- Top of page
- Materials and methods
Objective To describe and compare serotype distribution and antibiotic susceptibility of invasive and nasopharyngeal isolates of Streptococcus pneumoniae from children in rural Mozambique.
Methods From August 2002 to July 2003, we prospectively obtained invasive pneumococcal isolates from children <15 years of age admitted to the paediatric ward of Manhiça District Hospital. During a cross-sectional study of children <5 years of age with mild illnesses, attending the outpatient department of the hospital in March and April 2003, we collected nasopharyngeal isolates. Serotypes and antibiotic susceptibilities were determined using standardized methods.
Results The two most common pneumococcal serotypes among invasive isolates were types 1 (40% of 88 isolates serotyped) and 5 (10%), but these types were rare among nasopharyngeal isolates. Compared with invasive isolates, nasopharyngeal isolates were more likely to be serotypes in the licensed seven-valent conjugate vaccine (49%vs. 20%, P < 0.01), to have intermediate-level penicillin resistance (52%vs. 14%, P < 0.01) and to be non-susceptible to trimethoprim–sulfamethoxazole (61%vs. 45%, P < 0.01). Recent receipt of antibiotics or sulfadoxine/pyrimethamine were associated with carriage of antibiotic non-susceptible isolates.
Conclusions These data indicate that a pneumococcal conjugate vaccine containing serotypes 1 and 5 could substantially reduce pneumococcal invasive disease among young children in rural Mozambique. Carriage surveys can overestimate potential coverage of the seven-valent pneumococcal conjugate vaccine in settings where serotypes 1 and 5 predominate.
Objectif Décrire et comparer la distribution des sérotypes et la sensibilité aux antibiotiques de souches invasives et nasopharyngales de S. pneumoniae chez les enfants en milieu rural au Mozambique.
Méthodes D'août 2002 a juillet 2003, nous avons prospectivement collecté des souches invasive de pneumocoques provenant d'enfants de moins de 15 ans admis dans le service pédiatrique de l'hôpital de district de Manhiça. Au cours d'une étude transversale sur des enfants de moins de 5 ans avec des maladies bénignes se présentant dans le département des malades ambulatoires de l'hôpital entre mars et avril 2003, nous avons collecté des souches nasopharyngales. Les sérotypes et la sensibilité aux antibiotiques ont été déterminés par des méthodes standards.
Résultats Les deux sérotypes de pneumocoques les plus courants parmi les souches invasives étaient le type 1 (40% sur 88 souches typées) le type 5 (10%). Mais ces 2 sérotypes étaient rares parmi les souches nasopharyngales. Comparées aux souches invasives, les souches nasopharyngales étaient plus probables: 1) de faire partie de sérotypes inclus dans le vaccin conjugué hepta-valent breveté (49%vs. 20%, P < 0,01), 2) d'avoir un niveau de résistance intermédiaire à la pénicilline (52%vs. 145, P < 0,01) et 3) d’être résistantes au trimethoprim/sulfamethoxazole (61%vs. 45%, P < 0,01). Les prises récentes d'antibiotiques ou de sulfadoxine/pyrimethamine résultaient en une association avec des souches résistantes aux antibiotiques.
Conclusions Ces données indiquent qu'un vaccin conjuguée comportant les sérotypes 1 et 5 pourrait substantiellement réduire les maladies à pneumocoques invasives chez les jeunes enfants en milieu rural au Mozambique. Les surveillances de porteurs peu surestimer la couverture potentielle du vaccin pneumococcal hepta-valent conjugué dans les endroits où les sérotypes 1 et 5 sont prédominantes.
Objetivo Describir y comparar la distribución de serotipos y la susceptibilidad antibiótica de aislados invasivos y nasofaríngeos de S. pneumoniae de niños de un área rural de Mozambique.
Métodos Entre Agosto 2002 y Julio 2003 obtuvimos, de forma prospectiva, aislados de neumococo invasivo de niños menores de 15 años admitidos en la sala pediátrica del Hospital Distrital de Manhiça. Durante un estudio de corte transversal en niños <5 años con enfermedades leves, que visitaron las consultas externas del hospital entre Marzo y Abril del 2003, se recogieron los aislados nasofaríngeos. Los serotipos y la susceptibilidad antibiótica se determinaron mediante métodos estandarizados.
Resultados Los dos serotipos más comunes entre los aislados invasivos fueron el tipo 1 (40% de 88 aislados serotipados) y el 5 (10%), pero estos serotipos eran raros entre los aislados nasofaríngeos. Comparados con aislados invasivos, los nasofaríngeos eran más a menudo de los serotipos incluidos en la vacuna licenciada siete-valente (49%vs. 20%, P < 0.01), tenían unos niveles de resistencia intermedia a la penicilina (52%vs. 14%, P < 0.01) y no eran susceptibles al cotrimoxazol (61%vs. 45%, P < 0.01). El haber recibido antibióticos recientemente, incluido el cotrimoxazol, estaba asociado a la presencia de aislados resistentes.
Conclusiones Estos datos indican que una vacuna neumocócica conjugada, que contenga los serotipos 1 y 5, podría reducir sustancialmente la enfermedad neumocócica invasiva en niños pequeños del Mozambique rural. Los estudios de portadores pueden sobreestimar el cubrimiento potencial de la vacuna conjugada seven-valente en lugares en los que predominen los serotipos 1 y 5.
- Top of page
- Materials and methods
Streptococcus pneumoniae, the pneumococcus, is a leading cause of bacteraemia, pneumonia and meningitis in children worldwide and is estimated to be responsible for nearly one million deaths annually in children <5 years of age. Pneumococcus ranks among the major killers in sub-Saharan Africa, together with malaria and diarrhoea (Greenwood 1999; Heikki 2001; Black et al. 2003).
The emergence and progressive spread of pneumococcal strains resistant to penicillin and other antibiotics is a major public concern around the world (Klugman 1990; Adam 2002; Appelbaum 2002). The high disease burden in sub-Saharan Africa, combined with limited access to health services and obstacles to providing appropriate treatment, make vaccination against pneumococcal disease a priority.
The incidence of pneumococcal disease was reduced in the United States after introduction of a seven-valent pneumococcal conjugate vaccine in 2000 (Whitney et al. 2003). This vaccine includes the serotypes 4, 6B, 9V, 14, 18C, 19F and 23F (Prevnar®, Wyeth, Philadelphia, PA, USA), which accounted for >80% of the serotypes that caused invasive disease among young children in the US before conjugate vaccine introduction (Robinson et al. 2001). In developing countries, where the burden of pneumococcal disease is much higher, serotype distribution may differ greatly from that in industrialized countries (Hausdorff et al. 2000; Hausdorff 2002). To increase potential coverage in a wider range of settings, expanded pneumococcal conjugate vaccines are under development (Poolman 2004). These new vaccines include serotypes in the seven-valent formulation as well as types 1 and 5, which are important causes of invasive pneumococcal infections in African and Latin American children (Scott et al. 1996; Hausdorff et al. 2000). Two clinical trials in developing countries demonstrated that a conjugate vaccine containing nine serotypes effectively prevented invasive pneumococcal disease in both HIV-uninfected (Klugman et al. 2003; Cutts et al. 2005), and HIV-infected children (Klugman et al. 2003), substantially reduced the burden of pneumonia (Klugman et al. 2003; Cutts et al. 2005) and reduced all-cause mortality in one setting (Cutts et al. 2005). These findings raised hopes for a conjugate vaccine for Africa. More data on pneumococcal serotype distribution in sub-Saharan Africa, especially from rural areas, are needed to demonstrate the potential impact of pneumococcal conjugate vaccines.
Because surveillance for invasive pneumococcal disease is expensive, requiring a laboratory infrastructure rarely available in many developing countries, alternative methods are needed to characterize circulating pneumococci. Nasopharyngeal carriage studies of healthy children have been widely used to provide an estimate of antibiotic resistance among pneumococci in the community (Huebner et al. 1998; Feikin et al. 2000; Rowe et al. 2000), but their utility for measuring the prevalence of disease-causing serotypes is questioned (Gordon et al. 2003). To estimate the potential coverage of conjugate vaccine formulations, we investigated the serotype distributions and antibiotic resistance among invasive pneumococci isolated from children in rural Mozambique. Also, we compared these pneumococci with others recovered from the nasopharynx of children with mild illnesses.
- Top of page
- Materials and methods
During the study period, a total of 127 cases of invasive pneumococcal disease among patients <15 years of age were identified at Manhiça District Hospital. The median age of the case patients was 19 months (range: 5 days to 14 years); 5 (4%) were younger than 2 months and 105 (83%) were between 2 and 59 months of age. Males accounted for 61 (48%) cases. Antibiotic susceptibility testing was performed for all isolates; 88 (69%) were recovered for serotyping after storage. The median age of cases for whom isolates were not recovered was 15 months, which was not significantly different from that of cases included (P = 0.34).
