Neonatal and Young Infant Sepsis in a Regional Hospital in Ghana

Background: Neonatal sepsis is a global health problem that mainly affects low- and middle-income countries. We have previously shown that early neonatal mortality is high at the Ho Teaching Hospital (HTH) of Ghana. We sought to determine the prevalence of neonatal sepsis, sepsis-related mortality, and bacterial species patterns in neonatal and young infant sepsis in this hospital. Methods: A hospital-based study was conducted in the hospital’s neonatal intensive care unit (NICU) from March to June 2018. Blood samples from 96 babies clinically diagnosed with or at risk of sepsis were cultured using the BACTEC 9050® machine. Clinical data including gravida, parity and antibiotic medication before delivery of mother and delivery type, gestation, birth weight and antibiotic medication status were collected for analysis. MALDI-TOF MS identified bacterial isolates, and their identities were confirmed via tuf gene sequence typing. The data were analyzed using GraphPad Prism 8.0.2. Results: Blood cultures were positive in 28 of the babies, with 14 and 12 representing early-onset and late-onset neonatal sepsis, respectively, and two cases of unknown sepsis type. Of the bacterial species that caused sepsis in the babies, coagulase-negative staphylococcus (CoNS) was the most prevalent isolate in 22 cases, followed by Klebsiella pneumoniae in two and Staphylococcus aureus, Streptococcus agalactiae, the Acinetobacter species, and Escherichia coli in the rest (one each). Of the CoNS, S. haemolyticus and S. epidermidis were the most prevalent species, found in eight and six cases, respectively. Thirteen neonates died, of whom seven had positive blood cultures, and two were referred. A case fatality rate of 7/26 was estimated. Neonatal mortality caused by Gram-negative bacterial infection was higher than that caused by Gram-positive bacteria. Conclusions: These data suggest a significant burden of sepsis among neonates and young infants and are associated with substantial morbidity and mortality at the HTH. There is a need to investigate risk factors associated with the increased sepsis rate in this hospital to inform measures to reduce the neonatal sepsis rate.


Introduction
Epidemiological estimates suggest that there were 1.7 million neonatal sepsis cases globally in 2010 and increased to 3.0 million cases in 2016 [1] [2]. Sepsis in the NICU remains one of the most significant causes of morbidity and mortality, especially for preterm newborns with very low birth weight and other congenital disorders requiring invasive therapy. Although neonatal sepsis can be community-acquired and hospital presenting, most neonatal sepsis cases are "hospital acquired" and "hospital presenting". Reports indicate that the frequency of healthcare-associated infections (HAIs) is at least two to three times higher in low-and middle-income countries (LMICs) and that infections cause 4% -56% of neonatal mortality in LMICs [3]. Another separate study also indicated that neonatal death accounted for 40% of total lives lost globally [4]. The Global Burden of Disease report listed neonatal sepsis as the fourth leading cause of death in infants. [5] Although neonatal morbidity and mortality caused by bloodstream infections also present significant problems in high-income countries (HICs), newborns in resource-limited countries are most severely affected [6]. In fact, in 2005, Lawn et al. estimated 99% of neonatal deaths in LMICs [7]. The proportion of this estimate attributed to bloodstream infection in neonates is mostly unknown for many developing countries. On the contrary, we have data from HICs on the deaths of neonates attributed to bloodstream infections. The current study has identified hospital-acquired neonatal infections and antimicrobial resistance as a major burden in Africa. The study recommends that more population-based neonatal infection studies and improved routine surveillance are needed to improve clinical care, plan health systems approach, and address antimicrobial resistance [8].
At the WHO's Sepsis Technical Experts meeting held in Switzerland in 2018, experts agreed that the current neonatal sepsis burden in LMICs is unclear. Thus, there exists a need to collect prospective clinical-based surveillance data on sepsis from LMICs [3]. Also, local data on the most common pathogens causing neonatal sepsis and their resistance patterns will help put measures in place to prevent and manage neonatal sepsis in LMICs [3]. Scientists consider this to be necessary because historical reviews have shown that the bacteriologic profile of microorganisms responsible for neonatal sepsis has evolved, underscoring the need for constant surveillance to identify predominant species. Knowledge of pathogens likely to cause sepsis at a particular setting serves to guide empiric therapy while awaiting culture and antimicrobial susceptibility results.
In LMICs, the prompt and accurate diagnosis of neonatal sepsis based on clinical and laboratory findings remains challenging. Some necessary laboratory tests-e.g. blood culture and adjunct tests such as haematological, C-reactive protein (CRP), and inflammatory markers-are not available in most hospital settings. The complex and dynamic disease pathophysiology often results in clinical signs and symptoms that are subtle and non-specific and overlap with non-infectious conditions [9] and other infectious diseases such as malaria and dengue [3]. Consequently, no consensus definition for neonatal sepsis exists [10]. However, any attempt to reduce neonatal mortality by reducing neonatal bloodstream infection will require local and regional knowledge on incidence, etiological patterns, and trends in neonatal infections.
In Ghana, neonatal mortality is a huge problem. The latest national survey conducted in 2011 estimated the country's prevalence of 32 deaths per 1000 live births, [11] with some regions having a higher prevalence than others.

