Bacterial profile and drug resistance patterns in neonates admitted with sepsis to a tertiary teaching hospital in Ethiopia


 Background : Tackling neonatal infections and antibiotic resistance remains a challenge in low-income countries where neonatal mortality is high and antibiotic resistance is growing. The aim of this study was to identify bacterial etiologies and their drug resistance patterns in neonates admitted with diagnosis of sepsis to neonatal intensive care unit at Jimma Medical Center in Ethiopia. Methods : A prospective longitudinal study was conducted from April to October 2018. A total of 313 clinical specimens (211 blood and 102 cerebrospinal fluid) were processed for 238 neonates suspected to have sepsis. Blood culture was done using BD BACTEC FX40 automated system. Bacterial identification and antibiotic susceptibility testing were done according to standard microbiological techniques. Results : Bacterial etiologies were isolated from 62.1% (131/211) and 3.9% (4/102) of blood and cerebrospinal fluid cultures respectively. The predominant bacteria isolated from blood were Coagulase negative Staphylococcus 27.5% (36/131), followed by S. aureus 20.6% (27/131), Klebsiella species 14.5% (19/131), and Acinetobacter species 10.7% (14/131). Nearly 85% of the isolates were multidrug resistant, predominantly observed in Gram-negative bacteria. Among locally available antibiotics, only meropenem for Gram-negatives and vancomycin for Gram-positives were found to be largely effective. Conclusion : Bacterial pathogens identified in the study were highly resistant to available antibiotics. Thus, an effort to reduce neonatal mortality in the setting should focus on improving diagnostics for neonatal infection and containment of antimicrobial resistance.


Background
Neonatal infection remains one of the most common causes of morbidity and mortality in newborns with over one million deaths annually; 99% of these deaths occur in resource-limited countries [1,2]. Neonatal sepsis can be classified based on the time of occurrence as early-onset and late-onset sepsis depending on the age at onset. Early-onset neonatal sepsis (EONS) is defined by its occurrence with seven days of life and is commonly caused by organisms acquired from the mother's genital tract during delivery. The most common causative organisms are Group B Streptococcus, Escherichia coli and Klebsiella pneumoniae. Late-onset neonatal sepsis (LONS) occurs after 7 days of life and is usually caused by pathogens acquired during hospitalization or delivery. The most common causative organisms are coagulase-negative Staphylococcus (CoNS), Staphylococcus aureus, Enterococcus species, and Enterobacteriaceae [ 1,3,4].
It is well documented that neonates with sepsis have an increased risk of developing meningitis, which on its own would deteriorates the prognosis of such neonates [5,6]. Data from both high-and low-income settings have shown that meningitis affects 0.3 to 3% of neonates with early onset neonatal sepsis [7,8] and up to 30% of cases with LONS [6, [9][10][11]. Hence, there should be high index of suspicion for bacterial meningitis whenever neonatal sepsis is diagnosed, especially in cases of LONS [9,10].
Early diagnosis and treatment of neonatal sepsis are essential to prevent severe and life-threatening complications. However, accurate diagnosis of neonatal infection remains a challenge, particularly in resource-limited settings, due to variable and non-specific clinical features and the difficulty of obtaining infection markers in the early stage [12]. Poor diagnostic facilities and lack of skilled health workforce in such settings add further challenge to achieve the goal of reducing neonatal sepsis globally. In real-life clinical practice, treatment is additionally challenging due to the absence of susceptibility patterns of bacterial etiologies of sepsis and the lack of accurate diagnostic markers [13,14]. Antibiotic treatment is also increasingly complicated by limited options and the emergence of high level and more extensive antimicrobial resistance against the existing antibiotics in several developing countries [15].
Tackling neonatal sepsis and antibiotic resistance is extremely challenging in lowincome countries where neonatal mortality is high and antibiotic resistance is growing [16]. In these countries, isolation of etiologies of sepsis as well as determining drug susceptibility patterns of the organisms is difficult. As a result, overutilization of antibiotics is inevitable [17,18]. Consequently, multidrug resistant (MDR) organisms have emerged as important causes of neonatal sepsis in these settings in recent years. These extremely drug resistant pathogens include extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae [19][20][21], methicillin-resistant S.aureus (MRSA) [21,22], and MDR Acinetobacter baumannii [23], among other MDR organisms. Different studies have shown that a high prevalence of sepsis due to MDR bacterial pathogens in intensive care settings dramatically increases mortality of newborns, especially in developing countries [2, 20,24].
Essential data on the burden of severe bacterial infections in neonates and bacterial causes are scarce in Ethiopia, and the magnitude of antibiotic resistance remains unclear. Thus, a better understanding of the driving forces of bacterial sepsis and transmission is needed to help the fight against neonatal sepsis and antibiotic resistance. Hence, this study aimed to determine the profile and drug resistance patterns of bacteria causing sepsis in neonates admitted to a tertiary teaching hospital in Ethiopia.

