I think we have given too little attention to the nature of the crystalloid solutions we use. Not just in neonatology, but apparently in the adult ICU also, (Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M: Association between a chloride-liberal vs chloride-restrictive intravenous fluid administration strategy and kidney injury in critically ill adults. JAMA : the journal of the American Medical Association 2012, 308(15):1566-1572) I’m a bit surprised this was published in JAMA, although it is interesting, it is not methodologically the best study, in fact it is not even a trial, it is a cohort study where the authors changed their practice in a very strict fashion, and recorded patient outcome data. They did this for a 6 months control period, then a 6 month intervention period.
First a bit about the results and the potential implications in neonatology, I will come back to the study design later.
Background: We know that normal saline has far more chloride than normal plasma, and that it’s use can lead to hyperchloremic acidosis for exactly this reason. Older patients often receive Ringers lactate, Hartmann solution or other lower chloride solutions, but there is very little use of these solutions in newborns. In my unit very often arterial lines in the past were filled with normal or half normal saline. I think neither of these are optimal, saline has relatively far too much chloride, and in fact we should avoid sodium in the first 3 or 4 days of life, as RCTs have shown worse outcomes when more sodium is administered.
Half normal saline is probably not a good idea, you give less sodium, but hypotonic solutions are problematic as when the blood is withdrawn into the catheter and syringe, there is a mixture created of the hypotonic solution and the blood which leads to lysis of the red cells. So although you give less sodium you end up giving more potassium and free hemoglobin when you re-infuse the dead space. We now often use a solution of sodium acetate in the arterial lines, which may be variably hypotonic, this has the advantage of giving less chloride and giving a buffer in its place, but there has been little study of acetate in the newborn.
To get back to this new study, the authors noted that high chloride administration in experimental animals causes renal dysfunction. So during the intervention period they changed their usual fluid administration practices, eliminating normal saline and 4% albumin and using lower chloride solutions (Hartmanns solution, plasma-lyte and 20% albumin) instead. They found significantly less acute kidney injury, and less need for renal replacement therapy (dialysis and the like).
I think this does not have direct implications for our practice, but should stimulate us to think a bit more about the fluids that we use, and whether they are optimal or not.
Finally to return to the study design, I think this is the worst of both worlds. I know it is a pain to design randomized controlled trials, and it can be horrendous getting them through the ethics review committees, but this study (which was approved by the ethics committee without the need for consent) could have been much more important, had much more impact and saved (or not depending on the findings) many future adults in ICU from kidney failure if they had randomized the subjects. As it is, people will discount this because they didn’t do it right; the patients were exposed to exactly the same risks as if they had been in an RCT, and people will go on using high chloride solutions, or will change to lower chloride when we really don’t know the probable effects.
An editorial in JAMA accompanying this study noted “A more definitive study examining the composition of intravenous fluids is now justified and necessary to scientifically test for potential adverse effect” how much better if they had done that straight away.
Neonatal Research 