Diabetic ketoacidosis (DKA) occurs in a setting of absolute or relative insulin deficiency. Estimated annual incidence is 4 to 8 episodes per 1000 patients with diabetes. Specific clinical settings should generate a high index of suspicion for the disorder. Interruptions of normal insulin delivery due to purposeful reduction in insulin dosage or interference with the delivery system (eg, kinking in pump tubing) are frequent precipitating events, as are reduced insulin sensitivity in the setting of systemic infection, myocardial infarction, burns, trauma, or pregnancy. In a significant percentage of patients, DKA is the presenting feature of diabetes. In these instances, clinical suspicion and accurate interpretation of the initial laboratory studies will usually lead to the correct diagnosis. Measurement of HbA1sb>C (hemoglobin A<sb>1sb>C) levels may help in assessing the chronicity of the diabetes. Mortality in DKA is less than 5% in experienced centers. Prognosis worsens at the extremes of age and in the presence of coma or hypotension.
DKA is characterized metabolically by two prominent features: hyperglycemia and ketoacidosis (see Chapter 17). Patients with DKA present with evidence of volume contraction (eg, dry mucous membranes, thirst, orthostatic hypotension) and labored breathing (Kussmaul respiration) related to the underlying acidosis. The breath often has a fruity odor, reflecting the presence of acetone. Patients may have abdominal pain mimicking an acute abdomen, nausea, and vomiting. The latter symptoms may be related to elevated gastrointestinal prostaglandins that accrue in the presence of insulin deficiency. Presentation may be dominated by symptoms of the precipitating illness (eg, urinary tract infection, pneumonia, or myocardial infarction).
Plasma glucose levels are elevated, usually to over 250 mg/dL. This reflects impairment in glucose utilization (see discussed earlier), increased gluconeogenesis and glycogenolysis, and reduced renal clearance of glucose in the setting of decreased glomerular filtration rate (GFR). Osmotic diuresis related to glucose excretion results in reduction in intravascular volume and depletion of total body water, sodium, potassium, phosphate, and magnesium. In general, the relative depletion of water is roughly twice that of the solutes it contains. Hypertonicity in the extracellular fluid compartment, although typically not as severe as that seen in hyperosmotic nonketotic coma (see later), can be significant. Calculated plasma osmolalities greater than 340 mOsm/kg are associated with coma. Plasma osmolality—rather than acidemia—correlates most closely with state of consciousness in DKA (Figure 24–3).
A: Relationship between state of consciousness and blood pH in patients with diabetic ketoacidosis. B: Relationship between state of consciousness and plasma osmolality in diabetic ketoacidosis. Note that the state of consciousness correlates with plasma osmolality rather than blood pH.
(Reproduced, with permission, from Fulop M, et al. Ketotic hyperosmolar coma. Lancet. 1973;2:635.)
Arterial blood pH is low and, in the absence of coexistent respiratory disease, is partially compensated by a reduction in Pco2. The acidosis is metabolic in origin and accompanied by an anion gap that is calculated by subtracting the combined concentrations of chloride and bicarbonate from serum sodium concentration. Anion gaps greater than 12 mEq/L are considered abnormal. Keto acids account for most of the unmeasured anions that generate the abnormal gap, although under conditions of extreme volume contraction and hypoperfusion, lactate accumulation may also contribute. Levels of serum and urinary ketones (measured using the nitroprusside reagent) are typically high in DKA. It should be recalled, however, that this reagent reacts strongly only with acetoacetate, less strongly with acetone (which is not a keto acid and does not contribute to the anion gap), and not at all with β-hydroxybutyrate. Thus, paradoxically, the most extreme levels of ketoacidosis may be accompanied by relatively modest levels of ketones measured by this method. As a corollary of this, resolution of severe DKA may be linked to transient increases in measurable ketone levels as β-hydroxybutyrate is converted to the more readily detectable acetoacetate.
