Management of diabetic ketoacidosis in children

Learning objectives.

By reading this article, you should be able to:

• •Recall the pathophysiology of diabetic ketoacidosis (DKA).
• •Diagnose a child with DKA accurately.
• •Explain the principles of management of a child with DKA
• •Identify and treat the most common and serious complications of DKA in children, including cerebral oedema.

Key points.

• •Diabetic ketoacidosis is a common cause of morbidity in children with diabetes.
• •Guidelines have recently changed and place a greater emphasis on replacement of fluids.
• •Cerebral oedema is a significant cause of mortality.
• •All children require close monitoring during treatment and may require transfer to the ICU.
• •Identifying and addressing precipitating factors are essential parts of management.

Diabetic ketoacidosis (DKA) is a serious and potentially life-threatening complication of Type 1 and rarely Type 2 diabetes. It remains a significant cause of morbidity in children with diabetes and has an estimated mortality rate of 0.15–0.5%.¹ Prompt recognition and management of DKA and its complications are vital.

Balancing adequate replacement of fluids to rehydrate the patient while minimising risk of cerebral oedema is key. The updated guidelines of the British Society for Paediatric Endocrinology and Diabetes (BSPED) and the International Society for Pediatric and Adolescent Diabetes (ISPAD) reflect new evidence and are summarised and evaluated. This review focuses on the treatment of patients aged 16 yrs and younger. It may be appropriate that patients aged 16–18 yrs are treated according to adult guidelines, depending on their place of admission.²

Pathophysiology

Diabetic ketoacidosis results from a deficiency in circulating insulin, which results in an accelerated catabolic state, increased glycogenolysis and gluconeogenesis, impaired peripheral glucose utilisation and increased lipolysis and ketogenesis. This condition ultimately leads to hyperglycaemia, hyperosmolality, ketonaemia and metabolic acidosis. An osmotic diuresis occurs because of increased urinary glucose concentrations with subsequent severe electrolyte loss and dehydration.

In children, DKA either results from insulin deficiency or increased requirements for insulin. Insulin deficiency may result from undiagnosed Type 1 diabetes, omission of insulin therapy or failure of an insulin pump whereas an increased insulin requirement may occur when counter-regulatory hormones, such as adrenaline (epinephrine), are increased as a result of stress triggered by sepsis or trauma, for example. In this case, an increased ‘sick day’ insulin dosing strategy should be implemented.

Diagnosis

The specific clinical features (Box 1) and metabolic derangement at presentation depend on the severity and duration of illness, which are often influenced by the age of the child.

Box 1.

Clinical signs of DKA in children

Nausea with or without vomiting

Anorexia

Dehydration (e.g. tachycardia and decreased skin turgor)

Kussmaul's breathing

Acetone breath

Dehydration

Abdominal pain, possibly focal in origin and mimicking appendicitis or an acute abdomen

Lethargy

Confusion/drowsiness

Coma

Diagnosis of diabetes and DKA in younger children is often later as a result of delay in symptoms, such as polydipsia, being recognised and reported. Capillary blood glucose should always be measured promptly in any unwell child and followed up by a formal laboratory measurement.

DKA as a first presentation of of Type 1 diabetes is more common in younger children, and in ethnic minority and low socioeconomic groups. It is important to recognise that the risk of developing DKA in any child may be strongly influenced by social and psychological factors (Box 2).³ It is vital that all clinicians consider the individual circumstances of each child with diabetes that they encounter to identify children at higher risk before an episode of DKA occurs.

Box 2.

Risk factors for DKA³

Younger age at diagnosis

Ethnic minority populations

Low socioeconomic groups

History of psychiatric disorder, including eating disorders or drug and alcohol misuse

Unstable family circumstances

Poor compliance with therapy or poor control of blood glucose

Concurrent illness (e.g. gastroenteritis with severe or persistent vomiting)

Peripubertal and adolescent girls

Compliance with medication is often a factor in the development of DKA. In an international study of more than 30,000 children, the use of an insulin pump has been shown to reduce the incidence of DKA compared with insulin injection therapy.⁴

If DKA is suspected based on clinical presentation, venous blood gas, glucose and ketone concentrations should be measured. The diagnostic criteria are as follows.

