Diabetic Ketoacidosis: “Sneaky” triggers and clinical pearls

Authors: Catherine Reynolds, MD (EM Resident Physician, UT Houston), Kathryn Fisher, MD (EM Resident Physician, UT Houston), and Hilary Fairbrother, MD (EM Attending Physician, UT Houston) // Edited by: Alex Koyfman, MD (@EMHighAK) and Brit Long, MD (@long_brit)

Clinical Case #1:  

The patient is a 45-year-old male with history of type 2 DM and HTN who presents to the emergency department with diffuse abdominal pain, nausea, and vomiting for two days. The patient reports that he was recently started on Invokana (canagliflozin) about two weeks ago for his diabetes, and prior to that he was managing his sugars with diet and exercise alone. The patient has had difficulty keeping any food down over the last two days, but denies shortness of breath, chest pain, or fever. He reports his vomit is non-bilious and non-bloody. His initial finger stick is 84 mg/dL, and your intern states that he “knows that this is not DKA” as the serum glucose is normal.

Vital signs showed RR 28, HR 110, BP 152/86, SpO2 98% on room air, T 98.4 F. Labs on presentation reveal a Na of 132 mmol/L, K of 5 mmol/L, Cl of 98 mmol/L, bicarbonate of 10 mmol/L, glucose of 84 mg/dL, and venous pH of 7.1. Venous lactate is within normal limits, and serum beta-hydroxybutyrate is positive.

The patient is started on dextrose-containing fluids, an insulin drip, and admitted to the ICU for DKA.

What was the cause of the patient’s euglycemic DKA?

Clinical Case #2:

The patient is a 62-year-old male with a history of DM, CAD, HTN, HLD, an NSTEMI, and 2 subsequent cardiac stents.  He presents to the ED with diffuse abdominal pain, nausea, vomiting, and diarrhea for two days.  The patient is on Humalog and Lantus but he hasn’t taken the medications over the last two days because he hasn’t been eating.  He denies shortness of breath, chest pain, or fever.  Your intern states that he “knowns that this is DKA” as the serum glucose is 420 mg/dL.

Vital signs showed RR 28, HR = 120, BP 148/78, SpO2 97% on room air, T 99.0 F.  Labs on presentation reveal a Na of 129 mmol/L, K of 4.0 mmol/L, Cl of 90 mmol/L, bicarbonate of 12 mmol/L, glucose of 402 mg/dL, venous pH of 7.0.  Venous lactate is 2.5, serum beta-hydroxybutyrate is positive.

The patient is started on an insulin drip and admitted to the ICU for DKA.

Just before the patient is due to move to the ICU, the nurse comes to you and states that he feels that the patient looks worse, that the patient’s wife is asking for more pain medication, and a repeat venous lactate is 4.5 despite the patient’s successful initial resuscitation for DKA.

Is there something besides DKA making this patient sicker than anticipated?


Diabetic ketoacidosis (DKA) is an acute, life-threatening complication of diabetes, with a mortality rate of 2-5% in developed countries.[1] It is traditionally characterized by hyperglycemia, the production of ketones, systemic acidosis, and dehydration, although in some circumstances patients may have euglycemia.[2] Diabetic patients typically experience DKA when they have a relative insulin deficiency coupled with an excess production of counter regulatory hormones driving ketone production from fatty acid metabolism.[2]

Most commonly, patients in DKA are triggered by infection (30%), a new diagnosis of diabetes (25%), or noncompliance with medication (20%).[1] However, these common causes of DKA don’t cover all of the possible triggers–listed in this article are rare but important causes of DKA that can be very difficult to treat if they are not recognized on admission.

Sneaky DKA Triggers


Sodium-glucose co-transporter (SGLT2) inhibitors such as canagliflozin, dapagliflozin, and empagliflozin represent the newest class of oral agents (often used as a second-line in conjunction with other oral medications) in the treatment of type 2 diabetes mellitus. [3] They act by reversible inhibition of the sodium-glucose SLGT2 transporter in the proximal tubule of the kidney, which is responsible for ~90% of glucose reabsorption. [3] Inhibition reduces blood glucose through increased urinary glucose excretion.

