Streptococcus pneumoniae is a major cause of community-acquired pneumonia and meningitis worldwide, particularly in low- and middle-income countries (1). Although all age groups are affected, the highest incidence rates of pneumococcal disease occur in young children and older adults (1, 2). Pneumococcal bacteria also frequently colonize the respiratory tract of asymptomactic carriers. Nasal carriage and related pneumococcal disease have been shown to be influenced by the presence of respiratory viruses, particularly during influenza seasons (3, 4).

Prevention of pneumococcal pneumonia involves the 23-valent pneumococcal polysaccharides vaccines (PPSV23) for adults and pneumococcal conjugate vaccines (PCV) for children. Lao PDR introduced the 13-valent PCV (PCV13) into the national childhood vaccination program in October 2013, using a 3 + 0 schedule at 6, 10 and 14 weeks of age (5). Catch-up vaccination was offered to infants up to 12 months old during the first roll-out. National coverage of the third dose of PCV13 among infants was estimated at 89% in 2020 (6).

Most previous studies on S. pneumoniae burden and serotype distribution in LMICs, including in Lao PDR, have been based on healthcare facility surveillance in infants and young children (5, 7, 8). To better understand S. pnuemoniae epidemiology and transmission, including risk factors for carriage or infection, and the impact of vaccination on circulating serotypes, prospective community-based studies, across all age groups, are also needed.

The longitudinal surveillance study of respiratory infectious agents in metropolitan Vientiane, Lao PDR (LaCoRIS) is a community-based cohort study that monitored the etiology and burden of influenza-like illness among residents of this region from 2015 to 2019. The findings of the first year of surveillance, covering over 20 respiratory pathogens, have been published elsewhere (9), with S. pneumoniae the most comon bacterial agent detected in cases of influenza-like illness (ILI) (9). In this paper, we focus on S. pneumoniae as the main outcome of interest, to report on prevalence, risk factors and serotype distribution of pneumococcal carriage among ILI cases in the LaCoRIS study during the three surveillance years.

Full details on the LaCoRIS study design, including recruitment and surveillance of households in the cohort, are reported elsewhere (9). In brief, participants were recruited from randomly selected households in 25 villages in metropolitan Vientiane. Participant enrolment into the cohort was conducted at two main timepoints. The first enrolment phase began in April 2015, during which a total of 4,855 individuals from 995 households were recruited and followed-up for ~1 year (until 31 May 2016), as reported previously (9). The cohort study then resumed in March 2017 (i.e., approximately a year after the end of the first surveillance period), with a second enrolment phase conducted to initiate a further 2 years of ILI surveillance. For this second phase, all previously recruited households were invited to participate, along with additional, randomly selected households within the same villages, to account for loss of follow-up. Thus, ILI surveillance was conducted among an open cohort over three surveillance years i.e., year 1 (from 28/04/2015 to 31/05/2016), year 2 (from 01/03/2017 to 28/02/2018), and year 3 (from 01/03/2018 to 28/02/2019).

During these surveillance periods, active case finding for ILI was conducted through weekly phone calls to each participating household. Individuals fulfilling the WHO case definition for ILI were visited to administer a questionnaire and collect a nasopharyngeal swab, after obtaining informed written consent (National Ethics Committee 060/NECHR; 1 January 2017). Follow-up questionnaires were administered to ILI cases ~40 days after the sample collection and case investigation, to obtain information on clinical outcomes (e.g., hospitalization or sequelae) that had since occurred in relation to the ILI episode.

The FTD^® Respiratory pathogens 21 Plus technology was used for real-time Polymerase Chain Reaction (qPCR) testing of swabs, as detailed previously (9). Detection of S. pneumoniae with this assay targets the lyt A gene, a largely conserved gene encoding the major autolytic enzyme of pneumococcus. This approach is considered a gold standard among culture-independent S. pneumoniae assays, with reported sensitivity and specificity of 100 and 99.5%, respectively (10), with a detection limit of one plasmid copy per 10μl reaction (11). S. pneumoniae-positive samples were then screened for 40 different serotypes using a multiplex qPCR assay, following published protocols (12). This assay has shown high sensitivity for the included serotypes, able to detect < 100 colony forming units (CFU) per reaction for the majority (91%) of strains tested (12).

