Simultaneous colonization of the human nasopharynx with more than one strain or serotype of S. pneumoniae
is an intriguing event. It sets the stage for pneumococcal evolution through horizontal exchange of genes, which may result in the acquisition of antibiotic resistance or capsule switch 
. In addition, the study of cocolonization offers insight into the interaction between different strains or serotypes of S. pneumoniae
. This study shows that PCV7 selection pressure did not affect the cocolonization rate up to three years after the introduction of the vaccine. Therefore, opportunities for horizontal gene transfer between pneumococci in the nasopharynx will continue to occur at the same frequency in the vaccine era as in the prevaccine era and new strains may emerge under vaccine selection pressure as has been described 
One might have expected that cocolonization would involve more frequently the prevalent serotypes with high incidence of acquisition and long duration of carriage 
. However, in this study, rare serotypes predominated among cocolonizing pneumococcal strains. For example, serotype 19F, one of the most carried serotypes, was relatively rare among cocolonizing strains already in the prevaccine era and serotype 19A did not emerge among cocolonizing strains in the vaccine era as it did among single colonizing strains. The observed predominance of rare serotypes during cocolonization suggests that simultaneous colonization is not simply a consequence of cumulative exposure to transmission opportunities. This hypothesis is also supported by the observed lack of an association of cocolonization with day care usage, a history of recurrent otitis media and recent antibiotic exposure.
It has been shown that pneumococcal serotypes differ for their competitiveness in colonizing the nasopharynx 
; some serotypes, such as 6B can outcompete others, such as 23F. In our study, colonization density was significantly higher in nasopharyngeal swabs containing multiple pneumococcal strains as compared to those swabs with a single strain. Also, the median ratio of two pneumococcal strains in a given sample was balanced (median 1.3). Therefore, cocolonizing strains seem to co-exist rather than to compete with each other for example for resources (i.e. nutrients, space) or by production of bacteriocins. Margolis and colleagues recently showed in a rat colonization study that sequential cocolonization with two pneumococcal strains can lead to stable co-existence and a higher density for cocolonization compared to single colonization 
. Whether the tendency for co-existence between some pneumococcal serotypes, as described here, may also apply to co-existence with other pathogens such as H. influenzae
and S. aureus
needs to be studied 
Serotypes, which predominated in multiple colonization in this study, did not include serotypes covered by the seven-, ten- or thirteen-valent conjugated polysaccharide vaccines, and the prevalence of these rare serotypes increased in the vaccine era (). Therefore, we can not exclude that rates of multiple colonization may increase in the future with the emergence of these serotypes. The follow-up period of three years after introduction of PCV7 may have been too short to observe such a trend.
A limitation of this study is that the study population was restricted to patients with acute otitis media or pneumonia. The conclusions drawn from this study may therefore not apply to healthy persons. Also, the epidemiology of cocolonization may be different in populations with overall higher carriage rates than observed in our study 
. Multiplex PCR for the detection of pneumococcal serotypes has also been used for the detection of cocolonization 
. Whilst multiplex PCR for serotype cannot detect cocolonization by different pneumococcal strains of the same serotype the method used in this study, plyNCR
RFLP and T-RFLP, will miss cocolonization by two serotypes with an identical RFLP type. We compared the performance of multiplex PCR and T-/RFLP for the detection of cocolonization in 141 consecutive nasopharyngeal swabs (from time period 4) and found a slightly higher sensitivity for the T-/RFLP assay than multiplex PCR: cocolonization was detected by both methods in six samples, by T-/RFLP only in 8 samples, and by multiplex PCR only in one sample (). Therefore, we believe that this study did not underestimate the cocolonization rate in comparison to studies using multiplex PCR.
Comparison between multiplex PCR and plyNCR RFLP/T-RFLP for the detection of multiple colonization in 141 nasopharyngeal swabs collected during study period 4 for which culture revealed growth of a pneumococcal strain.
In summary, this study showed that pneumococcal cocolonization rates have not been affected by the selection pressure of PCV up to 3 years after the introduction of universal vaccination. Multiple colonization involves more often serotypes that have been of low prevalence in the prevaccine era, but may now emerge under vaccine selection pressure. This may lead to an increase in the cocolonization rate in the future, with an increasing opportunity for pneumococcal evolution through horizontal gene exchange.