Paediatric DKA Severity Interpreter
Paediatric DKA Severity Interpreter
Diabetic ketoacidosis in a child is graded by venous pH and bicarbonate, and the grade is only the first half of the reading. Effective osmolality and corrected sodium tell you whether this is ordinary DKA, a hyperosmolar state, or both at once — and the features associated with cerebral injury tell you what to watch, which is not the same as telling you to slow the fluids. This page grades and interprets. It is not a fluid protocol: the paediatric maintenance fluid calculator and the paediatric dehydration fluid deficit calculator are the pages for that.
Grade the acidosis, then read the osmolality
pH + bicarbonate → ISPAD gradeAn eight-year-old with known type 1 diabetes. Venous pH 7.15, bicarbonate 8 mmol/L, sodium 134 mmol/L, glucose 28 mmol/L. Alert and orientated, no headache.
The ISPAD grades, and the two derived numbers that change what the grade means
Mild — venous pH below 7.3 or bicarbonate below 18 mmol/L.
Moderate — venous pH below 7.2 or bicarbonate below 10 mmol/L.
Severe — venous pH below 7.1 or bicarbonate below 5 mmol/L.
Effective osmolality (mOsm/kg) = 2 × plasma Na + plasma glucose (mmol/L). Normal 275 to 295.
Corrected sodium (mmol/L) = measured Na + 1.6 × ([glucose − 5.6] ÷ 5.6).
Hyperglycaemic hyperosmolar state — glucose above 33.3 mmol/L (600 mg/dL), venous pH above 7.25, bicarbonate above 15 mmol/L, small ketonuria with absent to mild ketonaemia, and effective osmolality above 320 mOsm/kg.
- pH or bicarbonate
- not both. Each grade is met if either measurement crosses it, so the grade is whichever of the two is worse. A child with a pH of 7.18 and a bicarbonate of 4 is severe, not moderate. This catches people out because the two do not move together: respiratory compensation holds the pH up while the bicarbonate falls, so in a child who is still compensating well the bicarbonate is the more honest number and in an exhausted child the pH overtakes it
- venous
- load-bearing. The criteria are written for venous pH. Arterial blood reads about 0.02 to 0.04 pH units higher, which is small in absolute terms and large relative to the 0.1 gaps between the grades — enough to read a severe child as moderate. Use venous, and convert if you have to
- effective osmolality
- 2 × Na + glucose, and deliberately not the full calculated osmolality, because urea crosses cell membranes freely and so contributes nothing to the osmotic gradient that moves water in and out of brain cells. Adding urea in, as the standard osmolality formula does, inflates the number in a dehydrated child and makes a hyperosmolar state look worse than it is
- 1.6 and 5.6
- the correction factor and the reference glucose in the sodium correction. For every 5.6 mmol/L (100 mg/dL) of glucose above normal, sodium is diluted by about 1.6 mmol/L. A competing factor of 2.4 (the Hillier coefficient) exists and gives a higher corrected sodium; ISPAD prints 1.6 and this page follows it. The corrected sodium is what tells you whether the sodium is really low or only looks low
- 320 mOsm/kg
- the effective osmolality above which a hyperosmolar state is defined, against a normal range of 275 to 295. It is reached mostly by glucose: at a sodium of 140, an effective osmolality of 320 needs a glucose of 40 mmol/L. That is why a hyperosmolar state and DKA can coexist and why the osmolality has to be calculated rather than assumed from the pH
Worked example
An eight-year-old with known type 1 diabetes. Venous pH 7.15, bicarbonate 8 mmol/L, sodium 134 mmol/L, glucose 28 mmol/L. Alert and orientated, no headache.
