Mild DKA in the ED: Who Can Skip the Insulin Drip

A 29-year-old warehouse worker with type 1 diabetes ran out of insulin four days ago and arrives vomiting and short of breath. His glucose is 410 mg/dL, beta-hydroxybutyrate 4.8 mmol/L, venous pH 7.21, bicarbonate 13 mmol/L and potassium 4.9 mmol/L. He is alert, his pressure holds after the first liter, and the chest film and urine are clean. The order set wants an insulin drip, and the insulin drip wants an ICU bed the hospital does not have tonight.

He may not need either. The 2024 hyperglycemic crises consensus allows subcutaneous rapid-acting insulin for uncomplicated mild or moderate DKA, given in the ED or a step-down unit [1]. The work is in choosing the patient and running the checks.

Confirm it is DKA by the 2024 criteria

The 2024 consensus from the ADA, EASD, JBDS, AACE and DTS replaced the 2009 statement [1]. All three parts of the diagnosis must be present [1]:

• D: glucose of 200 mg/dL or more, or a known history of diabetes at any glucose.
• K: beta-hydroxybutyrate of 3.0 mmol/L or more. Urine ketones of 2+ or more are the fallback; the consensus recommends measuring blood beta-hydroxybutyrate.
• A: pH below 7.3, bicarbonate below 18 mmol/L, or both.

Two changes matter at triage. The glucose threshold dropped from above 250 mg/dL to 200 mg/dL, and a known history of diabetes now meets the criterion at any glucose [1]. About 10% of patients present with euglycemic DKA, a glucose below 200 mg/dL with ketoacidosis [1]. A patient on an SGLT2 inhibitor with a normal fingerstick still needs a ketone level.

The criteria also sort out who is in the room. In one academic ED, 1,676 adults screened positive by point-of-care glucose and ketones or were given a DKA diagnosis [2]. Only 53% met laboratory criteria. Of 740 who screened positive but did not meet criteria, 31% were still diagnosed with DKA. The rest included plain hyperglycemia, starvation ketosis, HHS and infection [2]. The authors read this as DKA being partly a clinical diagnosis. For pathway selection, the safer reading is to write down which criteria the patient met before choosing a route.

Grade severity before you choose the route

The consensus grades severity on ketones, acid-base status and mental status [1]:

| | Mild | Moderate | Severe | |---|---|---|---| | Beta-hydroxybutyrate | 3.0 to 6.0 mmol/L | 3.0 to 6.0 mmol/L | above 6.0 mmol/L | | Acidosis | pH above 7.25 to below 7.30, or bicarbonate 15 to 18 | pH 7.0 to 7.25, bicarbonate 10 to below 15 | pH below 7.0, bicarbonate below 10 | | Mental status | Alert | Alert or drowsy | Stupor or coma | | Suggested level of care | Regular or observation unit | Step-down or intermediate care | ICU |

Not every variable has to fit one column, and the consensus leaves the level of care to clinical judgment [1]. It is explicit on one point: subcutaneous rapid-acting insulin is not recommended for severe or complicated DKA, or for HHS [1]. The consensus sends to the ICU patients with severe DKA or HHS, altered mental status, or critical illness as the precipitating cause, such as myocardial infarction, gastrointestinal bleeding or sepsis [1]. The patient in the opening scene is moderate, alert and has no such precipitant. He is a candidate.

What the evidence for subcutaneous insulin shows

The randomized evidence is small. A 2016 Cochrane review found 5 trials with 201 participants, all with mild or moderate DKA, comparing subcutaneous lispro or aspart with intravenous regular insulin [3]. No deaths were reported. For lispro (2 trials, 90 patients), time to resolution differed by 0.2 hours (95% CI -1.7 to 2.1, very low-quality evidence). Hypoglycemia occurred in 6 of 80 lispro patients and 9 of 76 on intravenous insulin (risk ratio 0.59, 95% CI 0.23 to 1.52). The reviewers found neither advantages nor disadvantages, on mostly low- to very low-quality evidence [3].

The most detailed emergency department data come from one center. The SQuID protocol (subcutaneous insulin in DKA) at an urban academic ED treated 78 mild-to-moderate patients subcutaneously and 99 with a non-ICU insulin infusion [4]. Rescue dextrose for hypoglycemia was no more frequent, and median ED length of stay was 3.0 hours shorter. The follow-up study widened eligibility and admitted patients to a regular floor [5]. Of 84 patients, 62 were treated on the protocol. Rescue dextrose was needed in 8% versus 18% on intravenous insulin, a difference that was not statistically significant. Time on protocol was 10.4 hours shorter. ICU admission for mild-to-moderate DKA fell from 54% to under 21% after the protocol started [5]. A retrospective multicenter cohort of 257 patients with mild-to-moderate DKA found no difference in time to resolution between subcutaneous and intravenous protocols, and more hypoglycemia with the intravenous protocols [7].