All statistical analyses were conducted in R version 4.2.2. Incidence of ILI and 95% confidence intervals (CI) were estimated using complex survey analysis methods, adjusting for the demographic structure (age and sex) of the Vientiane urban population and clustering in the survey design, as detailed previously (9, 13). The prevalence of pneumococcal detection among ILI cases, and the proportional distribution of serotype categories among pneumococcal-positive ILI cases, were calculated and assessed for associations with age-group, surveillance year, and climatic season (wet vs. dry months) using chi-squared tests. Binomial generalized linear mixed models (GLMMs), adjusting for random effects at household-level, were used to further investigate risk factors for pneumococcal positivity among ILI cases using the “lme4” R package (14). Bivariable GLMM analyses tested for associations with each independent variable separately, while multivariable GLMMs included age group, sex, and any covariates which showed associations with P < 0.05 in the bivariable models. We were unable to adjust for individual level random effects in these GLMMs as the majority of study subjects with ILI experienced only a single episode throughout the surveillance period (Models specifying two nested levels of random effects, i.e., for both households and individuals, displayed poor convergence and singularity issues, indicating that these were overly complex for the dataset). However, to assess whether the results of our GLMM analyses may have been influenced by individual-level random effects which could not be adjusted for in the models, we also repeated these analyses with a subset of the data which included only a single ILI episode per subject. Simpson's Diversity Indices with 95% Cis were calculated using the “simboot” package (15) to assess for differences in diversity of serotypes between age groups and surveillance years, following Løchen et al. (16).

Throughout the study period, a total of 1,142 households (5,690 individuals) were recruited. Of 995 households (4,885 individuals) enrolled in the first surveillance period (2015/16) (9), 936 households (94.1%) enrolled for the second phase (2017–19), along with an additional 147 new households (765 individuals). The mean duration of ILI surveillance for each household was 148 weeks (range 51–141 weeks), with a total follow-up size of 16,104 person-years across all households and surveillance years.

Few studies have reported on pneumococcal detection and serotypes among ILI cases, particularly among cases identified through community-based surveillance as opposed to surveillance at healthcare facilities. In this study, we describe the prevalence and risk factors of nasopharyngeal pneumococcal detection among subjects with ILI in Lao PDR. In addition to higher prevalence in younger age groups, pneumococcal positivity was significantly associated with passive smoking and concurrent detection of Hemophilus influenzae. Detection of PCV13 serotypes was observed to decrease over the study period, although this trend was only significant in ILI cases under 5 y.

The inverse relationship between age and pneumococcal carriage has been widely reported elsewhere, including in studies among healthy children in Lao PDR (7) and other countries in Southeast Asia (17).

While the high prevalence of pneumococcal carriage in young children could implicate S. pneumoniae as an important etiological pathogen of ILI in this population, it is important to acknowledge that the clinical significance of pneumococcal detection among our study participants is unclear. S. pneumoniae bacteria commonly colonize the nasopharynx without leading to invasive disease, and ILI symptoms can be caused by a diverse range of viral and bacterial pathogens. We cannot provide evidence of a temporal association between ILI occurrence and pneumococcal acquisition/infection during the study period because only a single nasopharyngeal specimen was collected during the acute phase of ILI. Furthermore, we did not collect nasopharyngeal swabs in study participants without ILI. Thus, we refer to pneumococcal-positivity among ILI cases in our study as “carriage”, rather than pneumococcal infection or disease. While it is possible that some cases may have represented bacterial pneumonia initially presenting as ILI, our follow-up data on hospitalizations do not provide evidence support this. Hospitalizations among ILI cases were low, and not associated with S. pneumoniae detection. Moreover, all three pneumococcal-positive cases who were briefly admitted were also positive for at least one respiratory virus.

Consistently with other studies (18, 19), we identified passive smoking as a risk factor for pneumococcal detection. Exposure to cigarette smoke has been shown to impair nasal mucociliary clearance and increase pneumococcal adherence to respiratory epithelia (20–22), which may explain the associations observed in our study. Carriage of Hemophilus influenzae was also significantly associated with increased risk of pneumococcal positivity. A synergistic association between S. pneumoniae and H. influenzae colonization has been reported among Australian Aboriginal (23) and Italian children (24). The latter study also found that children vaccinated with PCV were even more likely to be colonized by H. influenzae. Commensal coexistence of these bacteria within a biofilm in the human nasopharynx has been suggested as a mechanism enabling their persistence in the host (25).

