CONTENTS

• Rapid Reference 🚀
• Getting started
  • Evaluating anion gap & ketoacidosis
  • Definition & severity of DKA
  • Evaluating the cause of DKA
• Core components of DKA resuscitation
  • Fluid administration
  • Electrolyte management
  • Insulin infusion
  • Long-acting, basal insulin
  • Management of severe or refractory ketoacidosis
  • NAGMA management
  • Monitoring & management of DKA recurrence
• Special situations
  • DKA in a hemodialysis patient
  • Euglycemic DKA
  • DKA with an insulin pump
  • Intubating a DKA patient
  • Cerebral edema
  • IV access
• Podcast
• Questions & discussion

---

rapid reference

(back to contents)

---

DKA management checklist ✅

diagnostic evaluation (more)

• Minimum evaluation for a patient with DKA: Electrolytes including Ca/Mg/Phos, complete blood count with differential, urinalysis, EKG, pregnancy test as appropriate.
• If unclear whether patient has DKA: beta-hydroxybutyrate & lactate levels.
• If the cause of DKA is unclear: blood cultures +/- urine culture, chest X-ray, perhaps CT abdomen/pelvis to evaluate for septic focus, possibly lipase (noting that DKA itself can increase lipase; 14578269), troponin if genuine suspicion for ischemia.

crystalloid (more)

• 1st: Bolus with lactated Ringers (LR) if substantial volume depletion (which is usually the case).
• 2nd: Infuse LR at ~150-200 ml/hr, until glucose <300 mg/dL (<16.6 mM).
• 3rd: When glucose <300 mg/dL, cut LR infusion in half and add a D10W infusion (e.g., 100 ml/hr D10W plus 100 ml/hr LR).

electrolyte repletion (more)

• Check electrolytes q4hr (including magnesium & phosphate).
• Target K>5.3 mM, if renal function preserved.
• Target Mg >2 mg/dL (>0.8 mM).
• Replete phosphate as needed.

insulin infusion (more)

• Getting started:
  • Hold insulin if K <3.3 mM.
  • Most patients: start insulin at 0.1 U/kg/hr (up to a max of 15 U/hr).
  • For severe acidosis (e.g., bicarbonate <5 mM), consider a 10 unit IV insulin bolus followed by an infusion at 0.2 U/kg/hr.
• Continue insulin infusion per protocol, until ALL the following criteria are met:
  • Resolution of ketoacidosis (anion gap <10-12 mM, in the absence of renal failure).
  • Bicarbonate >18 mM.
  • Patient received full dose of basal insulin >2 hours previously.
  • Glycemic control is reasonable.
  • Patient feels better & tolerating PO.

basal insulin (more)

• Start basal insulin early (well before the anion gap has closed).
• Patients should receive their full daily requirement of basal, long-acting insulin as a single dose of glargine:
  • New diagnosis of diabetes: start 0.25 U/kg glargine.
  • Most patients: provide their entire daily requirement of basal insulin.
• Schedule basal glargine q24 hours.

management of non-anion-gap metabolic acidosis (more)

• NAGMA often emerges during the course of a DKA resuscitation.
• Signs of developing NAGMA:
  • Anion gap is falling, but bicarbonate is not rising appropriately.
  • (Na – Cl – 10) << 20.
• NAGMA often requires treatment with IV bicarbonate to achieve a bicarbonate level >18 mM, to expedite discontinuation of the insulin infusion in a timely fashion.

---

evaluating anion gap & ketoacidosis

(back to contents)

---

three ways to evaluate for ketoacidosis

(#1) anion gap

• Anion Gap = (Na – Cl – Bicarbonate)
  • Using this formula, an elevated anion gap is above 10-12 mEq/L.(23833313)
  • Please don't correct for albumin, glucose, or potassium. Don't make this unnecessarily complicated.
• Anion gap may be elevated due to a variety of causes (with the differential diagnosis explored here). Therefore, an elevated anion gap does not necessarily imply DKA! This is especially true among patients with chronic renal failure, who may have a chronically elevated anion gap.

(#2) urinary dipstick for ketones

• The urinary ketone dipstick tests for acetoacetate.
• This test has a high sensitivity for DKA (98-99%), with urinary ketones are generally being ≧2+.(32771260, 10459090) False negatives may occur in patients with highly acidic urine.(32409703)
• The specificity of a positive measurement of urinary ketones is low, so a positive urinary measurement of ketones doesn't establish a diagnosis of DKA.(32763063) For example, starvation ketoacidosis is a more common cause of urinary ketones in most contexts.

(#3) blood beta-hydroxybutyrate level

• Beta-hydroxybutyrate level is the gold standard for defining the presence and extent of ketoacidosis in DKA.
• Rough interpretation:(32771260)
  • 0-0.6 mM: Normal
  • 0.6-1 mM: Mild ketosis, may consider adjustment of insulin regimen. Among patients who initially presented with DKA, a reduction of the beta-hydroxybutyrate level below <1 mM indicates resolution of the DKA.(32409703)
  • 1-3 mM: Moderate ketosis, medical intervention is warranted. There is a risk of progression to DKA.
  • >3 mM: Consistent with DKA.(10030312, 18184896, 32771260)
  • >6 mM: Severe DKA.

clinical approach to anion gap & ketoacidosis

• Obvious DKA: In some cases the history and physical examination are strongly suggestive of DKA. In this situation, an elevated anion gap with positive urinary ketones may be sufficient to reach a diagnosis of DKA.
• Complicated cases: In confusing situations, it's helpful to simultaneously measure electrolytes, lactate, and a beta-hydroxybutyrate level. This can help sort out the precise etiology of the elevated anion gap, for example:
  • A strongly elevated beta-hydroxybutyrate level would support a diagnosis of DKA.
  • A markedly elevated lactate level with mildly elevated beta-hydroxybutyrate level might suggest an underlying disease process (e.g., sepsis or shock) that may be causing a mild amount of ketoacidosis.
  • An elevated anion gap with normal lactate and beta-hydroxybutyrate levels implies an alternative cause of the anion gap (e.g. certain intoxications).

---

definition & severity of DKA

(back to contents)

---

definition of DKA

• Many definitions of DKA may be found in the literature, most of which are antiquated. The Canadian DKA guidelines are therefore correct in asserting that “there are no definitive criteria for the diagnosis of DKA.”(24070967)
• My preferred definition of DKA is any patient with diabetes plus a significantly elevated serum beta-hydroxybutyrate level (>3 mM/L). (10030312, 18184896, 32771260)
• Please note the following:
  • DKA patients can have a normal glucose (euglycemic DKA, more on this below).
  • DKA patients can have a normal pH and a normal bicarbonate (this usually occurs due to a combination of ketoacidosis plus metabolic alkalosis from vomiting).

differential diagnosis of DKA

• Causes of ketoacidosis include starvation ketoacidosis, alcoholic ketoacidosis, and diabetic ketoacidosis.
• Clinical history is paramount in sorting these out. In patients with diabetes and alcoholism, it may be nearly impossible to sort out diabetic ketoacidosis versus alcoholic ketoacidosis (in this situation, the safest approach is often to treat the patient as if they have DKA).

severity of DKA

• As a rough gradation, the following benchmarks can be helpful:(32771260, British guidelines)
  • Severe DKA: serum bicarbonate <5 mM (or pH < 7.0-7.1, or beta-hydroxybutyrate >6 mM).
  • Moderate DKA: serum bicarbonate 5-10 mM (or pH ~7.1-7.2).
  • Mild DKA: serum bicarbonate >10 mM (or pH >7.2).

---

evaluating the cause of DKA

(back to contents)

---

precipitating cause

DKA is occasionally the initial manifestation of diabetes, but it usually occurs in the context of known diabetes plus a trigger. This is especially true of patients with type-II DM, who don't generally require exogenous insulin but may develop DKA in the context of physiologic stress. Most triggers of DKA are benign (e.g., nonadherence, viral gastroenteritis). However, DKA can be caused by any source of physiologic stress. Occasionally, DKA is the presentation of a serious underlying problem, especially sepsis. Common triggers of DKA include:

• Absolute insulin deficiency:
  • Insulin nonadherence.
  • Inadequate dosing of basal insulin.
  • Insulin pump failure.
  • New diagnosis of diabetes.
• Infection (e.g., gastroenteritis, pneumonia, urinary tract infection, diabetic foot infection).
• Pancreatitis.
• Pregnancy.
• Trauma, surgery.
• Substance abuse or alcoholism.
• Medications:
  • Steroid.
  • Sympathomimetics.
  • SGLT-2 inhibitors.
  • Atypical antipsychotics.
  • HIV protease inhibitors.
  • Checkpoint inhibitors (e.g., pembrolizumab, nivolumab).
  • Pentamidine.(32409703)
  • Anti-calcineurin immunosuppressives.(10743693)

evaluation for the cause of DKA

• History and physical examination are the key here. If there is clear history of nonadherence, a big workup isn't necessary.
• Infectious trigger?
  • DKA itself may cause leukocytosis, so a WBC elevation alone is nonspecific.
  • Infection is suggested by fever, marked left-shift, or severe leukocytosis (>20,000-25,000).(3101715)
• Primary abdominal problem?
  • DKA itself can cause abdominal pain. This creates diagnosis confusion – we must sort out whether the pain is due to DKA, or whether the pain represents an underlying problem (appendicitis, cholecystitis, etc). This may be sorted out in two ways:
  • (#1) Severe pain with only mild ketoacidosis argues against DKA causing the pain.(12040551)
  • (#2) When in doubt about the need for an abdominal CT scan, aggressively treat the DKA and follow serial abdominal examinations. If the abdominal pain is due to DKA, it will resolve as the ketoacidosis improves. If pain fails to resolve or gets worse, then further investigation is warranted.
• Primary neurologic problem?
  • DKA itself may cause mental status changes, but this usually occurs when the calculated serum osmolality is >320 mOsm/kg. Abnormal mental status despite normal serum osmolality should trigger suspicion for a primary neurologic problem (e.g., meningitis, intracranial hemorrhage).(25905280)
  • Another sign of a primary neurologic problem is if the mental status doesn't improve with treatment of the DKA. (Noting also that if mental status deteriorates during therapy, the possibility of cerebral edema should also be considered.)

---

fluid administration

(back to contents)

---

#1) start with fluid boluses

• DKA patients are often profoundly volume depleted (e.g., due to vomiting, reduced oral intake, and osmotic diuresis). Hypovolemia triggers the release of stress hormones (e.g., catecholamines, cortisol) which cause insulin resistance and thereby exacerbate the DKA. So prompt reversal of hypovolemia is important.
• Most patients will require ~2-4 liters of crystalloid up front in the form of boluses (e.g. ~1,000 ml/hr).
  • For young DKA patients with normal cardiorenal function, if the patient's heart rate is >100 b/m then they probably need more fluid.
  • ⚠️ Ultrasound-guided fluid resuscitation is useful for patients with heart failure, or patients on hemodialysis.
• Balanced crystalloid is preferred here (e.g., LR or plasmalyte) if possible (further discussion below).