Serum sodium levels may be high, normal, or low, but in all instances total body sodium is depressed. Estimates of depletion range from 7 to 10 mEq/kg body weight. As blood glucose levels rise in DKA, they create an osmotic gradient that draws water, as well as intracellular solutes, into the extracellular space. This results in moderate hyponatremia, which can be corrected to account for the dilutional effect of the transmembrane flux of water by adding 1.6 mEq/L (more recent studies suggest that the correction factor is closer to 2.4 mEq/L) to the sodium concentration for every 100 mg/dL increment in plasma glucose above a basal concentration of 100 mg/dL (see below). The decrease in serum sodium partially offsets the increase in tonicity that accompanies the elevation in plasma glucose. This results in a net increase in plasma osmolality of 2 mOsm/kg H2O per 100 mg/dL elevation in plasma glucose:
Total body potassium levels are also severely depleted in DKA to an average of 5 to 7 mEq/kg body weight. This results from a number of factors, including exchange of intracellular potassium for extracellular hydrogen ion, impaired movement of K+ into cells in the insulinopenic state, increased urinary potassium excretion secondary to the osmotic diuresis and, in those instances where intravascular volume contraction is present, secondary hyperaldosteronism. Serum potassium levels may be high, normal, or low depending on the severity and duration of DKA, the status of extracellular fluid volume, and the adequacy of renal perfusion and excretory function. A low serum potassium at presentation generally indicates severe potassium deficiency and, in the presence of adequate renal function, is an indication for early and aggressive repletion (see below). Potassium depletion can result in muscle weakness and cardiac arrhythmias, including ventricular fibrillation.
The H+ excess in DKA titrates endogenous buffer systems including serum bicarbonate, resulting in reduction in concentrations of the latter. Chloride levels may also be low, reflecting the osmotic diuresis alluded to above. Ketone bodies have been estimated to account for one-third to one-half of the osmotic diuresis seen in DKA. Electrolyte depletion is further aggravated by the obligate cation (eg, sodium) excretion required to maintain electrical neutrality. In patients who have maintained adequate hydration during the development of DKA or in those who are aggressively resuscitated with normal saline, chloride levels may be elevated and the anion gap narrowed. This reflects the enhanced clearance of the keto anions in the kidney, converting the system from an anion gap acidosis to a hyperchloremic, nongap acidosis (ie, the hydrogen ion excess persists despite clearance of the anion). Because the excreted keto anions represent a lost source of bicarbonate regeneration, correction of the hyperchloremic acidosis may proceed slowly. The hyperchloremic acidosis is inconsequential from a clinical standpoint.
Total body magnesium and phosphate levels are also depleted by the osmotic diuresis in DKA. Phosphate depletion is amplified by diffusion of the anion from the intracellular to the extracellular compartment in the absence of insulin. Phosphate depletion can result in muscle weakness, rhabdomyolysis, hemolytic anemia, respiratory distress, and altered tissue oxygenation (due to reduction in 2,3-diphosphoglycerate levels in the red blood cell).
Treatment of DKA is focused on two major objectives. The first is restoration of normal tonicity, intravascular volume, and solute homeostasis. The second is correction of the insulinopenic state with suppression of counterregulatory hormone secretion, glucose production, and ketogenesis and improved utilization of glucose in target tissues. The steps outlined in Table 24–7 provide a general approach to the management of this disorder.
Table 24–7 Management of Diabetic Ketoacidosis. |Favorite Table|Download (.pdf)
Table 24–7 Management of Diabetic Ketoacidosis.
