JC: Balanced fluids vs Saline on the ICU. The SMART trial. St Emlyn’s

There has a been a little flutter of activity in the #FOAMed world this week about two trials published in the NEJM on the subject of balanced fluids in the care of critically ill patients, and also on admitted patients in US hospitals. The debate about balanced fluids vs. saline is fairly long standing, the view here at St Emlyn’s has been that balanced fluids (e.g. Ringers, Hartmans, Plasmalyte) offer a more physiological approach to resuscitation and fluid replacement. There is also clear evidence that saline causes a hyperchloraemic acidosis and some limited evidence that it may increase the need for renal replacement therapy.

However, the picture from an EBM perspective is hardly definitive which in a way is a bit of a surprise as we use literally tons of this stuff! IV Fluids are arguably one of the most common interventions we deliver and yet the evidence base is weak at best.

A few years ago we reviewed the SPLIT trial^{1 2}which was a randomised controlled trial comparing plasmalyte against saline. It showed no really significant difference but may have been too small to be definitive. That group is now working on the PLUS trial  ³ which is a much larger and hopefully definitive RCT on the use of balanced solutions in the ICU.

What then in the meantime? Well there is more evidence out this week in the NEJM. There are two trials from the US looking at the use of balanced vs. saline solutions. One on ICU⁴ and the other in ward admitted patients⁵. This blog deals with the ICU paper. The abstract is below, but as we always say it’s important that you read the full paper yourself.

What kind of study is this?

It’s an interesting design to be honest. The trials was designed to run across 5 ICUs with randomisation done by the site the patient happened to be on. So if you are admitted to site A then you get Saline, site B Plasmalyte and so on. Every month they swapped around and started doing the opposite. So it’s the site that is randomised and not the patients. Technically this is a randomised controlled trial, but don’t mistake it for an individual randomisation trial. There are significant differences and biases that may result.

Paul Young who authored the SPLIT trial and is working on the PLUS trial also noticed this about  the randomisation which can also add bias to a study.

It seems problematic to me that #SALTED has post randomisation eligibility criteria. How do we know fluid choice did not alter chance of hospitalization or the kind of people hospitalized? Perhaps I have missed something.

— Paul Young (@DogICUma) March 2, 2018

Tell me about the patients.

This is an ICU study. Great you say, I know what ICU is……, but do you. Admission rates to ICU, patient populations and groups vary enormously around the world. In this paper the predicted in hospital mortality for the ICU admitted patients was about 9-10%. That’s considerably lower than in the UK and thus may be a rather different population of patients to our practice here.

In terms of numbers and power then this trial is far better than past attempts at answering this question. 15802 patients took part in the study which is vastly superior to any previous trial. Those numbers bring significant power to the study and the ability to detect small differences in outcome.

What were the main outcomes?

The main outcome was the need for Major Kidney Events (such as renal replacement therapy (RRT)). That’s a reasonable outcome as it has been one of the major concerns about the use of saline, but it’s probably not as patient centred as we would like. It’s also a combined outcome that can cause confusion about what’s really important here. IN this case it’s a combo of death, RRT or creatinine rise. That’s OK and relevant but I’m most interested in mortality outcomes and longer term function amongst ICU patients as opposed to whether they required a specific therapy. I do understand that RRT is really important and it is a proxy marker for badness, but still it’s not the same.

They did look at mortality too and they looked at some sensible and a-priori sub-group analyses around conditions such as sepsis and in higher and lower mortality bands.

Was this a blinded trial?

No. The cluster design meant that everyone knew what the patients were getting. Such a design is easier to deliver but it does have problems like unblinding. You might think it would not make a huge difference, but it’s exactly the sort of thing that might differences to decsions about starting RRT.

What did they find?

The headline figure is that there is a small but statistically significant difference in Major Kidney Events in favour of balanced solutions. It’s not a huge difference 14.3% vs 15.4% but if that were extrapolated across the global ICU population it would have a significant impact.

Secondary outcomes also leaned towards a benefit to balanced solutions. Overall mortality differences and rates of long term renal function and RRT were not statistically significant but in favour of balanced solutions. Notably in septic patients the authos found a statistically significant difference, with a 25.2% vs 29.4% rate of mortality, but this should be treated with caution as it is a sub group and not a primary aim of the study.

It’s also interesting to note that the volumes of fluid infused was actually pretty small for most patients (reflecting the fact that they are probably less sick than a UK cohort). In a subgroup analysis of patients with greater volumes of infusion then the differences in outcome were again in favour of balanced fluids and with a greater treatment effect.

So what now?

Here in Virchester we happily submit to our confirmation bias that this proves what we’ve said all along. In that regard we think along the lines of Josh Farkas (read his excellent blog here). Balanced fluids are better, but wait. It’s all too easy to believe the outcomes of a study that fits with your a-priori bias. Yes there appears to be a benefit associated with the use of balanced solutions but the difference is small and the trial design could have led to some bias introduced into the data.

Having said that this is further evidence that the patho-physiological argument around balanced fluids is unlikely to be harmful and may be beneficial. On that basis we are going to carry on using balanced solutions until we hear otherwise.

Finally you really should read John Myburgh’s editorial on this. It clarifies a lot of the concerns about the state of fluid resuscitiaton and the need for patient centred outcomes in critical care research⁶. I strongly recommend you read it.

You should also follow Paul Young on twitter if you are interested. His twitter based analysis of this paper is excellent, and I’ll leave you with his final thoughts (which pretty much agree with ours)

#8. Bottom line for me is that the effect of these fluids on 'patient centred' end points still need to be established in high quality double blind RCTs like #PLUStrial and #BASICStrial but that outside the conduct of these trials favouring balanced crystalloids is reasonable

— Paul Young (@DogICUma) February 28, 2018

You should also read the following blogs on the same topic.

• SMART on WICS⁷
• The case of the unbalanced solution⁸
• Josh Farkas PulmCrit⁹
• Synopsis of both trials over at Rebel EM by Salim Rezaei¹⁰

vb

S

Before you go please don’t forget to…

• Subscribe to the blog (look top right for the link)
• Subscribe to our PODCAST on iTunes
• Follow us on twitter @stemlyns
• PLEASE Like us on Facebook
• Find out more about the St.Emlyn’s team

References