WIP 89 - Variations in Physician Prescribing Practices for Low-Value Care in Infants Hospitalized with Bronchiolitis: A Retrospective Cohort Study from a Single Canadian Center

Publication Number: WIP 89.7491

Zachary Dionisopoulos, University of Toronto Temerty Faculty of Medicine, Montreal, PQ, Canada; Francesca del Giorgio, McGill University Faculty of Medicine and Health Sciences, Montreal, PQ, Canada; Layla Dehbidi Assadzadeh, CHU Sainte-Justine, Montreal, PQ, Canada; Nick Barrowman, University of Ottawa Faculty of Medicine, Ottawa, ON, Canada; Mei Han, Children's Hospital of Eastern Ontario Research Institute, Ottawa, ON, Canada; Lamia Hayawi, Children's Hospital, Ottawa, ON, Canada; Louis Philip Benoit Bouvrette, CHU Sainte-Justine, Montreal, PQ, Canada; Tamara Perez, Universite de Montreal Faculty of Medicine, Montreal, PQ, Canada; Olivier Drouin, CHU Sainte-Justine, Montreal, PQ, Canada

Background: Low-value care (LVC) is the use of health services for which harms and costs outweigh benefits. Bronchiolitis is among the most common and costly causes of hospitalization in children and LVC remains common in bronchiolitis care despite high-quality evidence discouraging its use. Canadian data are scarce examining inter-physician LVC prescription practice variation for infants with bronchiolitis.

Objective: To estimate inter-physician variation in LVC for infants hospitalized with bronchiolitis at a large tertiary Canadian pediatric hospital.

Design/Methods: This is a five-year retrospective cohort study (2017 – 2022) of infants (1 – 12 months) hospitalized with bronchiolitis identified through ward discharge summary diagnoses (n = 2533 patients). Individual physician data was collected on prescribing practices for 5 low-value health services: 1) antibiotics, 2) chest x-ray, 3) narrow viral testing (Influenza, SARS-CoV-2, Respiratory Syncytial Virus), 4) short-acting beta agonists, 5) systemic corticosteroids (last two available only for wards physicians). Physicians who treated ≥20 patients with bronchiolitis were included (n = 126 physicians). The primary outcome is the proportion of patients prescribed each LVC service. Raw and adjusted proportions will be calculated for individual physician LVC use. The adjusted model will use a mixed-effects model including physician-specific random effects and adjustments for age, sex, comorbidity status and triage severity scores. The within-physician distribution of proportions of LVC prescriptions will be compared between higher-acuity (any intensive care unit [ICU] stay) and low-acuity (no ICU stay) infant cohorts. Physician characteristics influencing LVC prescription practices, such as specialty and years of practice, will also be examined. Data analysis will be completed by December 2024, and a manuscript will be finalized for publication in January 2025. Internal review board approval was obtained through the study site.