Hospital-Acquired Diabetic Ketoacidosis.
Case #1: A 46-year-old Hindi-speaking resident of a skilled nursing facility (SNF) with a history of end-stage renal disease (ESRD), hypertension, poorly controlled type 1 diabetes complicated by polyneuropathy, drug-induced hepatitis, and duodenal ulcer, presented to the emergency department (ED) with asymptomatic post-dialysis hypertension, with a blood pressure of 246/100. He reported taking his regular blood pressure medications without any clear trigger for this hypertensive episode. He was treated in the ED with hydralazine and clonidine, with mild improvement in blood pressure. His chemistry panel included a serum glucose of 292 mg/dl and a calculated normal anion gap of 11. He was then discharged back to his SNF, as he remained asymptomatic, with recommendations to resume his usual antihypertensive medications.
The patient was placed in the ED hallway waiting for COVID-19 test results, as asymptomatic testing is required for all patients returning to SNFs. No nurse was assigned to the patient in the hallway, and the patient was given a multitude of small carbohydrate-containing meals (turkey sandwich, juice, crackers) by several ED staff. During the patient's 16-hour ED course, he did not receive any basal insulin; he only received 5 units of aspart and 10 units of regular insulin. When a nurse later passed the patient, and he expressed feelings of hunger, she checked his point-of-care glucose, which only read HI (>600 mg/dL), so a specimen was sent to the laboratory. Results included a serum glucose of 787 mg/dL with a markedly elevated anion gap of 22. The patient was started on a continuous insulin infusion and transitioned to subcutaneous insulin as he recovered from diabetic ketoacidosis (DKA).
Case #2: A 49-year-old male patient with history of poorly controlled type 1 diabetes, homelessness, injection drug use, and mental health concerns presented to the ED with nausea, and left thumb pain, swelling and discoloration for two days. He was seen by an orthopedic surgeon and was admitted for treatment of infection and hyperglycemia. Only correctional (short-acting) insulin was ordered. He was seen by the Inpatient Glycemic Team the next morning, which noted hyperglycemia, a slightly elevated anion gap of 14, and serum ketones. An insulin drip was started for treatment of DKA. After a few hours, basal (long-acting) insulin was administered, and the infusion was stopped. Treatment continued for his thumb cellulitis.
By Dahlia Zuidema, PharmD, Berit Bagley, MSN, and Charity L Tan MSN
In the first case, a patient with type 1 diabetes and ESRD presented to the ED with hypertensive urgency, which was treated appropriately. However, his transfer back to the SNF where he lived was delayed, and the ED staff (having officially discharged the patient) failed to attend to his need for basal insulin, allowing him to develop diabetic ketoacidosis (DKA). In the second case, a patient with type 1 diabetes, homelessness, and inconsistent access to food and housing, presented to the ED with a thumb infection and hyperglycemia. Basal insulin was withheld, presumably because of his poor oral intake and unknown outpatient dosing, allowing this patient also to develop DKA within an acute care hospital.
Type 1 diabetes is an autoimmune disease in which beta cells of the pancreatic islets of Langerhans do not produce the hormone insulin; therefore, exogeneous insulin is always required, even when patients are not eating or drinking. As highlighted in other PSNet commentaries, physiologic insulin replacement includes basal (long-acting) insulin to cover the body’s sugar production, nutritional (short-acting) insulin to cover carbohydrate intake, and a correction to account for any mismatch. The use of sliding-scale algorithms to cover hyperglycemia in the setting of type 1 diabetes is ineffective and can result in DKA, as these two cases demonstrate.
