Dynamics of nasopharyngeal bacterial colonisation in HIV-exposed young infants in Tanzania

Authors

  • G. D. Kinabo,

    1. Department of Pediatrics, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • A. van der Ven,

    1. Department of Internal Medicine, Radboud University Medical Center, Nijmegen, The Netherlands
    2. Nijmegen Institute for Infection, Inflammation and Immunity, Radboud University Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • L. J. Msuya,

    1. Department of Pediatrics, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • A. M. Shayo,

    1. Department of Pediatrics, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • W. Schimana,

    1. Department of Pediatrics, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • A. Ndaro,

    1. Biotechnology Laboratory, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • H. A. G. H. van Asten,

    1. Department of International Health, Radboud University Medical Center, Nijmegen, The Netherlands
    2. Nijmegen Institute for Infection, Inflammation and Immunity, Radboud University Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • W. M. V. Dolmans,

    1. Department of International Health, Radboud University Medical Center, Nijmegen, The Netherlands
    2. Nijmegen Institute for Infection, Inflammation and Immunity, Radboud University Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • A. Warris,

    Corresponding author
    1. Department of Pediatrics, Radboud University Medical Center, Nijmegen, The Netherlands
    2. Nijmegen Institute for Infection, Inflammation and Immunity, Radboud University Medical Centre, Nijmegen, The Netherlands
    • Department of Pediatrics, Kilimanjaro Christian Medical Centre, Moshi, Tanzania
    Search for more papers by this author
  • P. W. M. Hermans

    1. Department of Pediatrics, Radboud University Medical Center, Nijmegen, The Netherlands
    2. Nijmegen Institute for Infection, Inflammation and Immunity, Radboud University Medical Centre, Nijmegen, The Netherlands
    3. Nijmegen Center for Molecular Life Sciences, Radboud University Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author

Corresponding Author Adilia Warris, Department of Pediatrics, 804; Radboud University Medical Center; P.O. Box 9101; 6500 HB Nijmegen; the Netherlands. Tel.: +31-243614430; Fax: +31-243616428; E-mail: a.warris@cukz.umcn.nl

Abstract

Objectives

To estimate the prevalence of nasopharyngeal bacterial colonisation (NPBC) patterns in young Tanzanian HIV-exposed infants and to analyse the influence of maternal NPBC and of the infant's HIV status on the NPBC pattern.

Methods

Longitudinal cohort study of neonates born to HIV-infected mothers visiting Kilimanjaro Christian Medical Centre, Tanzania, between 2005 and 2009. Demographic and clinical data and nasopharyngeal bacterial cultures were obtained at the age of 6 weeks, 3 and 6 months, and at one time point, a paired mother–infant nasopharyngeal swab was taken.

Results

Four hundred and twenty-two swabs were taken from 338 eligible infants. At 6 weeks of age, colonisation rates were 66% for Staphylococcus aureus, 56% for Streptococcus pneumoniae, 50% for Moraxella catarrhalis and 14% for Haemophilus influenzae. Colonisation with S. aureus diminished over time and was more common in HIV-infected infants. S. pneumoniae and H. influenzae colonisation rose over time and was more prevalent in HIV-uninfected children. Co-colonisation of S. pneumoniae with H. influenzae or M. catarrhalis was mostly noticed in HIV-infected infants. S. pneumoniae and M.catarrhalis colonisation of the mother was a risk factor for colonisation in HIV-uninfected infants, while maternal S. aureus colonisation was a risk factor for colonisation in HIV-infected infants. Among the 104 S. pneumoniae isolates, 19F was most prevalent, and 57 (55%) displayed capsular serotypes represented in the 13-valent pneumococcal conjugate vaccine.

Conclusions

NPBC was common in Tanzanian HIV-exposed infants. The significant prevalence of pneumococcal vaccine serotypes colonising this paediatric population justifies the use of the 13-valent pneumococcal vaccine to reduce the burden of pneumococcal invasive disease.

Abstract

Objectifs

Estimer la prévalence des modes de colonisation bactérienne rhinopharyngée (CBRP) chez les jeunes nourrissons tanzaniens exposés au VIH et analyser l'influence de la CBRP maternelle et du statut VIH de l'enfant sur le mode de CBRP.

Méthodes

Etude de cohorte longitudinale de nouveau-nés de mères séropositives visitant le Centre Médical Chrétien de Kilimandjaro, en Tanzanie entre 2005 et 2009. Les données démographiques et cliniques et les cultures bactériennes rhinopharyngées ont été obtenues à l’âge de 6 semaines, 3 et 6 mois et à un moment donné, un échantillon rhinopharyngé apparié mère-enfant par écouvillonnage a été prélevé.

Résultats

422 prélèvements ont été effectués chez 338 nourrissons admissibles. A 6 semaines d’âge, les taux de colonisation étaient de 66% pour Staphylococcus aureus, 56% pour Streptococcus pneumoniae, 50% pour Moraxella catarrhalis et 14% pour Haemophilus influenzae. La colonisation par S. aureus diminuait au fil du temps et était plus fréquente chez les nourrissons infectés par le VIH. La colonisation par S. pneumoniae et H. influenzae augmentait au fil du temps et était plus fréquente chez les enfants non infectés par le VIH. La co-colonisation par S. pneumoniae et H. influenzae ou M. catarrhalis était surtout constatée chez les nourrissons infectés par le VIH. La colonisation par S. pneumoniae et M. catarrhalis de la mère était un facteur de risque pour la colonisation chez les nourrissons séronégatifs, tandis que la colonisation par S. aureus de la mère était un risque de colonisation pour les enfants séropositifs. Parmi les 104 isolats de S. pneumoniae, le sérotype 19F était le plus répandu et 57 (55%) isolats présentaient des sérotypes capsulaires présents dans le vaccin conjugué 13-valent contre le pneumocoque.

Conclusions

La CBRP était fréquente chez les nourrissons exposés au VIH en Tanzanie. La prévalence importante des sérotypes vaccinaux contre le pneumocoque colonisant cette population pédiatrique justifie l'utilisation du vaccin 13-valent contre le pneumocoque pour réduire la charge de la maladie invasive à pneumocoques.

Abstract

Objetivos

Calcular los patrones de prevalencia de la colonización nasofaríngea por bacterias (CNB) en neonatos expuestos al VIH en Tanzania y analizar la influencia de la CNB materna y del estatus de VIH de los neonatos sobre el patrón de CNB.

Métodos

Estudio longitudinal de cohortes de neonatos nacidos de madres infectadas con VIH que se visitaron en el Centro Médico Cristiano de Kilimanjaro, Tanzania entre el 2005 y el 2009. Se obtuvieron datos demográficos y clínicos, se tomaron muestras nasofaríngeas para realizar cultivos a la edad de 6 semanas, 3 y 6 meses y en un momento dado se tomó una muestra pareada madre-hijo de frotis nasofaríngeo.

Resultados

Se realizaron 422 frotis a 338 neonatos eligibles. A las 6 semanas de edad, la tasa de colonización era del 66% para Staphylococcus aureus, 56% para Streptococcus pneumoniae, 50% para Moraxella catarrhalis y 14% para Haemophilus influenzae. La colonización con S. aureus disminuyó a lo largo del tiempo y era más común entre los neonatos infectados con VIH. La colonización con S. pneumoniae y H. influenzae aumentó a lo largo del tiempo y era más prevalente en niños no infectados con VIH. La colonización conjunta de S. pneumoniae con H. influenzae o M. catarrhalis se observaba principalmente en los neonatos infectados con VIH. La colonización por S. pneumoniae y M. catarrhalis de la madre era un factor de riesgo para la colonización de los neonatos no infectados con VIH, mientras que la colonización materna con S. aureus era un factor de riesgo para la colonización de neonatos infectados con VIH. Entre los 104 aislados de S. pneumoniae, el 19F era el más prevalente, y 57 (55%) presentaban serotipos capsulares representados en la vacuna de neumococo conjugada 13-valente.

Conclusiones

La CNB era común entre neontaos Tanzanos expuestos al VIH. Una prevalencia significativa de los serotipos presentes en la vacuna de neumococo entre los que colonizan a esta población pediátrica justifica el uso de la vacuna de neumococo conjugada 13-valente para reducir la carga de la enfermedad neumocócica invasiva.

Introduction

The nasopharynx of children is frequently colonised by a wide variety of microorganisms, including bacteria such as Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus and Moraxella catarrhalis. Asymptomatic carriage of these bacteria is common and considered to be a crucial step towards the development of invasive disease such as sepsis, meningitis and pneumonia (Nguyen et al. 1999; Bogaert et al. 2004a). As a result of a B-cell dysfunction in association with the primary T-cell dysfunction, children with HIV infection are particularly susceptible to systemic diseases caused by these bacteria, in particular S. pneumoniae (Farley et al. 1994; Madhi et al. 2000; Bliss et al. 2008).

Reported nasopharyngeal colonisation rates of S. pneumoniae in HIV-infected versus non-infected children and HIV-exposed versus unexposed children vary considerably and even show contrasting results (Rusen et al. 1997; Leibovitz et al. 1999; Polack et al. 2000; Cardoso et al. 2006). S. aureus colonisation is consistently more frequent in HIV-infected patients than in uninfected patients, including children (Raviglione et al. 1990; Rusen et al. 1997; Nguyen et al. 1999; Sissolak et al. 2002; D'Avila et al. 2008). The inverse relationship between the nasopharyngeal carriage of S. pneumoniae and S. aureus in HIV-uninfected children (Bogaert et al. 2004b; Regev-Yochay et al. 2004) was not observed in HIV-infected children (McNally et al. 2006; Madhi et al. 2007). From 2002 onwards, it was advised that HIV-exposed neonates should receive co-trimoxazole prophylaxis until proved HIV negative. One would expect less colonisation in this group, but Gill et al. reported that although there is reduction in colonisation rates, it was not significantly different from those without prophylaxis (Gill et al. 2008).

Few studies have been carried out on the natural history of nasopharyngeal bacterial colonisation (NPBC) in young HIV-exposed infants. This is particularly true in sub-Saharan Africa (SSA) where the prevalence of HIV is high and the burden of disease caused by these bacteria is significant, even among HIV-unexposed children (Crump et al. 2011; Obaro et al. 2011). We therefore performed a longitudinal cohort study of neonates born to HIV-infected Tanzanian women to analyse the dynamics of NPBC by S. pneumoniae, S. aureus, H. influenzae and M. catarrhalis in HIV-infected and HIV-uninfected infants between 6 weeks and 6 months of age. Paired mother–infants nasopharyngeal swabs and known risk factors for colonisation were also analysed, and serotyping of all isolated S. pneumoniae was performed to estimate the impact of eventual introduction of the 13-valent pneumococcal conjugate vaccine.

Methods

Study population and data collection

Infants born to HIV-infected mothers seen at the Reproductive and Child Health (RCH) clinic at Kilimanjaro Christian Medical Centre (KCMC), which serves as a tertiary referral hospital in the northern part of Tanzania, were included between September 2005 and December 2009. Informed consent was obtained from the mother or caretakers when visiting the RCH clinic with their infant at the age of 6 weeks. Data were stored in case record forms (CRF) and included information regarding living conditions, family size, way of cooking, feeding practices, maternal antiretroviral prophylaxis or therapy, the medical condition of the mother (such as tuberculosis, any hospital admission and use of antibiotics) and neonatal antiretroviral prophylaxis. CRF data of the infants covered the various visits and included results of physical examination, weight, length, immunisation status and medication used. All neonates were prescribed co-trimoxazole prophylaxis from the age of 6 weeks, which was stopped if they were confirmed HIV negative. At the age of 6 weeks, venous blood was taken for HIV-RNA real-time PCR (M2000 RT, Abbott, Illinois USA). A second HIV-RNA PCR was performed in initially breast-fed babies if breast feeding was stopped at least 6 weeks before the evaluation day. In an infant never breast-fed, a single negative PCR test at the age of 6 weeks was regarded to exclude HIV infection. Infants who were confirmed HIV positive and met criteria for starting HAART were taken out of the study and started on HAART at the Pediatric Infectious Diseases Clinic. Pneumococcal vaccination of infants has not been introduced in Tanzania yet, and none of the included infants received it. As from 2009, H. influenzae B vaccine is given to children under the age of five in Tanzania, but none of the included infants had received this vaccine yet.

Bacteriological procedures

Nasopharyngeal and nasal swabs were taken at the age of 6 weeks, 3 and 6 months, in combination with a single swab from mothers or caretaker when the child was 3 or 6 months of age. Because of limited personnel available in the laboratory at KCMC, swabs could only be taken on two days each week. A nasopharyngeal swab was taken by touching the back of the nasopharynx using a flexible swab (Copan 168C) transnasally. The swab was placed directly in Amies transport medium and kept at room temperature for at maximum of 6 h. The swabs were sent to the Biotechnology Laboratory of the KCMC and plated immediately after arrival on a blood agar plate containing 5% sheep blood (Becton Dickinson, BD) and a chocolate agar plate (BD). The tip of the swabs was placed in 15–20% glycerol/PBS and stored at −80 °C for later use. The plates were incubated for 24 h at 35 °C in a CO2-rich environment. Bacterial growth of S. pneumoniae, M. catarrhalis, H. influenzae and S. aureus was determined by standard procedures daily. For nasal specimens, the swab was rubbed gently through the nose and placed directly in Amies transport medium of dehydrated phenol red mannitol broth (BD) and kept at room temperature for a maximum of 6 h. The swabs were placed overnight in a 35 °C incubator. In case, the medium turned yellow, the tubes were cultured on a blood agar plate (BD) and incubated for 24 h in a 35 °C incubator. The bacteria grown at plates were swabbed using a loop and put in 15–20% glycerol/PBS. The stock was stored at – 80 °C.

All nasopharyngeal swabs were sent to the Laboratory of Paediatric Diseases in Nijmegen for serotyping of the S. pneumoniae isolates. The vials containing the bacterial cultures from the nasopharyngeal swabs were used to inoculate blood agar plates, which were subsequently incubated overnight in a 37 °C incubator containing 5% CO2. The next day, single colonies of presumed S. pneumoniae were picked and grown in liquid GM17 medium. These cultures were used to prepare a −80 °C stock in 10% skim milk/10% glycerol and for DNA isolation (QIAamp DNA Blood mini kit). To confirm the strains to be S. pneumoniae and to characterise their capsular serotypes, a multiplex PCR was performed (Pai et al. 2006; Saha et al. 2008). PCR-negative specimens were serotyped by the capsular swelling method (Quellung reaction; Statens Serum Institute, Denmark). In case strains were Quellung reaction negative, they were re-cultured on blood agar plates, and optochin discs (Brunschwig Chemie, the Netherlands) were placed on top. The plates were then incubated overnight in a 37 °C incubator containing 5% CO2. In case an inhibition zone was present, the strains were considered to be non-typeable S. pneumoniae isolates.

Ethical considerations

The study was conducted in accordance with the research ethics regulations of the Tumaini University. Ethical clearance through the IRB of KCMC was obtained prior to the start of this study. Informed consent was obtained prior to inclusion into the study by the principal investigator or the research nurse from the parent(s) or caretakers of the infants. Access to research data was only allowed to involved investigators, and no patient names were used in the electronic database.

Statistical analysis

Data were entered into spss 16.0 for Windows. Data analysis was carried out using PASW Statistics 18 (SPSS, Chicago, IL, USA). Frequency tables and cross tabulations for different colonising bacteria were made according to age, and bacteria concurrent colonisation without HIV status and with HIV status was carried out. Co-colonisation was analysed by cross-tabulation stratified for HIV status, and without HIV status, between different bacteria isolated, and expected count. The chi-squared test was used to establish P values; where expected numbers were smaller than five, Fisher's exact test was used. As six comparisons were made between four different colonising bacteria, a P value < 0.01 was regarded to indicate a statistical significance. Backward binary multivariate logistic regression was used to determine relationships between risk factors and colonisation with specific bacteria at each specific age. All factors with P = 0.25 in univariate analysis were included as candidate predictors in backward stepwise binary logistic regression modelling; covariates were removed from the model if P < 0.10; the model included a constant. Variables included in the multivariate analysis were as follows: HIV status, having siblings below 10 years of age, having siblings above 10 years of age, residence (urban vs. rural), cooking practices (using wood vs. electricity), breast feeding, hospitalisation and use of antibiotics apart from the co-trimoxazole prophylaxis. As infants were not systematically swabbed on all subsequent visits, due to availability of swabbing facilities on two days of the week only, repeat measurement analysis was not possible.