Study Population, Participant Selection, and Recruitment
All the neonates admitted at the HTH's NICU and babies' unit during the study period totalled 186. Those with clinical diagnoses of sepsis (using the Hospital's guidelines for diagnosing neonatal sepsis) or at risk of sepsis (121 babies) qualified for a blood culture test (Figure 1). At risk of sepsis neonates includes those whose mothers had premature rupture of membranes (PROM), preterm rupture of membranes, and prolonged rupture of membranes. Also, premature babies with very low birth weights and those with prolonged (>7 days) use of central intravenous catheters were considered at risk of sepsis. Babies on antibiotic medication before admission were included. Babies older than 60 days were excluded. Three neonates died before an attempt was made to take their blood samples and were excluded. The convenience sample collection method was used. The clinicians referred the babies eligible for the study. A research protocol approved by the Research Ethics Committee of the University of Health and Allied Sciences, Ghana, was used by the principal investigator (PI) to obtain consent from the mothers. In cases where a patient has to be immediately managed with antibiotic initiation where the PI was not available (usually late in the night), the blood samples were put into culture bottles. Late consent was obtained within 24 hours. We have taken samples on all occasions because blood culture is a routine procedure for every hospital's suspected sepsis case. The blood samples were taken from 118 babies. However, only 96 babies whose mothers gave consent for their babies to participate in the study were enrolled and their data analyzed ( Figure 1). Figure 1. Flow chart for the study selection process and study results.

Study Procedure
Before enrolment, the study was explained to each mother by the PI, and her consent was obtained. In the case of mothers who declined, the blood culture Open Journal of Pediatrics tests were still conducted on a humanitarian basis. However, the results and bacterial isolates were not included in the study analyses and further laboratory analyses. Blood samples were collected from the babies (0.5 -2 ml) for culture and sensitivity testing. A butterfly syringe with a needle gauze size of 23G or a syringe with a 1 -5 ml barrel was used. The blood samples were collected by doctors who were assisted by trained nurses. These samples were collected using aseptic procedures from new venipuncture sites and inoculated directly into BACTEC Peds Plus/F (Becton Dickinson Company, Maryland, USA) blood culture bottles. The samples were transported from the ward immediately or within two hours of collection and were placed in the BACTEC TM 9050 blood culture instrument. The participants were assigned study identification numbers (IDs), which were used throughout the study. Information regarding the neonates and young infants (e.g. age, gender, delivery type, gestation period, birth weight, Apgar scores at 1 min and 5 min, antibiotics medication status) and their mothers (e.g. age, gravida, parity, number of antenatal visits, marital status, educational level) were collected from clinical notes. The babies' body temperatures were taken using electronic digital thermometers. Babies were followed from admission to discharge or decease.

