Etiology of Early-Onset Neonatal Sepsis and Antibiotic Resistance in Bukavu, Democratic Republic of the Congo

Abstract

Neonatal sepsis is a leading cause of neonatal mortality [1, 2], and more than 99% of cases occur in low- and-middle income countries [3]. Fleischmann-Struzek and coworkers (2018) estimated that 11%–19% of the 3.0 million neonates who develop neonatal sepsis die annually [4]. In the Democratic Republic of the Congo (DRC), the early neonatal mortality rate is estimated to be between 14 [5] and 28 per 1000 live births [2], and 15.6% of cases are caused by neonatal sepsis [6].

Neonatal sepsis is a systemic infection, most often caused by bacteria [7], and can be classified as either early-onset neonatal sepsis (EOS), which occurs within less than 72 hours of life, or late-onset neonatal sepsis (LOS), which present after the first 72 hours of life up to the age of 3 months [8]. Pathogens that cause EOS are considered to be vertically transmitted from the maternal vagina to the fetus/newborn after breaking of the fetal membranes by an ascending infection or during passage through the birth channel. In contrast, pathogens that cause LOS are considered to be hospital or community acquired [8, 9].

The most common organisms that cause EOS in high-income countries are group B streptococci (GBS, Streptococcus agalactiae) and Escherichia coli, accounting for more than 70% of cases [7, 10, 11]. Most international guidelines that are in line with World Health Organization (WHO) recommendations for empirical antibiotic treatment are based on the common antibiotic susceptibility of the predominant pathogens that cause EOS in high-income countries [12]. These guidelines recommend hospitalization and antibiotic therapy with a combination of gentamicin and benzylpenicillin or ampicillin for at least 7–10 days for the management of serious bacterial infection in infants aged less than 2 months [13]. However, studies that document the pathogens that caus EOS in sub-Saharan Africa are very limited and so are data on antibiotic susceptibility. For example, a recent systematic review and meta-analysis found reports on only 90 identified pathogens isolated from cases of neonatal sepsis in Central Africa in the period 2008–2018 [14]. As such, data on etiology and antibiotic resistance are urgently needed to evaluate the adequacy of the empirical treatment and to tailor antibiotic regimens to local resistance patterns.

In the current study, we aimed to identify the pathogens causing EOS in Bukavu (DRC) in order to assess their antibiotic susceptibility patterns and evaluate the applicability of the WHO guidelines for the management of EOS in the DRC.

METHODS

Study Population and Design

This was a descriptive cross-sectional study carried out from June 2017 until July 2018 at the Provincial General Reference Hospital of Bukavu (HPGRB). The HPGRB’s neonatal intensive care unit (NICU) serves neonates born inside the hospital as well as neonates referred from other health centers in the Bukavu area. Clinical data, that is, history taken from the parents and physical examination report, were abstracted from the registers of the NICU and from the laboratory records. Neonates were considered as having possible early-onset neonatal sepsis (pEOS) when they had 1 of the following signs, according to the WHO recommendations: incapacity to feed, fever, hypothermia, tachypnea, severe chest indrawing, nasal flaring, grunting, lethargy, reduction of movements, poor capillary refill time, bulging fontanelle, convulsions, jaundice, skin pustules, and/or unconsciousness [9, 15]. These newborns were admitted to the NICU where general resuscitation measures were taken and blood was sampled to assess C-reactive protein and for hemoculture. A standard antibiotic regimen that consisted of a combination of ampicillin, cefotaxime (as an additive), and gentamicin was started according to the synopsis of international guidelines for the management of EOS [16] that are in line with WHO recommendations [13]. In case there was no clinical improvement, an empirical regimen of amikacin in combination with clindamycin or benzylpenicillin (based on availability) was initiated.

A minimum of 0.5 mL of blood was collected from each neonate admitted for suspicion of pEOS in the first 72 hours after birth for hemoculture and for the assessment of C-reactive protein, except from neonates who began antibiotics before the admission or who had a congenital malformation or were undergoing a surgical procedure.

