SummaryBackground Serotype-specific estimates of pneumococcal invasiveness used in pneumococcal carriage transmission models to predict changes in disease incidence post-vaccination are largely derived from high-income settings. We conducted a systematic review of carriage prevalence and invasive pneumococcal disease (IPD) incidence to calculate case–carrier ratios (CCRs) in different income settings. Methods We conducted a systematic search of MEDLINE, Embase, and Global Health databases on March 14, 2022, to identify publications on pneumococcal carriage prevalence or IPD incidence; we requested individual-level data from authors of relevant texts. Serotype-specific CCRs, calculated as IPD incidence divided by carriage prevalence, were pooled across settings using random effects meta-analyses, stratified by before versus after pneumococcal conjugate vaccine (PCV) introduction, country income group, age group, sex, and HIV status. Findings We identified 80 publications from 18 countries (13 upper-middle-income countries UMICS or high-income countries HICs, five low-income countries LICs or lower-middle-income countries LMICs) reporting carriage prevalence or IPD incidence in overlapping geographical areas, time periods, and age groups. We calculated CCRs for more than 70 serotypes, stratified by age group, income setting, and pre-PCV versus post-PCV introduction. In children younger than 5 years, pre-PCV CCRs for non-vaccine serotypes not included in the 13-valent PCV were higher in LICs and LMICs than in UMICs and HICs (177 95% CI 124–251 vs 103 60–176, respectively). Post-PCV CCRs for non-PCV13 serotypes dropped in UMICs and HICs (26 22–30) but not in LICs and LMICs (173 139–216). Pre-PCV versus post-PCV changes varied by serotype and age group. CCRs were lowest in 5–14-year-olds and were higher in HIV-positive than HIV-negative individuals. There were no differences in CCRs by sex. Interpretation Pneumococcal invasiveness varies by serotype, age group, country income group, HIV status, and over time; however, substantial variation remains unexplained. Our CCRs represent the most representative estimates of invasiveness currently available for use in statistical or mathematical prediction models of disease incidence, where only carriage prevalence data are available. Funding Wellcome Trust.
Gallagher et al. (Sun,) studied this question.