We conducted a retrospective cohort study using the OHDSI OMOP-CDM at Korea University Medical Center. The data comprised three tertiary referral centers in the Seoul-Gyeonggi metropolitan area: Korea University Anam Hospital, Korea University Guro Hospital, and Korea University Ansan Hospital. These hospitals collectively provide 2949 beds and serve approximately 830,000 hospitalized patients annually.

The Korea University Anam Hospital IRB Committee (IRB No. 2022AN0400) approved the study protocol and waived the requirements for informed consent.

The OMOP CDM is a structured data model using a standardized vocabulary. All medical and cost data were extracted, transformed, and loaded into OMOP CDM version 5.3 following OHDSI protocols. This process included diagnoses, drug exposures, and comprehensive medical practice records. Stringent data verification procedures were implemented using the Automated Characterization of Health Information at Large-Scale Longitudinal Evidence Systems, complemented by thorough checks performed by both data analysts and clinicians.

The EMR data from the three tertiary centers were transformed into the OMOP CDM using a standardized process. This transformation enabled us to capture a wide range of clinical variables, including patient demographics, gestational age, birth weight, palivizumab prophylaxis, and comorbid conditions. We also conducted a detailed assessment of data completeness and addressed missing values—for example, missing gestational age records were noted in a small subset of patients and appropriately handled during the analysis

We searched the CDM database for cases of primary RSV infections. Subsequently, using concept IDs for RSV infections, we identified cases of health outcomes recorded in the CDM database. The primary outcome under consideration was admission to the ICU during hospitalization.

We conducted an analysis using the OMOP-CDM database, which comprised data from 33,674 children admitted to the Korea University Medical Center between August 2008 and December 2022.

The eligibility criteria for RSV testing in hospitalized children were defined based on the following parameters: (1) age between 0 and 9 years; (2) hospitalization for a minimum of 1 day; and (3) undergoing RSV testing through multiplex polymerase chain reaction (PCR). Children were eligible for inclusion in the study only if they underwent RSV testing by multiplex PCR. Only those cases with PCR-confirmed RSV infection were included in the analysis, ensuring high diagnostic accuracy. Antigen testing was not performed and therefore not considered in our inclusion criteria, minimizing variability in the diagnostic approach and reinforcing the reliability of our cohort selection. Subsequently, the RSV-positive group was categorized as follows: (1) children admitted to the ICU (either neonatal ICU or pediatric ICU) during hospitalization, referred to as the ICU group; and (2) children who were not admitted to the ICU throughout hospitalization, referred to as the non-ICU group.

Following a comprehensive literature review to identify potential risk factors for RSV infection leading to ICU admission, our analysis incorporated specific clinical data extracted from the CDM database. This data included demographic characteristics (age and sex), pre-existing medical conditions, gestational age, birth weight, and palivizumab prophylaxis, all collected during the RSV infection period. However, certain variables were excluded due to their absence in the electronic medical record (EMR) or inability to be translated into the CDM format. These excluded variables were as follows: length of observation, socioeconomic factors, presence of siblings, and maternal age.

Statistical analyses were performed using SAS software (version 9.4; SAS Institute, Cary, NC, USA). Descriptive statistics are presented as mean ± standard deviation (SD), median, counts, and percentages, as appropriate for the data type. Categorical variables were assessed using the chi-squared test, with RSV infection or ICU admission as the dependent variable and other variables as independent factors. Continuous variables were analyzed using a t test.

To identify risk factors associated with RSV infection and ICU admission, we employed a multivariate logistic regression model to compute odds ratios (ORs) and corresponding 95% confidence intervals (CIs). Statistical significance was set at p < 0.05. In the case of multicollinearity between gestational age and birth weight, two distinct regression analyses were conducted, with each analysis incorporating either gestational age or birth weight as an independent variable.