1. Description of the problem

Diabetic ketoacidosis (DKA) is a serious and potentially life-threatening complication of diabetes mellitus, usually type 1 insulin-dependent diabetes. It can occasionally be seen in patients with type 2 diabetes in the presence of stress, trauma or infection. It is characterized by hyperglycemia, elevated blood ketones with metabolic acidosis and dehydration with body water and electrolyte depletion.

DKA may occur from infancy to late adulthood. Important differences exist in management due to the almost exclusive complication of cerebral edema in children and adolescents.

Clinical features

Polydipsia, polyuria and weight loss are the predominant presenting symptoms. Weight loss is not only due to loss of body water via diuresis or vomiting but may be accelerated by poor appetite and starvation. Nocturia is a common feature in childhood DKA. Other nonspecific symptoms include fever, fatigue, nausea, vomiting and abdominal pain. These may not always be related to a primary precipitating event (e.g. urinary tract infection, pancreatitis).

Continue Reading

In adults infection is the most common precipitating event. Coexisting medical and surgical problems are more likely in adult patients. In contrast, acute DKA is the primary presentation of childhood diabetes in about 20% of cases. Noncompliance with insulin therapy is common in patients with recurrent admissions.

Dehydration occurs over days to weeks following the glucose-driven osmotic diuresis. Clinical estimation of degree of dehydration is poor, and there is frequently a poor correlation between markers of dehydration and shock and degree of hyperglycemia / acidosis. There is an increased risk of thrombosis in both children and adults, which is partly explained by hyperosmolality and dehydration.

Tachypnea with Kussmal’s breathing is universal when severe metabolic acidosis is present. The breath also has a characteristic ketone odor. Signs of compensated shock (tachycardia, poor capillary perfusion) are common, but notably hypotension is rare in pediatric DKA. Here, even hypertension may be documented in over 50% of cases on presentation despite significant dehydration.

Adult patients are more likely to present with signs of hypovolemic shock secondary to sepsis as infection is a common precipitating event. In elderly patients attention should be paid to the possibility of myocardial infarction, especially if chest pain is a presenting feature. Body temperature may vary depending on underlying inter-current infection. Hypothermia may occur in adolescent and adult patients with severe DKA and coma if environmental exposure to cold is significant.

Confusion, disorientation and agitation are frequently seen in severe cases; however, deep unresponsive coma should suggest cerebral edema (children) or as a secondary underlying cerebral disorder (e.g. meningitis, cerebral thrombosis). Stroke is not an uncommon finding in childhood DKA if cerebral involvement occurs. Although some authors suggest a causal link between coma and severity of acidosis, a depressed level of consciousness is usually transient and responsive to osmotherapy if associated with cerebral edema.

Key management points

Frequent and meticulous monitoring is required during the first 48 hours of treatment. This is especially relevant in pediatric DKA due to the risk of cerebral edema, which is greatest during the first 12 hours of therapy.

Management strategies are generally based on national or international consensus guidelines due to lack of high-level evidence. Thus, although some issues are controversial, the main key points in management are as follows:

1) Rapid correction of hypovolemia / shock with slow rehydration over 48 hrs

2) Insulin therapy (low-dose insulin infusion) with resolution of blood ketonemia

3) Correction of electrolyte abnormalities

4) Identify and treat the precipitating cause

5) Prevention and early treatment of cerebral edema (children and adolescents)

2. Emergency Management

Rapid reversal of shock and slow rehydration

Hypovolemia shock with hypotension should be treated by rapid restoration of intravascular volume using isotonic crystalloid solutions such as 0.9% saline. In the first two hours 1-2 L of fluid may be required to correct hypovolemia. In children, the initial fluid bolus of 10 ml/kg is usually adequate within the first hour (maximum 20 ml/kg) as hypotension is unusual. There is an increased risk of cerebral edema with overly aggressive fluid resuscitation (>40 ml/kg in first 4 hours). If hypotension persists despite 20 ml/kg, there should be a low threshold for use of inotropes.

Adolescent patients are also at risk for development of cerebral edema but may still require large fluid volumes if septic shock is present. In this scenario small-volume resuscitation with hypertonic saline (5 ml/kg 3% saline) may be prudent to avoid large volumes of fluid replacement.

