Diabetic Ketoacidosis and Hyperosmolar Hyperglycaemic State (DKA / HHS)

Meet the patient

A 19-year-old with known type 1 diabetes is brought to the emergency department drowsy and breathless at 4am. She has been vomiting for two days with a chest infection, omitted her long-acting insulin because she "could not keep anything down", and is now breathing deeply and rapidly with a sweet, chemical smell on her breath.[1][17]

The two questions that decide her first hour are the two that decide every hyperglycaemic emergency: is this DKA, HHS, or a mimic? (the venous gas and ketones answer within minutes) and is the potassium safe to start insulin? (the blood gas answers that before any infusion runs). Hold both questions and the bundle below falls into place.[1][17]

One defect, two faces — and the trap of mistaking one for the other

DKA and HHS are not two diseases; they are two expressions of one defect — insulin deficiency plus a counter-regulatory surge — read through the degree of ketogenesis. The continuum runs from pure DKA (absolute insulin deficiency, rampant ketogenesis) through mixed pictures to pure HHS (relative insulin deficiency, enough residual insulin to suppress lipolysis but not hepatic glucose output).[1][17]

DKA is defined by the triad of hyperglycaemia, ketosis, and high-anion-gap metabolic acidosis from accumulation of ketoacids (beta-hydroxybutyrate and acetoacetate). It is the commonest cause of death in type 1 diabetes under thirty and is largely preventable.[1][17]

HHS (formerly hyperosmolar non-ketotic coma, HONK) is defined by severe hyperglycaemia, hyperosmolality, and profound dehydration without significant ketoacidosis. It classically affects older type 2 patients and carries a far higher mortality than DKA — driven by age, comorbidity, and thromboembolism, not by the biochemistry.[1][9]

The classic trap: the two overlap. A type 2 patient may present with HHS-range hyperosmolality yet still have ketosis, and the management fork — fluid rate, insulin dose, anticoagulation — depends on which problem dominates. The teaching line that earns marks: acidosis drives the insulin rate, dehydration drives the fluid rate.[1][5]

The numbers that make the diagnosis — the triad, the bands, and HHS

Two test panels — a venous gas and a beta-hydroxybutyrate — settle almost every case at the bedside. Reproduce the thresholds verbatim; examiners treat them as fixed.[1][17]

[3] [5]

The DKA triad:[3]

• Hyperglycaemia — blood glucose over 250 mg/dL (about 13.9 mmol/L); recent guidelines de-emphasise this criterion because euglycaemic DKA — glucose under 250 mg/dL — is increasingly recognised, especially with SGLT2 inhibitors.[2][3]
• Ketosis — elevated blood ketones (beta-hydroxybutyrate, serum preferred) or urine ketones.[3]
• Acidosis — pH below 7.30, serum bicarbonate below 18 mEq/L, anion gap over 10.[3]

Severity — graded mild, moderate, or severe, and this triages location from ward to HDU to ICU:[9][18]

• Ketoacidosis severity is categorised mild, moderate, and severe based on serum pH and bicarbonate — clinically, DKA and HHS differ only by the degree of dehydration and the severity of the metabolic acidosis.[9][18]
• Mild DKA — the shallowest derangement, alert; moderate DKA — deeper derangement, alert or drowsy; severe DKA — the deepest acidosis, stupor or coma; manage severe DKA in HDU or ICU.[18]

HHS criteria (JBDS): marked hyperglycaemia 30 mmol/L or over, osmolality 320 mOsm/kg or over calculated as 2 times sodium plus glucose plus urea, without significant ketonaemia (3.0 mmol/L or under) and without significant acidosis (pH over 7.30, bicarbonate 15 mmol/L or over) — with marked hypovolaemia and altered cognition.[5]

Euglycaemic DKA — the ketoacidosis picture with a glucose under 250 mg/dL — is increasingly seen with SGLT2 inhibitors, and also occurs against a background of starvation, chronic liver disease, pregnancy, infection, or alcohol use. It is the one presentation where a near-normal glucose actively conceals the diagnosis.[2][3]

The four I's and an S — find the precipitant

DKA and HHS are symptoms; the precipitant is the diagnosis. Find it or the crisis recurs. Every unwell diabetic gets an active hunt, clustered once so the list stays:[1]

