CME: Managing diabetic ketoacidosis--a delicate balance

June 1, 2000

Diabetic ketoacidosis is one of the most common, and dangerous, complications of both insulin- and noninsulin-dependent diabetes. The patient's life depends on a complex therapeutic juggling act to restore metabolic, acid-base, fluid, and electrolyte balances.


Managing diabetic ketoacidosis— a delicate balance

Jump to:Choose article section...CME LEARNING OBJECTIVES What is DKA? How DKA happens The first step in management: Is it DKA? Differential diagnosis of DKA Clinical features The physical exam Laboratory tests The controversy over fluids DKA flow sheet Replacing electrolytes Insulin therapy

By Waseem Hafeez, MB, BS, and Patricia Vuguin, MD

Diabetic ketoacidosis is one of the most common, and dangerous, complications of both insulin- and noninsulin-dependent diabetes. The patient's life depends on a complex therapeutic juggling act to restore metabolic, acid-base, fluid, and electrolyte balances.

A 4-year-old girl with a five-day history of vomiting and abdominal pain is brought to your office by her parents. The severity of her symptoms prompted her parents to seek care on each of the last three days at another office, where she was treated unsuccessfully for acute gastroenteritis. On further questioning, the family mentions that the child has lost considerable weight over the last month and has been extremely thirsty. She recently developed enuresis at night. The youngster looks ill and dehydrated and is in need of immediate hospitalization.

This scenario resembles the common presentation of new-onset diabetes and should suggest a diagnosis of diabetic ketoacidosis (DKA) to any pediatrician who evaluates a child with similar complaints. Insulin-dependent diabetes mellitus (type 1, IDDM) is one of the most frequent chronic diseases in children in the United States, and current evidence suggests that both IDDM and noninsulin-dependent diabetes mellitus (type 2, NIDDM) are increasing globally. About 13,000 children in the US develop diabetes mellitus each year. The estimated prevalence is 1.2 per 1,000, and the annual incidence is 18.2 per 100,000. Type 2 diabetes accounts for 16% of all new cases of diabetes and 33% of the cases among children and adolescents between 10 and 19 years of age. It occurs most often in African-American children who are obese and have a strong family history of type 2 diabetes.1

Diabetic ketoacidosis, one of the most common acute metabolic complications of diabetes mellitus, is a life-threatening catabolic state that occurs in the context of insulin deficiency.2 Initially thought to occur only in type 1 diabetes, it is now recognized as a complication of type 2 diabetes as well.3

DKA accounts for 14% of hospital admissions for diabetes and is the most common cause of hospital admission in diabetics under 20 years of age.4 Around 20% to 30% of episodes of DKA occur in newly diagnosed diabetics, and the mortality rate is 7% to 8% in major medical centers. 5

Successful management of DKA requires a thorough understanding of the factors that precipitate it and the pathophysiology of the metabolic, acid-base, fluid, and electrolyte imbalances that develop. It also requires considerable skill and a meticulous approach to evaluation and treatment. Our purpose is to offer clear guidelines to help pediatricians identify this potentially devastating condition and understand the basis of treatment and the controversies surrounding it.



After reviewing this article the physician should be able to:


What is DKA?

Diabetic ketoacidosis is defined as hyperglycemia (serum glucose greater than 300 mg/dL) and metabolic acidosis (pH less than 7.30 and HCO3 less than 15 mEq/L) accompanied by ketonemia, ketonuria (acetone, ß-hydroxybutyrate, and acetoacetate), and glucosuria. In addition, the patient may have water and electrolyte imbalances with varying degrees of dehydration. The clinical presentation varies and may mimic gastroenteritis, acute appendicitis, hyperventilation, or intoxication. In one study, 20% of patients with DKA were misdiagnosed at initial presentation and hospitalized for a primary diagnosis other than type 1 diabetes.5

In patients with established type 1 diabetes the most common cause of DKA is failure to take insulin. Other common precipitating factors include infection, lack of education about diabetes, stress, psychological problems, and cocaine use.1,6 Emotional stress is responsible for DKA in 20% to 30% of patients with recurrent ketoacidosis.

How DKA happens

Figure 1 outlines the pathogenesis of DKA. Insulin deficiency accompanied by a compensatory increase in the counter-regulatory hormones (epinephrine, norepinephrine, cortisol, growth hormone and glucagon) results in failure to deliver an adequate supply of glucose into the cells.2,7 This catabolic state affects three major tissues: muscle, adipose tissue, and the liver.



In the muscle cells, inability to receive fuel substrates causes the cells to act as they do in starvation. Proteolysis accelerates, resulting in the release of large amounts of amino acids that are converted into glucose by the liver.

