(P-190) Characterizing Real-world Treatment Patterns Among Patients Hospitalized Due to sars-cov-2

Location: Poster Gallery-Informatics

Presenting Author(s)

RV

Rohit Vashsiht

Clinical Data Scientist
1. Bakar Computational Health Sciences Institute, University of California, San Francisco, California, USA, California, United States

Co-Author(s)

AP

Ayan Patel

Lead Data Scientist
1. Bakar Computational Health Sciences Institute, University of California, San Francisco, California, USA, California, United States
LD

Lisa Dahm, n/a

DIRECTOR, HEALTH DATA AND ANALYTICS
2. Center for Data-driven Insights and Innovation, University of California Health, Oakland, California, USA, California, United States
CW

Charles Wilson

LEAD DATA ARCHITECT
2. Center for Data-driven Insights and Innovation, University of California Health, Oakland, California, USA, United States
AB

Atul J. Butte, n/a

Priscilla Chan and Mark Zuckerberg Distinguished Prof, Director (BCHI), Chief Data Scientist (UC)
1. Bakar Computational Health Sciences Institute, University of California, San Francisco, California, USA, California, United States

Background: Over 90 drugs among 15 drug classes are under evaluation for treatment of COVID-19 per NIH guidelines. Understanding the real-world use of the drugs administered to COVID-19 patients hospitalized with or without ICU will be critical for the development of meaningful comparative effectiveness studies.

Objectives: To characterize drug treatments initiated among patients who were directly admitted to ICU compared to those who did not require ICU at the time of hospital admission due to SARS-CoV-2.

Methods: Multicenter cohort study using EHRs of 426,987 patients receiving COVID-19 related care across the University of California Health system between March 01, 2020, to January 08, 2021. COVID-19+ patients were identified based on laboratory-confirmed RT-PCR test. In-patient visits within +/- 14 days of COVID-19+ diagnosis was defined as hospitalization. The age and sex adjusted odds ratio of initiating a particular drug treatment at the time of hospitalization with or without requiring ICU was calculated using logistic regression.

Results: Across UC Health, 31,828 of the 426,987 patients were tested positive for SARS-CoV-2. Among the 31,828 patients, 4,277 were hospitalized with 685 requiring ICU, and 666 requiring ICU at a later stage of hospitalization. Patients requiring ICU at a later stage were not considered for this analysis thus resulting into 3,611 total in-patient admissions (2926 without ICU and 685 requiring ICU). Among 3,611 patients, 3,278 (2,618 without ICU and 660 with ICU) had drug information available for 14 of the 15 drug classes considered in this analysis. Almost half of the 3,278 patients (45.45%) started on a combination of two or more drug classes reflecting a considerable amount of heterogeneity in treatment initiation. Low molecular weight heparins (LMWH) were administered to 36.45% of the patients, followed by corticosteroids (10.89%) and ACE inhibitors (1.9%) at the time of admission. Initiating drug treatment with LMWH (OR 0.85; 95%CI [0.71-1.02]) or ACE inhibitors (OR 0.47; 95%CI [0.21-1.04]) did not vary between patients hospitalized with or without requiring ICUs. Initiating drug treatment with corticosteroids was less likely among patients not requiring ICUs (OR 0.62; 95%CI [0.45-0.85]), whereas the combination of corticosteroids and LMWH was very likely to be used among patients requiring ICU (OR 2.27; 95%CI [1.81-2.84]) at the time of admission.

Conclusions: Corticosteroids or their combination with LMWH is more likely to be used among patients requiring ICU compared to those not requiring ICU at the time of admission. Accounting for drug combinations used as starting treatment among hospitalized patients will be imperative to conduct meaningful real-world comparative effectiveness studies of COVID-19.