Clinical and epidemiological characteristics of children admitted with fever in emergency department with or without sepsis

Introduction: Sepsis is a major cause of childhood death worldwide. In developing countries, epidemiological data about sepsis is scarce. This study describes and compares the frequency of etiological agents and initial sites of infection in children with or without sepsis, identifying risk factors and assessing outcomes. Methodology: Clinical and demographic data from patients < 13 years of age with reported fever in a pediatric emergency department were collected and registered in forms. Patients were classified as with or without sepsis according to Goldstein et al.’s criteria [6]. Results: Of 254 patients, 120 (47%) did and 134 (53%) did not meet the sepsis definition. Overall, the median age (IQR) was 1.7 (0.8–3.9) years, and 153 (60%) were boys. Patients with sepsis were older (2.8 [1.1–5.3] versus 1.3 [0.6–2.9] years; p < 0.001) and had sickle-cell disease more frequently (7.6% versus 0.8%; p = 0.007). By multiple logistic regression, age and sickle-cell disease were independently associated with sepsis. The most frequent initial infections were pneumonia (43.7%), diarrhea (17.3%) and cellulitis/adenitis (13.0%). The frequency of these did not differ when patients with or without sepsis were compared. Etiology was established in 57 (22.4%) patients, 32 (26.7%), and 25 (18.7%) with or without sepsis, respectively. Four (3.3%) patients died in the sepsis subgroup, whereas none died in the other subgroup. Conclusions: Children who met the 2005 international consensus definition of sepsis showed differences in age and comorbidities (sickle-cell disease) upon admission and were more likely to die.


Introduction
According to data published by the World Health Organization (WHO), of the 6.3 million children who died in their first 5 years of life in 2013, 51.8% (3.257 million) died of infectious diseases. Pneumonia, diarrhea, and malaria were the leading infectious causes of death [1]. Sepsis represents the progressive underlying inflammatory pathway secondary to any infectious illness, and ultimately is responsible for most infectious disease-related deaths [2].
In immunocompetent children, fever is the earliest clinical sign of infection [3], and, at the same time, it is one of the most common symptoms reported in pediatric emergency departments (EDs) [4]. As such, the early identification of potential septic children by healthcare providers who assist patients with acute febrile illness has decisive importance [5].
In 2005, Goldstein et al. established specific sepsis criteria for pediatric patients, based on age-specific norms of vital signs and laboratory data [6]. Though Goldstein et al.'s definitions were designed for research purposes, they are often used clinically in developed nations for identification of children who have lifethreatening infections requiring urgent intervention. More importantly, they form the basis for the development of guidelines aimed at treating the various stages of the sepsis continuum [2].
The Surviving Sepsis Campaign established some essential guidelines for reducing the high mortality rates resulting from sepsis in the world. It emphasizes the need to get to know the characteristics of each region [7]. Local epidemiologic studies about pediatric sepsis have been undertaken, most of them in developed countries. However, the important differences between children in developing countries and those in developing ones are clear [8].
This study aimed to describe and compare the frequency of etiologic agents and initial sites of infection, identifying risk factors and assessing outcomes, in pediatric patients admitted with fever, with or without sepsis, by applying Goldstein et al.'s criteria [6] in an ED in a developing country.

