Modelling the effect of changes in vaccine effectiveness and transmission contact rates on pertussis epidemiology

• We studied pertussis transmission with an age structured model developed by us.
• We analyzed the effect of changes in vaccine effectiveness and contact rates.
• The model predicts different trends when these parameter changes are considered.
• Agreement between our results and reported data of different US states was found.
• This agreement suggests clues to explore possible causes of pertussis resurgence.

The incidence of the highly infectious respiratory disease named pertussis or whooping cough has been increasing for the past two decades in different countries, as in much of the highly vaccinated world. A decrease in vaccine effectiveness over time, especially when acellular vaccines were used for primary doses and boosters, and pathogen adaptation to the immunity conferred by vaccines have been proposed as possible causes of the resurgence. The contributions of these factors are not expected to be the same in different communities, and this could lead to different epidemiological trends. In fact, differences in the magnitude and dynamics of pertussis outbreaks as well as in the distribution of notified cases by age have been reported in various regions.Using an age-structured mathematical model designed by us, we evaluated how the changes in some of the parameters that could be related to the above proposed causes of disease resurgence – vaccine effectiveness and effective transmission rates – may impact on pertussis transmission.When a linear decrease in vaccine effectiveness (VE) was assayed, a sustained increase in pertussis incidence was detected mainly in infants and children. On the other hand, when changes in effective transmission rates (βij) were made, a dynamic effect evidenced by the presence of large peaks followed by deep valleys was detected. In this case, greater incidence in adolescents than in children was observed. These different trends in the disease dynamics due to modifications in VE or βij were verified in 18 possible scenarios that represent different epidemiological situations. Interestingly we found that both incidence trends produced by the model and their age distribution resemble the profiles obtained from data reported in several regions. The implications of these correlations are discussed.