We first conducted a search to identify currently available wP vaccines and their manufacturers, then conducted a systematic search for published articles that assessed the duration of immunity and effect of boosters of currently available vaccines. The study protocol was registered in PROSPERO (registration number CRD42018107309). We planned and developed the review following the Cochrane Handbook for Systematic Reviews of Interventions ¹² . We followed the 2020 Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) recommendations for reporting ^{13, 14} .

The review question, developed in accordance with the elements of the acronym PICO (Population, Intervention, Comparator and Outcomes), is presented in Box 1.

As there is no single structured source of information on wP vaccines currently in use, we first obtained the 2019 list of wP vaccines pre-qualified by the World Health Organization (WHO) ¹⁵ , which listed five wP vaccine manufacturers. Vaccines from these manufacturers are likely to be used by developing countries as they are commercially available through vaccine procurement funds such as United Nations Children's Fund (UNICEF) and the Pan American Health Organization (PAHO)’s Revolving Fund ⁴ . We then obtained the 2018 list of wP vaccine manufacturers from the “Developing Countries Vaccine Manufacturing Network” (DCVMN) from WHO ⁴ , which lists 13 manufacturers of wP vaccine, only three of which were also included in WHO’s pre-qualified list. The remaining 10 wP manufacturers likely produce wP vaccines for domestic use by National Immunization Programs, which are thus not commercialized or exported. Finally, we used a January 2018 Global Vaccine Market study commissioned by WHO ¹⁶ , which presents a map of global producers of vaccines that contain diphtheria and tetanus components. Producers that were not on the previous lists (WHO pre-qualified and DCVMN) and could be producers of wP were identified from the map and further verified through internet searches on the websites of each producer. Three more manufacturers were identified from this source.

We included all studies reporting primary data, in which the protection, immunity or duration of immunity conferred by a currently available wP vaccine was assessed. In keeping with the goals of the review, only studies in which immunological markers in a group of individuals were assessed over time after the primary series, or which compared three doses with four or five doses, were included.

Randomized controlled trials (RCTs) and observational studies were included. We categorized observational studies as case-control, cohort (including controlled and uncontrolled cohort studies), and cross-sectional study designs. To be eligible for review, studies had to target healthy children of either sex who had received wP vaccination by the age of six years, and also evaluate a primary schedule of three doses administered by six months of age and/or any additional booster doses. Eligible studies could have compared children vaccinated with wP to unvaccinated children, to children vaccinated with aP, to children vaccinated with different wP dosing schedules, or have no comparator.

The outcomes measured in the study could be clinical efficacy/effectiveness or immunogenicity levels, including seroconversion (% of individuals who seroconverted at a given threshold value), antibody levels (reported as described by the authors as geometric mean antibody concentrations/titers (GMC/GMT)), and/or serological measurements of antibody levels over time. Immunologic markers included anti-pertussis toxin (PT), anti-pertussis fimbriae (anti-PF), anti-pertussis agglutinating antibody (anti-Agg), and anti-filamentous hemagglutinin (anti-FHA).

Studies were excluded if they: assessed children with underlying disease; used fractionated vaccine doses; evaluated only acellular pertussis (aP) vaccine; evaluated only 1 or 2 doses; or had only one immunogenicity assessment after the 3rd primary dose, without any later assessment to evaluate the decay of antibody titers. We also excluded ecological and modelling studies, letters to the editor, recommendations, guidelines, and reviews.

Electronic searches were conducted in Medline (PubMed), Embase, Web of Science, Lilacs, and Central. A complementary search was conducted in the electronic library SciELO. Additionally, reference lists of selected articles and reviews were screened. No date, location, or language limits were placed on the searches which included publications between 30 April, 2019 and updated in September 21, 2021. Search strategies for each database are presented in Supplementary Table 1 ( Extended data ¹⁴ ).

The procedure for screening and selecting studies was carried out in two phases. In the first phase, three reviewers (AB, MQ and GP) independently screened all titles and abstracts for duplication and inclusion criteria. Then, screened articles were categorized as potentially eligible, unclear, or excluded. Citations on which the reviewers disagreed were discussed or assessed by a fourth reviewer (CT). In the second phase, two reviewers (AB and MQ) independently screened studies obtained for full reading to confirm that they met the eligibility criteria.