Methodology
This was a retrospective cohort study. Every patient admitted at the pediatric ED of the Federal University of Bahia Hospital between 1 September 2004 and 28 August 2005 had their medical chart reviewed by one member of the research team. At this ED, children come from the community of their own accord, and those under 13 years of age are seen. Patients who reported fever and had suspected or detected infection were identified, from whom patients with documented fever at physical examination were selected. The inclusion criteria comprised documented fever plus white blood cell count (WBC) collected upon admission plus respiratory rate plus heart rate measured at three or more distinct moments in the absence of fever. Patients with malformations, bronchopulmonary dysplasia, cardiac insufficiency, renal insufficiency, hepatic insufficiency, or medullary aplasia prior to hospital admission were excluded to ensure that vital signs were not abnormal due to underlying ongoing diseases. Demographic and clinical data upon admission and during evolution were registered in standardized forms.
Bacterial infections were searched for by performing blood, urine, and feces culture. Eleven viral causative agents of pneumonia were investigated in nasopharyngeal aspirates (NPA) collected from these patients. Rhinovirus, enterovirus, and human metapneumovirus were searched for by reverse transcriptase polymerase chain reaction (PCR) [9]. Parainfluenza viruses 1, 2, and 3, respiratory syncytial virus, influenza viruses A and B, and adenovirus were investigated by searching for viral antigen in the NPA by performing time-resolved fluoroimmunoassay with monoclonal antibodies. Infection by these respiratory viruses was also investigated by determining specific IgG increase in paired serum samples by enzyme-linked immunosorbent assay (ELISA) [10]. Human bocavirus infection was investigated by quantitative PCR of NPA and serum, by determining IgG increase in paired serum samples and by searching for IgM and IgG avidity by ELISA [11]. Streptococcus pneumoniae, Haemophilus influenza, and Moraxella catarrhalis infections were investigated by determining specific IgG increase tested by ELISA in paired serum samples [12]. PCR in the acute buffy coat was performed for the detection of pneumococcal DNA [13]. Mycoplasma pneumoniae infection was investigated by testing either the first or second serum sample for specific IgM by ELISA [14]. Chlamydia trachomatis infection was investigated by high IgG level determined by ELISA. IgG increase in paired serum samples or detection of specific IgA and IgM, all by microimmunofluorescence, were used to investigate Chlamydophila pneumoniae infection [15].
Serological markers of hepatitis A virus were investigated by determining specific IgM by ELISA [16]. Rotavirus (RV) antigen was searched for in stool specimens by a rapid latex assay (ROTA Rich, Richmond Diagnostics, Barcelona, Spain). Giardia lamblia infection was investigated by stool examination on fresh specimens or after preservation with polyvinyl alcohol or 10% formalin [17].
Children were classified as with or without sepsis by Goldstein et al.'s sepsis criteria [6]. For this purpose, axillary temperature > 38.5°C was defined as fever and < 36.0°C was defined as hypothermia. Tachypnea was defined as respiratory rate > 50 breaths/minute among children < 1 week of age, > 40 breaths/minute among children 1 week to 1 month of age, > 34 breaths/minute in children 1 month to 1 year of age, > 22 breaths/minute in children between 2 and 5 years of age, and > 18 breaths/minute in children between 6 and 12 years of age. Tachycardia was defined as heart rate > 180 beats/minute among children < 1 year of age, > 140 beats/minute in children between 2 and 6 years of age, and > 130 beats/minute in children 6 to 12 years of age. Bradycardia was defined as heart rate < 100 beats/minute among children 1 week to 1 month of age, and < 90 beats/minute in children between 1 month and 1 year of age. Abnormal WBC was defined as > 10% of immature neutrophils, or WBC count > 34 × 10 3/ mm 3 among children < 1 week of age, > 19.5 or < 5 × 10 3 /mm 3 in children between 1 week and 1 month of age, > 17.5 or < 5 × 10 3 /mm 3 in children 1 month to 1 year of age, > 15.5 or < 6 × 10 3 mm 3 in children between 2 and 5 years of age, and > 13.5 or 4.5 × 10 3 /mm 3 in children 6 to 12 years of age.
The risk factors studied included age, gender, malnutrition and severe malnutrition, sickle-cell disease, low birth weight, post-natal respiratory distress, and asthma. As part of the analysis, the software Anthro, versions 1.02 and 3.22 (WHO, Geneva, Switzerland), was used to perform the nutritional evaluation in accordance with the National Centre for Health Statistics, United States standard. A z score under -3.00 or -2.00 for the weight-for-age index defined malnutrition and severe malnutrition, respectively [18].
The primary outcome was death. Categorical variables were compared by using Chi-squared or Fisher's exact test as appropriate; continuous variables were assessed using Student's t test or the Mann-Whitney U test, taking into account the variable distribution. Multivariate logistic regression analysis by the enter method was used to assess independent association between sepsis and risk factors that significantly differed in the bivariate analysis. The multivariate analysis was performed in a model adjusted for age. Additionally, multivariate logistic regression analysis by the enter method was used to evaluate independent association between total WBC count sepsis in a model adjusted for age. Then, the receiver operating characteristic (ROC) curve was used to evaluate the ability of total WBC count (predictor variable) to differentiate sepsis from no sepsis (outcome variable). The statistical tests were two tailed, with a significance level of 0.05. The software SPSS version 9.0 (IBM, Armonk, USA) was used for the analysis. The study was approved by the ethics committee of the Federal University of Bahia.
The most frequent diagnoses were pneumonia (43.7%), diarrhea (17.3%), and cellulitis/adenitis (13.0%). All diagnoses are shown in Table 1. Table 2 demonstrates the comparison of each diagnosis frequency between cases with or without sepsis. No significant difference was found. Table 3 compares the frequency of potential risk factors between patients with or without sepsis. Notably, children with sepsis were older and had sickle-cell disease more frequently.
By multiple logistic regression, age was an independent risk factor for sepsis ( Figure 2) when WBC ≥ 17,250/mm 3 was the cutoff with best performance: sensitivity 72% and specificity 100%.
Overall, etiology was found in 57 (22.4%) cases: 32 (26.7%) with sepsis and 25 (18.7%) without sepsis (p = 0.1). The most frequent causative agents were S. pneumoniae (n = 9; 3.5%), parainfluenza virus (n = 9; 3.5%,) and respiratory syncytial virus (n = 8; 3.1%).    None of the cases with pneumococcal infection had sickle-cell disease. Table 4 depicts the comparison of the detected etiologies between children with and without sepsis. S. aureus infection was significantly more common among children with sepsis. Table 5 depicts the frequency of bacterial agents found by culture in each subgroup and per diagnosis.