Nasopharyngeal swabs were collected from 285 eligible children attending the outpatient clinic of the district hospital. The median age was 20 months (range: 1–59 months) and 162 (57%) were males. Pneumococci were recovered from 248 (87%) of these children. The prevalence of colonization decreased significantly with increasing age: 88 (92%) of 96 children <1 year, 115 (90%) of 128 children 1–2 years and 45 (72%) of 61 children 3–4 years were pneumococcal carriers (P < 0.01). Children as young as 1 month old were colonized (three of five infants of this age). Antimicrobial testing was performed for all nasopharyngeal isolates; 192 (77%) were recovered for serotyping after storage.
Pneumococcal carriage was more common among children who had LRTI symptoms at the time of enrolment [90 (93%) of 97] than among children without LRTI [156 (84%) of 185; P = 0.04]. Children living in cement block houses had lower prevalence of colonization (65%) than children in traditional mud or reed housing (88%; P = 0.01), and children who lived in a house where at least one adult smoked (78%) were less likely to be colonized than children who did not (91%; P = 0.02).
Among 88 invasive isolates available for serotyping, 8 (9%) were from CSF and 80 (91%) were from blood. Serotype 1 accounted for 6 (75%) of the isolates from CSF and 29 (36%) of the isolates from blood. We found 24 serotypes in the 88 invasive isolates. Five serotypes accounted for 61 (70%) of the invasive isolates: serotypes 1 (40%), 5 (10%), 14 (7%), 6A (7%) and 23F (6%). Only 18 [20%, 95%CI = (13, 30)] of the invasive isolates were serotypes included in the seven-valent conjugate vaccine, vs. 62 [70%, 95%CI = (60, 79)] in the nine-valent formulation which includes serotypes 1 and 5. An additional 13 (15%) were potentially cross-reactive serotypes (of the same serogroup as vaccine serotypes) for both seven and nine-valent vaccine formulations. Estimates of potential vaccine coverage were similar when analyses were limited to invasive isolates from children <2 years [23%, 95%CI = (13.36) for seven-valent and 65%, 95%CI = (52.77) for nine-valent formulations], and overall estimates did not change when invasive isolates from case-patients aged 5 years or older were excluded (data not shown).
There were 38 serotypes among 192 nasopharyngeal isolates tested. Five of these serotypes accounted for 54% of pneumococcal carriage: 19F (26%), 19A (8%), 23F (7%), 6A (7%) and 6B (6%). Serotypes 1 and 5 were each recovered from only one child. Ten nasopharyngeal isolates were nontypeable pneumococci. Serotypes and serogroups included in the seven-valent vaccine accounted for 49% [95%CI = (42, 56)] and 68% [95%CI = (61, 74)] of pneumococcal carriage.
Among invasive isolates, resistance to oxacillin was found in 23 (18% of 127) and of those tested for susceptibility (n = 13) 10 (77%) had intermediate resistance. Based on this result we estimated that 18 (14%) invasive isolates were non-susceptible to penicillin, compared with 128 (52% of 248) nasopharyngeal isolates (P < 0.01; Table 1). None were fully resistant to penicillin. Non-susceptibility to cotrimoxazole was found in 45% of invasive and 61% of nasopharyngeal isolates (P < 0.01). Resistance to penicillin and cotrimoxazole were highly correlated (P < 0.01). Resistance to erythromycin, chloramphenicol and cefotaxime was uncommon (<5%) or not observed. The proportions of antibiotic non-susceptibility were similar among serotyped and non-serotyped isolates (data not shown).
Table 1. Antibiotic susceptibility of invasive and nasopharyngeal isolates of Streptococcus pneumoniae from children in rural Mozambique
| ||Invasive isolates (N = 127)||Nasopharyngeal isolates (N = 248)|
|Intermediate [n (%)]||High [n (%)]||Range of MICs [n (%)]||Intermediate [n (%)]||High [n (%)]||Range of MICs (μg/ml)|
|*Penicillin||18 (14)||0–||0.094–0.250||128 (52)||0 –||0.094–0.380|
|Cotrimoxazole||14 (12)||39 (33)||–||91 (37)||59 (24)||0.75–32.0|
|Erythromycin|| ||1 (1)||–||0 –||4 (2)||1.0–256.0|
|Cefotaxime||Not tested||–||–||0 –||0 –||–|
|Chloramphenicol||1 (1)||3 (2)||–||0 –||4 (2)||8.0–12.0|
Figure 1 shows antibiotic susceptibility according to serotype. Serotype 19F isolates had the highest proportion of non-susceptibility to penicillin [42 (84%) of 50 nasopharyngeal isolates and one invasive isolate]. Among all serotyped isolates from invasive disease and carriage, serotypes in the seven-valent vaccine formulation were more likely to be non-susceptible to one or more antibiotics tested (90%) than non-vaccine serotypes (43%; P < 0.01).