#2) initial maintenance fluid infusion (if glucose >300 mg/dL or >17 mM)

• Once the patient is approaching a euvolemic state, a maintenance fluid infusion is generally started.
• The usual choice is an isotonic balanced crystalloid (lactated Ringer's or plasmalyte) at ~150-200 ml/hr.
• ⚠️ Patients with heart failure may tend to become overloaded with standard DKA protocols, so follow volume status with ultrasonography and consider using less fluid.
• If the glucose level is <300 mg/dL or <17 mM already (e.g., in euglycemic DKA), then skip this step.

#3) after the glucose falls <300 mg/dL or <17 mM, add dextrose

• As the glucose falls, dextrose must be added to the IV fluid to allow for ongoing insulin administration (since ongoing insulin administration is needed to correct the ketoacidosis).
• A nice strategy here is to drop and split:(32771260)
  • Cut the LR rate in half (e.g., from 200 ml/hr to 100 ml/hr).
  • Add a D10w infusion at an equal rate (e.g. 100 ml/hr LR plus 100 ml/hr D10W). Note that D10W is fine for peripheral IV infusion, it doesn’t require a central line. The D10W can actually be infused together with the LR using a single intravenous line, because these two fluids are compatible.
  • Combining LR with an equal volume of D10W effectively creates a solution of “D5 1/2 LR” (a solution which doesn't exist in pre-mixed bags). The advantage of giving the components separately is that it provides you greater control with regards to adjusting the amount of sodium you are giving versus the amount of dextrose. For example, if you want to give additional dextrose you can up-tirate the D10W infusion (without giving the patient more sodium and causing volume overload).
• An alternative approach is to switch to D5 1/2 NS at ~200 ml/hr.

balanced crystalloid versus normal saline

• Balanced crystalloid is generally preferred (e.g. lactated Ringers), as this will avoid worsening the patient's acidosis. Evidence supports the ability of balanced crystalloid to accelerate resolution of DKA.(33196806)
• The advantage of using normal saline is that it is available in preformulated bags containing potassium chloride. In some hospital units, this is a more convenient strategy for potassium repletion. Using normal saline for resuscitation is fine, particularly if this is the only way to appropriately replete the patient's potassium. However, clinicians should be aware that using saline will promote the development of NAGMA that may require active management with intravenous bicarbonate later on (as discussed below).

---

electrolyte management

(back to contents)

---

hypokalemia

• Hypokalemia is extremely problematic, because insulin cannot be given to patients with significant hypokalemia (since insulin will exacerbate the hypokalemia). Thus, hypokalemia impairs our ability to treat DKA.
• Hypokalemia may be treated with aggressive doses of IV potassium (e.g., 40 mEq/hr), with careful monitoring of electrolytes every hour.(24070967, 16191494, 28659865) To avoid damaging the veins, potassium may be given via a central line or via multiple peripheral lines (e.g., 20 mEq/hr simultaneously through two peripheral IVs).
  • (More on hypokalemia management here.)

hyperkalemia

• Hyperkalemia is less of a problem, because the usual DKA resuscitation will naturally reduce potassium.
• In severe hyperkalemia, IV insulin is indicated (e.g. 10 unit IV insulin bolus). IV calcium may also be indicated.
  • (More on hyperkalemia management here).

ongoing potassium repletion

• DKA resuscitation will cause the potassium to fall over time.
• Aggressive potassium repletion is generally needed, usually with repeated doses of IV potassium. Oral potassium can be used, but patients are often nauseous and unable to tolerate this.
• In the absence of renal failure, shoot for a potassium >5.3 mM (to avoid falling behind).
• In renal failure, be more conservative with potassium repletion.
• Potassium chloride is generally used. However, oral potassium citrate or IV potassium acetate may offer the advantage of reducing the chloride load and thus decreasing the tendency to develop NAGMA. .

magnesium repletion

• Maintaining a high-normal magnesium level may tend to protect against hypokalemia-induced arrhythmia, in case the potassium falls too low (isolated hypokalemia is usually well tolerated, whereas the combination of hypokalemia plus hypomagnesemia tends to cause torsade de pointes).
  • (More on hypomagnesemia here).

phosphate repletion

• Phosphate will drop during treatment, especially in patients with severe DKA.
• Follow the phosphate and replete if substantial hypophosphatemia occurs (<1-1.5 mg/dL or <0.3-05 mM).(32409703)
  • (More on hypophosphatemia here).

---

insulin infusion

(back to contents)

---

general concepts of using insulin in DKA

• The primary problem with DKA is ketoacidosis (not hyperglycemia). Therefore, our overall goal is to titrate insulin as needed to treat the ketoacidosis (figure above).
• Unfortunately, it's a bit more complicated than this. Glucose levels are easier to repeat than measurements of ketoacidosis (e.g., the anion gap). Thus, glucose levels are often used as a surrogate measurement of the biological efficacy of insulin (for example, during the initial phase of resuscitation, if the glucose level isn't falling, that indicates that insulin isn't working and should be up-titrated).
• Every hospital will have a DKA protocol, which can generally be followed. However, it's still useful to understand the broad strokes of how insulin is utilized in DKA, as described below.

