Introduction

Community acquired infection (CAI) is the leading indication for hospitalization in general paediatric wards in low and middle-income countries (LMICs) [1]. CAIs include all infections arising outside of the healthcare setting that are already present or incubating at the time of hospital admission. Common types of CAI include severe bacterial infections (bloodstream infection, pneumonia, meningitis, urinary tract infection, skin and soft tissue infection), gastroenteritis, respiratory virus infections, viral infection syndromes (e.g., herpes stomatitis, and varicella) and fungal infections.

Community-acquired pneumonia is the leading cause of child morbidity and mortality worldwide, with an incidence in LMICs of 0.22 episodes per child-year [2]. The main risk factors for development of CAP are young age, malnutrition, low birth weight and Human Immunodeficiency Virus (HIV) infection. Globally there has been a 30% decrease in the incidence of CAP from 2010 to 2015, mostly attributed to prevention of vertical HIV transmission and improved pneumococcal conjugate vaccine (PCV) uptake [3]. S. pneumoniae remains the most important bacterial pathogen, although the incidence has declined significantly following introduction of the PCV-7 and more recently PCV-13 [4]. Owing to the difficulty in obtaining bacteriological identification in cases of childhood CAP, antibiotic treatment is usually presumptive.

Respiratory Syncytial virus (RSV) is the most common viral pathogen in children and is especially prevalent in the winter months in temperate climates. In the Drakenstein study in South Africa, RSV was the most common viral pathogen identified in hospitalized children with respiratory illness, although the bacterium B. pertussis was also frequently isolated. Between 2011–2016 24.1% and 5.5% of 3645 hospitalized children aged of 0–59 months, tested positive for RSV and influenza virus infections, respectively. The large overlap in the clinical, laboratory and radiological findings in bacterial versus viral CAP, has led to recommendations for routine antibiotic use in young children with severe CAP [5, 6].

The World Health Organization (WHO) developed the Integrated Management of Childhood Illness guidelines (IMCI) for use in primary care facilities to assist with triage of ill-children based on the severity of their presenting symptoms and signs (S1 Appendix). For acutely ill children requiring transfer from primary care to hospital, the IMCI guidelines and the South African Essential Drug List (EDL) recommends a pre-transfer dose of intravenous or intramuscular third-generation cephalosporin e.g., ceftriaxone. Oral amoxicillin is first line therapy for non-severe pneumonia. Ampicillin plus gentamicin is recommended for infants less than 3 months of age with severe pneumonia and ceftriaxone for older children with severe pneumonia (S2 Appendix). Oral amoxicillin is the first line therapy for non-severe pneumonia as it provides adequate cover for the most common CAP pathogens (Streptococcus pneumoniae and Haemophilus influenzae). However, amoxicillin has little activity against enteric Gram negative bacteria e.g. salmonella, that are important causes of paediatric CAP in some parts of Africa [7].

Gastroenteritis (GE) is the second most common cause of childhood morbidity and mortality globally with an estimated 1.7 billion cases of diarrhoea in 2010 [8]. While rotavirus is the most prevalent GE pathogen worldwide, the incidence of rotaviral-associated GE death and hospitalization rate has substantially declined in South Africa and the continent since inclusion of the rotavirus vaccine in the national immunization schedule in 2009 [9–11]. Antibiotics are not routinely recommended for diarrhoeal disease as the causative pathogens are mostly viruses. Antibiotics are used in certain situations such as severe malnutrition and for infants less than 28 days (ampicillin and gentamicin). In cases of suspected dysentery, ceftriaxone or ciprofloxacin is the EDL- recommended treatment.

Febrile illnesses are a common presentation in young children requiring hospital admission. Severe bacterial infections including bloodstream infection (BSI) or bacteremia, meningitis, and urinary tract infections, typically present as a febrile illness in infants and young children. For this reason, most febrile infants undergo blood culture sampling, urine screening and in some cases, a lumbar puncture on hospital admission. BSI may occur in isolation (so-called primary BSI) or may be secondary to infection at another site e.g., S. pneumoniae BSI with community-acquired pneumonia (CAP) or meningitis, or E. coli BSI arising from a urinary tract infection (UTI). A study profiling trends in paediatric BSI at Tygerberg Hospital (2008–2013) showed that Gram negative infections predominated, with E. coli and K. pneumoniae being the most CA-BSI common pathogens [12].

