To estimate the prevalence of antibodies against Coxiella burnetii (Q fever) among children in eight villages in The Gambia, West Africa.
To estimate the prevalence of antibodies against Coxiella burnetii (Q fever) among children in eight villages in The Gambia, West Africa.
Sera of 796 children aged 1–15 years were tested for presence of antibodies against phase II of C. burnetii by ELISA.
IgG and/or IgM phase II antibodies against C. burnetii were detectable in 8.3% (66/796) of all serum samples analysed with significant differences in seroprevalence between villages. Highest prevalence was found in the age group 1–4 years.
Exposure to C. burnetii is considerable in the early years of life in The Gambia, and further studies are warranted to estimate the role of Q fever in acute febrile illness in The Gambia and elsewhere in Africa.
Estimer la prévalence des anticorps contre Coxiella burnetii (fièvre Q) chez les enfants de 8 villages de la Gambie, en Afrique de l’Ouest.
Les échantillons de sérum de 796 enfants âgés de 1 à 15 ans ont été testés par ELISA pour la présence d'anticorps contre la phase II de C. burnetii.
Les anticorps IgG et/ou IgM de phase II contre C. burnetii étaient détectables dans 8,3% (66/796) de tous les échantillons de sérum analysés avec des différences significatives dans la séroprévalence entre les villages. La plus haute prévalence a été observée dans le groupe d’âge de 1 à 4 ans.
L'exposition à C. burnetii est considérable dans les premières années de la vie en Gambie et des études supplémentaires sont nécessaires pour évaluer le rôle de la fièvre Q dans la maladie fébrile aiguë en Gambie et ailleurs en Afrique.
Calcular la prevalencia de anticuerpos frente a Coxiella burnetii (fiebre Q) en niños pertenecientes a 8 poblados en Gambia, África Occidental.
Mediante un ELISA en suero se determinó la presencia de anticuerpos frente a la fase II de C. burnetii en 796 niños con edades entre 1–15 años.
Se detectaron anticuerpos IgG y/o IgM en fase II frente C. burnetii en 8.3% (66/796) de todas las muestras de suero analizadas con diferencias significativas en la seroprevalencia entre poblados. La mayor prevalencia se encontró en el grupo de edad de 1–4 años.
La exposición a C. burnetii es considerable en los primeros años de vida en Gambia y se necesitarían más estudios para calcular el papel de la fiebre Q entre la enfermedad febril en Gambia así como en otros lugares de África.
Coxiella burnetii, the causative organism of Q fever, is found worldwide in domesticated and wild animals; infected sheep and goats are considered the primary sources of Q fever in humans. Infection in humans occurs by inhalation of contaminated aerosols and presents as subclinical seroconversion, acute febrile illness, pneumonia or hepatitis. Approximately 2% of acute Q fever patients develop chronic Q fever, a potentially lethal disease with endocarditis as the main presentation (ECDC 2010). A recent large epidemic of Q fever in the Netherlands, associated with goat farming, caused interest for Q fever as an emerging zoonotic disease (Dijkstra et al. 2012). Few reports of Q fever in children exist, suggesting that infections during childhood often remain asymptomatic (Maltezou & Raoult 2002), but one can also assume that Q fever diagnostic tests are seldom carried out in febrile children, especially in resource-poor settings. Rapid, presumptive treatment of febrile illness with antimalarial drugs has long been considered a key strategy for reducing child mortality in sub-Saharan Africa. However, non-malarial fevers now constitute the major proportion of febrile episodes in Africa (Gething et al. 2010). Reyburn et al. (2004) showed that presumptively treated malaria slide-negative patients had a higher mortality than did slide-positive patients. This could partly be due to patients with bacterial infections being treated with antimalarial drugs instead of antibacterial drugs. Insight into non-malarial causes of fever is therefore needed.
There is limited information on Q fever in Africa but in many rural areas, there is close contact of people with animals, which may facilitate transmission of C. burnetii from animal to man. In Ghana, IgG antibodies against phase II of C. burnetii (indicating past exposure) were observed in 17% of 219 2-year-old children and 9% of 158 adults (Kobbe et al. 2008). In Mali, 40% of 156 mainly adult febrile patients had antibodies against C. burnetii, with 10% of positives having a serological profile suggesting acute infection (Steinmann et al. 2005). A recent study in Tanzania among hospitalised febrile patients showed that 13.5% had acute Q fever or rickettsial infection (Prabhu et al. 2011).
To serve as a pilot study for a larger project assessing causes of fever in Africa, we tested banked serum samples from children in The Gambia for antibodies against C. burnetii in order to gain insight into the prevalence of C. burnetii infection in The Gambia.
