A registrar/consultant-level reference: pathophysiology, expected blood gas trajectory, age-stratified assessment and fluid/insulin/potassium protocols, cerebral oedema recognition and treatment, and management of the child newly diagnosed with T1DM who is not in DKA β cross-checked against CHQ, RCH, ISPAD 2022 and current trial evidence.
Read first. Decision-support, not a substitute for senior/PICU/endocrinology input or your local Metro South protocol. Confirm every fluid, insulin and potassium order against CREDD / the current CHQ flowchart / local formulary before prescribing. DKA management should be senior-led from the outset β this is not a junior-doctor-alone task.
DKA is a biochemical triad: hyperglycaemia + ketosis + metabolic acidosis. All three must be present β hyperglycaemia alone is not DKA, and ketosis with a normal pH is not DKA (though it may be pre-DKA).
Ketoacidosis with a near-normal glucose β seen in partially treated DKA, prolonged vomiting/starvation, pregnancy, and increasingly with SGLT2-inhibitor use (relevant in adolescents, off-label/T2DM). Do not exclude DKA on glucose alone if the clinical picture fits.
Leading cause of morbidity/mortality
DKA remains the leading cause of death in children with T1DM, overwhelmingly via cerebral oedema.
Severity classification (ISPAD / CHQ)
Mild
pH 7.2β7.3 Β· HCOββ» 10β18
Assume ~5% fluid deficit.
Moderate
pH 7.1β7.2 Β· HCOββ» 5β10
Assume ~7% fluid deficit.
Severe
pH <7.1 Β· HCOββ» <5
Assume ~10% fluid deficit. PICU involvement.
Note: RCH stratifies primarily by clinical/estimated dehydration (mild β€4%, moderate 4β7%, severe β₯7%) rather than pH bands alone β see Β§14. Use whichever your site's fluid calculator is built around, and don't mix the two systems within one calculation.
2
Pathophysiology
this explains every number you'll chase
DKA results from a relative or absolute deficiency of circulating insulin combined with a surge of counter-regulatory hormones (glucagon, catecholamines, cortisol, growth hormone). This combination unleashes catabolism that was normally held in check by insulin.
The hyperglycaemia arm
Without insulin, the liver ramps up glycogenolysis and gluconeogenesis while peripheral tissues cannot take up glucose. Once glucose exceeds the renal threshold (~10β12 mmol/L) glycosuria causes an osmotic diuresis β obligate loss of water, sodium, potassium, phosphate and magnesium. This is the direct cause of the profound dehydration and total-body electrolyte deficits, even when serum levels look deceptively normal.
The ketoacidosis arm
Insulin deficiency + counter-regulatory excess activates hormone-sensitive lipase, releasing free fatty acids that the liver converts to acetoacetate and Ξ²-hydroxybutyrate via Ξ²-oxidation. These ketoacids consume bicarbonate buffering capacity, producing a high-anion-gap metabolic acidosis. Respiratory compensation gives the classic Kussmaul breathing (deep, sighing) as the child blows off COβ.
Why potassium is the practical crux
Total-body potassium is always depleted (osmotic diuresis, vomiting, secondary hyperaldosteronism) β but serum potassium may be normal or even high at presentation because acidosis drives KβΊ out of cells (HβΊ/KβΊ exchange) and insulin deficiency prevents its cellular uptake. Starting insulin reverses both β KβΊ rushes back into cells β so serum potassium can fall precipitously and dangerously fast once treatment starts. This is why potassium goes into the fluid before or with insulin, not after.
Why cerebral oedema happens β the current model
The mechanism is incompletely understood, but the leading model implicates cerebral hypoperfusion/ischaemia during established DKA (from dehydration and hypocapnia) followed by a reperfusion/vasogenic injury during treatment, rather than a simple "too much fluid, too fast" osmotic effect. This is why the 2018 PECARN FLUID trial β which found no difference in neurological outcome between fast/slow rehydration or 0.9%/0.45% saline β reframed decades of practice: the injury process appears to begin before treatment starts in many cases, not to be caused by the treatment itself. See Β§14 for the trial detail.
