Clinical features of neonatal sepsis in an emergency department of a tertiary care center in Karachi, Pakistan
Article information
Abstract
Purpose
Despite the high burden of neonatal sepsis in emergency departments (EDs), most studies focus on inpatient settings or post-discharge outcomes, leaving a gap in understanding infection signs within EDs. We aimed to identify early signs of neonatal sepsis in an ED of a low- or middle-income country.
Methods
This retrospective cross-sectional study was conducted on neonates with sepsis who visited the Aga Khan University Hospital ED in Karachi, Pakistan, from January 1 through December 31, 2021. We collected and analyzed patient data, covering demographics, clinical presentation, laboratory findings, and outcomes, and categorized them into 4 groups: “neonatal jaundice with suspected sepsis,” “suspected sepsis,” “sepsis,” and “septic shock.”
Results
Among 368 neonates with sepsis (median age, 7.0 days [interquartile range, 3.0-16.0]), 59.6% were aged 7 days or younger, 48.9% were preterm born, and 54.3% were delivered via cesarean sections. The most common presenting symptom was fever (33.7%), followed by jaundice (20.9%), vomiting or loose stools (19.6%), and dyspnea (19.3%). The septic shock group had the highest median concentrations of C-reactive protein and blood urea nitrogen, and the lowest median values of systolic blood pressure, hemoglobin concentration, and platelet count. The group also reported the highest frequencies of oxygen therapy, intubation, blood transfusion, hospitalization to the neonatal intensive care unit, and in-hospital mortality. Blood cultures were true positive in 17 (4.6%), with Staphylococcus aureus most common, followed by Klebsiella pneumoniae, Streptococcus pyogenes, and Burkholderia cepacia.
Conclusion
We found that early signs of neonatal sepsis in the Pakistani ED are neonatal age (< 7 days), true positivity of blood culture, and laboratory markers. These diagnostic approaches in low- or middle-income countries can help diagnose and treat sepsis early on, in the setting of unaffordability or unavailability of more sensitive and specific diagnostic tests.
Introduction
Sepsis is the sixth and eighth common causes of death in newborns and infants, respectively (1). Neonatal sepsis can lead to life-threatening organ dysfunction (2). In countries with the highest neonatal mortalities, infection contributes to approximately half of neonatal deaths, most of which occur shortly after birth (3). The 2016-2017 Global Burden of Disease study estimated that there are about 1,300,000 annual cases of neonatal sepsis worldwide, leading to an alarming figure of about 203,000 deaths solely due to sepsis (4,5). In addition, neonatal sepsis is a leading cause of neonatal mortality in low- or middle-income countries (LMICs), comprising 4.8% of global neonatal deaths. This highlights the alarming gap in neonatal health outcomes worldwide, with many infants suffering due to inadequate healthcare and infection control practices (6,7). Although neonates are vulnerable to sepsis with most deaths occurring in LMICs, data on incidence are often unavailable or unreliable in the geographic locations (8).
Early-onset neonatal sepsis, which occurs within 72 hours of birth, is typically due to the pathogens acquired during the peripartum period (2,9,10). The condition is commonly caused by gram-positive bacteria, such as Streptococcus agalactiae, Escherichia coli, Staphylococcus aureus, Enterococcus spp., and Streptococcus pneumoniae (11). Late-onset neonatal sepsis, which occurs after 72 hours in neonatal intensive care units (NICUs) or after 7 days in full-term neonates, is generally caused by hospital-acquired gram-negative bacteria, such as E. coli, coagulase-negative staphylococci, Klebsiella pneumoniae, and Acinetobacter spp. (3,11-13).
Despite the high burden of neonatal sepsis in emergency departments (EDs), most existing studies have focused on inpatient settings or re-hospitalization rates as signs of morbidity following early neonatal discharge (14,15). There is a gap in signs of neonatal sepsis within EDs. This lack of information restricts our understanding of how neonates with sepsis present in EDs where diagnostic tools are most effective for early intervention to reduce the morbidity and mortality.
The objective of our study was to identify early signs of neonatal sepsis in terms of clinical features, inflammatory biomarkers, multiorgan dysfunction, and birth history to facilitate early recognition and intervention in an ED. Such signs would be highly impactful in LMICs, where the mortality rate is high, and diagnosis is primarily based on clinical features or routine blood works due to the unavailability or unaffordability of resources for neonatal intensive care.
Methods
1. Study design and setting
This retrospective cross-sectional study was conducted at the ED of Aga Khan University Hospital, a tertiary care hospital in Karachi, Pakistan. The hospital operates as a major referral center, with approximately 15,000-18,000 annual pediatric emergency visits. The ED is a 65-bed, well-equipped facility serving adult and pediatric patients, of which pediatric section has 23 dedicated beds, is staffed by trained pediatric emergency medicine faculty providing 24/7 coverage, and receives approximately 1,500 neonates annually. The study was approved by the ethical review committee of Aga Khan University Hospital with a waiver for informed consent (ERC no. 2021-5786-16806).
