OBJECTIVES : After the introduction of the seven-valent pneumococcal conjugate vaccine (PCV7) in the year 2000, the incidence of invasive pneumococcal disease (IPD) in the US declined by 80 percent in vaccinated children and 30 percent in unvaccinated adults. Subsequently, US vaccine policy switched to the 13-valent PCV and added adult PCV13 vaccination. This work explores the accuracy of a transmission dynamic model developed in 2009 to predict long-term IPD due to changes in US immunization practices.

METHODS : The differential equation model simulates the transfer and acquisition of asymptomatic carriage of pneumococci and the development of IPD among individuals aged <2, 2–4, 5–17, 18-49, 50-64, and ≥65 years. Pneumococcal serotypes were stratified into 3 categories: PCV7-type (4, 6B, 9V, 14, 18C, 19F, and 23F), PCV6-type (1, 3, 5, 6A, 7F, and 19A), and non-PCV-type (all others). Model parameters were calibrated using US IPD surveillance data from 1998-2006. Model results were compared to observed epidemiology to determine its long-term accuracy to predict the incidence of IMD.

RESULTS : After adjusting model parameters for PCV13 efficacy and adult vaccine coverage, modeled IPD closely replicated observed IPD from 2000 through 2016. Observed baseline pre-vaccine incidence for children <2 years of age was 192 cases/100,000 and 13.5 cases in 2016, versus 18.5 cases estimated by the model. Similarly, observed versus modeled cases in the ≥65-year-old age group were 24 and 23.6 cases.

CONCLUSIONS : This epidemiologic model accurately estimates the observed US IPD surveillance data 17 years after initial introduction of PCV, accounting for direct and indirect benefits of vaccination. The work demonstrates that well-constructed mathematical models can accurately replicate real-world scenarios where key input parameters can be varied to predict the impact of alternate scenarios, providing insights to inform public health policymaking.