In the adipose tissue, severe insulin deficiency activates hormone-sensitive lipoprotein lipase, releasing large amounts of fatty acids into the plasma.8 The fatty acids are delivered to the liver where a combination of glucagon excess and insulin deficiency transforms them into ketone bodies. Accumulation of these ketoacids (acetoacetic acid, acetone, ß-hydroxybutyrate acid) leads to metabolic acidosis. In an attempt to excrete the resulting excess CO2, lung ventilation increases, causing characteristic Kussmaul breathing.

In the liver excess glucose production through glycogenolysis and gluconeogenesis leads to hyperglycemia. As the serum glucose level rises above the renal threshold (about 180mg/dL), glucosuria develops. Hyperglycemia induces osmotic diuresis, with increased urinary losses of water and electrolytes, resulting in dehydration, acidosis, and hyperosmolality.9

As a defense against cell shrinkage, hyperosmolality causes the cells, especially the brain cells, to accumulate osmoprotective molecules (myoinositol, taurine, and glutamate). The blood-brain barrier is permeable to water by simple diffusion but prevents the passage of solutes into the brain interstitium. Glucose-induced hypertonicity may compromise the tight junctions of the cerebral endothelium so that cells with added osmols rapidly take in water, increasing their volume.10 As dehydration, acidosis, and hyperosmolality progress and glucose and oxygen diminish, consciousness becomes impaired and the patient ultimately becomes comatose. Continued brain swelling increases intracranial pressure (ICP) and causes cerebral vasoconstriction, which leads to hypoxia, thrombosis, infarction, or, ultimately, brain herniation.

The first step in management: Is it DKA?

Management of patients with DKA encompasses five steps:

  • Diagnosis, including clinical features, physical exam, and laboratory investigations

  • Fluid resuscitation and electrolyte management

  • Insulin therapy

  • Continuous monitoring and record keeping

  • Recognition and management of cerebral edema.

In order to meet the diagnostic criteria for DKA, patients must have all three components:

  • Diabetes: serum glucose greater than 300 mg/dL

  • Ketosis: serum ketosis and ketonuria

  • Acidosis: serum pH less than 7.30 and HCO3 less than 15 mEq/L.

Patients with poor oral intake may have blood sugar levels less than 300 mg/dL yet have ketosis and acidosis while other diabetic patients may present in hyperglycemia without significant acidosis.

In a known diabetic the diagnosis of DKA is presumed and easily confirmed, though in new-onset type 1 diabetes it may be difficult. The differential diagnosis is varied (Table 1), but a few bedside tests can confirm DKA. Figure 2 outlines the clinical pathway for diagnosing DKA.



Differential diagnosis of DKA

Polyuria, polydipsiaDiabetes insipidus, psychogenic water drinking, UTI



Clinical features

The presentation of DKA varies greatly with age, precipitating factors, degree of acidosis and hyperglycemia, and the presence of cerebral edema. The physician should always consider DKA as a differential diagnosis in any ill-appearing child. The patient or parent may not volunteer the classic history of polyphagia, polyuria, polydipsia, and recent weight loss unless asked specifically about these complaints, which usually precede the acute illness by a few days, weeks, or months. Younger patients have a faster onset of symptoms. Nausea, vomiting, and abdominal pain are common, and varying degrees of dehydration are usually present. The patient may be quite alert given the severe levels of dehydration and acidosis. As many as 10% of patients present in coma.11

Nausea and vomiting are often the first symptoms of DKA, and as the ketoacidosis worsens, vomiting may become severe. Lethargy, irritability, personality changes, and headache in association with nausea and vomiting are clues to increased ICP.

Abdominal pain occurs in 50% of younger children and usually results from gastric distention (gastroparesis) or stretching of the liver capsule rather than from a cause requiring surgery, such as appendicitis.12 Often the pain subsides within four to six hours of starting therapy.

In previously diagnosed diabetics, obtain a history of precipitating factors such as stress, infection, dietary indiscretion, reduced fluid intake, and compliance with insulin therapy (be sure to ask about the current dose schedule and the usual level of glycemic control.) In younger children with vomiting and reduced oral intake parents sometimes withhold insulin, and this may trigger DKA.

The physical exam

A complete physical examination is essential to determine whether there is a focus of infection and to assess the degree of dehydration and cerebral edema. Adhering to the basic ABCs in assessing and resuscitating any ill-appearing child is critical. Assess the patency of the Airway and determine the efficacy of Breathing by evaluating skin color, respiratory rate, air entry, use of accessory muscles, and pulse oximetry readings. Evaluate Circulation by means of heart rate, capillary refill, distal pulses, and blood pressure.

Look for signs of dehydration, such as tachycardia, hypotension, tachypnea, delayed capillary refill, dry mucous membranes, sunken eyes, depressed fontanels, decreased skin turgor, drowsiness, or lethargy. Urinary output is not a reliable indicator because of osmotic diuresis.

Cerebral edema and increased ICP are ominous indications of impending brain herniation. Look for clinical signs such as deteriorating level of consciousness, papilledema, and Cushing's triad (hypertension, bradycardia, and Cheyne-Stokes respiration). As herniation progresses the patient develops unequal pupils with posturing, initially decorticate (flexion) and finally decerebrate (extension).

