Clin Microbiol Infect 2011; 17: 907–914
A prospective cohort study of preschool healthy children (3–6 years old) from two distinct socio-economic settings in the Brussels area, Belgium, was conducted during the years 2006–2008. The objectives were to evaluate nasopharyngeal colonization by Streptococcus pneumoniae, Staphylococcus aureus, Moraxella catarrhalis and Haemophilus influenzae at the time of PCV7 vaccine introduction and to assess the socio-economic level impact on flora composition and antibiotic resistance. Three hundred and thirty-three children were included and a total of 830 nasopharyngeal samples were collected together with epidemiological data. Pneumococcal serotypes and antibiotic resistance profiles were determined. Risk factors for carriage and bacterial associations were analysed by multivariate logistic regression. Carriage rates were high for all pathogens. Fifty per cent of the children were colonized at least once with S. aureus, 69% with S. pneumoniae, 67% with M. catarrhalis and 83% with H. influenzae. PCV7 uptake was higher among children from a higher socio-economic setting and S. pneumoniae serotypes varied accordingly. Children from lower socio-economic schools were more likely to carry M. catarrhalis, S. aureus and antibiotic-resistant S. pneumoniae, including a high proportion of non-typeable pneumococcal strains. Positive associations between S. pneumoniae and H. Influenza, between H. influenzae and M. catarrhalis and between H. influenzae and S. aureus were detected. Our study indicates that nasopharynx flora composition is influenced not only by age but also by socio-economic settings. A child’s nasopharynx might represent a unique dynamic environment modulated by intricate interactions between bacterial species, host immune system and PCV7 immunization.
Streptococcus pneumoniae, Staphylococcus aureus, Haemophilus influenzae, and Moraxella catarrhalis are aetiologic agents of a wide range of diseases, from minor infection to life-threatening invasive diseases, but are also carried without clinical consequences [1–4]. Carriage is the first step in disease development  and allows horizontal spreading within the community. The nasopharynx of young children presents specific characteristics: (i) an immature immunological response, (ii) increased microbial exchanges due to enhanced contacts, and (iii) high susceptibility to viral respiratory infections [5–8]. Thus, a child’s nasopharynx represents a major reservoir of specific bacterial species that undergo complex interactions and evolution in terms of resistance and virulence [9–11].
The composition of this niche is influenced by environmental factors, genetic background and socio-economic factors [12–14]. Among these, socio-economic determinants are difficult to accurately assess in prospective studies. Antimicrobial agent consumption and conjugate vaccine use also constitute major selective pressures influencing the nasopharyngeal reservoir [12–15]. The seven-valent pneumococcal conjugate vaccine (PCV7) reduces vaccine pneumococcal serotypes (VT) from children’s nasopharynx, creating a vacant niche that may be filled in with non-vaccine types (NVT), which are usually more susceptible to antibiotics [15–17]. However, it is observed that, with time, antibiotic-resistant NVT are being selected by antibiotic use . In Belgium, PCV7 has been available (not free of charge) since October 2004 and provided free with a 2 + 1 schedule (two priming doses at 2 and 4 months of age followed by a booster dose at 12 months) since January 2007 for children under 2 years old. No data regarding nasopharyngeal flora evolution in Belgian children since PCV7 introduction are available yet.
In this study, we evaluated carriage rates and associations of four clinically relevant potential bacterial pathogens (S. pneumoniae, S. aureus, H. influenzae and M. catarrhalis) at the time of PCV7 introduction. Nasopharynx flora modifications were followed during the school year in two socio-economically distinct preschool populations (3–6 years old) in the Brussels area, Belgium. We also assessed associations between different epidemiological risk factors and both carriage and antibiotic resistance rates.
Patients and Methods
We carried out a prospective cohort study on healthy children (3–6 year-old) from 11 preschools of the Brussels area, Belgium (five schools in 2006–2007, six other schools in 2007–2008). In Belgium, parents are free to choose which school their child will attend. Based on the financial and educational level of the families attending a given school, a ranking of schools is made and a percentage (12.5%) of the schools showing the lowest socio-economic index is labeled as ‘positive discrimination’ (PD). Children attending a PD school live in more crowded conditions and less comfortable housing. Their parents received less formal education and show high unemployment rates and low incomes. Detailed information regarding our study was provided to parents and school staff. The Ethics Committee of the Hôpital Universitaire des Enfants Reine Fabiola approved the study. Signed informed consents were obtained from parents.
