The laboratory evaluation of patients suspected to have DKA includes determination of the blood glucose, plasma or urinary ketones, serum electrolyte concentration, blood urea nitrogen (BUN), creatinine, osmolarity, baseline calcium and phosphorus, and, if infection is suspected, complete blood count and blood culture. Bedside determination of the blood glucose with a glucose-monitoring device and evaluation of the urine for glucose and ketones should be performed as quickly as possible, and treatment should be initiated without waiting for the results of the laboratory assessment to become available. A baseline blood gas measurement should also be made to determine the pH and PCO₂. While venous blood gas measurements suffice in most episodes of DKA, an arterial blood gas measurement should be considered in patients who are suspected to have incomplete respiratory compensation or those with hemodynamic instability.

The initial blood glucose level is characteristically, but not invariably, elevated. Hyperglycemia beyond the renal threshold for glucose filtration indicates reduced glomerular filtration. Although the definition of DKA presupposes hyperglycemia, ketoacidosis with normal or even low glucose levels may occur in patients with known diabetes who have taken insulin recently. Poor oral intake or vomiting may also present with near-normal glucose concentrations.

Measurement of serum electrolytes in DKA reveals a low bicarbonate level and increased anion-gap metabolic acidosis, primarily due to the unmeasured ketoacids, including β-hydroxybutyrate and acetoacetate, as well as lactate in situations of severe tissue hypoxia. Most urine dipsticks that utilize a nitroprusside reaction to measure ketones detect only acetoacetate. In normal individuals, the ratio of β-hydroxybutyrate to acetoacetate is ˜1:1. However, this ratio may increase to 10:1 or more in DKA. Therefore, although useful to diagnose ketosis, these strips are not helpful in gauging the severity of DKA or response to treatment. Patients in DKA typically have an anion gap of 20-30 mEq/L. An anion gap >35 mEq/L suggests a concomitant lactic acidosis (3).

Serum electrolyte concentrations are almost always abnormal in DKA and may not accurately reflect total body electrolyte disturbances. Because of the osmotic flux of water from the intracellular space to the extracellular space in the presence of hyperglycemia, serum sodium will be reduced. The actual sodium concentration may be determined by adding 1.6 mEq/L for each 100 mg/dL rise in glucose concentration >100 mg/dL (5). Furthermore, extreme hypertriglyceridemia may cause the serum sodium concentration to be falsely lowered (dilutional hyponatremia) (6). A normal or elevated sodium level in the setting of severe DKA suggests extreme free water losses. The degree of elevation of the BUN and creatinine, as well as the hematocrit, may also indicate the extent of dehydration (and the possibility of renal damage).

The initial serum potassium may be low, normal, or high, depending on the degree of acidosis and the quantitative urinary losses. However, total body potassium stores are almost always depleted from excessive urinary losses, and normal or low K levels at presentation may be associated with severe hypokalemia once treatment for DKA is initiated. Phosphate, like potassium, is depleted in the setting of DKA, and serum phosphate levels may not accurately reflect total body stores. Serum phosphate level may also decrease during therapy.

A rough estimation of degree of dehydration should be made to calculate the fluid deficit and facilitate rehydration therapy even though such clinical approximations are subjective and typically inaccurate in estimating the actual degree of dehydration. Furthermore, the dehydration of DKA is hyperosmolar, resulting in a translocation of intracellular water to the extracellular compartment, thus masking the degree of dehydration. Significant total body fluid losses can occur before evidence of changes in vital signs or physical examination. Serum BUN and hematocrit may be useful markers of dehydration. As in other forms of hypertonic dehydration, too rapid fluid administration should be avoided in children in DKA. The fluid deficit should be corrected gradually, over 48 hours. The rare DKA patient who is hemodynamically unstable at the presentation should receive a 10 mL/kg normal saline or Ringer’s lactate bolus. Only one additional 10 mL/kg isotonic fluid bolus should be given if hemodynamic instability has not resolved. Thereafter, inotropic support, invasive monitoring, and an echocardiogram are required to guide hemodynamic management.