Laboratory tests

Initial laboratory investigations include blood chemistry, urinalysis, hematology, an electrocardiogram, and tests for new-onset type 1 diabetes.

Blood chemistry tests. The following blood chemistry tests should be ordered:

  • Sodium. Either hyponatremia or hypernatremia can be present initially depending on the patient's state of hydration, water balance, and ADH secretion. There is total body sodium depletion, and the measured serum sodium may be spuriously low because of increased serum glucose, lipids, and proteins. To correct for pseudohyponatremia add 1.6 mEq/L to the serum sodium measurement for every 100 mg/dL serum glucose above 100 mg/dL.

  • Potassium. Serum potassium (K+) is usually normal or elevated despite depletion of total body potassium caused by insulin deficiency, dehydration, and acidosis. Insulin therapy increases the transport of potassium from extracellular fluid to intracellular fluid, reducing serum potassium levels, and each increase of 0.1 in the serum pH decreases serum K+ by 0.6 mEq/L.

  • Bicarbonate and pH. In severe acidosis, pH may drop below 7.00 and HCO3 below 5 mEq/L.

  • Glucose. A bedside glucose measurement can give a quick estimate of hyperglycemia, but most glucometers cannot detect levels higher than 600 mg/dL. Standard chemistry tubes can degrade glucose, so special plasma separator tubes should be used for accuracy if a delay in the laboratory turnaround time is expected.

  • Effective osmolality(Eosm) and anion gap. Effective osmolality can either be measured in the laboratory or calculated using the formula:


EGlucose mg/dL

+BUN mg/dL


The normal range is 285 to 295 mOsm/L. An osmolal gap, indicating the presence of osmotically active particles (such as glucose), is present when the difference between the measured and calculated osmolality is greater than 10.

An anion gap, also indicative of osmotically active particles, can be calculated from the unmeasured anions and cations using the formula:

(Na + K) - (Cl + HCO3) = Normal 12 mEq/L ± 2

  • Calcium/magnesium/phosphate. The serum phosphate level is usually decreased, and phosphate should be replaced early by administering potassium phosphate. Since phosphate can lower serum concentrations of calcium and magnesium, their levels should also be monitored.

  • Glycosylated Hgb. Hgb A1C reflects the average glycemic control over the past three months. It increases at a steady rate as glucose levels increase. Normal values vary by lab but usually are 4% to 6 %. In poorly controlled diabetics, values are above 9%.

  • Amylase/lipase. Amylase is often elevated in DKA, but the elevation is usually of no clinical consequence as it does not reflect pancreatitis or intra-abdominal pathology.13 When fractionated, the amylase is found to be predominantly salivary amylase. If pancreatitis is suspected, serum lipase levels should be measured.

Urinalysis shows glucosuria and ketonuria. The presence of leukocytes, red blood cells, leukesterase, nitrites, and bacteria suggests a urinary tract infection. Take a sample of catheterized urine in infants and mid-stream clean catch urine in older children if UTI is suspected. The dipstick method of detecting urinary ketones uses the nitroprusside reaction that measures acetoacetate (A) but not ß-hydroxybutyrate (B) levels. In DKA the A:B ratio may increase to 1:30, and the test therefore may be falsely negative despite the presence of ketonemia.5 In such cases, treating DKA increases ketonuria.

Hematology. The white blood cell count is usually 15,000 to 20,000/µL, which may result from the effects of counterregulatory hormones, not infection. A WBC differential with a shift to immature forms may indicate infection.14

Electrocardiogram. This quick bedside test can detect hypo- or hyperkalemia. Obtain a baseline 12-lead ECG since monitor tracings are unreliable.

Tests for new-onset type 1 diabetes. Before starting insulin therapy, measure insulin level, glutamic acid decarboxylase (GAD) level, islet-cell antibodies, and anti-insulin autoantibodies. These tests can differentiate between new-onset type 1 diabetes and type 2 diabetes. The islet-cell antibodies—islet-cell cytoplasmic antibody (ICA) and islet-cell surface antibody(ICSA)—insulin antibody, and GAD are present in up to 80% of patients at the time of initial diagnosis of type 1 diabetes and are responsible for ß-cell destruction. They may become undetectable after one or two years.15 Thyroid function tests—TSH and free T4—also should be done since type 1 diabetes is associated with autoimmune thyroiditis in 20% to 30% of Caucasian children.

Other tests. When clinically indicated, cultures of blood, urine, and throat and a chest X-ray should be done depending on the suspected source of infection. Patients in DKA should be continuously monitored, including serial ECGs to monitor potassium levels as reflected in heart function.

