Background: Evaluation of children with fever without localising signs (FWLS) has barely changed in the USA since 1993 despite reduced invasive disease after the introduction of Haemophilus influenzae type b conjugate vaccine and conjugate pneumococcal vaccine (PCV7). PCV7 is now recommended in the UK for children under 2 years of age, and new NICE guidelines have been issued for managing feverish children in the UK in anticipation of PCV7’s efficacy. We compared rates of bacterial infections in children aged 3–36 months with FWLS in the pre- and post-PCV7 eras to define current trends and evaluate existing guidelines.
Methods: We identified all paediatric blood cultures performed in an emergency department before and after PCV7. We subsequently identified all children aged 3–36 months with FWLS and reviewed their medical records.
Results: We identified 148 patients with FWLS in the pre-PCV7 period and 275 patients after PCV7. There were 17 positive cultures before PCV7 (10 pathogens and seven contaminants) and 14 positive cultures (but only one pathogen) after PCV7. This represented a 94.6% decrease overall (p = 0.009) and a 100% decrease in Streptococcus pneumoniae. Rates of urinary tract infections (UTIs) were unchanged (6.8% vs 7.6%); UTIs are now the most prevalent bacterial infection in this group. Over 50% of patients still received empirical antibiotics.
Conclusions: Based on our data, the emphasis in managing children with FWLS should be on diagnosing UTI. Guidelines for evaluating children with FWLS in countries using PCV7 should emulate the NICE model and reflect the trends identified in this study.
Statistics from Altmetric.com
Fever accounts for up to 17% of all paediatric emergency department visits, and historically has been the most frequent chief complaint in children under 3 years of age.1 2 Twenty per cent of these children have fever without a localising source of infection (FWLS).3 Prior to 1990, 3–12% of children under 3 years of age with FWLS had occult bacteraemia.3–6 Patients with occult bacteraemia can develop other serious bacterial infections (SBI), including sepsis, meningitis, pneumonia, osteomyelitis and septic arthritis.4 Urinary tract infections (UTIs) frequently present as FWLS in children: 3–4% of boys under the age of 1 year and 8–9% of girls younger than 2 years with FWLS have UTIs.7–9
What is already known on this topic
Rates of invasive pneumococcal disease have been greatly reduced following the introduction of PCV7.
Historically, children with fever without a localising source of infection are also at risk of urinary tract infections in addition to occult bacteraemia.
What this study adds
Urinary tract infection is the most common bacterial infection identified in children with fever without a localising source of infection (FWLS).
This study validates the UK guidelines that focus on diagnosing urinary tract infection for the management of children with FWLS.
US guidelines that still concentrate on the detection and treatment of occult bacteraemia should be changed as this invasive disease is now rare.
In evaluating a child with FWLS, clinical and laboratory data are used to estimate the risk of SBI. Guidelines proposed in 1993 recommended that a child aged 3–36 months with a fever of 39.0°C or more and a white blood cell count of ⩾15 000/mm3 should have a blood culture performed and receive antibiotics and that selected groups should be evaluation for UTI.6
Since 1993, several major developments have significantly changed the aetiology of FWLS in children. The first was the introduction of the Haemophilus influenzae type b (Hib) conjugate vaccine, which reduced the incidence of invasive disease by over 90% in industrialised countries.10–12 After Hib, most cases of occult bacteraemia were caused by Streptococcus pneumoniae. Prior to the implementation of the pneumococcal conjugate vaccine (PCV7; Prevnar, Wyeth Pharmaceuticals, Collegeville, Pennsylvania), the seven serotypes covered by the vaccine caused 80% of invasive infections in the USA and Canada.13 Since 2000, when PCV7 was approved and implemented in the USA, there has been a 66% decrease in the rates of pneumococcal bacteraemia and a 50% reduction in other invasive disease.14 15 Most changes have been in children younger than 3 years of age.16 17 Given the rates of invasive disease in the UK, PCV7 was approved for use there for all children under 2 years beginning in 2006.18
Despite the fact that it was anticipated that PCV7 would eliminate most pneumococcal occult bacteraemia in the USA, clinical recommendations for children with FWLS have been largely unchanged since 1993.19–21 In contrast, the National Institute for Health and Clinical Excellence (NICE) in the UK recently issued new guidelines for the management of feverish children.22 The NICE guidelines recommend more conservative management of these children with FWLS since the advent of PCV7, and call for investigation of the urine, no blood cultures and no empirical antibiotics.
