As in Incident command...

IC established!
We'll have several different sections reporting in - recent research, local topics, or highlighting areas of the Sponsor Hospital Council of Greater Bridgeport protocols.

*** Keep in mind - this website does not replace your protocols, and these posts do not reflect SHCGB or Bridgeport Hospital policies. This is a place to discuss research, controversies, or discuss possible future protocols. When in doubt, check your current protocols through the official source.

Wednesday, April 3, 2013

Hyperoxia during CPR associated with improved survival

There has been some great discussion on the web recently about the potential danger of hyperoxia. Mike McEvoy was interviewed on EMS 12-lead about the emergency and ICU evidence, and a vigorous discussion took place on EMTLIFE about the same topic. Rogue Medic weighed in as well, asking the question "How many hundreds of thousands of patient have we killed with oxygen and our refusal to require evidence of improved outcomes?" 

In the midst of this heated dialogue about reactive oxygen species, a recent study was published that may be an important addition to this discussion. 

The new paper
The study, "Increasing arterial oxygen partial pressure during cardiopulmonary resuscitation is associated with improved rates of hospital admission," was conducted in an Austrian EMS system. The ambulances are staffed with physicians, and are equipped with portable ABG analyzers as well. They retrospectively analyzed all non-trauma cardiac arrest calls over a 7-year span, and found 145 patients that had received ABG analysis of the PaO2 during the code. The ABGs were obtained after compressions, intubation, and 100% oxygen had been started. 


After dividing the patients into low, intermediate, and high levels of PaO2, they examined which patients had survived to hospital admission (HA), as well as the "cerebral performance category"(CPC) of the longer-term survivors. About half of the 145 cardiac arrest patients with an ABG had ROSC.



It turned out that patients with intermediate (61-300 mm Hg) or high (> 300 mm Hg) levels of PaO2 were incrementally more likely to survive to hospital admission. This positive association did not extend to showing a significant improvement in cerebral performance in survivors, however, despite a suggestive trend.

How does this fit with prior studies?
One retrospective study conducted in the ICU demonstrated an association between post-ROSC hyperoxia and increased mortality, while a similar trial showed no consistent association. No other trials have looked at pre-ROSC PaO2, however.

The accompanying editorial attempts to explain this seeming paradox - that hyperoxia pre-ROSC increases survial, but worsens survival post-ROSC - but I wonder if it is premature to try and reconcile these studies. Given all the known limitations of retrospectively-obtained data, and of this trial in particular, perhaps we should await controlled trials that more clearly define the role of oxygen levels and survival. As the authors note, "Reasons for the benefit of higher oxygen tensions during CPR can more easily be hypothesized than explained." Given the conflicting data, it might behoove us to proceed cautiously in modifying the targets for oxygen delivery in cardiac arrest.

7 comments:

  1. I have not yet read the article, but the sample size raises some questions. How did they choose which patients had ABG draws?

    I would also like to see a comparison of ABGs from patients with compression-only CPR and CPR with ventilations (continued through to survival and neurological outcomes, as well).

    And then there is the possibility of improved outcomes with stutter CPR.

    http://emcrit.org/podcasts/future-of-cpr/

    The only reason I wrote hundreds of thousands is that I can't spell billions.*

    * Cut and pasted from Carl Sagan. He wasn't using it.

    ReplyDelete
  2. Indeed, how did they choose which patients to get an ABG? It likely wasn't random. Patient who got ABGs had bystander CPR more often, and more interventions. Also they were more often men.

    Most significantly was the ROSC rates - 30% for the non-ABG patients, but 50% for the (+) ABG cohort. This suggests some baseline differences.

    Nonetheless, this is the data we have, not the data we want, or need. Until we get some large, randomized human clinical trials, we're going to be stuck with it.

    ReplyDelete
  3. It reminds me of the high dose epi studies of the early 1990s. A short term benefit in ROSC doesn't seem to translate into a good neurological out come at discharge.

    I'd say that the data is so skimpy that there is no reason at all to change clinical practice at this point.

    ReplyDelete
  4. Retrospective data can be used to draw whatever conclusions the author is looking for. Agreed with what you've all said so far. Definitely baseline differences between ABG & non-ABG patients. Interesting nonetheless though.

    ReplyDelete
  5. Replies
    1. Not sure my comments are getting through your CAPTCHA system...

      Here's what I (tried) to write on your post:
      *******

      The importance of this study is that it may suggest that hyperoxia may be beneficial at different points in critical illness, and that we need to be careful before acting too broadly.

      Perhaps there is an analog to the past studies that looked at the treatment of sepsis. Intial studies of sepsis, using aggressive fluid resus and pressors, were conducted in the ICU, hours after admission. Although these studies were based on strong physiologic evidence, the clinical studies were negative. However, when these same intervention were started in the ED, minutes after arrival, we found a huge drop in mortality. Timing matters.

      And so it may be with oxygen. It doesn't help that most of the studies out there are retrospective, and so are suggestive, but they require confirmation with prospective trials. Some "common-sense" changes in practice have already been implemented, but we need to be careful before extending this reasoning to all clinical situations.

      Delete