Small is Beautiful Emergency Medicine
N Engl J Med 2019;380:2529-40
(Hard copy available in hospital library, online access via library website from next month. Or register at NEJM.org to get free access to 3 'closed' papers a month.
Application of High-Sensitivity Troponin in Suspected Myocardial Infarction. Neumann JT, Twerenbold R, Ojeda F et al. June 27th 2019
I can't remember how I came across this study, but I read it hoping it may inform what I currently do in ED - which is use the 2015 European Society of Cardiology guidelines (here, well worth a read, figure 2 sums up biomarker use). Well, most of the time. Gestalt still plays a role.
I'm not doing a 'full' appraisal this time - so feel free to offer one via the comments. In short, it's a part Roche and Abbott bankrolled (I'm pretty sure) mega study that aggregates prospectively obtained individual patient data from 15 US/Australasian/European studies. These patients were 'suspected of having MI' clinically, and had 2 or more troponins done (I or T, high sensitivity). STEMI patients were excluded. The study includes data for 22,650 patients, 9604 in the derivation study, and 13407 in the validation study, obviously this is key in a paper studying a test. There was a 15% prevalence of NSTEMI overall - I don't know what 'our' point prevalence is but this sounds about right.
The aim of the study is great - it aspires to be able to do away with rigid troponin-taking protocols (e.g. time 0 and time +1/2/3 whatever hours) and risk stratify your patient whatever data you have for them. You do need 2 results however. They grouped these results into 'early' - 0 and <2 hours, and 'late' - 0 and >2 hours 1 minute.
Their stats are bound to be complicated given study size, and I don't really understand how they got there, so I can't question their process. But essentially they looked at multiple cutoff ranges to define 'low risk' and 'high risk' of death or MI at 30 days. Most of the conglomerated studies recorded this - some went out as far as 2 years. But death or MI at 30 days is certainly an outcome of interest to a risk-stratifying emergency physician.
They then developed hard to understand (for me, anyway) 'compasses' or wheels so you can take a patient's initial troponin, interval change, and work outwards to a 'proportion of patients who are low risk if below those figures' and eventually a risk of death or MI at a month. Your patient will 'fit' into either the low, or the high, risk compass. I think, from playing around with various hypothetical patients.
www.compass-mi.com is their website with an algorithm where you can plug in your data sets. This is helpful, and easier to do than trying to fit your patient onto one of the compasses, but it's confused me further as it seems to only make them 'low risk' if the results are below the figures you have for your patient... it gives a risk of being high risk, rather than stratifying them for you, if that makes any sense at all.
Perhaps it's my simple ED brain, but this isn't terribly helpful and won't impact my practice. Maybe it should - I've invited 2 cardiologists to comment.
I'd love to know your thoughts as an ED team member too, please click on 'comment' just below the title, I'll validate it once submitted and add to the thread - they aren't published immediately to reduce spam.
In 2019's relaunched journal club, we'll go after answers to clinical queries, interspersed with 'classic' or recommended papers. Recently, the above question has been raised regarding treatment of ED patients. My practice habit is to use balanced crystalloid (Plasma-Lyte in our ED) for sick patients, and normal saline for the less sick patients. Based on having seen large volume fluid resuscitation with saline cause hyperchloraemic metabolic acidosis.
The question has recently been raised again due to a shortage of Plasma-Lyte in our trust.
So two papers for you to peruse and consider, recommended by one of our intensivists. There's significant ED experience/expertise on this topic - please add your thoughts & other references via the comments!
My appraisal & thoughts:
critically ill patients paper
Strenths and weaknesses
- Big (15000 patients) study, in a single big academic centre (probably the only avoidable weakness of this study). Lightly patient - relevant primary outcomes stated prior to the study (death from any cause, initiation of renal replacement therapy (RRT), or persisting renal dysfunction at discharge or 30 days post ICU admission).
- Secondary outcomes were also stated prior to the study.
- Randomised by month, including ED stay prior to ICU admission. Not blinded, but results extraction from patient records was.
- Power calculations were done, then adjusted pre-study, seem appropriate.
- 'Off protocol' permitted for patients with hyperkalaemia (as balanced crystalloids contain K+) and traumatic brain injury (where balanced crystalloids may worsen cerebral oedema). Intention to treat analysis.
- No industry involvement declared, and 2 types of balanced crystalloid used. Would be good to have more specific results, study approach maybe means greater generalisability of results, as apply to more than one 'brand' balanced crystalloid.
- Really readable & clear paper. Authors should be commended. I recommend you read it as an example of a well constructed study.
Approx. 10% less hyperchloraemia and 10% less bicarb drop in the balanced crystalloid group, with the effect increasing as the amount of fluid given increased. Interesting but not a patient-oriented outcome. However...
- Study found an odds ratio of 0.9 in reduction of primary outcomes listed (95% CI 0.84-0.99) giving balanced crystalloid instead of normal saline.
- Breaking this down a bit further it's 0.8% mortality reduction, 0.4% reduction in RRT, and 0.2% reduction in persistant renal dysfunction. All statistically significant. One patient per 94 admitted could avoid these outcomes if balanced crystalloid were used in all patients.
- Very similar outcomes using various predetermined sensitivity analyses. Authors excluded various groups from analysis who could affect result validity - for example those patients in ICU in a 'crossover month' where they were exposed to both crystalloids.
- The NNT seems high, and 0.8% mortality reduction sounds a small effect. But the outcomes are so important - death obviously, but RRT and persistent renal dysfunction are highly resource consuming, invasive, and quality of life reducing, that this result matters.
- A litre of normal saline costs £3.59 according to the BNF - but the price your trust pays may be very different as it's all negotiated...
- A litre of Plasmalyte cost is not in the BNF/ on medicines.org/on the manufacturer website.
- A conversation with a pharmacist at our trust clarified that the trust negotiated price for Plasma-Lyte is comparable to that of the trust price for normal saline.
- In sepsis - the outcomes were more marked. There was a 4% difference between in hospital mortalities, favouring balanced crystalloid. This was a secondary outcome, so don't hang on to hard to that figure, but it was predetermined (i.e. there was no data-mining looking for positive correlation in this study).
non critically ill patients paper - the ED paper
Strengths & weaknesses - are essentially the same as above. Single centre is key weakness here. 13,000 patients - who were then admitted, but not to ICU, hence not 'critically ill'
- primary outcome was hospital free days (alive after discharge before day 28).
- secondary outcomes were essentially the same as the primary ones in the study above.
- median fluids given in ED was 1 litre. Excluded those who received less than 500mls in the ED stay. That's quite a lot for 'noncritically ill' patients. They may have just reached their ceiling of care, but I think this is less likely in US practice than the UK.
- was 95% Ringer's lactate used in this study, 5% Plasma-Lyte. We don't have Ringer's lactate in our ED. This may matter... but probably doesn't.
- no difference in length of hospital admission.
- Similar, perhaps slightly greater effect, on the secondary outcomes - death/RRT/persistent renal dysfunction - OR 0.82, 95% CI 0.7-0.95. NNT of 111.
Bottom line - I'll continue use balanced resuscitation fluids in adult resuscitation, excepting hyperkalaemia and traumatic brain injury, and DKA till the protocol changes. I think I'll start using them in majors too, as there appears to be a mortality benefit, once availability is no longer an issue.