INTRODUCTION —
Diabetic ketoacidosis (DKA) and hyperosmolar hyperglycemic state (HHS, also known as hyperosmotic hyperglycemic nonketotic state [HHNK]) are two of the most serious acute complications of diabetes. DKA and HHS often occur together (mixed DKA/HHS). Ketoacidosis with mild hyperglycemia or even normal blood glucose ("normoglycemic" DKA) has become more common with the increased use of sodium-glucose cotransporter 2 [SGLT2] inhibitors.
The treatment of DKA in adults will be reviewed here. The epidemiology, pathogenesis, clinical features, evaluation, and diagnosis of DKA and HHS are discussed separately, as is the treatment of HHS in adults. DKA in children is also reviewed separately.
●(See "Hyperosmolar hyperglycemic state in adults: Treatment".)
●(See "Diabetic ketoacidosis in children: Clinical features and diagnosis".)
●(See "Diabetic ketoacidosis in children: Treatment and complications".)
DISTINGUISHING DKA FROM HHS —
Diabetic ketoacidosis (DKA) and hyperglycemic hyperosmolar state (HHS) differ clinically according to the presence of ketoacidosis and, usually, the degree of hyperglycemia (table 1) [1-3]. However, approximately one-third of patients have a mixed presentation of DKA and HHS.
●In DKA, metabolic acidosis, ketonemia, and hyperglycemia are typically the major findings. The serum glucose concentration is generally below 800 mg/dL (44.4 mmol/L) and often in the 350 to 500 mg/dL (19.4 to 27.8 mmol/L) range [2-4]. However, serum glucose concentrations may exceed 900 mg/dL (50 mmol/L) in patients with DKA, usually in association with coma [3,5], or may be normal or minimally elevated (<200 mg/dL [11.1 mmol/L]) in patients with normoglycemic DKA. Normoglycemic DKA occurs more often in patients with poor oral intake, those treated with insulin prior to arrival in the emergency department, pregnant women, those who use sodium-glucose cotransporter 2 [SGLT2] inhibitors, and insulin pump-treated patients in whom insulin delivery is interrupted due to catheter or pump failure.
●In HHS, ketoacid accumulation is mild or absent, the serum glucose concentration frequently exceeds 1000 mg/dL (56 mmol/L), the plasma osmolality may reach 380 mOsmol/kg, and neurologic abnormalities are frequently present (including coma in 25 to 50 percent of cases) [2,4,6].
The typical total body deficits of water and electrolytes in DKA and HHS are compared in the table (table 2). (See "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Clinical features, evaluation, and diagnosis", section on 'Serum glucose' and "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Clinical features, evaluation, and diagnosis", section on 'Diagnostic criteria'.)
TREATMENT
Overview and protocols — The treatment of diabetic ketoacidosis (DKA) (algorithm 1) includes correcting the fluid and electrolyte abnormalities that are typically present (hyperosmolality, hypovolemia, metabolic acidosis, and potassium depletion) and administering insulin [7].
●Determine the site of care – For the vast majority of patients, treatment of DKA should occur in the emergency room or inpatient setting where volume and electrolyte repletion and insulin therapy can be administered safely. Outpatient management may be a reasonable option for selected patients with mild DKA (eg, hyperglycemia and positive ketones without nausea or vomiting) who have a known diagnosis of diabetes and a clear and readily reversible cause of DKA (eg, occlusion of insulin pump tubing). Such patients must have adequate insulin and glucose testing supplies; detailed instructions for insulin administration, fluid intake, and glucose monitoring; and close telephone follow-up with clinic staff.
●Assess vital signs, cardiorespiratory status, and mental status – Initial evaluation should include assessment of the patient's cardiovascular, respiratory, and mental status. For patients who present with stupor or coma, a Glasgow Coma Scale score should be determined (table 3). In patients with a score ≤8, endotracheal intubation is usually required for airway protection. (See "Stupor and coma in adults", section on 'Management'.)
●Treat the volume depletion and electrolyte abnormalities – The first step in the treatment of DKA is infusion of isotonic fluid (saline or buffered crystalloid) to expand extracellular volume and stabilize cardiovascular status (table 4). Volume expansion also increases insulin responsiveness by lowering the plasma osmolality, reducing vasoconstriction and improving perfusion, and reducing stress hormone levels [8,9]. The next step is correction of the potassium deficit (if present). Potassium repletion may inform the choice of fluid replacement; the osmotic effect of potassium repletion must be considered since potassium is as osmotically active as sodium. (See 'Fluid replacement' below and 'Potassium replacement' below.)
●Administer insulin – Low-dose intravenous (IV) insulin should be administered to all patients with moderate to severe DKA who have a serum potassium ≥3.5 mEq/L. If the serum potassium is <3.5 mEq/L, insulin therapy should be withheld until potassium replacement has increased the serum potassium concentration above 3.5 mEq/L. This delay in insulin initiation is necessary because insulin will drive potassium into cells and worsen hypokalemia, which could trigger cardiac arrhythmias. IV regular insulin and rapid-acting insulin analogs are equally effective in treating DKA. Subcutaneous rather than IV insulin may be used in individuals with uncomplicated mild to moderate DKA. (See 'Insulin' below.)