Results

Within a period of 4 years and 3 months, 338 infants born to HIV-infected mothers were eligible, of whom 29 (8.6%) were HIV infected. Between follow-up at 3 and 6 months, seven infants died, 19 were prescribed anti-retroviral treatment (ART), and 16 were lost to follow-up.

Prevalence of colonisation by age

A total of 422 swabs for culture were taken: 156 at age 6 weeks, 130 at age 3 months and 136 at age 6 months. As shown in Table 1, high prevalences (>50%) of colonisation with S. pneumoniae and M. catarrhalis were found at 6 weeks, 3 and 6 months of age. At 6 weeks, S. aureus was most common (66%). Colonisation by S. pneumoniae, M. catarrhalis and H. influenzae increased with age, while that by S. aureus decreased with age.

Table 1. Prevalence of nasopharyngeal and nasal bacterial colonisation by age
 AGE
6 weeks3 months6 months
Number of swabs156130136
S. pneumoniae 56%76%75%
S. aureus 66%36%24%
H. influenzae 14%19%27%
M. catarrhalis 50%58%67%
Nasal S. aureus58%45%33%

Colonisation by HIV status

In HIV-uninfected infants, colonisation with S. pneumoniae was more frequent at 6 weeks, 3 and 6 months of age, than in HIV-infected infants, reaching statistical significance at 6 months (P = 0.017; Figure 1). In HIV-infected infants, S. aureus colonisation was more frequent than in HIV-uninfected infants at each of the three time points; the difference was statistically significant at 6 months of age (P = 0.003). Nasal S. aureus colonisation was also more common in HIV-infected infants but the difference was only statistically significant at 3 months (P = 0.017).

Figure 1.

Prevalence of colonisation by age and HIV status. Represented are nasopharyngeal colonisation rates of S. pneumoniae, S. aureus, H. influenzae and M. catarrhalis, as well as nasal S. aureus colonisation in HIV-infected (blue) and non-infected (red) infants at age of 6 weeks, 3 and 6 months (first, second and third pair bars, respectively). In HIV-infected infants, nasopharyngeal colonisation with S. pneumoniae was significantly lower at 6 months (P = 0.017), while S. aureus colonisation was significantly higher at 6 months of age (P = 0.003). Nasal S. aureus colonisation was higher in HIV-infected infants, but the difference was only statistically significant at 3 months (P = 0.017).

Lumping together colonisation by any of the bacteria examined (data not shown), we found that at age of 6 weeks, 3 and 6 months, the mean number of concurrent bacteria isolated was approximately two in both HIV-infected and uninfected infants.

Co-colonisation in HIV-infected and uninfected infants

Co-colonisation of the various bacteria stratified for HIV status and age is shown in Table 2. At the age of 6 weeks in the HIV-uninfected infants, S. aureus carriage was 66% and H. influenzae carriage was 14%. The expected dual carriage of the two pathogens was 11.8 cases, whereas we observed seven cases (P = 0.01). The remaining differences at 6 weeks were not statistically significant at the P < 0.01 level that was used to correct for multiple comparisons. For nasal S. aureus and nasopharyngeal S. aureus, the expected concurrent colonisation (i.e. if colonisation by one of the two did not influence the other) was 49.2 cases, whereas we observed 60 (46.4%; P = 0.001). At the age of 3 months, in the HIV-uninfected infants, less bacterial concurrent colonisation of S. pneumoniae and S. aureus was observed than expected (27.7 vs. 22 cases; P = 0.004). The remaining differences at 3 months were not statistically significant. At the age of 6 months, there was more concurrent colonisation of S. pneumoniae and H. influenzae than expected: we expected 23.8 cases, but observed 29 cases (P = 0.006). Also, more than expected nasopharyngeal S. aureus colonisation concurrent with nasal S. aureus colonisation was seen (one case observed vs. 6.2 expected; P = 0.011).

