Sub Saharan Africa contributes significantly to the global mortality of children aged less than 5 years with mortality rates of 100-250 per 1000 [1]. It is now well established that invasive bacteria disease is the leading cause of childhood mortality in sub-Saharan Africa and most of these infections are preventable by the use of vaccines which are already licensed and in routine use in most developed countries. Few health facilities in Africa have the microbiologic laboratories with capacity to identify invasive bacterial infection. Thus the clinical importance of bacterial infections and associated mortality is often not recognized with most febrile illnesses being attributed to malaria. Randomized controlled efficacy trials in Gambia [2, 3]. and South Africa [4] and effectiveness studies in Malawi [5] uncovered substantial disease burden from Haemophilus influenzae type b (Hib) and Streptococcus pneumoniae, while population based surveillance studies in Kenya have demonstrated substantial disease caused by S. pneumonia [6]. While S. pneumoniae has been identified as the leading cause of bacteremia or invasive bacteria disease in The Gambia [7], Kenya [6], Mozambique [8], this was not the case in Ghana [9] and Malawi [10] where non-typhi Salmonella (NTS) predominated. Availability of national epidemiologic data will strengthen advocacy for the introduction of relevant vaccines.

Although Nigeria is the most populated country in the sub region and records high infant and under-5 mortality rates of 97 and 189 per 1000 respectively [11], data on the etiologic agents of invasive bacteria disease in children are sparse.

Primary health care system is poorly developed in the country and chemists, pharmacies, private clinics and hospitals are often the first port of call for parents/guardians of children seeking care for their children. Over-the-counter antibiotic use is very prevalent and this is likely to impact the outcome of any studies aimed at the determination of the causes of bacteria infection in children. It is possible that this healthcare seeking behavior may of itself modify the spectrum of prevalent bacteria pathogens and also the spectrum of clinical disease presentation. Previous reports of causes of bacteremia from different parts of Nigeria utilized conventional blood culture techniques with the use of human blood for the preparation of blood agar and suboptimal methods for the detection of fastidious organisms like S.pneumoniae. Limitations of this approach have been reviewed [12]. Against this background, we sought to introduce surveillance blood cultures using an automated system, providing access to the facility from 7 hospitals and clinics in the Abuja area. Using data arising from this pilot surveillance, we report the commonly identified causes of community-acquired bacteremia in children aged 2 months-5 years in an urban and semi-urban setting with reference to vaccine-preventable diseases.

From September 2008 to November 2009, children who presented with an acute febrile illness at the enrolling facilities were screened and blood sample collected for bacterial culture. 1,287 children were screened and enrolled, of these 227 children were outside the age range and 91 subjects had incomplete clinical data. These were excluded from further analysis. Thus 969 children aged between 2 months -5 years with a median age of 17 months (+/- 13.9 months) constitute the subject of this pilot report.

Bacteria was isolated from the blood stream of 204 children and of these, 93 were considered skin contaminants and not clinically relevant. Thus significant bacteremia occurred in 111/969 (11.5%) children.

The etiology of bacteremia is summarized in Figure 1. Salmonella spp were the leading cause of bacteremia (28.5%) with S.typhi predominating at 20.9% and non-typhi salmonella (7.6%). S. aureus accounted for 20.9% while Acinetobacter and S.pneumoniae accounted for 12.3% and 7.6% respectively. In five subjects alpha hemolytic streptococcus was isolated and considered clinically significant as these children had underlying congenital cardiac disease.

Concordance between results from the site laboratory and the reference lab was 100% for non-typhi Salmonellae (NTS), Candida, Escherichia coli and Pseudomonas. The concordance was lowest for alpha hemolytic streptococcus as a number of these were initially mis-identified as S. pneumoniae. The identification assigned by the reference laboratory was used for final analysis.

The two Haemophilus influenzae isolates were not recovered following primary storage and so we were unable to type these organisms. Of the eight confirmed S.pneumoniae isolates, three were 19F; two were 23F and one each of serotypes 1, 5 and 6B.

Antimicrobial activity in serum

Approximately 10% of samples from each enrollment site were randomly selected for testing. Prevalence of serum antimicrobial activity was highest in children who presented to the National Hospital (88.9%), the major tertiary referral center in Abuja and was lowest in those who presented to Zankli Medical Center (40%) and Nyanya General Hospital (45%). Data summarized in Figure 2.