Epidemiological data collection
Clinical and demographic data were collected through standardized questionnaires. Questionnaires given at enrollment addressed gender, date of birth, address, parents’ smoking habits, number of siblings, parent’s educational level, number of rooms in the family home, day-care centre (DCC) attendance prior to preschool, PCV7 vaccine status, and history of chronic illness. Another questionnaire, given prior to each sampling, addressed antimicrobial agents use (oral/topical) within the preceding 15 days and 6 months. Medical record reviews and phone calls were performed to obtain missing information.
Nasopharyngeal aspirates were collected three times during the school years (autumn, winter and spring in 2006–2007 and 2007–2008). We chose to use aspirates rather than swabs. Samples were considered as nasal when the choanes were not passed. Samples were plated within 6 h after sampling.
Bacterial isolations and antimicrobial susceptibility testing
Samples were plated onto Columbia 5% sheep blood agar and chocolate agar plates for 24 h in 5% CO2 atmosphere at 35°C. The rest of the sample was then inoculated in Brain Heart Infusion broth supplemented with 7.5% NaCl. The enrichment broth was subcultured on a chromogenic medium for S. aureus detection (CHROMagar Microbiology, Paris, France). Bacterial species were identified by phenotypic methods. Antibiotic susceptibility of S. pneumoniae and oxacillin susceptibility of S. aureus were determined by the disk diffusion method (Rosco Diagnostica, Taasrup, Denmark) as recommended by the Clinical and Laboratory Standards Institute. Minimal inhibitory concentration (MIC) for penicillin, erythromycin and clindamycin was determined by E-test (AB Biodisk, Solna, Sweden). Penicillin susceptible, intermediate and resistant MIC breakpoints were ≤0.06, 0.12–1.0 and ≥2.0, respectively. Multidrug-resistant (MDR) strains were defined as resistant (including penicillin non-susceptible) to more than three antibiotic classes. S. aureus isolates showing an inhibition zone smaller than 20 mm surrounding the oxacillin disk were considered as methicillin-resistant Staphylococcus aureus (MRSA). H. influenzae and M. catarrhalis were screened for beta-lactamase production using a cefinase disk test assay (Becton Dickinson Microbiology Systems, Cockeysville, MD, USA).
S. pneumoniae were serotyped using the Quellung reaction (Statens Serum Institut, Copenhagen, Denmark). Factor sera were available for all serogroups; specific sera for serotype determination were available for vaccine types and some relevant serotypes (6A, 19A, 23A, 23B and 7F). Non-typeable pneumococci were confirmed by bile solubility and lytA PCR detection .
To evaluate the accuracy of children’s socio-economic level and PD label, we analysed the associations between mother’s and father’s educational levels, number of rooms in the home and attending PD schools, using Cramer’s V. Children had between one and three samples analysed for the presence of bacterial species. We identified risk factors associated with carriage of one bacterial species in one sample (point prevalence). The chi-squared test was applied to compare proportions. Analysed factors included in the multivariate analysis were: age, gender, passive smoking, PD schools attendance, day-care centre (DCC) prior to preschool attendance, family size and PCV7 status. Studies on S. pneumoniae acquisition and clearance show a mean duration of carriage ranging from 2 to 10 weeks . In our study, the inter-sampling period was approximately 12 weeks. Therefore, we considered the results of repeated samples as independent. Logistic regression models were fitted to identify risk factors associated with bacterial carriage. From these models, adjusted odds ratios (OR) and 95% confidence intervals were derived; corresponding p-values were those from Wald’s test (p <0.05 was considered as significant). Goodness of fit was checked using Hosmer and Lemeshow’s test. Due to missing data, antibiotic consumption is not included in the multivariate model. Univariate analysis was used to assess this variable. Statistical analyses were performed using Stata software version 10 (StataCorp LP, College station, Texas, USA).