The controversy over fluids

The greatest controversy in the current management of DKA concerns how best to administer fluid resuscitation and prevent increased ICP. The optimal approach to initial management is to correct hypovolemia and shock, correct acidosis, reduce blood sugar, replace electrolytes, and prevent rapid changes in intracerebral osmolality that may precipitate or worsen cerebral edema.

Figures 3 and 4 show an overview and a more detailed clinical guideline for managing DKA. Every emergency department should develop such practice guidelines for quick reference. The physician must assess the patient's condition repeatedly and use his or her clinical judgment to correct any deterioration rapidly. Ongoing monitoring and record keeping are essential. Table 2 shows a sample monitoring flow sheet.




DKA flow sheet

Vital signs30 min, until stableGlucoseHourly during insulin drip
Neurology checks30 min, until stablePotassiumHourly until stable, then q4h
  Blood gas-pHHourly until pH 7.2, then q2h until pH 7.3
  Sodium, HCOEvery 2 hr until stable, then q4h
  Ketonuria/glucosuriaEvery urination until resolved
  Calcium, phosphorus, magnesiumEvery 12 hr
  ECGEvery hr if K
 Fluids—IV type @ rate/POVolDoseVolKetmg/dLmEq/LmEq/LmEq/L 
0-1 hr          
2-4 hr          

5-8 hr

9-16 hr          



After quickly assessing the patient, determining the level of dehydration, and prioritizing treatment, place the patient on a cardiac monitor, pulse oximeter, and blood pressure monitor. Insert an intravenous line and draw blood for the appropriate tests as described above.

During the first hour, give isotonic crystalloid solutions: 0.9 % NaCl (normal saline [NS]) or Ringers Lactate (RL). If the patient is in shock, rapidly infuse 10 to 20 mL/kg NS, reassessing the patient's status often until tissue perfusion is restored. If more than 10 to 20 mL/kg is required, it usually indicates third spacing of fluids, septic shock, or pancreatitis.2 Most patients are about 10% dehydrated. An NS bolus of 10 mL/kg given over one hour usually corrects the hypovolemia.

Calculate subsequent fluid administration to replace the total deficit gradually over the next 36 to 48 hours. Current guidelines prescribe gradual fluid replacement over 48 hours for patients with signs of increasing ICP, initial blood glucose greater than 1,000 mg/dL, serum osmolality greater than 320, and hypernatremia greater than 150 mEq/L.16

The fluid needed to replace the deficit is calculated as 10 mL/kg for each 1% dehydration and may be given as 0.45% NaCl (1/2 NS). Deduct the fluid given during the initial phase of resuscitation from the total amount of replacement fluid. Half the deficit replacement may be given over the next eight hours and the remaining fluid over the subsequent 16 to 36 hours. Urinary losses subside as the level of glucosuria declines and usually do not need to be replaced.

Daily maintenance fluids should be calculated and given evenly over the next 24 hours. The total fluid should not exceed 4.0 L/m2/24 hr.17 In children under 6 months of age, give 0.2% to 0.33% NaCl after the initial resuscitation phase. Subtract fluids used in any other drips—that is, insulin or bicarbonate drips—from the total fluids.

The cautious approach to fluid management after the initial resuscitation phase, which is the current standard, stems from studies indicating that patients who received more than 4.0 L/m2 of fluid in 24 hours were at greater risk of developing cerebral edema and increased ICP.17 More recent literature has not confirmed these findings.18

Replacing electrolytes

Sodium is the major extracellular cation and potassium is the major intracellular cation. Both are essential for maintaining intravascular volume, serum osmolality, and other cell functions. The approximate accumulated losses of electrolytes in DKA are as follows:9

  • Sodium—6 mEq/kg (range 5 to 13 mEq/kg)

  • Potassium—5 mEq/kg (range 4 to 6 mEq/kg)

  • Chloride—4 mEq/kg (range 3 to 9 mEq/kg)

  • Phosphate—3 mEq/kg (range 2 to 5 mEq/kg)

Since there is a close relation between serum glucose and serum sodium, the sodium concentration should rise by 1.6 mEq/L for every 100 mg/dL drop in serum glucose.19 If the measured sodium fails to increase at the expected rate or falls, it raises concern about overhydration and excessive free water, necessitating close monitoring of the patient for increased ICP.20 Fewer than 10% of patients have a corrected serum sodium (Na+c) greater than 150 mEq/L. These few need extremely gradual rehydration over 48 hours with more prolonged use of isotonic fluids.

The presence of severe anion gap metabolic acidosis, hyperglycemia, and impaired renal tubular reabsorption all contribute to elevated serum potassium at presentation, which is common in patients with DKA. An initial finding of hypokalemia (usually caused by vomiting, diarrhea, and increased urinary losses) is rare but extremely worrisome because rehydration and correction of hemoconcentration will likely cause a further drop in serum K+. Hypokalemia is also made worse by insulin therapy, which corrects the acidosis and causes a rapid intracellular influx of potassium. The ensuing severe hypokalemia in an already hypokalemic patient may cause life- threatening arrhythmias or respiratory arrest. Potassium replacement must be started at the same time as insulin therapy in hypokalemic patients rather than afterwards, as in hyperkalemia.