Managing DKA requires frequent clinical and laboratory monitoring and the identification and treatment of any precipitating events, including infection. Sodium-glucose cotransporter 2 (SGLT2) inhibitors, which can precipitate DKA, should be discontinued. Permanent discontinuation of the SGLT2 inhibitor should be strongly considered. (See 'Monitoring' below and "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Clinical features, evaluation, and diagnosis", section on 'Precipitating factors'.)
Our approach outlined below is based upon clinical experience and is largely in agreement with consensus guidelines from the American Diabetes Association (ADA), Joint British Diabetes Societies for Inpatient Care (JBDS), American Association of Clinical Endocrinology (AACE), and Diabetes Technology Society (DTS) for the management of hyperglycemic crises (algorithm 1) [4,10,11].
Fluid replacement — In patients with DKA, we recommend IV electrolyte and fluid replacement to correct both hypovolemia and hyperosmolality.
Initial choice of fluid
●IV isotonic fluid – Isotonic saline (0.9 percent sodium chloride [NaCl]) or isotonic buffered crystalloid (eg, Lactated Ringer) should be used for volume repletion. Buffered crystalloid may reduce time to DKA resolution and reduces the degree of hyperchloremic, non-anion gap metabolic acidosis that often results from high volume isotone saline administration.
No trials have directly compared saline with buffered crystalloid specifically in patients with DKA. In a subgroup analysis of two cluster-randomized trials evaluating choice of isotonic fluid in an emergency or critical care setting, adults with DKA who received buffered crystalloids (n = 94) had a shorter time to DKA resolution compared with those who received saline (n = 78; median 13 versus 16.9 hours, respectively) [12]. A subsequent cohort study and meta-analysis both reached similar conclusions [13,14].
●Adding dextrose to IV fluids – For patients who present with an initial serum glucose <250 mg/dL [13.9 mmol/L]), dextrose is added to IV fluids at the initiation of therapy. Such patients require both insulin and glucose to treat the ketoacidosis and prevent hypoglycemia, respectively.
Initial rate of fluid administration — The optimal rate of initial isotonic saline infusion depends on the volume status of the patient:
●In patients with hypovolemic shock, isotonic fluid should initially be infused as quickly as possible. (See "Treatment of severe hypovolemia or hypovolemic shock in adults".)
●In hypovolemic patients without shock or heart or kidney failure, isotonic fluid is infused at a rate of 15 to 20 mL/kg lean body weight per hour (approximately 1000 mL/hour in an average-sized person) for the first few hours, with a maximum of <50 mL/kg in the first four hours (algorithm 1) [1].
●In patients with mild hypovolemia or euvolemia, isotonic fluid is infused at a lower rate, guided by clinical assessment.
The goal is to correct estimated deficits within the first 24 to 48 hours. Osmolality should not be reduced too rapidly, because this may precipitate cerebral edema. Adequacy of fluid replacement is judged by frequent hemodynamic and laboratory monitoring. (See 'Monitoring' below and 'Cerebral edema' below.)
Subsequent fluid management
●Switching to hypotonic fluid – After the second or third hour of fluid administration, optimal fluid replacement depends upon the volume and hydration status, serum electrolyte levels, and urine output. The most appropriate IV fluid composition is determined by the sodium concentration "corrected" for the degree of hyperglycemia. The "corrected" sodium concentration can be approximated by adding 2 mEq/L to the plasma sodium concentration for each 100 mg/dL (5.5 mmol/L) increase above 100 mg/dL (5.5 mmol/L) (calculator 1).
If the "corrected" serum sodium concentration is [1]:
•<135 mEq/L, isotonic fluid infusion should be continued at a rate of approximately 250 to 500 mL/hour
•≥135 mEq/L, isotonic fluid infusion is generally switched to one-half isotonic saline at a rate of 250 to 500 mL/hour to provide electrolyte-free water
●Adding dextrose to IV fluids – For patients who present with hyperglycemic DKA, we add dextrose (5 to 10 percent) to the saline solution when the serum glucose declines to <250 mg/dL (13.9 mmol/L) (algorithm 1).
Special considerations
High-dose potassium replacement — Major potassium replacement at rates >20 to 30 mEq/hr is rarely required. If such high infusion rates are needed, the potassium chloride (KCl) should be added to one-half isotonic (0.45 percent NaCl) rather than isotonic (0.9 percent NaCl) saline. Potassium salts have an osmotic effect equivalent to sodium salts, and adding potassium to isotonic fluids generates a hypertonic solution. (See 'Osmotic effect of potassium salts' below.)
Reduced kidney or cardiac function — In patients with reduced kidney or cardiac function, more frequent monitoring must be performed to avoid iatrogenic fluid overload [4,9,15]. Rather than continuous isotonic fluid infusion, such patients may be managed with repeated, small-volume fluid boluses (eg, 250 mL) [4].
Potassium replacement — Potassium replacement is initiated immediately if the serum potassium is ≤5.0 mEq/L, provided urine output is adequate (urine output approximately ≥50 mL/hour or 0.5 mL/kg/hour) (algorithm 1). Almost all patients with DKA have a substantial potassium deficit, usually due to urinary losses generated by the glucose-driven osmotic diuresis and secondary hyperaldosteronism. Despite the total body potassium deficit, the serum potassium concentration is usually normal or, in approximately one-third of cases, elevated at presentation. This is largely due to insulin deficiency, hyperosmolality, and acidosis, which cause potassium movement out of the cells [16].