All three studies are observational, and the SQuID ICU figure is a before-and-after comparison. The protocol also depends on staffing: dosing and glucose checks every 1 to 2 hours [1,4]. A department that cannot meet that rhythm should not adopt the pathway.

Running the subcutaneous pathway

The route of insulin changes. The rest of the consensus does not [1]:

• Fluids. Isotonic saline or a balanced crystalloid at 500 to 1,000 mL/hr for the first 2 to 4 hours, with smaller boluses (for example 250 mL) in older adults and in patients with heart failure or end-stage kidney disease on dialysis.
• Potassium. Start replacement once potassium is below 5.0 mmol/L, aiming for 4 to 5 mmol/L; 20 to 30 mmol per liter of fluid covers most patients. If the first potassium is below 3.5 mmol/L, give 10 mmol/hr and hold insulin until it rises above 3.5. Recheck 2 hours after the first insulin dose, then every 4 hours. For the patient in the scene, at 4.9, potassium goes into the first bag of maintenance fluid.
• Insulin. Rapid-acting analog subcutaneously every 1 to 2 hours [1]. SQuID added glargine as a protocol step [5]. The consensus data on early low-dose basal insulin (0.15 to 0.3 U/kg) come from patients on insulin infusions [1].
• Dextrose. When glucose falls below 250 mg/dL, add 5% to 10% dextrose and keep treating the ketones.
• Bicarbonate. Consider it only when pH is below 7.0.
• Resolution. Ketones below 0.6 mmol/L plus venous pH 7.3 or higher or bicarbonate 18 mmol/L or higher. The consensus says not to use the anion gap, which saline can hold open, or urine ketones [1].

The fluid choice is not settled. The consensus cites faster resolution with balanced crystalloids and accepts either fluid [1]. In a subgroup of 172 DKA patients from two cluster randomized trials, median time to resolution was 13.0 hours with balanced crystalloids and 16.9 hours with saline [8]. The only dedicated ED randomized trial, a 52-patient Canadian pilot built to test feasibility, found 15.7 hours with Ringer's lactate and 12.7 hours with saline [6]. The balance leans toward balanced crystalloids, and saline remains acceptable.

For the potassium side of the problem, see protecting the heart in hyperkalemia. When the precipitant is infection, the sepsis bundle runs alongside this one.

Key Takeaways

• DKA needs all three: glucose 200 mg/dL or more or known diabetes, beta-hydroxybutyrate 3.0 mmol/L or more, and pH below 7.3 or bicarbonate below 18.
• Subcutaneous rapid-acting insulin is an option for uncomplicated mild or moderate DKA. It is not an option for pH below 7.0, bicarbonate below 10, beta-hydroxybutyrate above 6.0, altered mental status, HHS, or a critical-illness precipitant.
• If the first potassium is below 3.5 mmol/L, hold insulin and give potassium at 10 mmol/hr until it is above 3.5.
• Call resolution on ketones below 0.6 mmol/L plus pH 7.3 or higher or bicarbonate 18 or higher, not on the anion gap.

FAQ

Can mild DKA be treated without an insulin drip?

Yes, if it is uncomplicated. The 2024 consensus supports subcutaneous rapid-acting insulin analogs every 1 to 2 hours for mild or moderate DKA, delivered in the ED or a step-down unit [1]. A Cochrane review of 5 small trials found neither advantages nor disadvantages compared with intravenous insulin, on low-quality evidence [3].

What are the 2024 DKA diagnostic criteria?

Glucose of 200 mg/dL or more or a history of diabetes, beta-hydroxybutyrate of 3.0 mmol/L or more, and pH below 7.3 or bicarbonate below 18 mmol/L [1]. All three must be present. Urine ketones of 2+ are the fallback when blood ketones are not available.

When is DKA considered resolved?

When ketones are below 0.6 mmol/L and venous pH is 7.3 or higher or bicarbonate is 18 mmol/L or higher [1]. Glucose should ideally be below 200 mg/dL. Do not use anion gap closure or urine ketones to call it.