For reasons that are not completely clear, patients taking SGLT2 inhibitors are at increased risk of ketosis. Due to the renal excretion of glucose, patients often present euglycemic with blood glucose < 250 mg/dl and often < 100 mg/dl, thus not the typical hyperglycemic DKA that we all expect. [4] Patients still have the typical symptoms of DKA including nausea, vomiting, abdominal pain, and ketonemia. The elevated glucosuria can occur within a few days of initiation of the medication and can persist for up to 10 days after discontinuation of the drug. [5] As patients are euglycemic, it is important to add dextrose containing fluids initially to resuscitation efforts in conjunction with insulin. It may even be necessary to decrease the dose of insulin to prevent hypoglycemia. [6]

A meta-analysis performed by Monami et al. and released in August 2017 collected randomized controlled trials that looked at the incidence of ketosis in patients with type 2 diabetes mellitus who were prescribed SGLT-2 inhibitors. The researchers noted that while there were reported cases of DKA, when the drug was properly prescribed (at least 100 mg, PO, daily of canagliflozin, at least 5 mg, PO, daily of dapagliflozin, at least 10 mg, PO, daily of empagliflozin), the risk of DKA was negligible in patients with Type II diabetes (less than 0.01 %). [7]


Glucocorticoids alter carbohydrate metabolism by opposing insulin action, decreasing glucose utilization in the body, and increasing lipolysis.[8] In a healthy patient taking glucocorticoids, extra insulin will be produced to correct hyperglycemia. However, if the patient lacks normal insulin production or has increased insulin resistance, they will become hyperglycemic and may risk developing DKA. [9] These patients can be difficult to identify, as they may be presenting with new-onset DM at the time of DKA.


DKA is a well-known cause of stroke, especially in children and adolescents. However, the converse is also true, as the stress reactions during cerebrovascular accidents may cause DKA in diabetic patients.[10] DKA during a cerebrovascular incident is especially important to the prudent emergency medicine physician, as DKA causes activation of the coagulation system and may adversely affect the outcome of a patient suffering from a CVA.[10] Because DKA and CVA can have many overlapping symptoms, DKA should always be ruled out in the diabetic patient presenting with typical stroke symptoms.


Diabetic ketoacidosis in pregnant patients puts both patient and fetus at risk. Typically, DKA develops in patients who are in later stages of pregnancy and have newly diagnosed diabetes. In normal pregnancy physiology, patients have increased minute ventilation, which is compensated by increased renal excretion of bicarbonate.[11] Thus, pregnant patients have a lowered buffering capacity when exposed to any acidotic state and are at risk of developing DKA more rapidly and at a lower glucose level compared to their non-pregnant counterparts.[11] These patients can be difficult to identify, as they will likely present with vague symptoms of pregnancy such as abdominal pain and nausea.


Second-generation antipsychotics such as olanzapine, risperidone, and clozapine have been implicated in the development of DKA even in patients without a known prior history of diabetes.[14]  Though drugs differ in their propensity to lead to DKA, the risk is highest in patients using clozapine and olanzapine, often prescribed for the treatment of schizophrenia.[12] Some patients develop isolated hyperglycemia without DKA, while others progress to DKA. The mechanism is poorly understood but thought to be related to decreased pancreatic beta cell activity or decreased glycogen synthesis.[13] Patients who develop DKA often have symptoms within the first 6 months of treatment.[12] Signs and symptoms of DKA in this population are typical with abdominal pain, anorexia, nausea, vomiting, and hyperglycemic anion gap metabolic acidosis with ketonemia.


Thyrotoxicosis leads to both increased glucose absorption and production from glycogen, lactate, and amino acids. Thyroid hormone increases lipolysis, leading to increased fatty acid metabolism and ketone production. Thyrotoxicosis can precipitate DKA in patients through the decreased effectiveness of insulin and counter-regulatory hormones in patients who have type 1 diabetes, and patients with type 1 diabetes have a higher risk of developing autoimmune thyroid disease. If a patient has persistent tachycardia in the setting of DKA after adequate fluid resuscitation in the absence of suspected infectious etiology, consider underlying hyperthyroidism as both the DKA trigger and the cause of tachycardia and possibly high output cardiac failure.[15]

Weird Infections

Roughly 30% of patients in DKA are triggered by infection. Most commonly, these infections are found in the urinary or GI tract, but other, less common infections can also trigger DKA and may be more difficult to diagnose.[16] DKA may be triggered by acute pancreatitis, possibly due to a profound transient hyperlipidemia. These patients will have abdominal pain that can often be difficult to distinguish from the vague symptoms of a DKA with no clear infectious etiology. With acute pancreatitis, the diagnosis is further complicated as lipase and amylase are typically elevated in DKA, and Ranson’s prognostic criteria is not applicable to assess the severity. Additionally, acute pancreatitis is more likely to be associated with severe DKA with marked acidosis and hyperglycemia. Infections that require definitive surgical management, such as cholecystitis or appendicitis require an additional clinical challenge as they may present as primarily generalized abdominal pain, yet they require antibiotics and a surgical consult that cannot be missed. Diabetic patients are relatively immunocompromised so they may not present with a profound leukocytosis, even in these conditions. However, these less common infections should be ruled out in any DKA patient with a focused history and physical exam.