Serial calculations of actual (as opposed to measured) serum sodium concentrations are recommended, as sodium levels that fail to rise with treatment may signify excessive free water accumulation and an increased risk of cerebral edema (8,9). It is unusual to provide fluids in excess of 1.5-2 times the daily maintenance rate. The initial fluid replacement (first 6 hours) generally utilizes isotonic fluids. Subsequent fluids should contain at least 0.45% saline to ensure a gradual decline in serum osmolarity and prevent excessive free water accumulation (10). Hyperchloremic metabolic acidosis may develop with continued use of large volumes of normal saline. There are no data to support the use of colloids over crystalloids for this patient group (3).

Early potassium replacement is important to correct the deficit. With the initiation of insulin therapy and correction
of acidosis, serum potassium levels may drop precipitously as potassium shifts back from the extracellular to the intracellular compartment. In hypokalemic patients, replace potassium at the time of initial volume expansion and prior to starting insulin infusion. Otherwise, potassium should be administered only after adequate renal function is determined, particularly in patients who are hyperkalemic. Electrocardiographic monitoring facilitates early recognition of hyperkalemia (peaked T waves), hypokalemia (flat or inverted T waves), and the development of potentially dangerous cardiac dysrhythmias. Potassium is usually given at a dose of 40 mEq/L of fluid, either as potassium chloride or in combination with potassium phosphate. The latter has the additional advantage of replacing the phosphate deficit. Potassium can also be repleted as potassium acetate, which may theoretically provide buffering equivalents for the metabolic acidosis. The relative risks and benefits of these infusion alternatives in the treatment of DKA in children have not been rigorously studied (3,7). It should be noted that while hypophosphatemia in DKA may predispose to rhabdomyolysis and hemolysis and while low 2,3-diphosphoglycerate levels in DKA may impair tissue delivery of oxygen, complications of hypophosphatemia in DKA are rare, and phosphate replacement may precipitate acute hypocalcemia. Serum calcium should be monitored if phosphate is given. It is recommended to replace phosphate in patients with severe hypophosphatemia and unexplained weakness (3).

Acidosis in DKA reverses with fluid replacement and insulin therapy. Restoration of circulation improves renal perfusion, resulting in acid excretion. Improved tissue perfusion also limits the production of lactic acid in hypoperfused tissues. Insulin administration stops ketoacid production. Acidosis in DKA usually resolves later than hyperglycemia. However, it is important to continue insulin therapy until acidosis resolves while supporting the patient with glucose to prevent hypoglycemia.

Bicarbonate, once considered routine therapy in severe DKA, is no longer thought to be beneficial and is no longer recommended in the treatment of uncomplicated DKA. Most often, dramatic clinical improvement results simply from initial expansion of extracellular fluid volume, reestablishment of adequate peripheral perfusion, and prompt insulin administration. Bicarbonate therapy may lead to paradoxical worsening of CNS acidosis, hypokalemia, sodium overload, late alkalosis, and, theoretically, impaired tissue perfusion due to leftward shift of the oxyhemoglobin dissociation curve. Finally, bicarbonate has been associated, in multivariate analysis, with a higher risk of development of cerebral edema (8). Bicarbonate may still be considered in the setting of severe acidosis with impaired cardiac contractility and vasomotor tone. It may be given in a slow infusion of 1-2 mEq/kg over 60 minutes.

Insulin therapy, at a dose of 0.05-0.1 U/kg/h, should be initiated immediately after the patient has received initial volume expansion. Because insulin adsorbs to intravenous tubing, 20 mL of the insulin solution should be flushed through the tubing set before connecting to the patient. The insulin infusion can then be started immediately, and there is no need to allow the insulin to “dwell” in the intravenous tubing after the 20-mL flush (11). Intravenous insulin bolus is contraindicated because it may increase the risk of cerebral edema. As with fluid replacement, the aim of therapy is gradual correction of blood glucose (i.e., a reduction of the blood glucose by 50-100 mg/dL/h). However, the serum glucose concentration often falls significantly with initial rehydration alone owing to increased glomerular filtration from improved renal perfusion. The dose of insulin usually remains at 0.1 U/kg/h until acidosis resolves. Because insulin infusion often results in lowering of the blood glucose concentration before the decrease in ketoacidosis, it is important not to decrease the rate of insulin infusion, but instead to add dextrose to the rehydration fluids to prevent hypoglycemia. Dextrose should be added to the intravenous fluid solution when the serum glucose level falls below 250 mg/dL and should then be titrated to maintain blood glucose at about 200 mg/dL until DKA resolves. This may be accomplished with the simultaneous use of two intravenous solutions that differ only in the dextrose concentration (10% and 0%). Independent manipulation of each infusion allows the dextrose infusion to be varied quickly and efficiently (the so-called “two-bag” system) (12).