|(1) 1-2 L of normal saline over the first hour. Repeat if clinically significant volume contraction persists after the first hour.|
|(2) Change to half-normal saline, 500-1000 mL/h, depending on volume status. Continue for about 4 h. Decrease rate to 250 mL/h as intravascular volume returns to normal.|
|(3) Convert fluids to D5W when plasma glucose falls to 250 mg/dL.|
|(1) Administer 10-20 U of regular insulin IV.|
|(2) Mix 50 U of regular insulin in 500 mL of normal saline (1 U/10 mL). Discard first 50 mL of infusion to accommodate insulin binding to tubing. Administer through piggyback line along with parenteral fluids at a rate of 0.1 U/kg/h.|
|(3) Double the infusion rate after 2 h if there is no improvement in plasma glucose levels.|
|(1) Administer supplemental potassium chloride once renal function is established; provide 20 mEq/L of fluids for patients who are initially normokalemic, 40 mEq/L for those who are hypokalemic at presentation. In the latter case, hold insulin until serum potassium levels begin to increase.|
|(2) Gauge subsequent replacement based on serum K+ measurements at 2-h intervals.|
|(1) Sodium bicarbonate only for patients with blood pH less than 7.0.|
|(2) Add 1 ampule of sodium bicarbonate (44 mEq) to 500 mL of D5W or half-normal saline. Administer over 1 h.|
Because depletion of intracellular and extracellular fluids may be severe in DKA (typically in the range of 5-10 L), early and aggressive resuscitation with fluids is mandatory. This is usually initiated with administration of 1 to 2 L of isotonic normal saline (0.9% NaCl) over the first hour of therapy. As intravascular volume is restored, renal perfusion increases, with a consequent increase in renal clearance of glucose and a fall in plasma glucose levels. If volume contraction is severe, a second liter of normal saline can be administered. If not, half-normal saline (0.45% NaCl) can be initiated at a rate of 250 to 500 mL/h depending on intravascular volume status. Because water is typically lost in excess of solute in DKA, half-normal saline addresses both the volume depletion and the hypertonicity. It has been suggested that approximately half of the total fluid deficit should be corrected within the first 5 hours of therapy. Half-normal saline can be continued until intravascular volume has been restored or plasma glucose levels fall to 250 mg/dL, at which point D5W should be started. The latter maneuver reduces the likelihood of insulin-induced hypoglycemia and avoids the theoretical complication of cerebral edema due to osmotically induced fluid shifts from plasma into the central nervous system. This complication is, in fact, seen rarely in adults and uncommonly in children with DKA. It is important to account for ongoing urinary volume and electrolyte losses in assessing fluid requirements.
Once fluid resuscitation has been initiated, insulin should be administered. Only short-acting insulin should be used. A number of different insulin regimens have demonstrated efficacy in the treatment of DKA; however, a commonly used regimen includes a loading dose (10-20 U) of regular insulin intravenously followed by a continuous infusion at a rate of 0.1 U/kg/h. The need for a loading dose is controversial and may not be required in the majority of cases. It is not recommended for children with ketoacidosis. If intravenous access is problematic, maintenance insulin can be given intramuscularly (0.1 U/kg/h). This regimen provides plasma insulin levels in a physiologic range (100-150 mU/mL) with minimal risk of hypoglycemia or hypokalemia. It restores plasma glucose levels at rates equivalent to those obtained with regimens using higher insulin doses. Plasma glucose levels should fall at a rate of 50 to 100 mg/dL/h. Failure to achieve this end point over a 2-hour period should lead to doubling of the infusion rate with reevaluation an hour later. When plasma glucose concentrations reach 250 mg/dL, D5W is begun to prevent hypoglycemia (see above). Some diabetologists recommend a coincident reduction in insulin dose (to 0.05-0.1 U/kg/h) at this point. The insulin infusion is continued to suppress ketogenesis and allow restoration of normal acid-base balance.
As noted above, total body potassium stores are depleted in DKA (∼3-4 mEq/kg), and plasma K+ levels fall with treatment. Repletion of K+ is almost always indicated in management of DKA (one notable exception being DKA that occurs in the setting of chronic renal insufficiency); however, the timing of repletion varies as a function of the plasma K+ level. If the initial K+ level is less than 4 mEq/L, K+ depletion is severe, and repletion should begin with the first administration of parenteral fluids if renal function is adequate. Twenty milliequivalents of potassium chloride can be added to the first liter of normal saline if the serum K+ is in the 3.5 to 4 mEq/L range; 40 mEq should be added for K+ levels less than 3.5 mEq/L. Particular attention should be devoted to patients in this latter state, because K+ levels may plummet to very low levels with initiation of insulin therapy. To avoid this, insulin therapy should be postponed in this group until K+ repletion has begun and serum K+ levels are on the rise. The general goal of therapy should be to keep the K+ in a near-normal range. This may require several hundred milliequivalents of potassium chloride administered over several days.
