ICU and You · Podcast Sessions · Number 59
Nicole Glaser on the DKA that does badly on the DKA pathway
Free to listen to, with a transcript on the show's own site. It is built around a single case and worked through in order, so it survives being listened to in pieces better than most.
Listen first. It is not a test. Do not take notes, do not try to hold the numbers, and do not stop to look anything up. Let it wash over you on a drive or a walk and accept that most of it will not stick. That is not a failure of attention; it is how listening works.
What a podcast does that no paper can is let you hear somebody think. The hesitations, the qualifications, the places where an expert says plainly that we do not know — none of that survives into print, and it is most of what you are actually there for. You are picking up how somebody holds a problem, not a set of facts.
Then talk about it. Bring one thing to a ward round, argue with somebody about it over coffee, disagree with the guest out loud. An episode discussed once is worth three listened to alone.
And if something matters enough to act on, look it up properly afterwards. A podcast is a way in, not a source.
A registrar takes a transport call. Fifteen-year-old, known type 1 diabetes, no insulin for a couple of days, drowsy. Glucose 800, pH 6.9, ketones in the urine, sodium 155. One 20 mL/kg bolus given. The referring hospital asks whether to start insulin.
Say yes, and over the next hour the glucose will fall, the osmolality will fall, and the patient will go into shock while receiving intravenous fluid. That is the episode.
Nicole Glaser is a paediatric endocrinologist at the University of California at Davis who has spent a career on paediatric DKA — clinical work, basic science, and several international guidelines, including the ISPAD revision she was working on when this was recorded. The hosts are two paediatric intensivists. The whole episode is one case, worked through in order.
And its subject is a presentation most of us have met without naming. DKA is the most protocol-driven thing we do, and for good reason: garden-variety DKA does well on the standard pathway. Glaser's point is that a particular subgroup does badly on it — and that the reason they do badly is that somebody put them on it.
These are the kids that are likely to not do well on the standard protocol… I think that's often the biggest mistake in taking care of these kids, is that they do get sort of started on the regular DKA protocol. And it's not until sometime down the line that people notice that they actually need something a little different, or are developing complications.
Hyperosmolar DKA is exactly what the name says: a child who meets the criteria for DKA and also has the hyperosmolar state that characterises hyperglycaemic hyperosmolar syndrome. HHS is defined by a glucose above 600 with an osmolality above 320 and without significant ketosis or acidosis. Hyperosmolar DKA has that same severe hyperglycaemia and hyperosmolarity, plus the ketosis and acidosis.
The host asks the obvious practical question — does a transport call with a glucose above 600 mean you pivot? And the answer is no, and this is the part worth carrying:
“Not everyone who has a glucose above 600 necessarily has hyperosmolar DKA. It really depends on the osmolality.” A glucose above 600 with a low sodium is not especially hyperosmolar and does not carry these worries.
In the case, the measured sodium is 155 and the corrected sodium is 166. That is what makes him hyperosmolar, and it is arrived at by looking at the glucose and the sodium together. Look at one without the other and you will misclassify him in both directions.
A typical child with DKA is about 7% dehydrated. A typical child with HHS is 12% to 15% — roughly double. So a single 20 mL/kg bolus, which would be unremarkable in DKA, is almost certainly not enough here. Glaser would expect at least two boluses of that volume and quite possibly more, with ongoing rates at least one and a half to twice what you would usually run.
Her practical anchor, when the host presses on where the number comes from: go through the calculations assuming 12% dehydration, work out the volume, and base the rate on that. Not a fixed multiple of maintenance — an actual sum, for this patient.
Then the piece of advice that runs directly against what we teach about DKA:
We do not usually replace urine output in DKA. In hyperosmolar DKA we often should, because at a glucose of 900 or 1,000 it takes a very long time for the level to fall far enough to stop the osmotic diuresis. The polyuria goes on for hours after you have stopped expecting it.
The test is the balance chart rather than a rule: if the infusion rate is high and the urine output is still so brisk that you are not making steady positive gains, replace it. Glaser would reach for a catheter and one-to-one replacement at exactly that point and not before.