Laboratory Test Processes
The inoculated blood culture bottles were loaded into the BACTEC TM 9050 instrument per the manufacturer's guidelines. The culture bottles were incubated at 35˚C with continuous agitation for a predetermined period for the maximum recovery of organisms. The positive culture bottles were sub-cultured onto 5% -10% sheep blood agar. Pure growth cultures were harvested and stored at −20˚C in labelled sterile screwed-cap vials until they were shipped to Germany for further analyses.
All the bacterial isolates were identified in the Department of Infectious Diseases and Microbiology of the University of Lübeck, Germany, using the MALDI-TOF Biotyper® (Bruker Daltonik, Massachusetts, USA). The bacterial isolates were revived from glycerol-preserved stocks by seeding them on a 5% -10% sheep blood agar plate and incubated at 37˚C for 24 hours or until visible growth was observed on the plate. The bacterial isolates were spotted from a single colony onto a MALDI-TOF MS 48-well target plate per the manufacturer's instructions and identified by the machine. The results were confirmed with tuf gene sequence typing as described by Hwang et al. [14]. Briefly, PCR amplification of the tuf gene was performed on a C1000 Touch TM Thermal Cycler (BioRad) by applying a set of primers 5'-GCCAGTTGAGGACGTATTCT-3' and 5'-CCATT TCAGTACCTTCTGGTAA-3', which amplify a 412 bp fragment of the tuf gene.

Ethics Approval
The

Statistical Analyses
Statistical analyses were done with Graph pad prism software, version 8.0.2. Data with missing values were cleaned. Frequencies of the various characteristics were determined and then converted to percentages.

Maternal Demographics
Data of 83 mothers were analyzed in this study. A summary of the mothers' demographics is listed in

Babies' Demographic, Blood Culture Results, and General Mortality
Data of 96 babies were analyzed in this study. The majority (81/95) of the babies were within the neonatal age (0 -24 days), with most of them 0 -7 days old (72), as shown in Table 2(a). A small proportion (14) of the babies admitted were infants.
The total positive blood cultures (28/96) are shown in Table 1, with 14 and 12 representing EONS and LONS, respectively. The general mortality among neonates and infants was 28 deaths out of the 186 admissions during the study period.
Two of the neonates were referred to a national teaching hospital, and therefore, there was no data on their admission outcomes. More (5/14) babies presented with EONS died compared with those with LONS (2/12), as shown in Table 2(b).  Data presented as frequency and percentages, N = Total numbers of subjects in category; CS = Caesarean section; SVD = Spontaneous vaginal delivery; ELBW = Extremely Low Birth Weight (<1000 g); VLBW = Very Low Birth Weight (>1000 <1500 g); LBW = Low Birth Weight (>1500 <2500 g); NBW = Normal Birth Weight (>2500 g); Macrosomia (>4000 g); EONS = early-onset neonatal sepsis (<3 days from birth); LONS = late-onset neonatal sepsis (>3 days from birth). Referred 0 1 N = 28, but two babies presented with LONS were referred to a national hospital.
Antibiotics had been administered to 22/96 of the neonates before their blood samples were taken. The data analyzed show that apart from one neonate born in the HTH, 21 neonates on antibiotics were referrals from the Volta Region districts. Table 3 shows the referral hospitals, blood culture results, and common antibiotics given to these babies before referral. The routine antibiotics used in the HTH for neonates and infants diagnosed with sepsis are the combination of either amikacin or gentamicin with flucloxacillin. These were the antibiotics used for the studied babies clinically diagnosed with sepsis. Metronidazole is given to babies with suspected anaerobic infections. Figure 2 shows the antibiotics supplied to the HTH's NICU, babies' wards and children ward from 2017 to 2019. More metronidazole was given to children within the age range 1 -6 years old admitted (clinically diagnosed or suspected with anaerobic infections) to the Children Ward than was given to neonates.   blood culture-proven sepsis was obtained for 13 of them. Meconium aspiration syndrome has been reported to be strongly associated with EONS [15]. In this study, 13/161 of the total diagnoses made were meconium aspiration syndrome.