The administration of bicarbonate in the setting of DKA has been controversial. Acidosis, in addition to increasing ventilatory work (Kussmaul respiration), may also suppress cardiac contractile function. Therefore, restoration of normal pH would seem to make sense in the setting of DKA. However, there is considerable risk associated with the use of sodium bicarbonate in this setting, including paradoxical acidification of the central nervous system due to the selective diffusion of CO2 versus HCO3 across the blood–brain barrier and an increase in intracellular acidosis, which may worsen rather than ameliorate cardiac function. Volume overload related to the high tonicity (44.6-50 mEq/50 mL) of the bicarbonate solution, hypokalemia resulting from overly rapid correction of the acidosis, hypernatremia, and rebound alkalosis are also potential complications of bicarbonate therapy. In general, pH of 7.0 or greater is not life-threatening to the average patient with DKA and will resolve with appropriate volume expansion and insulin therapy. For pH less than 7.0, many clinicians would argue for a limited administration of sodium bicarbonate. If bicarbonate is used, careful patient monitoring looking for alterations in mental status or cardiac decompensation is indicated. The goal of therapy should be to maintain pH greater than 7.0, not to return pH to normal.
Similarly, phosphate administration, once considered a key component in the management of DKA (estimated deficit ∼5-7 mmol/kg), has come under closer scrutiny. Phosphate depletion definitely occurs in DKA for the reasons outlined above, and in the past repletion of phosphate (much of it as potassium phosphate salts) had been advocated to forestall the development of muscle weakness and hemolysis and to promote tissue oxygenation through generation of 2,3-diphosphoglycerate in erythrocytes. However, the administration of phosphate salts has been associated with the development of hypocalcemia and deposition of calcium phosphate precipitates in soft tissues, including the vasculature. Thus, in general, parenteral phosphate repletion is not routinely provided for patients with DKA unless plasma phosphate falls to very low levels (<1 mmol/L). In this case, 2 mL of a mixture of KH2PO4 and K2HPO4 solution, containing 3 mmol of elemental phosphorus and 4 mEq of potassium, may be added to 1 L of fluids and introduced over 6 to 8 hours. In no instance should all K+ repletion be in the form of potassium phosphate salts. In general, renewal of food ingestion and insulin therapy complete restoration of total body phosphate stores and return plasma phosphate levels to normal over a period of several days.
In pediatric patients (age <20 years) the need for volume expansion needs to be weighed against the potential risk of cerebral edema secondary to aggressive fluid administration, although this is controversial (see below). A recent recommendation suggests 10 to 20 mL/kg/h of normal saline during the first 1 to 2 hours with limitation of total fluid administration to 50 mL/kg over the first 4 hours. The remaining fluid deficit is corrected over the next 48 hours. Normal saline or half-normal saline (depending on serum Na+ levels) usually accomplishes this with a rate of 5 mL/kg/h. The decrease in serum osmolality should not exceed 3 mOsm/kg H2O/h. An insulin bolus prior to initiating the insulin infusion (0.1 U/kg/h) is usually not required in children.
Finally, it is necessary to actively seek out and treat the precipitants of DKA when they are identified. This includes appropriate cultures of urine and blood (and cerebrospinal fluid, if indicated) and empiric antibiotic therapy directed against the most likely pathogenic organisms (pending the results of the cultures). The presence of fever is typically a good marker for infection or other inflammatory process because it is not a feature of DKA per se. Elevated white blood cell counts, on the other hand, are frequently seen with DKA alone. Hyper-amylasemia is common but rarely reflects pancreatitis—the amylase is usually of salivary origin. Other precipitants should also be sought. Myocardial infarction, which is often clinically silent in diabetic patients, is an uncommon but life-threatening precipitant of DKA in patients with established diabetes.
