The Serotype Distribution among Healthy Carriers before Vaccination Is Essential for Predicting the Impact of Pneumococcal Conjugate Vaccine on Invasive Disease

Pneumococcal conjugate vaccines (PCVs) have substantially reduced morbidity and mortality of pneumococcal disease. The impact of the 7-valent PCV on all-serotype invasive pneumococcal disease (IPD) among children was reported to vary between high-income countries. We investigate the ability to predict this heterogeneity from pre-vaccination data. We propose a parsimonious model that predicts the impact of PCVs from the odds of vaccine serotype (VT) among carriers and IPD cases in the pre-PCV period, assuming that VT are eliminated in a mature PCV programme, that full serotype replacement occurs in carriage and that invasiveness of the NVT group is unchanged. We test model performance against the reported impact of PCV7 on childhood IPD in high-income countries from a recent meta-analysis. The odds of pre-PCV7 VT IPD, PCV schedule, PCV coverage and whether a catch up campaign was used for introduction was gathered from the same analysis. We conducted a literature review and meta-analysis to obtain the odds of pre-PCV7 VT carriage in the respective settings. The model predicted the reported impact on childhood IPD of mature PCV programmes; the ratio of predicted and observed incidence risk ratios was close to 1 in all settings. In the high income settings studied differences in schedule, coverage, and catch up campaigns were not associated with the observed heterogeneity in impact of PCV7 on childhood all-serotype IPD. The pre-PCV7 proportion of VT IPD alone also had limited predictive value. The pre-PCV7 proportion of VT carriage and IPD are the main determinants for the impact of PCV7 on childhood IPD and can be combined in a simple model to provide predictions of the vaccine preventable burden of IPD.Author Summary: Pneumococcal vaccines (PCVs) that protect children against 7, 10 and 13 of the most pathogenic pneumococcal serotypes have substantially reduced childhood morbidity and mortality. A recent analysis that evaluated the impact of the 7 valent PCV in multiple high income settings in North America, Europe and Oceania found that the magnitude of all-serotype invasive pneumococcal disease reduction varied greatly between settings (24%-83%). We explored potential sources for that variation, including differences in disease epidemiology before vaccination, vaccine coverage, vaccine schedules and the use of catch-up campaigns for introduction. We find that differences in reported disease impact among mature PCV programmes are likely to be unrelated to the differences in the vaccine programme but can be predicted from a simple model based on pre-vaccination epidemiology, in particular the proportion of vaccine serotypes detected among patients with invasive pneumococcal disease and the proportion of vaccine serotypes that are found in the nasopharynx of healthy individuals. This model presents a useful tool to estimate the potential impact of PCVs (as a relative rate reduction), highlights the essential role of pre-vaccination carriage in healthy individuals for disease impact of PCVs and can estimate the prevented burden of disease where disease surveillance is unavailable.