(#1) insulin infusion: getting started

• Unless the patient is hypokalemic (K <3.3 mM), insulin should be started immediately.(32409703)
• Insulin bolus (10 units IV) is generally not recommended. However, an insulin bolus may be helpful in the following situations:
  • (1) There is a delay in receiving an insulin infusion from the pharmacy. The main advantage of an insulin bolus is that this can usually be given immediately (most units have 10-unit insulin vials immediately available), whereas an insulin infusion needs to be mixed up in pharmacy).
  • (2) For patients with severe acidosis (e.g., bicarbonate <5-10 mM), an insulin bolus will help immediately establish a therapeutic insulin level.
• An insulin infusion is usually started at 0.1 U/kg/hour (up to a max of 15 units/hour in morbid obesity). However, for patients with severe acidosis (e.g., bicarbonate <5 mEq/L) or marked insulin resistance (with high chronic insulin requirements), higher doses will often be needed (e.g. 0.2-0.3 U/kg/hr).

(#2) up-titration of insulin infusion, if needed

• The insulin infusion should be up-titrated as needed, with a goal of dropping the glucose by 50-70 mg/dL (2.8-3.9 mM) per hour.
• Occasionally, if the patient's anion gap isn't clearing, you might need to simultaneously increase both the insulin infusion rate and the glucose infusion rate. (Remember, the insulin is being used to clear the ketoacidosis.)

(#3) cut back on the insulin infusion, but don't stop it

• Once the glucose falls to ~250 mg/dL (14 mM) the insulin infusion rate is typically reduced considerably (to ~0.05 U/kg/hr).(32771260, British guidelines)
• Avoid stopping the insulin infusion entirely, if possible. Hypoglycemia may generally be managed by the use of additional IV dextrose and down-titration of insulin (rather than shutting the insulin off entirely).

(#4) stop the insulin infusion only after the following criteria are met:

• (a) Resolution of ketoacidosis (anion gap < 10-12 mEq/L).
  • An exception here is a patient with end-stage renal disease, who may chronically have an elevated anion gap due to uremia which never normalizes. In this situation, normalization of the beta-hydroxybutyrate level (<0.6 mM) is a more useful way to determine that ketoacidosis has resolved.
• (b) The patient isn't significantly acidotic (bicarbonate ≧18 mEq/L).(32771260, British guidelines)
  • Acidosis increases insulin resistance, so if the patient remains acidemic then there is an increased risk that the anion gap will open up.
  • Many patients will develop a NAGMA, leading to a persistent acidosis that doesn't respond to insulin. This may be treated with IV bicarbonate as described below.
• (c) The patient has received the full daily dose of long-acting insulin >2 hours previously.
• (d) Glucose is reasonably well controlled (e.g., <250 mg/dL or <14 mM).
• (e) The patient should ideally be hungry (this is an excellent sign suggesting that the ketoacidosis has resolved).
  • If the insulin infusion is stopped and the patient doesn't eat anything or receive any IV glucose, this increases the risk of recurrent DKA.
  • An exception can be made for patients with gastroenteritis or diabetic gastroparesis, who may not be hungry for several days. In this situation, the insulin infusion can be stopped, but patients should remain on low-dose intravenous glucose (e.g. D5W at 50-75 ml/hr). If the patient's glucose level increases, they should be treated with PRN short-acting insulin. Ongoing administration of carbohydrate plus PRN insulin will help prevent DKA recurrence.

(#5) start meal-associated & PRN insulin when the infusion is stopped

• Start meal-associated and sliding-scale insulin.
  • If the patient isn't already on a prescribed regimen of meal-associated insulin, a dose of ~0.08 U/kg rapid-acting insulin per meal may be reasonable (i.e., about one third of the daily basal insulin requirement). Follow glucose carefully and titrate to effect.
• Encourage patients to eat. Carbohydrate intake (along with meal-associated and sliding-scale insulin) is important at this point, to prevent recurrent DKA.

---

long-acting basal insulin

(back to contents)

---

concept of early basal insulin

• Long-acting insulin should be started early (well in advance of discontinuing the infusion). Glargine has a delayed onset compared to some older forms of insulin (e.g., NPH), so the traditional two-hour overlap may not work well with glargine.(28372715) Early initiation of long-acting insulin facilitates transitioning off the insulin infusion, reduces the incidence of hyperglycemia, and might decrease hospital length of stay.(28183452)
  • (More on the rationale & evidence for early insulin here.)

step #1: determine the total daily requirement of long-acting insulin

• For patients on home subcutaneous insulin, this will be equal to the patient's home basal dose:
  • For patients on once-daily long-acting insulin (e.g., glargine), that's their basal dose
  • For patients on twice-daily basal insulin, add up all the basal doses given during a day.
• For patients naive to insulin, a starting dose of 0.25 units/kg daily of glargine (Lantus) may be used.(British guidelines)
• For patients on an insulin pump:
  • Some patients have a backup dosing regimen of long-acting insulin (e.g., a certain dose of s.q. long-acting insulin to use if their pump malfunctions). You can use that as their daily basal insulin dose.
  • The basal insulin requirement may also be calculated from the pump's basal rate (e.g., multiply the basal rate times 24 hours to obtain the total daily basal requirement).
  • Note that the patient's pump should be stopped and removed (more on this below)
• ⚠️ Don't calculate the patient's daily insulin requirement based on how much insulin they are receiving via the insulin infusion. Sick DKA patients are receiving lots of IV dextrose and they are acidotic, which will temporarily increase their insulin requirements.