Several factors can contribute to the development or recurrence of DKA, including housing and food insecurity, substance use, low health literacy, chronically poor glucose control with a hemoglobin A1c (A1c) of 8.5% or greater, psychiatric disorders such as depression, and a history of non-adherence to insulin, for either economic or behavioral reasons.1,2,3 For example, Case #1 involved a non-English speaking patient with the potential of having low health literacy, limited recent involvement in management of his disease, and possibly miscommunication regarding the timing of prior insulin doses.1 Patients with chronic kidney disease (CKD) are in a chronic state of metabolic acidosis, which makes diagnosing DKA challenging. The incidence of DKA in this population is less than in patients with preserved kidney function as insulin clearance is decreased by CKD.4
Case #2 involved a patient with higher insulin needs due to infection, which is associated with increased secretion of endogenous steroids, glucagon, and epinephrine. These stress hormones drive gluconeogenesis, which can result in ketone formation in patients with type 1 diabetes who are unable to increase endogenous insulin production. As in Case #1, low health literacy and erratic oral intake may have increased the risk of hospital-acquired DKA (HADKA).
The prevalence of both type 1 and type 2 diabetes is increasing, and ED visits for these patients are common. The Centers for Disease Control and Prevention (CDC) estimates that 37.3 million Americans have diabetes in 2022, representing 11.3% of the US population.5 About 23% of adults with diabetes are undiagnosed and are therefore at risk for unsuspected DKA.5 For all patients with type 1 diabetes and many patients with long-standing type 2 diabetes, insulin is the primary therapy. When patients are acutely ill from trauma, infection, or cardiorespiratory disease, they are under constant physiologic stress resulting in higher concentrations of counterregulatory hormones (e.g., glucagon, cortisol, catecholamines), causing hyperglycemia. Pre-admission insulin doses may be insufficient to maintain blood glucose in target range, leading to DKA, even in patients with type 2 diabetes. Knowing the patient’s correct insulin regimen, frequency of dosing, and time of last dose can be helpful in determining insulin dosing plans and glucose monitoring needs when patients present to the hospital. Addressing access and cost concerns may reveal potential barriers causing insulin omission and underdosing.
The diagnostic criteria for DKA include severe hyperglycemia (i.e., glucose >250 mg/dl), metabolic acidosis, and ketonemia (i.e., beta-hydroxybutyrate >3 mmol/L).1,4 DKA should be in the differential diagnosis for all patients with type 1 diabetes whose acidosis exceeds normal limits, or their own baseline value. The beta-hydroxybutyrate level is useful for diagnosing DKA even among patients with CKD on hemodialysis;6however, beta-hydroxybutyrate levels tend to be lower among ESRD patients with DKA than among patients with preserved renal function and DKA, despite higher glucose levels and greater anion gaps among the former patients.4 Typically, an anion gap greater than 20 is indicative of DKA in hemodialysis-dependent patients.6
The three pillars of DKA management are insulin therapy, fluid resuscitation, and electrolyte replacement (i.e., potassium, magnesium, and phosphorus).4 In addition, the causative factor of DKA should be addressed.4 In Case #1, the primary causative factors were the lack of nutritional insulin with food eaten and the lack of basal insulin for at least 16 hours during the patient’s extended stay in the ED while awaiting COVID test results. In case #2, the primary causative factors were the lack of basal insulin overnight and the patient’s underlying bacterial infection.
Two recent studies have reported on the incidence of HADKA.7,8 In a retrospective study by Ranasinghe and Acharya7 from two Australian hospitals with 47,622 diabetes-related admissions, 940 patients had documented DKA, of which 12 were hospital-acquired. Most of these patients had type 2 diabetes (75%, n=9), were elderly, and their mean A1c was 8.4% (while the threshold for diabetes is 6.5% or greater). The most common reason for HADKA was related to inadequate insulin administration, including withholding insulin, insulin dose reduction, and interrupted insulin infusion. The remaining cases were caused by infection, steroid-induced hyperglycemia, acute stress from cardiac disease with cardiogenic shock, and use of sodium-glucose cotransporter-2 (SGLT2) inhibitors, all of which are known to increase the risk of DKA.8 In an English retrospective study by Williams and colleagues,9 25 patients were diagnosed with HADKA over a period of six years (i.e., 2016-2022). Nearly half (48%) of these patients had type 2 diabetes. Again, the most common causes of HADKA were insulin administration errors, omission, and inappropriate dose reductions.