• (i)Venous pH <7.3 or serum bicarbonate <15 mmol L⁻¹

and.

• (ii)Ketonaemia (blood β-hydroxybutyrate ≥3 mmol L⁻¹ or moderate or large (≥2+) ketonuria)²

Hyperglycaemia with a concentration greater than 11 mmol L⁻¹ is present in the majority of cases of DKA. Occasionally, children may present with ‘euglycaemic’ DKA, which may occur if they have recently injected insulin before admission or have taken sodium–glucose transporter-2 (SGLT-2) inhibitors. Current guidelines are that these patients should be treated in the same way as patients with hyperglycaemic DKA.⁵

Near-patient testing of blood ketone concentration is ideal. Urinary ketones are helpful to make the diagnosis but are not useful in monitoring response to treatment. Urinary ketone values of ++ are typically equivalent to blood β-hydroxybutyrate concentration of >3.0 mmol L⁻¹. Point-of-care ketone tests are accurate up to 3–5 mmol L⁻¹.²

Once a diagnosis of DKA has been made, an assessment of severity should be the next step. The degree of acidosis provides an indication of the extent of dehydration and thus guides fluid replacement, as specified by the updated BSPED guidelines (Table 1).²

Table 1.

Assessment of the severity of DKA.

Degree of acidosis	Degree of dehydration (%)
Mild: venous pH <7.3 or bicarbonate <15 mmol L−1	5
Moderate: pH <7.2 or bicarbonate <10 mmol L−1	5
Severe: pH <7.1 or bicarbonate <5 mmol L−1	10

It is also important at this stage to make the differentiation between DKA and a hyperosmolar hyperglycaemic state (HHS). A very high blood glucose concentration (>30 mmol L⁻¹) in a child with little or no acidosis or ketones, and altered consciousness or seizures (occurs in approximately 50%), should raise the index of suspicion of HHS. The strategy for fluid replacement is more aggressive in HHS because dehydration is more pronounced and insulin is commenced later. This highlights the importance of differentiating between DKA and HHS.⁵

Management

Diabetic ketoacidosis is a medical emergency, and the management should be directly led by a senior paediatrician. An ‘ABC’ approach is essential with early i.v. access and placement of a nasogastric tube, especially if there are concerns with level of consciousness or vomiting. Figure 1 summarises the initial updated management plan according to the BSPED guidelines.²

Fig 1.

Summary of management of DKA in children according to the updated BSPED guidelines 2021.

The overall aims of DKA management are to.

• (i)Restore circulating volume
• (ii)Switch off ketogenesis and make glucose available for intracellular processes
• (iii)Closely monitor for complications, particularly cerebral oedema
• (iv)Investigate and treat the underlying cause; consider antibiotics

Fluids

Regimens for fluids aim to rehydrate the child effectively whilst minimising the risk of cerebral oedema. Historically, there has been a tendency to restrict fluid replacement in DKA, particularly in children, because of the potential risk of cerebral oedema. This approach is evolving, partly as a result of evidence generated by the Pediatric Emergency Care Applied Research Network (PECARN) study. This multicentre RCT compared acute and long-term neurological outcomes in more than 1,300 episodes of DKA in children treated with fluids at slower or more rapid rates, using either saline 0.45% or saline 0.9%. There was no significant difference in the frequency of altered mental status or diagnosis of cerebral oedema and no difference in long-term neurocognitive outcomes.⁶

Amended guidelines from BSPED and ISPAD published in 2021 and 2018, respectively, reflect this new evidence. Meticulous management of fluid balance remains key, with an increased emphasis on the importance of treating shock and replacing circulating volume with greater volumes of fluids (40 ml kg⁻¹) before vasopressor drugs are considered. However, the BSPED guidelines do advise giving only 10 ml kg⁻¹ initially if shock is suspected, in line with the Resuscitation Council UK (2021) guidelines.⁷ In contrast, the ISPAD guidance recommends 20 ml kg⁻¹ as an initial bolus. This more cautious approach is supported by the National Institute for Health and Care Excellence (NICE) guidelines (2022).⁸

Strict recording of fluid input and output is imperative when managing DKA. According to the BSPED, ISPAD and NICE guidelines, fluid deficits should be replaced in the first 48 h of hospital admission.^2,5,8 The overall requirement is calculated by adding the fluid deficit to the fluid maintenance requirements and subtracting any volume already given (excluding boluses used for shock) (Box 3). Ideal body weight should be used to avoid overhydration in patients with obesity.