Neurological state is normal, so the cerebral injury branch does not apply
Glucose 28 mmol/L is not above 33.3, so the hyperosmolar criteria cannot be met whatever the osmolality
Effective osmolality = 2 × 134 + 28 = 296 mOsm/kg — one above the top of the 275 to 295 normal range, and nowhere near the 320 that defines a hyperosmolar state
Severe? pH 7.15 is not below 7.10, and bicarbonate 8 is not below 5. Neither limb is met
Moderate? pH 7.15 is below 7.20. One limb is enough, so the grade is moderate — note that the bicarbonate of 8 is also below 10, so both limbs agree here
Corrected sodium = 134 + 1.6 × ((28 − 5.6) ÷ 5.6) = 134 + 1.6 × 4.0 = 140.4 mmol/L. The bracket is exactly 4.0, so this one is not a rounding — the measured sodium of 134 looks low and is not
So: moderate DKA, no hyperosmolar component, a sodium that is normal once the glucose is accounted for, and a child who needs fluids, insulin after the fluids, and potassium from the outset
ISPAD severity grading — either measurement alone is enough
| Grade | Venous pH | or | Bicarbonate |
|---|---|---|---|
| Mild | Below 7.3 | or | Below 18 mmol/L |
| Moderate | Below 7.2 | or | Below 10 mmol/L |
| Severe | Below 7.1 | or | Below 5 mmol/L |
Features associated with cerebral injury, and what each one should change
| Feature | Changes the monitoring | Changes the fluid rate |
|---|---|---|
| Age under 5 years | Yes — hourly neurological observations | No |
| New-onset diabetes | Yes | No |
| Longer duration of symptoms before presentation | Yes | No |
| More severe acidosis at presentation | Yes | No |
| Greater hypocapnia than the acidosis explains | Yes | No |
| Raised urea at presentation | Yes | No |
| Bicarbonate given to correct the acidosis | This is a treatment, not a feature — do not give it | — |
DKA against hyperglycaemic hyperosmolar state
| DKA | Hyperosmolar state | |
|---|---|---|
| Glucose | Above 11 mmol/L | Above 33.3 mmol/L |
| Venous pH | Below 7.3 | Above 7.25 |
| Bicarbonate | Below 18 mmol/L | Above 15 mmol/L |
| Ketones | Beta-hydroxybutyrate 3 mmol/L or more, or moderate to large ketonuria | Small ketonuria, absent to mild ketonaemia |
| Effective osmolality | Variable | Above 320 mOsm/kg |
| Typical fluid deficit | 5 to 10% of body weight | 12 to 15% of body weight |
| Insulin | After fluids, at a fixed rate | Later and lower — glucose falls with rehydration alone |
Why the grade is the easy part
Grading paediatric diabetic ketoacidosis is arithmetically trivial and clinically useful: venous pH below 7.3 or bicarbonate below 18 is mild, below 7.2 or below 10 is moderate, below 7.1 or below 5 is severe, and either measurement alone carries the grade. The grade drives the level of care, the frequency of observation and, in most units, whether the child goes to a high-dependency bed. It does not drive the fluid rate, and the widespread belief that it should is the reason this page exists in the form it does.
Two derived numbers change what the grade means. Effective osmolality — twice the measured sodium plus the glucose in mmol/L — identifies the child who is hyperosmolar as well as, or instead of, acidotic, and that child has a much larger fluid deficit and needs insulin later and more slowly. Urea is deliberately excluded from the calculation because it crosses cell membranes freely and exerts no osmotic pull on brain cells, so including it, as the standard osmolality formula does, inflates the figure in exactly the dehydrated children in whom the answer matters. Corrected sodium — measured sodium plus 1.6 for every 5.6 mmol/L of glucose above normal — separates a sodium that is genuinely low from one that only looks low because glucose has pulled water into the vascular space.
The contentious part is fluid rate, and the evidence has moved. For two decades the fear of cerebral injury drove cautious rehydration, on the basis of retrospective associations between rapid fluid administration, falling serum sodium and cerebral oedema. A large randomised trial of fluid rate and fluid sodium content found no difference between the arms in altered mental status, in clinically diagnosed cerebral injury, or in long-term neurocognitive outcome. The current guideline’s position is that a range of protocols can be used safely, that clinicians should not unnecessarily restrict fluid when the child needs circulatory volume, and that promoting a rise in serum sodium need not be a routine focus of treatment — and that if the sodium does need changing, it is the sodium content of the fluid that should be adjusted rather than the rate. What remains firmly associated with cerebral injury is giving bicarbonate to correct the acidosis, which is why no protocol does.
What the risk features do change is monitoring. Younger age, new-onset diabetes, a longer duration of symptoms, more severe acidosis, greater hypocapnia than the acidosis explains and a raised urea all identify children who need hourly neurological observation and a low threshold for treating a suspected cerebral injury before imaging. The bedside rule for making that diagnosis — one diagnostic criterion, two major criteria, or one major plus two minor — runs at about 92 per cent sensitivity with a 4 per cent false-positive rate, and is deliberately structured so that the headache and vomiting almost every child with DKA has cannot trigger it alone. For the fluids themselves, the paediatric maintenance fluid calculator and the paediatric dehydration fluid deficit calculator are the pages; for the sodium, the sodium correction for hyperglycaemia calculator; for the ketones, the beta-hydroxybutyrate unit converter.
Frequently asked questions
What pH counts as severe DKA in a child?
A venous pH below 7.1, or a serum bicarbonate below 5 mmol/L — either one alone is sufficient. The full ISPAD grading is mild below 7.3 or below 18 mmol/L, moderate below 7.2 or below 10 mmol/L, and severe below 7.1 or below 5 mmol/L. Two practical points. All the thresholds are strict, so a pH of exactly 7.10 grades as moderate. And the criteria are written for venous pH; an arterial sample reads roughly 0.02 to 0.04 higher, which is enough to move a child down a grade, so an arterial gas should be converted before it is graded.
Why use effective osmolality rather than calculated osmolality in DKA?
Because urea crosses cell membranes freely and therefore exerts no osmotic force across them. Effective osmolality — 2 × sodium + glucose, both in mmol/L — measures only the solutes that actually hold water in or out of cells, which is the quantity relevant to brain volume and to the risk of shifting it too quickly. The standard calculated osmolality adds urea in, and in a dehydrated child with a high urea that inflates the figure and makes a hyperosmolar state look worse than it is. The normal effective osmolality is 275 to 295 mOsm/kg and the hyperosmolar threshold is above 320.
Does a high risk of cerebral oedema mean the fluids should be slowed?
