Diabetic ketoacidosis — branching viva

Opening framework

My framework has three layers. First, the diagnosis: this drowsy, dehydrated child breathing deeply with ketones on her breath has diabetic ketoacidosis, and I confirm it on the triad of hyperglycaemia, a venous pH below 7.3 or bicarbonate below 15, and ketonaemia. Second, the physiology that makes it dangerous: she is more dehydrated and more potassium-depleted than she looks, and correcting her too fast risks cerebral oedema. Third, the management in a fixed safety order: cautious fluids, then insulin by infusion an hour or two later at 0.05 to 0.1 units/kg/h with no bolus, early potassium, glucose-guided dextrose, and relentless neurological monitoring. [1]

Diagnosis and severity grading

For this five-year-old the diagnosis is DKA: her glucose is high, her ketones are 5 mmol/L, and her venous gas shows a pH of 7.05 and a bicarbonate of 5 mmol/L. By the ISPAD grading this is severe DKA, because the pH is below 7.1 and the bicarbonate below 5, and her young age and probable new-onset diabetes make her a high-risk child for cerebral oedema. She needs high-dependency or intensive-care nursing with hourly observations from the outset, and in a peripheral hospital I would involve the retrieval service early. [1] [3]

The fluid and insulin plan and its order

She is dehydrated but not shocked, so she does not need a bolus; a bolus of 10 mL/kg isotonic saline is only for the shocked child. I would begin steady rehydration with isotonic saline giving maintenance plus a modest deficit evenly over 48 hours, and start insulin only after fluids are running, an hour or two in, at 0.05 to 0.1 units/kg/h by continuous infusion. The two absolute rules are fluids before insulin and never a bolus of insulin, because insulin into a hypovolaemic, potassium-depleted child causes a dangerous osmotic and electrolyte shift. The PECARN FLUID trial reassures me that a sensible fluid regimen does not worsen neurological outcome. [1] [2]

Why the potassium is misleading

Her serum potassium will read normal or high, but that hides a large total-body deficit, because the acidosis and lack of insulin have shifted potassium out of the cells and it has been lost in the urine. As soon as I start insulin and fluids, potassium will move back into the cells and the serum level will drop, so I add potassium to the fluids at 40 mmol/L once it is below 5.5 mmol/L and she has passed urine, and I monitor it hourly. If she were frankly hypokalaemic, I would correct the potassium before starting insulin. [1]

The complication to watch for and its treatment

Cerebral oedema is the complication I watch for, and it is the leading cause of death in paediatric DKA. It usually appears four to twelve hours into treatment, and I would suspect it from a headache, a falling conscious level, irritability, incontinence, a new neurological sign, or a rising blood pressure with a slowing heart rate. I would treat it immediately with hypertonic saline or mannitol, reduce the fluid rate, elevate the head of the bed, and call intensive care — treating on clinical grounds, because a scan is to look for other causes and must never delay treatment, and outcome depends on how fast it is treated. [3] [9]

Branch — the pump user with unexplained ketosis

If instead this were an established pump user with unexplained ketosis, I would treat that as a pump failure until proven otherwise, because a pump delivers only rapid-acting insulin and provides no long-acting depot, so an occluded line or failed site can precipitate ketoacidosis within a couple of hours, sometimes with a glucose that is only modestly raised. The response is to give insulin by pen or syringe, revert to injections while the pump problem is sorted, and manage the ketoacidosis on its severity. This is also how euglycaemic DKA is recognised — the diagnosis rests on the ketones and acidosis, not the glucose. [1] [8]

Branch — persisting acidosis with a normalising glucose

If the glucose normalises but the acidosis persists, I would not stop the insulin, because its job is to clear the ketones, and I would instead add dextrose to the fluids so the insulin can keep running. A persistently high anion gap means insufficient insulin or an unresolved precipitant, not resolution. I would also consider a hyperchloraemic acidosis from large-volume saline, which lowers the bicarbonate without ongoing ketosis and needs no more insulin — bedside beta-hydroxybutyrate tells the two apart. The Nallasamy trial supports a lower insulin rate of 0.05 units/kg/h as effective with fewer episodes of hypoglycaemia and hypokalaemia. [1] [6]

Branch — resolution and transition to subcutaneous insulin

I would move her to subcutaneous insulin only once DKA has resolved: a venous pH above 7.3, a bicarbonate above 15 mmol/L, a closed anion gap with cleared ketones, and the child able to eat. I give the first subcutaneous dose before a meal and overlap it with the intravenous insulin for a short period so there is no gap in cover. For a new-onset child like this, the transition is also the start of her life with diabetes, so it triggers structured education for the family in insulin, glucose monitoring, and sick-day rules, with psychological support for a diagnosis they did not expect. [1] [8]