Population and study participation
One thousand three hundred and forty-seven healthy children from 11 schools were invited to take part in the study. Twenty-one per cent of the invited PD children (205/963) enrolled, in comparison to 37% of the invited HL children (141/384). A total of 346 children participated in our study. The proportion of PD children (60%) was similar over the two study periods. PD children were distributed in 42 different classrooms (mean three to eight pupils per classroom) in two PD schools in 2006–2007 (71 children) and five in 2007–2008 (134 children). HL children were distributed in 18 different classrooms (mean four to nine pupils per classroom) belonging to three HL schools in 2006–2007 (47 children) and one in 2007–2008 (94 children). The difference between PD and HL schools in study participation rate was significant (p <0.0001). The adequacy of the PD school was confirmed by a significant correlation between parents’ educational levels and number of rooms in the home (Cramer’s V = 0.42, 0.33, 0.43, respectively). Among the 346 children enrolled, 13 could not be sampled and were excluded. The mean age of the 333 included children was 4.2 years and 50% were girls (Table S1). PD schools’ attendees were less likely to be immunized with PCV7 and more frequently cared for at home prior to preschool than HL schools’ attendees (Table S1).
Samples and carriage of nasopharyngeal pathogens
A total of 830 aspirates were performed (746 nasopharyngeal and 84 nasal). Sixty-one per cent (204/333) of the children were sampled three times, 27% (89/333) two times and 12% (40/333) only one time. Over the sequential samples, 83% of the children were colonized at least once with H. influenzae, 69% with S. pneumoniae, 67% with M. catarrhalis and 50% with S. aureus (data not shown). The distribution of S. pneumoniae, H. influenzae, S. aureus and M. catarrhalis within samples and their associations are presented in Table 1. On the one hand, H. influenzae was the most prevalent bacterial species (prevalence in the samples 60.9%), followed by S. pneumoniae and M. catarrhalis (42.5% and 41.9%, respectively) and S. aureus (35.5%) (Table 1a). On the other hand, isolation of S. pneumoniae with H. influenzae alone or together with M. catarrhalis, forming the classical otitis media (OMA) triad, was very common (10 and 11% of the samples, respectively) (Table 1a). Fifteen per cent of the samples did not contain any of the four studied species and were mostly nasal (p 0.001) (data not shown). Positive associations were significantly found for H. influenzae with the three other pathogens (S. pneumoniae, M. catarrhalis or S. aureus) (Table 1b).
|Bacterial species||Number of samples (%) (n = 830)||% of samples containing SP (n = 353)||% of samples containing HI (n = 505)||% of samples containing SA (n = 286)||% of samples containing MC (n = 348)|
|SP + HI + SA + MC||46 (6)||13||9||16||13|
|SP + HI + MC||88 (11)||25||17||–||25|
|SP + HI + SA||48 (6)||13||9||16||–|
|SP + SA + MC||10 (1)||3||–||3||3|
|HI + SA + MC||46 (6)||–||9||15||12|
|SP + HI||85 (10)||24||17||–||–|
|SP + SA||16 (2)||5||–||6||–|
|SP + MC||21 (3)||6||–||–||6|
|HI + SA||53 (6)||–||9||16||–|
|HI + MC||72 (8)||–||13||–||20|
|SA + MC||19 (2)||–||–||7||5|
|Bacterial species||Number of samples (%) (n = 830)||% of samples with SP (n = 353)||% of samples with HI (n = 505)||% of samples with SA (n = 286)||% of samples with MC (n = 348)||OR (95% CI)||p-value|
|SP + SA||120 (14)||33||–||42||–||0.9 (0.7–1.2)||0.6|
|SP + HI||267 (32)||74||53||–||–||2.8 (2.0–3.8)||<0.001|
|SP + MC||164 (20)||45||–||–||47||1.2 (0.9–1.6)||0.3|
|HI + SA||193 (23)||–||38||67||–||1.8 (1.3–2.5)||0.001|
|HI + MC||252 (30)||–||50||–||72||2.3 (1.7–3.2)||<0.001|
|SA + MC||121 (14)||–||–||42||35||1.0 (0.7–1.3)||0.9|
Risk factors for carriage
Young age (under 4 years old) was a risk factor for M. catarrhalis carriage while S. aureus was more likely to be isolated in older children (Table 2). PD school attendees were more likely to carry M. catarrhalis and S. aureus than HL school attendees. No differences in S. pneumoniae and H. influenzae carriage between the two populations were observed. Presence of other children in the house was negatively associated with M. catarrhalis carriage and positively with S. pneumoniae carriage. Smokers in the home were not a major risk factor as only M. catarrhalis was more frequently isolated from children with smoking parents. By contrast, attending a DCC prior to preschool decreased the risk of carrying M. catarrhalis. PCV7 immunization had no effect on the carriage rate of these four main pathogens.
