Evaluating Trends in Antibiotic Non-Susceptible Invasive Pneumococcal Disease Using Multiple Imputation by Chained Equations and Random Forest Models to Impute Missing Data, United States 2018–2023
- Presentation Day/Time: Thursday, April 23, Noon
- Presenter: Stephen Mugel, PhD, MS, National Center for Immunization and Respiratory Diseases, Division of Bacterial Diseases
The Issue
- Invasive pneumococcal disease (IPD) causes significant morbidity, and antibiotic non-susceptibility can complicate treatment. Pneumococcal conjugate vaccines (PCVs) have reduced IPD and non-susceptible IPD (NS-IPD) incidence attributed to vaccine-covered serotypes. However, missing data present challenges for estimating their impact.
What We Did
- We identified U.S. IPD cases during 2018–2023 and determined serotype and non-susceptibility by whole genome sequencing. We used machine learning to fill in missing data gaps on non-susceptibility and estimated annual NS-IPD (IPD with resistance to one or more antibiotics) incidence per 100,000 population.
What We Found
- Overall NS-IPD and vaccine serotype-NS-IPD rates declined among older adults and remained low among young children.
What This Means
- New imputation models addressed missing serotype and non-susceptibility data, enabling incidence estimates crucial for monitoring NS-IPD after new PCV introduction.
