Factors associated with neonatal sepsis among neonates admitted in Kibungo Referral Hospital, Rwanda

More than one million neonatal deaths occur every year worldwide, of which 99% take place in low-income countries. In Rwanda, nearly 71% of neonatal deaths are preventable and among these, 10% are due to neonatal sepsis. Nevertheless, limited information exists on neonatal sepsis and its associated factors in Rwanda. The objectives of the study were to find prevalence and factors associated with neonatal sepsis among neonates admitted in Kibungo Referral Hospital, Ngoma District, Rwanda. We used a retrospective cross-sectional study design reviewing a subset of neonatal, maternal and laboratory records from Kibungo Hospital in 2017. Data were reviewed and collected from March to May, 2018. Logistic regression and odds ratios were calculated to identify the factors associated with neonatal sepsis at 95% CI, p < 0.05. Of the 972 total neonates’ medical records from 2017, we randomly selected 422 of which 12.8% (n = 54) had neonatal sepsis. When blood cultures were positive, 62% grew Klebsiella pneumoniae. Among neonates with sepsis, 38 (70%) recovered while 16 (30%) died. Neonatal sepsis was strongly associated with neonatal age less than or equal to three days (aOR: 2.769, 95% CI 1.312–5.843; p = 0.008); and gestational age less than 37 weeks (aOR: 4.149; CI 1.1878–9.167; p ≤ 0.001). Increased use of blood cultures including sensitivity testing, routine surface cultures of the neonatology and maternity wards facilities, and systematic ward cleaning are all important approaches to prevent and treat neonatal infections in additional to regular neonatal sepsis evaluations.


Study population and sample size
Of the 972 neonatal medical records at Kibungo Hospital from 2017, we collected neonatal and maternal data from a random sample of 422.We also collected all environmental laboratory cultures sent from the neonatology and maternity wards during 2017.

Data collection and procedures
Clinical data including neonatal and maternal characteristics were extracted from neonatal files.Environment swabbing culture results were collected from laboratory records.A Microsoft Excel data collection tool was developed to record neonatal, maternal and ward environmental data.Data collection was conducted in a threemonth period from March to May, 2018.
Maternal data were: age, fever > 38 °C, number of previous pregnancies, place of delivery, educational level, employment status, health insurance, obstructed labor, premature rupture of membrane (PROM), foul smelling amniotic fluid.Neonatal data were: post-natal age, sex; birth weight; mode of delivery; place of birth; birth asphyxia (yes/no); gestational age in weeks; Apgar score at 10 min; resuscitation after birth (yes/no); sepsis (yes/ no), and if yes, time of onset, bacterial isolate, and sepsis outcome (died or recovered).The determination of sepsis was relied on physicians' classification of whether the neonate had sepsis as final diagnosis.
The environment cultures data were gathered from laboratory records.Laboratory records were reviewed to collect information related to environmental samples collected for bacteriological cultures, sample sources and identified bacteria species.

Data analysis
Stata Statistical Software was used for analysis: version 13.College Station, TX: Stata Corp LP https:// www.stata.com/.Descriptive statistics, including frequencies and percentages, were calculated to characterize the study population in terms of socio-demographic and other relevant variables.Analytical statistics were performed for bivariate and multivariable logistic regression.For the bivariate logistic regression, the association between independent and dependent variables was found at 95% confidence interval with p value < 0.05.Multivariable logistic regression was done for the variables that showed significant association at the bivariate analysis, and p value ≤ 0.05 was considered statistically significant.

Ethical approval
The study obtained ethical approval from the ethical committee of Kibungo Referral Hospital with a reference letter (Ref No:14/117/RJ/H1-1/2/2018). And all data were collected in accordance with all relevant guidelines and in the Declaration of Helsinki 46 .The secondary data from a programmatic management health system were used hence the informed consent and assent were waived by the ethical committee of Kibungo Referral Hospital.Findings of the study were submitted to Kibungo Hospital and the University of Rwanda, College of Medicine and Health Sciences, School of Public Health.1).

Environmental bacterial colonization monitoring
Environmental cultures reviewed were collected from surfaces of neonatology and maternity equipment such as incubators, lamps, nasal aspirator and oxygen concentrators, and showed the presence of various bacterial species in the neonatal and maternity wards: Enterobacter cloacae, Klebsiella pneumoniae, Serratia species, Providencia stuartii and Escherichia coli.

Bivariate analysis of neonatal characteristics
After bivariate analysis, neonates with a younger post-natal age group (p = 0.022) and younger gestational age (p = 0.031) were found to be significantly associated with neonatal sepsis (Table 3) and therefore, they were considered for multivariate logistic regression analysis.Association of Apgar score with neonatal sepsis was not statistically significant (p = 0.771).

Bivariate analysis of maternal characteristics
After the bivariate analysis of the maternal characteristics, it was seen that no maternal characteristic was statistically significant and eligible for the multivariable logistic regression (Table 4).