Initial potassium replacement
●Serum potassium <3.5 mEq/L – If the initial serum potassium is <3.5 mEq/L, insulin should not be administered until the potassium has been raised above this threshold. IV potassium chloride (KCl; 10 to 20 mEq/hour, which usually requires 10 to 20 mEq/L added to each liter of IV fluid) should be given. Patients with marked hypokalemia may require more aggressive potassium replacement (eg, 30 mEq/hour, potentially requiring central venous access) to raise the serum potassium concentration above 3.5 mEq/L [17-19]. If needed, potassium infusion rates >20 to 30 mEq/hr are highly irritating to peripheral veins and generally must be infused into a large central vein or via multiple peripheral veins. Such high infusion rates also usually require cardiac rhythm monitoring.
●Serum potassium 3.5 to 5.0 mEq/L – If the initial serum potassium is 3.5 to 5.0 mEq/L, IV KCl (10 to 20 mEq) is added to each liter of IV fluid. Adjust potassium replacement to maintain the serum potassium concentration in the range of 4 to 5 mEq/L.
●Serum potassium >5.0 mEq/L – If the initial serum potassium concentration is >5.0 mEq/L, then potassium replacement should be delayed until the serum concentration has fallen below this level. Serum potassium should be monitored every two hours.
Insulin administration rapidly reverses the altered potassium distribution and can result in a dramatic fall in the serum potassium concentration, despite potassium replacement [17,18]. However, potassium replacement must be done cautiously, particularly if kidney function is decreased and/or urine output remains below 50 mL/hour. Careful monitoring of the serum potassium is essential. (See 'Monitoring' below and "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Epidemiology and pathogenesis", section on 'Potassium'.)
Osmotic effect of potassium salts — Potassium salts added to IV fluids have the same osmotic effect as sodium salts, and this should be considered when determining the potential impact of IV fluid infusion on osmolality. As an example, 40 mEq of KCl added to 1 L of fluid generates 80 mOsmol/L of electrolyte osmolarity. The addition of 40 mEq of potassium to 1 L of one-half isotonic saline creates a solution with an osmolarity of 234 mOsmol/L (77 mEq NaCl and 40 mEq KCl), which is osmotically equal to three-quarters isotonic saline. (The osmolarity of isotonic saline is 308 mOsmol/L.) If 40 mEq of KCl is added to isotonic saline, the final osmolarity will be approximately 388 mOsmol/L. However, KCl will not have the same extracellular fluid (ECF) expansion effect as NaCl, because most of the potassium will shift into cells very rapidly. (See "Maintenance and replacement fluid therapy in adults", section on 'Choice of replacement fluid'.)
Insulin
Initiating insulin treatment
●Timing of insulin initiation – We recommend initiating insulin treatment immediately in all patients with DKA who have a serum potassium ≥3.5 mmol/L. The only indication for delaying the initiation of insulin therapy is a serum potassium <3.5 mEq/L since insulin will worsen hypokalemia by driving potassium into cells. Patients with an initial serum potassium <3.5 mEq/L should receive fluid and potassium replacement prior to treatment with insulin. Insulin therapy should be withheld until the serum potassium is >3.5 mEq/L to avoid complications such as cardiac arrhythmias, cardiac arrest, and respiratory muscle weakness [1,17,18]. (See 'Fluid replacement' above and 'Potassium replacement' above.)
●Effects on glucose and ketoacidemia – Insulin therapy lowers the serum glucose concentration (primarily by decreasing hepatic glucose production and also by enhancing peripheral utilization [20]), diminishes ketone production (by reducing both lipolysis and glucagon secretion), and may augment ketone utilization. Inhibition of lipolysis and ketogenesis requires a much lower level of insulin than that required to reduce the serum glucose concentration. Therefore, if the administered dose of insulin is reducing the glucose concentration, it should be sufficient to stop ketone generation [20-22]. (See "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Epidemiology and pathogenesis", section on 'Pathogenesis'.)
Moderate to severe DKA
Intravenous insulin — In moderate to severe DKA, treatment can be initiated with a fixed-rate continuous infusion of regular insulin of 0.1 units/kg per hour (equivalent to 7 units/hour in a 70-kg patient) (algorithm 1) [21,23-26]. Alternatively, a variable rate insulin infusion may be administered using a nurse-driven protocol. IV regular insulin and rapid-acting insulin analogs are equally effective in treating DKA [27]. The choice of IV insulin is based on institutional preferences, clinician experience, and cost concerns. We generally prefer regular insulin, rather than rapid-acting or ultra rapid-acting insulin analogs, due to its similar availability when given intravenously and much lower cost. (See "General principles of insulin therapy in diabetes mellitus", section on 'Human insulins'.)
●Insulin bolus if infusion is delayed – If initiating the insulin infusion is delayed (eg, due to difficulty with venous access), an IV or intramuscular (IM) bolus of regular insulin (0.1 units/kg body weight) should be administered to raise insulin levels rapidly, followed by continuous infusion.
●Role of long-acting, subcutaneous insulin – For acute management of DKA, expert approaches differ in the use of long-acting, subcutaneous basal insulin. Some experts avoid subcutaneous basal insulin during IV insulin infusion due to possible increased risk of hypoglycemia and/or hypokalemia. However, other experts administer long-acting insulin during insulin infusion based on evidence that it accelerates DKA resolution and reduces length of hospital stay [4]. All patients require subcutaneous basal insulin prior to the discontinuation of IV insulin. (See 'Converting to subcutaneous insulin' below.)
