Treatment and complications of diabetic ketoacidosis in children INTRODUCTION Diabetic ketoacidosis (DKA) is the leading cause of morbidity and

d mortality in children with type 1 diabetes mellitus (DM), ranging from 0.15 percent to 0.31 percent [1-3]. In addition, DKA also can occur in children with type 2 DM; this presentation is most common among youth of African-American descent [4-8]. (See "Classification of diabetes mellitus and genetic diabetic syndromes".) The management of DKA in children, which is based upon the large collective experience of children with type 1 DM, will be reviewed here (table 1). There is limited experience in the management and outcomes of DKA in children with type 2 DM, although the same principles should apply. The clinical manifestations and diagnosis of DKA in children and the pathogenesis of DKA are discussed elsewhere. (See "Clinical features and diagnosis of diabetic ketoacidosis in children" and "Epidemiology and pathogenesis of diabetic ketoacidosis and hyperosmolar hyperglycemic state".) DEFINITION Consensus statements from the European Society for Paediatric Endocrinology/Lawson Wilkins Pediatric Endocrine Society (ESPE/LWPES) in 2004, the American Diabetes Association (ADA) in 2006, and the International Society for Pediatric and Adolescent Diabetes (ISPAD) in 2007 defined the following biochemical criteria for the diagnosis of DKA [9-12]:

Hyperglycemia, defined as a blood glucose of >200 mg/dL (11 mmol/L) AND Metabolic acidosis, defined as a venous pH <7.3 and/or a plasma bicarbonate <15 meq/L (15 mmol/L)

These abnormalities are accompanied by disturbances in fluid and electrolyte balance, resulting in volume depletion and mild to moderate serum hyperosmolality. The clinical manifestations of DKA are related to the degree of hyperosmolality, volume depletion, and severity of acidosis [13]. (See "Clinical features and diagnosis of diabetic ketoacidosis in children".) TREATMENT The approach and principles of management are the same for all children with DKA regardless of the severity of DKA. However, based upon the severity of presentation, the clinician should determine the appropriate clinical setting in which to treat the child. As an example, a mild presentation of DKA without vomiting may be managed in an ambulatory setting under close supervision and monitoring by an experienced diabetes team. On the other hand, a patient with severe DKA will need a level of care found in a pediatric intensive care unit. The clinician must individualize the treatment plan based on the child's physical and laboratory findings, and treatment will need to be adjusted over time for each child. No protocol or formula should be blindly used in the management of DKA. Assessment of severity The following clinical and laboratory findings have been used by the authors to estimate the severity of DKA at presentation:

Neurologic status Severe neurologic compromise at presentation is a poor prognostic indicator, primarily because such patients are at increased risk for developing cerebral edema during therapy. This was illustrated in a retrospective multicenter study of 61 children with DKA and cerebral edema; all patients who either died or survived in a persistent vegetative state presented with Glasgow coma score 7 (score of 6 to 7 includes an abnormal or absent purposeful response to pain) (table 2) [14]. The Glasgow coma score is not adequately sensitive to identify patients early in the course of developing cerebral edema. As a result, we suggest additional monitoring of all patients for the development of more subtle signs of cerebral edema (table 3). (See 'Cerebral edema' below.)

Acid-base status The venous pH and serum bicarbonate concentration directly reflect the severity of the acidosis (table 4). The respiratory rate also may be helpful, because the magnitude of the respiratory compensation is directly related to the severity of the acidosis. The anion gap is a useful measure of the severity of the metabolic acidosis (because it reflects both ketosis and lactic acidosis). A high anion gap may also reflect decreased renal perfusion, which limits ketoacid excretion. Alternatively, measurement of plasma beta-hydroxybutyrate may be available and permits direct monitoring of the degree of ketoacidemia.

Volume status Children with moderate or severe DKA are estimated to present with a 7 to 10 percent fluid deficit, as discussed below. (See 'Fluid and electrolyte deficits and repletion' below.) Duration of symptoms A long duration of symptoms, as well as depressed level of consciousness or compromised circulation, is evidence of severe DKA and should prompt close monitoring for potential complications of DKA, such as cerebral edema [9-11].

