Neonatal sepsis in Sana’a city, Yemen: a predominance of Burkholderia cepacia

Background Neonatal sepsis is a global concern with increasing morbidity and mortality. The burden of neonatal sepsis is highest in developing countries, especially in those lacking proper surveillance systems. The causative pathogens and their drug-resistance levels vary between countries with emergence of multidrug resistance organisms. Thus, accurate records on the recent trends of organisms causing neonatal sepsis will provide vital information for appropriate intervention. We aimed to investigate neonatal sepsis, identify its associated factors and causative pathogens and to assess the antibiotic susceptibility patterns in Sana’a city, Yemen. Methods A cross-sectional study was conducted on neonates admitted to intensive care units of six hospitals in Sana’a city, Yemen, in the period from January 15, to March 30, 2020. Natal and prenatal medical data were collected using well-structured questionnaire. Neonates were subjected to sepsis work-up including blood culture, complete blood count and C-reactive protein. Organisms were identified by Gram staining and analyzed by the VITEK II system for bacterial bio-typing and antibiotic susceptibility testing. Findings Of the 199-neonates with suspected neonatal sepsis, 154 (77.38%) had culture-proven sepsis. Early-onset neonatal sepsis (EOS) was higher (50.25%; 100/199) than late-onset neonatal sepsis (LOS) (27.13%; 54/199). Multivariable analysis identified vaginal delivery as an independent risk factor for neonatal sepsis p = 0.005. Majority of isolated bacteria (74.39%) were gram-negative with Burkholderia cepacia (39%) and Klebsiella oxytoca (13%) being the most common pathogens of EOS and LOS. The most common gram-positive pathogens were Staphylococcus haemolyticus (9.1%) and Staphylococcus epidermidis (7.1%). B. cepacia showed multidrug resistance except for cefepime. All Klebsiella species isolates (100%) and most Pantoea species (93%) were ESBL and carbapenemase positive. All Escherichia coli and Acinetobacter baumannii isolates were ESBL positive. A significant number of gram-positive bacteria showed resistance to vancomycin. Conclusion The study findings show a high proportion of neonatal sepsis among neonates admitted to hospitals in Sana’a city with antibiotic-resistant B. cepacia being the single most common pathogen causing EOS and LOS. Findings also emphasize the emerging threat of multidrug-resistant bacteria in neonatal units and will help develop evidence-based management of neonatal sepsis in Yemen.

Pathogens causing neonatal sepsis are either acquired from the maternal flora or postnatally from the hospital or community [7]. Bacteria are the most common pathogens, and they vary in the distribution between countries, regions and according to the disease onset EOS versus LOS [7]. The commonly involved bacteria in developing countries include Klebsiella species, Staphylococcus aureus, E. coli, and Group B Streptococcus [8,9]. Recently, B. cepacia, a bacterial pathogen that has intrinsic drug resistance and ability to live inside drug products [10,11], has emerged as a significant pathogen causing neonatal sepsis [12][13][14][15][16].
Yemen is suffering of an ongoing war that started in 2015. The war has destroyed the health system, rendering health facilities either not functioning or partially functioning [17]. There is a paucity of information about neonatal sepsis including its types, causative pathogens patterns and antibiotic susceptibility. Therefore, this study aimed to investigate neonatal sepsis, identify its associated factors and causative pathogens, and to assess the antibiotic susceptibility patterns in Sana'a city, Yemen. Findings from this study will fill gaps of knowledge about neonatal sepsis in Yemen, helping public health authorities to develop an effective control strategy to combat the disease, and guide Yemeni physicians to implement an evidence-based approach for the clinical management of neonatal sepsis.

Methods
All methods were carried out in accordance with the relevant guidelines and regulations.

Study design and subjects
This was a cross-sectional study conducted on neonates admitted to neonatal units in six major hospitals in Sana'a city; Al-Kuwait, C-Plas, Al-Gumhory, Al-Sabeen, AL-Thawra, and Family hospitals, in the period from January 15 to March 30, 2020. Neonates who were admitted for at least 72 h with suspected sepsis during the study period were included. Neonates with congenital anomalies and hemolytic jaundice were excluded from the study.