• Infection — the commonest, 30 to 50 percent of episodes — pneumonia, urinary tract infection, gastroenteritis, cellulitis, influenza, COVID-19, sepsis.
• Insulin omitted or pump failure — missed injections (watch eating disorders and "diabulimia" in adolescents), a faulty pen, an expired vial, a dislodged pump catheter.
• Infarction and inflammation — myocardial infarction, stroke, pancreatitis, trauma, surgery, burns.
• Introduction of a new drug — SGLT2 inhibitors (euglycaemic DKA), corticosteroids, thiazides, sympathomimetics, atypical antipsychotics, pentamidine.
• Substance and other — alcohol, cocaine, pregnancy, severe stress, eating disorders.[1]

The SGLT2 inhibitors deserve their own line. Dapagliflozin, empagliflozin, and canagliflozin cause glucosuria that lowers the glucose while ketogenesis runs unchecked — so the patient arrives acidotic with a "normal" sugar. Risk peaks with surgery, fasting, low-carbohydrate diets, alcohol, and type 1 diabetes (a boxed warning).[2][18]

One defect, six consequences — the pathophysiology

The unifying mechanism is insulin deficiency plus a counter-regulatory surge (glucagon, catecholamines, cortisol, growth hormone) released by the precipitant. Six downstream consequences explain every clinical and biochemical feature, and each maps to a treatment.[17]

1. Hyperglycaemia. Insulin deficiency and the counter-regulatory surge leave hepatic glucose output unopposed and peripheral uptake impaired — both are characterised by insulinopenia and severe hyperglycaemia — and once glucose exceeds the renal threshold it spills into the urine.[1][3]

2. Osmotic diuresis and dehydration. Glucosuria drags water, sodium, potassium, and phosphate with it. Water loss exceeds sodium loss, so the free-water deficit can reach roughly 6 litres in DKA and 9 to 12 litres in HHS. This single mechanism explains the hypernatraemia, the hyperosmolality, and the flat volume status.[1]

3. Ketogenesis — why DKA is "keto". With insulin absent, hormone-sensitive lipase runs unchecked: triglyceride breaks to free fatty acids and glycerol. In the liver, the carnitine shuttle ferries fatty acids in for beta-oxidation to acetyl-CoA. Insulin normally makes malonyl-CoA, which inhibits the shuttle; insulin absence removes the brake, and glucagon accelerates the pathway through HMG-CoA to the circulating ketoacids.[17]

• Beta-hydroxybutyrate — the predominant ketoacid in DKA (ratio 3:1 to 10:1 with acetoacetate).
• Acetoacetate — the parent compound, modest in level.
• Acetone — volatile, exhaled via the lungs, responsible for the fruity breath; not an acid, contributes little to the anion gap.[1]

Everyone forgets: urine ketone strips lie. Nitroprusside strips detect only acetoacetate, so they underestimate severity and can paradoxically rise during recovery as beta-hydroxybutyrate converts back. Only blood beta-hydroxybutyrate is reliable — measure it for diagnosis and for monitoring clearance.[1][17]

[1]

4. High-anion-gap metabolic acidosis. Ketoacids are strong acids that dissociate and release hydrogen ions buffered by bicarbonate; for every mole of ketoanion, one mole of bicarbonate is consumed. The unmeasured ketoanion widens the gap — DKA is defined by an anion gap over 10 mEq/L.[3]

5. Total-body potassium depletion behind a normal serum. This is the killing electrolyte, and it gets its own section below. At presentation the serum potassium is normal or high, yet the total-body deficit is large, because acidosis shifts potassium out of cells and osmotic diuresis dumps it in the urine — and the current UK guideline revision was driven specifically by the high prevalence of hypokalaemia during treatment.[1][4]

6. Water and sodium — the corrected sodium. Hyperglycaemia draws water out of cells and dilutes serum sodium, so the measured sodium looks low. Corrected sodium equals measured sodium plus 0.4 times (glucose minus 5.5) mmol/L.[1]

Why HHS is so much drier. HHS develops over days to weeks, allowing sustained osmotic diuresis to strip free water while residual insulin keeps ketogenesis switched off. The slow onset also lets the brain adapt, so patients tolerate very high osmolalities — until decompensation tips them into coma, seizures, focal deficits, or thrombosis.[1][7]

Clinical presentation — read the bedside, smell the breath

DKA announces itself fast — hours to one or two days — with a syndrome every final-prof candidate must recognise cold.[1]