The majority of patients in DKA will require rehydration with 1/2 NS with potassium acetate, or phosphate as their deficit fluid replacement solution. After the initial fluid bolus, if the patient is urinating and serum K+ is less than 6.0 mEq/L, add 20 to 60 mEq/L of potassium to the replacement fluid.21 The goal is to maintain a serum K+ level of 4.0 to 5.0 mEq/L.

Replacement potassium may be given half as potassium acetate (KAc) and half as potassium phosphate (KPO4). Acetate is converted to bicarbonate in the liver and can help correct metabolic acidosis. Using potassium chloride in fluid replacement is not recommended because it causes hyperchloremic metabolic acidosis, which may be associated with a lower rate of recovery from DKA, although studies have not confirmed this.22

Since phosphate is also depleted significantly in patients with DKA and can substitute for chloride, it is used routinely (in the form of KPO4) to replenish stores of 2,3-diphosphoglycerate (DPG). Recently hypophosphatemia has been found to lead to insulin resistance, and the use of phosphate has been associated with symptomatic hypocalcemia and hypomagnesemia.9 Despite its theoretical advantages—preventing insulin resistance caused by hypophosphatemia and avoiding hyperchloremic acidosis by substituting for chloride—routine use of phosphate has shown no clinical benefit in treating DKA.23

Magnesium deficiency also occurs during DKA but is not in itself clinically significant. Symptomatic hypocalcemia resulting from reduction of ionized calcium stores by phosphate therapy cannot be corrected without replacing magnesium, however.22

The use of alkali therapy in DKA is controversial. Severe acidosis (pH less than 7.0) impairs myocardial contractility, produces peripheral vasodilatation and subsequent hypotension, and eventually may result in shock. Several studies have shown that giving bicarbonate to patients with serum pH between 6.85 and 7.18 failed to change, or even delayed, the rate of metabolic correction. Opponents of bicarbonate administration list the following reasons for their reluctance to use it:12

  • In patients with hypophosphatemia and low levels of 2,3-DPG, a rise in pH decreases tissue oxygenation by causing lactic acidosis.

  • Bicarbonate makes cerebral acidosis worse because it crosses the blood-brain barrier much more slowly than carbon dioxide.

  • Alkalosis may exacerbate hypokalemia by accelerating the intracellular transport of potassium, causing dysrhythmias. It also increases ketogenesis, thereby delaying recovery from DKA.

In light of the controversial aspects of bicarbonate administration, we prefer to use bicarbonate sparingly. We infuse sodium bicarbonate (NaHCO3) at 1 mEq/kg over two hours only if the pH is below 7.0 and discontinue it once pH approaches 7.10.

Insulin therapy

Short-acting (regular) insulin is the only intravenous form available currently and should be given by infusion pump. Start the infusion at the end of the first hour after rechecking the patient's serum glucose. An initial insulin bolus is optional and rarely used nowadays.

Once hypovolemia has been corrected, serum glucose may decrease by as much as 30% to 50% as a result of corrected hemoconcentration and improved urinary output.24 Glucosuria accounts for 15% to 20% of the drop during initial rehydration and insulin therapy.25 Acidosis gets worse initially as improved perfusion mobilizes more lactic acid.

The insulin drip is started at 0.1 U/kg/h (short-acting insulin at 0.05 U/kg/h if the glucose level is greater than 1,000 mg/dL or the patient is under 3 years of age). The aim is to correct acidosis at a rate of 0.03 pH units per hour and decrease glucose by 100 mg/dL/h. If the expected rate of improvement is not achieved, change the rate of infusion and check the concentration of the prepared drip or make a new solution using a new vial of insulin.

Since insulin binds to the IV tubing, run off about 50 to 100 mL to saturate all the binding sites in the system before connecting the tubing to the patient. When glucose falls below 300 mg/dL, add:

5% dextrose 0.45 NS solution (D5-1/2 NS) and continue the insulin drip until pH is greater than 7.25. At this point, reduce insulin by titrating to maintain a glucose level of 150 to 250 mg/dL.2 Prepare to switch to subcutaneous insulin if pH is greater than 7.30, HCO3 is greater than 15 mEq/L, and the patient is tolerating oral fluids.

A common mistake is to decrease the insulin drip when hyperglycemia has improved but the pH or bicarbonate is still low.

Generally the glucose concentration falls in four to six hours while acidosis improves to a pH greater than 7.30 in eight to 12 hours.22 Serum glucose may decline more slowly in the presence of infection, and the insulin drip may have to be increased. Ketonuria may take several days to resolve, even in well-managed patients, because of the ongoing conversion of ß-hydroxybutyrate to acetone. Table 3 presents a summary of the phases of DKA therapy.



