Pneumococcal Carriage in Burkina Faso After 13-Valent Pneumococcal Conjugate Vaccine Introduction and Before a Schedule Change

We examined changes in the pneumococcal carriage prevalence among healthy persons aged ≥1 month in Burkina Faso between 3 and 6 years after 13-valent pneumococcal conjugate vaccine (PCV13) introduction and before a change in the PCV13 schedule.

Streptococcus pneumoniae (pneumococcus) is the most common bacterial cause of deaths among children aged <5 years globally [1].In 2015, there were an estimated 9.2 million pneumococcal infections and 318 000 pneumococcus-related deaths among children aged <5 years worldwide; approximately half of the pneumococcus-related deaths occurred in Africa [2].
Widespread use of pneumococcal conjugate vaccines (PCVs) has led to decreased incidence in pneumococcal disease and declines in vaccine-type (VT) carriage among vaccinated children, resulting in decreased transmission of VTs from vaccinated children to unvaccinated children and adults (ie, herd protection) [2][3][4][5][6].The World Health Organization (WHO) recommends the inclusion of PCVs in routine childhood immunization programs using a 3-dose schedule, administered as 3 primary doses without a booster (3 + 0) or 2 primary doses with a booster (2 + 1) depending on programmatic considerations of the country.The 2 + 1 schedule has potential benefits over the 3 + 0 schedule due to the booster dose inducing higher antibody levels in the second year of life, which may provide greater herd effects due to a longer duration of protection against carriage acquisition among vaccinated children [7].
Burkina Faso is a low-income country in sub-Saharan West Africa located entirely in the African meningitis belt and experiences hyperendemic rates of bacterial meningitis [8].Following the introduction of Haemophilus influenzae serotype b vaccine in 2006 and meningococcal A (MenA) conjugate vaccine in 2010, S pneumoniae became the predominant cause of bacterial meningitis among all ages [8,9].Burkina Faso introduced 13-valent PCV (PCV13) into the childhood routine immunization program in October 2013 using a 3 + 0 schedule with doses administered at 2, 3, and 4 months of age with no catch-up campaigns.The WHO/United Nations Children's Fund national coverage estimate of the third dose of PCV13 has been >90% for most years following introduction [10].
Pneumococcal Carriage in Burkina Faso After PCV13 Introduction and Before a Schedule Change • OFID • 1
We conducted a cross-sectional community pneumococcal carriage study among persons aged ≥1 month approximately 6 years post-PCV13 introduction to establish a baseline before the PCV13 schedule change and to examine changes in overall and VT pneumococcal carriage from the aforementioned 2017 study to 2020.

Study Design and Participants
In March 2020, we conducted a cross-sectional, age-stratified community pneumococcal carriage study in Bobo-Dioulasso, a city located in Western Burkina Faso.We used the same recruiting methods as previous pneumococcal carriage studies conducted in Bobo-Dioulasso pre-and post-PCV13 introduction (2008,2015,2017) [11,16] and compared with data from the 2017 survey.Target sample size for each study (2017 and 2020) was 1000 persons aged ≥1 month with 200 participants each in 5 age groups: 1-11 months, 1 year, 2-4 years, 5-14 years, and ≥15 years.
We randomly selected 10 sectors among 21 eligible sectors (military and industrial sectors excluded) with 2 back-up sectors; in each sector, 20 crossroads were randomly selected with 10 back-up crossroads.At each crossroad, a street was randomly selected, and households were visited starting with the first on the left.The aim was to recruit 1 person in each age group at each crossroad to reach the target sample size.If there was >1 person in the same age group living in the same household, 1 was randomly selected.People living in the same household, but of separate age groups, were eligible for inclusion.The inclusion criteria for participants were resident of Bobo-Dioulasso, aged ≥1 month, and informed consent of an adult participant or of the parent/guardian of a child aged <18 years.The exclusion criteria were residence outside of Bobo-Dioulasso, severe malnutrition, or severe underlying disease.