Table 2. Frequency of bacterial co-colonisation by age and HIV statusThumbnail image of

In HIV-infected infants (being on co-trimoxazole prophylaxis), at 3 months of age, 11 cases of concurrent colonisation of nasal S. aureus and nasopharyngeal S. aureus were seen, while only 6.8 cases were expected (P = 0.003; Table 2).

Risk factors for colonisation by different bacteria at different ages

Risk factors for bacterial colonisation were examined using backward stepwise logistic regression analysis. All samples were included in the multivariate analysis if all variables used in the model were available; this resulted in 142 (91%) of 156 samples at 6 weeks, 110 (85%) of 130 samples at 3 months and 128 (94%) of 136 samples at 6 months of age. The ORs with 95% confidence intervals (CI) are shown in Table 3.

Table 3. Risk factors for colonisation at 6 weeks, 3 and 6 months (backward stepwise logistic regression method)
Risk factor for colonisationP-valueaOR95% CI
  1. a

    P-value based on Fisher's exact test.

S. pneumoniae
At 6 Weeks
Sibling > 10 years0.052.11.0–4.4
At 3 months
Sibling < 10 years0.013.31.3–8.7
Hospitalisation0.020.20.07–0.8
At 6 months
HIV0.070.40.15–1.07
Hospitalisation0.070.40.14–1.09
S. aureus
At 6 Weeks
Residence Urban0.012.51.2–5.2
At 3 months
Sibling > 10 years0.010.320.13–0.78
At 6 months
HIV0.0034.11. 6–10.6
H. influenzae
At 3 months
Sibling < 10 years0.043.31.1–9.9
Sibling > 10 years0.062.71.0–7.4
 At 6 months
Antibiotic treatment0.082.40.9–6.0
M. catarrhalis
At 3 months
Breast feeding0.022.81.2–6.6
Hospitalisation0.0050.140.03–0.55
At 6 months
Hospitalisation0.040.370.14–0.96
Nasal S. aureus
At 6 weeks
Sibling < 10 years0.0082.81.1–5. 9
Rural<0.0010.20.1–0.4
At 3 months
HIV0.034.11.2–14.1
At 6 months
Breast feeding0.0093. 61.4–9.4

At 6 weeks of age, living in an urban area was a risk factor for S. aureus (OR 2.5, 95% CI 1.2 – 5.2; P = 0.01), and having siblings in the family below 10 years of age was a risk factor for S. aureus (OR 2.8, 95% CI 1.3 – 5. 9; P = 0.008). However, living in a rural area was highly protective for S. aureus colonisation (OR 0.2, 95% CI 0.1 – 0.4; P < 0.0001).

At 3 months of age, significant risk factors for S. pneumoniae colonisation were as follows: having siblings below 10 years of age (OR 3.4, 95% CI 1.3 – 8. 7; P = 0.01), while hospitalisation was negatively associated with colonisation (OR 0.2, 95% CI 0.07 – 0.8; P = 0.020). Having siblings less than 10 years of age was a risk factor for H. influenzae colonisation (OR 3.3, 95% CI 1.1 – 9.9; P = 0.04). Also, at 3 months of age, being HIV infected was a risk factor for nasal S. aureus colonisation (OR 4.1, 95% CI 1.2 – 1.4; P = 0.03), and breast feeding was a risk factor for M. catarrhalis colonisation (OR 2.8; 95% CI 1.2–6.6).

At 6 months of age, being HIV-infected was a risk factor for S. aureus colonisation (OR 4.1, 95% CI 1. 6 – 10.5; P = 0.003), and breast feeding was a risk factors for M. catarrhalis colonisation at 3 months of age (OR 2.8, 95% CI 1.2 – 6.6; P = 0.02). Hospitalisation at the age of 3 and 6 months reduced the risk of M. catarrhalis colonisation (OR 0.1, 95% CI 0.03 – 0.5; P = 0.065, and OR 0.37, 95% CI 0.14 – 0.96; P = 0.04).

Paired mother–infant colonisation patterns

As shown in Table 4, an increased risk of the infant being colonised if the mother was colonised was only observed for nasal S. aureus (OR 2.9, 95% CI 1.3–6.4; P = 0.01). This could not be stratified for HIV status because of low numbers of mother–infant pairs with known HIV status in both.

