Three years after PCV7 implementation in the NIP for all newborns, parents of 11,005 children were invited to participate. A total of 1,045 (9.5%) parents were interested in participating, among which 892 families were assessed for eligibility; 153 families were not assessed because the enrollment target had already been achieved. Of the 892 families, 233 were excluded. The most frequent reason for exclusion was household presence of older siblings who had received a pneumococcal conjugate vaccine (78%). In total, 329 children 11 months of age and 330 children 24 months of age and 1 parent for each of the 24-month-old children (n = 324) were enrolled from February 9 through July 9, 2009.

The mean age of vaccinees was slightly lower (4 weeks at 11 months, and 1 week at 24 months of age; both p<0.001) than that of the pre-PCV7 controls. At 11 months, the proportion of siblings <5 years of age and the exposure to smoking indoors was lower for vaccinees (both p<0.001). At 24 months, the proportion of male participants and the proportion of children who had received antimicrobial drugs within 1 month before the swab was taken was higher among vaccinees (both p = 0.03). Also, the months of sampling differed between both studies (p<0.001; Table 1).

Three years after NIP implementation of PCV7 for all newborns in the Netherlands, PCV7-serotype carriage of S. pneumoniae was reduced 80%–90% among vaccinated children at 11 and 24 months of age. Among parents of vaccinated children, carriage of PCV7 serotypes had almost disappeared. This impressive reduction of PCV7-serotype carriage in infants is larger than that observed in clinical trials, which showed 50%–60% reduction in PCV7 rates after conjugate vaccination, and should be attributed to herd effects (13,18,19). Herd effects would also account for the disappearance of PCV7 serotypes in parents. This large effect might in part be a result of a high pneumococcal vaccine uptake because 94.4% of all 2-year-old children in the Netherlands have been fully vaccinated (20). Our data confirm the major role of young infants in the transmission of pneumococci in the community.

Herd effects may also have contributed to the reported unexpectedly high reductions of otitis media (by 43%) (21) and all-cause pneumonia (by 33%) (22) in young children in the United States since PCV7 introduction. These reductions exceed overall vaccine efficacy found in randomized controlled trials: 6%–9% reduction of otitis media (9,23), and 4% reduction of all-cause pneumonia (24). However, a recent US study on community-acquired pneumonia (with radiographic confirmation) found no consistent reductions in pneumonia rates among children and adults, except for children <1 year of age (25). Whether this finding is the result of replacement disease by other nonvaccine pneumococcal serotypes, other pathogens, or other causes remains to be evaluated. Nasopharyngeal serotype replacement remains a potential drawback of vaccination with pneumococcal conjugate vaccines.

Increased rates of carriage of nonvaccine serotypes were also observed in this study. In vaccinated infants and their parents; serotypes 19A and 11A were the most frequently carried serotypes in the Netherlands. In the United States, multidrug resistant serotype 19A has become a frequent cause of IPD as well as of otitis media in children (26,27). There is ongoing debate about the actual role of PCV7 introduction and the increase in serotype 19A; antimicrobial drug pressure and secular trends have been emphasized (28). In our study, however, post hoc susceptibility testing of all 19A isolates showed a low prevalence of nonsusceptible strains among controls and vaccinees. In addition, our group previously reported a significant increase in serotype 19A carriage after PCV7 vaccinations in a study conducted in a randomized controlled setting (29), excluding secular trends and indicating a direct role of PCV7. A trend toward lower carriage rates of serotype 6A and higher carriage rates of serotype 6C was observed in both age groups, suggesting PCV7 cross-protection for serotype 6A but not for serotype 6C, in line with other carriage studies (30). However, serotypes 11A and 6C have not yet been reported as a frequent cause of IPD in the Netherlands (12).