●Expected glucose response – Regular insulin infusion should decrease the serum glucose concentration by approximately 50 to 70 mg/dL (2.8 to 3.9 mmol/L) per hour [20,24-26]. Higher doses do not generally produce a more prominent glucose-lowering effect, probably because the insulin receptors are fully saturated by the lower doses [23]. However, if the serum glucose does not fall by at least 50 to 70 mg/dL (2.8 to 3.9 mmol/L) from the initial value in the first hour, check the IV access to ensure that the insulin is being delivered and that no IV line filters that may bind insulin have been inserted into the line. After these possibilities are eliminated, the insulin infusion rate should be doubled every hour until a steady decline in serum glucose of this magnitude is achieved.
The fall in serum glucose is the result of both insulin activity and volume repletion. Volume repletion alone can initially reduce the serum glucose by 35 to 70 mg/dL (1.9 to 3.9 mmol/L) per hour due to the combination of ECF expansion, reduction of plasma osmolality, increased urinary losses resulting from improved kidney perfusion, and a reduction in stress hormone levels [25,28].
Insulin titration and dextrose administration — When the serum glucose is <250 mg/dL (13.9 mmol/L), add 5 to 10 percent dextrose to the IV fluid, and decrease the insulin infusion rate to 0.05 units/kg per hour (or according to the variable rate protocol) [7,9,23]. If possible, do not allow the serum glucose to fall rapidly to below 200 mg/dL (11.1 mmol/L), because this may promote the development of cerebral edema. (See 'Cerebral edema' below and "Diabetic ketoacidosis in children: Cerebral injury (cerebral edema)".)
Serum glucose should be maintained between 150 and 200 mg/dL (8.3 and 11.1 mmol/L) until resolution of DKA. (See 'Resolution criteria' below.)
Uncomplicated mild to moderate DKA
Subcutaneous insulin — Patients with uncomplicated, mild to moderate DKA (table 1) can be safely treated with subcutaneous, rapid-acting insulin analogs on a general medical floor or in the emergency department, provided adequate staffing is available to carefully monitor the patient and check capillary blood glucose with a reliable glucose meter, typically every hour. We do not recommend the use of continuous glucose monitoring (CGM) in this setting due to a lack of supporting evidence.
●Insulin bolus – An initial bolus of rapid-acting insulin 0.1 units/kg body weight should be administered.
●Initial insulin regimen – Following the initial bolus, either of the following rapid-acting insulin regimens may be used:
•0.1 units/kg given every hour
or
•0.2 units/kg given every two hours
In patients with uncomplicated DKA, IM, subcutaneous, and IV insulin therapy show similar efficacy and safety [29-33]. Subcutaneous administration is less painful than IM, and trial data support the safety of rapid-acting insulin analogs (eg, insulin lispro, aspart) given every one or two hours (algorithm 1) [29,30]. In patients with mild DKA, intermediate- or long-acting insulin can be administered at the initiation of treatment, along with rapid-acting insulin.
Insulin titration and dextrose administration — When the serum glucose is <250 mg/dL (13.9 mmol/L), do both of the following:
●Reduce the insulin dose to 0.05 units/kg every hour or 0.1 units/kg every two hours.
●Add 5 to 10 percent dextrose to the IV fluid. For patients who initially present with serum glucose <250 mg/dL, dextrose should be added immediately to IV fluids and initiated concurrently with insulin therapy.
Normoglycemic DKA (glucose <200 mg/dL [11.1 mmol/L]) — For patients who present with normoglycemic DKA, initial treatment is similar to that for patients with mild to moderate DKA. However, our approach differs as follows:
●Dextrose (5 to 10 percent) should be added immediately to IV fluids and initiated concurrently with insulin therapy.
●We do not administer a rapid-acting insulin bolus.
●The initial insulin dose is 0.05 units/kg every hour or 0.1 units/kg every two hours.
If normoglycemic DKA is associated with SGLT2 inhibitor use, the SGLT2 inhibitor should be stopped immediately. For people who develop DKA during SGLT2 inhibitor treatment, restarting the agent after DKA resolution is not recommended [4].
Method of glucose measurement — Serum glucose measurements should be done with hospital-approved bedside devices or in the chemistry laboratory and not with CGM devices. CGM devices measure interstitial rather than circulating glucose concentrations. As a result, if glucose is rapidly changing, changes in CGM-derived values may have a 10- to 15-minute delay relative to venous glucose levels. Further, CGM may be less accurate in the setting of severe hyper- or hypoglycemia, volume depletion, large volume shifts, and/or acidosis. (See "Glucose monitoring in the ambulatory management of nonpregnant adults with diabetes mellitus", section on 'CGM systems'.)
Bicarbonate and metabolic acidosis
Indications for bicarbonate (rarely administered) — Although the indications for sodium bicarbonate therapy are controversial [34], selected patients may benefit from cautious alkali therapy. They include:
●Arterial pH <7.0 – In patients with an arterial pH <7.0, decreased cardiac contractility and vasodilatation can impair tissue perfusion [35,36]. At an arterial pH ≥7.0, most experts agree that bicarbonate therapy is not necessary since insulin therapy and volume expansion will largely reverse the metabolic acidosis [37].