We set out to investigate the frequency of occult bacteraemia, UTI and antibiotic administration as part of the assessment of children aged 3–36 months with FWLS in the pre- and post-PCV7 eras in an effort to determine what management is appropriate in these patients.
This research project was reviewed and approved by the Duke University Institutional Review Board. Using the clinical microbiology laboratory database, we identified all children less than 3 years of age seen in the urgent/emergency care setting at Duke University Hospital (Durham, North Carolina) who had a blood culture performed in the years before (1997–1999) and after (2001-mid 2004) PCV7 institution. The year 2000 was specifically excluded because PCV7 was licensed for use by the FDA on 7 February 2000; we therefore chose to exclude the entire year due to the transitional phase of implementation. Our post-PCV7 group belonged to the period when routine administration of PCV7 for all infants and young children was recommended.23
Children under 3 months of age were excluded because they are not fully immunised against S pneumoniae and have a different risk profile for occult bacteraemia and SBI. We analysed the records of the subgroup of children aged 3–36 months who presented with FWLS. Patients were determined to have FWLS if they presented without significant underlying illness or past medical history, were non-toxic appearing, and had no apparent source of infection after a thorough physical examination as determined by emergency department medical staff. Patients with otitis media were excluded. Patients were considered febrile if they had an examination temperature of 39.0°C or higher recorded in the emergency department without regard for anti-pyretic administration. Patients who were afebrile in the emergency department but had a history of documented fever were considered to be febrile to the degree reported. Patients with incomplete medical records were excluded. Complete records included documented fever, presenting symptoms, physical findings and culture results. More than one culture may be included per patient in the study period if they were performed at separate visits. Rectal temperatures were recorded if possible.
Clinical presentation, laboratory tests and disease outcomes were compared between the two groups. Electronic records were reviewed and included physician dictations, discharge summaries, information on prior and subsequent visits, and all laboratory tests. We recorded the age, sex, past medical history, duration of fever, presenting symptoms, maximum temperature, temperature upon examination, and results of a complete physical examination. We also recorded laboratory tests including complete blood cell count with differential and all blood culture results as well as any radiological studies, lumbar punctures, urinalysis, urine cultures, direct respiratory viral antigen tests and respiratory viral cultures. Blood cultures considered pathogens were those that isolated organisms consistently reported to cause disease in children: S pneumoniae, Haemophilus influenzae, Moraxella catarrhalis, Staphylococcus aureus, Enterococcus sp, Salmonella sp, Neisseria meningitidis and group A β-haemolytic streptococci. Organisms isolated that were considered contaminants included coagulase-negative Staphylococci, Bacillus sp, mixed oropharyngeal flora, diphtheroids and Streptococcus viridans. Only pathogens were considered occult bacteraemia. A positive urine culture was defined as bacterial growth from a urine specimen obtained by catheter or a clean voided specimen with 10 000 CFU/ml or greater of a pathogenic organism.
For the two groups, antibiotic administration was analysed as well as patient outcomes (admission, discharge, revisit or call back) after the emergency department visit.
Statistical comparisons were performed using a two-tailed t test and χ2 testing for all numerical and categorical values, respectively. Significance was set at p⩽0.05. We used an open-source, web based program (www.openepi.com).