Box 3.

Example of the calculation of fluids required in DKA in the first 48 h

Requirement = (deficit – resuscitation (unless shocked)) + maintenance
5-yr-old female (18 kg) presents with moderate DKA, with pH 7.14 and 5% dehydration:
Initial fluids bolus: 10 ml kg−1 crystalloid fluid for dehydration=180 ml.
Deficit	Weight × deficit × 10
	18 × 5 × 10=900 ml
	Minus initial bolus (900–180)=720 ml
	Infusion over 48 h=15 ml h−1
Maintenance: using Holliday–Segar method	100 ml kg−1 for first 10 kg=100 × 10=1,000 ml
	50 ml kg−1 for next 10 kg=50 × 8=400 ml
	Total maintenance=1,400 ml per 24 h
	Infusion over 24 h=58.3 ml h−1

Total fluids= 15 +58.3 ml h ⁻¹ = 73.3 ml h ⁻¹

Controversies

The changes to recommended guidelines have been questioned, given the long-standing concerns regarding the development of cerebral oedema. The possibility that restrictive regimens increase the possibility of acute kidney injury (AKI) or cerebral hypoperfusion, thus potentially exacerbating the development of cerebral oedema, has been one of the main triggers for a change. However, some have highlighted the possible limitations of the BSPED and ISPAD guidelines, given their reliance on expert opinion and unpublished audit data, and the PECARN study because children with a Glasgow Coma Scale (GCS) <11 were not included. Indeed, very few patients in the study had a GCS <15; thus, the sickest children were excluded, and there were no incidences of brain injury, suggesting that it was not sufficiently powered to assess this risk. Furthermore, some authors point out that the metabolic derangements in DKA are associated with much lower mortality rates compared with the same abnormalities in non-DKA-related conditions assessed using the Pediatric Index of Mortality (PIM) 3 score.⁹ Moreover, there is no consensus on shock and its severity in children, which is further confounded in the context of DKA. Acidosis can lead to tachypnoea hypocapnia and tachycardia; these occur commonly in children with DKA and do not necessarily indicate shock. Furthermore, peripheral capillary refill time is not a reliable indicator of shock in the context of DKA, as hypocapnia may cause peripheral vasoconstriction and a misleading prolonged capillary refill time.⁸ Children with DKA may therefore receive increased volumes of fluids if shock is misdiagnosed.⁹ Nevertheless, the PECARN trial provides valuable evidence negating the risk of negative neurological outcomes with more liberal fluid regimens and has heralded a shift in the focus of management of DKA in children. Furthermore, a recent retrospective cohort study conducted over 30 months showed no significant difference in rates of admission to a paediatric intensive care unit (PICU) or cerebral oedema between a group of 150 children with DKA treated according to 2015 or 2021 guidelines, despite the latter group receiving more fluids (average 28%).¹⁰

The role of vasogenic factors leading to cerebral ischaemia has also gained support, thus superseding the importance of osmotic changes in the development of cerebral oedema. Cerebral oedema has been identified in some children before treatment has started.^11,12 Recent studies have shown a positive correlation between cerebral oedema on MRI scans and Paco₂ or degree of hyperventilation, dehydration and urea concentration, thus suggesting that both osmotic and vasogenic changes contribute to cerebral oedema in children with DKA. Moreover, it has been observed that absolute glucose concentration or its rate of change had no effect on cerebral outcomes.13, 14, 15

Overall, it is clear that the pathogenesis of cerebral oedema is complex and is likely to be multifactorial involving vasogenic, osmotic, inflammatory and ischaemic mechanisms. This is supported by evidence of children presenting with brain injury before treatment has been commenced.¹¹ More liberal fluid regimens have been used without harm being demonstrated.⁶ However, many dispute whether this evidence is sufficient. Consequently, clinicians are advised to use their own local guidance and clinical judgement and to remain extremely vigilant when treating children with DKA.

