DKA and Cerebral Oedema – Do we really know the cause?

Subscribe to Resus

Get the latest updates on our Conferences PLUS our Webcasts and Education Newsletters. And it's FREE!

  • We respect your privacy
  • This field is for validation purposes and should be left unchanged.

Register Now:

Learn More

Cerebral oedema is the most feared complication in children presenting with Diabetic Ketoacidosis (DKA). It occurs in about 1% of cases but has a mortality rate of up to 90% (Waldorf J et al Diabetes Care 2006; 29:1150-9). Patients will have a decreased conscious state and may also have cranial nerve palsies, headache and/or bradycardia and hypertension.

Its incidence has remained the same since it was described in 1936 and although we have clues as to what may contribute to it, and we know that some patients have subclinical cerebral oedema at presentation(Krane et al NEJM 1985;312:1147-51), we still can’t predict who will get it, nor greatly affect its high rate of mortality.

There are theories of causative factors, most of which are vasogenic or osmotically based, but there is really no great evidence out there. The studies are small or retrospective, or both.


One theory relates to osmolytes accumulating in brain cells. These are the compounds that maintain normal cell volumes. As extracellular osmolality decreases rapidly with treatment, water flows rapidly onto these cells causing the brain to swell.

Another theory relates to Na+ / H+ exchanger, such that a correction of acidosis  results in Na and water passing onto the brain cells, resulting in oedema.

How much has been proven to be the actual mechanism? By my reading, not much.


We believe that the following increase the chance of developing cerebral oedema:

  1. Newly diagnosed diabetes
  2. Duration of symptoms
  3. Rapid change in Osmolality (2xNa + U + Glc):                                                                      – ie rapid decrease in Glucose and a decrease in Sodium
  4. Initial pH < 7.1……………. and in more recent studies (Glaser N et al NEJM 2001;344:264-269)
  5. Decrease in CO2
  6. Increased Urea
  7. If treated with Bicarbonate


FLUIDS: If the patient is hypotensive and crashing then fluid resuscitation is necessary. No question about it. However, how much we give may matter. In some of the work published there is an indication that more than 40ml/kg given over the first 4 hours may predispose to cerebral oedema.

We seem to be conditioned to give 20ml/kg to every child. My approach is a little less aggressive. I aim at 5-10ml/kg. I then review and give it a little time ( that may only be minutes) before more fluids are given. The point is, unless the child is crashing we don’t have to give large volumes, we can pull back a little.

INSULIN AND GLUCOSE LEVELS: Large drops in glucose, decrease osmolality. If this is done rapidly, it can, according to the osmotic theory cause cerebral oedema. Hoorn et al (J pediatr 2007:150: 467-473), found that a drop in osmolality from 304 +  5 to 290 + 5, was associated with increased cerebral oedema.


The evidence isn’t great. What we see is that the more acidotic you are, the more deranged your electrolytes, the more abnormal your physiology, the worst you do. The sicker kids did badly. Common sense.

Firstly it’s important to say that it’s very rare that you will see one of these cases unless you are working in an ED that sees lots of kids. That’s why its even more important to know what to do, because its not that frequent.

If a child is sick and haemodynamically crashing, then I give fluids. I don’t give 20ml/kg of normal saline right away. I will give 5-10ml/kg and then review. I can always give more if needed.

I don’t give a bolus of insulin, I start and infusion an wait for gentle lowering of the blood sugar i.e.., < 5mmol/L drop per hour. Things improve over hours. Nothing needs to happen quickly here.

If I think a child has cerebral oedema, mannitol 1g/kg IV is the treatment of choice. You must discuss these patients with endocrine and ICU. If the conscious state deteriorates to the point where protection of the airway is needed, or if airway control is necessary to paralyse and lower intracranial pressure( if this indeed assists in this case), then beware not to hyperventilate the patient. Hypocapnia from mechanical ventilation has been associated with this condition(Glaser N et al NENM 2001;344:264-269)

Cerebral oedema a difficult condition to treat and the outcome is not good. We don’t really know if our treatment contributes to the development. However, we need to do everything we can to minimise any possibility of causing the problem.

Peter Kas

Dr Peter Kas

Emergency Physician, Educator. Key Interests: Resuscitation, Airway, Emergency Cardiology, Clinical Examination. Creator


  1. Muthiah on 13/06/2017 at 8:16 am

    Isn’t hyperventilation a strategy to reduce ICP by decreasing blood volume in cerebral veins? If so, how is hyperventilation harmful in dka in children?

  2. admin on 21/06/2017 at 4:09 pm

    Hyperventilation has been a strategy used to decrease ICP by directly decreasing PCO2 and raising vasoconstriction–in the past. You are correct. However this hyperventilation is done very short term and in special circumstances i.e.., in the acute deteriorating patient where we are quickly reducing pressures prior to theatre. We now ventilate to normal CO2 levels in all patients for a number of reasons:

    1. Prolonged hyperventilation when discontinued, allowing normocapnia to be restored quickly results in a REBOUND phenomenon causing an INCREASE in CEREBRAL BLOOD FLOW and thus an increase in intracranial pressure.

    2. Overzealous hyperventilation at pCO2 around 20-25mmHg can induce CEREBRAL ISCHAEMIA via vasoconstriction.

    3. No study has shown any benefit in terms of survival with hyperventilation, in fact the contrary is shown.

Leave a Comment