Laboratory Procedures

Hemoculture Procedures and Identification

Neonatal blood samples were added to hemoculture bottles (BACT/ALERT PF plus, BioMérieux) that were incubated aerobically at 37°C for up to 7 days. In case bacterial growth was noticed, a small volume of hemoculture fluid was inoculated onto tryptic soy agar plates with 5% sheep blood (Tryptic Soy Agar, Becton Dickinson, Erembodegem, Belgium; sheep blood from animals kept on campus; blood agar plates) that were incubated aerobically for 2–7 days at 37°C. Isolates were preserved in soft agar tubes, before shipment to the Laboratory Bacteriology Research (Ghent University, Ghent, Belgium). The colonies from the soft agar were regrown on blood agar plates and identified at the Department of Laboratory Medicine (Ghent University Hospital, Ghent, Belgium) using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker, Bremen, Germany) using the direct spot method according to the manufacturer’s protocol.

Antibiotic Susceptibility Testing

The methodologies used for susceptibility testing with the disk diffusion method are consistent with those recommended by the European Committee on Antimicrobial Susceptibility Testing [17].

Ethical Approval

This research was approved by the review board committee of the Catholic University of Bukavu and endorsed by the Provincial Health Ministry in the DRC and by the Ghent University Hospital Ethical Committee. A parent or close relative accompanied each neonate and provided agreement for participation in the study, and written informed consent was signed.

RESULTS

The flow chart for the hemocultures and isolate collection is presented in Figure 1. A total of 61 strains were isolated from 59 neonates. Of the 61 strains that were preserved and shipped, 50 (82.0%), from 48 neonates, could be regrown (Figure 1). All strains that could not be recultured were found to be gram-negative in the HPGRB laboratory.

Figure 1.

Flow chart for hemocultures and isolate collection. Abbreviations: HPGRB, Provincial General Reference Hospital of Bukavu; LBR, Laboratory Bacteriology Research; NICU, neonatal intensive care unit; pEOS, possible early-onset neonatal sepsis. aIsolated from the blood of 48 neonates (ie, 2 neonates had a polymicrobial infection).

Open in new tabDownload slide

The clinical characteristics of the 150 neonates with pEOS are listed in the Supplementary Materials. Nearly all neonates (133 of 150, 88.7%) were admitted to the NICU between 24 and 72 hours after birth. The female-to-male ratio was about 1:1. Approximately two-thirds of neonates were referred to the HPGRB. Less than half (46.0%) were born preterm, and more than half (54.7%) weighed less than 2500 g at birth. Hypothermia (47.3%), hypotonia (34.0%), and incapacity to feed (8.7%) were the most frequently reported symptoms. Respiratory distress was the most frequently reported clinical sign at admission to the NICU (62.0 %). The EOS case fatality rate was 25.3% (38 of 150 patients), and mortality was not associated with first-line treatment with antibiotic to which the pathogen was resistant (odds ratio, 4.7; P, .14).

The identified isolates are listed Table 1. The majority of isolates (82%) were gram-negative species. The most prevalent species were Enterobacter cloacae complex (42.0%), Klebsiella pneumoniae (18.0%), and Serratia marcescens (12.0%).

The antibiotic susceptibility of E. cloacae complex, K. pneumoniae, and S. marcescens isolates to different antibiotics is shown in Table 2. The majority of these isolates had an ESBL phenotype, that is, all E. cloacae complex, 8 of 9 (89%) K. pneumoniae, and 3 of 6 (50%) S. marcescens isolates. Of the other gram-negatives, only the Proteus mirabilis isolate had an ESBL phenotype.

DISCUSSION

Ours is one of the few microbiological studies of EOS in the DRC. A recent systematic review and meta-analysis on the etiology and antimicrobial resistance in neonates in sub-Saharan Africa reported on 90 cases from Central Africa, highlighting the absolute lack of data in this region [14]. Our data are in line with these meta-data. We found that the majority of pathogens (82%) were gram-negatives, which is in line with the reported 76% in the study by Okomo and colleagues [14]. Enterobacter cloacae complex (42%), K. pneumoniae (18%), and S. marcescens (12%) were identified as the most prevalent pathogens causing EOS in our study, whereas the meta-data reported Klebsiella (42%), Escherichia coli (19%), and Staphylococcus aureus (14%) to be the most prevalent pathogens. However, it should be noted that data for EOS and LOS are merged in the publication by Okomo et al (2019). In the meta-analysis, Enterobacter was found in only 4% of the cases, and Serratia was absent in Central Africa [14]. Serratia, however, is found in a minority of cases in eastern, southern, and western Africa [14]. In contrast, Bunduki et al (2019) [19] documented the etiology of 18 cases of EOS in Butembo (DRC) and found Streptococcus agalactiae (n = 6) to be the most prevalent pathogen, followed by S. aureus (n = 5), E. coli (n = 3), and K. pneumoniae (n = 2).