Hyperlactemia is often present as a marker for sepsis as it is usually not elevated in uncomplicated childhood ketoacidosis. Although many different fluid protocols are available for adult patients, a 10% loss of body water (7 L in a 70-kg patient) can be replaced at a rate of about 14 ml/kg/hr in the 1st hour, 7 ml/kg/hr over the next 4 hours, and 4 ml/kg/hr in the next 6 hours.

In contrast, fluid replacement rates in children should generally be lower, between 3 and 5 ml/kg/hour after the initial 10- to 20-ml/kg fluid bolus and not include replacement of urine output. In severe cases where the risk of cerebral edema is high, fluid may need to be restricted to 50-60% maintenance with maximum 5% rehydration per day (50 ml/kg/day).

Fluid replacement rates should be modified according to the severity of volume deficit. Caution should be used in the elderly, pregnant women and patients with cardiac or renal disease, where fluid loads may not be tolerated. Renal impairment with oliguria may result in persisting acidosis as about 40% of the acid load generated from ketone production is excreted via the kidneys. Renal replacement therapy may occasionally be required if anuria occurs or in frankly hyperosmolar states to minimize rapid changes in plasma osmolality.

The most widely used resuscitation fluid in DKA is 0.9% saline. There is, however, no strong evidence supporting a preference for crystalloid solution. Although hyperchloremia is aggravated by use of large volumes of 0.9% saline, there is no information to show this is detrimental, specifically in DKA. Ketone clearance is maximized under acidemic conditions and may be increased when alkalinization is attempted. More physiologically pH-balanced solutions, like Plasmalyte, reduce the incidence of hyperchloremia in DKA, but evidence that this is beneficial is lacking. Hartman’s and Ringer’s solutions, although low in chloride, contain large quantities of lactate, for which its impact on ketone body production and clearance remains unknown.

There is no clear consensus if a change in fluid type to a more hypotonic rehydration fluid (e.g., 0.45% saline) is neither required nor necessary, especially in children, where hypotonic fluids may promote cerebral edema. In patients with a high risk of cerebral edema hypotonic fluids such as 0.45% saline should be avoided. However, there are observational case series demonstrating the safe use of hypotonic fluids, provided careful and regular monitoring of electrolytes is performed. Monitoring change in corrected sodium is useful in pediatrics as a guide to adequacy of fluid replacement.

Insulin therapy and resolution of ketoacidosis

The main aim of insulin in DKA is not to primarily reduce blood glucose but to inhibit lipolysis and prevent the ongoing formation of more ketoacids. This can be safely achieved using a low-dose insulin infusion in both adults and children. Higher doses of insulin are associated with an increased risk of cerebral edema in pediatric ketoacidosis. Insulin therapy should be commenced only after the initial fluid resuscitation (after 1st hour) as glucose levels fall significantly during volume expansion.

Current consensus is that low-dose intravenous infusion at a rate of 0.05 to 0.1 units/kg/hr is adequate for the resolution of ketoacidosis. Insulin infusion rates should be titrated to blood ketones (beta-hydroxybutyrate) as the primary goal.

In adults mean beta-hydroxybutyrate levels are about 7 mmol/L (range 4-12) and should fall at a rate of 1 mmol/hr. Thus, resolution of ketoacidosis should occur within 8 to 12 hours of insulin therapy. Persisting metabolic acidosis with resolution of ketonemia should prompt a search for hyperchloremia or other causes of a raised anion gap. The confounding effect of chloride can easily be determined by partitioning the base excess, with the chloride component equal to Na – Cl – 32 mEq/L.

In children beta-hydroxybutyrate levels are usually lower and rarely exceed 10 mmol/L. Occasionally higher insulin infusion rates are required (0.2 to 0.3 u/kg/hr) if insulin resistance is present, typically in patients with sepsis. The rate of insulin can be reduced once beta-hydroxybutyrate levels are below 1 mmol/L.

Changes in glucose can be controlled by modifying intravenous glucose intake rather than changing insulin infusion rate. Glucose is usually added to replacement fluid when blood glucose falls below 12 to 14 mmol/L. The rate of glucose fall should be about 3 mmol/hr to minimize rapid osmolality shifts.