Meningitis is another common indication for laboratory investigation and hospitalization in children. The diagnosis of meningitis in children relies on clinical suspicion and the correct interpretation of appropriate laboratory test results, if available. Aseptic meningitis remains the predominant type of meningitis reported in the literature [13] although this could partly be attributed to the administration of pre-hospital transfer antibiotics given according to the IMCI guidelines. Empiric treatment of suspected meningitis is ampicillin plus cefotaxime in infants < 3 months of age, with ceftriaxone alone used for older infants and children. Urinary tract Infections are most common in the first 6 months of life. Symptoms of UTI are often non-specific, especially in infants, therefore it is important to exclude the diagnosis in all febrile children. Diagnosis is made on urine specimens taken aseptically. Risk factors for development of UTI include anatomical urinary tract abnormalities and female gender. Common UTI pathogens include the Enterobacteriaceae family with E.coli as the most common pathogen overall (70–90% of cases). Staphylococcus aureus and Candida albicans are other important pathogens [14]. For non-severe urinary tract infections in infants and children > 3 months old, the recommended therapy is oral amoxicillin-clavulanic acid. Intravenous amoxicillin-clavulanic acid or ceftriaxone was the EDL-recommended treatment for severe UTI and UTI in infants <3 months of age.

In high-income settings, laboratory investigations are routinely submitted to identify the causative CAI pathogen/s (bacterial, viral, fungal or other), however in many parts of Africa, microbiological diagnostic services are unavailable and the diagnosis of CAI is made clinically. Conducting laboratory tests may assist with clinical decision-making including the need for, type and duration of antimicrobial therapy. Although some data exists on certain CAI pathogens in South African children, there is limited information on the CAI burden, clinical management and outcome of hospitalized children. In addition, the recent introduction of several key vaccines targeting CAI pathogens (rotavirus, pneumococcal conjugate vaccine) has changed the profile of vaccine preventable diseases and paediatric hospitalizations. Understanding the burden of CAI and the profile of causative pathogens may assist with developing guidelines for appropriate investigation and antimicrobial management, ultimately contributing to improved childhood health outcomes. In this study we describe the burden, spectrum, pathogen profile, clinical management and outcome of children hospitalized for CAI in a general paediatric ward. In addition, we analyze the cost and appropriateness of diagnostic laboratory testing performed and empiric antimicrobial therapy prescribed for CAI in a cohort of hospitalized South African children.

Methods

Results

A total of 451 children were admitted to the general paediatric ward over the 6-month study period. Of these admissions, 364/451 (81%) were hospitalized with a primary diagnosis of one/more CAIs. The spectrum of CAI episodes included community-acquired pneumonia—CAP (54%; 197), diarrhoeal disease (14%; 51), mixed infections with >1 infection type (12%; 43), bloodstream infection (BSI) (9%; 33), meningitis (5%; 17), UTI (2%; 8), and other infection types (4%; 15)Among children with CAI, the median age was 4.8 months (IQR 1.5–17.5), with a male predominance and an HIV-infection prevalence of 6.0% (22/364) Severe acute malnutrition (SAM) was present in 29/364 (8%). Underlying medical and surgical conditions were frequent (46/364; 12%), including neurological disorders/cerebral palsy (17; 5%), congenital cardiac lesions (7; 2%), chronic lung disease (7; 2%), chromosomal abnormalities (6; 2%), chronic kidney disease (4; 1%), sickle cell disease (3; 1%) and suspected primary immune deficiency (2; 1%). More than half of the children (225/ 364; 62%) with CAI were transferred in from other health facilities; 139 (38%) came directly from home. Of the referred children, 152 (68%) had received pre-hospital empiric antibiotic therapy. (Among the 22/364 (6%) of children living with HIV, 14 (64%) were already on antiretroviral therapy (ART) but only 6 (43%) had achieved virological suppression (HIV RNA <50 copies/ml); their median CD4 cell count was 451 (IQR: 371–1501). Pneumonia was the most frequent admission diagnosis 15 (68%) among HIV-infected children, and three-quarters (17/22; 75%) had received recent or current treatment for confirmed tuberculosis disease (pulmonary, meningitis or disseminated TB); 15/17 (88%) for drug-susceptible TB. (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics and outcomes of children hospitalized with community-acquired infection (n = 364).