Aggressive resuscitation with isotonic or hypotonic fluids is a theoretical but uncommon cause of fluid overload during management of DKA. Careful attention to the cardiovascular examination, chest x-ray, and urine output should aid in preventing this complication.
Hypoglycemia is relatively rare in the current era given the low doses of insulin used in management and appropriate initiation of glucose-containing fluids as plasma glucose levels fall below 250 mg/dL.
Cerebral edema due to rapid correction of plasma hypertonicity has usually been reported with plasma glucose levels below 250 mg/dL. Clinically significant cerebral edema is relatively uncommon in adult patients. Milder forms of cerebral edema have been noted in many patients being treated for DKA but have not been strongly correlated with changes in extracellular tonicity. At present, in a symptomatic adult patient, it would appear prudent to treat hypertonicity exceeding 340 mOsm/kg aggressively with hypotonic fluids to avoid complications related to plasma hyperviscosity. Further correction from that point to normal plasma osmolality (about 285 mOsm/kg) can probably be accomplished in slower fashion over several days. Cerebral edema occurs in 1% to 2% of children with DKA, frequently with devastating results. Approximately one-third of children with clinically significant cerebral edema die during the acute illness, and another third sustain permanent neurologic impairment. The predilection for young children may reflect, in part, immaturity of the autoregulatory mechanism that governs cerebral blood flow. There is increased risk in children less than 5 years of age, and those with low Pco2 or high blood urea nitrogen at presentation. Cerebral edema in children may be associated with high initial rates of fluid resuscitation (>4 L/m22/d) and rapid falls in plasma sodium (or corrected sodium) concentration, although it can occur in clinical settings without an apparent cause, and mild degrees of cerebral edema have been noted in DKA even prior to initiation of therapy. In the absence of definitive trial data to guide therapy, lower rates of fluid administration (<2.5 L/m2/d) with volume resuscitation spread over a longer time interval would seem appropriate if the clinical situation permits. When signs of cerebral edema appear—deterioration in level of consciousness, focal neurologic signs, hypotension or bradycardia, sudden decline in urine output after an initial period of apparent recovery following treatment for DKA—fluid administration should be reduced and mannitol (0.2-1 g/kg given intravenously over 30 minutes) should be administered with repetition at hourly intervals based on response. CT or MRI scan of the brain can be done once therapy has been initiated to confirm the diagnosis. Hyperventilation has not been shown to alter the course of this complication once it develops.
Patients with DKA are also prone to develop acute respiratory distress syndrome, presumably reflecting the sequelae of a damaged pulmonary endothelium and elevated capillary hydrostatic pressures following fluid resuscitation. Patients who present with rales at the time of initial diagnosis may be at higher risk for the development of this complication. Patients may also be at increased risk for development of pancreatitis as well as systemic infection, including fungal infections (eg, mucormycosis).
Abdominal pain and gastric stasis seen in DKA may put a semi-stuporous patient at risk for aspiration. Up to 25% of patients with DKA have emesis that may be guaiac positive. The latter finding appears to result from hemorrhagic gastritis. Patients who are believed to be at risk with regard to airway protection should have a nasogastric tube in place for evacuation of stomach contents.
Finally, patients with DKA are at risk for recurrence of the disorder if insulin is withdrawn prematurely. The current infusion protocols, because they raise plasma insulin only to physiologic levels, have a very short half-life for control of blood glucose and ketogenesis. Premature cessation of insulin therapy before depot insulin (eg, NPH or glargine insulin) can exert its effect may allow the patient to regress into ketoacidosis. To preclude this possibility, subcutaneous regular and intermediate-acting insulin should be provided on the morning when feeding is to be resumed. The insulin drip should be continued for 1 hour following this injection to provide coverage until the depot insulin becomes effective.