Diagnosis of neonatal sepsis
Neonatal sepsis was suspected based on the presence of clinical signs or risk factors according to the international recommendations [18] and confirmed as neonatal sepsis if blood culture was positive [18,19]. Neonatal sepsis was then classified into EOS and LOS [19].

Questionnaire
Bio and socio-demographic data were collected using a standard questionnaire by face-to-face interviews with the neonate's mothers. The health condition of the mothers and clinical manifestations of the neonates were collected by physicians using standard tools. Written informed consent was obtained from guardians of all neonates after explaining the study objectives. All procedures described in this study were approved by the Research and Ethics Committee of the Faculty of Medicine and Health Sciences, University of Science and Technology, Sana'a (Approval No. EAU/UST122).

Laboratory investigations
Laboratory investigations were performed according to standard microbiological techniques [20]. Under aseptic conditions, trained nurses collected blood samples that were used for laboratory investigations of complete blood counts, C-reactive proteins and blood culture. For culture, at least 1 ml of blood was inoculated into BacT/ Alert PF plus culture bottle (BIOMERIEUX, France, LOT 4053532) and incubated until the BacT/Alert instrument (BACTEC 9050, Becton Dickinson) signals it either as positive or negative. All positive samples were sub-cultured on choclate agar, blood agar, and MacConkey agar and incubated at 37 °C for 24-48 h. Gram-staining was conducted to differentiate between gram-positive and gram-negative bacteria. A sufficient number of colonies of pure culture was used to suspend the microorganism in 3.0 ml of sterile saline test tube. Pure bacterial suspension was added to bacterial specific identification and sensitivity testing kit device and analyzed by the VITEK II system for bacterial bio-typing and antibiotic susceptibility patterns as instructed in the product information manuals (BIOMERIEUX). VITEK ® GN ID identification card (lot 2410933203) was used to characterize gramnegative bacteria and VITEK ® GP ID identification card (lot 2420938203) was used to characterize gram-positive bacteria. Different cards were used to perform antibiotic susceptibility tests; AST-GN 87 (lot; 6770912203), AST-GN 72(lot; 5921083103) and AST GN75 (lot; 5951129403) for gram-negative bacteria and AST-GP67 (lot; 1321137103) for gram-positive bacteria. All procedures were conducted for routine diagnostic and therapeutic reasons.

Statistical analysis
Data were analyzed using IBM SPSS Statistics for Mac, version 23.0 (IBM Corp., Armonk, NY, USA). Categorical variables were presented in frequencies. The association between independent and dependent variables was tested using Pearson's Chi-squared with reporting odds ratio (OR) and 95% confidence interval (CI). Fisher's exact test was used when applicable. Multivariable analysis using the binary logistic regression model was conducted, including all variables, and the adjusted OR with its corresponding 95%CI was reported. A p-value of < 0.05 was considered significant.

Ethical consideration
The study protocol was approved by the Research and Ethics Committee of the Faculty of Medicine and Health Sciences, University of Science and Technology, Sana'a, Yemen (Approval No. EAU/UST122).

Characteristics of the patients
A total of 199 neonates admitted to referral hospitals in Sana'a city, Yemen, were enrolled in this study. Nearly half of neonates (84, 42.2%) were preterm, majority of them (70, 83.3%) were moderates to late preterm. More than half of neonates (109, 54.8%) were low birth weight. A total of 113 neonates (56.8%) were born by vaginal delivery. The age of 133 neonates (66.8%) was less than 72 h on admission. Tachypnea, lethargy and poor feeding were the most common clinical manifestations observed among neonates (Table 1).