Arterial Disease (Myocardial Infarction and Mesenteric Ischemia)

Ischemia is a known trigger of DKA, but the converse is also true as diabetes contributes to the development of arterial ischemia. Both myocardial infarction and mesenteric ischemia are rare but serious causes of DKA. Diabetes is a known hypercoagulable state, and acute hyperglycemia is thought to increase this secondary to its action on clotting factors. Patients with diabetes have accelerated atherosclerosis and plaque rupture leading to an increased risk of myocardial infarction (MI). MI accounts for less than 5% of all cases of DKA; however, it represents a disproportionately high mortality rate with DKA.[17]  In the setting of DKA, myocardium glucose uptake is inhibited, further contributing to cardiac pump failure. Mesenteric ischemia is also a rare but potentially fatal trigger for DKA. Patients with mesenteric ischemia may present with sepsis or septic shock secondary to necrosed bowel. While diffuse abdominal pain in DKA is common and often attributed to DKA, it is critical to recognize this surgical emergency and to consider underlying etiologies.[18]

Substance Abuse

Substance abuse can trigger DKA through counter-regulatory hormones release and medication noncompliance during drug use. Best documented in cocaine use, the drug increases catecholamine levels by stimulating the release of epinephrine and norepinephrine, and possibly increasing the concentrations of corticotropin and cortisol. [19] Although it’s easy to anchor when patients present with drug use, it’s important to consider the possibility that their use may trigger a serious diabetic crisis.


This article was written in Houston, TX during the time that Hurricane Harvey hit. We never initially considered a hurricane to be a sneaky DKA trigger, but found that many patients were emergently displaced from their homes and without their medications, including insulin or the supplies to self-administer this injectable drug.  Many patients presented in the days following the floods with typical DKA symptoms. While medication non-compliance is a typical DKA trigger, the inciting factor in this case was “sneaky,” and has been previously documented in natural disasters that overwhelmed medical resources. [20]

Case conclusion #1

Patient was admitted to the ICU and found to have an initial troponin I of 5.24 ng/mL and EKG showed new right bundle branch block with nonspecific repolarization abnormalities. He never developed chest pain. His acidosis resolved with appropriate DKA management, his tachypnea resolved and his abdominal pain and nausea improved. Troponins down-trended. Two days later he underwent coronary angiography showing triple-vessel disease. He subsequently underwent triple-vessel coronary artery bypass grafting and was discharged home on a new insulin regimen without restarting his canagliflozin [21].

Case Conclusion #2

Patient was admitted to the ICU and had persistent abdominal pain. His blood pressure dropped to 80/60, HR increased to 120’s despite initial fluid resuscitation and lactate rose to 6.6. Broad spectrum antibiotics were initiated. Further imaging was ordered to assess for concurrent infection or ischemia. CT scan showed pneumatosis intestinalis and portomesenteric vein gas consistent with the diagnosis of mesenteric ischemia. Subsequent CT angiography was without vascular occlusion or thrombus. He was diagnosed with non-occlusive mesenteric ischemia and necrotic bowel. He went for immediate bowel resection and returned to the ICU for post-operative care. He was discharged on post-operative day 4 on insulin.


Key Points

-Always rule out serious life-threatening conditions that may have precipitated a patient’s DKA.  DKA is never simple, and the EM physician’s job is not over with the diagnosis and initiation of the insulin drip.

-Consider DKA in populations with diabetes risk factors, even in patients with no known history of diabetes; particularly in patients suffering CVAs, pregnant patients, patients on high risk medications (olanzapine, risperidone or clozapine) or substance use disorders.

-Make sure to consider DKA on the differential even in euglycemic patients, particularly those on the novel SGLT2 inhibitors.

Persistent tachycardia in DKA in aseptic patients after adequate volume resuscitation may be due to hyperthyroidism.

-Make sure to consider possible underlying infections or ischemia in DKA patients who present with diffuse abdominal pain as attributing this pain solely to DKA may mask the underlying pathology.

Consider cardiac ischemia as MI can be a trigger of DKA, but DKA can also act as a cardiac stress test.

-Consider DKA in the setting of natural and manmade disasters where the supply of insulin may become unavailable to patients.

This post is sponsored by www.ERdocFinder.com, a supporter of FOAM and medical education, who with their sponsorship are making FOAM material more accessible to emergency physicians around the world.

Screen Shot 2017-03-12 at 5.28.13 PM

References/Further Reading:

[1] Lebovitz HE: Diabetic ketoacidosis. Lancet 1995; 345: 767-772.

[2] Kitabchi, Abbas E. et al. “Hyperglycemic Crises in Adult Patients With Diabetes.” Diabetes Care. American Diabetes Association. July 2009.