Data and Specimen Collection
Data collection methods have been previously described [11].In brief, surveyors collected information on household characteristics and demographics from all consenting participants.Vaccination history was collected from children aged <5 years in 2017 and <7 years in 2020 using vaccination cards.The study team reviewed immunization registers in local health posts to collect vaccine histories for children without vaccination cards during enrollment.Enrolled participants were given appointment cards to visit a district hospital for clinical specimen collection.
At the district hospital, informed consent for clinical specimen collection was obtained and a second questionnaire was administered that included questions on the health history of participants.Trained nurses collected nasopharyngeal swabs from all participants and oropharyngeal swabs from participants aged ≥5 years [17].Swabs were immediately placed into cryotubes containing 1 mL skim milk, tryptone, glucose, and glycerol (STGG) transport medium and placed in coolers with ice packs.Specimens were transported to Centre Muraz, a national reference laboratory in Bobo-Dioulasso, within 4-6 hours.

Laboratory Methods
When specimens arrived at Centre Muraz, inoculated STGG was vortexed for 10-20 seconds before being stored in a −80°C freezer.For nasopharyngeal and oropharyngeal swab analysis, 200 µL of swab-inoculated STGG media was transferred to 5.0 mL Todd Hewitt broth containing 0.5% yeast extract (THY) and 1 mL of rabbit serum and incubated at 35°C-37°C for 6 hours.Cultured broth was plated on sheep blood agar and incubated in 5% carbon dioxide at 35°C-37°C.After 18-24 hours of incubation, plates were examined for the appearance of α-hemolytic colonies resembling streptococci.Pneumococci were identified by susceptibility to optochin and bile solubility test.S pneumoniae isolates were inoculated in a preservation medium STGG and stored at −80°C.
Pneumococcal serotypes were determined using published sequential multiplex polymerase chain reaction (PCR) assay [18].Quality control of the bacterial identification results obtained by Centre Muraz was performed at the Centers for Disease Control and Prevention (CDC) Streptococcus Laboratory in Atlanta, Georgia, for 20% of negative samples in 2017 and all negative samples in 2020.Pneumococcal isolates determined to be nontypeable (NT) or for which the serotype was unresolved by multiplex PCR were tested by Quellung reaction at CDC.Additionally, 20% of serotyped isolates were sent to CDC for quality control of the results obtained by Centre Muraz.
Data analyses.We conducted descriptive analyses of participants by study year and age group.We used χ 2 tests and Fisher exact tests to compare proportions.Changes in the overall, VT, and NVT carriage prevalence were examined between 2017 and 2020 among all participants and by age group.The carriage prevalence of individual serotypes in 2017 was compared to prevalence in 2020 for all age groups and all children aged <5 years combined; individuals carrying >1 serotype were counted in the numerator for each serotype.Crude prevalence ratios (PRs) were calculated using standard methods with 2017 serving as the reference period.We reviewed literature to identify potential confounders associated with pneumococcal carriage (inside cooking location, presence of other children aged <5 years in the household, antibiotic use in the past 2 weeks, and illness in the past 2 weeks) for a priori inclusion in the adjusted model [11,[19][20][21][22].We also looked at participant characteristics that changed from 2017 and 2020 (Table 1) and whether there were any significant differences  1) to inform our model selection.Adjusted prevalence ratios (aPRs) were modeled using logbinomial regression.Poisson regression using robust error variance was used if log-binomial models failed to converge [23,24].Data analyses were performed using SAS software (version 9.4).P values <.05 were considered statistically significant.

Patient Consent Statement
Prior to enrollment at the household, the study objectives were explained to participants in French or the local language, and written consent was obtained from all participants.At the district hospital, informed consent was obtained prior to the collection of clinical specimens.The study documents were approved by the Burkina Faso ethical committee.This activity was reviewed by CDC, deemed not research, and was conducted consistent with applicable federal law and CDC policy (see,

Overall Pneumococcal Carriage by Household and Participant Characteristics
In 2017 and 2020, households with presence of other children aged <5 years or with ≥4 persons sharing a room and participants with a cold/runny nose or cough had a significantly higher pneumococcal carriage prevalence than those without (Supplementary Table 1).Households using gas as a fuel source had a significantly lower pneumococcal carriage prevalence than households not using gas in both study years.There were no significant changes in the overall pneumococcal carriage prevalence by household and participant characteristics from 2017 to 2020.