Insulin

Insulin therapy stops ketone production and promotes metabolism of existing ketoacids to produce bicarbonate. A low-dose i.v. infusion of insulin at 0.05 units kg⁻¹ h⁻¹ should be sufficient in most cases, unless DKA is severe. This will also reduce the risk of subsequent hypoglycaemia especially in children aged <5 yrs. A bolus dose of insulin should not be used at any stage during treatment, and if an insulin pump is in place, this should be stopped before starting an i.v. infusion of insulin. For children on long-acting insulin, it is generally recommended to continue this alongside the i.v. infusion.²

The infusion should be continued until the resolution of DKA (pH >7.3; serum bicarbonate >15 mmol L⁻¹; ketones <1 mmol L⁻¹). It is important to note that this will take longer than the resolution of blood glucose concentration, and glucose concentration per se should not be used as an indicator for stopping insulin therapy unless it is extremely low (see section on hypoglycaemia below).

Sodium

Increased serum glucose affects the measured sodium concentration through osmotic forces. As glucose is largely restricted to the extracellular space, water is drawn out of cells by osmosis, resulting in a dilutional hyponatraemia. When DKA is treated and blood glucose concentrations fall, then sodium concentration is expected to increase. A failure of sodium concentrations to increase, or indeed a decrease, may be a risk factor for cerebral oedema, as it is an indicator of free water in the circulation; it should be acted on promptly (Fig. 1).² Conversely, hypernatremia may protect against cerebral oedema. It is recommended that corrected sodium concentrations, defined as (corrected sodium [mmol L⁻¹]=measured Na⁺+[[plasma glucose–5.6]/3.5]), should be monitored rather than serum sodium concentration to help quantify the true sodium and water deficit.^2,5

Precipitating factors

At the time of presentation, there should be a full examination of the patient and investigation of the precipitating cause of DKA, such as an underlying infection or adherence with medication.

Antibiotics should be given to a child with signs of infection, with or without a fever. There should be a high index of suspicion for sepsis in a child with fever or hypothermia, decompensated shock, increased lactate or persisting acidosis.⁸

Troubleshooting

It is important to regularly monitor the response to treatment. This includes routine observations in addition to a capillary glucose, venous pH, bicarbonate, Paco₂ and blood ketone concentrations at least every 2–4 h or more frequently if indicated. A lack of response to treatment should mandate an urgent review by a senior clinician and a full reassessment of the child. Several causes should be considered (Box 4).² If the blood ketone concentrations are not falling within 4–6 h, the insulin infusion may be increased to 0.1 units kg⁻¹ h⁻¹.

Box 4.

Reasons for persisting acidosis in DKA

•Insufficient insulin to switch off ketones (including incorrect insulin infusion; i.v. cannula blocked)

•Inadequate resuscitation (incorrect assessment of child or fluid calculation error)

•Sepsis

•Hyperchloraemic acidosis

•Salicylates or other drugs (prescription or recreational)

A persisting acidosis may or may not result from high plasma ketone concentrations, and other causes of a low pH should be excluded (e.g. increased lactate from result of sepsis or insufficient circulating volume). Sodium bicarbonate should never be used to treat an acidosis in DKA unless there is evidence of poor cardiac function secondary to refractory hyperkalaemia or severe acidosis, and it should only be considered in consultation with a paediatric intensivist.⁸

Complications

Hyperchloraemic metabolic acidosis

Hyperchloraemia may develop as a result of large amounts of chloride-rich fluids and preferential renal excretion of ketones over chloride, which will exacerbate the existing acidosis of a child with DKA. According to Stewart's strong ion theory, electrical neutrality must be maintained by balancing positively and negatively charged ions and weak acids. Bicarbonate competes with other negatively charged ions to buffer the solution adequately. If the chloride concentration is high, there is less space for bicarbonate to occupy, and its buffering capacity is compromised. Chloride is therefore acidifying when it is in excess. It is important to distinguish the cause of an ongoing acidosis between a persisting ketosis and hyperchloraemia. The formula (base excess attributable to chloride=[sodium–chloride]–32) can be used.⁵