Figure 1. Comparison of serotype distribution for invasive vs. nasopharyngeal isolates, by susceptibility. Note: Non-susceptible refers to intermediate or full resistance to one or more antibiotics tested (including oxacillin for invasive isolates).
Download figure to PowerPoint
Among children colonized with pneumococci, carriage of cotrimoxazole-non-susceptible isolates was associated with receipt of sulfadoxine–pyrimethamine in the 30 or 90 days prior to swab collection (Table 2), and receipt of sulfadoxine–pyrimethamine in the previous 30 days remained significant after controlling for age <2 years, gender and hospitalization in the previous 90 days. We found a similar pattern for carriage of penicillin–non-susceptible pneumococci, which tended to be associated with receipt of a beta-lactam antibiotic (mainly ampicillin; OR: 1.7; 95% CI: 0.8–3.8) or sulfadoxine–pyrimethamine (OR: 1.8; 95% CI: 0.6–5.1) in the 30 days prior to swab collection. None of the other variables were significantly associated with carriage of penicillin–non-susceptible pneumococci.
Table 2. Factors associated with nasopharyngeal carriage of cotrimoxazole-non-susceptible Streptococcus pneumoniae among colonized children
|Characteristic||Cotrimoxazole non-susceptible* [N = 150 (60)]||Cotrimoxazole susceptible [N = 98 (40)]||OR (95% CI)|| Adjusted* [OR (95% CI)]|
| <2||94 (60)||62 (40)||0.97 (0.58–1.65)||0.98 (0.57–1.70)|
| 2–4||56 (61)||36 (39)||Referent|| |
| Male||83 (61)||54 (39)||1.01 (0.61–1.68)||1.04 (0.61–1.77)|
| Female||67 (60)||44 (40)||Referent|| |
|Family member works in South Africa|
| Yes||55 (63)||32 (37)||1.22 (0.71–2.09)|| |
| No||93 (58)||66 (42)||Referent|| |
|Child hospitalized in previous 90 days|
| Yes||21 (78)||6 (22)||2.46 (0.96–6.34)||1.76 (0.65–4.77)|
| No||128 (59)||90 (41)||Referent|| |
|Health center visit in previous 30 days|
| Yes||81 (63)||47 (37)||1.29 (0.78–2.53)|| |
| No||68 (57)||51 (43)||Referent|| |
|LRTI at time of swab collection|
| Yes||53 (60)||36 (40)||0.98 (0.57–1.67)|| |
| No||93 (60)||62 (40)||Referent|| |
|Received cotrimoxazole in previous 30 days|
| Yes||8 (62)||5 (38)||1.05 (0.33–3.30)|| |
| No||142 (60)||93 (40)||Referent|| |
|Received sulfadoxine–pyrimethamine in previous 30 days|
| Yes||16 (94)||1 (6)||11.69 (1.52–89.7)||9.82 (1.26–76.7)|
| No||130 (58)||95 (42)||Referent|| |
|Received sulfadoxine–pyrimethamine in previous 90 days|
| Yes||36 (73)||13 (26)||2.09 (1.04–4.19)|| |
| No||110 (57)||83 (43)||Referent|| |
- Top of page
- Materials and methods
The most frequently isolated serotypes causing invasive pneumococcal disease in rural Mozambican children during this study were serotypes 1, which accounted for 40% and serotype 5, which comprised 10% of all invasive isolates serotyped. Thus a vaccine including these two serotypes, like the formulation used in clinical trials in South Africa and The Gambia (Klugman et al. 2003; Cutts et al. 2005), would provide much greater coverage in this population than the currently licensed seven-valent formulation (70%vs. 20%).
The predominance of serotype 1 and to a lesser extent serotype 5, among invasive isolates from young children in this rural African setting is similar to the distribution described in Kenyan children (Scott et al. 1998). In a review of geographical differences in serotype distribution, serotype 1 was the third most common serotype among invasive isolates from young children in several studies in Africa (Hausdorff et al. 2000), and it has been described as the most commonly isolated serotype in Egypt (31% of all serotypes isolated), Rwanda (22%) (Hausdorff et al. 2000), Malawi (27%) (Gordon et al. 2003) and Kenya (23%) (Berkley et al. 2005). In Mozambique, the proportion of serotype 1 isolates was even higher than in these other African studies. The prevalence of serotype 1 may reflect an outbreak of a virulent clone rather than endemic transmission and needs to be further investigated. Interestingly, surveillance in Mali during a 12-month period found that serotype 5 accounted for 50% of all invasive serotypes, while serotype 1 accounted for just 2% (Campbell et al. 2004). Multiyear surveillance may help distinguish transient increases from endemic disease burden.