2. Study population and definitions
We included all neonates who visited the ED from January 1 through December 31, 2021. Cases of neonatal sepsis were classified into 4 groups: “neonatal jaundice with suspected sepsis (NJSS),” defined as jaundice accompanied by fever and nonspecific symptoms, prompting a workup for both jaundice and sepsis; “suspected sepsis,” similar signs and symptoms as the former but without jaundice; “sepsis,” fever, lethargy, feeding reluctance, diarrhea, vomiting, respiratory distress, or seizures; and “septic shock,” requirement of fluid resuscitation or inotropic support due to low systolic blood pressure. The on-call pediatric ED faculty adhered to standardized criteria to diagnose neonatal sepsis. The NJSS group underwent an additional diagnostic workup for jaundice, such as total serum bilirubin, blood typing, direct Coombs’s test, and reticulocyte count.
Blood culture (BC) was performed using BACT/ALERT PF Plus (bioMérieux, Inc.). During the study period, a blood sample for BC (≥ 1 mL) was obtained by phlebotomy using sterile technique and inoculated into an an aerobic bottle. Pathogenic bacteria were defined as S. aureus, K. pneumoniae, Streptococcus pyogenes, Burkholderia cepacia, Serratia marcescens, Candida pelliculosa, Acinetobacter spp., and Aspergillus fumigatus. The other bacteria were regarded as contaminants. If no growth was reported after 5 days of BC, it was considered negative.
3. Study protocol
For stable neonates with suspected or definite sepsis, an initial assessment by the on-call faculty was followed by a septic workup, such as complete blood count, C-reactive protein (CRP), biochemistry panel, and BC. Third-generation cephalosporin and aminoglycoside were empirically administered per the institutional protocol. This regimen was adjusted to clinical response. For example, for neonates failing to improve within 48 hours, the initial combination was switched to vancomycin and meropenem, suspecting a nosocomial infection. In cases of septic shock, we performed additional investigations, such as coagulation profile, renal function tests, electrolytes, blood gas analysis, and plain radiography, with urine culture or lumbar puncture conducted if indicated. Based on the clinical stability, neonates were hospitalized to either the ward or the NICU.
4. Data collection and analysis
Data were recorded in a standardized online form, capturing patient demographics, presenting complaints, vital signs, birth history, prior hospitalizations, investigations, medications, ED length of stay (≥ 6 hours), and disposition. IBM SPSS ver. 21 (IBM Corp.) was used for data analysis. Continuous variables were summarized using medians with interquartile ranges, and categorical variables as numbers and percentages. Among the 4 groups, continuous variables were compared using 1-way analyses of variance or Kruskal-Wallis tests according to normal distribution, and categorical variables using chi-square tests.
Results
A total of 368 neonates were included in the study, comprising 226 males (61.4%) and 142 females (38.6%; Table 1). Approximately half of the patients were born term or post-term or via cesarean section. The most common presenting symptom was fever (124 [33.7%]), followed by jaundice (77 [20.9%]), vomiting or loose stool (72 [19.6%]), and dyspnea (71 [19.3%]; Figure).
Neonatal and maternal characteristics (N = 368)
Clinical features of neonatal sepsis. This graph illustrates the frequency of common clinical features observed in the neonates diagnosed with sepsis (n = 368), arranged from least to most prevalent: failure to pass stool (5 [1.4%]), irritability (6 [1.6%]), altered level of consciousness (12 [3.3%]), dysuria (13 [3.5%]), abdominal distension (21 [5.7%]), increased crying (30 [8.2%]), feeding difficulties (35 [9.5%]), dyspnea (71 [19.3%]), vomiting or loose stools (72 [19.6%]), jaundice (77 [20.9%]), and fever (124 [33.7%]). The list is mutually inclusive.
We found that the NJSS or septic shock group had a higher proportion of neonates aged 7 days or younger (P < 0.001; Table 2). At triage, the septic shock group showed the lowest median systolic blood pressure and the highest concentration of CRP. NICU hospitalization was higher in the sepsis and septic shock groups than in the other 2 groups. In-hospital mortality was highest in the septic shock group.
Distribution of birth weight, gestational age, triage vital signs, and C-reactive protein across the neonatal sepsis categories
Neonates progressing to sepsis exhibited significant declines in hemoglobin concentration and in platelet count, as well as a rise in blood urea nitrogen concentration (Table 3). In terms of the therapeutic procedures, the septic shock group showed the highest percentages of oxygen therapy, intubation, and transfusion, whereas the NJSS group exhibited the highest percentage of phototherapy. Regarding the antibiotic therapy, the NJSS or suspected sepsis group tended to receive cefotaxime and amikacin more frequently than the other groups. An inverse pattern was observed in the use of vancomycin and meropenem.
Laboratory findings, procedures, antibiotic use, and outcomes in neonatal sepsis
BCs were positive in 56 neonates (15.2%), of whom 17 (4.6%) had true positive cultures, most commonly in the septic shock group (27.3%; Table 3). The most commonly isolated bacteria were S. aureus, followed by K. pneumoniae, S. pyogenes, and B. cepacia (Appendix, https://doi.org/10.22470/pemj.2025.01319).