Moderate and severe DKA The underlying principles of management are to administer insulin and to correct the fluid and electrolyte abnormalities of metabolic acidosis, hypovolemia, potassium, and perhaps phosphate depletion. During initial therapy, the patient should be carefully monitored for signs of cerebral edema. (See 'Cerebral edema' below.) Fluid and electrolyte deficits and repletion Studies estimating water and electrolyte losses in DKA were conducted in the 1940s and 1950s. Most included adults, but one was a detailed study of a 10-year-old female [15-17]. The data from the available studies are consistent with the following average losses in severe DKA:

Water 70 (range 30 to 100) mL/kg Sodium 5 to 13 meq/kg Potassium 6 to 7 meq/kg

It is difficult to clinically assess the degree of dehydration in children presenting with DKA [18]. The ESPE/LWPES consensus recommended that initial fluid management of children with moderate or severe DKA be based upon the assumption that patients present with a 7 to 10 percent fluid deficit [9,10]. This recommendation is consistent with the above studies that assessed fluid and electrolyte losses. However, fluid replacement should be performed gradually, as discussed below. Urinary losses from the glucose osmotic diuresis (due to osmotic action of glucose and ketones in the urine) and gastrointestinal losses from vomiting and/or diarrhea, if present, result in extracellular volume depletion. At the time of presentation, clinical estimates of the extracellular fluid volume deficit in moderate to severe DKA are usually in the range of 7 to 10 percent [9,10,18]. Hypovolemic shock is a rare occurrence in DKA. The patient in shock should be evaluated for other causes of shock. Evaluation of the degree of dehydration in the child with DKA is difficult even for the experienced clinician. Many of the clinical findings used to assess volume status are unreliable, such as the condition of the patient's oropharynx, which is almost always dry because of hyperventilation and mouth breathing. Skin turgor is also frequently unreliable because the fluid losses are from both the extracellular and intracellular spaces. In addition, the degree of extracellular fluid loss is in part masked because hyperglycemia results in a shift of water from the intracellular to the extracellular fluid compartment [18]. This effect is rapidly reversed with insulin therapy. (See "Clinical features and diagnosis of diabetic ketoacidosis in children", section on 'Signs and symptoms'.) Elevated urine specific gravity, which is a common sign of volume depletion in children, cannot be used as a measure of hypovolemia in patients with DKA because glucose and, to a lesser degree, ketones raise the specific gravity. Elevations in the blood urea nitrogen and hematocrit remain useful laboratory studies to confirm hypovolemia in DKA and to follow its correction. Fluid repletion The goals of initial volume expansion are [9,10]:

To restore the effective circulating volume by replacing sodium and water loss To restore glomerular filtration rate to enhance clearance of ketones and glucose from the blood

Most experts recommend accomplishing the volume expansion gradually and with isotonic fluids, because these approaches might reduce the risk for cerebral edema [10-12]. The evidence underlying this possible association is weak and conflicting. In one clinical study in children, high volume of fluid administration during the first four hours of treatment was one of several risk factors for cerebral edema [19]. However, several other clinical studies failed to show such an association [20-22]. Nonetheless, gradual correction of the deficit and the use of isotonic fluids seems prudent. In patients who are not markedly hypovolemic, this approach may even result in earlier reversal of the acidosis [23,24]. Initial volume expansion The volume and rate of administration depend upon the effective circulating volume. Fluid repletion in moderate to severe DKA is usually begun with an infusion of 10 mL/kg over one hour. If and only if the effective circulating volume is still compromised, an additional infusion of 10 mL/kg can be given over the next hour. We generally do not give more than 20 mL/kg in total boluses unless the patient's cardiovascular status is compromised. (See "Clinical features and diagnosis of diabetic ketoacidosis in children", section on 'Signs and symptoms'.) The initial volume expansion should be with a solution which is near isotonic, such as isotonic saline [9,10]. In our practice, the initial volume expansion is done with Ringer's Lactate. Subsequently, the deficit is corrected with Ringer's lactate to which potassium has been added. (See 'Metabolic acidosis' below.) Subsequent fluid administration Once the child is hemodynamically stable, subsequent volume expansion should be given more slowly. The rate of fluid administration should not exceed 1.5 to 2 times the maintenance rate and should not include urinary losses. The hourly rate of fluid administration should never exceed two times the maintenance rate (about 3000 cc/m2/day) unless there is objective evidence of shock; excessive fluids may increase the risk for cerebral edema. (See "Maintenance fluid therapy in children" and 'Cerebral edema' below.) The solution used for this ongoing volume repletion should initially consist of isotonic saline (normal saline or lactated Ringers) for approximately four to six hours. For most patients, 40 mEq/L of potassium salts should be added to the solution to correct the total body potassium deficit. (See 'Serum potassium' below.) In our practice, we use lactated Ringers to which 35 mEq/L of potassium is added (20 mEq/L as potassium phosphate and 15 mEq/L as potassium chloride) throughout the period of rehydration. After the first four to six hours of treatment, some clinicians will reduce the sodium concentration to not less than one-half isotonic, provided that the serum sodium concentration is rising appropriately as the serum glucose concentration falls, and that the patient's circulatory and mental status are stable. The choice between one-half isotonic and isotonic saline depends on the patient's circulatory status and the rise in serum sodium as the hyperglycemia is corrected. Inclusion of potassium salts in the replacement fluid should be based on serum potassium concentrations. Once the ketoacidosis is predominantly cleared and oral fluids are permitted, hourly rates of fluid administration can be liberalized to achieve full hydration over 48 to 72 hours [9,10]. The rate over the first 24 hours should not exceed 1.5 to 2 times the usual rate of administration of maintenance fluid. Based upon body surface area, the total fluid intake should be no greater than 3500 mL/m2 for 24 hours, including the initial fluid bolus as well as any oral intake. Insulin After the initial fluid bolus is complete, an insulin infusion is begun at a rate of 0.1 unit/kg per hour [9,10]. A lower dose of 0.05 unit/kg per hour may be used initially in younger children who may be more sensitive to insulin. Do not give an initial bolus of insulin. A study with a case-control design suggests that the risk of cerebral edema is decreased by delaying insulin administration for one hour or more after initiation of fluid therapy [19]. The insulin can be mixed in one-half isotonic saline and administered in a syringe infusion pump to control the rate of insulin administration. The solution should be concentrated as much as possible, because the tubing and syringe can bind insulin, and should be flushed through the tubing of the insulin infusion syringe. As an example, 50 units of short-acting ("regular") insulin are added to 50 mL of one-half isotonic saline, providing

