To evaluate pneumococcal carriage in Italian children, a cross-sectional study was conducted between November 2011 and April 2012 in children aged 0-5 years in two Italian cities, Bologna (Emilia-Romagna region) and Milan (Lombardy region), located in regions with different pneumococcal vaccination policies. The Emilia-Romagna region actively provided pneumococcal immunization to all infants since 2006; while, the Lombardy region initially offered vaccination only to disease-specific risk groups of children. Since 2009, Lombardy has offered vaccination to all infants upon the parents’ or pediatricians’ request [13]. As expected, different vaccine coverage rates were observed in the two regions in 2008, with 95% PCV7 coverage in Emilia-Romagna and 27% in Lombardy [14]. Vaccination coverage in Lombardy increased substantially starting in 2009 and reached almost 80% at the time of this study (A. Piatti, personal communication). Children attending routine visits in the first years of life were recruited for the study from the pediatric outpatient clinics of the S. Orsola-Malpighi University Hospital in Bologna and a pediatric practice associated with Hospital Maggiore Policlinico in Milan. Children suffering from respiratory tract infections, other respiratory problems, or chronic illnesses were excluded from the study.

A structured questionnaire was filled in by the pediatrician and was used to collect the children’s demographic data (age, gender, number of family members, presence of young siblings under 6 years, daycare attendance, and presence of at least one smoking parent) and their PCV7/PCV13 vaccination status. Children who had received all the recommended doses for their age, according to the Italian schedule (2+1 doses at 3, 5, and 11/13 months), or who had received a catch-up dose at >1 year of age, were considered completely vaccinated. Children who had not completed the recommended vaccination schedule were classified as partially vaccinated.

The study was approved by the Ethics Committee of S. Orsola-Malpighi University Hospital in Bologna (reference number: 1946/2011) and the Ethics Committee of Hospital Maggiore Policlinico in Milan, according to the Italian legislation. Parents or guardians of the participating children provided written informed consent. The procedures followed were in accordance with the European Statements for Good Clinical Practice and the Declaration of Helsinki of the World Medical Association.

Nasopharyngeal specimens were obtained using a nylon flocked swab (eSwab, Copan, Brescia, Italy), according to the WHO pneumococcal colonization detection protocol [20]. Swabs were streaked onto 5% sheep blood Columbia agar plates supplemented with 5 µg/mL gentamicin within 18 hours of collection and were incubated overnight at 35°C in 5% CO₂-enriched air. Swabs were then stored at -80°C in milk-tryptone-glucose-glycerol (STGG) medium (www.cdc.gov/ncidod/biotech/files/pcr-pneumo-carriage-March2010.pdf).

Pneumococcal isolates were detected by colony morphology and identification was confirmed by optochin susceptibility and bile solubility testing. Individual isolates were serotyped by latex agglutination and the Quellung reaction using commercially available antisera (Statens Serum Institut, Copenhagen, Denmark).

Susceptibilities to penicillin, ceftriaxone, erythromycin, clindamycin, tetracycline, and chloramphenicol were determined by the antimicrobial gradient strip diffusion method (Etest, bioMérieux, Durham, USA). The breakpoints proposed by the EUCAST (http://www.eucast.org/clinical_breakpoints/. Accessed 2013 May 31) were used. Briefly, for penicillin the meningitis criteria were: susceptible, MIC ≤ 0.06 mg/L; resistant, MIC > 0.06 mg/L. The non-meningitis criteria were: susceptible, MIC ≤ 0.06 mg/L; resistant, MIC > 2 mg/L. For the purpose of this study, all S. pneumoniae isolates with a penicillin MIC > 0.06 mg/L were defined as non-susceptible. For ceftriaxone, the breakpoints used were: susceptible, MIC ≤ 0.5 mg/L; resistant, MIC > 2 mg/L. Isolates that were not susceptible to antimicrobials were further analyzed by Multi Locus Sequence Typing (MLST) following the procedures described at the MLST website (http://spneumoniae.mlst.net. Accessed 2013 Jun 20). For each serotype, complete allelic profiles were determined for representative isolates from each of the antibiotic susceptibility groups. Partial MLST profiles were determined for the remaining isolates using a previously described strategy [21]. If the xpt and ddl alleles corresponded to those obtained for the representative isolate, the same Sequence Type (ST) was assigned. If the xpt and ddl alleles were divergent, a complete allelic profile was obtained. The software eBURST, available from the MLST website, was used to compare the STs obtained with those available at the MLST website and to define clonal complexes (CCs). Pneumococcal Molecular Epidemiology Network clones (PMEN website: http://www.sph.emory.edu/PMEN/. Accessed 2013 Jun 20) were also identified.

