Conceived and designed the experiments: PT CT NPD NJW FN DG. Performed the experiments: PT CT AJ NH. Analyzed the data: PT SJL. Wrote the paper: PT.
The authors have read the journal's policy and have the following conflicts: Co-author Nicholas White is a PLoS ONE Editorial Board member. This does not alter the authors' adherence to all the PLoS ONE policies on sharing data and materials.
Pneumococcal disease is a major cause of childhood death. Almost a third of the world's children live in Southeast Asia, but there are few data from the region on pneumococcal colonization or disease. Our aim was to document the dynamics of pneumococcal carriage in a rural SE Asian birth cohort.
We studied 234 Karen mother-infant pairs in Northwestern Thailand. Infants were followed from birth and nasopharyngeal swabs were taken from mother and infant at monthly intervals until 24 months old.
8,386 swabs were cultured and 4,396 pneumococci characterized. Infants became colonized early (median 45.5 days; 95% confidence interval [CI] 44.5-46.0) and by 24 months had a median of seven (range 0–15) carriage episodes. Maternal smoking and young children in the house were associated with earlier colonization (hazard ratio [HR] 1.5 (95% CI 1.1–2.1) and 1.4 (95% CI 1.0–1.9)). For the four commonest serotypes and non-typeable pneumococci, previous exposure to homologous or heterologous serotypes resulted in an extended interval to reacquisition of the same serotype. Previous colonization by serotypes 14 and 19F was also associated with reduced carriage duration if subsequently reacquired (HR [first reacquisition] 4.1 (95% CI 1.4–12.6) and 2.6 (1.5–4.7)). Mothers acquired pneumococci less frequently, and carried them for shorter periods, than infants (acquisition rate 0.5 vs. 1.1 /100 person-days, p<0.001; median duration 31.0 vs. 60.5 days, p = 0.001). 55.8% of pneumococci from infants were vaccine serotypes (13-valent pneumococcal conjugate vaccine, PCV13), compared with 27.5% from mothers (p<0.001). Non-typeable pneumococcal carriage was common, being carried at least once by 55.1% of infants and 32.0% of mothers.
Pneumococcal carriage frequency and duration are influenced by previous exposure to both homologous and heterologous serotypes. These data will inform vaccination strategies in this population.
To explore colonization dynamics and modifiers, we designed an intense pneumococcal carriage study of mother-infant pairs from birth for 24 months. We nested the study within a longitudinal study designed to establish the epidemiology and etiology of pneumonia in a cohort of refugee children aged <2 years. We were particularly interested to understand the behavior of serotypes included in the current conjugate vaccines, as they are the most prevalent serotypes causing disease and conjugate vaccines are known to interfere with pneumococcal transmission by perturbing nasopharyngeal colonization.
The study was carried out in Maela, a long-term camp for displaced persons in rural Northwest Thailand. Maela has a population of ∼40,000 displaced persons from Myanmar, predominantly of Karen ethnicity, who live in a 4 km2 area. Camp residents receive WHO EPI (Expanded Program on Immunization) immunizations.
Between October 2007 and November 2008, all pregnant women attending the SMRU antenatal clinic at 28–30 weeks gestation were invited to consent to their infant's participation in a pneumonia cohort study. Using sealed opaque envelopes containing an allocation code, women were randomly allocated to the pneumococcal carriage sub-cohort at enrolment. For this sub-cohort, women had a nasopharyngeal swab (NPS) taken at delivery and both infant and mother had a NPS taken at monthly surveillance visits from 1–24 months of age. Samples of blood were collected from the mother and umbilical cord at delivery and infant blood samples were taken at each monthly visit.