Correction of electrolyte abnormalities

Regular monitoring of electrolytes is required 2-4 hrly in the early phase of treatment, the most important of which is potassium. Intravenous potassium should be added to rehydration fluid (20-40 mmol/L) if serum K is below 5.5 mmol/L and modified according to the rate of K change following insulin therapy. Hyperkalemia is rare in DKA unless renal impairment is severe or other causes are present, such as rhabdomyolysis.

Routine phosphate replacement is not usually warranted unless severe (< 0.4 mmol/L) and associated with symptoms. Hypocalcemia is a frequent complication of intravenous phosphate treatment. If indicated, 20 mmol/L of phosphate can be added to replacement fluid for a slow correction of hypophosphatemia.

Identify and treat precipitating cause

Infection is an important precipitating cause, with urinary tract infection and pneumonia being most common. A thorough search for infective site should be performed and appropriate antibiotics used. In adults comorbid disease (myocardial infarction, alcoholism, liver disease, etc.) play an important role as potential triggers of DKA. A precipitating cause is not identified in about a third of patients.

Prevention and treatment of cerebral edema

Cerebral edema is a potentially serious and devastating complication of pediatric and adolescent DKA. Using strict criteria, cerebral edema is rare, affecting <1% of pediatric cases. Unexplained alteration in level of consciousness is not infrequent in severe DKA (pH < 7.1). The risk of cerebral edema in childhood is approximately 4/1000 episodes of DKA in previously diagnosed diabetics and 12/1000 in newly diagnosed patients.

Patients may present with deep coma (20%) or more typically as a progression in alteration of level of consciousness develops over the first 4 to 12 hours following therapy. There is no strong evidence to suggest depth of coma is directly related to low pH, as many patients may have a pH < 7.1 without coma.

A fall in the level of consciousness should suggest the possibility of cerebral edema. This may be subtle and include agitation, confusion and delirium. An early warning of cerebral edema is a progressive rise in pCO2 if pH fails to increase. When pH is <7.1, a pCO2 above 2.5 to 3 kPa suggests compromise of central respiratory drive.

There is no universally accepted hypothesis that explains the occurrence of cerebral edema in childhood DKA. There are underlying factors related to the primary disease (raised plasma and brain ketones, prolonged hyperosmolar state, hypocapnia with cerebral vasoconstriction, abnormal brain metabolites) and factors associated with therapy (insulin use, fluid rehydration, rapid correction of tonicity or acidosis) that play important roles in its pathogenesis (Table I).

Table I.
Younger age (< 1yr)
pCO2 < 2 kPa or15 mmHg
Admission pH <7.1
Large-volume fluid resuscitation (>40 ml/kg fluid in 1st 4 hours)
Rapid fall in corrected sodium
Fall in effective plasma tonicity
Early use of insulin (1st hour of therapy) or high-dose insulin
Bicarbonate therapy
Raised plasma urea

More importantly, the delay in treatment increases the likelihood of poor outcome; thus, early recognition is crucial.

Treatment of cerebral edema

Prompt treatment with osmotherapy (3- to 5-ml/kg bolus of 3% hypertonic saline over 10 to 20 minutes) should improve neurological status rapidly if cerebral edema is present.

Lack of response to osmotherapy should suggest an alternate diagnosis (e.g. meningitis, cerebral stroke) for which neuro-imaging may be required (CT scan or MRI). Neuro-imaging is not routinely necessary if cerebral edema is suspected as CT scans are frequently reported as normal despite marked coma. There are two possible reasons for this: 1) lack of established norms to quantify how rapidly brain volume should increase following rehydration therapy and 2) the mechanism of coma in DKA may not be exclusively related to direct brain swelling. Frank cerebral edema is a late and usual terminal finding.

Osmotherapy can be repeated if necessary with further boluses of hypertonic saline. This can be titrated by monitoring for a fall in effective osmolality or more simply by following the temporal profile in glucose-corrected sodium. Glucose-corrected sodium should increase by about 5 mmol/L in the first 8 hours. In refractory cases an infusion of hypertonic saline may be useful even if absolute hypernatremia is present.

Mannitol as the primary osmotic agent is not generally warranted as it promotes further obligatory diuresis and its effect on plasma osmolality is difficult to quantify, especially if repeated boluses are needed. Additionally, hypertonic saline has the benefit of volume expansion and preservation of intravascular volume.