Data were checked for outliers, duplicate records, and distribution of the variables. Age, computed by date of examination minus date of birth, was analyzed as a categorical (0-5; 6-23; 24-35; and 36-71 months) variable. For each variable, missing values were tabulated by age, sex, categories of outcome (colonization status), exposure (vaccination status), and recruitment center. We examined the association between colonization status and vaccination exposure (unvaccinated was the reference category). Potential risk factors for nasopharyngeal carriage of S. pneumoniae were identified by univariate analysis. Odds ratios (ORs) and 95% confidence intervals (CIs) for colonization were computed using logistic random effect regression models, with the recruitment center as the random effect. Multivariate analysis was adjusted for other covariates like age, day-care centers attendance, number of siblings younger than 6 years, and smoking status of parents. The effect of potential confounding factors was estimated by fitting a model with and without the variable of interest and by comparing the adjusted and crude estimates. Records with missing values for a variable were excluded for each analysis involving that variable. Statistical analysis was carried out in STATA 10 software (StataCorp).

A total of 571 children, 270 from Bologna and 301 from Milan, aged 0-5 years, were enrolled in the study. The mean age was 35 months (SD 17.15) and 316 (55.3%) participants were male. Overall, 78.9% (451/571) of children attended day care centers, and this percentage was higher (97%) in the 24-71 month old group. Only 30.8% (176/571) of children had siblings aged <6 years and 37.1% (212/571) had at least one smoking parent. The only difference observed between the children populations of the two cities with regard to demographic characteristics was a higher presence of young siblings in Milan, which was borderline significant (p= 0.051).

The vaccination status was known for 531 children: 81.2% (431/531) had received at least one dose of PCV7 or PCV13 and 74.9% (398/531) had completed the recommended vaccination schedule for their age. Specifically, 57.3% (228/398) received PCV7, 27.1% (108/398) PCV13, and 15.6% (62/398) a combination of the two vaccines. According to the Italian schedule [16], children vaccinated with a combination of vaccines receive 2 or 3 doses of PCV7 followed by 1 dose of PCV13 (48/62 children, 77.4%) or 1 or 2 doses of PCV7 followed by 2 doses of PCV13 (14/62 children, 22.6%). By age-adjusted analysis, a statistically significant difference was found in the proportion of vaccinated children between Bologna and Milan. In Bologna, 98.8% of children had received at least one vaccine dose and 97.8% were completely vaccinated compared with 68.1% and 57.5%, respectively, in Milan (p < 0.001).

The overall pneumococcal carriage rate was 32.9% (188/571) and there was no significant difference between Bologna and Milan (34.4% and 31.5%, respectively, p= 0.5). The colonization rate increased from 7.1% in the younger age group (0-5 months) to 33.3% (6-23 months), 33.0% (24-35 months), and 35.1% (36-71 months) in the older aged groups (Table 1). Among colonized children, 71.8% (135/188) were completely vaccinated with PCV7 or PCV13 or a combination of the two, while 5.3% (10/188) were partially vaccinated. Among the non-colonized children, 68.6% (263/383) were completely vaccinated and 6.0% (23/383) were partially vaccinated.

In univariate analysis, colonization was associated with age, day-care center attendance, presence of young siblings in the family, and prior vaccination with PCV7 (Table 1). Unexpectedly, the presence of at least one smoking parent was significantly associated with a lower carriage rate, both in univariate and multivariate analyses (Table 1). Further analysis was performed only on children from Milan, for whom information about which parent was the smoker in the household (if mother, father, or both) was available, and indicated that a smoking mother was associated with a higher risk of pneumococcal carriage in children (OR: 3.66; 95%CI: 1.53-8.73, p= 0.003). The multivariate logistic regression model showed children 6-35 months of age were the most prone to pneumococcal colonization (6-23 months, OR: 3.75, 95%CI: 2.19-6.43, p<0.001; 24-35 months, OR: 3.15, 95%CI: 2.36-4.22, p<0.001) (Table 1).