DISCUSSION
We examined changes between 2 pneumococcal carriage studies conducted approximately 3 and 6 years after PCV13 introduction in Burkina Faso and before a change in vaccine schedule.We found no significant change in overall pneumococcal carriage (60.6% to 59.3%) whereas VT carriage decreased significantly (21.6% to 15.9%) in participants of all ages, although by age group, declines were only significant among children aged 5-14 years.Despite increased PCV13 coverage over time, in 2020, VT carriage was observed in 1 in 5 children aged <5 years and remains high in all age groups compared to middle-and high-income countries using schedules with a booster dose [14,25].A previous report comparing carriage studies conducted in Bobo-Dioulasso in 2015 (1 year post-PCV13 introduction) and 2017 (3 years post-PCV13 introduction, also included in the current study) showed reductions in VT carriage among pneumococcal carriers aged <5 years, although there were differences in the laboratory methods used for pneumococcal detection between the studies conducted in 2015 and 2017 [11].Between 2017 and 2020, we observed no further significant reduction in VT carriage among children aged <5 years.Persistent VT carriage has been observed in other sub-Saharan countries using the 3 + 0 schedule.In the Gambia, 5 years post-PCV13 introduction (preceded by 7-valent PCV introduction 2 years prior), VT carriage was 13.5%-14.4% in children aged <5 years [26]; in Malawi, 7 years post-PCV13 introduction, VT carriage was 15.7%-16.7%among children aged 3-8 years [22].In Mozambique, VT carriage was 14.5% in human immunodeficiency virus (HIV)uninfected children and 21.0% in HIV-infected children 3 years after 10-valent PCV introduction [27].Persistent VT carriage in younger children results in less pronounced indirect effects on carriage and disease for older ages.Preliminary meningitis surveillance data from Burkina Faso from 2018 through 2020 showed no significant declines in VT pneumococcal meningitis incidence for all age groups combined, and serotype 1, a PCV13 serotype rarely detected in carriage, remained the predominant serotype causing pneumococcal meningitis among older children and adults [28].High residual VT carriage and disease found in sub-Saharan countries, including Burkina Faso, prompted evaluations of strategies to optimize PCV impact including alternative schedules such as those with a booster.In 2019, WHO issued a new position paper stating that the 2 + 1 schedule could potentially provide a longer duration of protection than 3 + 0, although data are still limited [7].
The aforementioned report comparing results from early post-PCV13 introduction (2015 vs 2017) in Bobo-Dioulasso found no clear evidence of reductions in VT carriage in older children or adults [11].The reductions in VT carriage we observed in children aged 5-14 years (some of whom were PCV13 vaccinated in 2020) and adults (nonsignificant after adjustment) align with previous studies suggesting that it may take several years to develop herd protection.In Malawi, reductions in VT carriage among PCV13-unvaccinated HIV-infected adults were seen at 7 years post-PCV13 introduction using a 3 + 0 schedule (VT carriage 15.2% and 8.9%, 3.6 and 7.1 years post-PCV13 introduction, respectively) [22].Additionally, a systematic review of studies in high-and middle-income countries found that the mean time to reach a 50% and 90% reduction in invasive pneumococcal disease (IPD) among unvaccinated populations was approximately 2-4 years and 9-10 years (depending on PCV product) post-PCV introduction, respectively [6].Thus, the observed VT carriage reductions in persons aged ≥5 years could be explained by additional years of sustained PCV13 use in the community, with additional birth cohorts receiving PCV13.
In 2020, serotypes 3, 19A, and 19F were the most common VTs identified, with significant increases in the carriage prevalence of serotype 19A from 2017 (1.0%) to 2020 (2.7%) in children aged <5 years.In South Africa, serotypes 19A and 3 were the most identified VTs in IPD patients of all ages in the vaccine era, though neither serotype was significantly associated with increased in-hospital death [29].Serotype 19A has not been a major cause of pneumococcal meningitis in Burkina Faso [9], but additional monitoring for carriage and disease is warranted, especially after the PCV13 schedule change.
As expected, we observed increases in NVT carriage from 2017 to 2020 in children <5 years of age.Serotype replacement following PCV introduction has been well documented [26,[30][31][32].Specifically, we observed significant increases in serotypes 23B and 23A carriage among children aged <5 years and participants of all ages and serotype 10A among children aged <5 years only.Serotype 35B also remained as the most common NVT in participants of all ages and children aged <5 years in 2020.From 2011 to 2017, few pneumococcal meningitis cases caused by these NVTs have been reported [9].A systematic review estimating the IPD potential of individual serotypes in children aged <5 years in the PCV era found that serotypes 23B and 35B had a lower IPD potential when serotype 19A was used as the reference [4]; however, additional monitoring of serotype 23B and 35B carriage and disease is needed since some countries have reported increases in IPD due to these serotypes [33,34].
Our study is subject to several limitations.First, we did not account for the study cluster design or include sampling weights because we did not use a true probabilistic sampling design [35].While sectors and crossroads were randomly selected, back-up sectors and crossroads had a chance of being selected twice.We compared the carriage prevalence in sectors selected during the first round with the back-up sectors and found no differences in the carriage prevalence when stratified by age group.Additionally, at each crossroad the first household on the left was visited and surveyors moved consecutively to subsequent households, possibly leading to a bias if participants living closest to crossroads were more or less likely to be pneumococcal carriers.Second, fewer adult males (2017: 19.5%; 2020: 35.2%) were enrolled; however, a systematic review found no association between participant sex and pneumococcal carriage [19].Third, there could be residual unmeasured confounders contributing to significant reductions in overall pneumococcal carriage in children aged 5-14 years, therefore leading to significant reductions in VT carriage in this age group.Last, since we applied the same sample size for each age group, we may have been underpowered to detect significant changes in VT carriage in adults.
We observed sustained direct effects and the first evidence of indirect effects of PCV13 introduction on VT carriage in Burkina Faso.Despite declines in VT carriage from 2017 to 2020 among participants of all ages, there were no further significant reductions in children aged <5 years observed early post-PCV13 introduction [11], and VT carriage remained high.Approximately 6 years post-PCV13 introduction and in the context of high PCV13 coverage, 1 in 5 children aged <5 years still carries a VT pneumococci.Our study provides baseline VT carriage estimates before the PCV13 schedule change that occurred in June 2021.Additional monitoring of VT carriage is needed in the years following the schedule change to assess the impact.Evidence generated from future studies in Burkina Faso may inform pneumococcal vaccination policy decision-making for other countries in Africa using the 3 + 0 schedule.