step #2: give the full dose of basal insulin

• Provide the entire day's worth of basal insulin (typically in the form of glargine).
• If the glargine is given at an inopportune time (e.g. it’s given in the evening and the patient prefers taking it in the morning), the timing can be slowly shifted each day to meet the patient's preference.

common pitfalls with long-acting insulin

• Practitioners who are nervous about giving early glargine may sometimes give a reduced dose, which leads to tremendous confusion. Please give patients their full home-dose of basal insulin. Critical illness causes insulin resistance, so patients may have a tendency to require more insulin (not less).
• Some patients are on twice daily glargine (for reasons which aren't entirely clear to me). If such patients are continued on twice daily glargine, the insulin infusion shouldn't be stopped until after they receive their second dose of glargine. Alternatively, both doses can be compiled into a single daily dose (this is preferred as it may accelerate weaning off the insulin infusion).
• Glargine must be ordered “q24 hours,” rather than “daily.” If the glargine is electronically ordered as “daily,” then it may default to every morning at 9 AM – which will cause some patients to receive their daily dose in the evening, and then another dose the following morning. Please check and double-check the glargine dose and when it is scheduled to be delivered.

---

management of severe or refractory ketoacidosis

(back to contents)

---

(#1) initial management of patients with severe ketoacidosis

• Some patients will present with profound ketoacidosis (e.g. pH<6.9 or bicarb <5 mM). Patients generally tolerate this surprisingly well.
• Avoid giving bicarbonate during the initial resuscitative phase (for management of ketoacidosis). The proper treatment of ketoacidosis is insulin. Thus, if the patient has a severe ketoacidosis which requires aggressive management, the most effective strategy is to increase the insulin dose (usually along with administration of additional glucose and potassium):
  • a) Don't wait for the insulin to arrive from pharmacy: bolus 10 units IV immediately.
  • b) Consider starting an insulin infusion at 0.2 U/kg/hr in the sickest patients.
• High-flow nasal cannula (HFNC) may be helpful to support respiratory compensation for metabolic acidosis.
  • High-flow nasal cannula is a safe way to support the patient's breathing (note that patients are not good BiPAP candidates, due to a tendency to vomit). High-flow nasal cannula may reduce the anatomic dead space, thereby reducing the work of breathing and avoiding respiratory fatigue.
  • High-flow nasal cannula may be set as follows:
    • FiO2 titrated to achieve a saturation >92% (usually a low FiO2 will be needed e.g., 30-40%)
    • Increase the flow rate as high as the patient can tolerate (e.g., 60 liters/minute). The flow rate is what does the work of reducing dead space and thereby blowing off CO2. If the patient is very sick and air-hungry, they will tolerate high flow rates.
• More on the management of severe acidosis here.

(#2) refractory ketoacidosis: if the anion gap is not closing

• If the anion gap isn't closing, consider the following possibilities:
  • Inadequate fluid resuscitation.
  • Inadequately low insulin dose.
  • Malfunction of insulin infusion (e.g., line infiltration or infusion pump error).
  • Festering, underlying problem which hasn't been addressed.
• Interventions if the anion gap isn't closing:
  • Evaluate fluid status (e.g. with ultrasonography), provide additional crystalloid resuscitation if necessary.
  • Consider increasing the insulin infusion rate (which may require a simultaneous increase in dextrose administration).
  • Reevaluate for a missed underlying problem.
  • Consider checking beta-hydroxybutyrate & lactate levels (to exclude an occult lactic acidosis).
  • Make sure the infusion pump and IV catheter are functioning properly.

---

NAGMA management

(back to contents)

---

the problem with non-anion-gap metabolic acidosis (NAGMA)

• NAGMA commonly develops in patients with DKA, often worsening during resuscitation.
• Causes of NAGMA include the following:
  • Resuscitation with normal saline or half-normal saline.
  • Excretion of ketoacid in the urine (once ketoacid is in the sewer system, it can no longer be converted back into bicarbonate).
• NAGMA is problematic, because it makes it harder to transition patients off the insulin infusion.
  • A residual acidosis will increase insulin resistance, thereby increasing the risk of recurrent DKA after stopping the insulin infusion (more on this here).
  • If the NAGMA is severe, then it may delay the discontinuation of the insulin infusion entirely.

diagnosis of NAGMA

Predicted Final Bicarb = (Na – Cl – 10)

 

• Developing NAGMA may be revealed by the following:
  • The anion gap is closing, but the patient's bicarbonate remains low.
  • The predicted final bicarbonate is a rough estimate of where the bicarbonate will end up after all the ketoacid is converted into bicarbonate (above). If the predicted final bicarbonate is falling over time to well under 20 mM, this suggests NAGMA.

