The frequency of S. pneumoniae NP carriage was investigated among children attending two distinct settings in a large metropolitan area in Brazil. An overall prevalence of 49.2% was found, and it was considerably higher at the DCC when compared to the PH (58.8% vs. 42.1%). The most common characteristics associated with pneumococcal colonization observed in this study (day care attendance and cohabitating with young siblings) were also found in recently published studies carried out in Taiwan  and in the USA , reinforcing our findings. Other factors might have accounted for the high prevalence rate in the present study, such as the low socioeconomic status of the analyzed population and crowding .
Data from the SIREVA II project (2010) revealed that the most prevalent serotypes associated with IPD among children less than 5 years old in Latin America were 14, 6B, 19A, 1, 23F, 6A and 19F, accounting for approximately 60% of these diseases . Although we have not analyzed invasive isolates, these serotypes, except for serotype 1 which was not detected, represented 56.2% of the isolates recovered from the nasopharynx of children included in this study. Such observations highlight the importance of pneumococcal carriage, since frequency of isolation was very similar to the one observed for IPD, and previous colonization is known to be a step usually required for the development of invasive infections.
Considering that the currently available conjugate vaccines against S. pneumoniae may contribute to eliminate the asymptomatic carriage, PCV7 and PCV10 would have an identical theoretical impact (45.5%) in the analyzed population, since the additional serotypes represented in the PCV10 (1, 5 and 7F) were not detected. Regarding PCV13, which contains serotypes 3, 6A and 19A in addition to those provided in PCV10 and is also approved for use in Brazil, the projected coverage for pneumococci recovered in the present study was 59.5%. Nevertheless, in Brazil, this vaccine is currently available only in private clinics. By setting, this impact would be considerably lower among children attending the DCC when compared to those at the PH, as for PCV10 (34.4% vs. 56.7%) and for PCV13 (47.5% vs. 71.7%).
The reason behind this very low estimated coverage in the DCC is the prevalence of serotypes 15C, 17F and 11A, ranked from 2nd to 4th position respectively at this scenario, which are not included in any of the available conjugate vaccine formulations. These serotypes have also been found in the nasopharynx of unvaccinated children aged less than 5 years in a previous Brazilian study, but they are rarely associated with invasive diseases in Latin America and other regions of the world [7, 20]. Interestingly, serotype 15B/C, together with 6C and 19A, was one of the serotypes most commonly isolated from children aged <7 years in primary care practices in Massachusetts, USA, where serotype replacement post-PCV7 is considered essentially complete . These serotypes are already circulating among the children included in the present study, mostly at the DCC where they account for almost 25% of the pneumococcal isolates, and this finding can predict serotype replacement post-PCV10 in our region.
We have also observed association between serotypes, settings and penicillin non-susceptibility, since serotypes 6B and 19F were strongly associated with children at the DCC and the PH, respectively, whereas penicillin non-susceptibility was more common among pneumococci isolated from children at the PH, being frequently related to serotypes 14 and 23F.
As previously demonstrated , colonization rates tend to be higher during respiratory tract infections. Although, we have not observed this fact globally in the present study, serotype 19F carriage was strongly associated with children who presented respiratory symptoms, such as coryza/sneezing and cough/expectoration. Also, it was possible to observe that PNSP colonization was very frequent among children presenting all the symptoms assessed, both general and respiratory.
Despite the strong association of PNSP with serotypes 14 and 23F, non-susceptibility to penicillin was also found among serotypes 6A, 6B, 10A, and 19F. These results are in accordance with data in the literature, showing that most PNSP isolates belong to serotypes 6A, 6B, 9V, 14, 15A, 19F, 19A, and 23F [7, 21, 22]. Although this association is well-documented, non-susceptibility to penicillin has already been found in other serotypes, such as 3, 6C, and 18C , and in the present study, serotype 10A.
In theory, vaccine coverage projected for isolates recovered in the present study could potentially reduce penicillin non-susceptibility rates from 27.3% to 7.4% or 5% with the use of PCV10 or PCV13, respectively. However, other authors have described no or low impact on PNSP carriage post-PCV7 immunization [23, 24]. Also, in a recent study carried out in Korea after optional use of the PCV7, rates of penicillin non-susceptibility were found to increase significantly and were strongly associated with non-vaccine serotypes 6A and 19A .
High rates of non-susceptibility to sulphamethoxazole-trimethoprim have already been described among invasive and carriage isolates recovered from Brazilian children and adolescents, corroborating the findings of the present study [7, 26, 27].
The low rate of erythromycin resistance observed in the present study was similar to that obtained for isolates recovered from individuals under 15 years of age with pneumonia , but considerably lower to that (15.0%) reported in 2010 on the SIREVA II project among children with invasive infections aged < 5 years .
Despite high rates of non-susceptibility to penicillin and sulphamethoxazole-trimethoprim, the pneumococcal isolates recovered from nasopharynx of the children analyzed in this study showed a high degree of susceptibility for the majority of the antimicrobial agents tested and, as previously reported [19, 20], this may also reflect the characteristics of the pneumococcal isolates associated with diseases in our region.