In all, one hundred twenty six sera samples were tested for antimicrobial activity (Table 1). Of these, forty eight tested negative and seventy eight were positive. Of the samples that tested negative, forty six were negative for bacterial growth and two tested positive for bacterial growth (S. pneumoniae 1 and coagulase negative staphylococcus). Of the samples that tested positive for antimicrobial activity, eight had positive bacterial culture (coagulase negative Staphylococcus 5, E.coli 1, S.typhi -1 and S. pneumoniae-1).

Bacterial Culture Result	Negative Antimicrobial Activity	Positive Antimicrobial Activity
Culture Negative	46	70
Culture Positive	CNS*-1 S. pneumoniae- 1	E.coli - 1 CNS- 5 S.pneumoniae- 1 S.typhi- 1

CNS*- Coagulase negative Staphylococcus

This pilot study has demonstrated the public health significance of etiologic diagnosis of acute febrile illness in a setting where clinical malaria is over-diagnosed, non-prescription antibiotics are prevalent and empiric prescription antibiotic by physicians is the standard clinical practice. This pilot study was sited at Abuja as it is the federal capital city and it is strategically located for access to both urban and rural populations.

The spectrum of agents of bacteremia from this pilot study differs significantly from reports from the Gambia from a similar age group of children [7]. In our cohort, S. aureus and S. typhi were the leading cause. While in the Gambia, the pneumococcus is the leading cause of bacteremia and non-typhi salmonella is much more prevalent without any cases of S. typhi. The relatively less significant role of pneumococcal bacteremia is likely to be a significant epidemiologic finding secondary to the high incidence of non-prescription antibiotic use. This is most likely, given that all our pneumococcal isolates are penicillin susceptible and the penicillins are the most readily available oral antibiotics in this setting. It is unlikely that we missed clinical isolates, since we utilized a culture system similar to that which was used in other studies in Africa which have reported S. pneumoniae as the leading cause of bacteremia in this age group. The same explanation may also hold true for the scarcity of H. influenzae from blood cultures. Although H. influenzae type b (Hib) vaccine is not included in the National Immunization Program, it is often available in most private hospitals and offered to those who can afford it. However, it is not likely that it would be that widely used to have impacted disease incidence in the wider community. The relative burden of meningitis caused by these two pathogens may be more significant. More intense surveillance for meningitis is planned in the next phase of this surveillance.

In a recent study from Ibadan [16] which utilized conventional culture methods, of 24 cases of definite meningitis 9 were caused by the pneumococcus and 11 by Hib only 4 were caused by Klebsiella spp; among pneumonia cases 9 were caused by pneumococcus and 2 by Hib. Interestingly the bacteria isolates from bacteremia cases are similar to those in the current report. So the spectrum of the bacteria recorded in the Abuja project may be related to the prevalent disease spectrum. Also a recent project undertaken just south of Ibadan indicates high pneumococcal carriage rates (90%) in young infants aged < 9 months. (Adetifa I and Adegbola RA- personal communication). This is similar to the carriage rates in The Gambian villages. There is no reason to believe that in a well conducted population study the pneumococcus and Hib may be as important in Nigeria as in The Gambia or Kenya as causes of pneumonia and meningitis.

There may be a distinct difference in disease epidemiology due to the multiple slum settlements on the outskirt of Abuja which is more readily affordable to the general population than Abuja city. Sanitary conditions in these settlements are usually poor or non-existent. The relatively high incidence of S.typhi is very concerning and is an important public health issue. Currently, no country in sub Saharan Africa implements routine immunization with typhoid vaccine and reports from several countries suggest that Non-typhi salmonella may be a more important problem than S.typhi and there are efforts towards developing vaccines for NTS, while the currently available typhoid vaccines remain unused due to lack of convincing disease burden evidence. Our pilot data suggest that, at least in this setting there may well be a need for implementing routine immunization with typhoid vaccine. It is noteworthy that the current age spectrum included in this study did not include school age children 5-14 yrs, a subgroup that often has a higher incidence of typhoid fever [17]; thus the observation in our report is likely to be an underestimate.