Dr. Chet's Take

I have spent more than 25 years as an emergency department attending, and the insulin drip has long been one of the most reliable ways to turn a stable patient into a boarder. This article is right about where the real decision sits. The question was never whether subcutaneous insulin lowers glucose. It was whether we can pick the right patient and check him often enough. The consensus has now said, in print, that a moderate DKA patient does not need an ICU bed just to receive insulin. That changes the conversation with the intensivist and with the bed board, and it gives the emergency physician a document to point to at two in the morning. It does not change the physiology. The potassium rule still decides who gets insulin first, and the ketones still decide when the work is done.

That being said, the evidence is thinner than the enthusiasm. The randomized trials are small and old, and the most detailed emergency department data come from one academic center that built and studied its own protocol. The honest answer is that I do not know how this performs in a department where one nurse holds four rooms and the fingerstick every hour slips to every three. A protocol that depends on a clock fails when the clock is not kept. Before you copy the pathway, count your nurses at 0300. The safety in those studies came from the checks, and the checks come from staffing. I would also want to know how many of the subcutaneous patients bounced back within a week, a number the published abstracts do not report.

If you are leading an emergency department, write the consensus exclusions into the order set as hard stops, so nobody at the end of a long shift has to remember them. Build the beta-hydroxybutyrate into the admission panel so resolution has a number attached to it. Then audit the first 30 patients for rescue dextrose, missed checks and return visits before you call the pathway a success. If you are the attending, grade the severity in your note before you pick the route, and say in the note why the patient qualifies. Teach residents to name the column on the severity table out loud at the bedside. The patient who looks well on arrival still has a potassium that can fall fast.

— Chester Shermer, MD, FACEP | Emergency Medicine, 25+ Years Clinical Experience | State Surgeon

Continue Your Training

If you're an emergency physician (or any clinician treating patients daily) trying to understand how AI will actually impact your clinical practice — not just the hype — I put together a free practical guide. You can download the AI in EM Survival Guide here.

Browse All Courses at Global MedOps Command

Relevant Reading on Global MedOps Command:

How to Avoid Becoming an AI Casualty — Dr. Shermer's guide to navigating AI tools in clinical and operational settings without compromising judgment or patient outcomes.

Emergency Department Efficiency Playbook — Practical systems for throughput, triage optimization, and operational efficiency.

Read more from Dr. Shermer on Medium

Connect with Dr. Shermer: LinkedIn — Chester "Chet" Shermer, MD, FACEP.

References

1. Umpierrez GE, Davis GM, ElSayed NA, et al. Hyperglycemic Crises in Adults With Diabetes: A Consensus Report. Diabetes Care. 2024;47(8):1257-1275. PubMed 39052901 · Co-published in Diabetologia. 2024;67(8):1455-1479. PubMed 38907161
2. Griffey RT, Haas N, Schneider RM, et al. How Often Are Emergency Patients Diagnosed With Diabetic Ketoacidosis Despite Not Meeting Laboratory Criteria? Acad Emerg Med. 2026;33(5):e70315. PubMed 42089584
3. Andrade-Castellanos CA, Colunga-Lozano LE, Delgado-Figueroa N, et al. Subcutaneous rapid-acting insulin analogues for diabetic ketoacidosis. Cochrane Database Syst Rev. 2016;(1):CD011281. PubMed 26798030
4. Griffey RT, Schneider RM, Girardi M, et al. The SQuID protocol (subcutaneous insulin in diabetic ketoacidosis): Impacts on ED operational metrics. Acad Emerg Med. 2023;30(8):800-808. PubMed 36775281
5. Griffey RT, Schneider RM, Girardi M, et al. SQuID (subcutaneous insulin in diabetic ketoacidosis) II: Clinical and operational effectiveness. Acad Emerg Med. 2025;32(1):61-71. PubMed 39308229
6. Yan JW, Slim A, Van Aarsen K, et al. Balanced crystalloids (RInger's lactate) versus normal Saline in adults with diabetic Ketoacidosis in the Emergency Department (BRISK-ED): a pilot randomised controlled trial. Emerg Med J. 2024;41(2):103-111. PubMed 38050056
7. Stuhr K, LeeMaster R, Hickman AW, et al. Subcutaneous Insulin Versus Traditional Intravenous Insulin Infusion in Treatment of Mild to Moderate Diabetic Ketoacidosis. J Emerg Med. 2023;65(3):e221-e228. PubMed 37689412
8. Self WH, Evans CS, Jenkins CA, et al. Clinical Effects of Balanced Crystalloids vs Saline in Adults With Diabetic Ketoacidosis: A Subgroup Analysis of Cluster Randomized Clinical Trials. JAMA Netw Open. 2020;3(11):e2024596. PubMed 33196806

Related simulation training: Books by Dr. Shermer