3
Expected blood gas & biochemical trajectory
know what "on track" looks like
At presentation
pH
ββ
high anion gap acidosis
HCOββ»
ββ
buffered by ketoacids
pCOβ
β
Kussmaul compensation
Anion gap
ββ
>12β16; unmeasured ketoacids
Glucose
ββ
usually >20 mmol/L
NaβΊ (measured)
β/nl
dilutional β see corrected Na
KβΊ (serum)
nl/β
despite total-body deficit
Urea/Cr
β
pre-renal, dehydration
Lactate
mild β
hypoperfusion contributory
Phosphate
β (evolving)
falls further with insulin
Corrected sodium β always calculate it
Corrected NaβΊ = measured NaβΊ + 2.4 Γ [(glucose β 5.5) / 5.5] (mmol/L, glucose in mmol/L). Hyperglycaemia osmotically pulls water into the vascular space, diluting measured sodium β the corrected value tells you the true sodium/free-water status and should rise (or stay stable) as glucose falls with treatment.
Expected trajectory over the first 24β48 hours
Parameter
Expected pattern with appropriate treatment
Concerning deviation
Glucose
Falls ~2β5 mmol/L/hr with fluids Β± insulin; add dextrose once β€15 mmol/L to allow insulin to continue
Failure to rise, or a fall β early warning sign for cerebral oedema risk (Β§8)
pH / HCOββ»
Gradual correction over 24β48 h as ketoacids clear
Rapid overcorrection with bicarbonate β avoid (Β§7)
Anion gap
Progressively normalises as ketosis resolves
Persistently wide gap despite treatment β reassess insulin delivery/dose, look for a second process
Serum KβΊ
Falls after insulin starts β replacement should keep it in range
Rapid/severe fall β reduce insulin 50%, increase KβΊ replacement, discuss central access with critical care
Serum phosphate
Falls further with insulin (intracellular shift)
Severe hypophosphataemia (<0.3 mmol/L) β muscle weakness, rhabdomyolysis risk; consider replacement per ISPAD 2022
Why chloride rises (hyperchloraemic acidosis)
Large volumes of 0.9% saline (Clβ» 154 mmol/L) can produce a non-anion-gap hyperchloraemic metabolic acidosis as ketoacidosis resolves β the anion gap normalises but the pH/HCOββ» may lag, which can be mistaken for treatment failure. This is a recognised, usually self-limiting phenomenon that resolves once IV fluids stop; balanced solutions (Plasma-Lyte/Hartmann's) may reduce its magnitude but have not shown outcome benefit in paediatric trials (Β§14).
4
Precipitants & risk factors by age
New-onset T1DM
15β70% of new diagnoses present in DKA (varies by region/access to care)
Risk β with age <2 years, no first-degree relative with T1DM, lower socioeconomic status, delayed diagnosis/access to care
Established T1DM
Insulin omission β deliberate or inadvertent (pump failure/occlusion, expired insulin, psychosocial factors) is the leading cause
Intercurrent illness/infection (increased insulin requirement not met)
Peripubertal/adolescent girls, psychiatric comorbidity, disordered eating, difficult family circumstances β recognised higher-risk groups
Insulin pump users β rapid ketosis on occlusion (no depot of long-acting insulin)
5
Assessment β history, examination, red flags
History
Polyuria, polydipsia, weight loss, fatigue (classic triad β may be missed, especially in toddlers/infants)
Severe acidosis (pH <7.1) or very high corrected osmolality
6
Investigations
Confirm diagnosis
Venous blood gas (pH, HCOββ», lactate), blood glucose, blood or urine ketones (Ξ²-hydroxybutyrate preferred β correlates better with severity/resolution than urine ketones).
Baseline biochemistry
U&E (NaβΊ, KβΊ, urea, creatinine), phosphate, magnesium, calcium, FBC (leucocytosis common even without infection β stress response, don't over-interpret alone).
Precipitant work-up
Directed by history/exam β blood/urine culture, CXR, HbA1c (new diagnosis or assessment of recent control).
New diagnosis
Diabetes autoantibodies (GAD, IA-2, ZnT8, islet cell), coeliac serology, thyroid function β sent at diagnosis, does not delay treatment.
ECG
Consider if severe hyper/hypokalaemia suspected β peaked T waves (hyperK) or flattened T/U waves (hypoK, especially once treatment underway).
Monitoring frequency
Hourly: vital signs, GCS/neuro obs, glucose. Every 2β4 hours: VBG, electrolytes (more frequent β hourly β in severe DKA, age <5 years, or if KβΊ/glucose unstable).
7
The management protocol
fluids β (wait) β insulin β potassium throughout
Core sequence: resuscitate if shocked β calculate and begin fluid deficit replacement β start insulin infusion ~1 hour after fluids begin β add potassium from the start (once levels known) β add dextrose once glucose falls to ~15 mmol/L, all while monitoring hourly for cerebral oedema.
