Journal Club was back this week and we had a great group followed by a team ED night out which was great for the wellness and team cohesiveness ahead of what promises to be a busy winter in NHS-land!
The paper this month was initially flagged up to me by my awesome colleague Steve Goodacre who is ‘Mr Chest pain’ round these parts. Thanks Steve. It made me think about th types of patient we see with Acute MI, and what we do for them.
Clinical Question: Does giving oxygen to patients with Suspected Acute MI improve their mortality?
Title of the paper: Oxygen Therapy in suspected Myocardial Infarction.
Published: NEJM 2017
Population: Adults >30 presenting with symptoms and investigations consistent with Myocardial infarction
Comparison: Ambient Air
Outcome: 1 year Mortality
This was a multicentre, open label, registry based controlled trial. Patients were recruited at sites across Sweden and could present the ambulance service, the Emergency Department or to the Coronary care unit with symptoms of ACS.
They specifically included patients who were
Had Chest pain or SOB symptoms for <6 hours And ECG changes consistent with MI or a positive Troponin AND Oxygen saturation of >90%
Had to be Swedish as they were being followed up on the SWEDEHEART registry (a national registry of all the Swedes and their healthcare data – i assume like a centralised computer database of the entire population
Ongoing oxygen therapy
Oxygen prior to enrolment WAS allowed, but had to have been for <20 minutes and had to have a 10 minute washout period prior to enrolment. This is entirely reasonable and exactly the sort of patient we want to know about, I think. Although this group is a much narrower than 'all comers with suspected MI' given that we know they have ECG changes or they have a positive troponin, that being said I reckon this cohort represents who i want to know about. I assume for reasons in line with 'first do no harm' they set a lower limit of oxygen saturation at 90, as it would be hard to justify no giving oxygen below this level except in pretty specific circumstances. Now the patients were all Swedish. I've nothing against the Swedes, BUT, this cohort doesn't feel similar to my MI patients. The patient had characteristics are fascinating. Mean BMI was only 27, diabetes rates 17%, and smoking 21%. Now these characteristics were spread similarly across the groups so the randomisation spread the confounders fairly BUT think about your patients.. I bet they're larger and smokier....
(other Swedes and football teams are available, however I reckon this is the best combo!)
There was some great stuff methodologically, with computer based randomisation and intention to treat analysis being a highlight. They also made a pragmatic adaptation to their power calculation which we liked – adding extra patients to the number ‘required’ to detect a 20% Relative risk reduction from a 14% baseline mortality, knowing that some of the patients would go off protocol and need oxygen therapy due to the natural history of the disease.
The intervention and the comparitor were open label and were clearly different – 6 litres oxygen via an open face mask compared to ambient air. They do discuss the idea of a ‘sham mask’ but were worried about CO2 re-breathing, and also note that there is no compressed air on Swedish ambulances. Given the end point studied was mortality I think its reasonably that this was the approach they took. If it were subjective measure such as use of oxygen to change a breathlessness score then this would potentially have influenced the outcome.
Results: They enrolled 6629 patients across the two groups with no difference detected in mortality at 1 year (5% Vs 5.1% and a HR of 0.97 (0.79-1.21)). This is interesting in that they hypothesised a 20% relative reduction, but from a baseline mortality of 14%. This was based on the SWEDEHEART registry data for years preceding the trial. However, it’s worth noting this was all comers with MI. Which include those who have sats <90 and those who'd had a cardiac arrest, both of which were part of the exclusions for this trial, selecting out a sub group of patients who on the surface of things are much sicker than the patients studied here. I wonder then, how many patients they would have needed to detect a difference given a far lower baseline mortality? Our conclusion was similar to the authors. There doesn't seem to be any benefit in giving oxygen to patients with presumed norm-oxia and suspected MI. (But there also doesn't seem to be any harm ensuing either!) We'll be turning the oxygen down and off in patients on it with normal saturation in supected MI, however we wont get cross if they happen to be on it. Make sure you have a listen to the RCEMLearning team discussing the paper here and the awesome as always BottomLine here
Thanks for reading and keep fighting the good fight