Sub-Saharan Africa continues to have high under-five mortality rates fuelled by the burdens of malnutrition and multiple childhood infections.^1,2,3 Gram-negative Enterobacteriaceae infections are of particular concern because of an increasing incidence and the emergence of broad-spectrum antibiotic resistance, limiting the available treatment options.^4,5 The extent of Gram-negative Enterobacteriaceae infections has been well documented in adults, but this is not the case in children, especially in populations with high HIV and malnutrition rates.^6,7

Severely malnourished children have a threefold higher case fatality than children with normal nutritional status.⁸ Although the current standard management guidelines for children hospitalised with severe acute malnutrition (SAM) recommend empiric antibiotics (ampicillin and gentamicin or ceftriaxone for severe illness), these antibiotics may be less effective in treating Enterobacteriaceae infections.^9,10 Understanding the incidence and antimicrobial resistance patterns of Gram-negative isolates from malnourished children will better inform antibiotic usage in malnourished children and will influence the choice of antibiotics in this vulnerable population.

Invasive bacterial enteric pathogens, including Shigella, Campylobacter and Enteroaggregative-adherent E. coli (EAEC), have been associated with gut inflammation and barrier disruption triggering chronic diarrhoea that can lead to malabsorption and malnutrition.¹¹

Multiple factors such as organ or stem cell transplant, intensive care unit admission, prolonged hospitalisation, surgery and care in long-term care facilities have been highlighted as risk factors for infection with drug-resistant Enterobacteriaceae in adults.¹²

In a review summarising six studies of carbapenem-resistant Enterobacteriaceae infections, 64 isolates were cultured from 63 children. The median age of infected children was 1 year (0–17 years), males (67%) were affected more than females. Many children (53%) were hospitalised to an intencive care unit (ICU) and probably had hospital-acquired infections. A total of 87% had been hospitalised for more than 48 h. Comorbidities identified included pulmonary disease, prematurity, leukaemia or solid tumours, cardiac disease, necrotising enterocolitis, solid organ or stem cell receipt.¹³

Extended-spectrum β-lactamase (ESBL) producing Enterobacteriaceae have emerged as a major cause of healthcare-associated infections with significant morbidity and mortality in children.^12,13.

In populations with high HIV endemicity, children are hospitalised with poor nutritional status and opportunistic infections, including tuberculosis. The relationship to Enterobacteriaceae infection has not been completely described and explained.

There are limited therapeutic options available for multidrug resistant Gram-negative bacterial infections in children.⁷ Ideal treatment regimens are hampered by a scarcity of randomised trials assessing the efficacy of various drug options in children with community-acquired Enterobacteriaceae infections.^14,15 Morbidity and mortality associated with these infections are high. Reports on epidemiology and risk factors for infection in children are limited.^7,12,13,15

This study aimed to describe the clinical characteristics of children who were hospitalised at King Edward VIII Hospital (KEH), Durban, from whom an Enterobacteriaceae was cultured.

A retrospective, descriptive, observational study of all children hospitalised at KEH, who cultured an Enterobacteriaceae from any site, was conducted from the period January 2016 to 31 December 2017. All hospitalised children are admitted to an acute admission ward under the care of a specialist paediatrician. Decisions on specimen collection for specific microbiological cultures were made at the discretion of the treating clinician.

The National Health Laboratory Service (NHLS) was used to access all positive Enterobacteriaceae cultures, from any site, in children between the ages of > 28 days and < 13 years for the study period. This study excluded neonates (who have a higher Gram-negative burden of infections) and reflects a general hospitalised paediatric population with an expected lower burden of such infections.

The patient files of all children who had positive Enterobacteriaceae cultures were sourced from the KEH archives, and the following data were extracted: age, gender, final clinical diagnosis, HIV status, nutritional status, clinical outcome, mortality and length of stay (LOS). Anthropometric measurements from the patient files were used to classify nutritional status using World Health Organisation (WHO) growth standards. Patients were classified: no malnutrition (NAM), moderate acute malnutrition (MAM), severe acute malnutrition (SAM) (with or without oedema) and overweight for age (OWFA).¹⁶

The HIV status of infants was determined from the last verified HIV test carried out within the health system and results were also verified on the NHLS information system. They were categorised as: HIV-infected, HIV-uninfected, HIV-exposed uninfected and HIV-exposed unknown. For the exposed group without confirmation of HIV results, these patients were categorised as unknown (no confirmation of HIV results). Patients infected with HIV were further categorised as receiving antiretroviral treatment (ART) or not. Patients with an unknown HIV result (107/5946; 1.8%) were excluded from the non-descriptive analysis.

Information of the sites of culture, types of organisms and antibiotic sensitivity were extracted and verified from the NHLS information system. All data were extracted into an excel spreadsheet.

Data from the study cohort was compared with routinely collected data on all children hospitalised to the paediatric medical wards at KEH during the study period.

A total of 5946 children above 1 month were hospitalised to King Edward Hospital Paediatric Department during the study period 01 January 2016 and 31 December 2017. In this period, 207 of these hospitalised children (3.5%) had positive Enterobacteriaceae cultures, of which 59.4% were community-acquired (defined as positive cultures less than 48 h after admission) and 40.6% were hospital-acquired defined as positive cultures more than 48 h after admission.