1 unit per mL of infusate. The syringe is then "piggybacked" into the patient's indwelling intravenous catheter as close as possible to the venous site. Within 60 minutes, steady state serum insulin levels are achieved (100 to 200 microU/mL), which offset insulin resistance, suppress glucose and ketone production, and stimulate peripheral glucose and ketone metabolism [25,26]. In addition, volume expansion will lower the serum glucose by dilution. (See "Epidemiology and pathogenesis of diabetic ketoacidosis and hyperosmolar hyperglycemic state".) In most patients, the hyperglycemia corrects before the ketoacidosis. Normalization of the serum glucose concentration may not always be accompanied with an improvement in the metabolic acidosis. The insulin infusion rate should only be reduced when the ketoacidosis is corrected. If the ketoacidosis persists, the patient should be reassessed. Possible explanations are severe insulin resistance due to infection, incorrect preparation of the insulin infusion, and decreased insulin delivery due to adhesion of insulin to tubing. We recommend preparing a new insulin syringe for the pump to rule out the possibility of an erroneous dilution of insulin. The insulin dose in such patients usually needs to be increased. When the serum glucose concentration decreases to 250 to 300 mg/dL (13.9 to 16.7 mmol/L), the intravenous fluid infusion should be changed to 5 percent dextrose in isotonic saline or lactated Ringer's solution. This allows continued administration of insulin, which is often necessary to correct the residual ketoacidosis [11]. If the serum glucose falls below 250 mg/dL (13.9 mmol/L) before complete resolution of the ketoacidosis, the concentration of dextrose in the intravenous solution should be increased to up to 10 to 12.5 percent. Some institutions use a "two bag system" to adjust the dextrose concentration of the intravenous fluids between 0 and 10 percent [23,27]. In this technique, two bags of the selected saline solution are prepared, one of which contains 10 percent dextrose and the other 0 percent dextrose. By adjusting the relative rates of fluid administration from each bag, the rate of fluid administration can be maintained constant, while the variable rate of glucose infusion can be achieved to respond to changes in the patient's serum glucose concentrations. The purpose of adding dextrose to the intravenous solution is to prevent hypoglycemia while continuing to administer insulin to correct ketoacidosis. In our practice, we establish steady rates of insulin and intravenous fluids during the first six to eight hours of insulin therapy, which allows for a gradual decrease in serum glucose concentrations (about 50 to 100 mg/dL per hour or 3 to 5 mmol/L per hour). Once the serum glucose reaches 200 to 300 mg/dL, dextrose is added to the IV solution (usually as 5 percent dextrose) to permit the continued insulin administration necessary to clear the ketoacidosis while minimizing the risk of hypoglycemia. In younger patients with increased insulin sensitivity it may be necessary to decrease the insulin infusion rate to 0.05 units/kg/hour as long as the ketoacidosis continues to improve. For patient safety reasons, it is advisable to keep serum glucose concentrations around 150 to 200 mg/dL (8.3 to 11.1 mmol/L) for younger children; or 100 to 150 mg/dL in older children (5.5 to 8.3 mmol/L), before switching to subcutaneous insulin. In unusual circumstances, and especially if facilities to administer intravenous insulin are not readily available, subcutaneous or intramuscular insulin can be used as initial therapy [28,29]. However, with subcutaneous administration, the absorption of insulin may be inconsistent, particularly in the setting of volume depletion and secondary sympathetic activation which can decrease local perfusion [30]. Serum sodium The initial sodium concentration will depend on the net sodium and water losses prior to hospitalization, the degree of hyperglycemia and the degree of lipemia (severe hyperlipidemia causes pseudohyponatremia). (See "Clinical features and diagnosis of diabetic ketoacidosis in children".) It must be kept in mind that diabetic ketosis is a form of hypertonic dehydration, and care must be taken in rehydrating the patient. During rehydration, the clinician must serially assess the serum sodium and employ anticipatory planning in the fluid replacement plan. Reversing the hyperglycemia with fluid expansion and insulin administration will lower the plasma osmolality, cause water to move from the extracellular fluid into the cells, and raise the serum sodium concentration. Failure of the serum sodium to rise appropriately could be an early sign that the patient is at risk for cerebral edema, as suggested by a few studies [31-35]. A causal association between the rate or concentration of sodium administered during treatment for DKA and the development of cerebral edema has not been established [12]. If the serum sodium fails to rise appropriately, consideration should be given to increasing the sodium concentration in the infused fluids and decreasing their rate of administration. The patient should be carefully monitored for early signs of cerebral edema. (See 'Cerebral edema' below and "Cerebral edema in children with diabetic ketoacidosis".) In our practice, we measure serum sodium hourly for the first three to four hours and then every two hours, to assess the anticipated changes of serum sodium described above. When clinically appropriate, the frequency of measurement can be reduced to every four to six hours. With appropriate replacement of fluids and electrolytes in a child who is otherwise improving, the serum sodium concentration should gradually increase. The expected rise in serum sodium (Na) attributable to the decrease in serum glucose can be calculated according to the following formula [36]: Corrected serum Na = Measured serum Na + [SG 42] where SG is the increment above normal in the serum glucose concentration (in mg/dL). The SG should be divided by 2.3 if measured in mmol/L. Thus, serum Na should rise by approximately 2.4 meq/L for every 100 mg/dL fall in plasma glucose. Some clinicians plot the corrected sodium concentration on a nomogram (figure 1) to facilitate tracking of changes in serum sodium during insulin treatment and to help identify patients in whom the serum sodium does not rise appropriately. Serum potassium Both renal and gastrointestinal losses can contribute to an often marked degree of potassium depletion in DKA. These losses will tend to produce hypokalemia. On the other hand, the combination of insulin deficiency, which impairs potassium entry into the cells, and hyperosmolality, which pulls water and potassium out of the cells, tend to raise the serum potassium. Ketoacidosis itself appears to have little effect on transcellular potassium movement. The evidence supporting this conclusion and why this might occur are discussed elsewhere. (See "Potassium balance in acid-base disorders".) Because of these counteracting effects, the serum potassium at the time of presentation can be normal, increased, or decreased. Regardless of the initial level, insulin therapy drives potassium into cells, resulting in a fall in the serum potassium concentration. Thus, potassium replacement will almost always be required within one to two hours of the initiation of fluid and insulin therapy in children with DKA who do not have renal