Multivariable analysis of characteristics associated with sepsis
Multivariable analysis was done only for the neonate characteristics that showed significant association with neonatal sepsis in bivariate analysis (neonate age and gestational weeks).Strong association with sepsis was found with neonatal age less or equal to three days (aOR: 2.769; 95% CI 1.312-5.843;p = 0.008), and gestational weeks less than 37 weeks (aOR: 4.149; CI 1.1878-9.167;p ≤ 0.001) (Table 3).

Discussion
The purpose of this study was to identify the risk factors associated with neonatal sepsis among neonates admitted in Kibungo Referral Hospital, Ngoma District, Rwanda during the calendar year of 2017.Neonatal sepsis prevalence was 12.8%, of whom 29.6% died.Decreasing deaths for newborns and mothers is a global priority to achieve Sustainable Development Goals, and to implement United Nations Global Strategy for Women, Children and Adolescent health 5 .Though Rwanda is among the few countries in Africa that achieved the fourth Millennium Development Goal to reduce child mortality, mortality rate among neonates is still of great concern.Neonatal sepsis is still a leading cause of neonatal morbidity and mortality in Rwanda, including in the study area 47,48 .The 12.8% neonatal sepsis prevalence found in this study is almost similar to results reported in the Northwest Ethiopia (11.7%) 49 .But it is lower compared to reports from Ethiopia (33.8%),Tanzania (24%) 50 .The difference in neonatal sepsis prevalence may be due to the difference in the definitions of sepsis and the study settings in terms of infection prevention and control, staffing, funding, and policy.
In this study, early onset neonatal sepsis was slightly more common (57.4%) than late onset neonatal sepsis (42.6%).On the other hand, our study results contrasts with what was reported in a research conducted at Mansoura Hospital in Egypt, where early onset sepsis was lower at 44.2% compare to late onset sepsis at 55.8% 51 .This could be due to the differences in the characteristics of mothers and the setting where the study was conducted.
The sepsis fatality rate in this study was 29.6%, which is high and may reflect inadequate management of sepsis, perhaps linked to delays in diagnosis and treatment.This sepsis fatality rate is similar to the findings of a study from Tehran (27.4%) 52 .Blood cultures were not routinely performed in suspected cases, with only 25 of the 54 cases of suspected sepsis confirmed by culture.This could be due to the long interval between sending and receiving the culture results that might push physicians to treat suspected sepsis empirically.The predominance of Klebsiella pneumoniae was noted and the sensitivity pattern was similar to the findings from the research conducted in India where the isolated bacteria were sensitive to Meropenem and Imipenem with almost half www.nature.com/scientificreports/ of the isolated Klebsiella pneumonia bacteria species were resistant to Ciprofloxacin 53 .The findings of current study are different from a study done in Sudan that reported susceptibility of Klebsiella pneumoniae of 87% to Ciprofloxacin, and 81% to meropenem, which was caused by a variation of antimicrobial susceptibility patterns in bacteria 54 .Multivariable analysis revealed that younger gestational age and postnatal age were both statistically significantly associated with neonatal sepsis.Neonates aged 3 days or less were more likely to develop sepsis which is comparable with the study done in Ethiopia neonates 55,56 .Premature rupture of membrane, place of delivery, intrapartum fever, Apgar Score < 7 at 10 min, low birth weight, meconium stained amniotic fluid, foul smelling amniotic fluid and assisted ventilation were not found to be associated factors, in contrast to the studies done in Mexico and Ethiopia which, of note, were conducted in urban settings 19,57 .
In this study, environmental cultures showed the presence of Enterobacter cloacae, Klebsiella pneumoniae, Serratia species, Providencia stuartii and Escherichia coli in the neonatal and maternity wards.The study done in Morocco, Brazil, Austria and India on neonatology environment reported the presence of different bacteria including Klebsiella pneumoniae, coagulase-negative staphylococci and other Enterobacteriaceae 37,[58][59][60] .The presence of those bacteria isolates in the neonatal and maternity settings highlights the need to regularly monitor their environment and execute infection control by improving hygiene and sanitation.

Limitation of the study
One of the study limitation was the fact that data were collected in a single rural referral hospital in Rwanda.Therefore, the study results may not be directly applicable to other settings.Additionally, the use of secondary data may have limited the available data to assess for statistical association with neonatal sepsis in the study area.

Conclusion
The study highlighted that post-natal age ≤ 3 days, and gestation age < 37 weeks were significantly associated with neonatal sepsis.Improving the use of blood culture and consistent tailoring of antibiotics based on antibiotics susceptibility testing could enhance the management of neonatal sepsis.

Table 3 .
Bivariate and multivariate analysis of neonatal characteristics associated with sepsis at Kibungo Hospital.*: Statistical significance at p < 0.05, **: statistical significant at p < 0.01.cOR crude odds ratio, aOR adjusted odds ratio.Significant values are in [bold].

Table 4 .
Bivariate analysis of maternal characteristics associated with sepsis at Kibungo Hospital.