|Factors||S. pneumoniae||S. aureus||H. influenzae||M. catarrhalis|
|OR (95% CI)||p-value||OR (95% CI)||p-value||OR (95% CI)||p-value||OR (95% CI)||p-value|
|Age <4 yearsold||1.2 (0.9–1.7)||0.19||0.51 (0.4–0.7)||<0.001||1.4 (0.9–1.9)||0.05||1.7 (1.2–2.3)||0.001|
|Male gender||1.2 (0.9–1.6)||0.24||1.1 (0.8–1.6)||0.4||1.3 (0.9–1.8)||0.1||1.0 (0.8–1.4)||0.7|
|Smoking parents||0.8 (0.6–1.3)||0.37||1.0 (0.7–1.4)||0.9||0.8 (0.6–1.3)||0.4||1.4 (1.1–2.0)||0.03|
|Children cared for at home before preschool||1.2 (0.8–1.6)||0.35||0.8 (0.6–1.1)||0.19||1.0 (0.7–1.4)||0.9||1.6 (1.1–2.2)||0.009|
|Immunized with PCV7||0.9 (0.6–1.4)||0.79||0.8 (0.5–1.2)||0.28||1.2 (0.8–1.8)||0.4||1.2 (0.8–1.7)||0.4|
|Other children at home||1.4 (1.0–2.0)||0.08||1.2 (0.8–1.7)||0.35||1.0 (0.7–1.5)||0.8||0.6 (0.4–0.7)||0.002|
|PD school attendance||0.8 (0.6–1.1)||0.21||1.5 (1.0–2.1)||0.02||0.9 (0.6–1.3)||0.5||1.9 (1.4–2.7)||0.002|
Thirty-one serotypes were identified among 362 S. pneumoniae isolates (nine samples showed two different serotypes) (Table 3). The five most prevalent serogroups/types were 6B, 19F, 23F, 6A and non-typeable pneumococci (NTP). VT isolates were significantly more commonly recovered from children attending PD than HL schools (35% vs. 23%, p <0.05). The proportion of VT also decreased from 34% to 26% between 2006–2007 and 2007–2008 (p 0.08); NVT isolates increased from 35% to 45% (p 0.06) (data not shown). Distribution of NVT and NTP was uneven between the school types, with serogroups 15, 38 and NTP predominating in PD schools and serogroups 23A, 35, 1 and 34 in HL schools (Table 3). These serogroups clustered neither in time nor in space (data not shown).
|Serogroups/serotypes||Number of isolates||Number of isolates in PD schools||Number of isolates in HL schools||p-value|
|Total VT (%)||106 (29)||68 (35)||38 (23)||<0.05|
|Total NVT (%)||256 (71)||125 (65)||131 (77)||0.47|
|TOTAL||362a||193 (53)||169 (47)|
Antimicrobial susceptibility patterns and risk factors associated with resistant strains
Pneumococcal resistance rates tended to be higher in PD than in HL schools (Table 4), although attending a PD school was the only risk factor identified for carrying penicillin-reduced susceptibility pneumococci (PRSP; OR, 5.5; 95% CI, 2.2–13.6; p <0.001) (data not shown). Eleven per cent of the S. pneumoniae isolates (n = 362) were PRSP (maximum MIC of 1.5 mg/L), 5% of which were NTP (Table 4). MDR S. pneumoniae, representing 7% of the isolates, was also more frequent in PD schools (OR, 2.3; 95% CI, 0.9–5.8; p 0.05) (data not shown). Resistance was associated with VT (19F, 6B and 14), 19A and NTP serogroups. PRSP could not be correlated with antibiotic usage, while erythromycin resistance, present in 24% of S. pneumoniae isolates, was associated with previous antibiotic consumption (p 0.04) (data not shown).
