The SQuID Protocol: SQ Insulin in DKA?

Background: DKA is traditionally treated with fluid resuscitation, electrolyte replacement, and intravenous infusions of insulin. However, it is unclear if all degrees (mild, moderate, severe) of DKA require the same intensive treatment. Mild to moderate DKA represents a subgroup of patients that often require admission to the ICU due to hospital policies not allowing insulin infusions outside of this clinical setting. One solution to decrease ICU admission is to treat these patients in the ED until their anion gap is closed and the patient can be transitioned from IV insulin infusion to SQ insulin.  This is not an optimal solution as these patients still require long stays in the ED, EDs are overcrowded as it is, the staffing is not the same in an ED compared to an ICU (don’t have the same nursing to patient ratios), all while more patients continue to come into the ED with other emergent conditions. A more optimal, and promising, solution may be the use of SQ insulin in mild to moderate DKA.

Paper: Griffey RT et al. The SQuID Protocol (Subcutaneous Insulin in /diabetic Ketoacidosis): Impacts on ED Operational Metrics. Acad Emerg Med 2023. PMID: 36775281

Clinical Question: In adult patients with mild to moderate DKA, does a subcutaneous insulin protocol reduce ED length of stay compared to a traditional IV infusion DKA protocol?

What They Did:

  • Implemented the SQuID (Subcutaneous Insulin in Diabetic Ketoacidosis) protocol for adults with mild to moderate DKA in an urban academic ED
  • Prospectively derived, quasi experimental (pre-post) study
  • Authors used retrospectively obtained electronic data in evaluating their outcome measures
  • DKA severity was stratified as mild, moderate, or severe
  • Not explicitly stated in the manuscript but it seems reading between the lines that initiating SQuID protocol or traditional insulin infusion was left to judgement of clinician once mild to moderate DKA was diagnosed

SQuID Protocol

Traditional” IV Infusion DKA Protocol


  • Fidelity: Frequency of required q2hr glucose checks
  • Safety: Proportion of patients administered rescue dextrose for hypoglycemia
  • Operational: ED LOS and ICU Admission Rate
    • ED LOS was the primary outcome of this trial


  • Adult ED patients meeting criteria for Mild to Moderate DKA
    • Hyperglycemia (Glucose >300)
    • Positive ketone test (≥1.1mmol/L)
    • Presence of an anion gap
    • Discharge diagnosis of DKA and having received insulin


  • Severe DKA
    • HCO3 <10mmol/L
    • Arterial pH < 7.0
  • Patients <18 years of age
  • Pregnancy
  • Concomitant serious infections
  • Concerns for MI
  • AMS
  • Active comorbidities (ESRD, CHF, Active use of inmmunosuppressants)
  • Need for surgical procedure
  • Determination by the ED or inpatient team that a patient was too ill for the designated floor


  • 177 mild to moderate DKA adult patients
    • SQuID: 78pts
    • Traditional cohort: 99pts
    • Historical controls:
      • Preintervention: 163pts (Nov 1st, 2020 to May 31st, 2021)
      • Pre-COVID: 161pts (Aug 1st, 2019 to Feb 28th 2020)
    • Fidelity (Frequency of q2hr Glucose Checks)
      • SQuID: 1.0 tests/hr (0.8 to 1.1)
      • Traditional: 1.0 tests/hr (0.8 to 1.1)
      • Pre-Intervention: 1.0 tests/hr (0.9 to 1.2); No Reduction/Increase 0.0%
      • Pre-COVID: 1.0 tests/hr (0.8 to 1.1); No Reduction/Increase 0.0%
      • High fidelity for patients on the SQuID pathway
      • Performance of blood glucose checks exceeded the q2hr testing requirement: Median 1.0 tests/hr
    • Safety (Proportion of patients administered rescue dextrose for hypoglycemia)
      • SQuID: 2.7%
      • Traditional: 3.6%
      • Pre-Intervention: 2.3%; Decrease by 0.4%
      • Pre-COVID: 0%; Increase by 2.7%
      • No statistical differences in safety between groups
    • Operational Impact
      • Median ED LOS (PRIMARY OUTCOME):
        • SQuID: 8.9hrs (6.5 to 11.5hrs)
        • Traditional: 11.9hrs (9.6 to 18.6hrs)
        • Pre-Intervention: 10.3hrs (7.7 to 12.9hrs); Reduction by 1.4hrs
        • Pre-COVID: 12.5hrs (8.7 to 19.3hrs); Reduction by 3.6hrs
        • ED LOS was shorter for the SQuID cohort
      • ICU Admissions:
        • SQuID & Traditional (Post Intervention): 42.9%
        • Pre-Intervention: 46.0%; Reduction by 3.1%
        • Pre-COVID: 49.1%; Reduction by 6.2%
        • Not Statistically Significant