• Hyperglycaemic symptoms — polyuria, polydipsia, weight loss, nocturia, often days of worsening control first.
• Gastrointestinal — nausea, vomiting, and diffuse abdominal pain, especially in children (the "pseudo-appendicitis" that resolves with treatment).
• Respiratory — Kussmaul breathing, deep and sighing, the compensation for metabolic acidosis.
• Breath — the fruity, nail-varnish-remover, pear-drop smell of acetone, often missed; ask a colleague to confirm it.
• Dehydration — dry mucous membranes, reduced skin turgor, sunken eyes, tachycardia, hypotension, oliguria, a postural drop.
• Neurological — lethargy, drowsiness, confusion; coma is uncommon in adults at presentation, and its presence should prompt a search for another cause.
• Features of the precipitant — fever, cough, dysuria, chest pain, focal deficit, a surgical source.[1]

HHS creeps in over days to weeks in an older patient with type 2 diabetes, comorbidity, and polypharmacy. The picture is dominated by profound dehydration, an altered conscious level proportional to osmolality, seizures, and focal deficits that mimic stroke — but no Kussmaul breathing and no acetone breath, because ketogenesis is suppressed. Coma is unusual if osmolality is below 320.[1][5]

Euglycaemic DKA hides behind the glucose. Suspect it whenever a patient on an SGLT2 inhibitor — or starved, pregnant, postoperative, or alcoholic — presents with nausea, vomiting, malaise, or fatigue with a near-normal sugar. Check ketones and a venous gas in any unwell diabetic, regardless of the glucose.[2]

The MUDPILES differential — and the coma-in-a-diabetic fork

DKA is one high-anion-gap acidosis among many. The bedside task is to separate it from the rest of MUDPILES, from HHS, and from the other causes of coma in a diabetic — because the treatment forks at each.[17]

[2] [3] [5]

The single discriminator: a high anion gap with ketosis is DKA at any glucose — under 250 mg/dL it is euglycaemic DKA, alcoholic, or starvation ketosis; glucose 30 mmol/L or over with osmolality 320 or over and minimal ketones is HHS. When the gap is high without ketones, work through the toxin and renal causes.[2][5]

Coma in a diabetic — the bedside fork

A comatose diabetic is hypoglycaemic until proven otherwise — then DKA or HHS, then everything else. Run the fork in this order at the bedside:[1][9]

• Hypoglycaemic coma — sudden onset, sweating, tremor, tachycardia, low glucose, no ketones, normal or shallow breathing. Treat immediately with IV dextrose or parenteral glucagon without waiting — hypoglycaemia is one of the three diabetic emergencies requiring prompt recognition and treatment.[9]
• DKA or HHS — ketones and a venous gas are diagnostic.
• Lactic acidosis (metformin, sepsis) — high gap, low or normal glucose.
• Other coma — stroke, post-ictal, head injury, alcohol, opiate (pinpoint pupils, give naloxone), hepatic or uraemic encephalopathy.[1]

Bedside assessment — how sick, how dry, and why

Examination answers three questions: how sick is the patient, how dehydrated are they, and what is the precipitant. The diagnostic sign is rarely on the skin; the value is in the resuscitation targets and the hunt for the cause.[1]

ABCDE, in order:[1]

• Airway and Breathing — rate and depth; Kussmaul breathing is deep and sighing, the respiratory compensation for metabolic acidosis, and dyspnoea is a common presenting symptom of DKA. If the pCO2 is higher than expected for the degree of acidosis, suspect exhaustion or aspiration — prepare to intubate.[3]
• Circulation — heart rate, blood pressure lying and sitting (a postural drop means over 10 percent volume loss), capillary refill, peripheral temperature, JVP. Cool peripheries with hypotension is critical.
• Disability — GCS, pupils, point-of-care glucose, capillary beta-hydroxybutyrate, focal deficit (especially in HHS, the stroke mimic).
• Exposure — search the skin, feet, lines, abdomen, and calf for the precipitant; calf swelling raises DVT, common in HHS.[1]

Degree of dehydration at the bedside:[1]

• Mild (under 3 percent) — not clinically apparent.
• Moderate (3 to 5 percent) — dry mucous membranes, reduced skin turgor.
• Severe (over 5 percent; in DKA around 6 percent, in HHS up to 12 percent) — sunken eyes, tachycardia, hypotension, oliguria, cool peripheries, postural drop, altered mental state.[1]

Targets that confirm adequate resuscitation: improving clinical and cognitive status, urine output 0.5 mL/kg/h or more, and — in HHS — osmolality falling gradually toward the resolution target with glucose held 10 to 15 mmol/L over the first day; in DKA, a falling ketone and improving pH trend on bedside monitoring.[5][6]