Figure 1 illustrates the most common organisms and the sites from which they were cultured. Escherichia coli (109; 44.5%) and Klebsiella spp. (59; 24.1%) were the most commonly isolated organisms. The positive cultures were obtained from urine (135/234; 57.7%), stool (33/234; 14.1%), blood (33/234; 14.1%), sputum (21/234; 9%), skin swabs (10/234; 4.3%), CSF (1; 0.4%) and one site (1; 0.4%), which was unknown E. coli was isolated predominately from urine and Klebsiella spp. from sputum. Extended-spectrum β-lactamase producing organisms were found in 51 patients, 33 (65.0%) of whom were malnourished and 18 (35.0%) HIV-infected. Only two of the 207 patients cultured extensively drug-resistant organisms (defined as non-susceptibility to at least one agent in three or more antimicrobial categories). These were both Klebsiella spp. cultured from urine. Of the 59 isolates that were cultured ESBL producers, these were: Klebsiella spp. (27/59; 45.8%), E. coli (26/59; 44.1%) and the remaining 10% were Shigella (2/59), both from stool samples and Morganella (1/59) cultured from sputum.

This study evaluated the clinical characteristics of hospitalised children with a positive Enterobacteriaceae culture. Human immunodeficiency virus and malnutrition were significant risk factors for Enterobacteriaceae infection.

The association of HIV infection and increased risk for Enterobacteriaceae infection in children is not well understood, especially in the context of improved ART access.^1,2 Human immunodeficiency virus-exposed infants may have lower body weight, an ill mother and thus may be at risk for poor growth trajectories and increased infections. Higher infection rates with Enteropathogenic E. coli (EPEC), Cryptosporidium and increased gut colonisation with ESBL Enterobacteriaceae have been found in HIV-infected children.^17,18,19,20 In our study, the incidence of Enterobacteriaceae infection was higher in children who were yet to be initiated on ART compared with those already on lifelong ART. Being ART naïve is probably associated with a compromised immune system. The independent association of HIV and malnutrition with Enterobacteriaceae infection in this study highlights the contributing role of these factors to infection risk.

In this cohort, children with SAM with or without oedema were more susceptible to Enterobacteriaceae infection. Several studies have described that patients with malnutrition are more vulnerable to Enterobacteriaceae infections.^21,22,23,24 Environmental enteric dysfunction (EED), villous blunting, reduction in mucus-secreting goblet cells, inflammation, reduced microbial diversity and increased potent pathogenic Enterobacteriaceae in children with malnutrition have been postulated as potential causes and warrant further study.^{25,26,27,28,29}

In keeping with other studies, the authors found that diarrhoeal disease was strongly associated with Enterobacteriaceae infection.^24,30,31 These findings suggest that Enterobacteriaceae infection, malnutritions and diarrhoea are possibly intertwined.

The incidence of Enterobacteriaceae infection in this study (3.5%) is similar to reports from other studies.^15,32,33 The higher incidence of Enterobacteriaceae infections in children under five has also been previously reported, and the role of perineal hygiene and toilet training are considered possible contributory factors.³⁴

Positive cultures for Enterobacteriaceae infections were predominantly from urinary samples. Several studies have demonstrated that Enterobacteriaceae organisms are the dominant Gram-negative organism causing UTIs in children.^35,36,37 Urinary tract infections in children often present with non-urinary tract symptoms and are frequently diagnosed during investigations for sepsis, diarrhoea or malnutrition.³⁸ The association between UTIs and malnutrition has been documented and the authors of this study confirms this association.

Enterobacteriaceae are the leading group of Gram-negative organisms causing UTIs in children, which has implications for the empiric choice of antibiotics.³⁹ The South African Standard Treatment Guidelines recommend using amoxicillin/clavulanic acid to treat UTIs and ampicillin and gentamicin to treat children with complicated SAM or oral amoxicillin for uncomplicated SAM.¹⁰ The emergence of increasing antibiotic drug resistance, especially in hospitalised children with malnutrition, is of concern. The guidelines may need to be revised.

Increasing antibiotic resistance amongst urinary tract isolates against ampicillin, amoxicillin and cotrimoxazole has been reported. Most isolates are still sensitive to third-generation cephalosporins, quinolones, macrolides and aminoglycosides.⁴⁰ Careful monitoring of antimicrobial resistance in high-risk sub-populations, such as severely malnourished children, is warranted.

Of the 207 Enterobacteriaceae cultured, 51 (24.6%) produced extended-spectrum beta-lactamases (ESBL) in mostly community-acquired infections.

This is in keeping with other studies, where an incidence of 10% – 35% was described.^7,17,33 A recent study demonstrated the correlation between the routine use of amoxicillin in children with uncomplicated SAM and increased ESBL production.⁴¹ Amoxicillin use is recommended as one of the WHO ‘10 steps’ in managing malnutrition.⁴² Children exposed to amoxicillin showed an increased ESBL production amongst isolates and a higher risk of transmission to siblings.⁴¹ Further research on the use of broad-spectrum antibiotics amongst malnourished children is required.^43,44,45

This study was conducted in an area with a high prevalence of HIV infection. Enterobacteriaceae infections are relatively common in hospitalised children, particularly in severely malnourished children with oedema and HIV-infected children not on ART. A high index of suspicion is necessary when caring for hospitalised children with these diagnoses. Revision of current antibiotic recommendations for hospitalised children with malnutrition may be necessary if current guidelines result in treatment failure related to antimicrobial resistance. Strategies to prevent malnutrition and HIV infection in children should be prioritised to reduce the risk of Enterobacteriaceae infections, mortality and prolonged hospital stay.