  • Asks a clinically important question
  • Clearly defined protocol that can be reproduced at other institutions
  • Evaluated historical pre-intervention and pre-COVID patients to ensure the findings were consistent
  • Performed a sensitivity analysis and quantile regression model to confirm findings of the primary outcome


  • SQuID protocol required more frequent glucose monitoring (median q1hr) which may be too much for medical floors and EDs
  • ED LOS and ICU admission is determined by subjective clinical decisions and hospital policy. This will likely vary from institution to institution
  • Single center study at a busy level 1 academic urban trauma center which may limit generalizability to smaller community hospitals
  • As this is a prospectively-derived quasi-experimental (pre-post) study there are most likely confounders not accounted for that only a RCT would be able to discern (Before and after design doesn’t take into account changes that may have occurred over time that weren’t studied but could affect outcomes)
  • Hypoglycemia events were a secondary outcome and therefore this trial was not powered appropriately for this outcome. Additionally compared to historical controls hypoglycemia events seem to be increased
  • Unclear why we don’t have ICU admission just in the SQuID group instead of a combination of the two (SQuID and Traditional IV Insulin)


  • The SQuID protocol did reduce ED LOS (by a few hours), reduced ICU admissions (although not statistically significant), and had equivalent safety compared to a traditional insulin infusion pathway in patients with mild to moderate DKA
  • SQ insulin for DKA was considered an existing standard of care rather than an experimental protocol based on the published literature
  • Overall patients in the SQuID protocol were younger, more likely to be female, had lower acuity on emergency severity index, and more likely to be discharged from the ED compared to patients in the traditional protocol
  • SQuID protocol will require significant education including education sessions not just for the ED, but hospitalists, ICU, and nurses throughout the hospital before safe implementation
    • There were several patients misclassified who actually had severe DKA or met exclusion criteria for which SQuID was started
    • A few cases of under-dosing or delayed administration of long-acting insulin which resulted in recurrence of hyperglycemia
    • Failure to initiate dextrose-containing fluids in a timely fashion

Author Conclusion: “In this single-center study at an academic ED, treatment of patients with MTM-severity DKA with a SQ insulin protocol was effective, demonstrated equivalent safety, and reduced ED length of stay.”

Clinical Take Home Point: For patients with mild to moderate DKA, the SQuID protocol (SQ insulin) has the potential to be an alternative treatment pathway to decrease ED LOS.  However, the ED LOS was only decreased by 3 hours, the ICU admission rate was only decreased by several percentage points, the need for hospital wide education to implement this protocol, the frequency of glucose monitoring (q1hrs), and unclear safety (secondary outcome that seems to be increased compared to historical controls) this protocol is not ready for primetime.


  1. Griffey RT et al. The SQuID Protocol (Subcutaneous Insulin in /diabetic Ketoacidosis): Impacts on ED Operational Metrics. Acad Emerg Med 2023. PMID: 36775281

Post Peer Reviewed By: Anand Swaminathan, MD (Twitter: @EMSwami)

Cite this article as: Salim Rezaie, "The SQuID Protocol: SQ Insulin in DKA?", REBEL EM blog, October 2, 2023. Available at:

Like this article?

Share on Facebook
Share on Twitter
Share on Linkdin
Share via Email

Want to support rebelem?