After therapy starts, reassess hourly for cerebral oedema — headache, behavioural change, irritability, incontinence, bradycardia, rising blood pressure, abnormal posturing, falling GCS, pupillary change. Any neurological deterioration during treatment of paediatric DKA is cerebral oedema until proven otherwise.[13]

Investigations — confirm, grade, find the cause, monitor

The goals are four: confirm the diagnosis, grade severity, find the precipitant, and monitor therapy. Do all four from the first blood draw.[1]

Venous blood gas (preferred over arterial — venous blood should be used rather than arterial unless respiratory problems dictate otherwise, and it is comparably useful for monitoring): pH below 7.3 in DKA, with the severity bands above; bicarbonate below 18; a negative base excess; and a low pCO2 reflecting respiratory compensation.[6]

Bedside capillary blood ketones are the method of choice for monitoring the response to treatment; if ketone measurement is unavailable, venous pH and bicarbonate with bedside glucose are used instead. Urine ketone strips detect only acetoacetate, lag treatment, and can paradoxically rise during recovery.[6]

The first blood set: electrolytes, phosphate, urea and creatinine, glucose, venous gas with lactate, ketones, full blood count, A1C, and an ECG for every patient — plus urinalysis, and amylase, lipase, transaminases, troponin, cultures and chest radiography as clinically indicated. Calculate the anion gap, the corrected sodium, and the osmolality from the first sample.[3]

Monitoring frequency during treatment:[1]

[1]

Fluids first, insulin second, potassium throughout — the first hour

The first hour is about circulation, not glucose. Resuscitation runs in a fixed order — aggressive rehydration, insulin therapy, electrolyte replacement, and treatment of the underlying precipitating event — and the order is the single most examinable fact in the topic:[1]

[4] [6]

ABCDE first. Protect the airway if GCS is under 8 and nurse in the left lateral position if vomiting; give oxygen if hypoxic and intubate if respiratory exhaustion supervenes. Insert two large-bore cannulae, draw the full blood set, and search the skin, feet, lines, abdomen, and calf for the precipitant.[1]

Fluids before insulin — and why it matters. Saline restores intravascular volume; in HHS the guideline is explicit that fluid replacement alone will lower the glucose and early use of insulin before fluids may be detrimental. Rehydration also improves tissue perfusion, which lowers the counter-regulatory surge.[1][7]

The adult JBDS regimen:[1]

• 0.9 percent sodium chloride, 1 L stat over the first hour — faster if the patient is in shock.
• Then 0.9 percent saline, 1 L every 1 to 2 hours until intravascular volume is restored.
• A typical adult schedule is 1 L in hour 1, then 1 L over 2 hours, 1 L over 2 hours, 1 L over 4 hours, 1 L over 6 hours, 1 L over 8 hours — roughly 6 L in the first 12 hours, then 8-hourly thereafter.[1]

When the blood glucose falls below 14 mmol/L, add 10 percent glucose so the fixed-rate insulin infusion can continue clearing ketones — and the updated JBDS guideline adds that de-escalating the insulin from 0.1 to 0.05 unit/kg/h should be considered at that point. The principle is non-negotiable: insulin keeps running to clear ketones even when the glucose is normal — you change the fluid, never stop the insulin.[4][6]

Avoid hypotonic fluids initially — rapid changes in osmolality during treatment are implicated in osmotic demyelination and neurological complications, so 0.9 percent saline remains the principal fluid, and an initial rise in sodium is expected and is not itself an indication for hypotonic fluids.[5][7]

The potassium trap — the killing electrolyte

Hypokalaemia is the leading preventable cause of death during DKA treatment, and it is entirely iatrogenic. Understand the mechanism and the trap closes.[4][17]

At presentation the serum potassium is normal or high, yet total-body potassium is profoundly depleted: insulin deficiency removes insulin's normal push of potassium into cells, acidosis shifts potassium out of cells in exchange for hydrogen, and osmotic diuresis dumps potassium in the urine. The updated JBDS guideline exists largely because of the high prevalence of hypokalaemia (with hypoglycaemia) under previous protocols.[1][4]

The moment you give insulin and saline, potassium is driven back into cells and washed into the urine — hypokalaemia is one of the two most prevalent treatment-related harms the current guideline was revised to reduce, and a patient treated without potassium cover can descend into ventricular ectopics, VT, VF, or asystole.[4]

Check the potassium before insulin starts, and choose the regimen from it:[1]

Practical limits: follow local protocols for the maximum peripheral potassium concentration and infusion rate, with continuous cardiac monitoring while potassium is running. Recheck the electrolytes frequently during correction and at any arrhythmia.[4]