Intubation and mechanical ventilation are usually only required in coma that is refractory to osmotherapy. Although controversy exists, it appears sensible to avoid severe hypocarbia after intubation since brain MRI findings suggest ischemia secondary to cerebral vasoconstriction as a potential mechanism of cerebral edema. If a response to osmotherapy is seen and plasma osmolality is increased and maintained, outcome is favorable. Prior to anesthetic induction and intubation, hypovolemia should be corrected as anesthetic agents may precipitate hypotension in the face of dehydration.

3. Diagnosis

The triad of raised blood glucose, raised blood ketones with acidosis and dehydration characterizes DKA.

Elevated blood glucose

Blood glucose is typically elevated in the range between 11 and 20 mmol/L. Glucose levels may vary considerably and are not related to the degree of dehydration or disease severity. Euglycemic ketoacidosis can occur if starvation and vomiting is a presenting feature.

Elevated blood ketones

Historically blood ketones were not easily measured in blood and thus surrogate markers of ketoacidosis were used (pH, bicarbonate base excess and anion gap). This was problematic because bicarbonate did not improve at the same rate as the anion gap, prompting many speculative theories concerning ketone distribution and clearance. Stewart’s physiochemical acid base theory has highlighted reasons for these inconsistencies.

The availability of commercial bedside blood ketone tests (beta-hydroxybutyrate) has allowed more accurate monitoring of this condition. Beta-hydroxybutyrate and acetoacetate are usually produced in a equimolar rate. In DKA, redox potential favors the predominant production of beta-hydroxybutyrate such that a ratio of up to 5:1 can occur. For this reason raised blood beta-hydroxybutyrate levels are used diagnostically and to monitor the resolution of ketoacidosis.

As a rough guide, serum bicarbonate of 18 mmol/L coincides with a blood beta-hydroxybutyrate concentration of 3.0 mmol/ in children and 3.8 mmol/L in adults with DKA. Mildly elevated blood beta-hydroxybutyrate levels >1 mmol/L suggest early ketoacidosis. Bedside BOH ketone sticks are sufficiently accurate in between 0.5 and 6 mmol/L.

Glycosuria and ketonuria are universally present; however, urinary keto-sticks that measure predominantly acetoacetate may be only mildly positive as they do not accurately reflect the beta-hydroxybutyrate component.

Diagnostic tests

Blood biochemistry

Venous measurements of glucose and blood gases and ketones are adequate. Plasma sodium can be low, normal or high on presentation despite the depletion of total body sodium. Plasma sodium is also is influenced by degree of hyperglycemia. Correction of sodium for glucose is required, especially in the first few hours of therapy when the rate of fall of glucose is greatest.

Plasma potassium is usually in the normal range but values > 5.5 mmol/L suggest a secondary cause such as renal impairment or increased potassium production (e.g. rhabdomyolysis). Phosphate is frequently in the lower range of normal but frank hypophosphatemia can occur. Urea is frequently elevated, especially relative to creatinine as a marker of dehydration.

Nonspecific elevation of liver function tests, amylase and lipase may be seen despite a lack of clinical evidence of abdominal pathology.

Full blood count

Hemoglobin and hematocrit may be marginally elevated. It has been suggested this is due to dehydration, but acidosis may also directly increase hematocrit. Anemia may occur but is usually related to other underlying pathology. Similarly, the white cell count may also be elevated even without a primary focus of infection. Thrombocytopenia secondary to rhabdomyolysis or venous thrombosis is well documented across all age groups.

Other tests

If infection is suspected a source should be identified. Investigations include urinalysis, chest x-ray, and blood culture. Other investigations are indicated if precipitating factors are identified, for example, and ECG and cardiac enzymes if myocardial infarction is suspected.

Differential diagnosis

The differential diagnosis includes conditions where metabolic acidosis and hyperglycemia occur.


