Abstract
Objective: Discharge is a pivotal transition of care moment; however, discharge antibiotic stewardship efforts have historically been limited. Our institution has a mandatory indication field for inpatient antibiotic orders. Using the last-ordered inpatient antibiotic indication (“Last Inpatient Indication”), we assessed the utility of this field compared to the ICD-10 in inferring clinician intent for discharge antibiotics, to forgo the need for manual chart review to assess antibiotic appropriateness. Methods: We extracted electronic medical record data on adult inpatient encounters in 2023 with ≤2 discharge antibiotics. We reviewed a random subset of 300 encounters to determine if the ICD-10 or Last Inpatient Indication had higher agreement with clinician intent for discharge antibiotics, as determined by chart notes. To facilitate comparison, we created a dictionary classifying ICD-10 indications. Results: We included 3,414 encounters. The most common discharge antibiotics were amoxicillin/clavulanate (24%) and ciprofloxacin (15%). The most common ICD-10s were non-infectious (48%) and sepsis (18%). In the subset of chart-reviewed encounters, the Last Inpatient Indication agreed with the documented clinician intent for discharge antibiotic more often than the ICD-10 (84% vs 28%). Applying institutional guidelines, we were able to use the Last Inpatient Indication to assess appropriate duration and choice of discharge antibiotics for select infections. Conclusions: The Last Inpatient Indication outperformed ICD-10s in inferring clinician intent for discharge antibiotics and can be used to efficiently assess antibiotic appropriateness without need for manual chart review. Usefulness of ICD-10s was limited by high percentage of non-infectious and sepsis codes.

This publication has 14 references indexed in Scilit: