Several clinical signs and symptoms, such as reduced consciousness, convulsions, cyanosis, tachypnea and slow capillary refill, have been identified as risk factors for serious bacterial infection (SBI) in high, middle and low-income settings.1 However, these so-called red flag signs or symptoms might only develop in the later stages of infection. Hence, recognizing SBI in children with febrile illness at an early stage of disease can be extremely challenging.2 In view of the potential of an innocent appearing febrile illness to rapidly develop into a serious, life-threatening bacterial infection with considerable morbidity and mortality, it is important to identify children at high risk for an SBI as early as possible. Therefore, diagnostic “biomarkers” can be useful to guide clinical decision-making.
Traditionally, a biomarker has been defined “as a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.”3 Biomarkers can be detected in many types of biosamples, varying from blood to central nervous system fluid (CSF) and urine. Most often, biomarkers used in routine pediatric care are proteins, but occasionally other types such as metabolites are used (eg, lactate in blood or CSF, nitrates in urine).
The perfect biomarker for infections should ideally discriminate between patients with and without infection and be able to distinguish (bacterial or viral) etiology (Table 1). In addition, a biomarker ought to be independent of the duration of febrile illness and of comorbidities and to be a predictor of severity of disease. Importantly, any promising new biomarker will need a diagnostic test that is readily affordable, that can be measured in samples obtained with minimally invasive procedures, and which has a rapid turnaround time.
Biomarkers for bacterial infections classically included total white blood cell count (WBC), absolute neutrophil count (ANC), C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR). Many other biomarkers have been studied over the last few decades: procalcitonin (PCT), interleukin-6 (IL-6), IL-8 and tumor necrosis factor-α (TNF-α) are just a few examples of relatively new biomarkers for infection.4–7
Here, we describe a summary of the host biomarkers currently used in clinical practice which aid differentiating bacterial from nonbacterial infections in febrile children. We summarize the advantages and disadvantages of using biomarkers in the context of pre-test and post-test probabilities. We also discuss the discovery of new biomarkers in the era of -omics, and the rise of multimodeling and machine learning to further improve management of children with acute febrile illness.
BIOMARKERS FOR INFECTION IN CHILDREN
CRP is currently one of the most frequently used biomarker(s) for infection worldwide. It is a short pentraxin, which is synthesized in the liver mostly in response to infection or inflammation following stimulation with IL-1β, IL-6 and TNF-α. CRP has several functions, such as complement activation via the classic pathway, modulation of the function of phagocytic cells, and augment cell-mediated cytotoxicity.8 It was first described in 1930 as a protein that precipitated with pneumococcal C-polysaccharide in patients with pneumococcal pneumonia.9 Induction of CRP was demonstrated 12 hours after the injection of endotoxin in the blood stream of healthy individuals, with levels subsequently plateauing between 20 to 72 hours and decreasing to normal values in 3–7 days.10
CRP levels should be low in healthy subjects. However, a rise in CRP can be triggered by many other disorders other than infection, such as rheumatologic disease, periodic fever syndromes, necrosis, trauma or malignancy.11 This has also been described in the neonatal period, where many factors can lead to a transient rise in CRP levels, making both initial high and low CRP levels difficult to interpret.12,13 Repeating a CRP level at 24 hours of life was introduced in the national guideline on the management of neonates suspected of early onset sepsis in the United Kingdom in an attempt to improve management of suspected early onset neonatal sepsis.14,15 However, as of yet no studies have shown the effectiveness or safety of this approach (Table 2).
PCT is the prohormone of calcitonin and is produced by C-cells in the thyroid gland. It was described as a new parameter of severe bacterial infection in 1993.16 The actual function of PCT in bacterial infection remains largely unknown. Rapid induction of PCT was demonstrated at only 3 hours after the injection of endotoxin in the blood stream of healthy individuals, with a peak level at 6 hours and a half-life of 25–30 hours.10
The use of PCT has been studied extensively in both children and neonates. PCT as a marker of neonatal bacterial infection is complicated by several factors. For example, a physiologic increase in PCT has been reported up to 48 hours postpartum.17 Additionally, similar to CRP, serum PCT concentrations can be elevated in neonates with respiratory distress syndrome, hemodynamic failure, perinatal asphyxia, intracranial hemorrhage, pneumothorax or after resuscitation.18 Chiesa et al developed a nomogram of normal PCT values in the neonatal period taking all these factors into account (Fig. 1), and these values proved useful in identifying septic neonates in the first three days of life.17,19,20
WBC and ANC
WBC and ANC have historically been used to determine which child is at risk for SBI. However, current insights show that the WBC offers a low sensitivity of 58% and a specificity of less than 73%.21 In children with fever without a source, both PCT and CRP performed better than WBC or a differential white cell count.22 Pratt et al10 compared the WBC, ANC and CRP in relation to the onset of fever and found that CRP had a better sensitivity and specificity than either WBC or ANC, regardless of the duration of fever. Notably, in this study all biomarkers performed better with a duration of fever of >12 hours.10 Furthermore, combining WBC or ANC with other biomarkers did not improve diagnostic accuracy.23
Erythrocyte Sedimentation Rate
The ESR might be the eldest known biomarker for infection and inflammation, dating back to the ancient Greeks.24 Disease entities with systemic inflammation, such as infections, malignancy, autoimmune disorders and autoinflammatory conditions, can elevate the ESR. Normal ESR values vary by age, and a value between 0 and 20 mm/h is considered to be normal after the neonatal period by most laboratories. The slow rise (>48 hours) and delayed lowering of ESR, in particular when compared with other acute phase reactants, make ESR suitable for monitoring inflammation in bone and joint infections and chronic inflammatory conditions, and it is believed it resembles the time to the complete resolution of inflammation more closely.24 ESR is also still commonly included in clinical algorithms for febrile children with a limp at risk for septic arthritis and osteomyelitis.25
Biomarkers and Diagnostic Accuracy
CRP and PCT were found to have near identical diagnostic performance in 2 large systematic reviews, both outperforming WBC.21,23 The pooled area under the receiver operating curve (AUC) for CRP was 0.81 (95% confidence interval, CI 0.78–0.84) and for PCT 0.84 (95% CI 0.80–0.87); for the emergency department (ED) setting AUCs were 0.85 (95% CI 0.0.81–0.87) for CRP and 0.84 ((95% CI 0.81–0.87) for PCT.21 In comparison, WBC had a pooled AUC of 0.70 (0.65–0.74). A recurring clinical dilemma for many clinicians is the question of which thresholds to use. For CRP, a cutoff of 20 mg/L is believed to have good rule-out value for SBI and a level of ≥80 mg/L good rule-in value. For PCT, a cutoff of 0.5 ng/mL is considered to have good rule-out value for SBI, whereas a level of ≥2.0 ng/mL has good rule-in value.23
Clinical studies have shown differing roles for the duration of fever on the diagnostic usefulness of both CRP and PCT, but it is generally felt that PCT might be more useful in children with a short duration of fever and in young febrile infants 90 days of age or less.10,21,26–29 Studies found that high levels of CRP and PCT were strongly predictive of SBI in children with fever, independent of duration of disease. To the contrary, low CRP levels could not be used to rule out or confirm SBI in children with a short duration of fever,7,10 and PCT was superior to CRP in detecting SBI at an earlier stage of the disease (Fig. 2).7 Furthermore, a low CRP after >24 hours of fever was useful for ruling out SBI.28
Many other biomarkers have been studied in both adults and children at risk for SBI.30 However, compared with CRP and PCT, most biomarkers do not have greater diagnostic accuracy.23 Some of these new biomarkers showed great promise in certain settings: for example, CD64, a protein expressed on the surface of neutrophils, showed great promise in sepsis in several adult studies.31 It also performed well as a diagnostic marker of neonatal sepsis,32–35 but then validated disappointingly in febrile children in the ED.36,37 This shows the importance of conducting not only biomarker discovery studies but also validation studies in a wide range of settings.
Previous studies already showed that combining biomarkers might lead to improved discrimination between children with viral and bacterial infections. For example, the Lab-score, using CRP, PCT and urinalysis in an easy-to-use algorithm, validated well in several studies.29,38,39 The Lab-score also confirms the important diagnostic role of urine analysis in children with fever without clear focus.29 This was reiterated in a large observational cohort study by De et al.40 In contrast, up to now there is insufficient evidence for the use of routine rapid viral testing (ie, influenza) to reduce antibiotic prescription or admission rate for children with fever without a source in the ED.41,42 A positive rapid test result, however, did result in a significant reduction of chest radiographs and laboratory tests.41 More recently, studies have focused on combinations of protein biomarkers that include markers of both viral disease and bacterial disease.43–45 One such immune assay, the so-called ImmunoXpert test, combines interferon gamma-induced protein 10 (IP-10) IP-10 and TNF-related apoptosis-inducing ligand (TRAIL), both markers of viral infection, and CRP. This assay had excellent discriminative ability for distinguishing bacterial and viral infections in a large prospective multicenter study (AUC 0.90, 95% CI 0.86–0.95).44 However, in this study, CRP had fairly similar discriminative ability with AUC 0.89 (95% CI 0.84–0.94); PCT had a slightly lower AUC of 0.84 (95% CI 0.79–0.90). In a validation cohort, the assay performed equally well with a positive likelihood ratio (LR) of 9.2 (95% CI 6.1–13.9) and negative LR 0.07 (95% CI 0.03–0.18).43 Another promising assay, the FebriDx test, combines human myxovirus resistance protein 1 (MxA), a biomarker for viral infection, and CRP and performed well in several adult studies aiding physicians in patients with upper respiratory tract infections in particular.