The seven pillars of definitive care

Once the first hour is run, definitive care rests on seven pillars — name them in this order in the viva.[17]

Pillar 1 — fixed-rate intravenous insulin infusion (FRIII). Weight-based fixed-rate insulin has replaced sliding scales: insulin is infused intravenously at a weight-based fixed rate until the ketosis has resolved.[6]

• Soluble insulin 0.1 unit/kg/hr IV — the classic randomised-trial regimen in DKA, and the rate from which the updated JBDS guideline de-escalates (to 0.05 unit/kg/hr, considered once glucose is below 14 mmol/L).[4][11]
• Targets: a steady fall in blood ketones with improving pH and bicarbonate on bedside ketone and venous gas monitoring.[6]
• Continue FRIII until ketosis has resolved and the patient is eating; in HHS, FRIII commences only once osmolality stops falling with fluid replacement, unless there is ketonaemia — and the euglycaemic-DKA guidance advises a slower transition to subcutaneous insulin to prevent relapse.[5][18]

Pillar 2 — fluids, continued. 0.9 percent saline remains the principal fluid to restore circulating volume and reverse dehydration; when the glucose falls below 14 mmol/L, add 10 percent glucose so the fixed-rate insulin can continue. Expect the sodium to rise initially — that alone is not an indication for hypotonic fluids.[6][7]

Pillar 3 — potassium, throughout. Replace potassium according to the serum level from the outset of therapy, guided by frequent electrolyte checks and cardiac monitoring where potassium is being infused — the guideline revision was driven by the high prevalence of hypokalaemia under previous protocols.[4][5]

Pillar 4 — find and treat the precipitant. Treatment always includes the underlying precipitating event: antibiotics after cultures for infection, cardiac care for infarction, the stroke pathway, and stopping the offending drug (for example an SGLT2 inhibitor in euglycaemic DKA) — the underlying etiology must be determined and treated.[2][3][5]

Pillar 5 — bicarbonate is usually avoided. In a randomised trial of severe DKA (arterial pH 6.9 to 7.14), bicarbonate did not speed the fall in glucose or ketones or the rise in pH or bicarbonate in blood or cerebrospinal fluid, and did not shorten time to pH 7.3 or bicarbonate 15 mEq/L. In children, bicarbonate treatment was the only therapeutic variable associated with cerebral oedema (relative risk 4.2). Reserve it for exceptional circumstances only.[10][13]

Pillar 6 — phosphate. Routine phosphate replacement is not recommended: in a randomised trial it raised serum levels but did not affect the duration of DKA, the insulin dose needed to correct the acidosis, or morbidity and mortality — phosphate is "not an essential part of the therapy for DKA in most patients". Evaluate phosphate among the baseline electrolytes and treat only selected patients with severe symptomatic hypophosphataemia.[3][11]

Pillar 7 — convert IV to subcutaneous insulin without a gap. The half-life of IV insulin is only about 5 minutes, so stopping the infusion before subcutaneous insulin is on board causes rebound ketoacidosis within hours. Always overlap.[6]

The conversion protocol:[6]

1. Continue (or start) the patient's long-acting basal insulin — JBDS says long-acting analogues such as insulin glargine or insulin detemir should be continued in usual doses — so the basal is on board while the IV insulin is still running.[6]
2. Continue IV insulin and glucose until the patient is eating, then cover the meal with rapid-acting insulin.
3. Stop the IV insulin only once subcutaneous insulin is established; check ketones after stopping and restart the infusion if they are rising.[6]
4. For new-onset diabetes, start the standard basal-bolus regimen with the diabetes specialist team.[6]

HHS — the same engine, a different gearbox

HHS shares the engine of DKA but needs a slower, gentler, and more anticoagulated hand. The differences are the examinable heart of the topic.[5]

[5] [7]

The single sentence for the viva: HHS is fluid-led, insulin-cautious, and anticoagulated; DKA is insulin-led, potassium-guarded, and anticoagulation is selective.[5][7]

Thromboembolism in HHS is addressed by a dedicated "prevention of harm" theme in the JBDS pathway — VTE prophylaxis sits alongside prevention of osmotic demyelination, fluid overload, and foot ulceration, unless contraindicated.[5]

The scenarios that bite — euglycaemic DKA, children, pregnancy, CKD

Euglycaemic DKA. Recognise glucose under 250 mg/dL with ketosis and high-anion-gap acidosis in a patient on an SGLT2 inhibitor, or starved, pregnant, postoperative, alcoholic, or with chronic liver disease or infection. Manage with IV fluids plus insulin plus glucose — the insulin to clear ketones, the glucose to avoid hypoglycaemia — and determine and treat the underlying etiology, including stopping the SGLT2 inhibitor; the updated JBDS guideline carries a dedicated section on SGLT2-inhibitor-associated euglycaemic ketoacidosis.[2][4]

