Antibody response to primary and booster immunization in infants born to mothers immunized with pertussis-containing vaccines during pregnancy versus unimmunized women: a systematic review and meta-analysis

In order to assess primary outcome, a mixed effects model will be used. This is a powerful tool that enables combing data from different studies. Using this model, we will adjust for co-variates that were shown to affect immune response to vaccination according to previous systematic literature review (Zimmermann et al. Clinical Microbiology Reviews 2019). The output of this model will be the geometric mean ratio (GMR) of antibody levels prior and after immunization. The GMR is the ratio of Abs in infants born to women immunized in pregnancy vs. unimmunized women (a GMR of ≥ 1 indicates no interference, a GMR of <1 interference). In addition, each individual dataset from each individual study included will be coded and given a specific code, to maintain the integrity of the original data by trial level. This variable will also serve as a grouping factor in the model. To address secondary outcomes 1 and 2, a chi-square test will be used to determine whether the percentage of protected infants (or pregnant women) is different in the 2 groups (those born to pertussis-immunized vs pertussis-unimmunized women). In this analysis, we will be using well-established sero-protective cut offs for specific vaccine-induced antibodies (e.g. for streptococcus pneumonia). This analysis is important as it will determine whether immunization against pertussis in pregnancy results in lower percentage of protected infants (e.g. 50% vs 70% in infants born to vaccinated mothers compared to unvaccinated mothers, p-value xx, chi-square test). To address secondary outcome 3, we will calculate the fold change in antibody levels after vs before immunization in infancy. This will be performed in the 2 groups (infants born to vaccinated vs unvaccinated mothers). The fold change values will be assessed for normality (preliminary results based on the already included studies showed normal distribution). If normal distribution is still maintained (as more datasets are added to the analysis) then t-test will be used to determine whether the fold change of antibody levels (after vs before immunization) is significant between infants born to pertussis- immunized vs -unimmunzed pregnant women (e.g. 5 vs 10 fold change in infants born to vaccinated vs unvaccinated women, p=yy, t-test). To address secondary outcome 4, a mixed-effect model will be used (based on Voysey et al, JAMA Pediatrics 2017; PMID: 28505244), in which co-varites that could have affected immune response to vaccination in infants will be entered. The aim of this analysis is to determine factors that can mitigate the effect of blunting. The output of this model is GMR which indicates the relative increase (fold rise) in antibody levels in response to vaccination associated with 1 unit change in a co-variate (examples of co-variates will be timing of initiation of primary vaccination series in infants, antibody levels at time of primary vaccination). Example of outcome of the model will be: the relative increase in antibody response is 10% for a 1-week older infant.