Observed changes in prevalence of serotype carriage may not be entirely random but may be directly related to the serotype capsule size, which in turn is related to the polysaccharide composition and metabolic costs of the capsule for the bacterium (31). Pneumococci with larger capsules are more resistant against nonopsonic phagocytosis and more commonly colonize young children. Our results agree with results of Weinberger et al., which show a significant increase in carriage of highly encapsulated serotypes such as 19A, 11A, 10A, and 35F (31). Furthermore, the serotype-specific capsule has been shown to be a major factor in the potential to cause IPD, independent of genetic background and temporal or geographic settings (32). This serotype-specific difference in disease potential has also been shown for mucosal infections, although the differences between serotypes were less apparent compared with differences in invasive potential (33). In addition, serotypes are independently associated with IPD severity (34,35). Harboe et al. showed that highly encapsulated and frequently carried serotypes such as 11A, 10A, and 19A have high mortality rates among healthy persons >5 years of age that are comparable to PCV7 serotypes such as 19F or 6B. Therefore, replacing serotypes in carriage may potentially cause equally severe disease. Furthermore, frequently carried serotypes are more likely to affect patients with concurrent illnesses than are infrequently carried serotypes with high invasive disease potential such as serotypes 1 and 7F (36). Replacement with highly encapsulated pneumococci may therefore substantially reduce vaccine benefits, especially for those who are older or have chronic disease (37). However, in addition to the prevalence in carriage and the disease potential of the serotype, the prevalence and severity of pneumococcal disease are also associated with genetic background and presence of drug-resistant clones (38) and depends on population or patient characteristics (36). As a consequence, it is crucial to monitor and critically evaluate all of these aforementioned aspects.

In contrast to several other carriage studies (11), our study found a significant reduction (20% at 11 months and 17% at 24 months of age) in overall pneumococcal carriage 3 years after the PC7 vaccination program began. Previously, we reported a 10% decline in overall pneumococcal carriage after PCV7 vaccination in a randomized controlled trial setting in the Netherlands after reduced-dose schedules and before implementation of PCV7 in the national immunization program for children (13). Herd effects may have contributed to the larger reduction in PCV7-serotype and overall pneumococcal carriage in the present surveillance study. We must, however, be cautious about ascribing the reported reduction in overall pneumococcal carriage in children to the introduction of PCV7 because this was an observational study and other unmeasured factors such as viral infections, seasonal variations, and temporal trends could not be taken into account (39). Among parents, overall pneumococcal carriage did not change, but low carriage rates in adults make it harder to detect significant changes. Because the same study procedures were followed by the same well-trained research nurse team and laboratory personnel in both studies, we do not think that the observed carriage reduction is an artifact.

For evaluation of our study results, some potential limitations should be taken into account. First, because our data are observational, we showed associations and no causalities between the introduction of PCV7 and changes in pneumococcal colonization. Although several potential confounders were measured and appeared to differ between the comparison groups, multivariate analysis showed that our results were quite robust. Second, the postvaccination data came from a cross-sectional cohort study including 2 separate age groups, whereas the pre-PCV7 control data were derived from a longitudinal study in which data were collected from the same children at age 12 and 24 months. However, we previously found that potential within-person dependency was not substantially affecting these carriage data, probably because of the large interval between carriage sample collections (13). Third, the study was not adequately powered to evaluate serotype-specific differences. Lastly, we used a single-colony method for serotyping in both studies. Currently, improved techniques for detection of multiserotype carriage, e.g., the newer, more sensitive PCRs, are available. Multiple serotype carriage methods might have revealed more nonvaccine strains in both studies, pointing to unmasking instead of true replacement after eradication of vaccine strains. The strengths of our study are the relatively high carriage rates found in both studies in the Netherlands compared with other Western countries, a high PCV7 uptake in the NIP, and the possibility to evaluate the effect of vaccination with PCV7 on pneumococcal carriage in adult contacts. Also, drug-resistant clones do not confound the results because antimicrobial drug use and consequent resistance are low in the Netherlands compared with other European countries (40).

Since 2009, pneumococcal vaccines with broader coverage have been licensed and will be introduced into vaccination programs worldwide. The effects of these broader coverage vaccines on potential shifts in pneumococcal serotypes in the nasopharynx are still largely unknown. To predict the long-term health and economic effects, close monitoring is warranted.

In conclusion, 3 years of vaccination with PCV7 has led to impressive shifts in serotype-specific carriage of S. pneumoniae in children and their parents. This finding indicates a major role of infants in transmission of pneumococci in the population.