The phases of DKA therapy

Recovery>~24/48 hrTolerating oral intakeSQ


Managing cerebral edema

One of the goals of DKA management is to prevent cerebral edema. Table 4 lists risk factors for this condition. Great controversy surrounds the etiology of cerebral complications during therapy for DKA. Half the patients with cerebral edema have a prodrome of headache, altered mental status, and signs of increased ICP. Studies have shown that therapy succeeds in only 50% of patients with evidence of raised ICP.26



Risk factors for cerebral edema

Age under 5 years

Prolonged poor glycemic control

Glucose reduced at rate >100 mg/dL/hr

Negative sodium trends and lower Eosm resulting from hyponatremia

Bicarbonate therapy

Fluid administration of more than 4l/m 2/24 hr

Early signs of raised intracranial pressure


During a first episode of DKA, an increase in ICP usually occurs about six to 12 hours after initiation of therapy.27 Cerebral edema is rare before 3.5 hours and after 22 hours.27 The incidence is estimated to be 0.7 to 1.0 per 100 episodes of DKA and is inversely related to age. It occurs rarely in patients over 20 years of age.28 Mortality is about 70%, and only 7% to 14% of patients recover fully.28

In a study of six children in DKA, all six showed evidence of subclinical cerebral edema.29 A subsequent study failed to confirm these findings, but the patients in the second study received a hypertonic infusion from the early stages of treatment.30 The rate of fluid administration and serum concentrations of sodium and glucose have all been implicated in cerebral edema.18,26

After conducting a five-year prospective study, investigators came to the following conclusions:31

  • Patients with DKA rarely require more than 20 mL/kg of resuscitation fluid

  • Replace fluid evenly over 48 hours

  • Use solutions of higher tonicity, such as 0.9% NaCl

  • Adjust fluid therapy as necessary to prevent a decrease in the Eosm. A decrease in serum sodium or its failure to rise as blood glucose falls is a sign of impending cerebral edema

  • Insulin affects the movement of glucose and electrolytes into the brain cells, thereby increasing cell size.

Closely monitor patients with persistent headache and evidence of increasing ICP. Patients with altered mental status, pupillary changes, seizures, and abnormal vital signs—tachycardia, bradycardia, hypotension, hypertension, or irregular respirations—should be managed emergently to rapidly reduce ICP:

  • Elevate the head of the bed to 30°

  • Infuse mannitol 1 g/kg immediately

  • Intubate the patient using rapid sequence induction and hyperventilate to PaCO2 25 to 30 torr

  • Restrict IV fluids to two thirds maintenance

  • Obtain an urgent head CT scan after the patient is stabilized and consider continuous ICP monitoring. Dexamethasone has no proven role in treating cerebral edema caused by DKA.7

The transition phase

Patients with mild hyperglycemia and ketosis and those in DKA whose hyperglycemia and acidosis are resolving should be weaned off the insulin drip and IV fluids when they are clinically stable, hyperglycemia and metabolic acidosis have been corrected, and they are tolerating substantial amounts of oral fluids. The majority of patients do not require IV fluids beyond 24 hours. Short-acting (regular or Humalog) insulin at a dosage of 0.1 to 0.25 U/kg is started subcutaneously about 15 to 30 minutes before a meal, and the insulin drip and dextrose infusion are discontinued 30 to 60 minutes after the first dose. In known diabetics, a regimen of intermediate- and short-acting insulin can be started at mealtime. In new-onset type 1 diabetes, the total daily insulin dose is 0.5 to 1.0 U/kg given approximately as follows:

  • A.M.—2/3 total insulin dose (2/3 intermediate-acting, 1/3 short-acting)

  • P.M.—1/3 total insulin dose (1/2 intermediate, 1/2 short-acting).

Admitting the patient to the hospital

All patients in DKA should be admitted to the hospital. Patients who meet at least one of the following criteria should be transferred to the pediatric intensive care unit:

  • Altered mental status

  • Evidence of circulatory collapse

  • Cardiac dysrhythmia

  • Initial glucose greater than 1,000 mg/dL

  • Severe acidosis with pH less than 7.0

  • Age under 3 years

Patients who meet the following criteria should be admitted to the regular inpatient service with close monitoring:

  • New-onset diabetes type 1 or 2 (patients need education and nutrition counseling)

  • Unable to tolerate oral fluids

  • No indication of clinical improvement (acidosis or hyperglycemia persists)

  • Evidence of infection

Caveats and pearls

Tables 5 and 6 list some caveats and pearls to remember in treating the child with DKA. The task is challenging and requires close attention to vital signs, the patient's response to fluids and insulin, and preventing cerebral complications. Careful monitoring and, when feasible, early involvement of a pediatric endocrinologist should help achieve a positive outcome.



