TA Evidence strategy lead, asthma BioPharmaceuticals Medical, AstraZeneca, Gaithersburg, US, United States
Background: Description of clinical outcomes of viral lung infection (VLI) could be challenging in real-world data (RWD) studies due to inconsistent use of diagnostic and procedure codes.
Objectives: We aimed to characterize cohorts of adults hospitalized with VLI and to describe the clinical burden of VLI using RWD.
Methods: Optum’s de-identified Clinformatics® Data Mart (CDM) and TriNetX were used to identify US study cohorts between June 2015 to December 2019 (pre COVID-19), and January 2020 to December 2021 (during COVID-19). Patients were included if any ICD-10 codes for the following were present in diagnosis positions one to five during in-patient encounters: enterovirus, adenovirus, coronavirus, metapneumovirus, parainfluenza, RSV, influenza, viral pneumonia, and unspecified acute lower respiratory infection. Records of ICU admission and invasive mechanical ventilation (IMV) or extracorporeal membrane oxygenation (ECMO) usage were reported. Code lists were developed and reviewed by clinical scientists. Outcomes were described within 30 and 60 days of hospitalization and were compared against data in literature.
Results: We identified 49,317 (CDM) and 67,780 (TriNetX) pre COVID-19 and 236,661 (CDM) and 225,350 (TriNetX) during COVID-19 adults hospitalized VLI patients. 39% and 46% were admitted to ICU and around 6% and 12% IMV/ECMO usage were recorded pre and during COVID-19 in CDM data. However, in TriNetX, recorded ICU admissions were half, around 19% pre and during COVID-19. IMV/ECMO usage in TriNetX were comparable to CDM with around 8% and 10% pre and during COVID-19 respectively. Both ICU admissions and IMV/ECMO usage rates were comparable within 30 and 60 days of hospitalization during each timeframe and were comparable with the literature.
Conclusions: Epidemiology data on VLI are heterogenous and identification of its outcome using RWD is challenging. However, comparability of IMV/ECMO usage rate across these two datasets in our study and literature data might indicate the relevance of the procedure and diagnostic code list used for these outcomes.