failure. In our experience, most patients present with potassium levels in the normal or high range. After initial IV expansion, the potassium concentration generally falls into the normokalemic or mildly hypokalemic range. It is uncommon for patients to present with hypokalemia before fluid expansion. However, should this be the case, more aggressive potassium replacement is indicated. Optimal therapy varies with the initial serum potassium. Regardless of the regimen, the serum potassium should be carefully monitored during therapy. In addition, electrocardiographic monitoring is recommended in patients with either hypokalemia or hyperkalemia.

If the patient is normokalemic, potassium replacement should be given with the start of insulin therapy (eg, adding 40 meq/L of potassium to the saline solution), as insulin will reduce the serum potassium. If the patient is markedly hypokalemic, potassium replacement should be started immediately, using an initial potassium concentration of 40 meq/L, concurrent with volume expansion. The serum potassium levels should be monitored hourly and the replacement adjusted as needed. The insulin infusion should be delayed and/or given at a reduced rate until the serum potassium is in the normal range.

If the patient is hyperkalemic, potassium replacement should be initiated when the serum potassium falls to normal and after verifying urine production.

Potassium replacement is usually given as potassium chloride. Potassium phosphate and potassium acetate also have been used. Potassium replacement therapy should continue with intravenous insulin and fluid therapy and the serum potassium should be carefully monitored. Metabolic acidosis Insulin and fluid repletion leads to partial correction of the acidosis seen in DKA. Insulin promotes the metabolism of ketoacid anions, resulting in the generation of bicarbonate, and stops the ongoing production of new ketoacids. Improved tissue perfusion corrects any lactic acidosis that might be present. Ketoacid anions have been called "potential bicarbonate," because their metabolism following the administration of insulin results in the generation of bicarbonate and reversal of the acidosis. However, bicarbonate cannot be regenerated from the ketoacid anions that were excreted in the urine. As a result of the urinary loss of "potential bicarbonate," almost all patients with DKA (except those advanced renal failure) develop a usually mild normal anion gap acidosis (also known as "non-gap acidosis") during treatment [37-39]. Suppose, for example, that a patient has a serum bicarbonate of 8 meq/L, an anion gap of 24 meq/L, and a ketone body concentration of 12 mM. Administration of insulin will convert all 12 mM of ketoacid anion back into bicarbonate; however, the serum bicarbonate concentration may only rise by about 6 meq/L (to 14 meq/L), with the remainder replenishing the cell and bone buffer stores. At this point, the patient has a metabolic acidosis with a normal anion gap of 12. If no ketoacid anions had been excreted in the urine (as in a dialysis patient), then insulin therapy would have returned both the anion gap and serum bicarbonate concentration to baseline. (See "The anion gap/HCO3 ratio in patients with a high anion gap metabolic acidosis".) There is substantial evidence that bicarbonate therapy should not be used in DKA. Controlled trials both in adults and children have been unable to demonstrate any clinical benefit from the routine administration of sodium bicarbonate [40,41]. In addition to lack of benefit, there are potential risks from bicarbonate therapy.