Figure 1 .
Figure 1.Pneumococcal carriage prevalence by serotype among all ages in 2017 (N = 1005) and 2020 (N = 1002), Bobo-Dioulasso, Burkina Faso.*P < .05. a χ 2 tests and Fisher exact tests were used to compare changes in individual vaccine serotype carriage prevalence in each age group between 2017 and 2020.b Sixteen participants in 2017 and 5 participants in 2020 were colonized with >1 serotype.

d
Children aged 5 and 6 years are included in the age group 5-14 years.There were 65 children 5 or 6 years old enrolled in 2020.

Table 1 . Demographic and Epidemiological Characteristics of Enrolled Participants by Study Year-Bobo-Dioulasso, Burkina Faso, 2017 and 2020
Multiple responses possible in 2017 study while 1 response only possible in 2020 study.P value not calculated due to differences in the question format.
a χ 2 tests were used to compare 2017 versus 2020 categorical responses.b One response missing from 2017 study (N = 1004).c Three responses missing from 2017 study (N = 1002).d Pneumococcal Carriage in Burkina Faso After PCV13 Introduction and Before a Schedule Change • OFID • 3 in pneumococcal carriage prevalence by those characteristics (Supplementary Table

Table 2 . Pneumococcal Carriage Prevalence and Prevalence Ratios by Age Group-Bobo-Dioulasso, Burkina Faso, 2017 and 2020
Abbreviations: CI, confidence interval; NT, nontypeable; NVT, nonvaccine type; PR, prevalence ratio; VT, vaccine type.a Standard methods for calculating risk ratios were used to obtain crude PRs, with 2017 serving as the reference period.b