GBS, which causes most EOS cases in industrialized countries [20], was found in only 1 case in our study, which is in line with the meta-analysis from Central Africa (2 of 90 cases) [14]. As we had only 1 GBS case among 660 neonates admitted to the NICU, the incidence of GBS EOS in Bukavu appears to be lower than the estimated 1.3 cases per 1000 live births in sub-Saharan Africa [21]. GBS vaccination has been suggested by the WHO as a potential control strategy in low- and-middle-income countries to reduce GBS EOS [22]. According to our results, GBS vaccination likely would not substantially reduce the burden of neonatal sepsis mortality in Bukavu.

In our study, we could document only 2 cases of EOS caused by S. aureus and 2 by E. coli. The latter pathogen causes most EOS mortalities in high-income countries [23].

In our study, the case fatality rate of neonates with culture-confirmed EOS was substantial (25.3%) and nearly 3 times higher compared with the 9.8% in low-income countries reported by Seale et al [24]. For the treatment of EOS in the HPGRB, ampicillin, gentamicin, and cefotaxime were used as first-line treatment based on the synopsis of international guidelines for the management of EOS [16], which are in line with WHO recommendations [13] and the American Academy of Pediatrics for the management of EOS [25]. In case no improvement was noticed in the first 72 hours, ampicillin was replaced by clindamycin.

In our study, all Enterobacter, Klebsiella, and Serratia isolates were resistant to ampicillin, as would be suspected, and to clindamycin because these species are intrinsically resistant to clindamycin [26–28]. Moreover, all Enterobacter and most Klebsiella and Serratia isolates were resistant to gentamicin and cefotaxime. These high resistance rates of the most prevalent pathogens in our study population toward the first-line antibiotics likely contributed substantially to the high EOS mortality rate in our study population.

During the study, a combination of amikacin with clindamycin or with benzylpenicillin was introduced. Amikacin was chosen because of its effectiveness in killing multidrug-resistant gram-negative aerobes. In the current study, all E. cloacae complex isolates and S. marcescens strains were susceptible to amikacin, as were more than half of the K. pneumoniae strains.

In this study, the Bacillus, Corynebacterium, and Staphylococcus epidermidis that were isolated could be contaminants rather than true pathogens [18], and cautious interpretation is warranted. We could not determine that the resistance to usual antibiotics was an independent risk factor of neonatal mortality. Although it is known that prevalent EOS pathogens may be vertically acquired from the vagina, data about bacterial vaginal colonization of sick neonates’ mothers were not collected during this study. Also, neonates who could not reach the referral hospital were missed.

The high prevalence of antibiotic resistance and ESBL phenotypes in our study population is worrisome. We did consider the possibility of an outbreak, but matrix-assisted laser desorption/ionization time-of-flight–based typing [29] (data not shown) and epidemiological data did not support this hypothesis. The widespread use of over-the-counter antibiotics within the community, fueled by a lack of knowledge by healthcare providers [30], likely is an important contributor. Our study clearly highlights the importance of a sustained monitoring system of (the antibiotic susceptibility patterns of) pathogens that cause neonatal sepsis in the DRC. The current challenges are to address the lack of quality-assured laboratories equipped for bacteriological monitoring and surveillance of EOS, the lack of adequate training in infection prevention, and appropriate control of access to essential antibiotics.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank Eric Hendwa, Faustin Kinunu, Faustin Masirika, the staff of the Provincial General Reference Hospital of Bukavu (HPGRB) neonatal intensive care unit, the staff of the HPGRB Department of Obstetrics and Gynecology, and all neonates and their parents for their helpful support of the study.

Financial support. This work was supported the Flemish Interuniversity Council (Vlaamse Interuniversitaire Raad (Flemish Interuniversities Council)-Universitaire Ontwikkelingssamenwerking (University Development Co-operation) [VLIR-UOS]; grant ZIUOS2012AP024).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

References