0 min
Resuscitate if shocked; baseline assessment
Shocked: 0.9% saline 10 mL/kg over 15 min, reassess, repeat to max 40 mL/kg (consider inotropes beyond this β discuss critical care).
Not shocked but clinically dehydrated: 0.9% saline 10 mL/kg over 30β60 min.
Document baseline GCS + cranial nerve exam. Weigh the child (do not estimate) β all fluid/insulin calculations are weight-based.
Send baseline bloods; check glucose & ketones at the bedside.
Why: resuscitation fluid restores perfusion but is not subtracted from the calculated deficit if the child was shocked (subtract only the "non-shock" 10 mL/kg bolus from the total deficit calculation, per most protocols β confirm against your local calculator).
Deficit calc
Calculate fluid deficit + maintenance
Total fluid = Deficit + Maintenance, replaced evenly over 48 hours (not faster β this is deliberate, unlike standard resuscitation).
Use 0.9% saline as the base fluid for at least the first 6 hours.
Do not use clinical dehydration signs alone to estimate the deficit β they are unreliable in DKA (hyperosmolality preserves intravascular volume). Use the pH/HCOββ»-based severity estimate.
~60 min
Start insulin infusion
IV insulin 0.05β0.1 units/kg/hr as a continuous infusion, started ~1 hour after fluids begin (never before, never as a bolus).
No IV bolus of insulin β increases hypoglycaemia and cerebral oedema risk without benefit.
ISPAD 2022 supports the lower end (0.05 units/kg/hr) in children <5 years, newly diagnosed, or with rapidly falling glucose β see Β§10.
Insulin's job is to switch off ketogenesis and lipolysis β not just to lower glucose. Do not stop it once glucose normalises; add dextrose instead.
Why wait 1 hour: fluids alone begin restoring renal perfusion and reducing counter-regulatory hormone drive; starting insulin promptly but not immediately allows initial volume expansion and avoids compounding early osmotic shifts.
With fluids
Potassium replacement
Add KCl 40 mmol/L to fluids once potassium level is known and urine output confirmed, increasing to max 60 mmol/L if needed to maintain normal range.
If KβΊ <3.0 mmol/L β replace potassium before starting insulin (risk of arrhythmia/respiratory muscle weakness).
If KβΊ >5.5 mmol/L or anuric β withhold potassium-containing fluid until levels fall/urine output established.
Repeat KβΊ within 1 hour of starting insulin β it is the parameter most likely to move fastest and most dangerously.
Hypokalaemia on treatment: reduce insulin infusion by 50%, increase KβΊ replacement, and discuss central access/higher-concentration replacement with paediatric critical care.
Glucose β€15
Add dextrose β do not stop insulin
Once BGL falls to β€15 mmol/L, change fluid to 0.9% saline + 5% dextrose (+ KCl as above) to allow the insulin infusion to continue treating the acidosis without causing hypoglycaemia.
If glucose keeps falling despite 5% dextrose β increase to 10% dextrose before reducing insulin.
Only reduce the insulin rate (not below 0.05 units/kg/hr) if glucose continues to fall despite 10% dextrose.
True hypoglycaemia (<4 mmol/L): 2 mL/kg 10% dextrose bolus, increase maintenance dextrose concentration; insulin may be paused briefly (β€1 hr) only if essential, remembering it is still needed to clear ketones.
Ongoing
Bicarbonate β almost never
Sodium bicarbonate is not routinely recommended β associated with increased cerebral oedema risk and does not improve outcomes; acidosis reverses with fluid and insulin alone.
Only purpose: improving cardiac contractility in severe shock/life-threatening hyperkalaemia β decision made with paediatric intensivist/endocrinologist at a tertiary centre.
Nil by mouth & antiemetics
Moderateβsevere DKA: NBM except ice to suck. If antiemetic needed, ondansetron only β other antiemetics sedate and can mask early cerebral oedema signs. Consider an NG tube for gastric paresis (non-mechanical delayed emptying is common in DKA). Offer oral fluids only after substantial improvement (BGL <15 mmol/L, improved consciousness, no vomiting) β proceed to oral intake and wind back IV fluids even if before the notional 48-hour mark.