Similar to a study among hospitalized children with acute respiratory infection in Lao PDR between 2013 and 2019 (5), VT strains were found in 41% of pneumococcal-positive ILI cases, and accounted for 38% of all pneumococcal isolates detected. The significant decrease in VT detection between 2015/16 and 2018/19 among ILI cases aged < 5y was likely due to the impact of vaccination; all ILI cases in this age group that occurred in 2015/16 were born prior to October 2013 (i.e., before PCV13 introduction in Lao PDR), while the ILI cases aged < 5y in 2018/19 were born during or after this month, and therefore more likely to have been vaccinated. However, the relative contributions of direct vs. indirect effects of PCV13 on VT carriage in young children cannot be ascertained through our data. The cohort study was designed to investigate the etiology and epidemiology of ILI in the community, rather than PCV13 effectiveness in young children, and data on vaccination status of study participants were not available. Nonetheless, the temporal decrease in the proportion of VT isolates also seen among older age-groups (who would not have been vaccinated), while not statistically significant, may point to an emerging indirect (herd) effect due to decreasing circulation of PCV13 serotypes. Cross-sectional carriage surveys pre- and post-PCV have similarly reported evidence of the indirect effect of PCV13 in Lao PDR (5, 8). We also note that, despite the temporal decline in VT detection, we did not observe a significant change in overall pneumococcal prevalence during the study period. This may represent preliminary evidence of serotype replacement, although longer term studies are needed to further support this.

Similar to our results, serotypes 6A/B, and 19F have been found as the most prevalent PCV13 serotypes among 12–23-month-old Laotian children both before and after PCV13 introduction (8). The most common NVT serotypes in our study were 11A, 35B, 15A and 20. Serotypes 15A and 11A have been identified as among the top 10 serotypes causing invasive pneumococcal disease post-vaccination in France and Norway, respectively (16). Further investigation is warranted on the need for broader pneumococcal vaccines in Lao PDR, including the PPSV23, PCV15 or PCV20 which have been introduced, for example in older age groups, in other countries.

Two key limitations of our study are that, as discussed above, we are unable to draw strong conclusions regarding clinical significance of pneumococcal carriage in our study population, or the precise role of PCV13 introduction on changes in serotype distributions. Indeed, these reflect challenges associated with S. pneumoniae surveillance in ILI cases more generally. The use of serological assays (26, 27), alongside nasopharyngeal qPCR testing, could help elucidate the clinical significance of pneumococcal detection in future ILI studies. Nonetheless, our study illustrates how samples and data from community-based ILI surveillance can still be leveraged, for example to identify risk factors, circulating serotypes, and emerging trends in which may indicate serotype replacement. Another limitation is that serotype data could not be obtained for 21.1% of pneumococcal-positive swabs. S. pneumoniae screening was based on PCR detection of the lyt A gene, an important virulence marker which is thought to play a role in nasopharyngeal colonization (28). Realtime PCR targeting of lyt A is considered one of the gold standards among culture-independent S. pneumoniae assays, showing high diagnostic sensitivity and specificity (10, 11, 28). However, lyt A positivity has also been reported for some non-encapsulated pneumococci, as well as a small number of non-pneumococcal isolates (e.g., Streptococcus pseudopneumoniae carrying lyt A homologs) (10). Thus, while it seems likely that most of our “non-typeable” samples represent encapsulated S. pneumoniae serotypes (i.e., which were not included in the multiplex typing PCR), it is possible that some were due to non-encapsulated strains or even non-pneumococci, which are of lower clinical and vaccine importance. We also acknowledge the possibility that variation in detection limits of the serotyping assay for different strains could have had some influence on the observed serotype and NT frequencies (12).

In conclusion, we found pneumococcal carriage among ILI cases in Vientiane to be associated with younger age, passive smoking, and concurrent H. influenzae colonization. Our findings contribute also to the growing body of evidence from low- and middle-income countries indicating the effect of PCV13 on serotype distributions. Analysis of S. pneumoniae carriage data among ILI cases detected through community-based surveillance, such as the LaCORIS study, can complement data from surveillance at health facilities, to provide a more complete picture of pneumococcal burden and circulating serotypes across all age-groups, including associations with other respiratory pathogens. Such evidence can help inform future studies aiming to evaluate the impact of vaccination on community transmission, and identify which serotypes may be important to consider in future vaccines.

The datasets presented in this article are not readily available because it is confidential information that will require the authorization of several study partners including the Funder, MoH, and Ethics Committee. Requests to access the datasets and R scripts should be directed to the corresponding author VSP.

The studies involving human participants were reviewed and approved by the Lao National Ethics Committee for Health Research in Lao PDR (060/NECHR; 1 January 2017). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

VSP designed the study. IK, PP, and BF coordinated and supervised data collection. JR analyzed the data. MS-E, JR, and VSP drafted the manuscript. All authors reviewed and approved the final manuscript.