Compared with invasive isolates, pneumococci recovered from the nasopharynx had widely different serotype distribution and antibiotic resistance. We found ubiquitous colonization among children <5 years, with moderate frequency of penicillin and cotrimoxazole non-susceptibility, consistent with results of other carriage studies in African populations (Huebner et al. 1998; Feikin et al. 2000; Rowe et al. 2000). Two variables associated with socioeconomic status in Mozambique, living in a cement block house or with an adult who smoked tobacco, were associated with lower prevalence of nasopharyngeal colonization among children. The predominance of serotypes 1 and 5 among invasive isolates greatly contributed to poor representativeness of nasopharyngeal isolates for estimating potential vaccine coverage. Serotypes 1 and 5 are rarely encountered in carriage studies (Brueggemann et al. 2003) and are usually susceptible to antibiotics. In regions where a high proportion of disease is caused by serotypes 1 and 5, carriage studies would likely overestimate the potential coverage of the seven-valent conjugate vaccine formulation. Our carriage study also overestimated the potential coverage of conjugate vaccine formulations because of the high prevalence of serotype 19F in carriage compared to invasive form (26%vs. <1%).
Because the majority of the non-susceptible pneumococci belonged to serotypes included in the conjugate vaccine formulations, an added benefit of pneumococcal vaccination may be to prevent antibiotic resistant infections. Increasing resistance to penicillin and cotrimoxazole in Mozambique may be expected because of the increased availability and use of certain antimicrobial agents. Resistance to cotrimoxazole among nasopharyngeal isolates was related to taking sulfadoxine–pyrimethamine, as reported by Feikin et al. (2000). Cotrimoxazole was still not widely used in this community for prophylaxis against HIV-opportunistic infections at the time of the survey. Because prophylactic use of cotrimoxazole is becoming more common, antibiotic resistance is expected to increase in this community. Periodic carriage studies may still be useful for monitoring trends in antimicrobial resistance among circulating pneumococci in the community, although they are likely to overestimate actual resistance among invasive infections. These carriage studies should be carried out in different seasons of the year in order to generalize our results.
This study has several limitations. HIV status of children was not known. HIV-infection greatly increases the risk of invasive pneumococcal disease in children and can affect susceptibility to certain pneumococcal serotypes (Scott et al. 1998; Karstaedt et al. 2000; Madhi et al. 2000; Westwood et al. 2000). HIV prevalence among women receiving prenatal care was 19% in Manhiça district in 2003 (unpublished data). As programmes to prevent mother-to-child transmission did not begin until the end of the surveillance period, HIV-infection may have been prevalent in the pediatric population. Additional limitations are the relatively small number of invasive isolates available after storage. Although subjects from whom isolates were recovered appeared to be similar to the rest of the sample, serotype distribution of non-serotyped isolates could have differed. In addition, the carriage study was limited to only a one-month period at the end of the rainy season among children attending a health facility. Carriage may change throughout the year, and may be different among healthy children in the community (Garcia-Rodriguez & Fresnadillo-Martinez 2002). This study highlights the limitations of carriage studies to inform potential coverage of pneumococcal vaccines.
Because of the limited number of invasive isolates, we included all documented cases of pneumococcal invasive disease in order to compare invasive vs. nasopharyngeal isolates. We may expect that most children not included in the DSS but seeking services provided by the Manhiça Hospital have been living in the area for some time and differences in terms of microbiological characteristics of the pneumococcal isolates are unlikely. Decisions regarding the introduction of conjugate vaccines in Africa depend not only upon licensure of pneumococcal vaccines with wide serotype coverage but also appreciation of disease burden and demand for the new vaccines. Regional data on disease burden are urgently needed, especially in view of promising results from South Africa (Klugman et al. 2003) and The Gambia (Cutts et al. 2005). Based on our findings, the nine-valent conjugate vaccine formulation would be expected to substantially reduce invasive pneumococcal disease among young children in Mozambique, and potentially in other parts of rural Africa, where the burden of pneumococcal disease is among the highest in the world.