Table 4. The number (and percentage) of children colonised with S. pneumoniae, S. aureus, H. influenza or M. catarrhalis at any time during the first 6 months of life in whom the mother was colonised as well with the same microorganisms at any time point, compared to colonised children from whom the mother was not colonised at any time point
BacteriaChild colonisation N (%)P-valueaOR95% CI
Mother colonisedMother not colonised
  1. a

    P-value Fisher's exact test.

S. pneumoniae 23/28 (82)57/89 (64)0.102.60.9–7.5
S. aureus 17/45 (38)16/72 (22)0.092.10.9–4.8
H. influenzae 11/35 (32)25/82 (31)1.001.00.4–2.5
M. catarrhalis 32/40 (80)49/77 (64)0.092.30.9–5.6
Nasal S. aureus29/59 (49)14/56 (25)0.012.91.3–6.4

Hospitalisation and invasive bacterial infections related to colonisation

Among 102 infants colonised at 6 weeks with S. pneumoniae, 13 (12.7%) were hospitalised between 3 and 6 months of age, while three of 67 (4.5%) non-colonised infants were hospitalised, the difference not being statistically significant (OR 3.1; 95% CI 0.8–17.6). For S. aureus colonisation at 6 weeks, the odds ratio for admission was 1.7 (95% CI 0.5–7.7), and for H. influenzae-colonised infants, the OR was 1.1 (95%CI 0.5–4.5).

Of 156 infants swabbed for colonisation at 6 weeks, four infants were reported to have an invasive bacterial infection (pneumonia and/or meningitis and/or septicaemia) between 3 and 6 months of age. The odds ratios for bacterial infection were 2.0, 1.7 and <0.1 for colonisation with S. pneumoniae, S. aureus or H. influenzae, respectively, but did not reach statistical significance.

Prevalence of pneumococcal serotypes

Serotyping of 108 isolates of S. pneumoniae (five of them from HIV-infected infants) was performed (Figure 2). Of four isolates, the serotype could not be established (non-typeable strains). Among the remaining strains, serotype 19F was the most common (N = 20, 19%), two of them from HIV-infected infants, followed by serotype 6B in 11 samples (10%). Of all 104 pneumococci serotyped, 57 (55%), 51 (49%) and 50 (48%) had serotypes represented in the 13-valent, 10-valent and seven-valent pneumococcal conjugate vaccine (PCV).

Figure 2.

Pneumococcal serotypes identified among 104 S. pneumoniae isolates. In blue, the vaccine serotypes as present in the 13-valent pneumococcal conjugate vaccine are depicted (57/104; 55% of the total number of isolates)

Discussion

In HIV-exposed infants, high nasopharyngeal bacterial colonisation was observed at 6 weeks of age with prevalences of 50% or more for S. aureus, S. pneumoniae, M. catarrhalis and nasal S. aureus. During the first 6 months of life, increasing prevalences were observed for S. pneumonia, H. influenza and M. catarrhalis, whereas S. aureus colonisation rates decreased substantially. These patterns were comparable for HIV-infected and HIV-uninfected infants, although higher S. aureus colonisation rates (both nasal and nasopharyngeal) and lower S. pneumoniae colonisation rates were observed in HIV-infected infants reaching significant differences at 3 months of age for nasal S. aureus and at 6 months of age for nasopharyngeal S. aureus and S. pneumoniae.

In healthy infants, S. aureus colonisation was very common among newborns, with rapidly falling colonisation rates during the first year of life, while pneumococcal colonisation rates are low at birth and rise with age (Lebon et al. 2008). We observed a comparable pattern in HIV-exposed infants, irrespective of being HIV infected or not. Our findings in HIV-infected infants are similar to those of studies in South Africa and Brazil (McNally et al. 2006; D'Avila et al. 2008), which showed that S. aureus was prevailing in HIV-infected children. Although colonisation rates of S. aureus in those HIV-infected children were comparable to our results, colonisation rates in HIV-uninfected children were substantially lower in those studies. An explanation might be the much wider age range compared with the young infants in our study.

