Written by Clay Smith
Spoon Feed
Hyperchloremia (an increase ≥5meq/L) in the first day of PICU admission was independently associated with 2.3 times greater odds of in-hospital death in critically ill children.
Why does this matter?
We know lower chloride, balanced IV fluid reduced major adverse kidney events at 30 days (MAKE-30) in both critically ill adults (SMART), and in ED patients (SALT-ED). The benefit in the composite MAKE-30 outcome was largely driven by mortality improvement in SMART. This study explored whether hyperchloremia might be associated with harm in critically ill children.
Chloride kills
This was a retrospective study of 1,935 children 90 days to <18 years admitted to the PICU, DKA excluded. They found that if the chloride increased ≥5meq/L in the first day (12% of patients), odds of in-hospital mortality increased 2.3 times after adjustment for known confounders. Although causality can’t be proven in this study, the benefit of balanced fluid has been proven in adults: SMART and SALT-ED. So, it seems that use of balanced, lower chloride fluids would be a good option in critically ill children as well. Of course, hyperchloremia could simply be a marker of receiving more IV fluid, which indicates greater illness severity. But adjustment for such confounders makes the 2.3 times greater odds for mortality in hyperchloremic children believable as an independent risk factor.
Source
Increase in chloride from baseline is independently associated with mortality in critically ill children. Intensive Care Med. 2018 Dec;44(12):2183-2191. doi: 10.1007/s00134-018-5424-1. Epub 2018 Oct 31.
Open in Read by QxMD
I have several comments about the analysis and conclusions of this research.
1. Although balanced fluids with lower (and more physiologic) chloride might be best, we must keep in mind that hypotonic fluids are associated with increased morbidity in hospitalized kids.
In addition, the analyses do not adequately adjust for potential confounding by fluid balance. They adjusted for fluid intake only for Day 1 in the PICU and did not adjust for fluids given before PICU admission or on subsequent days. Placing critically ill children in positive fluid balance is associated with adverse outcomes. It’s understandable that it may have been difficult to ascertain fluids before PICU admission, but they could have recorded fluid volumes after day 1 and adjusted for this.
As such, fluid balance before and during the PICU admission is an unmeasured confounder that, if adjusted for, might meaningfully change the aOR’s for chloride levels/elevation on Day 1 on mortality.
There were 71 deaths (the outcome of interest) and 16 covariates in their regression model (by my count in the Methods section, but it appears to be 21 in table 2), with at least 16 degrees of freedom for these covariates. A statistical concern is whether the model is overfitted to their dataset and thus not externally valid (no more than 4 outcomes per df rather than the 10-15 outcomes/df that is generally accepted for a stable model). The aOR’s of this study might thus be quite different in a cohort of similar patients.
I suggest "kills kids" is not the best term because it implies causality. At best the authors noted an association, but it could just as well be reverse causality: Greater illness results in lower chloride levels and greater risk of mortality.
This study nonetheless makes clear that we all need to reassess how much and what kind of fluids we are administering to ill children. For example, the evidence to date suggests that we should be volume-expanding kids with bloody diarrhea because if they turn out to have ETEC and develop HUS, morbidity and mortality are much improved.
Love JournalFeed!! Keep feeding us.
Don Arnold
Vanderbilt
Don, thanks for this comment and your statistical expertise and insightful critique. You’re absolutely right that the title was an overstatement and could have been more nuanced. It’s tough to be pithy and nuanced. Tomorrow’s article will touch on your comment about hypotonic fluids when we cover the AAP statement on this.