Monthly nasopharyngeal swab (NPS) specimens were collected from 955 infants between one and 24 months of age as part of a longitudinal cohort study of pneumonia epidemiology in Maela camp for displaced persons on the Thailand-Myanmar border [7]. Two hundred and thirty four infants and their mothers were randomly selected to be also included in a detailed pneumococcal carriage study (the “immunology” follow-up group) described elsewhere [8]. The remaining 721 infants formed the “routine” follow-up group (Figure 1). Pneumococcal vaccines were not available in the study population.

The WHO protocol for specimen collection was followed [5]. Briefly, Dacron-tipped flexible swabs (Medical Wire & Equipment, Corsham, UK) were used to sample the nasopharynx. Swab tips were immediately removed into sterile cryovials (Simport, Beloeil QC, Canada) containing 1mL STGG medium (skim milk-tryptone-glucose-glycerol medium; prepared in house) using 70% ethanol-cleaned scissors. These NPS-STGG specimens were transported to the laboratory in a cool box where they were briefly vortexed and subsequently frozen at -80°C within eight hours of collection.

On completion of the study follow-up, the specimen database was reviewed and all NPS from infants in the “routine” follow-up group, in whom a complete set of 24 monthly specimens had been obtained, were selected for assessment by the latex sweep method. A proportion of these NPS specimens had previously been cultured according to the WHO method (Figure 1). Laboratory technicians performing the latex sweep serotyping were blinded to the WHO culture results.

To compare pneumococcal acquisition rates and carriage durations by culture/serotyping method, a subset of infants were studied further. Infants from the “immunology” follow-up group (NPS previously processed by the WHO method and described in detail in [8]) were matched 1∶1 by gender and month of birth with a “routine” follow-up group infant (NPS prospectively processed by the latex sweep method) using the “vmatch” command in Stata/IC 12.1 (StataCorp, College Station TX, USA).

For the latex sweep method, 10 µL fully thawed NPS-STGG was streaked onto a 5% sheep blood-CNA agar plate (bioMerieux, Marcy L’Etoile, France) and incubated overnight at 36°C in 5% CO₂. All growth from the agar plate were picked using a sterile cotton swab (Medical Wire & Equipment, Corsham, UK) and suspended in sterile 0.85% saline to a density of at least 0.5 McFarland. This suspension was used immediately for serotyping using latex reagents prepared in-house [9], [10]. The suspension was first tested with antisera pools A-I and subsequently with all appropriate group, type, and factor antisera.

For the WHO method, 10 µL of fully thawed NPS-STGG was cultured as described above. All morphologically distinct alpha-haemolytic colonies were sub-cultured onto 5% sheep blood agar (Clinical Diagnostics, Bangkok, Thailand) S. pneumoniae was identified by colony morphology, Gram’s stain, and optochin disc susceptibility (Oxoid, Basingstoke, UK) [5]. Isolates with intermediate susceptibility to optochin, defined as a inhibition zone diameter of 7–13 mm, were subjected to tube bile solubility testing [11]. Isolates were serotyped by latex agglutination, using the same in-house reagents as used for the latex sweep method, with Quellung confirmation of equivocal results [9], [10], [12].

For both methods, colonisation density was assessed by estimating the number of pneumococcal colonies in each of the four agar plate quadrants: 1+ (growth in quadrant 1 (inoculation area/primary streak) but <10 colonies in quadrant 2), 2+ (>10 colonies in quadrant 2 but <10 colonies in quadrant 3), 3+ (>10 colonies in quadrant 3 but <10 colonies in quadrant 4), or 4+ (>10 colonies in quadrant 4).

The definitions of serotype acquisition, clearance, and duration employed were those used in the study of pneumococcal colonisation mother-infant pairs [8]. Briefly, acquisition was defined as the first appearance of a pneumococcal serotype in an infant’s NPS culture or when a serotype was re-cultured following clearance. Clearance was defined as the absence of a serotype from two consecutive NPS specimens. Serotype carriage episode duration was calculated as the interval between acquisition (mid-point between last negative and first positive swab) and clearance (mid-point between last positive and first of two consecutive negative swabs).

Data analyses were performed using Stata/IC 12.1. Comparisons of serotype detections by WHO culture or latex sweep serotyping were made by McNemar’s test. Continuous numerical variables were summarised by means and groups were compared by t-test. Serotype acquisition rates and carriage durations were calculated by survival analysis; groups were compared using the log-rank test.

