Bloodstream infection remains a major cause of morbidity and prolonged hospital stay in children . In neonates and infants, it is a life-threatening emergency, and therefore identifying the common bacteria and their susceptibility patterns will provide information necessary for timely intervention [25, 26]. Unfortunately there is limited information about the spectrum of pathogens causing infection and their antibiotic susceptibility in many under-resourced areas . Over the six months study period, this study showed a 25.9% positivity rate for isolated bacterial pathogens. This could most likely represent an under-estimation of the true prevalence of bacterial septicaemia as many of the infants and children cared for in NICU and the paediatric ward respectively received empirical treatment with antibiotics before blood was drawn for culture. Polage et al. iterated this fact in their study on laboratory use in Ghana where they concluded that physician perception regarding the value of diagnostic testing has been a major potential barrier to laboratory use resulting in empiricism and disproportionate antimicrobial administration. The estimated positivity rate for isolated bacterial pathogens from this study is however lower than rates found in University of Calabar Teaching Hospital (48.9%) and University of Port Harcourt Teaching Hospital (34.2%) in Nigeria [29, 30].
The spectrum of gram-negative bacteria responsible for neonatal sepsis in this study is similar to that reported by other authors [31–33] with the predominant isolates being Klebsiella and Escherichia coli. Klebsiella may be regarded as normal flora in many parts of the body. They colonize the skin, pharynx, and the gastrointestinal tract. They may, however, be associated with chorioamnionitis, urogenital infection or urinary tract infection in the mother. Neonates could therefore acquire these organisms before and/or during delivery from their mothers with infections at the time of delivery or shortly thereafter .
As reported in other studies [35, 36], the GPC were the predominant bacteria isolates in both early and late onset sepsis from NICU. The high proportions of CoNS isolated in both early onset (33.3%) and late onset (38.5%) at the NICU gives a suggestion of CoNS as an emerging pathogen in the unit . However, the diagnosis of CoNS in bloodstream infections is problematic. Generally, CoNS is not thought to be a cause of EOS, although it has been reported in studies reviewing organisms causing early infections [38, 39]. Conversely, CoNS is the predominant cause of LOS in low birth weight infants and infants who have indwelling catheters or other instrumentation [40, 41]. The isolation of CoNS in blood cultures is considered to be pathogenic if it is obtained from at least two separate sites, a process which may be difficult in infants . Therefore the combination of isolation of CoNS from the bloodstream with clinical symptoms of neonatal infection may be the best approach for determining pathogenicity of the CoNS [41, 42]. Since CoNS is the predominant bacteria colonizing the skin, it is very easy to contaminate the blood culture during the process of venipuncture. The optimal technique for obtaining blood cultures specifies povidone-iodine or chlorhexidine as skin-cleansing agents and venipuncture techniques that minimize contamination by skin flora . Generally, 70% alcohol serves as the major cleansing agent of the skin in resource-limited areas in view of perennial unavailability of povidone and chlorhexidine in both the NICU and paediatric wards. These factors suggest that the high proportion of the CoNS isolated is more likely to be contamination from the skin [26, 43].
Consistent with other studies [44, 45], Streptococcal species were recovered infrequently from neonates with early onset (EOS) in this study. There is an on-going debate about the importance of GBS infections in sub-Saharan Africa [46, 47]. Although most studies report that GBS infections are rare in Africa, studies in Malawi, Kenya, South Africa and Zimbabwe report that maternal colonization and EOS infections rates are similar to those reported from the United States and Europe . It has been hypothesized that the overwhelming nature of the infection in the neonate with early death, a predominantly out-born population, the infrequency of obtaining blood cultures in sick infants, and the fastidious culture requirements to identify GBS might interfere with its identification as a major pathogen in the neonatal period . Other authors suggest that the cause of EOS is gradually shifting from vertical transmission to acquisition of infection from immediate exposure to environment such as hands of health-care workers or unclean delivery practices [44, 45]. The yield of GBS from blood cultures is greatly influenced by many factors such as methodology of obtaining blood cultures, recent or intrapartum antibiotic use, and requirement for specialized media . Interpretation of results from this study should be done cautiously as most of these factors were not factored into the inclusion criteria per the design of the study.
In spite of the fact that late-onset (LOS) neonatal infection may occur through vertical transmission during passage through the birth canal leading to colonization and, later, to infection , the majority of the LOS in newborn infants is usually hospital-acquired infection . Although some of the blood cultures from babies greater than seven days of age were obtained from infants who had been admitted to the NICU immediately after birth and then became sick (representing hospital-acquired infection), many of the babies who had blood cultures drawn in the NICU were admitted directly from home or from the outpatient department after seven days of age. Infants, less than 30 days of age are also admitted to the NICU rather than the paediatric ward. Isolates from these admissions therefore, are more likely to represent community-acquired rather than hospital- associated late onset infections.
Other studies have reported non-typhoid Salmonella as the most frequent cause of bloodstream infection in older children [11, 51, 52] however the predominant Salmonella species isolated from the paediatric wards in this study was Salmonella typhi. It has been suggested that in addition to poor personal hygiene of caretakers and environmental contaminations, young children are more susceptible to Salmonella infection because they possess immature gut lymphoid tissues and have decreased gastric acidity [11, 53]. This finding underscores the need for increased awareness and knowledge about cleanliness and environmental sanitation for healthy living in the communities.
As observed from the current study, both the gram-positive (4% - 25%) and gram–negative (0% - 40%) bacteria showed very low sensitivity to Ampicillin and Cotrimoxazole. However the gram negative bacteria displayed high levels of susceptibility to the third-generation cephalosporins with the exception of Pseudomonas aeruginosa. Gentamicin was effective against Escherichia coli (80%), Acinetobacter (100%) and the CoNS (72%), but Klebsiella (33.3%), Pseudomonas (33.3%) and Coagulase positive Staphylococci (42.9%) were often resistant to Gentamicin.
The recommended empiric antibiotic therapy for EOS has been Ampicillin and an aminoglycoside, usually Gentamicin, while in LOS a third-generation cephalosporin in combination with Ampicillin is used [54, 55]. The noteworthy findings of this study are the susceptibilities of the GNR to Gentamicin and third-generation cephalosporins. Many authors report increasing GNR resistance to both Gentamicin and third-generation cephalosporins in developing countries which may be related to frequent and unrestricted use of third-generation cephalosporins [56–59]. It has been suggested that over-use of third-generation cephalosporins is also linked to increased risk of neonatal death , increased risk for necrotizing enterocolitis in low birth-weight infants  and increased risk for invasive candidiasis . Of particular interest are some Acinetobacter species that have emerged in Southeast Asia and Africa that are multi-drug resistant [63, 64]. Results from this study shows that the isolated GNRs are still fairly susceptible to third-generation cephalosporins. First-line empiric therapy for rule-out sepsis at the NICU in TTH currently utilizes a combination drug comprising Ampicillin and Gentamicin while the third-generation cephalosporins are only used very selectively, a practice that may reduce risk for multi-drug resistant bacterial strains.