|Antibiotic||Number of resistant isolates in PD schools (%)|
Total = 193
|Number of resistant isolates in HL schools (%)|
Total = 169
|Number of resistant isolates in both schools (%)|
Total = 362
serogroups (number of isolates)
|Penicillina||33 (17)||6 (3)||39 (11)||NTP (17), 19F (9), 19A (4), 6B (2), 14 (2)|
|Erythromycin||48 (25)||38 (22)||86 (24)||NTP (18), 6B (16), 19F (15), 19A (13), 23F (5)|
|Clindamycin||42 (22)||28 (16)||70 (19)||NTP (15), 19F (14), 19A (13), 14 (4), 9V (2)|
|STX-TMP||67 (35)||40 (24)||107 (30)||NTP (17), 6B (16), 19A (13), 23A (10), 23F (9), 19F (10)|
|Tetracyclin||30 (16)||14 (8)||44 (12)||6B (11), 19A (10), 19F (8), NTP (6), 23F (4)|
|MDR||18 (5)||7 (2)||25 (7)||NTP (6), 6B (6), 19A (5), 19F (4), 23F (2)|
Fourteen MRSA strains were isolated from 11 (3.3%) children and were preferentially recovered in HL schools (p 0.02). MRSA carriage was not associated with antibiotic consumption prior to sampling or history of chronic illness. Twenty-six per cent of H. influenzae and 99% of M. catarrhalis were beta-lactamase producers.
Better understanding of upper respiratory tract carriage has become an increasing priority since the nasopharyngeal niche has been recognized as a reservoir for major pathogens and antibiotic resistance . A high carriage rate of S. pneumoniae, H. influenzae, S. aureus and M. catarrhalis was found, even though our study included children older than the populations usually studied. Previous European studies focusing on similar age groups found lower carriage rates [14,22]. Our data might be explained by the use of aspirates rather than swabs as sampling technique as well as enrichment broth for S. aureus recovery. They could also be related to crowding persisting in preschools settings because overcrowding, such as that found in DCCs, is a well-known factor associated with increased nasopharyngeal bacterial colonization [7,23]. Age remained an important factor associated with carriage. Young children were more likely to be colonized with one of the otitis media triad; however, this was significant only for M. catarrhalis. By contrast, S. aureus carriage was more prevalent in older children. Although important in both socio-economic settings, M. catarrhalis and S. aureus carriage rates were higher in HL schools. A possible explanation might be the low DCC attendance prior to preschool for children from PD schools (31% vs. 62%). In contrary, S. pneumoniae and H. influenzae were surprisingly equally recovered from children from either type of school (Fig. 1), despite a marked difference of PCV7 uptake between the two populations (28% in HL schools vs. 11% in PD schools), which might result from the high cost of the vaccine. This difference most likely accounts for differences in serotype distribution. VT (mostly 6B) were more commonly recovered from PD than from HL school attendees (35% vs. 23%) and NVT were more prevalent in HL schools (70% vs. 42%). Surprisingly, serotype 1 (NVT) strains were isolated in nine children attending the same school and accounted for 6% of the pneumococci isolated in that particular school. Interestingly, this HL school had the highest PCV7 uptake. Serotype 1 is responsible for invasive diseases in children and is seldom found in asymptomatic carriers . These strains were isolated only during the 2007–2008 sampling campaign, and in the meantime an increase in serotype 1-associated invasive diseases (mainly complicated pneumonia) was observed in Belgium (G. Hanquet, personal communication). Thus, caution must be observed when considering this NVT as a replacement serotype. On the other hand, serotype 19A, which is recognized as the predominant replacement serotype in children , was equally recovered from vaccinated and non-vaccinated children. No increase of 19A prevalence over the two studied years was observed, despite an increase in vaccine uptake (from 8.5% to 24.1%). However, taking into account that the overall vaccination rate in Belgium was low at the time (36% of children aged 18–23 months, http://www.observatbru.be/documents/news-items/news-enquete-de-couverture-vaccinale-2006.xml) and that the vaccination scheme was incomplete for most of the children in our study, it is likely that, as observed in other countries where universal pneumococcal vaccination is adopted , serotype 19A might fill the nasopharyngeal niche. This is of major concern because this serotype shows both high invasive potential and high antibiotic resistance rate . In our study, serotype 19A and NTP showed multiple resistances, together with VT isolates as observed previously [19,25,26], NTP accounting for 44% of PRSP isolates. As polysaccharide-based vaccines are unable to directly have an impact on these strains and as NTP show a high frequency of transformation, they might constitute an emerging reservoir for horizontal transfer of resistance genes among streptococci. When restricted to typeable isolates, PRSP prevalence (7%; 95% CI, 4–9%) is higher than in pneumococcal invasive disease, in the same age group and for the same period in Belgium (4.5%) (G. Hanquet & J. Verhaegen, personal communication). However, it is lower than in the previous Belgian carriage study conducted in 2001 in children under 3 years attending a DCC (14%) . This might be due to a slight reduction in penicillin consumption in the community during the same period . Our study also reports MRSA circulating in a striking proportion (3.6%). MRSA carriage was preferentially isolated in higher socio-economic groups but was not associated with antibiotic consumption 2 weeks prior to sampling, or history of chronic illness. Unfortunately, these high resistance frequencies could not be correlated either with household contacts with healthcare workers or with frequently hospitalized guardians due to a lack of information. Nevertheless, these resistant pathogens are circulating in healthy children without any obvious risk factors and might represent a threat to the community.