In patients with DKA, evidence of benefit from bicarbonate therapy is lacking [38-40], and largely theoretical concerns persist about potential harms. In a randomized trial of 21 DKA patients with an admission arterial pH between 6.90 and 7.14 (mean 7.01), bicarbonate therapy did not change morbidity or mortality [38]. However, the study was small and limited to patients with an arterial pH ≥6.90, and the rate of rise in the arterial pH and serum bicarbonate did not differ between the bicarbonate and placebo groups. No trials have been performed to evaluate bicarbonate administration in DKA with pH values <6.90.
●Serum potassium >6.4 mEq/L or hyperkalemic changes on electrocardiogram (ECG) – In patients with potentially life-threatening hyperkalemia and acidemia, bicarbonate administration may drive potassium into cells, thereby lowering the serum potassium concentration. The exact potassium level that should trigger this intervention has not been defined; we generally administer sodium bicarbonate if the potassium level is >6.4 mEq/L or if hyperkalemic changes are evident on ECG [41]. (See "Treatment and prevention of hyperkalemia in adults".)
Dose and monitoring — For patients with pH <7.0, we give 100 mmol of sodium bicarbonate in 400 mL sterile water administered over two hours [4]. If the serum potassium is <5.0 mEq/L, we add 20 mEq of KCl. When the bicarbonate concentration increases, the serum potassium may fall and more aggressive KCl replacement may be required.
The venous pH and bicarbonate concentration should be monitored every two hours, and bicarbonate doses can be repeated until the pH rises above 7.0. (See 'Monitoring' below.)
Potential risks of bicarbonate — Bicarbonate administration is also controversial because of the following, potentially harmful effects:
●The administration of alkali may slow the rate of recovery of the ketosis [42,43]. In a study of seven patients, the three patients treated with bicarbonate had a rise in serum ketoacid anion levels and a six-hour delay in resolution of ketosis [42]. Animal studies indicate that bicarbonate infusion can accelerate ketogenesis. This may be due to a "braking effect" of acidemia on organic acid generation that is attenuated by any intervention that increases systemic pH [38].
●Alkali administration can lead to post-treatment metabolic alkalosis. Insulin accelerates the metabolism of ketoacid anions and consequently generates bicarbonate; thus, insulin alone is usually sufficient to correct the metabolic acidosis. (See "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Epidemiology and pathogenesis", section on 'Anion gap metabolic acidosis'.)
Phosphate repletion (rarely needed) — We do not recommend the routine use of phosphate replacement in the treatment of DKA.
However, if severe hypophosphatemia occurs (serum phosphate concentration <1 mg/dL or 0.32 mmol/L), phosphate replacement should be administered, especially if cardiac dysfunction, hemolytic anemia, and/or respiratory depression develop [44-48]. When needed, potassium or sodium phosphate 20 to 30 mEq can be added to 1 L of IV fluid.
Although whole-body phosphate depletion is common in uncontrolled diabetes mellitus, the serum phosphate concentration may initially be normal or elevated due to movement of phosphate out of the cells [45,49]. Like potassium, phosphate depletion and hypophosphatemia may be rapidly unmasked following the institution of insulin therapy and IV volume expansion. This frequently leads to asymptomatic hypophosphatemia, which gradually resolves. (See "Hypophosphatemia: Clinical manifestations of phosphate depletion".)
Prospective, randomized trials of patients with DKA have failed to show a beneficial effect of phosphate replacement on the duration of ketoacidosis, dose of insulin required, or the rate of morbidity or mortality [50-52]. In addition, phosphate replacement may have adverse effects, such as hypocalcemia and hypomagnesemia [50,53-55].
MONITORING
Monitoring schedule — A flow sheet of laboratory values and clinical parameters allows better visualization and evaluation of the clinical picture throughout treatment of diabetic ketoacidosis (DKA) (form 1).
●Hyperglycemia and metabolic acidosis – The serum glucose should initially be measured every hour until stable, while serum electrolytes, blood urea nitrogen (BUN), phosphorus, creatinine, and venous pH should be measured every two to four hours, depending upon disease severity and the clinical response [1,7]. Serum glucose measurement should be done with hospital-approved bedside devices or in the chemistry laboratory and not with continuous glucose monitoring (CGM) devices, which may not reflect rapidly changing levels and may be less accurate in the setting of volume depletion. (See 'Method of glucose measurement' above.)
Monitoring with arterial blood gases is unnecessary during the treatment of DKA; venous pH, which is approximately 0.03 units lower than arterial pH [56], is adequate to assess the response to therapy and avoids the pain and potential complications associated with repeated arterial punctures. If blood chemistry results are promptly available, an alternative to monitoring venous pH is to monitor the serum bicarbonate concentration (to assess correction of the metabolic acidosis).
●Ketonemia
•Blood or serum beta-hydroxybutyrate (preferred when available) – Blood or serum beta-hydroxybutyrate should be measured every two hours. Where available, bedside ketone meters that measure capillary blood beta-hydroxybutyrate are an alternative to measuring serum beta-hydroxybutyrate [4,57]. These devices are increasingly available, reliable, and convenient.
•Serum anion gap – When bedside meters and serum beta-hydroxybutyrate measurement are not available, monitoring venous pH and/or the venous bicarbonate with a calculated serum anion gap is sufficient. Accumulation of ketoacid anions (the sum of beta-hydroxybutyrate and acetoacetate) increases the anion gap above its baseline, and the increment reflects the sum of their concentrations in serum. The anion gap is calculated by subtracting the major measured anions (chloride and bicarbonate) from the major measured cation (sodium). The sodium concentration used for this calculation is the concentration reported by the laboratory, not the "corrected" sodium concentration.
