This was a retrospective, longitudinal, observational study utilizing Optum® electronic health records (EHR) data from October 1, 2015–June 30, 2023. The Optum® EHR database covers over 100 million unique individuals from all US Census regions and includes linked records from a broad range of healthcare settings and clinical fields. The Optum® sample is largely representative of healthcare utilizers and the general US population, across attributes including sex, age group, US region, race and ethnicity, and insurance coverage [18].

Empirical algorithms were used to define ARI episodes (the unit of analysis for this study) and identify ARI episodes with a diagnosis code. Specifically, a medically-attended ARI episode was defined as a period with one or a cluster of inpatient, emergency department (ED), or outpatient encounters with ARI diagnoses identified using International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes (Appendix A1 in Supplementary Material). The start date (i.e., index date) for an ARI episode was identified as the first inpatient admission or non-inpatient encounter with an ARI diagnosis, after (1) the first activity in the EHR, or (2) the end date of the previous ARI episode, whichever was later. The end date was the later date of the last non-inpatient ARI-related encounter or the date of discharge from the last ARI-related hospitalization. The 12-month period before the start date of an eligible ARI episode was defined as the baseline period (Appendix A1 in Supplementary Material). The window for whether an ARI episode was tested for RSV or other respiratory viruses was defined as a ± 7-day period around the ARI episode (Appendix A2 in Supplementary Material).

Adult patients were selected from the Optum® EHR database if they had ≥ 1 ARI diagnosis at age ≥ 50 years. For a given patient, an ARI episode was eligible for inclusion if the patient had ≥ 1 activity in the EHR database that was ≥ 12 months prior to the index date of the episode (as a proxy for continuous data availability during the 12-month baseline period), and the index date occurred between October 1, 2016 and May 26, 2023. The May 26, 2023 cutoff was used to account for the 28-day window following the start of an ARI episode to define an ARI episode, and the 7-day testing window prior to the end of data availability (June 30, 2023). Patients could contribute one or more eligible ARI episodes.

The primary outcome of this study was the percentage of all eligible ARI episodes that were tested for RSV. To contextualize these findings, the percentage of all eligible ARI episodes with at least one test for any of the following commonly tested respiratory viruses was evaluated: RSV, influenza, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), other coronaviruses (e.g., 229E, HKU1, NL63, OC43), human metapneumovirus, human rhinovirus/enterovirus, parainfluenza, adenovirus, and bocavirus. The percentage of eligible ARI episodes tested specifically for RSV, influenza, SARS-CoV-2, or using an RSV rapid antigen test was also assessed.

Tests of interest documented within the ± 7-day window around the ARI episode were identified (Appendix A2 in Supplementary Material). The percentage of ARI episodes tested each week were estimated as follows: for a given week (Sunday–Saturday), if the start date of the ARI episode was within that week, it contributed to the denominator of the weekly percentage estimation. An ARI episode eligible for the denominator estimation of a given week contributed to the numerator estimation of the week if a test of interest was undertaken within the testing window defined for that ARI episode.

In addition, for each epidemiological year (July 1–June 30 of the following year) and each RSV season (as defined by the CDC [14]), the following summary statistics were reported: (1) the number of ARI episodes; (2) the number and percentage of ARI episodes tested for RSV, influenza, SARS-CoV-2, and commonly tested respiratory viruses overall; and (3) the number and percentages of ARI episodes tested using specific testing approaches for RSV.

The analyses described above were further performed among subgroups based on most intensive care setting during the ARI episode (with the inpatient setting as the most intensive care setting, followed by the ED setting, then the outpatient setting).

RSV testing status and methodology were ascertained using RSV test procedure codes, which were based on Current Procedural Terminology and Healthcare Common Procedure Coding System codes, as well as laboratory tests used to identify PCR and rapid antigen testing methodologies (Table S1 in Supplementary Material).

Additional analyses were conducted to evaluate patient, HCP, and ARI characteristics associated with the likelihood of an ARI episode being tested for RSV.

A logistic regression model was fitted to identify patient characteristics (e.g., demographics, comorbidities), HCP characteristics (e.g., HCP specialty), and ARI characteristics (e.g., timing of ARI) associated with the likelihood of being tested for RSV. Covariates for healthcare organizations, defined as health systems or physician practice groups and composed of dummy variables, were included to control for potential differences in testing practices across healthcare organization networks. Generalized estimating equations (GEE) modeling was utilized to account for within-patient correlation between ARI episodes due to the inclusion of multiple ARI episodes from the same patient and from the same healthcare organization.

Regression analyses were performed overall and among subgroups based on most intensive care setting during the ARI episode (inpatient, ED, outpatient). Associations were reported using odds ratios (ORs), associated 95% confidence intervals (CIs), and p values for each covariate.

This study complied with all applicable laws regarding subject privacy. No direct subject contact or primary collection of individual human subject data occurred. The data are certified as de-identified by an independent statistical expert following Health Insurance Portability and Accountability Act (HIPAA) and California Consumer Privacy Act (CCPA; AB-375) statistical de-identification rules and managed according to Optum® customer data use agreements. Access to the database used in this study was granted through agreements established with Optum®. Therefore, informed consent, ethics committee, and institutional review board approval were not required.