In both the cases described above, a medical history with best possible medication history would have identified the delayed insulin orders and stimulated timely management of hyperglycemia. Indeed, omission of basal and/or nutritional insulin in a patient with type 1 diabetes is an event that should never happen in a hospital. An increasing number of guidelines, order sets and clinical decision support tools are available to prevent this process failure.
Approach to Improving Safety
Both the cases described above present opportunities for system improvements. The ED can be a hectic environment with many provider hand-offs and shift changes between bedside nurses. In both cases, there was a lack of provider knowledge and decision support with respect to type 1 diabetes, as patients with type 1 diabetes are dependent on an insulin regimen that includes basal, nutritional, and correctional components. The use of correctional insulin without a predetermined insulin treatment plan can lead to omission of insulin doses and misinterpretation of blood glucose values. The Multi-Center Medication Reconciliation Quality Improvement Study (MARQUIS), sponsored by AHRQ and organized by the Society for Hospital Medicine, supported high-quality medication reconciliation, where the goal is to prevent unintended medication discrepancies at transitions in care, as these inconsistencies may lead to adverse events.10 In many EDs, pharmacy technicians focus on performing medication reconciliation on patients with the highest risk of medication errors. Various approaches are used to obtain the most accurate information, including use of at least two sources, resolution of discrepancies between sources, asking open ended questions, asking about adherence, and for people with diabetes, asking about injectable products and tools for checking blood glucose.
The American Diabetes Association Standards of Care, Chapter 16.3, recommends consulting with a specialized diabetes or glucose management team to assist in the care of hospitalized patients with diabetes, whenever possible.11 For example, in some hospitals an Inpatient Glycemic Team is available for specialized diabetes-related questions every day.
The Joint Commission has also launched a Speak Up Campaign, which highlights the importance of patients with diabetes advocating for themselves.12 The Campaign provides five ways for these patients to be active in their care in the hospital, to help prevent adverse events:
- Find out how your diabetes will be managed. Specific suggestions include reminding caregivers about diabetes, wearing a diabetes identification tag, asking about the hospital’s diabetes care and glucose monitoring plan, asking for basal insulin orders to be written or updated, and telling the nurse immediately about symptoms of hypoglycemia or hyperglycemia.
- Ask what will happen with your medicines.
- Know what will happen with your diet.
- Avoid getting an infection.
- Find out what will happen when you go home. Specific suggestions include having your doctor explain what you can expect after leaving the hospital, asking about follow-up care, and making sure that instructions are understood.
Systems Changes Needed
Moving forward, it is important to improve the knowledge of clinicians treating insulin-dependent diabetic patients. Interprofessional communication about care plans or care transitions is essential, as nurses and pharmacists have critical roles in advocating for safe management of diabetes in seriously ill, hospitalized patients. Diabetes specialist teams should be consulted automatically for patients at high risk of DKA, including all hospitalized patients with type 1 diabetes and those with type 2 diabetes who have risk factors such as blood glucose over 250 mg/dL, trauma, infection, or shock. Medication reconciliation technicians should provide an early and complete medication history to care providers to assist clinical decision making.
To facilitate timely clinical decision-making, hospitals should create algorithmic decision support tools and standardized order sets for glycemic control in the ED and other high-risk settings. These decision support tools should incorporate the absolute need to give insulin to patients with type 1 diabetes and advise caution with withdrawal or omission of insulin among patients with type 2 diabetes who have increased physiological needs.
Other system improvements may include alerting providers when patients have an increasing anion gap or other early signs of DKA. When point-of-care testing reveals blood sugars over 250 mg/dL, ketones such as beta-hydroxybutyrate should be checked reflexively. In the two cases described above, reflexively testing beta-hydroxybutyrate may have triggered timelier diagnosis of DKA and thus earlier intervention.
Finally, for socially vulnerable patients with homelessness and psychiatric conditions, who often become frequent visitors to the ED, having an individualized, standing diabetes care plan may prevent delays, improve transitions of care, and ensure the availability of needed resources.