Paediatric DKA. Cerebral oedema is the feared complication — clinically apparent brain injury occurred in 0.9 percent of episodes in the PECARN FLUID trial, and declines in mental status (two consecutive GCS scores under 14) in 3.5 percent. In that trial, neither the rate of fluid administration nor the sodium chloride content (0.9 versus 0.45 percent) significantly influenced neurologic outcomes. Insulin runs as a low-dose IV infusion, and bicarbonate is not routine — in the case-control study, bicarbonate was the only therapeutic variable associated with cerebral oedema (relative risk 4.2).[11][12][13]

The FLUID trial (Kuppermann, 2018) randomised 1389 children in a two-by-two factorial of fast versus slow rehydration and 0.9 percent versus 0.45 percent saline, and found no difference in neurologic outcomes — refuting the long-held belief that aggressive fluids cause cerebral oedema within standard protocols.[12]

DKA in pregnancy. DKA in pregnancy is an obstetric emergency with risk of maternofetal death: in a modern cohort there were no maternal deaths but fetal demise in 17.2 percent of pregnancies, with high rates of pre-eclampsia, neonatal hypoglycaemia and neonatal intensive care admission; across a systematic review, stillbirth complicated 7 to 35 percent of pregnancies (most recent studies 7 to 31 percent), with preterm birth in 25 to 83 percent. Maintain a high index of suspicion in any pregnant woman with suggestive symptoms, and focus prenatal diabetes education on DKA prevention.[14][15]

HHS in the elderly type 2 patient. Insidious onset over days, often precipitated by infection, stroke, MI, or new drugs; coma, seizures, focal deficits, and profound dehydration that mimics stroke. Mortality among patients with HHS is around tenfold that of DKA, determined by the severity of dehydration, comorbidity, and age over 60. Correct gradually per the JBDS pathway and hunt hard for the precipitant.[5][9]

DKA with chronic kidney disease. Presenting potassium may be very high (no urinary loss) — use slower insulin, monitor hourly, and consider calcium resonium or dialysis for refractory hyperkalaemia. Insulin clearance is reduced, so doses are smaller and the duration longer. Watch for fluid overload with a central line and daily weights.[1]

Cerebral oedema — the paediatric killer

Cerebral oedema complicates a small fraction of paediatric DKA episodes and is devastating when it strikes, with significant morbidity and mortality — it is the disaster of the topic, and most cases are preventable.[12][13]

Risk factors (Glaser, 2001): low initial arterial pCO2 (relative risk 3.4 per 7.8 mm Hg fall), high serum urea nitrogen at presentation (relative risk 1.7 per 9 mg/dL rise), and treatment with bicarbonate (relative risk 4.2, the only therapeutic variable that remained significant). Of these, the one you control at the bedside is bicarbonate — avoid it.[13]

Clinical features, evolving 4 to 12 hours into treatment: headache, behavioural change, irritability, incontinence, bradycardia with rising blood pressure (Cushing's triad), abnormal posturing, falling GCS, pupillary change, and seizures.[1]

Emergency management:[16]

• Hyperosmolar therapy — mannitol, or 3 percent hypertonic saline, the latter an increasingly used, accepted alternative that has partly replaced mannitol in children's hospitals.[16]
• Support the airway and ventilation if they fail; raise the head of the bed.
• Reduce insulin and fluids, transfer to PICU or ICU, and exclude other causes of deterioration with urgent imaging.[16]

How patients with DKA and HHS come to harm — the preventable list

• Hypokalaemia from insulin given before the potassium is checked — a leading preventable death.[4]
• Cerebral oedema in a child treated with bicarbonate (relative risk 4.2 in the case-control study).[13]
• Rebound ketoacidosis from stopping the IV infusion before subcutaneous basal insulin is on board.[6]
• Hypoglycaemia from over-aggressive insulin — the harm the updated JBDS guideline was revised to reduce, with de-escalation of the insulin rate once glucose is below 14 mmol/L.[4]
• A thromboembolic event in HHS where VTE prophylaxis was omitted.[5]
• A surgical laparotomy for an "acute abdomen" that was DKA all along.[1]
• Euglycaemic DKA missed because the glucose looked normal and nobody checked ketones.[2]
• Recurrent DKA because the precipitant was never found and the sick-day rules were never taught.[1]