Overall, 189 pneumococcal isolates were obtained from 188 children, as 2 different isolates were obtained in one child. Five isolates were not viable upon storage and could not be serotyped. Of the 184 isolates, 31 different serotypes were identified, with 6C (19/184, 10.3%), 24F (16/184, 8.6%), and 19A (15/184, 8.1%) being the most common; in addition, 5.4% of the isolates (10/184) were non-typeable (NT) (Figure 1). Overall, 5.4% of the isolates were PCV7 serotypes, 18.0% were PCV13 serotypes, and 82.0% non-PCV13 serotypes (Figure 1). No significant difference was observed in the percentage of non-PCV13 serotypes between the children in Bologna and Milan (p= 0.856). By age-adjusted multivariate analysis, children who had received PCV7 were protected from the included serotypes; while, children who had received PCV13 or a combination of PCV7 and PCV13 were protected from the serotypes included in both vaccines. Vaccination with either PCV7 or PCV13 increased the risk of colonization with non-PCV13 serotypes (Table 2).

Within each serotype, the orange portion indicates isolates from Bologna and the blue portion isolates from Milan. The arrowheads indicate PCV13 serotypes.

The ranking order of serotypes was slightly shifted between the two populations (Table 3). In particular, the most prevalent serotypes in Bologna were 6C, 24F, and 19A followed by 10A; while, in Milan, serotypes 23B and 6C were the most prevalent, followed by 11A, 15B, 24F, and 19A. In addition, some serotypes were only recovered in one city, with 7F, 8, 10A, 12F, and 20 only in Bologna and 1, 6B, 18C, 29, and 34 only in Milan (Table 3, Figure 1). With the exception of 19A, which was the third most prevalent serotype, other PCV13 serotypes (1, 3, 6A, 6B, 7F, 14, 18C, and 19F) were poorly represented (0.5% to 3.2%, Table 3).

Using the EUCAST meningitis breakpoints for penicillin (MIC > 0.06 mg/L), 57 isolates out of 184 (30.9%) were penicillin non-susceptible. Of these, 11 (6.0%) showed high penicillin resistance (MIC> 1 mg/L). For the non-meningitis breakpoints, 30.4% (56/184) were intermediate to penicillin (MIC 0.12-2 mg/L), and only 1 isolate (MIC=3 mg/L) was fully resistant (MIC > 2 mg/L). Nine isolates out of 184 (4.8%) were non-susceptible to ceftriaxone (MIC >0.5 mg/L), 8 were intermediate, and only 1 isolate was fully resistant (MIC=3 mg/L). The rates of resistance to the other antibiotics were: erythromycin, 42.3% (78/184); clindamycin, 36.4% (67/184); tetracycline, 36.9% (68/184); and chloramphenicol, 1.6% (3/184). In addition, 23.9% (44/184) were non-susceptible to both penicillin and erythromycin. Non-PCV13 serotypes accounted for 75.4% and 70.8% of the penicillin and erythromycin non-susceptible isolates, respectively. The most prevalent serotypes (6C, 24F, and 19A) included mainly penicillin and/or erythromycin non-susceptible isolates, which accounted for 89.4% of 6C, 75.0% of 24F, and 80.0% of 19A isolates. A high rate of antibiotic non-susceptibility was also present in serotypes 23B, 15A, 19F, 3, 14, 33F, and in NT isolates (Figure 2).

Within each serotype, pneumococcal isolates were divided into 4 categories: susceptible to penicillin and erythromycin (green, PEN-S + ERY-S), non-susceptible to penicillin only (red, PEN-R), non-susceptible to erythromycin only (yellow, ERY-R), and non-susceptible to penicillin and erythromycin (striped, PEN-R + ERY-R).

The 93 isolates that were non-susceptible to penicillin and/or erythromycin were genotyped. A complete MLST profile was obtained for 54 isolates, while a partial MLST profile was used for 39 isolates to infer the ST [21]. In total, 47 STs were detected, clustered in 20 CCs and 2 singleton groups (Table 4). The most abundant CC was CC156, which included 7 STs clustered in different genetic lineages [22] and multiple serotypes (6B, 6C, 14, 19F, 23B, and 24F). The second most abundant CC was CC315, which included PMEN clone Poland^6B-20/ST315 and comprised only serotype 6C isolates, followed by CC230, which included Denmark¹⁴-32/ST230 and was composed of serotypes 19A and 24F. The non-susceptible NT isolates belonged to CC344, which included Norway^NT-42/ST344. Both the fully penicillin resistant isolate (MIC= 3 mg/L) and the ceftriaxone-resistant isolate (MIC= 3 mg/L) were serotype 19A and belonged to ST320/CC320, related to Taiwan^19F-14/ST236 (Table 4). The other isolates showing high penicillin resistance (MIC> 1 mg/L) belonged mainly to CC230 and CC156. Among the most common serotypes, 6C (5 STs, grouped in 4 different CCs and 1 Singleton group) and 19A (7 STs grouped in 3 different CCs) showed high genetic diversity.