Hyperglycemia frequently occurs in sepsis due to an increase in catabolic hormones such as cortical and catecholamines. A degree of insulin resistance may also occur, resulting in marginal elevation of blood ketones. In sepsis, it is unusual for ketones to be the predominant acidifying agent and other causes are often identified (raised lactate). Sepsis can, however, coexist with DKA. This should be suspected if blood ketone levels are high (>5 mmol/L) and glucose fails to normalize if intravenous glucose intake is limited

Starvation and alcoholic ketosis

Ketoacidosis may occur under stress conditions where counter-regulatory hormones are raised and a lack of insulin sensitivity exists. This results in lipolysis with an increased delivery of fatty acids to the liver for conversion to ketone bodies. Starvation and vomiting associated with alcohol abuse may lead to significant dehydration with ketoacids; however, blood glucose is not usually elevated.

Levels of beta-hydroxybutyrate may be particularly high in alcoholic ketosis as redox potential favors hepatic production of this ketone. Metabolic acidosis with mild elevation of beta-hydroxybutyrate (3-5 mmol/L) can also be seen in severe asthma where high-dose salbutamol is used. This is mediated by excessive beta-adrenergic receptor stimulation.

Inborn errors of metabolism

Congenital inborn errors in the ketone pathway defects can cause severe ketoacidosis, usually out of proportion to the precipitant stress or starvation episode. Presentation is usually in infancy as an autosomal recessive disorder. Pathogenesis is related to lack of the enzyme succinyl-CoA: 3-ketoacid CoA transferase and mitochondrial Acetoacetyl-CoA thiolase, which results in ketone production coupled with an inability to break ketones down. Treatment is rehydration with glucose provision to inhibit ketone production. Insulin may also be required to inhibit lipolysis. The clinical presentation may mimic DKA with severe acidosis and low pCO2 with Kussmaul’s breathing; however, glucose is not usually elevated or may be transient.

Confirmatory tests

In the modern era, point-of-care blood glucose and blood ketone testing should be readily available for diagnostic and monitoring purposes. Use of capillary blood glucose has shown promise in the early detection and treatment of DKA, with levels > 1 mmol/L suggesting the need for further investigation/treatment.


Under normal conditions there is a tight regulation of glycemia with the rapid distribution of glucose into liver (glycogen), kidneys, brain and other glucose-sensitive tissues. Under the anabolic action of insulin, glucose is also converted and stored in fat cells as even low levels inhibit the hormone lipase, preventing lipolysis.

In DKA, there is a loss (either absolute or relative) of insulin; thus, the actions of counter-regulatory hormones, including cortisol, glucagon and catecholamines, are unopposed. These promote glycolysis and gluconeogenesis, thus increasing blood glucose as it is mobilized from liver, muscle and peripheral tissues. Hyperglycemia leads to an osmotic diuresis and dehydration as the kidneys have a low threshold for reabsorbing glucose. A cycle of polydipsia and polyuria with progressive depletion of body water and electrolytes ensues.

Ketoacidosis occurs because the lipolytic enzyme lipase is no longer inhibited by the insulin. Fat cells are converted into circulating free fatty acids, which undergo metabolism in the liver into ketone bodies. These are beta-hydroxybutyrate and acetoacetate. Acetone is also produced via the decarboxylation of acetoacetate and is excreted via the lungs and kidneys; this is responsible for the characteristic ketotic odor of the breath in DKA. Acetone is non-ionized and does not contribute to the anion gap.


The epidemiology of DKA is complex due to wide variations in a number of factors, including socioeconomic status, ethnicity and genetics. In the past decade, rates between 4 and 8 episodes per 1000 patients have been reported, with the the younger and poorer at greatest risk. Socioeconomic factors have been implicated in patients with recurrent admissions, for which lack of compliance with insulin therapy is the predominant causal factor.

In the developed world, the majority of cases are managed in a non-critical care setting. In children, a higher proportion of cases are managed in pediatric critical care units but account for only <0.5 % of critical care admissions. 40% of diabetes-related admissions to pediatric intensive care in the United Kingdom are in children aged between 11 and 15 years.


The mortality for DKA across all age groups is about 1%, being highest at either extremes of age. In adults mortality is more closely related to underlying diseases such as infections, sepsis and comorbid conditions rather than as a direct complication of hyperglycemia or ketoacidosis.

In contrast, outcome in childhood DKA is strongly associated with the development of cerebral edema. This accounts for about 80% of all diabetes-related deaths in this age group. The survival rate for diabetes complicated by cerebral edema is between 70% and 80% Severe neurological disability in survivors is well documented, especially if treatment is delayed and coma score is low.

What’s the evidence?