Prognosis, disposition, and recurrence

Mortality is set by the syndrome, not the biochemistry. Overall mortality among children and adults with DKA is under 1 percent in good centres; mortality in HHS is about tenfold higher — driven by age, comorbidity, and the severity of dehydration.[9][17]

Where to manage:[1]

• Mild DKA — diabetes or admissions ward with hourly ketone and glucose monitoring.
• Moderate DKA — admissions unit with close monitoring; HDU if potassium is borderline, the patient is elderly, or there is cardiac disease.
• Severe DKA — HDU or ICU.
• All paediatric DKA — paediatric ward with senior review; PICU if mental status declines, GCS is reduced, or cerebral oedema is suspected.[12]
• HHS — HDU or ICU for most elderly patients (comorbidity, fluid balance, anticoagulation).[1]

Recurrence is the rule unless you interrupt it. Twenty-five to fifty percent of DKA is recurrent, driven by intermittent insulin omission, an eating disorder, or inadequate education. Education, psychological support, a structured diabetes review, and sick-day rules — never stop insulin, check ketones when unwell, drink fluids, call for help if vomiting — cut recurrence.[1]

The trials and the guidelines

The evidence base is small but decisive; name the trials and what each changed.[1]

[17] [10] [13] [12] [11] [2]

Regional deltas — the framework is global, the fluids and doses differ slightly:[1]

[4] [5] [3] [8] [1] [1]

Controversies to handle calmly: bicarbonate at very low pH (the randomised trial found no benefit even at pH 6.9 to 7.14; reserve for exceptional circumstances); routine phosphate (no outcome benefit in the randomised trial); ward versus ICU for moderate DKA (many centres manage mild-to-moderate on the admissions unit); and the mechanism of cerebral oedema — the FLUID trial shifted thinking away from fluid rate as the cause.[10][11][12]

The mantra, and the mnemonics

[17] [1] [4] [5] [6] [11]

The mantra: fluids, insulin, potassium — in that order; treat a low potassium before the insulin.[4][17]

Ward-round test — three stems, thirty seconds each

Stem 1 — the teenager from the top of the topic (answer)

The 19-year-old with type 1 diabetes, omitted basal insulin, vomiting, deep rapid breathing, and pear-drop breath. Potassium comes back at 3.1 mmol/L and the registrar has already drawn up the insulin. What do you do? Model: This is DKA with a treatment-threatening potassium. Defer the insulin — at 3.1 mmol/L, insulin would drive potassium lower and risk ventricular arrhythmia, and the current UK guideline was revised specifically because of the prevalence of hypokalaemia. Run 0.9 percent saline with potassium replacement under ECG monitoring, recheck the potassium, and start fixed-rate IV insulin once it is safe. Continue the usual long-acting basal insulin, send cultures, treat the chest infection, and monitor with bedside ketones. The preventable death here is hypokalaemia.[4][6]

Stem 2 — the older man with a glucose of 38 and no ketones (answer)

A 74-year-old with type 2 diabetes is brought in drowsy. Glucose is 38 mmol/L, osmolality 328 mOsm/kg, pH 7.36, ketones 0.4 mmol/L, sodium 154 mmol/L. The team plans FRIII 0.1 unit/kg/hr and asks whether to anticoagulate. What is right? Model: This is HHS, not DKA — glucose 30 or over, osmolality 320 or over, pH over 7.3, ketones 3.0 or under. Use 0.9 percent saline as the principal fluid and withhold insulin until the glucose is no longer falling on fluids alone (there is no significant ketonaemia), aiming to lower osmolality by 3 to 8 mOsm/kg/h with glucose kept 10 to 15 mmol/L in the first 24 hours, correcting over 24 to 72 hours, and preventing VTE because harm prevention is a dedicated theme of the pathway. Hunt for stroke, MI, and sepsis.[5][7]

Stem 3 — the postoperative patient on an SGLT2 inhibitor (answer)

A 52-year-old, two days after surgery and on empagliflozin, is vomiting and breathless. Glucose is 9 mmol/L, but the venous gas shows pH 7.18, bicarbonate 12, and ketones 4.2 mmol/L. What is this and what is the first move? Model: This is euglycaemic DKA — the SGLT2 inhibitor caused glucosuria that lowered the glucose while ketogenesis ran unchecked, so the near-normal glucose conceals a ketoacidosis. Stop the empagliflozin, give IV fluids with insulin and glucose (the insulin to clear ketones, the glucose to avoid hypoglycaemia — resuscitation with intravenous fluids, insulin, and glucose is the management), check and replace the potassium as for any DKA, and admit. Do not be reassured by the glucose.[2]