Study design
A prospective longitudinal study was conducted from April to October 2018. All neonates admitted to NICU with clinical diagnosis of sepsis and investigated with blood and/or CSF culture during the study period were included in the study.

Data collection
A structured case reporting format was used to collect background data and clinical information. One sample of 1-3 milliliters venous blood was collected from the neonates by staff nurses. Additionally, 2-3ml of cerebrospinal fluid (CSF), in the absence of contraindications, was also collected by resident physician from neonates suspected to have sepsis. The specimens were collected following the principles of aseptic specimen collection and were immediately transported to the microbiology laboratory of JMC for processing and analysis.

Isolation and identification of pathogens
The blood specimens were inoculated into a BD BACTEC Peds Plus/F bottle. The bottles were then incubated in the BD BACTEC™ FX40 automated culture machine for a maximum of five days until they were flagged as "negative" or "positive" for growth. Positive bottles were taken out from the machine and sub-cultured on Blood, Chocolate, and MacConkey agar (Oxoid, England). The CSF specimens, on the other hand, were directly inoculated on Blood, Chocolate and MacConkey agar (Oxoid, England) plates within 30 minutes of collection.
The chocolate and Blood agar plates from both blood and CSF specimens were incubated in a candle jar to create a condition of 5-10% CO 2 . All the plates were incubated at 35-37 o C aerobically for 18-24 hours. After overnight incubation, all the inoculated plates were inspected and organisms grown on the plates were identified according to the standard microbiological identification techniques [25].

Statistical analysis
The data were entered using Epi Data version 3.1 and exported to SPSS version 25 and Microsoft excel for analysis. Descriptive statistics were used to show the frequency of bacterial agents and their drug resistance pattern.  (Table 1).    all of which were MDR. E. coli and Enterobacter spp. isolated in this group were also found to be MDR. Though the sample size of this group is too small to determine any statistical association, it is highly possible that MDR organisms might have contributed to neonatal mortality at the hospital. It is also important to note here that 81% (13/16) of the deaths in culture-confirmed cases occurred in patients identified with Gram-negative bacterial sepsis.