management of NAGMA

• NAGMA should be treated with IV bicarbonate to achieve a bicarbonate level above 18-20 mEq/L prior to discontinuing the insulin infusion.
  • The bicarbonate deficit can be estimated using this formula. While the anion gap is still open, use the predicted final bicarbonate to get a rough concept of the bicarbonate deficit . Keep in mind, however, that you're only shooting for a bicarbonate of ~20 mEq/L (not 24 mEq/L). 100-150 mEq of bicarbonate is usually adequate.
  • If the patient is hyponatremic, then a couple of hypertonic bicarbonate ampules can be used (each ampule contains 50 mEq sodium bicarbonate in 50 ml water).
  • If the patient's sodium is normal or elevated, then isotonic bicarbonate may be used (e.g. one liter of D5W with three ampules of bicarbonate, to generate a 150 mEq/L bicarbonate solution, infused over 3-4 hours). This will cause the glucose to increase a bit, but that can actually be useful in closing the anion gap (because it will trigger an increase in the insulin infusion).
• The optimal time to treat NAGMA is often as the anion gap is beginning to close (e.g., when the anion gap is ~12-18 mEq/L). This facilitates prompt discontinuation of the insulin infusion (without needing to delay in order to separately treat the NAGMA).
• More on NAGMA and IV bicarbonate.

---

monitoring & management of DKA recurrence

(back to contents)

---

after the insulin infusion is stopped, patients should be monitored for recurrence of DKA:

• (1) Follow the glucose level.
  • Development of progressively severe hyperglycemia may be an early sign of recurrent DKA.
  • Since glucose levels are often checked frequently, skyrocketing glucose may pre-date the development of a widening anion gap by a few hours.
• (2) Consider repeat electrolytes:
  • A set of electrolytes ~6 hours after stopping the drip is a reasonable idea to make sure that the anion gap is remaining closed (if there is any doubt about this clinically).

management of recurrent DKA

• Causes of recurrent DKA (anion gap re-opens after stopping drip)
  • Insulin infusion was stopped despite not meeting all five of the criteria above.
  • Inadequate long-acting insulin dose.
  • Patient isn't eating enough (which causes insufficient meal-associated & PRN insulin doses)
  • Ongoing systemic inflammation (e.g. DKA caused by infection, with persistent infection).
• Treatment:
  • Restart the insulin infusion.
  • Continue long-acting insulin (consider up-titrating the dose).
  • Address any reversible causes of DKA.
  • Aggressively treat NAGMA to get the serum bicarb >20 mEq/L (this will improve insulin sensitivity).
  • Sometimes patients just need a bit longer on the insulin infusion (especially if they were severely ill on admission).

---

DKA in a patient on hemodialysis

(back to contents)

---

key physiologic differences

• Patients who are anuric on hemodialysis won't develop polyuria due to hyperglycemia, so they won't develop hypovolemia or potassium wasting. Consequently, they won't require aggressive volume or potassium repletion.
  • In some patients, hyperglycemia may osmotically pull water into the vasculature leading to hypervolemia! This can lead to volume overload with pulmonary edema, which may resolve following insulin administration (since the insulin causes a shift of glucose and water out of the vasculature and into the tissues).
• If you give excessive fluid or potassium, this will create a persistent problem that requires dialysis.
• Insulin clearance may be sluggish, so insulin may stick around longer than usual.

treatment pearls

• ⚠️ Standard DKA protocols will harm these patients.
• (1) Avoid aggressive volume administration.
  • Many patients may be euvolemic or only mildly hypovolemic (e.g., due to emesis or poor oral intake). Other patients may be hypervolemic (if they have missed hemodialysis recently).
  • There may be little or no need for volume resuscitation. Patients will inevitably get a couple liters of fluid with various drugs and infusions – this fluid alone may be adequate.
  • 💡 Fluids administered to the patient will accumulate over time, so you don't want the patient to be euvolemic within the first couple hours of resuscitation. If the patient is slightly hypovolemic initially, then accumulation of fluid with various IV medications will eventually bring them to a euvolemic state.
• (2) Avoid aggressive potassium administration.
• (3) Many patients just need insulin and dextrose.
  • For euvolemic patients, D10W may be superior to D5W to avoid causing hyponatremia. If central access happens to be available, then higher concentrations could be used (e.g., D20W). As discussed above, there is often no need for additional sodium administration (so there is no benefit from a fluid such as D5 half-normal saline).
  • Be conservative with insulin administration, as it may be cleared slowly. For patients who aren't severely acidotic, it may be wise to start the insulin at a slower rate than usual (e.g., 0.05 units/kg/hour).
• (4) The anion gap will never normalize!
  • Patients on dialysis will always have an elevated anion gap, due to uremia.
  • If beta-hydroxybutyrate is available, this is arguably the best way to determine the severity of the ketoacidosis. DKA resolution may correlate with a beta-hydroxybutyrate level which is below ~1 mM.(32409703)
  • If beta-hydroxybutyrate is not available, look at the patients prior anion gap values to determine a sense of the patient's baseline (often around ~12-16 mM). If the patient's anion gap falls to within this range and remains stable over time despite insulin administration, then the ketoacidosis has likely resolved.

role of hemodialysis

• Hemodialysis will remove ketoacid, replace bicarbonate, and basically fix everything. However, hemodialysis often isn't required (e.g., insulin alone may be adequate to improve hyperkalemia and acidosis).
• The risk of hemodialysis is that it may cause rapid osmotic shifts. For patients with severe hypertonicity (serum osmolality >>330 mOsm/kg), this could theoretically carry a risk of causing cerebral edema.
• Even if hemodialysis fixes everything, don't forget the insulin – the patient still needs insulin to prevent slipping back into DKA.