8
Cerebral oedema
rare (~1%) but the dominant cause of DKA mortality β act on suspicion, never wait for imaging
Recognise it
Typically 4β12 hours after starting treatment, but can occur any time (including before treatment starts, or up to 24β48 h in)
~40% of initial CT/MRI in confirmed cases are normal β this is a clinical diagnosis
Early warning biochemistry
Failure of corrected sodium to rise (or a fall) as glucose falls β should prompt heightened vigilance
Rapid fall in calculated osmolality
Development of hyponatraemia, or rapidly falling sodium, during therapy
Suspected cerebral oedema β treat immediately, do not wait for imaging or neurology
Urgently seek paediatric critical care advice (onsite or via RSQ) β in parallel with treatment, not instead of it.
Reduce IV fluid rate (commonly by ~β ) and elevate the head of the bed 30Β°.
Give Hypertonic saline 3% (2.5β5 mL/kg over 10β15 min)orMannitol (0.5β1 g/kg IV over 20 min) β use whichever is immediately available; do not delay treatment choosing between them. May repeat after 30 minβ2 hr if needed.
Consider intubation/ventilation only if impending respiratory failure β avoid aggressive hyperventilation (associated with worse outcomes in retrospective series); if intubated, target pCOβ ~35β40 mmHg near pre-intubation levels.
CT/MRI only after treatment has started, to exclude other pathology (haemorrhage, thrombosis) β never to confirm the diagnosis before treating.
Hypertonic saline vs mannitol
Both are reasonable first agents; large retrospective paediatric series have suggested possible higher mortality with hypertonic saline versus mannitol, but this may reflect selection (sicker children given saline when mannitol had already failed) rather than a true causal effect, and practice has increasingly shifted toward hypertonic saline over the past decade. The single highest-yield action is not delaying treatment while deciding which agent to use.
9
Other complications to watch for
Hypoglycaemia
From insulin without adequate dextrose once glucose falls β the most common treatment complication; prevented by protocolised hourly glucose checks and pre-emptive dextrose at β€15 mmol/L.
Hypokalaemia / hyperkalaemia
See Β§7 β the parameter needing the closest early monitoring; ECG changes (peaked T = hyperK; flat T/U waves = hypoK) can precede lab results.
Hyperchloraemic (non-anion-gap) acidosis
From large-volume 0.9% saline β usually self-limiting once IV fluids stop; don't mistake for treatment failure (Β§3).
Hypophosphataemia
Worsens with insulin (intracellular shift). Routine replacement is not universally recommended, but ISPAD 2022 supports considering it in severe DKA to prevent muscle weakness/rhabdomyolysis if readily available.
Venous thromboembolism
Associated with central venous catheter use β minimise central line duration/necessity where possible.
Acute kidney injury
Pre-renal from dehydration; usually resolves with rehydration β monitor urine output and creatinine trend.
10
Age-specific considerations
the same protocol, different risk weighting
Age group
Key considerations
<2 years
Highest cerebral oedema risk group. Diagnosis often delayed/missed (non-specific symptoms). Consider lower-dose insulin (0.05 units/kg/hr). Lower threshold for PICU admission. Confirm diagnosis carefully β congenital/neonatal diabetes is a differential in infants.
2β5 years
Still elevated cerebral oedema risk relative to older children. ISPAD 2022 supports lower-dose insulin in this group also, particularly if newly diagnosed or glucose falling rapidly. More frequent (hourly) neuro-observation.
School age (6β11 yr)
Standard-dose insulin (0.05β0.1 units/kg/hr) generally appropriate. Consider psychosocial factors in established T1DM presenting in DKA (supervision gaps, school insulin administration issues).
Adolescents (12+ yr)
Higher-dose insulin (up to 0.1 units/kg/hr) typically well tolerated. Screen for insulin omission as a deliberate behaviour (disordered eating/"diabulimia", psychosocial stress, non-adherence) β address underlying drivers, not just the acute episode. Consider pregnancy testing in females of reproductive age. Consider euglycaemic DKA if on SGLT2 inhibitors (off-label use, or T2DM overlap).
The ISPAD 2022 insulin-dosing update
Where earlier guidelines used a flat 0.1 units/kg/hr for all ages, ISPAD 2022 explicitly supports lower-dose insulin infusions (0.05 units/kg/hr) for children under 5 years, reflecting evidence that lower doses are equally effective at resolving ketoacidosis with a theoretically reduced osmotic-shift/cerebral-oedema risk in the highest-risk age group β while noting there's no evidence higher-dose infusions are unsafe in older children.