Excessive bicarbonate therapy to a near-normal pH can lead to a rapid rise in PCO2 (since there is less of an acidemic stimulus to hyperventilation), resulting in a paradoxical fall in cerebral pH as the lipid-soluble CO2 rapidly crosses the blood-brain barrier [9,10]. The administration of alkali may slow the rate of recovery of ketosis. One report evaluating seven patients found that the three who received bicarbonate had a rise in plasma ketoacid levels during bicarbonate infusion and also had a six-hour delay in improvement of ketosis [42]. Studies in animals suggested that bicarbonate therapy acts by increasing hepatic ketogenesis. The administration of bicarbonate can lead to a post-treatment metabolic alkalosis after insulin-induced metabolism of ketoacid anions results in the regeneration of bicarbonate and spontaneous correction of most of the metabolic acidosis. Bicarbonate therapy may be a risk factor for cerebral edema [22]. (See 'Cerebral edema' below.) The rapid correction of acidosis with bicarbonate therapy may result in hypokalemia [9,10]. The additional sodium load can further increase the degree of hyperosmolality before decreasing glucose levels with fluid and insulin therapy.

For all of these reasons, we do not use bicarbonate for the treatment of DKA unless other complications occur, such as cardiac arrest. However, the ADA and ESPE/LWPES consensus statements suggest that highly selected patients may benefit from cautious alkali therapy [9-11], in particular, patients with severe acidemia (arterial pH <6.90) in whom decreased cardiac contractility and vasodilatation can further impair tissue perfusion. Phosphate Cellular phosphate depletion is a common problem in uncontrolled diabetes mellitus. The serum phosphate concentration may initially be normal or elevated due to movement of phosphate out of the cells. As with hyperkalemia, the phosphate depletion is rapidly unmasked following the institution of insulin therapy, frequently leading to hypophosphatemia. Plasma phosphate levels frequently fall as low as 1 mg/dL during treatment of DKA, without evident adverse consequences. There have been concerns that low plasma phosphate levels could result in metabolic disturbances with particular concern about the erythrocyte 2,3-diphosphoglycerate concentration and its effect on tissue oxygenation [43]. However, hypophosphatemia has not been shown to affect tissue oxygenation in adult patients with DKA [44] and randomized trials of phosphate replacement in adults have failed to show clinical benefit [45,46].

Intravenous phosphate administration may induce hypocalcemia and hypomagnesemia, and is generally not necessary or sufficient to replace the total body phosphate deficit [47]. Therefore, we do not suggest phosphate replacement to address asymptomatic mild hypophosphatasemia. The phosphate deficit will be repleted after the DKA resolves and the patient resumes eating. The main indication for phosphate therapy is a serum phosphate concentration less than 1.0 mg/dL (0.32 mmol/L). When phosphate is given, careful monitoring of the serum calcium and magnesium is required [9,10]. In this case, phosphate can be given by using potassium phosphate salts instead of potassium chloride to provide potassium replacement. It is also reasonable to use potassium phosphate salts for part of the potassium replacement for patients with lesser degrees of hypophosphatemia. (See "Treatment of diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults", section on 'Phosphate depletion'.) Monitoring The initial evaluation of the child with DKA is discussed separately. (See "Clinical features and diagnosis of diabetic ketoacidosis in children".) Treatment of DKA requires close monitoring of the patient's clinical condition including changes in vital signs, neurologic status, fluid status, and metabolic state (eg, serum electrolytes, glucose, urea nitrogen, hematocrit, and venous pH) [9,10]. This should be accompanied by appropriate documentation that is easy to follow and is able to detect clinical changes that require medical attention. It is the authors' experience that a flow sheet of laboratory values and clinical parameters allows better visualization and evaluation of the clinical picture throughout treatment of DKA. Emphasis should be placed on the hourly evaluation of the patient's fluid status, changes in osmolality, glucose, and electrolytes, and changes in the patient's mental and neurologic status as outlined below:

Monitor blood glucose hourly during the initial 4 to 6 hours of fluid and insulin therapy. Subsequently, the frequency of blood glucose measurements should be dictated by the rapidity of correction of the metabolic acidosis and hyperglycemia. If capillary samples are obtained for the measurement of blood glucose, they should initially be compared to a venous measurement. These measurements should be used to adjust the rate of dextrose infusion by adjusting the dextrose concentration in the fluids (but not the rate of fluid or insulin infusion), until the metabolic acidosis due to ketosis is predominantly cleared. It is important to remember that insulin and fluids will not correct a metabolic acidosis due to hyperchloremia. (See 'Insulin' above.)

Electrolytes and venous pH should be initially evaluated hourly for the first three to four hours and then every two hours. When clinically appropriate, the frequency of measurement can be reduced to every four to six hours. These measurements are used to assess the anticipated changes of sodium, potassium, bicarbonate, and anion gap. (See 'Serum sodium' above.) Clinical parameters including heart rate, respiratory rate, blood pressure, oxygen saturation, and neurologic status should be monitored continuously. In patients with severe DKA or altered mental status, frequent neurologic examinations are recommended. Monitor for warning signs and symptoms of cerebral edema including headache, inappropriate decrease in heart rate, recurrence of vomiting, changes in neurologic status, rising blood pressure, and decreased oxygen saturation. (See 'Cerebral edema' below.)

Initiate electrocardiographic monitoring in patients with severe DKA or abnormal serum potassium concentrations. Evaluate the tracing for changes characteristic of hyperkalemia or hypokalemia. (See "ECG tutorial: Miscellaneous diagnoses", section on 'Electrolyte abnormalities'.) Measure and record fluid input and output accurately. If the patient is neurologically impaired or it is difficult to ascertain urine output, a urine catheter should be placed.

Discontinuing the insulin infusion The insulin infusion should continue at 0.05 to 0.1 units/kg per hour until the following conditions are met [9-11]:

Serum anion gap reduced to normal (12 2 meq/L) Venous pH is >7.30 or serum HCO3 is >15 meq/L Plasma glucose <200 mg/dL (11.1 mmol/L) Tolerating oral intake

During treatment for DKA, correction of the ketoacidosis does not necessarily result in complete correction of the metabolic acidosis. The residual acidosis is associated with a normal anion gap that is due to urinary loss of ketoacid anions which, as noted above, represents the loss of "potential bicarbonate." In this setting, insulin therapy will have no further effect on the acidosis. Thus, a persistent normal anion gap acidosis is not a contraindication for switching the patient to subcutaneous insulin, since there are no circulating ketoacids remaining. (See 'Metabolic acidosis' above.) The first subcutaneous injection should be given at an appropriate interval to allow for absorption prior to stopping insulin infusion. The onset of rapid-acting insulin (ie, insulin lispro) is approximately 15 minutes, whereas that of short-acting (regular) insulin is 30 to 60 minutes. Borderline DKA Patients with borderline DKA (venous pH >7.30, serum bicarbonate >16 meq/L), no neurologic impairment, and estimated volume deficit less than 3 percent (table 4) who are not vomiting may be managed in an ambulatory setting under the supervision of an experienced medical team. However, hospitalization may appropriate for young children (eg, <5 years of age) because of their sensitivity to insulin as compared with older children, and because of their increased risk for cerebral edema. Children of any age should be treated in hospital if the home environment does not provide close supervision and monitoring.