We tested 796 serum samples that were collected in a malaria study in July/August 2008 from children aged 1–15 years, after informed consent. Children were recruited from a cluster of 8-rural villages situated 5–15 km west of Farafenni in the North Bank Region of The Gambia. The design of the malaria study has been described before (Ceesay et al. 2010). In brief, study villages were part of the Farafenni demographic surveillance system. All subjects aged 1–15 years were eligible for enrolment. Recruitment was conducted at cross sectional surveys lasting 2–3 days in each study village during which eligible participants were invited to join the study. Overall, 36% (800/2228) of children aged 1–15 years were enrolled into the study across the eight villages. Blood samples from the children were collected at baseline and microscopically examined for malaria parasites. Then, study nurses followed up the health status of the cohort for 6 months to record any newly occurring episodes of fever. An appropriate course of antimalarial therapy was administered if subjects became malaria positive during follow-up. Remaining stored serum samples from the baseline malaria screening were used for the antibody study described in this manuscript.
For detection of IgG and IgM antibodies to C. burnetii phase II antigen, a commercial ELISA was used (Serion ELISA classic, Virion/Serion, Würzburg, Germany). IgM phase II antibodies were measured qualitatively, and samples with an optical density (OD) of >10% above OD cut-off were scored as positive; those with ODs of >10% below OD cut-off were considered as negative; in between samples were denoted as borderline and were retested. IgG antibodies were measured quantitatively and results generated from standard curves were reported in International Units/ml. In line with manufacturer's recommendations, samples with values of <20 IU/ml were considered negative, values of 20–30 IU/ml were scored as borderline and testing was repeated. Those that had values of >30 IU/ml were considered as positives.
Using identical kits, The Medical Research Council (MRC) laboratories in The Gambia and the National Institute for Public Health and the Environment in the Netherlands (RIVM) initially tested a set of 31 shared, anonymised samples in both laboratories, including negative samples, acute infections, follow-up samples, past infections and false-positive infections. This interlaboratory quality control showed high correlation of both IgM phase II (R2 = 0.923, P < 0.001) and IgG phase II (R2 = 0.862, P < 0.001) ELISA test results. Thereafter, the 796 study samples were tested at the MRC laboratories in duplicate.
An anonymised data file containing basic demographic data was extracted from the baseline malaria study dataset and linked to serological ELISA results based on an anonymous identifier. The study was reviewed and approved by the MRC The Gambia Scientific Coordinating Committee and The Gambia Government and MRC Joint Ethics Committee.
From the original 800 samples, 796 were available for the present study. There were 389 (48.9%) samples from boys and 407 (51.1%) samples from girls with a mean age of 6.4 years (Table 1). At the time of the baseline survey in 2008, 25 (3.1%) children tested positive for malaria parasites. During the follow-up period in the 2008 malaria season, 192 of 796 children (24.1%) were recorded with febrile illness. Malaria test results were available for 171 children with fever, of whom 20 (11.7%) were positive and 151 (88.3%) negative.
|Variable||Total population||IgG positive||IgG+IgM positive||IgM positive||Total positive|
|n (%)||n||n||n||Prevalence% (95% CI)|
|Age group (years)|
|1–4||299 (37.6)||28||1||2||10.4 (7.3–14.2)|
|5–9||308 (38.7)||20||2||0||7.1 (4.6–10.4)|
|10–15||189 (23.7)||9||0||4||6.9 (3.9–11.2)|
|Male||389 (48.9)||29||3||4||9.3 (6.7–12.4)|
|Female||407 (51.1)||28||0||2||7.4 (5.1–10.2)|
|ContehSokoto||106 (13.3)||2||0||0||1.9 (0.3–6.1)|
|Alkalikunda||118 (14.8)||3||0||3||5.1 (2.1–10.3)|
|Jumansareba||106 (13.3)||5||1||0||5.7 (2.3–11.4)|
|ContehNicci||121 (15.2)||9||0||1||8.3 (4.3–14.2)|
|Daru||72 (9.0)||6||0||0||8.3 (3.4–16.5)|
|Yallal||100 (12.6)||10||1||1||12.0 (6.7–19.5)|
|Jarjari||103 (12.9)||12||1||1||13.6 (8.0–21.3)|
|India||70 (8.8)||10||0||0||14.3 (7.5–24.0)|
|Presence of malaria parasites|
|No||771 (96.9)||55||3||6||8.3 (6.5–10.4)|
|Yes||25 (3.1)||2||0||0||8.0 (1.4–24.0)|
|Fever at follow-up|
|No||604 (75.9)||44||2||5||8.4 (6.4–10.9)|
|Yes||192 (24.1)||13||1||1||7.8 (4.6–12.3)|
Sera of 66 children tested positive for Q fever IgM and/or IgG antibodies giving an overall unadjusted, apparent prevalence of 8.3% (95% confidence interval (CI): 6.5%–10.4%). Of these 66 children, 36 (54.5%) were males and 30 (45.5%) were females (χ2 = 0.928, P = 0.335). Of the positives, 57 (86.4%) had IgG phase II antibodies only with a median concentration of 46.5 IU/ml (interquartile range 38.1–76.5). Six (9.1%) were positive for IgM phase II antibodies only. Three samples (4.5%) tested positive for both IgG and IgM antibodies, suggesting a recent C. burnetii infection. These three children, a girl of 2-year old and two boys of 5- and 6-year old, had negative malaria microscopy. IgG phase II antibody concentrations were 1071.7, 1000.0 and 50.4 IU/ml respectively. Unfortunately, clinical details were unavailable.