https://doi.org/10.1371/journal.pone.0272119.t001

Clinical haematology (White cell count, platelets), chemistry (C-reactive Protein), microbiology and virology laboratory test utilization rates were high with 360/364 (99%) children having had one/more laboratory test/s submitted. A total of 1206 tests were performed; patients who presented with pneumonia accounted for the most diagnostic tests submitted overall (522/1206, 43%). Of 274 blood cultures and 140 cerebrospinal fluid samples submitted, only 5% and 2% respectively yielded a pathogen.

Respiratory syncytial virus (RSV), adenovirus and parainfluenza virus predominated from shell vial culture tests, whereas RSV was the most frequent virus identified on respiratory virus polymerase chain reaction testing (RV PCR). Most children with CAI (303/364; 83%) were considered to be appropriately investigated; only 9/364 (3%) had inappropriate investigations and 52/364 (14%) had no investigations performed (which in most cases was deemed appropriate as they presented with probable viral illnesses).

In 323/364 (89%) patients, empiric antibiotic usage was EDL guideline compliant. Common CAI antibiotic treatment regimens included: ampicillin alone (53%); ampicillin plus gentamicin (25%) and ampicillin plus cefotaxime (20%). For 17/364 (5%) patients, prescriptions were not considered to be guideline compliant. This included administration of antibiotics in children presenting with acute gastroenteritis with no risk of a bacterial infection. Of those on empiric therapy (n = 340), antibiotics were stopped in 169/340 (50%) patients after clinical improvement and negative laboratory tests. De-escalation of antibiotic therapy, such as change from ampicillin plus cefotaxime to ampicillin plus gentamicin was noted in 38/340 (11%) of cases. In 39/340 (12%) children who needed antibiotic escalation, all patients received appropriate escalation based on clinical presentation and laboratory results (Tables 2 and 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Laboratory investigations and pathogen yield children with community-acquired infection.

https://doi.org/10.1371/journal.pone.0272119.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Infection diagnostics and antimicrobial therapy appropriateness and guideline compliance.

https://doi.org/10.1371/journal.pone.0272119.t003

Outcomes of paediatric hospitalization for CAI were generally favourable, however 3 children (1%) died from E.coli sepsis, pneumonia (with underlying cerebral palsy and hydrocephalus) and septic shock (underlying spastic cerebral palsy and disseminated TB). Fourteen children (3.8%) required re-admission to hospital within 30 days of discharge, with age ≤ 3 months and a history of preterm birth as risk factors in this group (Table 1).

In this study population, hospital costs for CAI was estimated to be just over ZAR 8 million. Community acquired pneumonia contributed 49% to the estimated cost followed by diarrheal disease at 15% (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. The estimated cost of hospitalization, investigation and treatment of paediatric community-acquired infections.

https://doi.org/10.1371/journal.pone.0272119.t004

Discussion

In this study, we characterized the epidemiology of CAI in South African children admitted to a general paediatric ward in Cape Town. The patient population was young, predominantly male (in keeping with published data on paediatric CAI patient demographics) and had a relatively low HIV prevalence rate when compared to previous SA hospital-based studies [12, 17]. Patients with CAI comprised 81% of all ward admissions, highlighting the heavy burden of infectious diseases in paediatric practice in South Africa. Common underlying factors in children hospitalized for CAI were prematurity, malnutrition, and young age.

Although a broad spectrum of CAI types was documented, pneumonia and GE remained the leading causes of paediatric hospitalization. Only one child with CAP had a confirmed S. pneumoniae infection. The steady decline in cases of invasive pneumococcal disease in South African children can be attributed to the introduction of the PCV vaccine, increased uptake of paediatric antiretroviral therapy and effective prevention of mother to child HIV transmission (PMTCT) [4, 18]. Prematurity was the most frequent underlying factor in children admitted with CAP. RSV was the most frequently isolated viral pathogen followed by adenovirus and CMV, as has been shown in previous studies [2, 5]. Protection of this vulnerable population from RSV infection is difficult, with Palivizumab immunoprophylaxis beyond the means of public sector patients in South Africa. Introduction of a RSV vaccine in future is likely to have a dramatic impact on the profile of paediatric admissions with CAP, particularly for ex-premature infants.

