SQuID Protocol for DKA: Impact on ED Length-of-Stay
/Clinical Question: What is the impact of a subcutaneous insulin protocol for mild-to-moderate DKA on ED length of stay and ICU admission rates?
Background
Diabetic ketoacidosis (DKA) is a common and resource-intensive condition contributing to significant morbidity and mortality among patients with diabetes mellitus. DKA accounts for approximately 223,000 ED visits annually or 8.9 ED visits per 1000 adults with diabetes based on 2018 data resulting in over 500,000 annual hospital days. This amounts to annual hospital costs of $5.1 billion. In most hospitals, DKA is managed by initiating patients on intravenous insulin infusions requiring admission to an intensive care unit (ICU) or stepdown level of care due to need for frequent laboratory testing and monitoring. With the advent of fast-acting subcutaneous insulin analogs over the past 20 years, multiple studies have demonstrated that subcutaneous insulin protocols for treatment for treatment of mild-to-moderate severity DKA are safe and cost-effective when compared to intravenous insulin infusion protocols in non-ICU settings. In the context of increasing ED and hospital crowding as well as limited ICU availability, this study aimed to evaluate the impact of a subcutaneous insulin (SQuID) protocol in the treatment of DKA on emergency department length-of-stay (EDLOS) and ICU admissions.
Study Design
This was a prospectively-derived single center pre-post study comparing the impact of the SQuID protocol with a traditional intravenous insulin infusion protocol for patients with mild-to-moderate DKA.
Inclusion and Exclusion Criteria
This study included patients with hyperglycemia (fingerstick blood glucose >300 mg/dL) and positive point-of-care ketone testing (>1.1 mmol/L). Patients with severe DKA (defined as HCO3 <10 mmol/L or pH <7.0) as well as patients less than 18 years old were excluded from analysis. Additional exclusion criteria from the SQuID protocol included the following:
Pregnancy
Concomitant serious infection
Concerns for myocardial infarction
Altered mental status
Active comorbidities (i.e., end-stage renal disease, congestive heart failure, active use of immunosuppressants)
Need for a surgical procedure
Determination by the ED or inpatient team that a patient was too ill for the designated floor (an inpatient observation unit run by hospitalist physicians)
Methods
The SQuID protocol was launched in June 2021 and data was captured on patients following a pre-planned 5-week washout period. In this protocol, all diabetic patients who presented to triage were screened with fingerstick blood glucose testing, and if above 300 mg/dL, point-of-care ketone testing was performed. If positive (>1.1 mmol/L), patients would automatically have additional laboratory studies drawn including a basic metabolic panel, whole blood potassium, and a venous blood gas and a best practice advisory (BPA) in the electronic medical record notified clinicians of potential candidates for the study pathway. Clinicians then determined if patients met laboratory criteria for DKA and assessed if patients were appropriate for the SQuID protocol versus the traditional intravenous insulin infusion protocol. After ED evaluation was complete, patients in the SQuID protocol were admitted to an inpatient observation unit staffed by hospitalists whereas patients in the traditional protocol were admitted to the ICU.
The study’s primary outcomes were the operational impacts of the SQuID protocol on EDLOS and ICU admission rates. The impact on EDLOS was evaluated by comparing EDLOS for the SQuID cohort compared to the insulin infusion cohort in the post-intervention period as well as two historical control periods – the pre-intervention period and the pre-COVID period. Impact on ICU admission rates was evaluated by comparing the proportion of patients with mild-to-moderate severity DKA admitted to the ICU in the post-intervention period compared to the historical control periods. Secondary outcomes included fidelity to and safety of the intervention. Fidelity to the protocol was evaluated by adherence to blood glucose checks at least every 2 hours. Safety of the intervention was evaluated by comparing the frequency of hypoglycemic events requiring rescue dextrose for the SQuID cohort compared to the traditional insulin infusion cohort. The SQuID protocol and study’s clinical pathway are provided below.
Results
During the 6-month study period from August 2021 to February 2022, 177 adult patients were enrolled (78 SQuID cohort, 99 traditional cohort) and retrospectively identified controls were identified in both the pre-intervention (163 patients) and pre-COVID (161 patients) periods. Of the 177 patients enrolled in the post-intervention period, 76 were admitted to an ICU whereas the remaining 101 patients were admitted to a medical floor. Of those admitted to a medical floor, 73 were managed on the SQuID protocol and 28 were managed on an insulin infusion.
Median EDLOS was significantly shorter for the SQuID cohort compared to control cohorts during the post-intervention period (-3.0 hrs, 95% CI -8.5 to -1.4), pre-intervention period (-1.4 hrs, 95% CI -3.1 to -0.1), and pre-COVID period (-3.6 hrs, 95% CI -7.5 to -1.8). ICU admission rates for mild-to-moderate severity DKA were lower in the post-intervention period (42.9%) compared to the pre-intervention (46.0%) and pre-COVID control periods (49.1%), though these reductions were not statistically significant. Fidelity to the SQuID protocol was high and there were no statistically significant differences in safety as measured by need for rescue dextrose (2.7% for the SQuID protocol vs. 3.6% for the traditional pathway).
Takeaway
This study demonstrated that a protocol using subcutaneous fast-acting insulin for the treatment of mild-to-moderate DKA was effective, safe, and reduced EDLOS by approximately 3 hours (median EDLOS 8.9 hrs in the SQuID cohort vs. 11.9 hrs in the traditional cohort). Given record levels of ED overcrowding and boarding with limited ICU availability across the US, alternative treatment pathways such as this to improve ED throughput are essential. Limitations of this study include differences between cohorts in the SQuID protocol and traditional pathway as well as this being a single-center study limiting its generalizability, especially with respect to variable operational design from hospital to hospital. Patients in the SQuID cohort were younger and more likely to have a lower acuity, but this was intentional in that the protocol is designed for healthier patients. Further, the authors speculate that this protocol may be appropriate for sicker patient populations though further research is needed.
At our institution, essentially all patients with DKA are treated with intravenous insulin infusions and are admitted to a stepdown level of care. Given hospital crowding and limited availability of stepdown beds, almost every DKA patient is treated in the ED as a boarder until they can be downgraded to floor status once they are taken off an insulin infusion. Our institution did recently develop an ED observation protocol using subcutaneous insulin for patients with mild DKA (pH >7.24 and HCO3 >14 mmol/L) though front-end protocols for channeling patients into this pathway are lacking. This study suggests that we can significantly increase ED throughput for patients with mild-to-moderate DKA if we were to adopt a similar SQuID protocol at our institution.
Authorship
Written by: Charlie Brower, MD, PGY-3, University of Cincinnati Department of Emergency Medicine
Editing and Posting: Jeffery Hill, MD MEd, Associate Professor, University of Cincinnati Department of Emergency Medicine
Cite As: SQuID Protocol for DKA: Impact on ED Length-of-Stay. TamingtheSRU. www.tamingthsru.com/blog/journal-club/squid. 3/12/2024