Seroprevalence in the age group 1–4 years (10.4%) was higher than the prevalence for the combined age groups 5–9 and 10–15 years (χ2 = 2.715, P = 0.050) (Table 1). There were significant differences between the eight villages with prevalence figures ranging from 1.9% (95% CI: 0.3–6.1) to 14.3% (95% CI: 7.5–24.0).
Our study showed considerable exposure to C. burnetii among young children in The Gambia with large differences in seroprevalence between nearby villages. Only basic demographic data of the children were available, and this limited the possibility for a proper risk factor analysis. However, it is remarkable that the highest prevalence of Q fever antibodies was observed among children aged 1–4 years. This suggests that C. burnetii infections are more prevalent among young children than generally assumed and that Q fever may be an underestimated cause of fever in small children. This raises diagnostic and therapeutic challenges for management of febrile illness because the paradigm in much of sub-Saharan Africa is that patients with febrile illness receive presumptive treatment with Artemisinin-based combination therapy (ACT), the currently recommended first line antimalarial drug. Studies in Australia (Parker et al. 2010), the Netherlands (Schimmer et al. 2012) and Spain (Cardeñosa et al. 2006) show increasing prevalence with increasing age. In contrast, in Mali (Steinmann et al. 2005), Zambia (Okabayashi et al. 1999) and an earlier study in the Netherlands (Richardus et al. 1987), seroprevalences in children and adults were not very different. However, it is difficult to compare seroprevalence estimates between studies because of differences in study populations, exposure, serological tests and cut-off values.
It is not clear why seroprevalence should be lower in the older children. One would expect persistent exposure in the natural environment with boosting of antibodies. However, it is common practice that goats and sheep are kept close to or sometimes in the house at night. This might create a certain level of contamination of the home environment if an animal is infected with C. burnetii. Pre-school children remain in and around the home for a large part of the day and thus experience higher exposure than the school-age children. Furthermore, consumption of raw goat milk by young children could play a role. Studies focusing on age-specific human–animal contact patterns and milk consumption patterns could shed light on this issue. Little information is available on the role of waning of antibodies against C. burnetii over time but halftimes could be several years (Teunis et al. 2013).
The variation of exposure between villages of residence might also be related to differences in human–animal contact patterns and density of small ruminants present in the village. No data were available on density of animals and infection status in the villages at the time of the survey. However, in 2012, a veterinary survey in the eight villages of the present study showed a seroprevalence in goats and sheep of 99/566 (17.5%) (International Trypanotolerance Centre, The Gambia, unpublished data). In order to confirm the hypothesis of transmission depending on animal density an integrated human–veterinary study to elucidate the sources of infection of humans would need to be carried out.
Our interlaboratory quality control showed high concordance of ELISA results. However, a common problem with the interpretation of serological surveys is that the tests are imperfect and the results provide only the apparent, not the true, seroprevalence. Specificity of the Serion ELISA that we used is close to 100%. Sensitivity strongly depends on the immunofluorescence (IF) titre that is used as the ‘gold standard’. With an IF cut-off of 1:64, sensitivity of the Serion ELISA has been reported as 82% for IgM II and 59% for IgG II (Blaauw et al. 2012). It is therefore likely that real seroprevalence is higher than we estimated with ELISA.
During follow-up of the children who were included in the present study, only 11.7% of those that developed fever had a positive test for malaria. This underscores the need for improved diagnosis in febrile patients to allow appropriate treatment for patients with bacterial infections and prevent complications, such as chronic infection. This would help reduce the costs of supplying unnecessary drugs for the management of malaria and reduce the risk of increased drug pressure leading to emergence and spread of antimalarial drug resistance. This not only applies to Q fever, but also to other underdiagnosed and neglected bacterial infections such as brucellosis, leptospirosis and rickettsiosis.
We thank Professor Umberto d'Alessandro for helpful advice with the study proposal and Samba Sowe for excellent support with laboratory assays. We gratefully acknowledge Drs Natalia Gomez-Escobar and Sanie Sesay's participation in the parent study from which samples analysed here were obtained. Professor David Conway provided information on the methodology of the 2008 malaria study from which the study samples were obtained.