Although rotavirus is the leading cause of GE globally, it was not isolated on stool from any patient in this cohort, possibly owing to the small number of stool specimens collected. Previous studies have highlighted a substantial decline in rotavirus GE associated hospitalization and mortality following introduction of the rotavirus vaccine to the SA-EPI in 2009 [9].

Aseptic meningitis remains the most common type of meningitis encountered in children, highlighted by the very low yield of CSF in this cohort (2%). Failure to isolate bacterial pathogens as a result of antibiotic usage prior to CSF sampling could not be excluded. Although the CSF yield was low, submission of specimens for bacterial culture and viral PCR is useful to identify viral causes of meningitis, where antibiotics can be safely discontinued.

The pathogen profile of UTIs in paediatric populations in South Africa is not well-described. Our study identified E.coli and K. pneumoniae as the most frequent uropathogens with many being ESBL-producers (38%). Of the patients treated for UTI, 50% received 3rd generation cephalosporins and only 3% received oral amoxillin-clavulanic acid. Regular review of empiric antibiotic guidelines for CA-UTI is needed considering the high rates of resistant pathogens encountered, even among CAI bacterial pathogens.

The most prevalent bloodstream pathogens in this study population were E. coli, pathogenic streptococci and non-typhoidal salmonella. This profile differed from a previous study (2009–2014) at the same hospital which noted that K. pneumoniae, S. aureus and E. coli were the most frequent paediatric bacteremia pathogens. The high rate of pre-transfer antibiotics (46%) in this cohort may have reduced the observed blood culture yield. Blood culture contamination rates in this study (7%) approximated that reported at another tertiary hospital in our province (6.6%), but were substantially lower than that reported from a Nigerian hospital [19]. The major blood culture contaminants in the studies were coagulase negative staphylococci, highlighting the importance of proper skin preparation before sampling of blood is done.

There is a need to produce local guidelines that emphasize judicious use of laboratory investigations to most effectively utilize the limited healthcare resources available in our middle-income country setting. Cautious use of laboratory investigations and antimicrobials will help reduce the overall cost of hospital management for CAI. In this study, the community acquired pneumonia and GE were leading contributors to CAI cost (49% and 15% respectively), with overall costs estimated at 8 million ZAR (approximately 500 000 USD).

Empiric antimicrobial prescriptions for CAI episodes at our institution were generally compliant with national guidelines (90%), including stewardship best practices such as prompt stopping of antibiotics, and de-escalation or escalation of antibiotics based on clinical picture and laboratory findings. An area for improved antimicrobial stewardship practice identified in this study is prescription of antibiotics to children with GE and no risk factors for bacterial infection. Furthermore, frequent review of empiric antibiotic guidelines (at least annually) is a clear priority given the high rates of antibiotic resistant pathogens encountered in paediatric CAI at our institution.

Limitations of this study include lack of generalizability (inclusion of data from a single, tertiary paediatric center), lack of data on the routine immunization status of the cohort, and the possible effect of pre-hospital administration of antibiotics on culture yield. Strengths of this study are the inclusion of detailed clinical and demographic data, access to advanced diagnostic microbiology and virology services, and inclusion of data seldom reported in studies of paediatric CAI from LMIC such as appropriateness of antimicrobial therapy, laboratory diagnostic use and cost of hospital management. Future studies should incorporate evaluation of antimicrobial and diagnostic test use for management of paediatric CAI, especially from LMIC where these data are seldom reported.

Conclusion

CAI was the most common reason for paediatric hospitalization, driving antimicrobial use, hospital and laboratory costs. Empiric antibiotic prescription practice for CAI was generally compliant with national guidelines. Improved diagnostic stewardship is needed to reduce culture contamination rates, improve pathogen yield, and decrease unnecessary antimicrobial use. Clinical outcomes of paediatric CAI episodes were generally favourable.

Supporting information

Acknowledgments

The authors thank the patients, their parents and the clinical staff of the Tygerberg Hospital general paediatric ward G7.