These children arrive with their tissues supremely dehydrated and their circulating volume relatively preserved — because the osmolality is holding water in the vessels. As one of the hosts puts it, if the osms are pulling everything into the intravascular space, it will plump up the vessel more than you would expect in someone that dehydrated. It is also why these patients can be profoundly dry and only look unwell for a day or two.
Now start insulin. The glucose falls, the osmolality falls, and the water that was being held in the circulation moves out into the tissues where it was always needed. The circulation contracts.
You can produce organ hypoperfusion, or frank shock, in a patient who is receiving intravenous fluid — because the fluid going in is not keeping pace with the volume leaving the circulation as the osmolality falls.
Which is why the fluid has to go in before the glucose comes down, and why the answer to that phone call is to resuscitate first and have a clear plan for when insulin starts, rather than to start it now because the protocol says so.
One of the hosts asks the question everybody would ask. These are the sickest DKA patients, the main treatment for cerebral oedema is more hyperosmolar therapy, and the patient is already hyperosmolar — so how bad is the cerebral oedema risk here?
Lower, as it turns out, or at least no higher. No study has directly compared the rates, but the level of hyperglycaemia at presentation has not been associated with cerebral oedema risk across multiple studies controlled for severity; and in HHS, where glucose levels run to 900 and 1,000, cerebral oedema is rare enough to be a case-report phenomenon, probably less frequent than in ordinary DKA. Glaser's estimate is that hyperosmolar DKA carries much the same risk as any other DKA.
And the reason follows from the mechanism above. If cerebral oedema arises from hypoperfusion followed by reperfusion, then a very high osmolality is protective, because it keeps the circulation full. The thing that makes these patients dangerous to treat is the same thing that is protecting their brain while you have not yet treated them.
Which leaves a genuinely hard clinical problem, and Glaser does not pretend otherwise: patients in a hyperosmolar state very frequently have altered mental status at presentation, nobody has a good explanation for why, and it looks exactly like the thing you are afraid of.
This is the part I would make a registrar read twice, because it is a case of the right advice surviving after its original justification collapsed.
We used to lower glucose and sodium slowly in DKA because we were afraid of cerebral oedema. That fear has, in Glaser's words, “pretty much been debunked” — studies show no association between fluid rate or fluid sodium content and cerebral oedema risk.
But in these patients you should still bring the glucose and sodium down gradually, for an entirely different reason: circulating volume and organ perfusion. Let them snap back to normal and the volume leaves the circulation faster than you are replacing it.
Which means the habit is right and the reasoning most people carry for it is wrong — and that matters, because a habit with the wrong reason attached gets applied in the wrong places and dropped in the right ones.
The working figure is a fall of 50 to 100 per hour, and she is admirably honest about its provenance: “I don't know that there's a real evidence base necessarily to support that specific number, but that's the one that's sort of been used.” She also points out that an early rapid fall on rehydration alone is not something to chase — that is reperfused kidneys clearing glucose, doing what kidneys do.
On the insulin itself: 0.05 units/kg/hour is reasonable in a child who is very hyperosmolar but only modestly acidotic, following beta-hydroxybutyrate to confirm the ketosis is still falling. At a pH of 6.9 it is a genuine toss-up — try the lower dose if you like, but move to the standard dose quickly if the ketones and the pH do not shift.
Electrolyte deficits in hyperosmolar DKA and HHS are far more extreme than in ordinary DKA, so these children are much more exposed to severe hypokalaemia. Two-hourly bloods, hourly if something is moving fast.
And then the catch:
Not every DKA protocol has a lot of monitoring of phosphate levels. But in a patient like this, you would definitely want to include that — and probably just as frequently as monitoring of potassium.
Hypophosphataemia here is linked to rhabdomyolysis and to haemolytic anaemia. A trainee following the standard protocol will not be looking for it, because the standard protocol does not ask them to. Alongside that: acute kidney injury, more severe than the AKI that attends ordinary DKA and occasionally reaching renal failure; thrombosis; and ischaemic intestinal necrosis. What she asks you to watch is the trend — the phosphate drifting down, the creatinine rising when it ought to be falling.
