Just as we were getting used to wearing our smug I-told-you-so grins about how very bad big pharma’s HES crystalloids are, the CRISTAL study comes along, suggesting colloids – even HES – are good for your shocked patients. But are they?
In light of recent findings, this trial is provocative. CRISTAL is well executed, but the design is slightly messy. So diving into it will take some explaining. Therefore the long post. Please bear with me. And please comment and discuss below.
Colloids or crystalloids for fluid resuscitation in critical care patients has been a long standing dispute. And the CRISTAL trial has been long under way. 9 years under way. It started recruting patients in 2005. It was designed for another world. It was designed to investigate if colloids in general were worse than crystalloids for ICU patients. They expected to find all colloids to be inferior, just as the then recently published Cochrane review had concluded.
Shaking things up
But to their surprise, they found no increased mortality from colloids. Actually, they found a tendency for increased survival with colloids that turned into a clear survival benefit after 90 days. But they lumped the ‘natural’ colloid together with all the dirty synthetic colloids – surely that’s skewing the results? And sure enough, albumin was skewing the results. In the subanalysis, albumine did worse than the synthetic colloids in this trial. It’s confusing. Let’s have a look.
Keep in mind the initial design of this study. As it finished in 2013, the world was a different place after 6S and CHEST. So even though subanalysis is not the perfect way to go, many were interested in seeing how each colloid had done, and in which patients. So CRISTAL split their patients into sepsis, trauma, and the rest (non-septic, non-traumatic hypotension). All colloids used at the centres involved were allowed. So they had HES, gelatins, dextran and albumine in the colloid group, sorted them into subgroups.
Half of the patients in CRISTAL were in septic shock, severe septic shock, so they should be perfect for proving the badness of HES. Also, CRISTAL got permission to include patients without consent, and could start treating them with colloids or crystalloids from the very start. These patients had very little pre-trial fluid loading.
The colloids given reflect very well on how colloids were used in the early 2000s: HES was given to 70% of the patients. Again, this big number would lead us to believe that any major HES badness should show up in the results. Second came gelatines, which were given to 35% of the patients. Albumin was given to 12% in the colloid group, but mostly due to hypoalbuminemia – just as common practice was in early 2000s. And therefore, almost the same percentage of patients received albumin in comparable amounts in the crystalloid group – evening out any big albumin effect. Dextran – well, forget about dextran in this trial. It was given to only 0.35% of the patients. So we’re really left with HES and gelatines – were HES has a big lead as colloid of choice.
So, with half the patient population in septic shock, and HES given overall to 70% of the patients – why couldn’t they measure any bad effect on the kidneys or on mortality from HES used in sepsis?
Here’s a graph over the lowest initial MAP of patients included. These patients were sick! A mean lowest MAP of 20-30!? If any kidneys should get hurt for being underperfused and then shocked with HES molecules, these kidneys should be right there. But they didn’t find any increased need for renal replacement therapy in the colloid group – nor in the HES subgroup. No increased HES mortality either.
We’ve also talked about the good effects of coloids – keeping the fluid in the vessels. But often the trials haven’t confirmed this belief. But CRISTAL seems to do so. CRISTAL found the colloid group to have fewer days on respirator, possibly due to less interstitial oedema, including less oedema in the lungs. Also they found less days on pressor, possibly due to the same factors, with more fluid kept intravasal, less leaking out.
The clinical difference
In CRISTAL, colloids weren’t given as daily supplements as in 6S and CHEST. In CRISTAL, colloids were only given as a response to strict criteria pointing to severe shock in need of fluid resuscitation. And even then, the amount of colloids given were at the discretion of the physician and colloids seem to have been largely abandoned when the initial shock phase was over. So, in CRISTAL, colloids are used pretty much as one would use them in clinical practice. These smaller amounts of colloids, and given over a shorter duration of time, might have contributed.
So CRISTAL is a more pragmatic trial. But therefore also a messier one. With CRISTAL, we might be back to doctor’s common sense and clinical judgement. But there are other trials to consider.
OK, enough with all the musings. What are we going to do in clinical life?! Well, 6S and CHEST are solid trials where the results still stand. And I will continue to be very judicious about colloid use, especially HES. But despite its design flaws, CRISTAL throws a little Molotov cocktail into the party, just to make us remember the world isn’t black and white – and the last word in the colloid vs. crystalloid debate isn’t said yet.
At JAMA’s full text page, there’s also a very interesting presentation by the main investigator Djillali Annane at the European Society of Intensive Care Medicine’s yearly meeting, ESICM 2013. It’s well worth watching for background info on the trial:
Also see our post on COLLOIDS VS CRYSTALLOIDS from Dec 3, 2012.