References

1. [1]Fayfman M, Pasquel FJ, Umpierrez GE. Management of Hyperglycemic Crises: Diabetic Ketoacidosis and Hyperglycemic Hyperosmolar State Med Clin North Am, 2017.PMID 28372715
2. [2]Long B, Lentz S, Koyfman A, Gottlieb M. Euglycemic diabetic ketoacidosis: Etiologies, evaluation, and management Am J Emerg Med, 2021.PMID 33626481
3. [3]Veauthier B, Levy-Grau B. Diabetic Ketoacidosis: Evaluation and Treatment Am Fam Physician, 2024.PMID 39556629
4. [4]Dhatariya KK; Joint British Diabetes Societies for Inpatient Care. The management of diabetic ketoacidosis in adults-An updated guideline from the Joint British Diabetes Society for Inpatient Care Diabet Med, 2022.PMID 35224769
5. [5]Mustafa OG, Haq M, Dashora U, Castro E, Dhatariya KK; Joint British Diabetes Societies (JBDS) for Inpatient Care Group. Management of Hyperosmolar Hyperglycaemic State (HHS) in Adults: An updated guideline from the Joint British Diabetes Societies (JBDS) for Inpatient Care Group Diabet Med, 2023.PMID 36370077
6. [6]Savage MW, Dhatariya KK, Kilvert A, Rayman G, Rees JA, Courtney CH, Hilton L, Dyer PH, Hamersley MS; Joint British Diabetes Societies. Joint British Diabetes Societies guideline for the management of diabetic ketoacidosis Diabet Med, 2011.PMID 21255074
7. [7]Scott AR; Joint British Diabetes Societies (JBDS) for Inpatient Care; JBDS hyperosmolar hyperglycaemic guidelines group. Management of hyperosmolar hyperglycaemic state in adults with diabetes Diabet Med, 2015.PMID 25980647
8. [8]Umpierrez GE, Davis GM, ElSayed NA, et al. Hyperglycemic Crises in Adults With Diabetes: A Consensus Report Diabetes Care, 2024.PMID 39052901
9. [9]Umpierrez G, Korytkowski M. Diabetic emergencies - ketoacidosis, hyperglycaemic hyperosmolar state and hypoglycaemia Nat Rev Endocrinol, 2016.PMID 26893262
10. [10]Morris LR, Murphy MB, Kitabchi AE. Bicarbonate therapy in severe diabetic ketoacidosis Ann Intern Med, 1986.PMID 3096181
11. [11]Wilson HK, Keuer SP, Lea AS, Boyd AE 3rd, Eknoyan G. Phosphate therapy in diabetic ketoacidosis Arch Intern Med, 1982.PMID 6802095
12. [12]Kuppermann N, Ghetti S, Schunk JE, et al. Clinical Trial of Fluid Infusion Rates for Pediatric Diabetic Ketoacidosis N Engl J Med, 2018.PMID 29897851
13. [13]Glaser N, Barnett P, McCaslin I, et al. Risk factors for cerebral edema in children with diabetic ketoacidosis. The Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics N Engl J Med, 2001.PMID 11172153
14. [14]Dhanasekaran M, Mohan S, Erickson D, Shah P, Szymanski L, Adrian V, Egan AM. Diabetic Ketoacidosis in Pregnancy: Clinical Risk Factors, Presentation, and Outcomes J Clin Endocrinol Metab, 2022.PMID 35917830
15. [15]Cozzi-Glaser GD, Davis AM, Bell M, Battarbee AN. Pregnancy outcomes following diabetic ketoacidosis: a systematic review Am J Obstet Gynecol MFM, 2025.PMID 40447103
16. [16]Decourcey DD, Steil GM, Wypij D, Agus MS. Increasing use of hypertonic saline over mannitol in the treatment of symptomatic cerebral edema in pediatric diabetic ketoacidosis: an 11-year retrospective analysis of mortality Pediatr Crit Care Med, 2013.PMID 23863818
17. [17]Kitabchi AE, Umpierrez GE, Miles JM, Fisher JN. Hyperglycemic crises in adult patients with diabetes Diabetes Care, 2009.PMID 19564476
18. [18]Chow E, Clement S, Garg R. Euglycemic diabetic ketoacidosis in the era of SGLT-2 inhibitors BMJ Open Diabetes Res Care, 2023.PMID 37797963