Using crystalloid fluids with a lower chloride content, such as Plasma-Lyte or compound sodium lactate, may reduce the likelihood of hyperchloraemia. There is no specific treatment for lowering chloride concentration; it will resolve spontaneously when the child is no longer requiring i.v. fluids and should not delay progressing the child to oral fluids and subcutaneous insulin if they are well.⁸

Hypokalaemia

It is essential to remember that when treating DKA, the body is depleted of potassium because of losses as a consequence of osmotic diuresis, vomiting and secondary hyperaldosteronism, which increases urinary potassium excretion. Potassium concentrations at presentation may be normal, increased or decreased. Movement of potassium from cells into the extracellular space will often mask the significant underlying potassium depletion.⁵

Insulin therapy drives potassium into cells, and its aldosterone-like effect leads to increased urinary potassium excretion. Concentrations may therefore decrease abruptly on starting insulin, potentially predisposing to cardiac arrhythmias. Adding potassium to maintenance fluids is therefore important unless the potassium concentration is high or the patient has not passed urine. Potassium should be replaced at a maximum rate of 0.5 mmol kg⁻¹ h⁻¹.⁵

Hypoglycaemia

Glucose should be closely monitored throughout treatment. If low, it may lead to symptoms, such as dizziness, confusion, sweating, seizures and loss of consciousness. Therefore, during treatment, if blood glucose decreases below 6 mmol L⁻¹, the glucose concentration of i.v. fluids should be increased. If blood glucose concentrations fall below 4 mmol L⁻¹, a bolus of dextrose 10% (2 ml kg⁻¹) should be given immediately, and the glucose concentration of the fluids should again be increased. During periods of particularly low glucose concentrations, the insulin infusion may be paused for up to an hour whilst being mindful that insulin is needed to switch off the process of ketogenesis.²

Cerebral oedema

Cerebral oedema is an uncommon but life-threatening complication of DKA and is more likely to occur in children than in adults. It remains the leading cause of death in DKA in children with a mortality of approximately 25%. Severe cerebral oedema occurs in 0.3–0.9% of episodes of DKA in children.¹ Vigilance should be high, particularly in children with risk factors or clinical features of cerebral oedema (Boxes 5 and 6).⁵

Box 5.

Risk factors for cerebral oedema in children with DKA

Risk factors for cerebral oedema
Younger age
Severe DKA
Increased serum urea concentrations
Severe hypocapnia
Attenuated increase in serum sodium
New onset diabetes
Treatment with sodium bicarbonate

Box 6.

Clinical features of cerebral oedema

Clinical features of cerebral oedema
Sudden onset of headache or progressively worsening or severe headache
Unexpected fall in heart rate
Hypertension
Change in neurological state, including restlessness, irritability, drowsiness and confusion
Any neurological deficit, such as cranial nerve palsies
Abnormal respiratory pattern (e.g. Cheyne–Stokes)

Cerebral oedema is a life-threatening emergency, and prompt treatment is needed if it is suspected

• (i)Call for help, and use the ‘ABCDE’ approach.
• (ii)Hypertonic saline 3% (3–5 ml kg⁻¹) or mannitol 0.5–1 g kg⁻¹ should be given immediately.
• (iii)Fluids should be reduced to half the maintenance rate.
• (iv)
  Neuroprotective strategies should be implemented including:

  • (a)
    Elevating the head of the bed to 30 degrees
  • (b)
    Avoiding hypotension to optimise cerebral perfusion pressure
• (v)In severe respiratory or neurological compromise, intubation may be considered.
• (vi)Once the child is stable, a head CT may be indicated to rule out other intracerebral events. This should not delay treatment.²

Tracheal intubation

Intubation of a child with DKA should be avoided, if at all possible, because of the risk of exacerbating cerebral oedema. Hypocapnia has often persisted long enough by presentation of DKA to be well compensated. Consequently, cerebral blood flow may be normal despite hypocapnia. At intubation, a rapid increase in Paco₂ may result in rapid cerebral vasodilatation, potentially exacerbating cerebral oedema and raised intracranial pressure.¹⁶