Commercially available fluids for oral hydration such as Rehydralyte (75 meq/L sodium and 20 meq/L potassium) can be used alone or in combination with initial parenteral fluid repletion. These fluids contain dextrose 2.5 g/dL (2.5 percent). A short-acting insulin (eg, recombinant modified human insulin, also known as "regular" insulin) can be given subcutaneously and repeated at three to six hour intervals as needed. The dose is adjusted depending upon the response. Close monitoring of serum electrolytes, glucose, and fluid balance is required. The patient should be transferred to an appropriate inpatient setting if there is an inadequate response or the clinical condition deteriorates. COMPLICATIONS AND MORTALITY Reported mortality rates for DKA are consistent in developed countries, ranging from 0.15 to 0.51 percent in national population studies in Canada, the United Kingdom, and the United States [1-3,48]. Cerebral edema accounts for the majority of deaths (60 to 90 percent) [22,49]. Other causes include aspiration pneumonia, multiple organ failure, gastric perforation, and traumatic hydrothorax [2]. Cerebral edema Cerebral edema occurs in 0.3 to 1 percent of children with DKA and has a high mortality rate of 21 to 24 percent [9-11,48]. Children who are younger, newly diagnosed with diabetes, or who present with severe acidosis or dehydration are at the greatest risk. Cerebral edema generally develops during treatment for DKA, and current protocols were strongly influenced by efforts to reduce risks. Nonetheless, cerebral edema can occur in optimally managed cases, and early identification and treatment of the problem is essential. Up to 20 percent of cases of cerebral edema occur before the initiation of therapy [22]. All children should be carefully monitored for early signs of cerebral edema throughout the course of treatment for DKA (table 3) [12]. The Glasgow coma scale can be used to evaluate for neurologic dysfunction, but may not be adequately sensitive to identify children early enough for effective intervention. More subtle neurological symptoms such as new onset or intensifying headache, altered consciousness, recurrent vomiting, age-inappropriate incontinence, irritability, or lethargy may be valuable as earlier indicators of cerebral edema [11,50]. Changes detectable by head computed tomography occur late in the development of cerebral edema. Therefore, the decision to treat should be based on clinical evaluation. Imaging may be useful to exclude other causes of neurological deterioration, but should not delay treatment. The pathophysiology, risk factors, diagnosis, and treatment of cerebral edema in children with DKA are discussed in detail separately. (See "Cerebral edema in children with diabetic ketoacidosis".) Cognitive impairment Preliminary evidence suggests that DKA may be associated with subsequent neurocognitive dysfunction, even in patients who did not have clinical evidence of cerebral edema [51,52]. There is insufficient information to determine if these effects are directly related to DKA or are a marker for other complications of type 1 diabetes, such as recurrent hypoglycemia. (See "Complications and screening in children and adolescents with type 1 diabetes mellitus", section on 'Neurologic sequelae'.) Venous thrombosis Children with DKA appear to be at increased risk for deep venous thrombosis, particularly in association with femoral central venous catheter placement [53,54]. It has been suggested that this may in part be due to a prothrombotic state associated with DKA [55]. Aspiration Children with DKA who present with an altered state of consciousness and vomiting are at increased risk for aspiration. Placement of a nasogastric tube should be performed and stomach contents emptied. If necessary, intubation also should be performed to protect the airway. Cardiac arrhythmia Cardiac arrhythmias may be seen with either hypokalemia or hyperkalemia. (See 'Serum potassium' above.) Pancreatic enzyme elevations Mild elevations in serum amylase and lipase are seen in about 40 percent of children with DKA, and are also common in adults with DKA [56]. In most cases, this does not reflect acute pancreatitis. The diagnosis of acute pancreatitis should be based on clinical findings and confirmed by CT scan. (See "Clinical features and diagnosis of diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults", section on 'Serum amylase and lipase'.) PREVENTION Attempts should be made to prevent DKA both before and after the diagnosis of diabetes mellitus has been established. Methods that may promote earlier diagnosis of diabetes include increasing awareness of both public and healthcare providers [57] and identifying high-risk individuals through family history, genetic, and immunologic screening [9,10]. (See "Prediction of type 1 diabetes mellitus" and "Prevention of type 1 diabetes mellitus".) Among children with known diabetes, a comprehensive diabetes program may reduce the rates of DKA [58,59]. Identification of the patients at greatest risk for recurrence and placement in a comprehensive program may decrease the rate of DKA. (See "Clinical features and diagnosis of diabetic ketoacidosis in children".) Worsening glycemic control often precedes the onset of DKA. Thus, the patient or caregiver should be taught to increase the frequency of testing blood glucose and measuring urine or blood ketones when the patient's blood glucose is greater than 240 mg/dL (13.3 mmol/L). INFORMATION FOR PATIENTS UpToDate offers two types of patient education materials, The Basics and Beyond the Basics. The Basics patient education pieces are written in plain language, at the 5th to 6th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-toread materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10th to 12th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon. Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on patient info and the keyword(s) of interest.)