In the total group of HIV-exposed infants, a high colonisation rate of nasopharyngeal S. pneumoniae was comparable to a very recent study in Kenya (Abdullahi et al. 2012). In contrast to our results, a twofold higher odds ratio of S. pneumoniae colonisation was found in a selected group of HIV-infected children in this and other studies (Rusen et al. 1997; O'Brien et al. 2009; Abdullahi et al. 2012). The difference to our findings could be due to the fact that HIV-exposed infants in our study were on co-trimoxazole prophylaxis until the infant tested negative for HIV by PCR. Due to the fact that almost 60% of the infants were still breast-fed at the age of 6 months, co-trimoxazole was given for a prolonged period of time. Support for this explanation can be found in the study performed by Gill and colleagues who reported a reduction in pneumococcal colonisation in HIV-exposed infants on co-trimoxazole prophylaxis as compared to those who were not (Gill et al. 2008). S. pneumoniae, S. aureus and H. influenzae are reported to cause more invasive disease in colonised HIV-infected children than colonised HIV-uninfected and HIV-unexposed children (Farley et al. 1994; Madhi et al. 2000), with higher case-fatality rates associated with these infections (Molyneux 2004; O'Brien et al. 2009). We observed a threefold higher incidence of hospital admission in infants colonised by S. pneumoniae, but the odds ratio did not reach significance for admission or invasive infections, most probably due to the low numbers.

We observed more concurrent colonisation in HIV-uninfected infants of S. pneumoniae and H. influenzae at age of 3 and 6 months than in HIV-infected children, but in HIV-infected infants, more concurrent colonisation at 6 months of S. pneumoniae and H. influenzae, as well as M. catarrhalis, was observed. Contrary to our observations, previous studies have demonstrated a negative association between S. pneumoniae and S. aureus (Bogaert et al. 2004b; O'Brien et al. 2009). A study from The Gambia reported that newborns have high rates of S. aureus carriage, and H. influenzae and M. catarrhalis acquired early and frequently co-carried with S. pneumoniae (Kwambana et al. 2011). The difference with our study is that all infants included in our study were HIV exposed, while in their study, HIV status was not checked. In other studies (Bogaert et al. 2004b; Regev-Yochay et al. 2004), carriage of S. pneumoniae and S. aureus was observed, but not a combined colonisation of S. pneumoniae and S. aureus (Regev-Yochay et al. 2004). The latter studies were carried out in a middle class population, probably with better immune status than our population. Another explanation for the differences with our study might be the fact that all our HIV-infected children were on co-trimoxazole prophylaxis.

Identified risk factors for colonisation included that HIV infection increased the risk of S. aureus nasopharyngeal colonisation (at the age of 6 months), and for nasal S. aureus (at the age of 3 months). Having siblings below 10 years was a risk factor for colonisation with S. pneumoniae and also H. influenzae, but breast feeding even from HIV-infected mothers was not. Sabirov and his colleagues reported protection of H. influenzae in children who were breast-fed as compared to those who were on formula feeding (Sabirov et al. 2009).

An increased risk of child colonisation, if the mother was colonised, was only observed for nasal S. aureus carriage. It has been shown that even if the mother is immunised against pneumococci, colonisation rate in her HIV-exposed infant is not reduced (Almeida et al. 2011).

Our pneumococcal serotyping data showed that serotype 19F was the most prevalent, followed by 6B and 23F, both represented in all the three PCVs. Of the additional serotypes 1, 5 and 7F in the 10-valent PCV compared to the seven-valent PCV, only serotype 1 was isolated (n = 1) in our study. Although serotypes 1, 5 and 7F are important causes of invasive pneumococcal disease in Africa (Campbell et al. 2004; Cutts et al. 2005), colonisation rates with these serotypes in African children can be very low (Abdullahi et al. 2012). As 55% of the S. pneumoniae serotypes were represented in the 13-valent PCV (49 and 48% in the 10- and seven-valent PCV, respectively), countries like Tanzania are expected to prevent a significant proportion of morbidity and mortality due to pneumococcal disease by implementing routine PCV in their current immunisation programme (Abdullahi et al. 2012; Tigoi et al. 2012).

Acknowledgement

We thank Mrs. C. Gaast-de Jongh for performing the serotyping of the S. pneumoniae isolates.

Ancillary