[3] Haas B, Eckstein N, Pfeifer V, Mayer P, Hass MD (2014). “Efficacy, safety and regulatory status of SGLT2 inhibitors: focus on canagliflozin”. Nutrition & Diabetes. 4 (11): 143

[4] Peters AL et al. Euglycemic Diabetic Ketoacidosis: A Potential Complication of Treatment With Sodium-Glucose Cotransporter 2 Inhibition. Diabetes Care 2015 Sep; 38(9): 1687-1693.

[5] Kelmenson, Daniel A., et al. “Euglycemic Diabetic Ketoacidosis With Prolonged Glucosuria Associated With the Sodium-Glucose Cotransporter-2 Canagliflozin.” Journal of Investigative Medicine High Impact Case Reports, vol. 5, no. 2, 2017, p. 1-3, doi:10.1177/2324709617712736.

[6] Taylor SI, Blau JE, Rother KI. SGLT2 Inhibitors May Predispose to Ketoacidosis. J Clin Endocrinol Metab 2015; 100:284

[7] Monami, Matteo et al. Effects of SGLT-2 inhibitors on diabetic ketoacidosis: A meta-analysis of randomised controlled trials. Diabetes Research and Clinical Practice, Volume 130, 53 – 60

[8] DK Agarwal, T Jeloka, AP Sharma, RK Sharma (2002) Steroid induced diabetes mellitus presenting as diabetic ketoacidosis. Indian J Nephrol 12: 122-123.

[9] Jaja, T, et al. “Steroid Induced Diabetic Ketoacidosis (DKA) in a 13 Year Old Female with Renal Disorder.” Pediatrics & Therapeutics, vol. 02, no. 01, 2012, doi:10.4172/2161-0665.1000109.

[10] Jovanovic, Aleksandar, et al. “Stroke and Diabetic Ketoacidosis – some diagnostic and therapeutic considerations.” Vascular Health and Risk Management, 2014, p. 201

[11] Kamalakannan, D, et al. “Diabetic Ketoacidosis in Pregnancy.” Postgraduate Medical Journal, BMJ Group, Aug. 2003, www.ncbi.nlm.nih.gov/pmc/articles/PMC1742779/.

[12] Vuk, et al. Diabetic Ketoacidosis associated with antipsychotic drugs: Case reports and a review of literature. Psychiatria Danubina, 2017; Vol. 29, No. 2, pp 121–135

[13] Kumar PNS, Thomas B. Hyperglycemia associated with olanzapine treatment. Indian Journal of Psychiatry. 2011;53(2):176-177. doi:10.4103/0019-5545.82562.

[14] Ramaswamy K, Kozma CM, Nasrallah H. Risk of diabetic ketoacidosis after exposure to risperidone or olanzapine. Drug Saf. 2007;30(7):589-99.

[15] Bhattacharyya A, Wiles PG. Diabetic ketoacidosis precipitated by thyrotoxicosis. Postgraduate Medical Journal 1999;75:291-293.

[16] Azoulay, E., et al. “Infection as a Trigger of Diabetic Ketoacidosis in Intensive Care–Unit Patients.” Clinical Infectious Diseases, vol. 32, no. 1, Jan. 2001, pp. 30–35, doi:10.1086/317554.

[17] Al-Mallah, M., et al. Positive Troponin in Diabetic Ketoacidosis without Evident Acute Coronary Syndrome Predicts Adverse Cardiac Events. Clin. Cardiol. 31, 67–71 (2008)

[18] Gocho, N. et al. “Non-occlusive Mesenteric Ischemia with Diabetic Ketoacidosis and Lactic Acidosis Following the Administration of a Sodium Glucose Co-transporter 2 Inhibitor.” Intern Med 55: 1755-1760, 2016)

[19] Warner, Elizabeth A., et al. “Diabetic Ketoacidosis Associated With Cocaine Use.” Archives of Internal Medicine, vol. 158, no. 16, 1998, p. 1799, doi:10.1001/archinte.158.16.1799.

[20] Cefalu, W. T. “The Hurricane Katrina Aftermath and Its Impact on Diabetes Care: Observations from ‘Ground Zero’: Lessons in Disaster Preparedness of People with Diabetes.” Diabetes Care, vol. 29, no. 1, Jan. 2006, pp. 158–160., doi:10.2337/diacare.29.1.158.

[21] Gil-Perdono, J., et al. “Diabetic Ketoacidosis Including myocardial infarction secondary to treatment with dapagliflozin: a case report” Clinical Case Reports 2017; 5(6): 809–811.

3 thoughts on “Diabetic Ketoacidosis: “Sneaky” triggers and clinical pearls”

Leave a Reply

Your email address will not be published. Required fields are marked *