On lines, the hosts are refreshingly conservative. Central line thrombosis rates in DKA are high, with some studies suggesting around 50%, so central access only if it is needed to resuscitate; arterial line with a low threshold if mental status or perfusion are shifting. And a heuristic worth keeping:
If they wouldn't notice you putting in an arterial line with full sterile technique, then you might want to consider that.
The patient's tolerance of the procedure is the assessment of their conscious state. Readers of the happily examined aphorism will recognise the shape of that.
One of the hosts names something I have watched happen for thirty years, and names it well. DKA management invites you to manage a flowsheet: the hourly sugars, the two-hourly electrolytes, the little columns. It is absorbing, it feels like vigilance, and it can be done from the desk.
We should be much more worried about the status of our patient than about whether the sugar fell by 101 versus 99 in the last hour.
Glaser's own answer to what she would emphasise is going to the bedside and re-evaluating the patient — looking at perfusion, at the vital signs, at mental status — because these patients need individualised care and you cannot individualise from a number.
There is a section of this episode that will land harder in Coffs Harbour than in Dallas. The host describes taking transport calls from hundreds of miles away, from a referring clinician who is extremely busy, and the possibility that if the retrieval team cannot come immediately, nobody will have another conversation about this child for three or four hours. He is uneasy about issuing prescriptions down a phone line with no reassessment built into them.
Glaser agrees, and her answer is not a pathway but a practice: get lots of detail, and talk to the outside hospital frequently, because the balance of acidosis, dehydration and hyperglycaemia is different in every one of these children and no single pathway covers them.
If you are the referring clinician in that conversation, the useful thing to say out loud is the corrected sodium, and then to agree explicitly when the next call will happen and what will have been measured by then.
Not, mostly, by the route we are trained to fear. Asked directly, Glaser says the case reports come back to shock and circulatory volume: treated initially on the standard DKA protocol, the fluid volume was not enough, and shock and renal failure and other organ injury followed on that basis. Beyond that, arrhythmias from rapid falls in potassium or phosphate, and the consequences of very low phosphate.
Her summary of the whole thing is the sentence to take away: once you recognise that this person has hyperosmolar DKA and that it needs treating differently from the usual protocol, that is more than half the battle.
The last question is about medicolegal considerations, and the answer is the best thing in the episode. Glaser reviews cases sent to her over concerns about malpractice, and says the most common thread is almost always the same one:
It wasn't really recognised that the patient had something different from garden-variety DKA… when you look back at what was happening with fluid balance during treatment, the patient wasn't really having a nice positive fluid balance. They were maybe staying neutral or even becoming more dehydrated, because they had very high urine output, and that wasn't noted or addressed.
Read that again as a description of a ward round rather than a lawsuit. Nobody did anything dramatic. The protocol was followed. The observations were charted. The urine output was measured and written down, in the correct column, by somebody competent. And the sum of those numbers — the one number nobody was actually looking at — was quietly going the wrong way for hours.
Which is as good an argument as I have heard for the oldest habit in intensive care: go and look at the patient, and add up the balance yourself.
Does your DKA protocol tell you what to do when the glucose is 800?
Go and read it — properly, this week, not from memory. Does it mention hyperosmolar DKA at all? Does it ask you to calculate a corrected sodium? Does it say anything about replacing urine output, or about phosphate? And if a registrar followed it exactly on the patient in this episode, where would it take them?
Tell me what yours says, and what it leaves out. I would like to know whether this is a gap in a few protocols or in most of them, and that is a question a lot of readers can answer in ten minutes each and nobody can answer alone.
The button opens a reply straight to me. If your device blocks it, write to icuandyou@icloud.com instead.
Heard something worth other people's time?
I take suggestions from anyone reading this. Tell me the show, the episode and why it was worth an hour, and I will listen. If you would rather write the review yourself, I will publish it here with your name on it — or without, entirely as you prefer.
The button opens a reply straight to me. If your device blocks it, write to icuandyou@icloud.com instead.