Serotypes 15B and 15C were considered to be a single serotype (15B/C) given the lability of their capsular polysaccharides [13]. Serotypes 6A and 6C were also combined (serotype 6A/C), since, at the time of the study, antisera-based serotyping could not distinguish between the two. Pneumococcal serotype 6D has also been described recently: this serotype may be identified as 6B using currently available antisera [14], [15]. There are no data regarding prevalence of this serotype in the study population, although given the absence of PCV use, carriage this serotype was unlikely to be common and thus antisera 6B isolates have been reported as such [16], [17]. Although commonly carried in the nasopharynx, non-typeable pneumococci were excluded from these comparisons, since the latex sweep method has no confirmatory test of pneumococcal identity except the presence of capsule.

Written informed consent was obtained from the mothers of all study infants prior to study enrolment. Ethical approval for this study was granted by the ethics committees of the Faculty of Tropical Medicine, Mahidol University, Thailand (MUTM-2009-306) and Oxford University, UK (OXTREC-031-06).

Three hundred and sixty four infants from the “routine” follow-up group had a complete 24 month NPS specimen set. A total of 8,736 swabs were cultured and latex sweep serotyping done (Figure 1). Typeable pneumococci were identified in 65.3% (5,699/8,736) of specimens. Multiple pneumococcal serotypes were found in 734 NPS specimens (12.9% of specimens from which pneumococci were cultured). In swabs with more than one serotype, the median number of serotypes detected was two and the maximum was four.

A total of 62 serotypes were identified and eight of these (19F, 23F, 6B, 6A/C, 14, 15B/C, 35C, 19A) accounted for two-thirds of all pneumococci detected (Figure 2). Almost two thirds (61.8%; 3,999/6,471) of the pneumococci detected were PCV13 serotypes and 42.4% (3,704/8,736) NPS contained at least one PCV13 serotype.

The ten most prevalent serotypes in these infants were identical to those in the 234 infants from the “immunology” follow-up group (excluding NT pneumococci), although the rank order varied [8].

One thousand one hundred and seven (12.7%) of the NPS specimens from these 364 infants had previously been cultured using the WHO protocol. At least one typeable pneumococcus was identified in 68.6% of NPS specimens by WHO culture and 64.0% by sweep method (p<0.001). 9.3% of specimens contained multiple serotypes by sweep method compared with 3.1% by WHO method (p<0.001). However, there were no differences in the proportion of specimens containing at least one PCV13 serotype or non-vaccine serotype (NVT) pneumococcus by method (Table 1).

In swabs containing at least one typeable pneumococcus by the WHO method, an identical serotype was identified in 85.0% (645/759) by the sweep method. Relaxed to agreement at the serogroup/type level, this proportion increased only slightly to 86.4% (656/759). Where both methods identified at least a single typeable pneumococcus, agreement was 91.2% (645/707; serotype level) and 92.8% (565/707; serogroup/type level), perhaps indicating that low density pneumococcal colonisation was missed by the latex sweep method. Serotype agreement between methods was lowest in specimens when there was light colonisation (defined as 1+) on the WHO culture plate (Table 2). Although not statistically significant at the serotype level, the likelihood of agreement at serogroup/type level was significantly higher if colonisation density was 2+ or greater (OR 1.65, 95% CI 1.06–2.57, p = 0.03).

One hundred “immunology” follow-up group infants whose NPS had been cultured by the WHO method were matched 1∶1 with “routine” follow-up group infants whose swabs were processed by the latex sweep method. All infants contributed 24 monthly swabs and all were colonised by pneumococci at least once.

In both groups, the median age at first pneumococcal acquisition was 46 days (p = 0.6; Figure 3). Considering individual serotype acquisition rates, there were no significant differences in rates of acquisition of commonly carried PCV13 serotypes. However, the non-vaccine serotype acquisition rate was higher in infants whose swabs were processed by the latex sweep method (Table 3). Durations of first carriage episodes, without stratification by serotype, were similar in both groups (“immunology”/WHO: median 90 days (IQR 31–151); “routine”/latex sweep: 62 days (IQR 31–121); p = 0.2; Figure 4).

Infants with swabs processed using the latex sweep method had a greater number of pneumococcal serotype carriage episodes than those whose swabs were cultured using the WHO method (mean 8.4 vs. 7.5, p = 0.03). There were no differences in the number of PCV13 or NVT serotype carriage episodes between the groups (PCV13 serotypes accounted for 51.4% (“immunology”/WHO) and 54.0% (“routine”/sweep) carriage episodes, p = 0.3).