Discussion
In this study, blood culture was able to identify etiologic agents in 63.5% of neonates suspected to have sepsis. This proportion is higher than that in findings from recent similar studies in different parts of Ethiopia; 46.6% in Gondar [26], 44.7% in Addis Ababa [27], and 29.4% in Asella [28]. Our finding is also higher than in similar studies from other low-and middle-income countries; 49.7% in Tanzania [14], 43.4% in India [29], 16.9% in Nepal [30], and 12.2% in Iran [31].
The high rate of culture positivity in our study may be explained by various reasons.
First of all, the participants were neonates admitted to NICU only, excluding neonates in relatively stable medical conditions admitted to regular pediatric ward.
Secondly, we used a highly sensitive automated blood culture system (BD BACTEC™ Blood Culture) [32] unlike in most of the other studies mentioned above where manual systems were used. Thirdly, the fact that we used only one blood specimen for culture, contamination might have also contributed to increased culture positivity. Nevertheless, in such low-income setting where studies on neonatal infections are limited, such a finding may also be reflecting a reality on the ground, since sample taking was part of routine clinical procedures and not investigator initiated. However, involved staffs do undergo regular trainings on appropriate and sterile sample-taking techniques.
Group B Streptococcus, E. coli and K. pneumoniae in EONS and CoNS, S. aureus, and Enterococcus spp. in LONS are major cause of neonatal sepsis [1,3]. In our study however, the common etiologies identified were CoNS, S. aureus, Acinetobacter spp., Klebsiella spp. and Citrobacter spp. These bacteria are mostly acquired from the hospital environments, health professionals, and medical devices [33,34].
Moreover, as majority of our study participants (84.1%) had early onset neonatal sepsis, the bacterial etiologies identified in our study somehow deviate from expectations [1,3,4]. One of the possible explanations is that about 68% of the neonates were delivered at the same hospital where high rate of hospital acquired infection by these organisms has been documented [35][36][37]. On the other hand, recent similar studies in Ethiopia have revealed that CoNS, S. aureus and Klebsiella spp. are major cause of neonatal sepsis in the country [26][27][28]. These evidences suggest that in settings where infection prevention and control practices are not properly implemented [38,39]; infections by these agents constitute a significant threat to neonates and other vulnerable populations.
Concerning to drug resistance, the isolates in the present study were resistant against commonly used antibiotics in the area and high frequency of MDR bacterial pathogens were observed predominantly among the Gram-negative bacteria. Recent studies in Ethiopia, India, Nepal, and China also showed high levels of resistance in Gram-negative organisms against ampicillin (85%, 78%, 100%, 80%), respectively, and ceftriaxone (57%, 100%, 100%, 50%), respectively [26,29,30,40]. Low level of resistance was observed against meropenem (17%) in Gram-negative bacteria and vancomycin (7%) in Gram-positive bacteria. A potential explanation is that these two antibiotics are expensive and not widely available in the setting studied and hence less frequently prescribed. Among the older antibiotics, only ciprofloxacin (48% resistance) and clindamycin (35% resistance) showed better in vitro efficacy against Gram-negative and Gram-positive bacteria respectively.
The MDR patterns in both Gram-negative and Gram-positive bacteria in our study (84.4%) is by far higher than national and global reports: Ethiopia (65%) [26], Jordan (69%) [23], and China (50%) [40]. Moreover, the high rates of MRSA and Acinetobacter spp. resistant to almost all antibiotics available in the setting are serious concerns for the healthcare system in the region and beyond. Unavailability of microbiologic diagnostic facilities, poor infection prevention and control practices, lack of antimicrobial quality control, lack of control on antimicrobial utilization and absence of proper antimicrobial resistance surveillance system in the country and at the hospital in particular might have contributed to this alarming rate of MDR pattern.
The proportion of culture confirmed meningitis in neonates with sepsis is only 4%.
As most of the neonates had EONS, the finding is concordant with existing global evidences [7,8]. However, this finding cannot be generalized to all neonates admitted to the hospital and other settings in Ethiopia as our study has excluded neonates admitted to regular pediatric wards.
Our study is one of few studies on etiologies and AMR patterns in neonatal sepsis in Ethiopia. We believe that the findings in this study complement what has already been done and could provide important data for policy-level intervention to tackle neonatal mortality and antimicrobial resistance. However, the fact that our study is limited to one facility, a tertiary teaching hospital, and that only neonates admitted to NICU were included may limit the generalizability of the findings. Moreover, only one blood sample was taken from the neonates for culture, contamination during specimen collection might have contributed to the high rate of culture positivity in our study.

Conclusions
This study has shown that MDR CoNS, S. aureus, and Gram-negative bacilli (Klebsiella, Acinetobacter, and Citrobacter spp.) were the leading causes of neonatal sepsis at the hospital. The emergence of MDR pathogens in the country, including resistance to reserve antimicrobials and the unmet target in reduction of neonatal mortality, deserve intervention at policy-level. This should be done through development of locally acceptable and applicable guidelines, adhering to standard evidenced based practices, commitment to rational use of antimicrobials, improving diagnostic facilities and implementing, nationwide AMR surveillance system. Future research should also be carried out focusing on identification of appropriate local empiric therapies with improved susceptibility profiles for timely and effective treatment.

Funding
The study was funded by the Center for International Health (CIH), German Academic Exchange Service (DAAD) through its Exceed program and the German Ministry of Economic Cooperation and Development. The funder had no any role in the design of the study and collection, analysis, and interpretation of data and in developing the manuscript.

Authors' contributions
EKG, SA, MB, MG, GKB, AW and GF conceptualized and designed the study protocol.
BE, MG, MB, EKG, SA, NW and GT coordinated and supervised data collection. MG and EKG drafted the manuscript and all the authors critically reviewed and approved the final manuscript in its current form and agreed to be accountable for all aspects of the work. Figure 1 Etiologies identified from blood culture in neonates admitted with sepsis to neonatal intensiv Distribution of MDR and non-MDR patterns of bacterial isolates in neonates admitted with sep