Most protocols for managing pediatric patients with diabetic ketoacidosis (DKA) are based on a theoretical association between fluid resuscitation and subsequent neurological decline. Although the evidence for an association between IV fluids and cerebral edema comes from retrospective reviews, for over 20 years, it is an accepted teaching principle of pediatric DKA.
Clinical Trial of Fluid Infusion Rates for Pediatric Diabetic Ketoacidosis, published just days ago in the New England Journal of Medicine, challenges this teaching with the first randomized controlled trial designed to investigate the relationship between IV fluids and cerebral edema. We review this publication and present a behind-the-scenes podcast interview with lead authors Dr. Nathan Kuppermann and Dr. Nicole Glaser from the Pediatric Emergency Care Applied Research Network (PECARN).
Podcast with Authors Dr. Nathan Kuppermann & Dr. Nicole Glaser
- Does the IV fluid infusion rate influence the neurological outcomes of children with DKA?
- Does the sodium content of IV fluids influence the neurological outcomes of children with DKA?
The trial was conducted across 13 U.S. emergency departments (EDs) in the PECARN collaborative, all of which were located in urban centers.
Enrolled patients were children 18 years old or younger who presented to the ED with DKA, defined as a blood glucose > 300 mg/dL and either a venous pH < 7.25 or a serum bicarbonate < 15 mmol/L.
Exclusion criteria included children with:
- Underlying disorders impacting mental status testing or neurocognitive evaluations
- Concomitant drug or alcohol use
- Head trauma
- GCS ≤11
- DKA already under treatment
- Pregnant women
- Factors where the treating provider felt that specific fluid and electrolyte therapies were needed
All patients received an initial 10 cc/kg bolus of 0.9% saline. This was standard DKA treatment for all participating centers. Consent occurred during the initial bolus, after which patients were randomized to 1 of 4 treatment rehydration arms as reported by this group in 2015.1 Patients were assumed to have a fluid deficit of either 5% or 10% of body weight. Patients in each group would have their deficit replaced either “fast” or “slow” and with either 0.45% or 0.9% saline. All 4 protocols were identical with regard to the other aspects of DKA treatment.
The primary study outcome was the neurologic decline to an abnormal GCS < 14 during DKA treatment. Previous work from this group used MRI to demonstrate a correlation between GCS and DKA-related cerebral edema.2 Patients who initially presented with a GCS < 14 were not included in the final analysis.
Secondary outcomes included neurological testing both during treatment and after recovery from DKA. Digit span testing and clinically apparent brain injury* during treatment were recorded, and memory testing was performed at ~3-month follow-up visits in children ≥3 years old.
* Clinically apparent brain injury was defined as deterioration in neurological status requiring either intubation or hyperosmolar therapy or resulting in death. A committee blinded to the treatment arms would then review medical records of these patients and either confirm or reject the diagnosis.
- A total of 1,389 DKA episodes in 1,255 patients underwent randomization. The demographic and clinical characteristics of the 4 groups were not statistically different.
- The vast majority (98%) of children presenting to the ED with DKA had an initial GCS of 14-15. Within this group, only 3.5% had a GCS that declined to less than 14 after starting treatment. The IV fluid treatment arm had no difference on the frequency, magnitude, or duration of GCS decline.
- There was no statistical difference between treatment arms with regard to the frequency of clinically-apparent brain injury. However, it was slightly higher in the slow rehydration arms.
- There was no significant difference in the rate of improvement in forward digit span testing, although it favored the more rapid rehydration group (p=0.06).
- Post-recovery memory testing outcomes were obtained for 855 DKA episodes, most of which were within 4 months of discharge. There were no significant neurocognitive testing differences among treatment arms.
Children with DKA were randomized to 1 of 4 treatment arms that varied in terms of fluid infusion rate and sodium content. Neither the rate nor the sodium content significantly influenced the neurological outcomes of children during treatment or at post-recovery follow up. Time to resolution of DKA and length of hospitalization were similar among treatment arms.
This study does highlights the need for future investigations into the true underlying etiology of clinically-apparent brain injury in children with DKA. By assuming a relationship between IV fluid administration and cerebral edema, an opportunity to identify other predictors (and potential interventions) has been missed. It may be, however, that these patients were more critically ill in the first-place – more dehydrated, more acidemic.
The paper also encourages us to re-visit our own hospital protocols for pediatric DKA. These are usually based off of a conservative approach to fluid administration in order to prevent us from CAUSING cerebral edema in the ED. It may turn out that this approach forces our hands into under-resuscitating these patients.
In at least some children with DKA, a more liberal approach to IV fluid resuscitation (with either a second 10 cc/kg bolus and/or faster maintenance IV fluid rates) is unlikely to cause cerebral edema. It may even improve neurological outcomes. Only time and future work from this PECARN team and others will tell.
- Glaser N, Ghetti S, Casper T, Dean J, Kuppermann N. Pediatric Diabetic Ketoacidosis, Fluid Therapy and Cerebral Injury: The Design of a Factorial Randomized Controlled Trial. Pediatr Diabetes. 2013;14(6):435-446. [PMC]
- Glaser NS, Marcin JP, Wootton-Gorges SL, et al. Correlation of Clinical and Biochemical Findings with Diabetic Ketoacidosis–Related Cerebral Edema in Children Using Magnetic Resonance Diffusion-Weighted Imaging. T. 2008;153(4):541-546.e1. doi:10.1016/j.jpeds.2008.04.048