Although S, pneumoniae did not feature as the leading cause of bacteremia, it accounts for 13% of all the deaths observed in this pilot study. The study from Ibadan [16] reported over 80% mortality with pneumococcal meningitis in children and only in one study has there been the observation of delayed mortality following invasive pneumococal infection. Whether these delayed deaths are associated with underlying conditions such as HIV, SCD or other immune defects have not been formally investigated. Nigeria is yet to implement routine immunization with the pneumococcal conjugate vaccine. The limited number of invasive pneumococcal isolates that were available from this pilot study suggests that 75% of invasive pneumococcal disease may be preventable with the 7-valent and 100% with the new 13-valent pneumococcal conjugate vaccine.

The rate of antibiotic use, as demonstrated by detection of residual antimicrobial activity in serum, in the community is high in most of the facilities and lowest at ZMC, private hospital. The clientele at this facility are at a higher socioeconomic status than those presenting to the government facilities and are therefore more likely to present early to a private health care provider. This is an interesting observation that warrants further evaluation since on a large scale this may well impact the pattern of etiologic agents identified in children who present to such a facility as opposed to Government establishments. Future surveillance will incorporate enrolment of children at private clinics that are staffed by trained pediatricians.

We have in the course of this pilot study provided increased awareness for physicians and health care providers on the need to carefully consider other diagnosis in children who present with an acute febrile illness, other than malaria and to seek etiologic diagnosis that will properly guide management, particularly, given the increasing reports of the decreasing incidence of malaria in the sub region¹⁸. We have also provided additional laboratory skills for the identification of bacteria for laboratory technicians and more importantly, we believe providing this diagnostic service will improve the quality of care and improve clinical outcome.

The etiologic agents and their susceptibility patterns observed in this pilot study and other recent studies raises concerns about the appropriateness of the current treatment guidelines recommended by the Integrated Management of Childhood Illness (IMCI). The first line treatment for most conditions with co-trimoxazole or the use of chloramphenicol for severe pneumonia requires revision given the high prevalence of resistance among the commonly isolated bacteria.

In setting up this pilot surveillance, we encountered a number of challenges which are worth mentioning briefly as these will inform the design of larger studies. First, we observed a high contamination rate due largely to the fact that physicians were not conversant with aseptic blood draw techniques in young children and we also had a rapid turnover of house physicians. Second, although motivation of personnel to participate in the study was initially strong, this waned over time as reflected in the decreased enrolment rate at the different sites. Third, a large number of the study questionnaire and patient sampling for cerebrospinal fluid in suspected cases of meningitis was incomplete. There was a general impression amongst the clinical staff that this clinical research project, while beneficial for patient management, increased their workload. These obstacles will be overcome in future studies which will aim to have dedicated salaried personnel for participant enrolment and obtaining relevant clinical specimens. This study has several limitations. Three of the most important include first, our inability to define a population census from where our participants were drawn and therefore unable to define incidence of bacteremia. While this was not the aim of this pilot study, clearly incidence figures would be very useful. However, this approach will be a challenge given the number of healthcare outlets available to the population. Second, we were only able to evaluate the role of underlying conditions such as SCD and HIV only in those who have an established diagnosis. Ideally we should have tested all participants for this condition to determine the role of these conditions in promoting the risk for specific bacteremia. This approach is planned in the next phase of our surveillance. Third, we were unable to provide systematic subject enrollment and this could have introduced some bias in sampling.

In conclusion, initial findings from the introduction of an automatic blood culture system in Abuja suggest that the spectrum of agents associated with community-acquired bacteremia in young children in central Nigeria may differ from that observed in most other African countries with a predominance of S.typhi. This may be due to environmental factors, antibiotic usage, patient selection or technical factors, and epidemiological studies are urgently needed. While this may in part be due to the multiple new settlements and satellite towns around Abuja which in general have poor water supply and sanitary facilities, there is also a high level of non-prescription antibiotic use.

There is a need to intensify surveillance at both rural and urban settlements, expand population surveillance to include school age children and also to determine if this pattern of infection is similar at other geographic locations in the country. A detailed surveillance study for the determination of the incidence of typhoid fever is being planned. Such surveillance activities will be critical for informing the need for introduction of typhoid fever vaccines in the medium term and improvement of infrastructures in the long term. This expanded surveillance should also provide additional data on the disease burden caused by S. pneumoniae which appears currently confounded by widespread antibiotic use.