Caveats in managing DKA

Cerebral edema occurs about six to 12 hours after initiation of therapy, rarely after 22 hours

Bicarbonate bolus may precipitate ventricular arrhythmias

Excessive use of phosphate may cause hypocalcemia

Serum pH <7.0 indicates severe acidosis and risk of cerebral edema

pH may drop further after volume replacement as a result of mobilization of lactic acid

Initial hypokalemia is extremely worrisome since a further drop in K+ is anticipated with insulin therapy

Discontinuing insulin drip when glucose drops before correcting acidosis delays recovery

Total fluids in 24 hours >4L/m2 increase chances of cerebral edema



Pearls in managing DKA

Patients with type 2 diabetes can present in DKA

Fever is absent even with infection

Leukocytosis is present even without infection

Treat infection empirically if it is considered a precipitating factor

Serum amylase is usually elevated in the absence of pancreatitis

The degree of acidosis may not correlate with the degree of hyperglycemia

Insulin adheres to the walls of glass and polyvinyl tubing

Subcutaneous insulin has no role in early management of DKA


The authors gratefully acknowledge the assistance of Jeffrey Avner, MD, Director, Division of Pediatric Emergency Medicine, and Paul Saenger, MD, Director, Division of Pediatric Endocrinology, Montefiore Medical Center, for reviewing the manuscript.


1. National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases: Diabetes in America. Baltimore, National Institutes of Health, 1995, p 1468

2. Fiordalisi I, Harris GD: Diabetic ketoacidemia, in Finberg L (ed): Saunders Manual of Pediatric Practice. Philadelphia, WB Saunders Company, 1998, pp 826-832

3. Pinnas-Hamiel O, Dolan LM, et al: Increased incidence of non-insulin-dependent diabetes mellitus among adolescents. J Pediatr 1995;128:608

4. Ganda OP, Marble A: Report of the National Commission on Diabetes to the Congress of the United States of America, vol 3, part 2. Washington, DC, US Department of Health, Education, and Welfare, 1976, pp 88-97

5. Faich GA, Fishbein HA, Ellis SE: The epidemiology of diabetic ketoacidosis: A population-based study. Am J Epidemiol 1983;117:551

6. Warner E, Greene G, et al: Diabetic ketoacidosis associated with cocaine use. Arch Intern Med 1998;158:1799

7. Krane EJ: Diabetic ketoacidosis: Biochemistry, physiology, treatment, and prevention. Pediatr Clin North Am 1987;34:935

8. Siperstein MD: Diabetic ketoacidosis and hyperosmolar coma. Endocr Metab Clin North Am 1992;21:415

9. Sperling MA: Diabetes mellitus, in Sperling MA (ed.): Pediatric Endocrinology. Philadelphia, WB Saunders, 1996, pp 229-263

10. Finberg L: Why do patients with diabetic ketoacidosis have cerebral swelling, and why does treatment sometimes makes it worse? (Editorial). Arch Pediatr Adolesc Med 1996;150:785

11. Hale DE: Endocrine emergencies, in Fleisher GR, Ludwig S (eds): Textbook of Pediatric Emergency Medicine, ed 3. Baltimore, Williams & Wilkins, 1993, pp 940-944

12. Cydulka R: Diabetes mellitus and disorders of glucose homeostasis, in Rosen P, Barkin R (eds): Emergency Medicine: Concepts and Clinical Practice, ed 4. St Louis, CV Mosby Inc, 1998, pp 2457-2478

13. Knight AH, Williams DN, Ellis G, et al: Significance of hyperamylasemia and abdominal pain in diabetes ketoacidosis. BMJ 1973;3:128

14. Slovis CM, Mork VG, Slovis RJ, et al: Diabetic ketoacidosis and infection: Leukocyte count and differential as early predictors of serious infection. Am J Emerg Med 1987;5:1

15. Zeigler AG, Herskowitz RD, Jackson RA, et al: Predicting type I diabetes. Diabetes Care 1990;13:762

16. Rosenbloom AL, Schatz DA: Diabetic ketoacidosis in childhood. Pediatr Ann 1994;23:284

17. Duck SC, Wyatt DT: Factors associated with brain herniation in the treatment of diabetic ketoacidosis. J Pediatr 1988;113:10

18. Hale PM, Rezvani I, Braunstein AW, et al: Factors predicting cerebral edema in young children with diabetic ketoacidosis with new onset type I diabetes. Acta Paediatr 1997;86:626

19. Wu RHK: Endocrine emergencies, in Crain EF, Gershel JC (eds.): Clinical Manual of Emergency Pediatrics, ed 3. New York, McGraw Hill, 1997, pp 144-148

20. Harris GD, Fiordalisi I, et al: Minimizing the risk of brain herniation during treatment of diabetic ketoacidosis: A retrospective and prospective study. J Pediatr 1990;117:22

21. Saladino RA: Diabetic ketoacidosis, in Barkin RM (ed): Pediatric Emergency Medicine. St Louis, CV Mosby Inc, 1997, pp 765-768

