Journal club this month featured the recently published SMART trial. A cracking paper on something we do all the time, every day.
Ok its the ‘critically ill’ bit that may pique your interest most but IV fluid is something we deliver lots of in ED.
If you’re keen there was a co-published paper from the same group in the same edition of NEJM – SALTed – looking at balanced crystolid vs saline in those not critically ill. (theres a review over on St Emlyns here)
So on with the review.
Clinical Question: Does Balanced crystalloid (e.g plasmalyte) offer reduced mortality or renal ‘badness’ than normal saline?
Title of paper: Balanced Crystalloids versus Saline in Critically Ill Adults.
Published: NEJM Feb 2018
Population: Adults admitted to ICU
Intervention: Balanced crystalloid (ringers lactate or plasmalyte)
Outcome: Composite outcome comprising Death, renal replacement therapy or persistent Creatinine rise of 200%. (I’m going to call this ‘MAKE’ for the rest of the review – Major adverse Kidney Event)
We wondered about the primary outcome measure. Composite end points are OK. If they are equivalent. We thought that death and the need for renal replacement therapy were bad things. but weren’t entirely convinced the creatinine rise was necessarily as bad as those other two. Is an increase from a baseline of 50 to 150 the same as needing dialysis? Its not very Patient orientated. We think. what do you think??
The first question to ask I wonder when I read a paper is to ask whether the patients reflect my own, first and foremost is it an ED study and where were the patients, is it UK based, is it multicentre?
The answer to the latter is ‘No’. 5 ICUs were involved but this in reality was a single centre study, they were the different ICUs on one campus. At our place we have a neuro-ICU, a POSSU, a general ICU and an HDU, I guess its this sort of set up. The site was in Nashville in the US. I’ve never been to Nashville (sadly) but I often worry about single centre studies in terms of external validity. There may be intrinsic differences between patients and practice at that site compared to how we do things in the Uk. A quick perusual of their website would tell you they may run things differently there, ICUs with receiving beds, treatment beds and step down beds.
The study was powered to detect a 12% relative difference in ‘MAKE’ between balanced crystalloids and saline with an increase in required sample size once some observational data on mortality in their own units was collected from 22% predicted to 15%) meaning an increase in sample size to 14000. This is a huge number of patients. The absolute difference they were trying to detect was small at 1.9% hence the big big numbers.
Randomisation was done by site rather than individual patients which makes logistical sense. With a change month by month, the doctors, patients and everyone else involved were aware of the allocation and treatment received. The alternative crystalloid was always available on site should it be required by the treating team at their discretion. The exclusions were relative with hyperkalaemia and traumatic brain injury quoted. Unfortunately there isnt a CONSORT diagram published in the write up so we arent clear on how many patients were screen for enrolment.
They recurited over 14000 patietns and 50% of them came through ED. So they arent all ED patients, but over 7000 were. with sepsis being the most common diagnosis, equally split across the groups.
They detected a 1.1% absolute difference in favour of balanced crystalloids versus saline. With Odds ratios of 0.91 (95% CI 0.84-0.99) quoted – although they did give results as marginal and conditional ratios. None of us at JC were clear on the reasons for using one or the other. I’ve googled it, and I still don’t understand! If any of you fine readers is a statistical whizz please let us know in the comments box!
Its interesting to see they detected a difference, but bear in mind this was lower than the difference they powered the study to detect. It may have been down to chance, given the sample size.
The video below is a cracking revision of risk and risk reduction – take the time to check it out.
Its interesting to see the cumulative volumes of fluid given and the mortality outcomes across the units. Cumulative median totals of around 1.5litres in the first 24 hours. And a little over 2 litres in the first 7 days. That feels low to me. The vast majority of patients I send to ICU will have had more than this. This probably goes back to the single centre thing. It makes external validity less likely in my opinion. The mortlaity at 30 days is also pretty low for ICU at around 10%. Perhaps their ICUs admit less unwell patients than ours, or maybe it means their surgical admissions are less sick, and what we would manage on a ward. I don’t know for sure but again, it rings alarm bells.
The secondary analyses are interesting with a bigger difference between the groups for ‘MAKE’ rates with an ARR of 5.1% in favour of balanced crystalloids in patients with a diagnosis of sepsis, but remember that’s a secondary end point, which the study isn’t powered to detect, but it could form the basis of future work and it looks interesting on the surface.
All in all this was a great apper for journal club with some trend towards benefit in the balanced crystalloids group. There are some concerns around external validity, whether these patients are our patients, whether the small difference they detected was down to chance given the stuidy was powered to detect a 1.9% difference, not a 1.1% difference. We dont know how many patients were screened and not includued and why, and we also had some concerns about the primary outcome measure. Make sure you read the paper for yourself and please form your own conclusions for you and your own patients. Let me know what you think too!