Discussion
This study demonstrated the high frequency of sepsis (47%) among feverish patients admitted in to a pediatric ED during the enrollment period. This finding is in accordance with Ganjoo et al.'s prospective study [19], which detected a frequency around 46% of sepsis among feverish children hospitalized in a tertiary hospital in India by applying the same definitions of Goldstein et al. [6]. This high frequency was expected because of the high sensitivity of these sepsis criteria [20]. However, considering that EDs are often the first point of contact for children with sepsis [21], it is important to use objective criteria to identify children at risk. In addition, our study demonstrated that children admitted with fever, which met the early sepsis criteria of Goldstein et al. [6], were more prone to die. This is in agreement with Ganjoo et al.'s data, which showed a statistically significant higher fatality rate among patients with sepsis [19].
Acute respiratory infections and diarrhea are the most frequent childhood illnesses and causes of attendance at health services in low-and middleincome countries [22]. Pneumonia was the most frequent diagnosis among feverish children in this study. The high frequency of this disease (43.7%) could be explained because the majority of patients seen were younger than 1 year of age (54%) and pneumonia is highly incident in this age group [1]. Furthermore, it is important to recognize that we analyzed the initial diagnoses performed by assistant pediatricians. Our finding are in accordance with Murphy et al.'s data, which detected signs suggestive of pneumonia in 49% of 2,128 children admitted with fever to an ED [23].
In contrast with literature data [24], older age was a significant and independent risk factor for sepsis when risk factors were assessed in the present study. It is important to note that the majority of the sepsis studies were conducted at intensive care units, where patients with severe sepsis and septic shock, and late stages of sepsis continuum, were included. Our finding highlights an important difference between these two populations, because in EDs, the vast majority of children present with benign and self-limited viral infections [25]. In addition, in our study, most of the patients were younger than 1 year of age. According to Bertille et al., parents show different attitudes with respect to children's fever. These authors noted that younger children visit pediatricians more frequently than older ones, even if with self-limited viral infections, overusing emergency care providers [26]. It is possible that in our study, older children had been taken to the ED presenting more severe diseases. Sickle-cell disease was another significant and independent risk factor for sepsis among patients with documented fever. This finding is due to the lack or reduction of spleen function, particularly during the first 6 years of life [27]. We should remember that sepsis may be the initial event in the presentation of asyet-undiagnosed sickle-cell disease (since the majority of such events occur before 3 years of age) [28,29]. It is possible that the high prevalence of this condition in Brazil [30] highlights this disease as a risk factor for sepsis regionally.
The infection in the two groups had different etiology; S. aureus infection was significantly more frequent in septic patients (3.3% versus 0%; p = 0.049). It is known that S. aureus bacteremia (SAB) is a common and important infection in pediatric patients. The exact incidence of SAB is difficult to ascertain, as prospective population-based surveillance studies are infrequently performed. Age remains one of the most consistent predictors of mortality, with children generally having lower SAB mortality rates than adults [31]. The place of SAB onset influences outcomes, with community-onset episodes having a lower mortality rate, probably secondary to the predominance of skin and soft-tissue infections [32]. There were no deaths among septic children with staphylococcal infection in this study. Interestingly, when the two groups were compared, no difference in the initial site of infection was found.
Methodological constraints in this investigation should be emphasized. This was a single-center study, only reporting data from a subset of the pediatric age group. The data were collected retrospectively, patients were evaluated by different observers, and standardization of evaluations could not be guaranteed. However, strict criteria for enrolling and grouping the cases were used, and those with potential confounding variables were excluded. Moreover, the study was performed in a teaching hospital where the same standardized procedures for assistance were used over the period of the study.

Conclusions
This study provides evidence that the fulfilment of the formal sepsis criteria is frequent among pediatric patients seen with documented fever in an ED. Furthermore, children with sepsis showed differences in clinical profile; they were older, in contrast to the literature data. Sickle-cell disease was a significant risk factor independently associated with sepsis. Importantly, patients who fulfilled Goldstein et al.'s sepsis criteria [6] were more prone to dying. Further studies are necessary to characterize how sepsis presents in developing countries.