Give the first subcutaneous insulin dose (rapid-acting with a meal, or basal) 15β30 min before stopping the IV infusion β never leave a gap, or ketosis/acidosis can rebound within hours.
New diagnosis
Start basal-bolus subcutaneous regimen (Β§13) once transitioning β this is also the natural point to begin structured diabetes education.
Known T1DM
Resume the child's usual regimen (adjusted with the diabetes team if the DKA reveals a need for dose/regimen change) and address the precipitant (pump issue, sick-day plan gap, psychosocial factors).
Ward monitoring
Continue 2β4-hourly BGL/ketones until stable; watch for rebound hyperglycaemia/ketosis if subcutaneous insulin was under-dosed at transition.
Before discharge
Diabetes educator review, sick-day management plan, glucagon/hypoglycaemia education, follow-up with paediatric endocrinology, psychosocial screen if insulin omission contributed.
12
HHS β the other hyperglycaemic emergency
rarer in children, more aggressive fluid strategy
HHS is extreme hyperglycaemia and hyperosmolality without significant ketosis/acidosis β historically a T2DM presentation, now rising in children/adolescents alongside paediatric T2DM, and also seen in neonatal diabetes and in T1DM patients with intellectual impairment who cannot signal thirst.
Diagnostic criteria
Marked hyperglycaemia (often glucose >30 mmol/L)
Venous pH >7.25 and/or HCOββ» >18 mmol/L (lactic acid may cause mild acidosis)
Effective osmolality markedly elevated; minimal or no ketosis
Altered consciousness common; frank cerebral oedema is rare in HHS (unlike DKA)
Key management differences from DKA
Dehydration is more severe and often under-recognised β polyuria/polydipsia may go unnoticed, especially in hot climates
More aggressive fluid replacement needed to expand intra- and extravascular volume and restore renal perfusion
Aim for a gradual, controlled decline in corrected sodium and osmolality β avoid rapid correction
Insulin is started later and at a lower dose than in DKA β glucose often falls substantially with fluids alone; adding insulin too early/fast risks a precipitous osmotic shift
Watch specifically for rhabdomyolysis (from hypophosphataemia), VTE (central line-associated), and malignant hyperthermia-like presentations
DKA/HHS overlap
Some children β including reported cases as young as 5 years β present with features of both (significant ketoacidosis and extreme hyperosmolality). Management follows an osmolality-guided fluid strategy with conservative correction targets and delayed, low-dose insulin, rather than a pure DKA or pure HHS protocol. This overlap group needs early senior/PICU involvement.
13
New-onset T1DM β not in DKA
subcutaneous insulin from the start
A child with new hyperglycaemia who is not acidotic (venous pH β₯7.3, HCOββ» β₯18) does not need IV insulin or an IV fluid deficit protocol β this group is managed with subcutaneous insulin from diagnosis, and admission is largely for education and safe initiation, not physiological resuscitation.
Enuresis (new, in a previously dry child) is a common trigger for presentation
Confirm ketosis status first
Check blood/urine ketones in every new hyperglycaemic presentation
If ketones are elevated (blood Ξ²-hydroxybutyrate β₯0.6 mmol/L) β start insulin as soon as possible to prevent progression to DKA, even before the full subcutaneous regimen is finalised
Confirm venous gas is not acidotic before committing to the non-DKA pathway
Starting a subcutaneous regimen
Basal-bolus (multiple daily injection) principle
Calculate the total daily dose (TDD) of insulin (typically starting around 0.5β1 unit/kg/day in a newly diagnosed, non-DKA child β lower initially if very young or highly insulin-sensitive). Split as ~40% long-acting basal insulin (once daily) and ~60% rapid-acting insulin, divided into 3 pre-meal boluses (~20% TDD each). Adjust daily based on pre-meal and bedtime glucose patterns with the diabetes team.
Alternative regimens
Twice-daily premixed or basal-bolus variants may be used depending on family/school factors, age, and local diabetes team preference β few regimens are specifically trial-validated in new-onset paediatric disease, so choice is individualised.
The "honeymoon period"
Many children enter a phase of partial endogenous insulin recovery lasting up to ~2 years post-diagnosis, with low insulin requirements (<0.5 units/kg/day) and easier glycaemic targets. Insulin doses will typically need down-titration during this phase β anticipate and counsel families, don't mistake it for "cure."
CGM at diagnosis
Increasingly initiated at or very soon after diagnosis (rather than after months of finger-prick monitoring) β supports both glycaemic pattern recognition and family confidence; local availability/funding varies.
