Dr Jonathan Round – PICU Consultant

Protocols save lives. They save lives because they stop professionals trying to work out for themselves what is best for each situation, and understanding for themselves the nuances of countless diseases.  To be able to do this is clearly unrealistic, and so protocols can be of help. They are drawn up by clever people, experts in their field, tested and revised. So for situations that are common, complex, fast paced or dangerous, protocols are the way forward.

There are, however, a few problems with protocols.

#1 This is not really a problem with the protocol but with the professional – you can choose the wrong protocol.

#2 Protocols are designed for the majority of patients, and this does not include those with co-morbidities and those who are at the extremes of illness. For such situations, there is really not a lot of research evidence, so clinicians are left to hope that the protocol continues to work even in critical illness or go back to basic pathophysiology and work out what to do from first principles. This is not easy if you have been brought up learning protocols.

#3 Protocols lead to atrophy of the brain. Not literally (arguably 5% saline does this a little), but it stops the user engaging with the ongoing disease processes, and appreciating the intricacies of successful management.

Protocols in DKA

Diabetes has seen a lot of protocols over the years, each prescribing different amounts of water, salt, glucose and insulin, looking at various combinations of physical signs and test results.

Ten years ago, the emphasis was on rapid correction of fluid loss, usually over 24 hours. As it was noted that deaths mostly occurred from cerebral oedema in the recovery phase, rehydration slowed to 48-72hrs.

Not wishing to add to the stockpile of diabetes protocols, I would only like to add a personal perspective. This is more about understanding of disease processes and helping the treating clinician understand how treatment is working. This breaks the presentation into three separate problems, each with its own signs and features.



Dehydration is difficult to identify in DKA. The issue is that the acidosis itself leads to tachycardia, peripheral vasoconstriction and a prolonged capillary refill.

If a recent weight cannot be obtained for comparison, the clinician should enquire about the length of symptoms or suggestions that the child is decompensating, such as a falling off of thirst. Clinical signs of actual use are sunken eyes and in an infant, a sunken fontanelle. Hypotension, raised lactate and a high creatinine are all signs of insufficient circulating volume, typically equivalent to over 10% dehydration.

Dehydration obviously needs correction with fluid. If the dehydration is bad enough to lead to shock it needs an isotonic fluid bolus for correction, even before rehydration begins.



Ketosis comes from the absence of insulin. A small but insufficient amount of insulin leads to hyperglycaemia, but not ketosis. Ketosis is the most potent metabolic reason making the body feel ill during DKA and causes the vomiting.

Clinical features of use to identify the degree of acidosis are tachypnoea, the (arterial) pH, base excess, ketone levels, anion gap and blood lactate.

Acidosis in DKA is typically caused initial by ketoacids (+/- lactate) and later chloride ions. In DKA, each ABG should be looked at carefully, with anion gap, Cl/Na ratio and lactate used to estimate the importance of each acid type.

If the acidosis is caused by lactate, then this reflects intravascular depletion and impending shock as a result of severe dehydration.

In initial phases, the acidosis is caused by ketones, and this can only be corrected with insulin. It takes a long time to correct this, usually 1-3 days, and only then if decent doses of insulin are administered.

Key mistakes happen with the management of ketosis. One is the expectation that the acidosis will disappear as fast as the hyperglycaemia, but this is not the case. Acidosis usually also persists when the ketoacidosis is replaced by chloroacidosis, and the clinician is worried his treatment is not working, when all that is needed is some time and renal function to deal with the chloride excess.

Unlike shock, hyperosmolar states and sepsis, ketosis is not dangerous. At very low (<7.0) pH, there is likely to be some myocardial dysfunction, but above this there are few measurable consequences of the acidosis.

The important principle here is initiation of insulin treatment, to stop acidosis worsening and to correct the glucose situation slowly. Maintain enough blood sugar to keep the insulin going at all times.


**Osmolar derangement**

This is more serious. During the pretreatment phase of DKA, sodium enters neurones to preserve cell volume in the face of the hyperosmolar serum on the other side of the cell membrane. As the blood glucose falls in treatment, there is an osmolar gradient promoting water flow into neurones, and this can lead to cerebral oedema if bad enough.

Here there are few physical signs of the hyperosmolar condition, but the blood glucose and corrected sodium should give you a strong clue. You can also measure the osmolality directly.

During the treatment phase, oedema is seen as a fall in conscious level. Any such fall should prompt immediate 5% saline as a bolus and a CT scan.

The key to managing hyperosmolar conditions is to be slow. The most important measure is the osmolality, which is made up of blood sodium and blood glucose. Formulae are easy to find elsewhere. It is essential that the osmolality drops by no more than 5 mmol/L in a day, which in some cases means reducing the insulin rate, giving 5% saline or 50% glucose. This is a particular problem when the glucose starts >50 mmol/L.



I lied – there is a fourth aspect of diabetes management. It is not always possible to identify the cause of the DKA, but sometimes it is bacterial sepsis. This is the other common cause of death in DKA in children. Staph Aureus sepsis was responsible for the my only patient death in all the paediatric diabetics I have ever seen. Diabetics are of course more likely to have bacterial invasive disease than those without diabetes. Any suggestion that the patient may be septic should lead to a broad spectrum antibiotic.

Clinical signs of use are a fever persisting after rehydration, vasodilation and a raised CRP. The WCC may of course be elevated by the acidosis and dehydration or acidosis can caused the fever as blood is diverted away from the skin.

End notes

This short reflective article is deliberately not meant to be another protocol. You will have seen no formulae or directives here – they can be found elsewhere. Instead, I have outlined the clinical principles behind the protocols, and given the reader a better understanding of what they are doing and why.

Dr Jonathan Round – Consultant Paediatric Intensivist

1 thought on “Diabetic Ketoacidosis: Principles behind Protocols”

  1. That is my genral approach because it makes sense physiologically but acknowledging that trials have failed to definitely prove that outcomes differ between high/low fluid or insulin alters outcome. Underpowered for a rare event?

    Another PICU consultant suggested giving replacement fluid based on urine output as long as shock not present to maintain neutral balance until diuretics resolved.

    I would argue that insulin is titrated to acidosis and give 50% glucose to prevent rapid falls in glucose/osmolality

    Too much saline rather than balanced solutions causes NAGMA and confuses biochemical picture.

    Many protocols hold insulin before fluid blouse but I think this is unnecessary if child K is ok on VBG and not anuric

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.