Take Home Points
- Develop and implement ED-specific clinical decision support tools and standardized order sets for diabetes management, including specific pathways to consult the inpatient glycemic team.
- Prioritize medication reconciliation in the ED for high-risk patients on insulin to improve communication, avoid medication errors, and increase provider awareness.
- Encourage patients to Speak Up and carry an updated medication list; this can be an important way to improve accuracy of medication reconciliation and prevent errors.
- Recognize that insulin omission or dose reduction in the setting of higher physiological need, such as infection, trauma, or shock, can lead to insulin deprivation and ketone formation.
- Develop system alerts to facilitate earlier recognition of hyperglycemia and early signs of DKA (e.g., increasing anion gap).
- Communicate with patients and caregivers about the goals of therapy and insulin dosing, especially around the time of procedures, dietary changes, and transitions of care; this is essential to improve patients’ understanding and provide support for necessary changes in home insulin regimens.
Dahlia Zuidema, PharmD, BC-ADM, CDCES
Assistant Clinical Professor, UCSF School of Pharmacy
Department of Pharmacy Services
UC Davis Health
Berit Bagley MSN, CDCES, BC-ADM
Inpatient Glycemic Team
Department of Endocrinology, Diabetes and Metabolism
UC Davis Health
Charity L Tan MSN, ACNP-BC, CDCES, BC-ADM
Inpatient Glycemic Team
Department of Endocrinology, Diabetes and Metabolism
UC Davis Health
- Ehrmann D, Kulzer B, Roos T, et al. Risk factors and prevention strategies for diabetic ketoacidosis in people with established type 1 diabetes. Lancet Diabetes Endocrinol. 2020;8(5):436-446. [Available at]
- Lyerla R, Johnson-Rabbett B, Shakally A, et al. Recurrent DKA results in high societal costs - a retrospective study identifying social predictors of recurrence for potential future intervention. Clin Diabetes Endocrinol. 2021;7(1):13. [Free full text]
- Fazeli Farsani S, Brodovicz K, Soleymanlou N, et al. Incidence and prevalence of diabetic ketoacidosis (DKA) among adults with type 1 diabetes mellitus (T1D): a systematic literature review. BMJ Open. 2017;7(7):e016587. [Free full text]
- Dhatariya KK; Joint British Diabetes Societies for Inpatient Care. The management of diabetic ketoacidosis in adults- an updated guideline from the Joint British Diabetes Society for Inpatient Care. Diabet Med. 2022;39(6):e14788. [Free full text]
- National Diabetes Statistics Report website. Centers for Disease Control and Prevention. Accessed December 5, 2022. [Available at]
- Seddik AA, Bashier A, Alhadari AK, et al. Challenges in management of diabetic ketoacidosis in hemodialysis patients, case presentation and review of literature. Diabetes Metab Syndr. 2019;13(4):2481-2487. [Available at]
- Ranasinghe U, Acharya S. Hospital-acquired diabetic ketoacidosis (HADKA) - an analysis comparing two major Australian hospitals. Int J Diabetes Clin Res. 2021;8:140. [Free full text]
- Musso G, Saba F, Cassader M, et al. Diabetic ketoacidosis with SGLT2 inhibitors. BMJ. 2020;371:m4147. [Available at]
- Williams DM, Taverner S, Watson R, et al. A series of patients with hospital-acquired diabetic ketoacidosis (HADKA): a descriptive analysis. Clin Med (Lond). 2022;22(6):549-552. [Free full text]
- MARQUIS Investigators. MARQUIS Implementation Manual: A Guide for Medication Reconciliation Quality Improvement. Society of Hospital Medicine; October 2014. [Available at]
- American Diabetes Association Professional Practice Committee; American Diabetes Association Professional Practice Committee:, Draznin B, et al. 16. Diabetes Care in the Hospital: Standards of Medical Care in Diabetes-2022. Diabetes Care. 2022;45(Suppl 1):S244-S253. [Free full text]
- Speak Up Campaign - Diabetes: Five ways to be active in your care at the hospital. The Joint Commission. Accessed December 5, 2022. [Available at]