Co-colonization relationships between S. pneumoniae and the three other main pathogens were analysed. About one-third of our population (30%, 100/333) carried at least once S. pneumoniae together with S. aureus, more than half (58%, 193/333) carried both S. pneumoniae and H. influenzae and 36% (119/333) carried the OMA triad (S. pneumoniae, H. influenzae and M. catarrhalis).
Scarce epidemiological data regarding co-colonization are available in the literature, although studies have shown an inverse relationship between S. pneumoniae and S. aureus carriage in children, especially when considering VT isolates [29,30]. Our study did not show negative associations between these two species, even when restricted to VT isolates (29% of our isolates). Lack of association was also observed in a fully vaccinated population and attributed to a low VT prevalence (1/3 of the isolates) . A similar observation in our study could be explained by a combination of this parameter together with the children’s age because resistance to pneumococcal colonization increases as children get older. The implication of the host immunological response has been emphasized by studies showing negative associations between S. pneumoniae and S. aureus in immunocompetent but not in HIV-infected children [32,33]. Our study revealed positive associations between S. pneumoniae and H. influenzae as well as between H. influenzae and M. catarrhalis and between H. influenza and S. aureus. It is likely that multiple interactions between bacterial species, host immune system, other nasopharynx colonizers such as viruses and other bacterial species that are usually not studied occur in children nasopharynx. These intricate relationships are difficult to assess and interpret, and may contribute to a unique dynamic environment that will be difficult to generalize to other epidemiological settings.
To conclude, the present prospective community-based survey demonstrates a high carriage rate of the main respiratory pathogens and S. aureus in children from 3 to 6 years old. Our study also proves that the socio-economic environment exerts influences on bacterial flora composition and on antibiotic resistance profiles.
We thank the children, their families, school directors and staff who collaborated in this project. The authors are grateful to Andrea Lossius Badosa, Sophie Blumental, Pierre-Alexandre Drèze, Oliver Luycks, Claire Nonhoff, Sandrine Roisin, Ariane Deplano and all the staff of the Microbiological Department of the Hôpital Erasme for their collaboration. We also acknowledge Jozef Vandeven and Jan Verhaegen for assistance in pneumococcal serotyping and Germaine Hanquet, Georges Casimir, Marc Struelens and Philippe Lepage for critical reading of the manuscript and support in the accomplishment of this work. Sarah Jourdain was supported by a grant from the Belgian Kids Fund. Pierre R. Smeesters is a ‘chargé de recherche’ FNRS and has been supported by an ESPID Fellowship Award. This work was supported by the paediatric department of the HUDERF, the ‘Iris Recherche’ fund the Belgian ‘FNRS’ and an ESCMID research grant to Anne Vergison.
Conflicts of interest: Anne Vergison has received funds for speaking and attending advisory boads and steering committees from GSK, Pfizer (Wyeth) and SPMSD. Pierre R. Smeesters has received research funding from Pfizer. Sarah Jourdain, Olivier Denis, Michèle Dramaix, Valérie Sputael, Xavier Malaviolle, Laurence Van Melderen: nothing to declare.