Data extraction, done by two independent reviewers, using abstraction forms developed for this review with Microsoft Excel (Microsoft Inc.), included country, funding source, study design, intervention details (vaccine and manufacturer, vaccination schedule, age of administration of each dose), study setting and period, measure of outcome including case definition and diagnostic criteria, baseline information on study population, methods for data analysis, and main results. Results extracted included descriptive study characterization, clinical results (absolute numbers or rates of pertussis cases), and serological results considering any anti-pertussis antibody type, including percentage of individuals who seroconverted and/or antibody GMT/GMC. Finally, information on pertussis vaccines and schedules in use in the country where each study was conducted was also obtained in WHO-reported immunization schedules by vaccine, by country ¹⁷ .

The methodological quality of studies was assessed using: a) the revised Cochrane tool for assessing risk of bias in randomized trials (Cochrane Risk of Bias tool) ¹⁸ , b) the Quality Assessment Tool for cohort or case-control, non-randomized trials studies with: The Risk Of Bias In Non-randomized Studies – of Interventions (ROBINS-I) ¹⁹ , and c) the Critical Appraisal tools for use in Systematic Reviews Checklist for Analytical Cross Sectional Studies from Joanna Briggs Institute ²⁰ . For the latter, considering the eight items in the checklist, we categorized the studies as presenting low risk (when scoring 7 or 8), moderate risk (scoring 6 or 5), or high risk (scoring 4 or lower).

Two authors performed the methodological assessments independently, and disagreements between reviewers were assessed and sources of divergence discussed until agreement was reached. When disagreement was not resolved, a third reviewer was used as an arbitrator.

We conducted descriptive analyses of manufacturers and vaccines identified and of published study characteristics including design, country, study period, population, vaccine and schedule used, endpoints considered, number of children enrolled and assessed, and study follow-up period.

For studies reporting immunological outcomes, we originally planned to conduct a meta-analysis to pool results from similar studies presenting antibody titers or seroconversion rates at similar points in time in relation to primary series and each additional booster dose. The meta-analysis was precluded; however, as the included studies were very heterogeneous in terms of the immunological markers and indicators considered, points in time in which they were assessed, and study designs. Thus, we have restricted out findings to a qualitative descriptive synthesis of the results reported by each study.

Results are presented by type of outcome. For efficacy or effectiveness, for studies with clinical endpoints, the main measure of interest is the vaccine’s effect in reducing pertussis, reported as number of events in each study group and absolute risk between vaccinated and unvaccinated as calculated by the review authors. For immunogenicity studies, results are presented either as GMT or GMC of anti-PT, anti-FHA, and/or as percent of individuals who seroconverted given a threshold as reported in the original study. All results are presented by dose of vaccine and timing of measurement of immune response, considering point in time after the primary vaccine series.

In total, we identified 18 manufacturers producing the currently available wP vaccines ( Table 1). Seven are located in India, and no other country reports more than one manufacturer; with South Korea being the only high-income country represented. Five of these manufacturers are in WHO’s pre-qualification list – four in India and one in South Korea. Although the WHO list included wP vaccine Quinvaxem, produced by Janssen Vaccines Corp. since 2011, the manufacturer discontinued its production in 2017 and the latest expiry date of batches supplied was December 2019. As this vaccine is currently no longer in use, we did not include it in our analysis.

A total of 2,648 non-duplicate references were identified, and, after screening based on titles and abstracts (first phase), 517 were retained for further assessment. Although the searches were done without language restriction, seven articles had to be excluded because they were written in languages not read by the authors (two Chinese ^{21, 22} , two German ^{23, 24} , onw Russian ²⁵ and two in Uzbek ^{26, 27} . After full texts were assessed (second phase), a total of 12 studies ^{28– 39} met the inclusion criteria. We present the complete PRISMA flow diagram in Figure 1. The complete list of excluded articles and reasons for exclusion is available upon request.

The included 12 studies report on research conducted in 2007–2020. They evaluated 13 vaccines from only six of the 18 vaccine manufacturers we identified as producing vaccines currently in use. The majority of the included studies (n=8) presented data from vaccines manufactured by Serum Institute of India Ltd ( Table 2, Supplementary Table 2, Extended data ¹⁴ ). Study designs included three RCT ^{28, 29, 32} , two uncontrolled prospective cohort ^{33, 34} , and seven cross-sectional studies ^{30, 31, 35– 39} . The studies were conducted in seven countries, none of which had introduced maternal pertussis vaccination at the time of study completion ( Table 2).

Detailed characteristics of the included studies are presented in Table 3. Only two studies assessed clinical outcomes ^{35, 36} , whereas the other studies assessed immunogenicity levels as main study outcomes. For studies that presented individual data over time (all except cross-sectional studies), most had a follow-up time of only one month after a given vaccine dose. Two of the RCT ^{28, 29} assessed the first booster dose (equivalent to fourth dose), reporting immunogenicity levels one month later. Three studies ^{34, 35, 39} assessed children who received a second booster dose (equivalent to fifth dose).