No — and this is the most important thing on this page. A randomised trial of fluid infusion rate and fluid sodium content found no significant difference between the arms in the frequency of altered mental status, in clinically diagnosed cerebral injury, or in long-term neurocognitive outcome. The guideline’s conclusion is that a range of fluid protocols can be used safely and that clinicians should not unnecessarily restrict fluid administration if clinical signs suggest the child needs circulatory volume. What the risk features change is the intensity of neurological monitoring and the threshold for treating a suspected cerebral injury. What does remain associated with increased risk is giving bicarbonate to correct the acidosis, which no current protocol does.
How do I tell DKA from hyperglycaemic hyperosmolar state in a child?
Four criteria define the hyperosmolar state: glucose above 33.3 mmol/L (600 mg/dL), venous pH above 7.25, bicarbonate above 15 mmol/L, and effective osmolality above 320 mOsm/kg, with only small ketonuria and absent to mild ketonaemia. DKA is the opposite pattern — acidotic and ketotic, with a glucose that is usually lower. The two can coexist, and this page tests the hyperosmolar criteria first for that reason. It matters because the management differs: the hyperosmolar child has a fluid deficit of 12 to 15 per cent rather than 5 to 10, needs fluid before insulin rather than with it, and needs insulin started later and at a lower rate, because glucose falls with rehydration alone.
What is corrected sodium for, and which factor should I use?
It tells you what the sodium would be without the dilution caused by hyperglycaemia, which is how you know whether a low measured sodium is real. ISPAD’s formula is measured sodium + 1.6 × ([glucose − 5.6] ÷ 5.6), in mmol/L — that is, 1.6 mmol/L of apparent sodium lost for every 5.6 mmol/L (100 mg/dL) of glucose above normal. A competing factor of 2.4 exists in the adult literature and gives a higher corrected value, particularly at very high glucose; the paediatric guideline prints 1.6. In treatment, a corrected sodium that rises or holds steady as the glucose falls is the expected pattern, but the guideline is explicit that promoting a rise in sodium need not be a routine focus of treatment, and that if the sodium does need changing it is the fluid’s sodium content that should change rather than the infusion rate.
Can a child have DKA with a normal glucose?
Yes. Euglycaemic diabetic ketoacidosis is well described in children who have been partly treated, who have been vomiting and starved, or who are taking an SGLT2 inhibitor, and the glucose can sit below the 11 mmol/L the diagnostic criteria require while the child is fully ketoacidotic. Because this page grades on pH and bicarbonate rather than on glucose, the severity grade it returns is still correct in that situation — but the diagnostic criteria as written are not met, and the reader has to know that the glucose limb can fail while the illness is present. Measure blood beta-hydroxybutyrate rather than relying on urine ketones, which lag and which measure acetoacetate rather than the dominant ketone.
Related calculators
References
- Glaser N, Fritsch M, Priyambada L, Rewers A, Cherubini V, Estrada S, Wolfsdorf JI, Codner E. ISPAD Clinical Practice Consensus Guidelines 2022: diabetic ketoacidosis and hyperglycemic hyperosmolar state. Pediatr Diabetes. 2022;23(7):835–856. doi:10.1111/pedi.13406. The source of every threshold on this page. Diagnostic criteria: glucose above 11 mmol/L (200 mg/dL), venous pH below 7.3 or bicarbonate below 18 mmol/L, and ketonaemia. Severity: mild below 7.3 or 18; moderate below 7.2 or 10; severe below 7.1 or 5. Effective osmolality = 2 × plasma Na + plasma glucose (mmol/L), normal 275–295. Corrected sodium = measured Na + 1.6([glucose − 5.6]/5.6). HHS: glucose above 33.3 mmol/L, venous pH above 7.25, bicarbonate above 15 mmol/L, effective osmolality above 320 mOsm/kg.
- Kuppermann N, Ghetti S, Schunk JE, et al; PECARN DKA FLUID Study Group. Clinical trial of fluid infusion rates for pediatric diabetic ketoacidosis. N Engl J Med. 2018;378(24):2275–2287. doi:10.1056/NEJMoa1716816. The trial behind the guideline’s position on fluid rate. Quoted in the 2022 guideline as showing "no significant differences in the frequency of either altered mental status or clinical diagnoses of cerebral injury in any of the treatment arms", with similar long-term neurocognitive outcomes in all groups.
- Muir AB, Quisling RG, Yang MCK, Rosenbloom AL. Cerebral edema in childhood diabetic ketoacidosis: natural history, radiographic findings, and early identification. Diabetes Care. 2004;27(7):1541–1546. doi:10.2337/diacare.27.7.1541. The bedside diagnostic rule reproduced on this page — one diagnostic criterion, two major criteria, or one major plus two minor, giving a sensitivity of 92% with a 4% false-positive rate.
- Hillier TA, Abbott RD, Barrett EJ. Hyponatremia: evaluating the correction factor for hyperglycemia. Am J Med. 1999;106(4):399–403. doi:10.1016/s0002-9343(99)00055-8. Cited only to name the competing correction factor of 2.4, which this page does not use; the paediatric guideline prints 1.6.
Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.