Discussion
In this multicentric cross-sectional study we report a high proportion of culture-confirmed neonatal sepsis, accounting for two-thirds (77.38%) of admitted neonates in the referral hospitals in Sana'a city in Yemen. EOS was more common than LOS and B. cepacia emerged as the predominant causative organism of both EOS and LOS. Majority of the isolated pathogens were resistant to commonly used antibiotics. Culture confirmed sepsis was high in our study. However, the proportion of culture-confirmed neonatal sepsis varied between studies from developing countries; 62.8% in Pakistan [21], 57% in Yemen [22], 45.9% in Egypt [23], 44.7% in Ethiopia [24], 24% in Tanzania [25] and 12.6% in Nepal [26]. The differences in the proportion of neonatal sepsis between countries may be due to several factors including sample size and the different used techniques.
In this study, EOS was about two times higher than LOS, which is consistent with other reports from developing countries [22,23,26,27]. However, it is   inconsistent with reports from developed countries where LOS is the predominant type of neonatal sepsis [28,29]. The predominance of EOS in developing countries can be attributed to low quality health services and poor hygiene. This is mainly due to illiteracy, ignorance, cultural beliefs, and prejudices [19]. Infection may be of the maternal genital tract or from delivery rooms or neonatal units [19,30]. Similar to previous studies neonates born by vaginal delivery were at a higher risk of neonatal sepsis compared with those delivered through Cesarean section [23,27]. Further, culture confirmed sepsis was high (89.9%)   In this study, gram-negative bacteria were the most common cause of EOS and LOS, which is consistent with previous reports from developing countries [8,9,22,31]. Unexpectedly, B. cepacia emerged as the most common cause of neonatal sepsis. It was found in all neonatal units. Moreover, it was resistant to aminoglycosides, fluoroquinolones and imipenem. B. cepacia has been reported to cause multiple hospital outbreaks and significant neonatal septicemia in different countries [12][13][14][15][16]. Also, it has been isolated from intravenous solutions, mouthwash, disinfectant and medical devices [31][32][33]. This organism is often overlooked and reported as Pseudomonas species [33,34]. To our knowledge, this is the first report of B. cepacia from Yemen.
Staphylococcus haemolyticus was the most frequently isolated CONS in this study. Most isolated S. haemolyticus showed variable degrees of antibiotic resistance, which is the nature of S. haemolyticus to acquire antibiotic resistance due to unusual genome plasticity [35]. So, this organism has a great ability to survive in the hospital environment, especially on medical devices.
Our results of antimicrobial resistance profile are consistent with other findings. Though some of the previous reports were based on cultures from different kinds of medical specimens and performed by conventional methods [8,24,28,[36][37][38][39]. Most of the isolated gram-negative and gram-positive pathogens were resistant to commonly used antibiotics, penicillins, cephalosporins and carbapenems. Increased antimicrobial resistance in developing countries is due to multiple factors including poverty, self-medication, unregulated supply and drug smuggling, misguided practice and inappropriate prescriptions [40,41]. Such factors may provoke changes in causative agents together with their change in antibiotic susceptibility patterns [40,41]. Markedly, majority of neonates in this study received antibiotics before sampling. The presence of ESBL and carbapenemase-positive gram-negative bacteria is an alarming sign both locally and globally and increases the burden of neonatal sepsis in Yemen.
Proven sepsis was associated with elevated CRP. Elevated CRP indicates activation of the immune system and is commonly used as an indicator of bacterial sepsis [18,19]. The variation in the total white cell counts and platelet counts is considered a hematological response to inflammation, but there was no association between the culture-positive and culture-negative groups regarding white cell or platelets counts. This may be explained by considering evolving neonatal immune system [18,19].

Conclusions
Culture-positive neonatal sepsis is high in Sana'a city with EOS representing two-thirds of the cases. B. cepacia followed by K. oxytoca and S. haemolyticus were the most common causes of both EOS and LOS. Majority of the isolated bacterial pathogens showed a high level of resistance to commonly used antibiotics. Vaginal delivery held high risk for developing neonatal sepsis. The study results emphasize the emergence of multidrug-resistant bacteria in the NICUs and might serve as a baseline for proper medical treatment of neonatal sepsis in Sana'a city.