Infant: first episode | Infant: all subsequent episodes | Mother: all episodes | |||||||
Serotype | No. carriage episodes | Acquisition age, days Median (95% CI) | Carriage duration, days Median (95% CI) | Serotype | No. carriage episodes | Carriage duration, days Median (95% CI) | Serotype | No. carriage episodes | Carriage duration, days Median (95% CI) |
|
35 | 46.0 (16.0–48.0) | 31.0 (30.5–61.0) |
|
160 | 33.5 (31.5–60.0) |
|
145 | 31.5 (31.0–50.5) |
|
23 | 47.0 (15.5–79.5) | 212.5 (77.5–242.5) |
|
146 | 91.5 (62.0–120.0) |
|
50 | 30.5 (30.0–31.0) |
|
21 | 44.5 (15.0–47.0) | 184.0 (61.5–241.0) |
|
114 | 89.5 (62.0–94.0) |
|
31 | 31.0 (30.0–31.5) |
|
14 | 44.0 (15.5–48.0) | 119.5 (60.5–152.5) |
|
105 | 62.0 (58.0–91.0) |
|
25 | 31.0 (30.0–59.0) |
|
9 | 45.5 |
121.0 (29.5–179.5) |
|
72 | 86.5 (61.0–92.0) |
|
23 | 45.5 (30.5–90.0) |
|
8 | 45.5 (15.5–77.0) | 60.0 (30.0–89.5) |
|
68 | 60.5 (48.5–90.5) |
|
23 | 31.0 (29.5–31.5) |
|
8 | 16.0 (12.5–83.0) | 61.5(30.0–151.0) |
|
60 | 63.0 (47.5–120.5) |
|
22 | 31.5 (30.5–59.5) |
|
8 | 45.5 (13.5–137.5) | 60.5 (30.0–62.5) |
|
46 | 62.0 (33.0–93.5) |
|
18 | 32.5 (29.5–48.0) |
|
7 | 15.5 (13.0–19.5) | 83.5 |
|
43 | 58.5 (31.0–62.5) |
|
18 | 31.5 (30.0–32.5) |
|
5 | 15.5 |
122.5 |
|
38 | 63.0 (31.5–134.0) |
|
16 | 31.5 (29.0–32.0) |
|
95 | 44.0 (16.0–46.5) | 116.0 (88.0–127.0) |
|
611 | 62.0 (61.0–75.5) |
|
223 | 31.0 (31.0–31.5) |
|
95 | 45.5 (44.0–46.0) | 61.0 (57.0–86.5) |
|
508 | 45.5 (32.5–57.5) |
|
316 | 31.0 (31.0–31.5) |
|
225 | 45.5 (43.5–46.0) | 63.0 (60.5–89.5) |
|
1,279 | 60.0 (57.0–60.5) |
|
684 | 31.0 (31.0–31.5) |
Duration calculated for 602 carriage episodes (acquisition date could not be calculated for carriage episodes including the first or second swab).
Unable to estimate 95% CI.
Excluding non-typeable (NT) pneumococci.
NPS were collected according to the WHO pneumococcal colonization detection protocol
A carriage episode was defined as the period of time between acquisition and clearance of a pneumococcal serotype. Acquisition was identified when a pneumococcal serotype was cultured from a swab for the first time or when a serotype was re-cultured following clearance as defined below. Carriage episodes commenced at the midpoint between the last negative swab and the first positive swab for the serotype. Following acquisition, clearance of the serotype from the nasopharynx was considered to have occurred when two consecutive swabs were culture negative for that serotype. Termination of the carriage episode was defined as the midpoint between the last positive swab and the first negative swab for the serotype (
Serotype | No. carriage episodes | Reacquisition number | Carriage duration | Time to reacquisition |
||
(first/reacquisitions) | HR (95% CI) |
|
HR (95% CI) |
|
||
|
195 (129/66) | 1st | 1.13 (0.79–1.63) | .5 | 0.18 (0.10 –0.35) | < .001 |
2nd | 0.75 (0.33–1.71) | .5 | 0.22 (0.07–0.63) | .005 | ||
3rd | 0.31 (0.04–2.46) | .3 | 0.40 (0.12–1.38) | .1 | ||
|
169 (112/57) | 1st | 2.64 (1.50–4.67) | .001 | 0.39 (0.14–1.08) | .07 |
2nd | 2.81 (1.09–7.22) | .03 | 0.35 (0.09–1.32) | .1 | ||
3rd | 1.39 (0.90–2.16) | .1 | 1.00 (0.19–5.16) | 1.0 | ||
|
135 (101/34) | 1st | 1.56 (0.94–2.60) | .08 | 0.50 (0.19–1.31) | .2 |
2nd | 3.20 (1.93–5.28) | < .001 | 0.19 (0.04–0.95) | .04 | ||
3rd | 0.76 (0.51–1.11) | .2 | 0.79 (0.08–7.99) | .8 | ||
|
119 (85/34) | 1st | 1.18 (0.62–2.23) | .6 | 0.30 (0.11–0.84) | .02 |
2nd | 1.21 (0.59–2.46) | .6 | 0.21 (0.05–0.84) | .03 | ||
3rd | - | - | - | - | ||
|
80 (71/9) | 1st | 4.12 (1.35–12.59) | .01 | 0.04 (0.04–0.35) | .004 |
2nd | - | - | - | - | ||
3rd | - | - | - | - |
Cox proportional hazards model.
Time from clearance of a serotype to subsequent reacquisition.
Parametric survival model (Weibull distribution).