•Serum or urine ketones (not appropriate for monitoring) – Although assessments of urinary or serum ketone levels by the nitroprusside method can be used for the initial diagnosis of ketoacidosis, they should not be used for monitoring resolution of DKA. Nitroprusside reacts mainly with acetoacetate, to a much lesser degree with acetone (which is not an acid), but not with beta-hydroxybutyrate. A positive nitroprusside test may persist for up to 36 hours after resolution of ketoacidosis due to a positive reaction with acetone, which is slowly eliminated, mainly via the lungs [58,59]. Since acetone is not an acid, a persistent nitroprusside reaction due to acetone does not indicate ketoacidosis. In addition, active treatment of ketoacidosis shifts the reaction between beta-hydroxybutyrate and acetoacetate toward acetoacetate. This may result in an increasingly positive nitroprusside test (due to higher acetoacetate concentrations) despite an overall improvement of the ketoacidosis (figure 1) [22].
Resolution criteria — The following criteria are used to define DKA resolution:
●Blood or serum beta-hydroxybutyrate <0.6 mmol/L or normalization of the serum anion gap (≤12 mEq/L).
●Venous pH ≥7.3 or serum bicarbonate ≥18 mmol/L – In patients who received isotonic saline, monitoring for resolution of acidosis may be complicated by the evolution a hyperchloremic, non-anion gap acidosis. Hyperchloremic acidosis does not indicate unresolved DKA. (See 'Hyperchloremic acidosis' below.)
●Serum glucose <200 mg/dL (11.1 mmol/L).
CONVERTING TO SUBCUTANEOUS INSULIN —
We initiate a multiple-dose (basal-bolus), subcutaneous insulin schedule when the ketoacidosis has resolved and the patient is able to eat. (See 'Resolution criteria' above.)
Designing an insulin regimen
●Prior diagnosis of diabetes – For patients with known diabetes who were previously being treated with insulin, their pre-diabetic ketoacidosis (DKA) insulin regimen may be restarted if preadmission glycemia was near or at target. For patients treated with continuous subcutaneous insulin infusion (insulin pump), the previous basal rate can be resumed. However, if the intravenous (IV) insulin requirements are significantly higher than their usual insulin requirements, it is reasonable to increase the basal rate temporarily. For those using partially automated insulin delivery (hybrid closed-loop) systems, it is also reasonable to stay in "manual mode" for the first 24 to 48 hours while insulin requirements return to baseline. If automated insulin delivery is immediately resumed, blood glucose levels likely will be higher for the first few days.
●New-onset diabetes
•Total daily dose – In patients with new-onset type 1 diabetes who presented with DKA, an initial total daily dose (TDD) of 0.5 to 0.6 units/kg of insulin per day is reasonable, until an optimal dose is established. In patients at increased risk of hypoglycemia, an initial TDD of 0.3 units/kg may be used. (See "Management of blood glucose in adults with type 1 diabetes mellitus" and "General principles of insulin therapy in diabetes mellitus".)
•Basal insulin – Approximately 40 to 60 percent of the TDD should be given as basal insulin, either as once- or twice-daily U-100 glargine or detemir (to be discontinued in the United States in 2024), or as twice-daily intermediate-acting insulin (neutral protamine Hagedorn [NPH]). The long-acting insulin can be given either at bedtime or in the morning; the NPH is usually given as approximately two-thirds of the dose in the morning and one-third at bedtime. In this setting, we do not use degludec or ultra long-acting U-300 insulin glargine, as these require at least three to four days to reach steady state due to their long half-life [60].
•Prandial insulin – The remainder of the TDD is given as short-acting or rapid-acting insulin, divided before meals. If NPH is the basal insulin used, a mid-day (pre-lunch) rapid-acting insulin may not be necessary. Frequent glucose monitoring or preferably continuous glucose monitoring (CGM) and comprehensive diabetes education are vital after initiating a new insulin regimen in treatment-naïve patients. (See "General principles of insulin therapy in diabetes mellitus" and "Management of blood glucose in adults with type 1 diabetes mellitus", section on 'Designing an MDI insulin regimen' and "Insulin therapy in type 2 diabetes mellitus".)
Timing of initial dose — The IV insulin infusion should be continued for one to two hours after initiating subcutaneous rapid-acting insulin. The first dose of basal insulin also should be administered before IV insulin is discontinued. If short- or long-acting insulin is initiated without rapid-acting insulin, the IV insulin infusion should be continued for two to four hours after subcutaneous insulin administration. Abrupt discontinuation of IV insulin acutely reduces insulin levels and may result in recurrence of hyperglycemia and/or ketoacidosis.
Basal insulin (eg, NPH, U-100 glargine, or detemir) can be administered either at the same time as the first injection of rapid-acting insulin (eg, in the morning before breakfast) or at bedtime the previous night.
COMPLICATIONS —
Hypoglycemia and hypokalemia are the most common complications of diabetic ketoacidosis (DKA) treatment. These complications have become much less common since low-dose intravenous (IV) insulin treatment and careful monitoring of serum potassium have been implemented [61]. Hyperglycemia may recur from interruption or discontinuation of IV insulin without adequate overlap coverage with subcutaneous insulin.