PRE AND POST-TEST PROBABILITY
Biomarkers can change the likelihood of children having a SBI, from a pre-test probability to a post-test probability (Fig. 3). The pre-test probability reflects the incidence of the disease of interest (ie, SBIs in our example) in the population at risk (eg, children with fever). The incidence, and thus pre-test probability, will change depending on the setting: for example, the incidence of SBI in primary care is much lower than in EDs,1,46,47 which in turn is much lower than intensive care departments. Next, clinical signs and symptoms, but also other predictors such as gender and the time of the day and seasonality, influence the probability, serving as a diagnostic test and resulting in a post-test probability. This Bayesian reasoning, using a pre-test probability and a test result resulting in a post-test probability, is important when deciding on when and how to use a test. As an example, in critically unwell children with fever at high risk of SBI in an intensive care department, no biomarker result is likely to change your decision on starting antibiotic treatment, but it might be useful in monitoring treatment response or deciding on duration of treatment. Similarly, in a well-appearing febrile infant with clear signs of an upper respiratory tract infection in primary care, one does not need additional tests to safely discharge a patient. In the first example, a negative biomarker test result, will unlikely lower the probability of SBI to a threshold where one would withhold antibiotics. In the second example, a positive biomarker result will only increase the post-test probability marginally, still well below the level for starting antibiotics or referral to the ED. Hence, biomarkers will be most useful in febrile children with remaining diagnostic uncertainty.
Generally, a diagnostic test result will be interpreted in a dichotomous manner, that is, as high or low, normal or abnormal, positive or negative, albeit the true thresholds for abnormality being elusive for many biomarkers. Traditionally, sensitivity and specificity have been the preferred methods of reporting diagnostic accuracy; however, for clinical decision-making LRs, a measure combining sensitivity and specificity, are more useful. A positive LR can, together with a high specificity, be used for ruling in an SBI. A negative LR, combined with a high sensitivity, is useful in ruling out SBI (Fig. 3). For example, an increased CRP value with a pooled positive LR of 3.10 changes the pre-test probability of 15% to a post-probability of 35%; likewise, a low CRP value with a pooled negative LR of 0.34 reduces the risk to 6%.21,23 This post-test probability then inversely resembles the number needed to test (eg, radiographs) or treat (eg, antibiotics).
Several clinical prediction models, such as the feverkidstool by Nijman et al,48–50 combine clinical signs and symptoms and biomarkers to provide clinicians with a post-test probability. Similarly, Classification and Regression Tree analysis presents physicians with decision trees that can guide physicians in establishing the risk of having SBI. For example, the step-by-step model for ruling out SBI in young febrile infants with fever without focus validated well in a study with infants from 11 European EDs. The step-by-step model reduced the risk of all types of SBI from 23% to 2.0% using ill appearance, age ≤ 21 days, and leukocyturia; adding PCT ≥ 0.5 ng/mL, CRP > 20 mg/L, and ANC > 10,000/mm3 then reduced this risk further to 1.1% (sensitivity 92% and negative predictive value 99.3%).51
A recent study by Verbakel et al52 showed that SBI was uncommon in children with low CRP < 20 mg/L and a clinical algorithm based on 4 vital signs and 7 clinical parameters (SBI 6/5,517; reducing the risk from 4.9% to 0.1%). Interestingly, this study positioned CRP as an initial triage tool before assessing clinical warning signs.
Recent advances in analytical and computational platforms have paved the way for the era of -omics and precision medicine in pediatric infectious diseases.