22. Ennis ED, Stahl EJ, Kreisberg RA: Diabetic ketoacidosis, in Porte D, Sherwin RS (eds.): Ellenberg and Rifkin's Diabetes Mellitus: Theory and Practice. Stamford, CT, Appleton & Lange, 1997, pp 827-844

23. Fisher JN, Kitabchi AE: A randomized study of phosphate therapy in the treatment of diabetic ketoacidosis. J Clin Endocrinol Metab 1983;57:117

24. Luzi L, Barrett EJ, Groop LC, et al: Metabolic effect of low-dose insulin therapy on glucose metabolism in diabetic ketoacidosis. Diabetes 1988;37:1470

25. King AJ, Cooke NJ, McCuish A, et al: Acid-base changes during treatment of diabetic ketoacidosis. Lancet 1974;1:478

26. Rosenbloom AL: Intracerebral crises during treatment of diabetic ketoacidosis.Diabetic Care 1990;13:22

27. Rosenbloom AL, Riley WJ, Weber FT, et al: Cerebral edema complicating diabetic ketoacidosis in childhood. J Pediatr 1980;96:357

28. Lebovitz HE: Diabetic ketoacidosis. Lancet 1995;345:767

29. Krane EJ, Rockoff MA, Wallman JK, et al: Subclinical brain swelling in children during treatment of diabetic ketoacidosis. N Engl J Med 1985;312:1147

30. Smedman L, Escobar R, Hesser U, et al: Subclinical cerebral oedema does not occur regularly during treatment for diabetic ketoacidosis. Acta Paediatr 1997:86:1172

31. Harris GD, Fiordalisi I: Physiological management of diabetic ketoacidosis: A 5-year prospective pediatric experience in 231 episodes. Arch Pediatr Adolesc Med 1994;148:1046

DR. HAFEEZ is Assistant Professor of Pediatrics, Albert Einstein College of Medicine, and Attending Physician, Division of Pediatric Emergency Medicine, Children's Hospital at Montefiore, Montefiore Medical Center, Bronx, NY.
DR. VUGUIN is Assistant Professor of Pediatrics, Albert Einstein College of Medicine, and Attending Physician, Division of Pediatric Endocrinology, Children's Hospital at Montefiore.


This activity has been planned and implemented in accordance with the Essentials and Standards of the Accreditation Council for Continuing Medical Education through the joint sponsorship of Jefferson Medical College and Medical Economics, Inc.

Jefferson Medical College of Thomas Jefferson University, as a member of the Consortium for Academic Continuing Medical Education, is accredited by the Accreditation Council for Continuing Medical Education to sponsor continuing medical education for physicians. All faculty/authors participating in continuing medical education activities sponsored by Jefferson Medical College are expected to disclose to the activity audience any real or apparent conflict(s) of interest related to the content of their article(s). Full disclosure of these relationships, if any, appears with the author affiliations on page 1 of the article.


This CME activity is designed for practicing pediatricians and other health-care professionals as a review of the latest information in the field. Its goal is to increase participants' ability to prevent, diagnose, and treat important pediatric problems.

Jefferson Medical College designates this continuing medical educational activity for a maximum of one hour of Category 1 credit towards the Physician's Recognition Award (PRA) of the American Medical Association. Each physician should claim only those hours of credit that he/she actually spent in the educational activity.

This credit is available for the period of June 15, 2000, to June 15, 2001. Forms received after June 15, 2001, cannot be processed.

Although forms will be processed when received, certificates for CME credits will be issued every four months, in March, July, and November. Interim requests for certificates can be made by contacting the Jefferson Office of Continuing Medical Education at 215-955-6992.


1. Each CME article is prefaced by learning objectives for participants to use to determine if the article relates to their individual learning needs.

2. Read the article carefully, paying particular attention to the tables and other illustrative materials.

3. Complete the CME Registration and Evaluation Form below. Type or print your full name and address in the space provided, and provide an evaluation of the activity as requested. In order for the form to be processed, all information must be complete and legible.

4. Send the completed form, with $20 payment if required (see Payment, below), to:
Office of Continuing Medical Education/JMC
Jefferson Alumni Hall
1020 Locust Street, Suite M32
Philadelphia, PA 19107-6799

5. Be sure to mail the Registration and Evaluation Form on or before June 15, 2001. After that date, this article will no longer be designated for credit and forms cannot be processed.


Jefferson Medical College, in accordance with accreditation requirements, asks the authors of CME articles to disclose any affiliations or financial interests they may have in any organization that may have an interest in any part of their article. The following information was received from the author of "ITP: How much treatment is enough?"

Waseem Hafeez, MB, BS, has nothing to disclose.
Patricia Vuguin, MD, has nothing to disclose.



Waseem Hafeez, Patricia Vuguin. CME: Managing diabetic ketoacidosis--a delicate balance. Contemporary Pediatrics 2000;6:72.