Cerebral edema — Cerebral edema in uncontrolled diabetes mellitus is primarily a disease of children, and almost all affected patients are younger than 20 years old [62]. Symptoms typically emerge within 12 to 24 hours of the initiation of treatment for DKA but may exist prior to the onset of therapy. Issues related to cerebral edema in DKA, including pathogenesis, are discussed in detail separately but will be briefly reviewed here. (See "Diabetic ketoacidosis in children: Cerebral injury (cerebral edema)".)
●Clinical features — DKA-associated cerebral edema has a mortality rate of approximately 30 percent [4]. Thus, careful monitoring for changes in mental or neurologic status that would permit early identification and therapy of cerebral edema is essential. Headache is the earliest clinical manifestation, followed by lethargy and decreased arousal. Neurologic deterioration may be rapid. Seizures, incontinence, pupillary changes, bradycardia, and respiratory arrest can develop. Symptoms progress if brainstem herniation occurs, and the rate of progression may be so rapid that clinically recognizable papilledema does not develop.
●Preventive measures – The following preventive measures may reduce the risk of cerebral edema in high-risk patients:
•Gradually replacing sodium and water deficits in patients with hyperosmolarity (algorithm 1).
•Adding dextrose to the IV fluids once the serum glucose level falls to <250 mg/dL (13.9 mmol/L). (See 'Initial choice of fluid' above.)
●Management – Data evaluating the outcome and treatment of cerebral edema in adults are not available. Recommendations for treatment are based upon clinical judgment in the absence of scientific evidence. Case reports and small series in children suggest benefit from prompt administration of mannitol (0.25 to 1 g/kg) and perhaps from hypertonic (3 percent) saline (5 to 10 mL/kg over 30 min) [62]. These interventions raise the plasma osmolality and generate an osmotic movement of water out of brain cells and a reduction in cerebral edema.
Hyperchloremic acidosis — In the absence of severe kidney disease, most patients treated with isotonic saline develop a hyperchloremic, normal anion gap acidosis ("non-gap" or "hyperchloremic acidosis") during the resolution phase of the ketoacidosis. This occurs for two reasons. First, IV volume expansion reverses volume contraction and improves kidney function, which accelerates the loss of ketoacid anions with sodium and potassium [63,64]. The loss of these ketoacid anion salts into the urine represents "potential" bicarbonate loss from the body. Second, isotonic saline has a chloride concentration of 154 mEq/L and does not contain any bicarbonate precursors. Therefore, volume expansion with isotonic saline generates an element of hyperchloremic metabolic acidosis. The hyperchloremic acidosis slowly resolves as the kidneys excrete ammonium chloride and regenerate bicarbonate. The risk of hyperchloremic acidosis is lower with buffered crystalloid than with isotonic saline because its chloride concentration is lower, and it contains "potential bicarbonate" (eg, acetate, lactate). There are generally no clinical sequelae of hyperchloremic acidosis in this setting.
Noncardiogenic pulmonary edema — Hypoxemia and rarely noncardiogenic pulmonary edema can complicate the treatment of DKA [65-68]. Hypoxemia is attributed to a reduction in colloid osmotic pressure that results in increased lung water content and decreased lung compliance [9]. Patients with DKA who are found to have a wide alveolar-arterial oxygen gradient and/or rales may be at higher risk for the development of pulmonary edema.
SOCIETY GUIDELINE LINKS —
Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Hyperglycemic emergencies".)
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-to-read 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 education: Diabetic ketoacidosis (The Basics)" and "Patient education: Hyperosmolar hyperglycemic state (The Basics)")
SUMMARY AND RECOMMENDATIONS
●General principles – The treatment of diabetic ketoacidosis (DKA) involves correcting of the fluid and electrolyte abnormalities that are typically present, including hyperosmolality, hypovolemia, metabolic acidosis, and potassium depletion, and administering insulin (algorithm 1 and table 1 and table 4). Frequent monitoring is essential, and underlying precipitating events should be identified and corrected. (See 'Overview and protocols' above.)
●Fluid replacement – Individuals with DKA require intravenous (IV) fluid replacement to correct both hypovolemia and hyperosmolality. Isotonic saline and isotonic buffered crystalloid (eg, Lactated Ringer) are both reasonable options. The optimal rate is guided by clinical assessment.
For patients who present with an initial serum glucose <250 mg/dL [13.9 mmol/L]), dextrose (5 to 10 percent) is added to IV fluids at the initiation of therapy.
•Initial rate of administration – Fluid replacement should correct estimated volume deficits within the first 24 to 48 hours, with care to avoid an overly rapid reduction in the serum osmolality. (See 'Fluid replacement' above.)
-Hypovolemia with shock – Isotonic fluid (0.9 percent saline or buffered crystalloid) should be infused as quickly as possible in patients with hypovolemic shock. (See "Treatment of severe hypovolemia or hypovolemic shock in adults", section on 'Nonhemorrhagic shock'.)
-Hypovolemia without shock – In hypovolemic patients without shock or heart failure, we suggest isotonic fluid infused at a rate of 15 to 20 mL/kg per hour (approximately 1000 mL/hour in an average-sized person) for the first few hours (Grade 2C).
-Mild hypovolemia or euvolemia – Isotonic fluid is infused at a lower rate than in hypovolemic patients without shock, guided by clinical assessment.
•Subsequent fluid management – After the fluid deficit is corrected, one-half isotonic (0.45 percent) saline is administered at a rate of approximately 250 to 500 mL/hour if the serum sodium (corrected for hyperglycemia) is normal or elevated; isotonic saline is continued at a rate of approximately 250 to 500 mL/hour if the serum sodium (corrected for hyperglycemia) is low. (See 'Subsequent fluid management' above.)