In the pre-PCV7 era of 1997–1999, there were 13 507 total visits to the paediatric emergency department with 1251 patients between the ages of 3 and 36 months having blood cultures obtained for any reason. Of those 1251 patients, 1184 had complete records (94.6%). In the post-PCV7 era of 2001 to mid-2004, there were 21 500 total visits and 2028 patients between the ages of 3 and 36 months who had blood cultures performed for any reason. Of those 2028 patients, 1994 patients had complete records (98.3%). The majority of incomplete records failed to include a dictated report of the emergency room visit. After examination of the records and application of the criteria for FWLS described above, in the pre-PCV7 era 148 of these patients fit the criteria for FWLS (11.8%). There were 275 patients in the post-PCV7 era with FWLS (13.6%). These data are shown in detail in table 1. These two groups of patients with FWLS were then analysed.
There were 17 positive blood cultures in the pre-PCV7 era: 10 were pathogens and seven were contaminants. The pathogens consisted of six S pneumoniae, one S aureus, one Streptococcus pyogenes and two M catarrhalis. The contaminants consisted of three coagulase-negative staphylococci, three Bacillus sp and one mixed oropharyngeal flora. In the post-PCV7 era, there were 14 positive blood cultures: only one was a pathogen and 13 were contaminants. The pathogen consisted of one Enterococcus sp. The contaminants consisted of nine coagulase-negative staphylococci, two S viridans and one diphtheroid. Before PCV7, the prevalence of S pneumoniae was 6/148 (6.7%; 95% CI 1.3% to 9.5%), while after PCV7, the prevalence was 0/275 (0%; 95% CI 0% to 1.3%). S pneumoniae was the most common pathogen accounting for 60% of pathogen cases before PCV7.
Occult bacteraemia affected 6.8% (95% CI 3.6% to 12.1%) of patients with FWLS before PCV7 and 0.4% (95% CI 0.0% to 2.2%) of patients with FWLS after PCV7, a 94.6% decrease in occult bacteraemia (p<0.001). S pneumoniae occult bacteraemia affected 4.1% of patients with FWLS in the pre-PCV7 era and 0% of patients with FWLS in the post-PCV7 era, representing a 100% decrease in S pneumoniae (p = 0.001). The percentage of patients with contaminant cultures did not change from before PCV7 to after PCV7 (4.7%). In the post-PCV7 era, the likelihood that a positive culture would be a contaminant increased from 41.2% to 92.9%.
As part of our analysis, we also investigated the results of urine testing in this group. These data are summarised in table 2. In the pre-PCV7 era, there were 10 positive urine cultures affecting 6.8% of patients with FWLS. In the post-PCV7 era, there were 21 positive urine cultures affecting 7.6% of patients with FWLS. These differences were not statistically significant (p = 0.9). Females comprised 60% and 52.4% of patients with positive urine cultures in the pre- and post-PCV7 eras, respectively. The average age of the children with positive cultures was 14±9 months before PCV7 and 8±6 months after PCV7 (p = 0.08). It is noteworthy that not all of these children had their urine investigated. Again, there were few differences between the two eras: approximately 75% and 77% of children with FWLS, respectively had a urinalysis before and after PCV7 and 55% and 61%, respectively, had a urine culture performed before and after PCV7.
Antibiotics were given to the majority of patients with FWLS in both eras. In the pre-PCV7 era, 60.8% of patients received antibiotics and in the post-PCV7 era, 57.2% received antibiotics. More than 90% in both groups received their antibiotics at their initial emergency department evaluation.
In this study, we described a sharp decrease in the frequency of bacteraemia in children with FWLS who had blood cultures performed. The pre-PCV7 published rate of occult bacteraemia in previously healthy non-toxic appearing children with FWLS was 1.32–6.1%.4 11 19 24 25 Our results support these observations as the pre-PCV7 rate of occult bacteraemia in our population was 6.7%. By 2001–2003, the published occult bacteraemia rate had dropped to 0.7–0.9%.26 27 Our results that showed the incidence of occult bacteraemia in children aged 3–36 months with FWLS who had a blood culture performed was 0.4%. Thus, in this group of patients, the incidence of bacteraemia with a pathogen has declined from one in every 14 children assessed to one in 275 children.
