A GENERALIZED MODELING FRAMEWORK FOR THE INDIRECT ESTIMATION OF HOSPITALIZATIONS ATTRIBUTABLE TO SPECIFIC RESPIRATORY VIRUSES INCORPORATING MULTI-PATHOGEN DYNAMICS AND PANDEMIC-DISRUPTED STUDY PERIODS
Author(s)
Robin Bruyndonckx, PhD1, Elisabeth J. Mesfun-Kap, Dr. Sc. Hum1, Pimnara Peerawaranun, MsC1, Giovanni Della Cioppa, PhD, MD2, Juana Magdalena, PhD2, Scott McDonald, PhD1, Laura Sarabia, PhD1, Javier R. Guinazu, MD2, Aleksandra Polkowska-Kramek, PhD1, Mike Penlington, PhD2, Oliver James Martyn, BSc, MPH3.
1P95, Leuven, Belgium, 2Vicebio Ltd, Louvain-la-Neuve, Belgium, 3Sanofi Pasteur, Copenhagen, Denmark.
1P95, Leuven, Belgium, 2Vicebio Ltd, Louvain-la-Neuve, Belgium, 3Sanofi Pasteur, Copenhagen, Denmark.
OBJECTIVES: Lower respiratory tract disease (LRTD) is a common cause for hospitalization, especially among infants and older adults. Estimating the proportion of these hospitalizations that is attributable to a specific respiratory virus is challenging. The recent COVID-19 pandemic and associated changes in countries’ testing approaches and hospitalization policies have added to the challenges.
METHODS: We developed a generalized modelling framework, building on Bruyndonckx et al., to estimate the hospitalization burden associated with specific respiratory pathogens using existing databases. The proposed model links the temporal variability in pathogen circulation with the variability in an outcome potentially associated with the pathogen. Data on pathogen circulation are obtained from national surveillance databases. Data on hospitalizations are derived from e.g., claims databases. The framework includes a standardized approach to (i) handling temporary deviations in testing approaches and hospitalization policies, (ii) standardizing surveillance-based pathogen counts, (iii) accounting for seasonality, and (iv) model selection. The framework was applied to Scottish data ranging from 2014 to 2024 (Electronic Communication of Surveillance Scotland and the Scottish Morbidity Records databases) and the results were compared with published observational studies. We focused on estimating the proportion of hospitalizations with ICD-10 codes related to LRTD that are attributable to respiratory syncytial virus (RSV) while accounting for important co-circulating pathogens (i.e., human metapneumovirus, parainfluenza virus serotype 3, influenza virus and SARS-CoV-2).
RESULTS: In Scotland, 2.3% (95% CI: 1.2%-3.4%) of LRTD-related hospitalizations were attributable to RSV. This estimate was minimally affected by adding rhinovirus as an additional co-circulating pathogen (2.3%; 95% CI: 1.2%-3.3%) or standardizing the pathogen counts (2.2%; 95% CI: 1.3%-3.1%). The results match observational studies conducted in older adult inpatients.
CONCLUSIONS: The developed framework presents a generalizable approach to estimate the proportion of hospitalizations attributable to specific respiratory pathogens. It facilitates multi-country analyses in settings with limited access to individual patient data.
METHODS: We developed a generalized modelling framework, building on Bruyndonckx et al., to estimate the hospitalization burden associated with specific respiratory pathogens using existing databases. The proposed model links the temporal variability in pathogen circulation with the variability in an outcome potentially associated with the pathogen. Data on pathogen circulation are obtained from national surveillance databases. Data on hospitalizations are derived from e.g., claims databases. The framework includes a standardized approach to (i) handling temporary deviations in testing approaches and hospitalization policies, (ii) standardizing surveillance-based pathogen counts, (iii) accounting for seasonality, and (iv) model selection. The framework was applied to Scottish data ranging from 2014 to 2024 (Electronic Communication of Surveillance Scotland and the Scottish Morbidity Records databases) and the results were compared with published observational studies. We focused on estimating the proportion of hospitalizations with ICD-10 codes related to LRTD that are attributable to respiratory syncytial virus (RSV) while accounting for important co-circulating pathogens (i.e., human metapneumovirus, parainfluenza virus serotype 3, influenza virus and SARS-CoV-2).
RESULTS: In Scotland, 2.3% (95% CI: 1.2%-3.4%) of LRTD-related hospitalizations were attributable to RSV. This estimate was minimally affected by adding rhinovirus as an additional co-circulating pathogen (2.3%; 95% CI: 1.2%-3.3%) or standardizing the pathogen counts (2.2%; 95% CI: 1.3%-3.1%). The results match observational studies conducted in older adult inpatients.
CONCLUSIONS: The developed framework presents a generalizable approach to estimate the proportion of hospitalizations attributable to specific respiratory pathogens. It facilitates multi-country analyses in settings with limited access to individual patient data.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
SA100
Topic
Methodological & Statistical Research, Real World Data & Information Systems, Study Approaches
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Respiratory-Related Disorders (Allergy, Asthma, Smoking, Other Respiratory)