog, with overestimated benefits and underestimated harms. Currently, however, the concept of CPR as an effective treatment is so deeply woven into our thinking that it would be impossible to stop doing it, or even to subject it to better scientific testing. Maybe in 50 years we will look back on our vigorous but naïve attempts to reverse death, and on the harms we did to those who were going to live anyway.


  1. You know I'm a big fan of yours, but I'm not so sure about this post. For one thing, it echos a recent blog by Dr. George Lundberg, former editor of JAMA. (Link: He is in your corner. However, he doesn't state the case so eloquently. In fact, he sounds a little crazy. He firmly states that he doesn't want CPR perform on himself.

    It may not be possible to completely divorce CPR from defibrillation either, especially in the hospital setting and even in many public places that have AEDs.

    What about you? If your heart stops today, would you not want CPR?

    1. Thanks, and that was quick. I had this on the back-burner for a while, and as you know, we discussed George Lundberg's blog. I agree with all your points, above.

      I don't like to personalise things, and to maintain my point that CPR is (at best) a lot less effective than most people think, and also more harmful than generally believed. I have also been careful not to state that it is definitely NOT effective.

      I know you will push for an answer, so my personal opinion is that squeezing someone's chest is unlikely to save their life, reversed their stopped heart, or maintain reasonable circulation and oxygenation. You and I have seen it: unless they revert with defib, or were not too bad to start with, they stay blue the whole time you are pumping. My answer: leave me alone.

  2. You don't mention children, in which survival is much better, especially with bradycardia.

    1. Thanks, and you are right: the literature reviewed (and most of the literature out there) is on adults. In fact, most people undergoing CPR, particularly in a hospital environment, are older and the risk of harms in this group is also going to be higher than in children.

      Having said that, do you have good evidence that CPR is helpful for bradycardia in children, or does CPR keep the doctors busy while the child responds to time and the drugs?

  3. I was intrigued by this post a while back, and I've been researching the subject since then.

    The grim statistics aside, why doesn't CPR work? The theories, the cardiac pump theory and the thoracic pump theory, both make some sense. It "should" work. Why doesn't it?

    All and any thoughts are welcome.

    1. THanks,
      I pay little attention to things that "sound good". So many things sound good, but aren't. It is part of our simplistic take on medical treatment. We underestimate the complexity of the situation, confounding, homeostasis, unintended consequences, etc. This is why real clinical tests are more important on theories and animal studies.

  4. Having in mind your nice comment on both Heisenberg principle and Hawthorn efects, I cannot agree with you. As intensivist, I would like to say: we have improved the results of OHCA patients, we have helped their families and we have changed ( a little) our ICU teams. Does not matter, what statistics tells us.

    1. Thanks Robert,
      For readers, OHCA menas Out of Hospital Cardiac Arrest.

      I usually warn doctors against pitting their own (subjective) observation against (objective) statistics. Granted, the studies and the numbers need to be accurate and unbiased, but medicine is full of doctors "seeing" what they believe to be true. Time after time, the real benefit ends up being far less than we perceived it to be.

      Your argument is similar to the argument I get from arthroscopists, urologists, cardiac stenters, spine surgeons, spine injecters, etc etc. When the numbers do not support our long-held beliefs and perception, we choose to not believe the numbers.

      As an intensivist, you may be interested in Scott Aberegg's blog:


Note: only a member of this blog may post a comment.