From the 2018 HVPAA National Conference
Emily Signor (University of Utah School of Medicine), Karli Edholm (University of Utah School of Medicine), katie lappe (University of Utah School of Medicine), Christy Hopkins (University of Utah), Nathan Hatton (University of Utah School of Medicine), Benjamin Gebhart (University of Utah School of Medicine), Heather Nyman (University of Utah School of Medicine), Stacy Johnson (University of Utah School of Medicine)
Background
Previously, our institution did not have a standardized protocol for diabetic ketoacidosis (DKA), and most patients were admitted to an intensive care unit (ICU).
Objectives
Our aim was to decrease the proportion of patients with non-severe DKA admitted to an ICU, to implement a DKA pathway without increase in adverse clinical or safety outcomes, and to decrease total direct hospital cost for DKA care.
Methods
We developed an EHR-driven care pathway for DKA at a 537-bed academic medical center using a best practice alert (BPA) decision support tool. Any patient ≥16 years presenting to the Emergency Department with lab criteria for DKA prompted a BPA linked to an order set including additional diagnostic studies, triage to admission location, and treatment initiation. Patients classified as severe DKA were started on an insulin infusion and admitted to ICU, while patients with mild-moderate DKA were given subcutaneous Lispro and admitted to the floor. Additional BPAs fired when hyperglycemia and anion gap resolved, and were also linked to order sets guiding subsequent therapy. Using an electronic database query, we identified patients with DKA and compared clinical, safety, and cost outcomes pre- and post-implementation of DKA pathway. Clinical outcomes included proportion treated with insulin drip, time to first insulin dose, time to hyperglycemia resolution, time anion gap closure, time to basal insulin initiation, ED length of stay (LOS), ICU LOS, hospital LOS, and proportion of patients seen by a diabetic educator prior to discharge. Safety outcomes included rates of recurrent DKA, treatment-induced hypokalemia, hypoglycemia, in-hospital mortality, 30-day ED return visit, and 30-day hospital readmission. Patient-specific costs were derived from a proprietary institutional costing tool.
Results
We included 107 patients in the 1-year pre-implementation period and 108 patients in the post-implementation period. ICU admission for DKA decreased from 67.3% to 41.7% (p<0.001) and diabetic educator consults increased from 44.9% to 63.9% (p=0.05). Hospital length of stay did not change (3.4 ± 3.0 days vs 3.4 ± 3.0 days, p=0.960). There was no difference in rate of hypoglycemia (31.8% to 24.1%, p=0.208). Return ED visits decreased significantly 12.2% to 2.8%, p=0.010). Total direct cost did not differ significantly between intervention groups. However, the small sample sizes may be underpowered to detect significant cost differences.
Conclusion
Our results demonstrate that non-severe DKA can be safely managed on a medicine floor with improved ICU utilization, increased diabetic education, and decreased ED return visits. Implementation of an EHR-driven diagnostic and treatment pathway does require education of the healthcare team, along with periodic adjustment of the trigger thresholds to improve specificity and ensure patient safety.
Implications for the Patient
By creating a DKA care pathway at our institution, we have been able to decrease ICU utilization for patients with non-severe DKA with similar clinical and safety outcomes