Discussion
To the authors’ knowledge, this study is the first to explore neonatal sepsis in EDs in LMICs. Karachi is a large and densely populated city with a wide range of healthcare facilities. In resource-limited settings, such as small clinics or secondary hospitals, it may be difficult to meet standardized guidelines of infection control practices or sterilization protocols. As a result, neonates in such an environment might be at higher risk of developing early-onset sepsis due to perinatal exposure to pathogens. This variability in care likely influences some of the trends that we observed in our study.
Our findings noted that most of the patients were 7 days or younger, consistent with the existing literature. Neonates at this age are more susceptible to neonatal sepsis, possibly due to perinatal exposure, unsterilized instrumentation, or immature immunity, relative to older neonates (2,15). AlGhamdi et al. (16) found that neonates frequently present to EDs, with the majority being under 2 weeks of age. In LMICs, most infections during the first week of life were related to environmental causes, such as unhygienic delivery practices (17). This is due to an immature immunity in the neonates, increasing susceptibility to infection (18).
We found that approximately half of the neonates with sepsis underwent cesarean sections. A possible association between the sepsis and cesarean delivery could be either due to unsterilized instruments or no exposure to vaginal flora, predisposing the neonates to pathogenic bacterial infection (19). In addition, the prolonged hospital stays at birth, which is associated with cesarean section, could also contribute development of sepsis (20).
In addition, gestational age at birth was also a risk factor. Our findings also showed that the preterm-born neonates had a higher susceptibility to sepsis, similar to what has been found in previous studies (21). A Ghanaian study found that delayed or insufficient breastfeeding could also lead to neonatal infections (22). This possible association was not analyzed in our study.
Our findings of high CRP concentration and thrombocytopenia were indicative of sepsis, consistent with an Indian study on the platelet count indices in neonatal sepsis (23). A high creatinine indicates deteriorating renal function and signaling progression towards sepsis (24). While BC remains the gold standard, we found CRP to be particularly useful for diagnosis. However, given its low sensitivity, the biomarker should be combined with other biomarkers, such as interleukins or cluster of differentiation 64, for diagnostic performance. For diagnostic purposes, Delanghe and Speeckaert (25) developed the ApoSAA score and found it to be highly sensitive for identifying neonatal sepsis. However, there is still a gap in the current literature on infection signs for neonatal sepsis in EDs, as most research has focused on inpatient care.
We found that contamination was a common issue in neonatal BC collection. Hashemizadeh et al. (26) found that contaminants accounted for 8.5% of samples for neonatal BCs. Contamination could be attributed to differences in sampling techniques, as shown by the fact that 26 Canadian NICUs where BC was performed for diagnosis of neonatal sepsis showed a significant practice variation in such techniques (27). Therefore, the best approach would be standardization of sampling techniques for BC, as well as using true positivity for diagnosis. The true positivity would help optimize antibiotic therapy. A study done at a Moroccan tertiary hospital found that true bacteremia cases showed a significantly different clinical course, based on whether appropriate antibiotics were given (28).
Limitations of this study include the retrospective study design, along with brief medical histories recorded during stay in the ED, limiting our ability to gather data on maternal history, antenatal history, reasons for caesarian sections, or Apgar scores. Furthermore, as a single-center study, it may not fully represent the diversity of patient populations or clinical practices across LMICs. This limitation identifies the need for multicenter research across EDs in LMICs to better understand the broader implications and trends in this topic.
To summarize, our study findings based on clinical and laboratory variables may help diagnose neonatal sepsis early on and assess its deterioration. Neonates in the first week of life, particularly those born preterm, were at a higher risk, reflected in triage vital signs and inflammatory biomarkers. With respect to deterioration, progression to septic shock was associated with worsening values of hemoglobin, platelet count, CRP, and blood urea nitrogen. BC has been proven to be a low-yield diagnostic test due to false positivity. In this context, we found that true positive BC contributes to more accurate diagnosis and more appropriate empirical antibiotic therapy. With respect to other therapeutic plans for neonatal sepsis in EDs, neonates diagnosed under the neonatal sepsis umbrella required ventilation, phototherapy, or transfusions. Inability to diagnose can lead to the use of inappropriate antibiotics or clinical deterioration, worsening morbidity and mortality, and a burden on resource-limited EDs in LMICs. Therefore, more research is needed to identify an affordable biomarker that can potentially help diagnose neonatal sepsis early on. We believe that this study will help reduce morbidity and mortality by contributing to earlier diagnosis and treatment in LMICs.
Notes
Author contributions
Conceptualization: SB, MAW, and AR
Methodology and Visualization: SB and MAW
Software and Formal analysis: AR
Validation, Supervision, and Project administration: SB
Investigation, Resources, and Data curation: NS, RJ, and WIV
Writing-original draft: MAW
Writing-review and editing: MAW and SB
All authors read and approved the final manuscript.
Conflicts of interest
No potential conflicts of interest relevant to this article were reported.
Funding sources
No funding source relevant to this article was reported.