Data were entered into an Access 2003 database (Microsoft) and the entire database was manually checked for errors by comparing with the original case record forms. Statistical analyses were carried out using Stata/IC 12.0 (StataCorp). Comparisons were made using the Wilcoxon rank-sum test. Categorical data were analyzed using the chi-squared test. Logistic regression models, with either study participant identifier included as a random-effect or robust standard errors to control for repeated observations within individuals, were used to determine temporal changes in serotype distributions, multiple serotype detection, and transmission. Odds ratios, adjusted for the number of NPS specimens per individual, were calculated to determine the likelihood of carriage of serotypes in infants or mothers. Time to pneumococcal acquisition and carriage duration were estimated by survival analysis, since some carriage episodes were censored. The log-rank test was used to compare groups. Survival models (Cox proportional hazards or parametric models using exponential or Weibull distributions) were used to assess potential predictors of acquisition and carriage duration
Written informed consent was obtained from the mothers prior to study enrolment. Ethical approval 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).
The bars indicate the number of isolates of each serotype and the dashed lines indicate the cumulative frequency (cumulative frequency of 67% is indicated by the vertical arrows). Dark gray bars highlight PCV13 serotypes.
Serotype | Overall N (%; rank) | Infant N (%; rank) | Mother N (%; rank) | Carriage in infants vs mothers OR (95% CI) |
|
660 (15.0; 1) | 354 (10.5; 3) | 306 (29.6; 1) | 2.8 (1.9–4.2) |
|
604 (13.7; 2) | 545 (16.2; 1) | 59 (5.7; 2) | 4.5 (2.9–7.0) |
|
471 (10.7; 3) | 432 (12.9; 2) | 39 (3.8; 3) | 6.3 (3.8–10.2) |
|
337 (7.7; 4) | 308 (9.2; 4) | 29 (2.8; 5) | 7.3 (4.2–12.8) |
|
216 (4.9; 5) | 188 (5.6; 5) | 28 (2.7; 6) | 4.7 (2.7–8.0) |
|
184 (4.2; 6) | 167 (5.0; 6) | 17 (1.7; 10) | 5.2 (2.8–9.7) |
|
171 (3.9; 7) | 152 (4.5; 7) | 19 (1.8; 9) | 5.7 (3.2–10.4) |
|
138 (3.1; 8) | 99 (2.9; 10) | 39 (3.8; 3) | 1.8 (1.0–3.2) |
|
117 (2.7; 9) | 108 (3.2; 8) | 9 (0.9; 18) | 6.7 (3.0–14.7) |
|
114 (2.6; 10) | 100 (3.0; 9) | 14 (1.4; 13) | 3.9 (2.0–7.6) |
|
2,162 (49.2%) | 1,878 (55.8%) | 284 (27.5%) | 9.0 (5.1–15.7) |
|
1,574 (35.8%) | 1,131 (33.6%) | 443 (42.9%) | 8.3 (4.4–15.6) |
|
4,396 | 3,363 | 1,033 | 4.9 (2.3–10.2) |
OR for the serotype being carried at any time point (adjusted for number of swabs collected per individual; all
Excluding non-typeable (NT) pneumococci.
Of 999 pregnant women recruited into the pneumonia cohort study, 250 were randomly allocated to the pneumococcal carriage sub-cohort. Nineteen women and their babies did not attend the first follow up visit (14 lost to follow up, four neonatal deaths, and one stillbirth) resulting in 231 mothers and 234 infants (three sets of twins) who had at least one surveillance NPS collected. These 234 mother-infant pairs are included in the analysis.