---

euglycemic DKA

(back to contents)

---

definition

• Definition of euglycemic DKA requires:
  • (#1) Ketoacidosis due to diabetes. The most unequivocal way to diagnose ketoacidosis is with a substantially elevated beta-hydroxybutyrate level (>3 mM). Alternatively, ketoacidosis may be inferred on the basis of urinary ketones, an elevated anion gap, and no alternative explanation for the anion gap elevation.
  • (#2) “Euglycemia” is defined as a glucose <250 mg/dL (<13.9 mM).(28924481) Of course, this is not an entirely normal glucose level.
• The definition of DKA is discussed further above. Note that DKA does not require the presence of acidemia (low pH). For example, coexisting metabolic alkalosis due to vomiting plus DKA may leave the patient with a normal pH and normal bicarbonate level.
• 🤯 Euglycemic DKA can occur with a normal glucose and a stone-cold normal blood gas (e.g., normal pH, normal bicarbonate, and normal pCO2).

more common causes of euglycemic DKA

• SGLT2 inhibitors (empagliflozin, canagliflozin, dapagliflozin) – usually in the presence of some additional stressor.
  • SGLT2 inhibitors block glucose reabsorption in the proximal nephron, promoting glucosuria. This may tend to induce a physiologic state mimicking starvation, which promotes ketoacidosis.
• Starvation, prolonged nausea/vomiting.
• Abdominal pathology (e.g., pancreatitis, gastroenteritis).
• Intoxication (especially alcohol).
• Liver disease.
• Sepsis.
• Pregnancy.
• Partial treatment with insulin before admission (either intentionally, or unintentionally via an insulin pump).
• Ketogenic diet.

clinical presentation

• Aside from the lack of hyperglycemia, patients present similarly to other DKA patients (e.g., with nausea, emesis, and abdominal discomfort).
• In situations where the DKA occurs in response to a primary stressor, clinical findings will reflect a melding of both conditions.
• Absence of hyperglycemia may cause both patients and clinicians alike to miss the diagnosis.

treatment: salient differences compared to the usual DKA resuscitation

• Treatment overall is very similar to DKA in general, with a few nuances:
• (1) Aggressive IV dextrose must be started immediately (e.g., D5 Lactated Ringers at ~250 ml/hr in mildly hypovolemic patients, or D10W at ~125 ml/hr in euvolemic patients).
• (2) Many patients weren't previously on insulin. Nonetheless, an insulin infusion is generally required to treat the DKA.(33626481) Additionally, early initiation of long-acting insulin is generally useful to prevent patients from slipping back into DKA (e.g., 0.25 units/kg glargine q24 hours).
• (3) Any SGLT2 inhibitors should be held until the DKA has resolved. These agents may be restarted later, at the discretion of the patient's endocrinologist.

---

DKA with an insulin pump

(back to contents)

---

• First, disconnect the insulin pump (including removal of the needle from the skin).(32763063) It's unclear whether the pump is working. The safest approach is to remove this variable from the equation until the patient is stabilized.(British guidelines)
• Otherwise, the treatment of DKA is essentially as for any other patient.
• If the patient and their endocrinologist decide to resume pump therapy, the transition from glargine back onto the pump can be made at a later date (following stabilization and ICU discharge).

---

intubating a DKA patient

(back to contents)

---

avoid intubation

• Whenever possible, avoid intubation.
• Intubating a patient due to altered mental status is usually a mistake. The mental status should improve over several hours, so careful observation is generally the best approach.
• Indications for intubation may include:
  • Frank inability to protect airway (e.g. gurgling, inability to control secretions).
  • Intubation needed to facilitate surgical procedure (e.g. patient has DKA plus perforated viscus).
  • Respiratory arrest or impending arrest (e.g. patient in extremis).
• If intubation is necessary (e.g. for a surgical procedure), it may be wise to delay it for a few hours to allow vigorous treatment of DKA first.

risks involved with intubation

1. Hemodynamic collapse: If hypovolemia isn't corrected prior to intubation
2. Vomiting/aspiration: These folks often have gastroparesis and illeus.
3. Decompensation of acidosis: Most patients have severe metabolic acidosis with a compensatory respiratory alkalosis. Paralysis takes away their respiratory compensation, potentially leading to profound acidosis.

mitigating the risks

• Hemodynamic collapse:
  • Volume resuscitate prior to intubation.
  • If necessary start a vasopressor infusion to establish MAP >75-80mm before the procedure.
  • Use hemodynamically stable induction drugs (e.g. ketamine).
• Avoid regurgitation
  • Visualize the stomach with ultrasound, if it's distended consider NG drainage prior to intubation.
  • If the patient is intermittently vomiting, encourage them to vomit immediately before anesthetic induction (while they can still protect their airway).
• Consider giving some bicarbonate prior to intubation if the bicarbonate level is <10 mEq/L.
  • For example, slowly push 2-3 ampules (100-150 mEq) of bicarbonate over 10-15 minutes, at least ~10 minutes prior to intubation.
  • Bicarbonate contains dissolved CO2, which the patient must blow off. In order to benefit from the bicarbonate, the patient should have enough time to blow off additional CO2 prior to intubation.
• Consider using mechanically controlled apneic ventilation (with BiPAP or a ventilator) during induction of anesthesia if you're adept at this. If you're not, then it's probably best to perform pure RSI to minimize risk of regurgitation (without any breaths interposed between paralytic and intubation).
• Use a relatively large ETT to minimize airway resistance (ideally nothing smaller than a 7.5-mm ETT).
• Use rocuronium, so that after intubation the patient will be paralyzed and sync perfectly with the ventilator.
• As soon as the ETT is secured, increase tidal volume & respiratory rate to hyperventilate the patient (thus restoring respiratory compensation).
  • Set the tidal volume at 8 cc/kg.
  • Crank the respiratory rate as high as possible without causing autoPEEP (will often end up around ~24-28 breaths/minute).
  • Shoot for a very high minute ventilation (e.g. 12-18 liters/minute).