●Potassium replacement – Most patients with DKA require IV potassium replacement. The dose depends on the initial serum potassium level (algorithm 1). In patients with high potassium (>5.0 mEq/L) and/or low urine output (eg, <50 mL/hour or 0.5 mL/kg/hour), potassium replacement should be delayed until potassium is ≤5.0 mEq/L and urine output ≥50 mL/hour.
Most patients with DKA have a substantial potassium deficit (due to urinary losses and secondary hyperaldosteronism) that may not be reflected in serum levels. (See 'Fluid replacement' above and 'Potassium replacement' above.)
●Insulin – In patients with initial serum potassium <3.5 mEq/L, insulin therapy should be delayed until the serum potassium is >3.5 mEq/L to avoid complications of hypokalemia. Aggressive potassium replacement is needed to avoid further delay in insulin therapy and increased risk of progressive acidosis (algorithm 1). (See 'Insulin' above.)
Insulin dosing is titrated based on hourly glucose measurement. We add dextrose to the IV fluids when the serum glucose is <250 mg/dL (13.9 mmol/L). In patients with DKA who present with an initial serum glucose <250 mg/dL, we add dextrose to the IV fluids at the initiation of therapy. (See 'Fluid replacement' above.)
•Moderate to severe DKA – Patients with moderate to severe DKA and a serum potassium ≥3.5 mEq/L require intensive insulin therapy (intravenous insulin infusion). We suggest IV regular insulin rather than rapid-acting insulin analogs (eg, insulin lispro, aspart, and glulisine) (Grade 2C), due to similar efficacy and much lower cost. Treatment can be initiated with a fixed-rate continuous infusion of regular insulin of 0.1 units/kg per hour (algorithm 1) [21,23-26]. The dose is doubled if the glucose does not fall by 50 to 70 mg/dL (2.8 to 3.9 mmol/L) in the first hour. Alternatively, a variable rate insulin infusion may be administered using a nurse-driven protocol. Serum glucose should be maintained between 150 and 200 mg/dL (8.3 and 11.1 mmol/L) until resolution of DKA. (See 'Moderate to severe DKA' above and 'Resolution criteria' above.)
•Mild DKA – For patients with mild DKA (table 1), subcutaneous rapid-acting insulin analogs may be used for initial treatment. Subcutaneous administration of insulin is safe only when adequate staffing is available to allow for close patient monitoring and frequent capillary blood glucose measurement with a reliable glucose meter, typically every hour. (See 'Uncomplicated mild to moderate DKA' above.)
•Normoglycemic DKA – For patients who present with normoglycemic DKA, initial treatment is similar to that for patients with mild DKA. Dextrose (5 to 10 percent) should be added immediately to IV fluids and initiated concurrently with insulin therapy. (See 'Normoglycemic DKA (glucose <200 mg/dL [11.1 mmol/L])' above.)
●Indications for bicarbonate – For most patients with DKA, we suggest not giving sodium bicarbonate (Grade 2C). However, in patients with severe acidosis (arterial pH <7.0), bicarbonate administration is reasonable. (See 'Bicarbonate and metabolic acidosis' above.)
●Indications for phosphate – For most patients, we suggest not administering phosphate (Grade 2B). Whole-body phosphate depletion is usually present, but routine phosphate administration does not appear to have clear benefit and can be associated with hypocalcemia and hypomagnesemia.
However, patients with severe hypophosphatemia (<1 mg/dL [0.32 mmol/L]) should receive phosphate. (See 'Phosphate repletion (rarely needed)' above.)
●Monitoring – Monitoring involves hourly glucose measurement until stable and basic chemistry profile, phosphorus, and venous pH every two to four hours. The course of ketoacidemia can be assessed by direct measurement of beta-hydroxybutyrate, the major circulating ketoacid, and/or measurement of the serum anion gap. In contrast, nitroprusside tablets or reagent sticks should not be used, because they react with acetoacetate and acetone but not with beta-hydroxybutyrate. (See 'Monitoring' above.)
●Potential complications – Cerebral edema is rare in adults but is associated with high rates of morbidity and mortality. Possible preventive measures in high-risk patients include gradual rather than rapid correction of fluid and sodium deficits and maintenance of a slightly elevated serum glucose until the patient is stable. (See 'Cerebral edema' above.)
●Converting to subcutaneous insulin – We initiate a multiple-dose (basal-bolus), subcutaneous insulin schedule when the ketoacidosis has resolved and the patient is able to eat. The IV insulin infusion should be continued for one to two hours after initiating subcutaneous rapid-acting insulin. The first dose of basal insulin also should be administered before IV insulin is discontinued. If short- or long-acting insulin is initiated without rapid-acting insulin, the IV insulin infusion should be continued for two to four hours after subcutaneous insulin administration. Abrupt discontinuation of IV insulin acutely reduces insulin levels and may result in recurrence of hyperglycemia and/or ketoacidosis.
When converting to subcutaneous insulin, we do not use ultra-long-acting insulins (eg, degludec, U-300 glargine) as the basal insulin; these have a long half-life and require two to three days to reach steady state. (See 'Converting to subcutaneous insulin' above.)
ACKNOWLEDGMENT —
The UpToDate editorial staff acknowledges Abbas Kitabchi, PhD, MD, FACP, MACE, who contributed to an earlier version of this topic review.