Methodological quality assessment of the included studies and reasons for judgement are presented in Supplementary Tables 3, 4 and 5 in the Extended data ¹⁴ . The quality of the 12 included studies varied significantly.

The two uncontrolled prospective cohort studies had serious ³³ or critical ³⁴ quality issues on Robins-I quality assessment tool (Supplementary Table 3, Extended data ¹⁴ ). Of the seven cross-sectional studies, two studies reporting clinical outcomes were assessed as of moderate overall quality ^{35, 36} . Cross-sectional studies reporting immunogenicity outcomes had heterogeneous quality assessments, with one of high risk ³⁷ , three moderate risk ^{31, 38, 39} , and two low risk ³⁰ based on the Joanna Briggs’ Institute checklist (Supplementary Table 4, Extended data) ¹⁴ . The 3 randomized controlled trials all had an overall high risk of bias (Supplementary Table 5, Extended data) ¹⁴ .

All the RCTs and uncontrolled prospective cohort studies had short follow-up periods ^{28, 29, 32– 34} , with a maximum of six weeks post-completion of the primary schedule (third wP dose). Five ^{30, 31, 37– 39} cross-sectional studies included children of older ages (six years of age and older), and only one of these ³⁹ evaluated children of 72 months of age who received a second booster dose ( Table 3).

We report study results separately for the 10 studies that reported immunogenicity outcomes and the two that reported clinical outcomes ( Table 4, panels A and B, respectivelly). For studies reporting on immunogenicity outcomes, GMC levels (EU/mL; 95% confidence interval [CI] or mean ± standard deviation [SD]), and seroconversion percentages (and 95% CI) at each timepoint are described in Table 4 (Panel A). One study ³³ evaluated two independent cohorts of individuals separately: although both cohorts received the same wP vaccine as the first booster dose, each cohort received a different vaccine in the primary series. We thus report results for these two cohorts separately, resulting in 13 sets of results in total.

We analyzed outcomes at five separate time points: up to 30 days post-third dose (primary series); pre-fourth dose/first booster (usually measured at 15-18 months); up to 30 days post-fourth dose; pre-fifth dose/second booster (usually measured at four-six years of age); and post-fifth dose, as depicted in Table 5.

The data show that immunogenicity levels, reported as mean GMT, were low right before the fourth (first booster) with all studies reporting levels below 11 EU/mL ^{33, 37, 39} , with three reporting values reaching above 40 EU/mL one month after the fourth-dose studies ^{29, 33} ( Figure 2). Similarly, low levels of GMT were reported prior to the fifth dose ^{34, 37– 39} , with the single study evaluating pre and post-fifth dose reporting levels of 8.4 EU/mL prior, and 30.2 EU/mL 30 days post booster ³⁴ .

When looking at seroconversion rates ( Figure 3), higher (over 75%) seroconversion was observed after completion of the three-dose primary regimen, after the first booster ^{28, 29, 33} , and after the second booster ³⁴ . The only two studies that reported low levels of serocoversion rates post-fourth dose ^{30, 31} are those that evaluated serocoversion post second booster later in time (at least one year after the booster), at three-five years of age ³⁰ , and one to two years after the booster ³¹ .

The results viewed in combination suggest that immunogenicity may wane substantially after the primary series and between booster doses.

How immunogenicity correlates with clinical outcomes is not totally clear. In the two cross-sectional studies reporting clinical outcomes ( Table 4, Panel B), infection rate of PCR-confirmed pertussis in children was heterogeneous, with 7.6% (6/79) post-third dose and 8.8% (8/91) post-fourth reported by Bailon et al. ³⁵ , and 37.5% (3/8) post-3rd dose and 50% (7/14) post-fourth reported by Dumaidi et al. ³⁶ .

All data underlying the results are available as part of the article and no additional source data are required.

Harvard Dataverse: Systematic review of immunogenicity and duration of immunity of currently licensed wP vaccines in children – Supplementary Files. https://doi.org/10.7910/DVN/XFRXGA ¹⁴

This project contains the following extended data files:

Harvard Dataverse: PRISMA flowchart and checklist for ‘Systematic review of immunogenicity and duration of immunity of currently licensed wP vaccines in children’. https://doi.org/10.7910/DVN/XFRXGA ¹⁴

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).