Pathogens other than S pneumoniae affected four out of 148 (2.7%) patients before PCV7 and one out of 275 patients (0.4%) after PCV7, compared to S pneumoniae bacteraemia which affected six out of 148 patients before PCV7 and none of 275 patients after PCV7. Thus, the 100% decrease in S pneumoniae bacteraemia was the major contributor to the overall reduction in occult bacteraemia.
We also found that antibiotic use in this population has not changed with over half of all patients in this age group still receiving antibiotics. It should also be queried whether antibiotic use improves outcomes for the few cases of occult bacteraemia that are found. The role of antibiotics in children with FWLS and/or occult bacteraemia has been controversial. A pre-PCV7 meta-analysis has suggested that there was a trend towards a reduced risk of SBI with oral antibiotic or intramuscular ceftriaxone administration, but the results were not statistically significance.28 The report found that 414 patients receiving empirical antibiotic therapy would only prevent one SBI. Our study suggests that the ratio is now about 1200:1, and we are not convinced that empirical antibiotics alter outcomes compared to watchful waiting to assess for localised symptoms of infection.
Our study indicates that urine cultures are extremely useful in this patient population, with the proportion of UTIs being equal to the proportion of bacteraemia cases in the pre-PCV7 era and outnumbering them more than 20-fold in the post-PCV7 era. The overall frequency of UTIs did not significantly change from the pre- to the post-PCV7 eras nor would anyone have expected a change. Due to the frequency of UTIs in this group and the implications of diagnosis, urine cultures are justified as part of the evaluation of children with FWLS. Given that only 75–77% of FWLS children had a urinalysis and only 55–61% had urine cultures performed, one is left to wonder how many UTIs were missed in these febrile children.
There are significant differences in the current guidelines and policies for evaluating children with FWLS in the USA and the UK. The American College of Emergency Physicians issued a clinical policy in 2003 that assumed that occult bacteraemia rates were between 1.5% and 2% and did not re-address the utility of blood cultures in these children after widespread implementation of PCV7.19 In contrast, the NICE guidelines for feverish children issued in the UK in May 2007 appear to have taken into account the anticipated decline in invasive pneumococcal disease.22 The NICE guidelines recommend that children with FWLS who are classified as “green-low risk” after examination by a paediatric specialist only require examination of the urine, with no blood culture performed and no empirical antibiotics administered. These recommendations appear to correlate with the current trends we have observed in our study. Coupled with the recently issued NICE guidelines on urinary tract infection in children,29 it is likely that the management of FWLS will result in fewer missed UTIs, far fewer blood cultures performed and significantly fewer antibiotics administered. Perhaps the widespread publication of reports covering several years of experience in the UK utilising this approach will facilitate an appropriate update of guidelines in the USA for children with FWLS.
Our findings should be analysed in light of certain limitations. One is the retrospective nature of the study and the relatively small number of patients from a single centre. However, other studies of a similarly selected group of children demonstrate the same trends. Secondly, we could not evaluate which patients in our cohort had been immunised with PCV7. However, CDC studies show that North Carolina exceeds national immunisation levels for all recommended childhood vaccines.30 We estimate that our regional PCV7 vaccination rate was at or better than the US national average of 73%. Another limitation is the inability to determine the number of febrile children who did not have a blood culture performed in the emergency department, which would change the denominator of children with FWLS.
In conclusion, we found a statistically significant decrease in the frequency of occult bacteraemia in children with FWLS aged 3–36 months in the post-PCV7 era. UTIs have not changed in frequency over time and thus are the major cause of SBI in these children with FWLS. Our study supports the recently issued NICE guidelines for the management of feverish children.
The authors would like to thank the following individuals for their assistance in the conception and execution of this project: Harmony Garges, MD, MPH, Laura Smith and Lizzie Harrell, PhD. We would also like to thank our colleagues in the Pediatric Infectious Diseases division for their input and Thomas Kinney for providing data important to our analysis.
Funding: This study was investigator funded.
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.