September 6th, 2012

Resuscitating Resuscitation


Patients at hospitals where resuscitation attempts lasted longer had higher survival rates according to an observational study involving patients with in-hospital cardiac arrests between 2000 and 2008 published in this week’s Lancet. Two of the study authors, Drs. Zachary Goldberger and Brahmajee Nallamothu answered  CardioExchange’s questions. 

What questions or comments do you have about this provocative study?

 It is widely believed that prolonged attempts at resuscitation are not beneficial. What was the impetus for conducting your study?

As you mention, many practitioners remain reluctant to continue resuscitation efforts for too long when return of spontaneous circulation does not occur early during the course of an arrest. The question there is how long is too long? Unfortunately, there are few empirical data to guide practitioners on this question. It is an extremely difficult decision to make during the course of an arrest because of its obvious implications – i.e., the patient dies when you stop.

How did you frame your analysis?

We initially focused our attention on survivors. Nearly 50% of the patients in our sample survived the arrest, and approximately 87% of those achieved a return of spontaneous circulation by 30 minutes. We were a bit surprised that this wasn’t a higher percentage, and this suggested that some patients only survive after longer resuscitation efforts. We then focused on non-survivors, and found that fewer than 23% of those who died were resuscitated for at least 30 minutes. So our message here was that there are clearly a non-trivial number of patients who need more than 30 minutes to achieve return of spontaneous circulation, and attempts in most patients are rarely extended for this long. We also found that the length of resuscitation attempts in non-survivors varied substantially across hospitals.

Your analysis is unusual because it focuses on a hospitals’ length of attempts in non-survivors to examine survival rates. Can you explain this?

This is an important point. We were able to arrive at our main finding largely because of variation among hospitals’ practice patterns for resuscitation duration in patients who ultimately didn’t survive (i.e., non-survivors), prior to pronouncing a death. In essence, we examined whether the predilection for “how long” a hospital attempts resuscitation is related to how likely a successful outcome would be in their patients. We found that a patient’s likelihood of survival was higher at hospitals that, on average, resuscitated non-survivors for a longer period of time.

Now it is important to note that hospitals in the quartile with the longest resuscitation duration had more than 50% longer attempts than those in the shortest quartile (25 minutes versus 16 minutes). While it is a seemingly small difference in time, these additional minutes have substantial implications in critically-ill patients when physicians are trying to evaluate clinical responses and provide additional treatments.

It seems particularly surprising, and encouraging that neurological outcomes did not differ between the hospital quartiles.

This is good news since a major concern is that prolonged efforts might be leading to higher rates of survival at the expense of worse neurological outcomes. One explanation is that these are in-hospital cardiac arrests, so chest compressions and other supportive measures were likely to be continuous during the entire event.

So, can your work recommend an optimal duration for which to resuscitate hospitalized patients?

That’s a great question, and it is likely that most readers will want to have a simple answer. But we don’t think we are there just yet. We need to be cautious about interpreting these findings. We cannot identify an optimal duration for all patients. How long to continue resuscitation efforts for any individual will depend on a number of patient and arrest-related factors. It will therefore continue to remain a bedside decision that requires clinical judgment. Furthermore, it needs to be stated that we identified an association. This may or may not be causal. For example, hospitals with longer attempts may better in other aspects of resuscitation care, such as their implementation of standardized protocols. We hope this paper will spur future work in in this area.

Ultimately the duration of resuscitative efforts comes down to a bedside decision. Overall, we hope that these data will make practitioners (including ourselves) more aware that some patients may benefit from more time, and – faced with a hospitalized in cardiac arrest – ask whether the patient in front of you might be one of those cases.

4 Responses to “Resuscitating Resuscitation”

  1. William DeMedio, MD says:

    My own anectdotal observation is that the longer a resuscitation attempt goes on the lower the likelihood of survival to discharge. The results surprised me.

  2. William,

    Thanks for your note. We absolutely agree. And our data show the same thing: the longer an attempt goes on, the lower the likelihood of survival to discharge. This is largely due to the fact that sicker patients are likely to take longer to resuscitate before a response is seen.

    So our question wasn’t aimed at whether or not someone who achieves ROSC at 15 minutes does better than someone who achieves it at 30 miuntes. We actually were trying to get at whether or not trying for longer (i.e., extending efforts from 15 minutes to 30 minutes) might make a difference. If clinicians were doing a perfect job at selecting those patients on whom to stop efforts at 15 minutes, we would have found little difference across the quartiles of hospitals stratified by their length of resusication attempts in non-survivors. But we didn’t, and so we think there may be some instances where patients might benefit from longer efforts. But obviously, this is just a first step (and a limited view at that). More empiric work is needed.

    Best, B

  3. Edgar Abovich, MD says:

    Cardiac arrest patients are a mixed bag with numerous medical problems not evenly distributed between the hospitals.It’s possible that you are comparing apples and oranges. The length of the efforts in non-survivors could vary depending on things like age, underlying medical problems, and even the type of physician in charge of resuscitation.
    The success may also depend on whether it occurred on an unmonitored or monitored floor. Sure there are patients who can benefit from longer resuscitation, we have all seen that, but it should also be left at the discretion of physician and not enforced by protocols.

  4. Ed,

    It is always tough to argue that observational data like these are definitive. As you correctly point out, the potential for residual confounding remains despite our multivariable modelling. Obviously, we should be cautious. But leaving such an important decision to discretion without an attempt to further understand these data I feel would be a mistake. For example, I believe that studies that examine the value of instituting at least a threshold of time for efforts or novel physiological predictors of prognosis obtained during the arrest (e.g., continuos waveform capnography) are possible to design and could help guide us. Leaving it to our gut instinct may be what we are left with at the end, but I think clinicians would welcome an exploration of “when long enough is long enough”.

    Best, B