Clinical Diagnoses of Neonates in the HTH NICU
Of this number, three were culture-proven sepsis. Considering these clinical diagnoses based on gestational type, the term babies were more specifically diagnosed (as shown in Table 4) than the preterm ones. The preterm babies were mostly diagnosed as premature and at risk of sepsis since their conditions were not clear-cut to make a diagnosis. In terms of the type of neonatal sepsis, with the exceptions of meningitis and sepsis, most of the babies presented with EONS were definitely diagnosed with various disease conditions than those with LONS.

Bacterial Species Isolation and Sepsis-Related Mortality among the Babies
Gram-positive bacterial species were more commonly isolated (24/28) than Gram-negative ones (4/28). Of the bacterial species that caused sepsis in the neonates, CoNS are the most prevalent isolates (22/28), followed by Klebsiella pneumoniae (2/28) and Staphylococcus aureus, Streptococcus agalactiae, Acinetobacter baumannii, and Escherichia coli (one each, as shown in Table 5). Of the CoNS, S. haemolyticus and S. epidermidis were the most represented species, with prevalence rates of 33% and 29%, respectively. Of note, other CoNS such as S. capitis, S. cohnii, S. hominis, and S. warneri were only presented in EONS

Discussion
We have previously determined neonatal mortality at the HTH of Ghana. We was 29% for our study and 22% for theirs. When the Ghanaians results are compared with that of a single-site hospital-based study performed in Nigeria, [17] which reported a 38% positive blood rate [17], it is evident that no difference between the positive blood culture rates reported from these two West African countries. A higher rate (78%) has been reported from an Ethiopian study [18]. Moreover, a South African study [19] has reported a relatively lower rate of 8%. However, the study was performed among babies presented with EONS only.
More Gram-positive bacterial isolates were cultured in this study. Although the sampling time was restricted to three months and our sample size was small, data presented herein suggests mortality caused by Gram-negative bacterial isolates appears to be higher (4/4) than that caused by Gram-positive bacteria (3/24).
However, it is a known fact and was buttressed by a previous study which suggested that neonatal bloodstream infection mortality caused by Gram-positive bacteria, especially CoNS, is lower than that caused by Gram-negative bacteria.
The former is positively associated with a prolonged hospital stay and increased hospital costs [33] [34] [35]. The exact mechanism(s) and pathophysiological differences among the bacterial species are not well understood. Our study, conducted in the HTH for just three months, suggests that the management of Gram-negative bacterial sepsis to improve neonates' conditions is challenging to the clinicians. To worsen the case for clinicians in this hospital, blood cultures are not routinely done due to the cost, which most clients cannot afford. However, blood culture is one of the essential investigations done in clinical microbiology laboratories. It has long been recognized as the "gold standard" for diagnosing bloodstream Infections. Not only will blood culture help in the isolation of offending pathogens, but it also allows susceptibility tests to be performed on isolates. Thus crucial decisions regarding antibiotics for managing patients with sepsis are based on blood culture results. Pediatrics in this hospital have to diagnose these neonates clinically and follow the hospital's protocols of treatment.

Conclusion
These data suggest a significant burden of sepsis among neonates and are associated with substantial morbidity and mortality at the HTH. CoNS were the frequent isolates seen in both EONS and LONS in this study. Despite the limited sample size, our data revealed that apart from S. epidermidis and S. haemolyticus, all other CoNS species isolated in this study were presented in EONS.
Moreover, this finding suggests more vertical transmissions than horizontal transmissions and called for further investigations to confirm the HTH's observation. We recommend that surveillance/clinical data at a regional level is important to allow comparisons and further studies stemming from limitations on current data.

Strength and Limitation of the Study
This study was the first attempt to understand the burden of neonatal deaths and the aetiology of those deaths in the HTH, where routine blood cultures are not done for neonates and infants diagnosed with sepsis. The limitation is that it was a single-centre study, performed for only three months and with a small