8,386 NPS were collected (73.6% of expected): 4,195 from mothers and 4,191 from infants (median 23 per individual; range 1–24 [infants], 2–25 [mothers]). The median mother-infant pair follow-up duration was 23.9 months post-delivery (728 days; range 28–760). Overall there were 2,188 carriage episodes identified (1,504 infant; 684 mother): infants had a median of seven (range 0–15) carriage episodes, compared with a median of two (range 0–16) in mothers (
Infant acquisition was extremely rapid and by the three month visit, 75.7% of infants had been colonized increasing to 97.0% by six months. All but one infant had acquired pneumococci at least once by the eleven month visit. The median age of first acquisition was 45.5 days (95% confidence interval [CI], 44.5–46.0), with an observed and modeled acquisition rate of 1.8 per 100 child-days (
We explored whether age at first acquisition was associated with household size, presence of other young children in the house, ethnic group, prematurity (delivery <37 weeks gestation), home delivery, season of birth, maternal pneumococcal colonization at birth, maternal smoking, or antibiotic consumption in the neonatal period (
The median duration of the infants' first ever carriage episode was 63.0 days (95% CI, 60.5–89.5). Duration of carriage of serotypes 19F and 23F was longer (
There were 272 instances of re-acquisition of a previously carried serotype. NT pneumococci, 19F, 23F, 6B and 14 accounted for ∼75% of these re-acquisitions. Controlling for age and carriage of heterologous serotypes, previous carriage of serotype 14 or 19F was associated with significantly reduced duration (HR>1) of subsequent carriage episodes of the same serotype, and there was a trend in the same direction for serotype 23F. Previous carriage was also associated with an increased interval to reacquisition (HR<1) of the same serotype (
The pneumococcal nasopharyngeal acquisition rate was significantly lower in mothers compared with infants (0.5 vs. 1.1 per 100 person-days;
There were 3,963 mother-infant swab pairs (swabs taken from both mother and infant on the same day), and in 233 (5.9%) a common serotype was identified. Identification of a common serotype in a mother-infant pair became less common as infant age increased (
A total of 4,396 pneumococci were isolated from all swabs taken, comprising 67 serotypes (3,363 [infants]; 1,033 [mothers]). In infants, eight serotypes (19F, 23F, NT, 6B, 14, 6A, 15B/C, 6C) accounted for 67.0% of the isolates (
In swabs where pneumococci were cultured, multiple serotypes were identified in 5.1% of infant swabs and 2.6% of mother swabs. Detection of more than one carried serotype became more common as age increased (infants
This is the largest longitudinal pneumococcal carriage study in children under two years, in terms of the number studied and the sampling frequency/duration of follow up. Adherence to the WHO detection protocol enables direct comparison with other similar studies. The high level of follow-up and the uniformity of swab frequency over the follow-up period permitted confident assessment of carriage prevalence and dynamics. The inclusion of consistent follow-up in the second year of life was important since the risk of invasive disease remains high
In this crowded refugee camp, infants were colonized frequently by pneumococci and the first acquisition occurred early in life. The median age of acquisition (45.5 days) is similar to published studies from Asia and Africa, although in individual studies from Papua New Guinea and The Gambia, infants were colonized even earlier (60% by 15 days and 50% by 33 days respectively)
Carriage duration varied by serotype, with serotypes 19F and 23F being carried for the longest periods in infants. For serotypes 14 and 19F, previous carriage of the serotype was associated with significantly increased interval to reacquisition and shorter duration of subsequent carriage of the same serotype. This trend was observed for 23F, although the reduction in carriage duration did not reach significance. These findings suggest that nasopharyngeal exposure to serotypes with immunogenic capsular polysaccharides (14/19F > 23F > 6B) result in an immune response capable of enhancing clearance if reacquisition occurs. The fact that NT carriage was also associated with delayed reacquisition of NT pneumococci implies that other factors, such as antibody responses to common surface proteins, are also involved in the regulation of pneumococcal nasopharyngeal colonization. Modeling of data from previous longitudinal carriage studies has demonstrated evidence for both serotype-dependant and independent effects on pneumococcal serotype acquisitions in young children, but not both in the same study population. Carriage of eight serotypes was modeled in Israeli infants aged 12–35 months, and a significantly lower risk of acquisition of serotypes 6A, 14, and 23F in those previously colonized was found
Eight serotypes accounted for two-thirds of the pneumococcal isolates from infants. Serogroups 6, 19, 23 and serotype 14 have been the most prevalent pneumococci in the majority of previous infant/childhood carriage studies, regardless of geographical location
There are some limitations to the study. Firstly, the WHO culture protocol underestimates the prevalence of multiple serotype carriage. We recently confirmed that while the predominant serotype is likely to be identified correctly, minor carried serotypes may be missed
In conclusion, we have documented the characteristics of pneumococcal carriage in mothers and infants in a rural SE Asian community. These data will inform vaccination strategies in this population. Determination of timing of transmission of pneumococci between mother and infant, and the impact of early colonization by a serotype on subsequent acquisitions, improve understanding of the complex process of early pneumococcal nasopharyngeal colonization. These observations will be supported by on-going molecular and immunologic studies of the pneumococcal isolates and the cohort, illuminating molecular changes in pneumococci in response to carriage and defining immune responses to pneumococcal antigens that may identify targets for new pneumococcal vaccines.
(PDF)
(PDF)
We are grateful to the women and infants for participating in the study, the staff of the SMRU clinic in Maela, and the SMRU laboratory team in Mae Sot. We also thank Peter Christensen for performing the serotype 6C PCR work.