Admission vs ambulatory pathway
Where a diabetes education/outpatient new-onset pathway exists, well children without DKA and with adequate family/social support may be suitable for ambulatory management with rapid specialist follow-up rather than routine admission β service-dependent; involve the local paediatric/endocrine team early.
Education & workup at diagnosis (not in DKA)
Immediate education
Insulin injection technique
Blood glucose/ketone monitoring
Hypoglycaemia recognition & treatment
Sick-day management principles
Baseline investigations
HbA1c (reflects recent control, prognostic)
Diabetes autoantibodies (GAD, IA-2, ZnT8, ICA)
Coeliac serology, thyroid function (associated autoimmunity)
C-peptide if type unclear
Team involvement
Diabetes educator
Dietitian
Psychology/social work (adjustment support)
Paediatric endocrinology follow-up arranged before discharge
14
Evidence & RCH vs CHQ comparison
The trial that reshaped fluid practice β PECARN FLUID (Kuppermann et al., NEJM 2018)
13-centre, 2Γ2 factorial RCT, 1255 children / 1389 DKA episodes, randomised to fast vs slow rehydration and 0.9% vs 0.45% saline. Result: no significant difference in neurological decline (GCS <14) or clinically apparent brain injury between any arm. This directly challenged the decades-old assumption that rapid/large-volume fluid resuscitation causes cerebral oedema, and supports the current model that oedema risk is driven more by the underlying severity of DKA itself (dehydration, hypocapnia, hypoperfusion) than by the specific fluid protocol chosen β provided fluids stay within a sensible, guideline-based range.
Practical takeaway: this does not license unrestricted rapid boluses β protocolised, deficit-based 48-hour replacement remains standard β but it should reduce anxiety about "getting the exact rate perfect" and refocus attention on early recognition of the sickest children (young age, severe acidosis, altered conscious state) as the real predictors of cerebral oedema risk.
Other current evidence points
ISPAD 2022 Consensus Guidelines β updated recommendation for lower-dose insulin infusion (0.05 units/kg/hr) in children <5 years; supports considering routine phosphate replacement in severe DKA where available; reinforces no role for bicarbonate outside extreme circumstances.
Balanced fluids (Plasma-Lyte) vs 0.9% saline β small paediatric RCTs (e.g. the SPinK trial) have explored whether lower-chloride balanced solutions reduce AKI/hyperchloraemic acidosis; results are promising but not yet practice-changing at a guideline level β 0.9% saline remains the standard first fluid.
Subcutaneous rapid-acting insulin for mild/moderate DKA β small prospective studies (e.g. insulin aspart 0.15 units/kg SC every 2 hours) show comparable resolution times to IV infusion in uncomplicated mildβmoderate DKA outside ICU, a option some centres are exploring to reduce ICU/HDU bed pressure β not yet standard CHQ/RCH practice and should only be used within a formal local protocol.
Clinical/estimated dehydration bands: mild β€4%, moderate 4β7%, severe β₯7%
Maintenance fluid
Reduced-rate formula (<10 kg 2 mL/kg/hr, 10β40 kg 1 mL/kg/hr, >40 kg 40 mL/hr)
Detailed maintenance table by weight (not simple Holliday-Segar) to a max of 70 kg
Initial fluid
0.9% saline; deficit + maintenance over 48 h
0.9% saline for β₯6 h; deficit + maintenance over 48 h
Insulin start
~1 h after fluids; 0.05β0.1 units/kg/hr
Comparable timing/dose range; age-based lower dosing consistent with ISPAD 2022
Potassium
Start when level known/urine output confirmed; up to 60 mmol/L
Start at 40 mmol/L, up to max 60 mmol/L; replace before insulin if KβΊ <3.0
Dextrose trigger
Add at BGL β€15 mmol/L
Change fluids at BGL β€15 mmol/L (consistent)
Cerebral oedema treatment
Hypertonic saline 3% or mannitol β whichever available first; avoid aggressive hyperventilation
Same principle; consistent escalation to PICU
Both guidelines derive from the same ISPAD evidence base. The practical difference to know is the deficit-stratification method (biochemical vs clinical) β use whichever your site's official fluid calculator is built on, and never mix pH-based and clinical-dehydration-based percentages in the same calculation.
15
Disposition & escalation
PICU / critical care from the outset
Severe DKA (pH <7.1, HCOββ» <5) β all patients with severe DKA and HHS require PICU admission