Basics topics (see "Patient information: Diabetic ketoacidosis (The Basics)")

SUMMARY AND RECOMMENDATIONS The following is a summary of the above discussion regarding the treatment of DKA and is consistent with consensus statements from the American Diabetes Association, European Society for Paediatric Endocrinology, the Lawson Wilkins Pediatric Endocrine Society, and the International Society for Pediatric and Adolescent Diabetes (table 1) [9-12]:

Obtain initial laboratory studies (electrolytes, serum glucose, venous pH, urea nitrogen, hematocrit, and a urine analysis; and blood beta-hydroxybutyrate if available). Continue to monitor blood glucose hourly during the initial 4 to 6 hours of fluid and insulin therapy. Subsequently, the frequency of blood glucose measurements should be dictated by the rapidity of correction of the metabolic acidosis and hyperglycemia. Measure electrolytes and venous pH hourly for the first three to four hours, then every two hours thereafter until the metabolic acidosis is cleared. (See 'Monitoring' above.)

We recommend careful evaluation and monitoring for signs and symptoms of cerebral edema in all children undergoing treatment for DKA (Grade 1C). Specific signs of increased intracranial pressure and changes detectable by head computed tomography often occur too late for effective intervention. Early symptoms that may indicate cerebral edema include new onset or intensifying headache, altered consciousness, and recurrent vomiting, ultimately progressing to bradycardia and opisthotonic posturing. (See 'Cerebral edema' above and "Cerebral edema in children with diabetic ketoacidosis".)

Initiate fluid repletion using an isotonic crystalloid solution at a rate of 10 mL/kg in the first hour. Repeat this regimen if the effective circulatory volume remains compromised. (See 'Fluid repletion' above.) After the initial fluid bolus is complete, an insulin infusion is begun at a rate of 0.1 unit/kg per hour. A lower dose of 0.05 unit/kg per hour may be used initially in younger children who may be more sensitive to insulin. An insulin bolus should NOT be given. Consider subcutaneous insulin when the venous pH is >7.30, the serum bicarbonate is >16 meq/L, or the anion gap is normal. (See 'Insulin' above.)

Add 5 percent dextrose to the intravenous fluid when the serum glucose decreases to 250-300 mg/dL (13.9 to 16.7 mmol/L). If the serum glucose falls below 250 mg/dL (13.9 mmol/L) before resolution of the ketoacidosis, increase the concentration of dextrose in the intravenous fluid up to 10 to 12.5 percent. (See 'Insulin' above.) After the initial fluid bolus, ongoing volume repletion should initially consist of isotonic saline (normal saline or lactated Ringers) at a rate that should not exceed 1.5 to 2 times maintenance and should not include urinary losses. The rate of intravenous fluid administration should be calculated to fully replete the patient's estimated fluid deficit over 48 hours. (See 'Subsequent fluid administration' above.)

After the first 4 to 6 hours of treatment, the sodium concentration may be reduced to not less than one-half isotonic saline, provided that the patient's circulatory status is stable and the serum sodium concentration is rising appropriately as the serum glucose concentration falls. The choice between one-half isotonic and isotonic saline depends on the patient's circulatory status and the rise in serum sodium as the hyperglycemia is corrected. The goal should be a slow decrease in serum osmolality to normal. (See 'Subsequent fluid administration' above and 'Serum sodium' above.)

Initial sodium concentrations will depend on the degree of net sodium and water losses prior to hospitalization, the degree of hyperglycemia and the degree of lipemia (pseudohyponatremia). With the correction of hyperglycemia, serum sodium concentrations should rise (and should be permitted to rise). Note that an increase in free water intake that attenuates or prevents the expected rise in serum sodium concentration may increase the risk of cerebral edema. During rehydration, the clinician should serially assess the serum sodium and employ anticipatory planning in fluid replacement. (See 'Serum sodium' above.)

All patients with DKA require potassium repletion, but the timing varies with serum potassium concentration. (See 'Serum potassium' above.) If the patient is normokalemic, potassium replacement should be given with the start of insulin therapy; starting concentrations of 40 meq/L KCl are reasonable. If the patient is hypokalemic, potassium replacement should be started immediately, as insulin will further reduce the serum potassium. The insulin infusion should be delayed and/or given at a reduced rate until the serum potassium level is in the normal range. One approach is to give potassium 40 mEq/L (40 mmol/L) added to normal saline and infused at a rate of 10 mL/kg over the first hour.

If the patient is hyperkalemic, potassium replacement should be initiated when the serum potassium falls to normal and after documentation of urine production. We recommend NOT treating with bicarbonate (Grade 1B). However, a few highly selected patients may benefit from cautious alkali therapy, including those with severe acidosis (arterial pH <6.90), particularly those with impairment of cardiac contractility or life-threatening hyperkalemia [9,10]. (See 'Metabolic acidosis' above.)