Some of the more common complications of DKA therapy are inadequate rehydration, hypoglycemia, hypokalemia, hyperchloremic acidosis, and cerebral edema. Of these, acute cerebral edema is the most serious, and accounts for 75%-87% of the mortality in DKA. Cerebral edema occurs in ˜0.5%-1% of episodes of pediatric DKA and carries a mortality rate of over 20% and a rate of permanent neurologic disability of over 25% (13). Cerebral edema is more likely to occur in patients at the time of first diagnosis of diabetes, in younger children, and in those presenting with the most severe degree of dehydration and metabolic derangement. It typically develops within the first 4-12 hours of treatment but may occur as late as 24-48 hours after the start of treatment. Symptoms and signs include headache, confusion, slurred speech, bradycardia, hypertension, and other signs of increased intracranial pressure. A bedside scoring system based on the patient’s neurologic status was proposed to diagnose cerebral edema (14). Although a number of causes of acute cerebral edema have been suggested from uncontrolled, retrospective studies, the etiology in many cases is not known. A common proposed mechanism involves a rapid decline in serum osmolality with treatment of DKA. On the basis of this hypothesis, previous recommendations for treatment of DKA advised limiting the rate of rehydration to <4 L/m²/day (15) and avoiding hypotonic fluids (16). Until the precise mechanisms that underlie the development and progression of cerebral edema are elucidated, it appears prudent to correct dehydration evenly over 48 hours (unless the patient is in shock), and monitor neurologic status and laboratory parameters closely.

Treatment of cerebral edema is aimed at lowering intracranial pressure. Reducing the rate of fluid administration by one-third is suggested. Prompt administration of IV mannitol (0.5-1 g/kg over 20 minutes) may be beneficial if given early in the course of cerebral edema. The dose may be repeated in 0.5-2 hours if there is no response to the initial dose. Hypertonic saline (3%), 5-10 mL/kg over 30 minutes, may be used as an alternative or in addition to mannitol. One should keep the patient’s head elevated and midline. Tracheal intubation may be needed for impending respiratory failure or airway protection. Although aggressive hyperventilation should be avoided as this decreases cerebral perfusion, care should be taken to match the level of hyperventilation the patient was spontaneously producing as sedating and ventilating the patient at a lower PaCO₂ than they were producing may worsen acidosis, cause cerebral vasodilation, and increase intracranial pressure if intracranial compliance is limited. Near infrared spectroscopy (NIRS) may be used to assess the ventilatory parameters and resulting PaCO₂ associated with a favorable cerebral O₂ supply/demand ratio. Intracranial imaging to exclude other pathologies, such as cerebral infarction or thrombosis, should be obtained but not at the expense of timely therapeutic interventions.

Other less common complications of DKA include thrombosis (including cerebral venous sinus thrombosis), a concern in children who require a central venous catheter for access; cardiac dysrhythmias, usually related to electrolyte disturbances; pulmonary edema; renal failure; pancreatitis; rhabdomyolysis; and infection, such as aspiration pneumonia, sepsis, and mucormycosis.

Infection is a common precipitating event for DKA. Infections should be appropriately worked up and treated. In cases of DKA in children with known diabetes, additional education on insulin omission, or sick day or pump failure management may be needed to prevent another episode of DKA.

Successful management of DKA requires meticulous attention to clinical and laboratory changes. Children with severe DKA or who are at increased risk for cerebral edema should be cared for in the ICU. Clinical monitoring is essential. Vital signs, perfusion, input/output balance, and neurologic status should be documented at least hourly. Cardiac monitoring is recommended owing to the risk of dysrhythmia. Laboratory monitoring should include venous blood glucose concentrations hourly and electrolytes and venous pH every 2-4 hours until normal. BUN and creatinine levels should be followed until they are normal, and calcium should be monitored, particularly if phosphate is administered. Lack of improvement in clinical and biochemical parameters with time suggests an occult infection or inadequate insulin or fluid replacement.