If intubation is deemed necessary, this decision should be in partnership with a paediatric intensivist or retrieval team because it is high risk and seldom performed. A sudden increase in Paco₂ and consequent acidosis must be avoided. However, a poor outcome has been demonstrated in children with DKA and cerebral oedema, whose tracheas have been intubated and who were subsequently hyperventilated to a Paco₂ <3.0 kPa.¹⁷ Consequently, in the absence of conclusive evidence for the ideal strategy, it is reasonable to avoid rapid changes in Paco₂ and aim for a slow return to normocapnia over a number of hours.

Other complications

Acute kidney injury is encountered commonly in children with DKA. It may be caused by intrinsic tubular dysfunction and poor renal perfusion. Acute kidney injury will often resolve with treatment for DKA.¹⁸

Venous thrombosis in the context of femoral line placement has been shown to be more common in children with DKA who have a central venous catheter (CVC) inserted as opposed to children with a CVC for a non-DKA-related indication.¹⁹ It is rare for a CVC to be required to treat a child with DKA even if presentation is severe. However, if a CVC is deemed necessary, thromboprophylaxis should be considered, and the line should be removed at the earliest possible opportunity.

Aspiration pneumonia is also a possibility in a child who is vomiting and obtunded, and a nasogastric tube should be inserted immediately to reduce the possibility of aspiration of gastric contents.

Where should children with DKA be managed?

All children with DKA require high levels of monitoring, particularly in the initial 48 h of treatment, and the hospital or ward to which they are admitted should reflect this. Nurses should be experienced in treating children with DKA; staffing ratios should be adequate and laboratories equipped to facilitate regular blood tests. An experienced consultant paediatrician should lead care.

The NICE guidelines recommend that any child less than 2 yrs old or with a pH <7.1 should receive one-to-one nursing, ideally in a paediatric high-dependency unit.⁸ Other indications for transfer include circulatory support, depressed level of consciousness and children with risk factors for cerebral oedema. As always, the risk of transferring a severely unwell patient should nevertheless be balanced with the overall benefit to the child. Local PICUs or paediatric retrieval teams are ideally placed to provide support and guidance regarding management of DKA in the child and where they should be treated.

Conclusions

Diabetic ketoacidosis is the most commonly encountered metabolic emergency and is a significant cause of morbidity and mortality in children with diabetes. The management of DKA in any child is not complete without establishing the underlying cause and addressing any contributing psychosocial factors to reduce the risk of further episodes.

Current guidance has been updated to reflect new evidence regarding fluids, which more aggressively targets fluid depletion while not appearing to increase the risk of cerebral oedema. However, questions remain as to the need to refocus the guidelines onto fluid depletion given that cerebral oedema, and not AKI or shock, is the most common cause of death in DKA, in addition to the robustness of the evidence used to underpin the updated guidance. Conversely, the pathophysiology of cerebral oedema is not fully understood, and it is important to recognise that cerebral hypoperfusion as a result of dehydration may play a part. The clinician must account for all these factors when treating children with DKA; vigilance for cerebral oedema and other important complications is paramount.

Declaration of interests

The authors declare that they have no conflicts of interest.

MCQs

The associated MCQs (to support CME/CPD activity) will be accessible at www.bjaed.org/cme/home by subscribers to BJA Education.

Biographies

Michelle Wright BSc (Hons) FRCA is a consultant in paediatric and adult anaesthesia at The Royal London Hospital. She was previously a fellow at the Children's Acute Transport Service at Great Ormond Street Hospital.

Suzanne Body BSc (Hons) FRCA is a specialty trainee in anaesthesia in North East London with paediatric experience from Evelina Children's Hospital and The Royal London Hospital. Her special interests include paediatrics and trauma.

Daniel Lutman FRCA FFICM is a consultant in paediatric anaesthesia and intensive care and paediatric transport retrieval medicine. He is head of the Children's Acute Transport Service at Great Ormond Street Hospital and Barts Health NHS Trust.

Matrix codes: 1A01, 2D01, 3D00