To the best of our knowledge, this is the first report of indirect protection of newborn infants against PCV-7 serotypes from Africa. We found significantly reduced carriage and risk of first acquisition of VT pneumococci within the first 8 weeks of life in infants born into communities where the whole population had been vaccinated with PCV-7 compared to those born in control communities where only those <30 months had been vaccinated. In parallel, overall carriage rate and risk of acquisition of non-vaccine serotypes (NVT) were increased in these infants compared to controls. A significant reduction in carriage and acquisition rate of VT was also demonstrated in both groups of villages when compared to findings in the previous baseline cohort study.
Protection against VT carriage has been demonstrated previously among American Indian infants living in PCV7-vaccinated communities compared to their counterparts living in control communities and this was attributed to reduced transmission of VT from vaccinated children 
. In addition, the proportion of isolates that were NVT was significantly higher in infants in vaccinated communities compared to controls. Indirect protection against VT carriage has also been demonstrated among children and adults living with a PCV-7-vaccinated child 
. The findings from these studies among American Indian children are similar to ours. The epidemiological characteristics of pneumococcal disease in the American Indian population is known to be similar to that seen in The Gambia, with high rates of carriage and invasive disease 
. However, our study has shown indirect protection of infants at a much younger age than in these studies and for the first time in a developing country setting.
Since pneumococcal transmission in the community is probably driven mainly by young children 
, as has also been shown previously in the Gambia 
, we expected to find little difference in carriage among the unvaccinated newborns. The absolute drop in VT carriage from baseline was about twice as much in infants from villages where the whole population had received PCV-7 than in infants from control villages. It is possible that a proportion of newborn infants in rural Gambia first acquire pneumococci from their mothers and that maternally acquired antibodies against VT (all mothers in vaccinated villages received one dose of PCV-7) could have played a role in the reduced VT carriage and acquisition rates seen in infants in vaccinated villages. Enhanced concentrations of IgG to specific pneumococcal serotypes have been demonstrated in Bangladesh 
and in Papua New Guinea 
among infants born to mothers who received pneumococcal polysaccharide vaccine in late pregnancy. A three-fold increase in anti-pneumococcal IgG in the colostrum was also demonstrated previously in Gambian mothers who received a pneumococcal polysaccharide vaccine in pregnancy 
. A similar study in the USA demonstrated enhanced IgG concentrations in infants of vaccinated pregnant women who received pneumococcal polysaccharide vaccine and lower nasal carriage of pneumococcal serotypes in the infants up to the age of 16 months 
In our study the proportion of isolates that were NVT was significantly higher in infants in vaccinated communities compared to those from control. Furthermore, though the risk of NVT acquisition was not significantly different between both study arms, it was significantly higher in each of the study arms when compared with findings from the baseline cohort study. Replacement carriage with NVT has been demonstrated in randomised controlled trials 
and observational studies 
. This trend in newborns in The Gambia is of concern given the increased frequency of NVT invasive disease which has been observed in some, but not all, populations where PCV-7 is routinely used 
. We noted no overall difference in the prevalence of VT or NVT serotypes by year since the start of the study. However the individual NVT serotypes 7F, 13 and 19A were all significantly more common in second year following vaccination, and the majority of the serotype 19A isolates were found only in second year in intervention villages. Serotype 19A has become the most frequently carried serotype and the leading cause of invasive disease and respiratory infections in the United States after the introduction of PCV7 
. We may have observed a true delayed replacement effect of 19A and this requires careful monitoring.
This study had some limitations. A follow up period of just 8 weeks, necessitated by the vaccination with PCV-7 of all infants in the study at this time, weakened the ability of the study to demonstrate whether the observed indirect effects would persist, disappear or even increase over time. The duration of carriage could not be assessed as most carriage losses were censored. We have used data from a longitudinal carriage study among infants born in the study communities nearly 2 years prior to the start of vaccination to make inferences about the extent of the indirect protection of newborns in the control villages. Though pneumococcal carriage rates have remained the same in Gambia in the years preceding PCV vaccination, temporal trends in the study villages could have reduced comparability between these data. However, whatever changes that might have occurred should have been distributed equally among the study arms because of randomisation and not affected the main conclusions of the study. The possible role of maternally acquired anti-pneumococcal antibodies in reducing carriage in infants in vaccinated villages could only be speculated on as this was not studied. The study was designed to detect differences in carriage of any VT or any NVT pneumococci. We conducted further analyses of individual serotypes. The chance of finding spurious effects is increased in these secondary analyses, due to multiple comparisons, and consequently these findings for individual serotypes need to be interpreted with caution.
In summary, this randomised controlled trial in rural Gambia has demonstrated that PCV-7 vaccination results in strong indirect protection against pneumococcal carriage of vaccine serotypes during the first 8 weeks of life. Comparison of the differences in pneumococcal carriage between communities where the whole population was vaccinated, communities where only under 30 month old children were vaccinated and the same villages prevaccination suggest that this strong indirect effect comes from both under 30 month and over 30 month old vaccinated populations. While evidence of replacement carriage with NVT necessitates the need for continued surveillance, the combined direct and indirect effects of PCV-7 enhance the recommendation for widespread implementation of PCVs into routine vaccination schedules in countries where they are needed most.