---

cerebral edema

(back to contents)

---

basics

• This is extraordinarily rare in the context of adult DKA (it's a much larger issue in pediatric DKA).
• Adults at risk for cerebral edema might include:
  • Younger patients (almost all affected are <25 years old).
  • Marked baseline hyperosmolarity (e.g., calculated serum osmolarity >~330 mOsm).
  • Lack of the normal, expected rise in Na level during DKA treatment.(32409703)

prevention of cerebral edema

• Previously, the following measures were recommended to avoid cerebral edema in patients at risk:
  • (1) Don't drop the glucose too fast or too low. Also avoid reducing the glucose below <200 mg/dL (<11 mM).
  • (2) Avoid hypotonic fluids (there is no rush in reducing the tonicity). Note that the patient's sodium level will often initially increase during resuscitation, as glucose and water enter the cells. This doesn't reflect an increase in serum osmolality – and it shouldn't be a trigger to administer free water.
  • (3) Avoid dropping the serum osmolality by more than 3 mOsm/kg/hour.(24070967)
• A large pediatric study found no difference in cerebral edema when patients were randomized to hypotonic vs. isotonic fluid at various rates.(29897851) This suggests that cerebral edema is largely an idiosyncratic response, which may not be substantially impacted by our treatments.
• Consequently, there is no need to routinely take the above precautions to prevent cerebral edema in adults.(British guidelines) However, in selected patients who are at unusually high risk for cerebral edema, such precautions may remain reasonable.

diagnosis and treatment

• Clinical features:
  • (1) Consider the diagnosis in at-risk patients, as described above.
  • (2) Look for the emergence of headache, altered mental status, and emesis during treatment.
• Treatment involves the use of hypertonic fluid (e.g., hypertonic saline or hypertonic bicarbonate) to raise the serum tonicity. When the diagnosis of cerebral edema is suspected, the serum tonicity should be raised immediately, without delaying for neuroimaging.(British guidelines)
• Neuroimaging should evaluate for alternative possibilities (including cerebral venous thrombosis or CVA).

---

IV access

(back to contents)

---

• The vast majority of patients with DKA can be treated with peripheral IV access, but for the sickest patients central access may be needed.
• Femoral site is often best for DKA patients:
  • In a severely acidotic patient with respiratory compensation, a pneumothorax would be poorly tolerated.
  • Patients may be delirious and unable to stay still enough to facilitate safe placement of a jugular/subclavian line.
  • The line will only be needed for 24-48 hours (until DKA resolves), so infection risk is minimal.

---

IBCC podcast on DKA

(back to contents)

---

Follow us on iTunes

---

questions & discussion

(back to contents)

---

To keep this page small and fast, questions & discussion about this post can be found on another page here.

• Missing an underlying cause of DKA, especially sepsis. The primary cause of death among patients admitted with DKA isn't the DKA itself, but rather associated conditions.
• Bolusing large volumes of normal saline will make the patient more acidotic. This is generally not a major problem, but it can be for the sickest patients who present with severe acidosis.
• Avoid intubation of DKA patient if possible (it's a trap!). If you do need to intubate, proceed with extreme caution & preparation.
• BiPAP should be avoided as well, as patients will often vomit. To provide some additional respiratory support, consider high-flow nasal cannula.
• Please don't measure a troponin on every DKA patient (older DKA patients will usually have a measurable troponin level, which may trigger unnecessary and harmful workups). Be a doctor. Check a troponin if you are genuinely concerned about ischemia, based on symptoms and EKG evaluation.
• Don't stop the insulin infusion until the patient meets criteria to do so.
• Don't exclude the diagnosis of DKA because a patient has normal glucose, or normal bicarb/pH. Remember, instead, to mind the gap.

Guide to emoji hyperlinks

• = Link to online calculator.
• = Link to Medscape monograph about a drug.
• = Link to IBCC section about a drug.
• = Link to IBCC section covering that topic.
• = Link to FOAMed site with related information.
• = Link to supplemental media.

Going further

• Guidelines
  • British guidelines (2021 update) – probably best guidelines currently available.
  • Italian guidelines 2020 (32771260)
  • Canadian 2013 guidelines.(24070967)
  • American Diabetes Association 2009 guidelines – these are outdated and not recommended.(19564476)
• PulmCrit DKA series
  • DKA I: the pearls
  • DKA II: Dominating the acidosis
  • Why ABG & VBG are unhelpful in DKA
• Euglycemic DKA
  • REBEL EM: Euglycemic DKA isn't a myth
  • EMDocs: Euglycemic DKA secondary to SGLT2 inhibitors
• Bicarb in DKA? See Chris Nickson on LITFL, Anand Swaminathan on EMDocs, Salim Rezai on RebelEM, Darrel Hughes on RebelEM.
• Insulin bolus? Darrel Hughes on RebelEM
• DKA myths by RebelEM, and also by Anand Swaminathan on EMDocs.
• More on DKA in ESRD: Great twitter thread by Jamie Willows here.