Biometrics, Multimodeling and Machine Learning
Intrauterine fetal heart rate monitoring is an established technique which has been used for the detection of fetal distress for various years. Similarly, neonatal heart rate characteristics monitoring can assist diagnosing neonatal sepsis in premature born infants.53 A reduced heart rate variability and transient decelerations were seen hours up to days before the onset of clinical signs of illness. Measurements of standard deviation, sample asymmetry and sample entropy were highly related to clinical illness.53 A few studies in adults have described computerized alerting systems that recognize the combination of hypotension and laboratory markers consistent with systemic inflammatory response syndrome.54 However, fewer data exist in patients in the ED setting and in children after the neonatal period. The interpatient variation and time of assessment in relation to disease onset in these settings are just a few examples which explain these discrepancies. Moreover, clinical data in electronic medical records usually are not written down in a standardized way and from a computational perspective, vital parameters are sparsely and irregularly sampled.55,56 When clinical data are entered in a structured way in electronic medical records, machine learning and deep learning can be promising new techniques to predict which children will develop or have an SBI using routinely captured clinical data.56
The Era of -Omics
Patient-specific signatures on a genetic, protein or metabolite level likely hold the key for the next generation of diagnostic tests, in particular in those cases where current diagnostics are not reliable enough. Theoretically, any biosample can be used for biomarker discovery studies and clinical application. The era of -omics could move future diagnostic decision-making away from the traditional manner of pattern recognition based on a constellation of clinical signs and symptom, toward a more molecular level-based taxonomy. For example, Herberg et al57 developed and validated a 2-gene RNA signature to distinguish viral and bacterial infections in a large cohort of children with severe infections. Similarly, Mahajan et al58 identified an RNA biosignature of 10 classifier to genes for detecting bacteremia in young febrile children <60-day old in the ED. Also, in children with RSV bronchiolitis, RNA biosignatures were associated with severity of disease and linked bacterial pathogens in the nasopharyngeal microbiome, in particular infections with Haemophilus influenzae and Streptococcus pneumoniae, with more severe disease.59,60 Hence, RNA biosignatures might also be useful in predicting severity of disease and need for hospitalization. Anderson et al61 described diagnostic RNA-based signatures to successfully differentiate between active and latent tuberculosis and other infections in African children. Up to now, RNA biosignatures are only used in research settings and do not play a role in routine diagnostics yet. However, these diagnostic tests are developing rapidly, as for example exemplified by the SeptiCyte test, a patented 4-gene RNA-based signature discriminating sepsis from systemic inflammatory response syndrome in ICU patients. Although originally developed and validated in adult populations,62,63 the test also proved useful in a pediatric ICU population.64
CONCLUSION AND DISCUSSION
In children with fever, biomarkers can help the physician in deciding on performing additional tests and imaging, starting antibiotic treatment or hospitalization. However, many physicians argue against the routine use of biomarkers such as CRP in febrile children, mostly using anecdotal evidence, stating that certain viral infections can have raised CRP levels, or that bacterial infections initially present with a low inflammatory response. However, in the overall population of children with fever presenting to the ED, both CRP and PCT have proven their diagnostic usefulness in predicting SBI.21,23 It is true that not all children with SBI present with high PCT or CRP levels, for example, secondary to a short onset of fever at the time of the first assessment; however, this only stresses the importance of interpreting biomarker results alongside the individual patient’s clinical signs and symptoms.
The clinical impact of diagnostic tools is not solely dependent on diagnostic accuracy, yet few studies assess the impact beyond the analytical or clinical performance.65 In 2 randomized trials, the initial prescribing rates of antibiotics among children with fever in EDs were not influenced by biomarkers, although in one study PCT levels did shorten the duration of antibiotics.66,67 Similarly, the availability of a rapid point-of-care CRP test hardly impacted clinical management compared with routine clinical practice in one clinical decision support trial.68 In contrast, introducing CRP point-of-care testing in primary care in Vietnam for nonsevere respiratory tract infections in children reduced initial antibiotic prescribing rates significantly (odds ratio 0.47, 95% CI 0.26–0.83), once again highlighting the importance of taking differences in clinical settings into consideration. Likewise, using a nomogram (Fig. 2) and PCT-guided decision-making in neonates suspected of early onset sepsis, Stocker et al were able to show a significant reduction of the duration of antibiotic therapy and of the hospital stay in an international intervention trial. There was a very low rate of reinfections and no study-related mortality in this study.69
CRP and PCT have been studied in a variety of clinical settings, from the pediatric ED to the intensive care and primary care, and in specific patient populations such as in patients with febrile neutropenia, inflammatory and autoimmune conditions. As a result, studies have reported differences in diagnostic accuracy of biomarkers in different clinical scenarios. This can make the interpretation of biomarker studies difficult, in particular in view of changing pre-test probabilities and of confounding factors influencing biomarker levels such as trauma or surgical procedures.
In our opinion, biomarkers should be used only in those cases where they truly influence clinical decision-making. Although the diagnostic accuracy of CRP and PCT are very similar, local preferences, costs and local population need to be taken into consideration. For example, in our hospital, the costs of PCT are significantly higher than CRP. At this point in time, we would not recommend using biomarker panels or combining CRP and PCT together routinely in the evaluation of febrile children.29,49
Despite all research efforts, the predictive value of a biomarker to discriminate viral and bacterial infection or to predict disease outcome and mortality at an individual level remains imperfect. Improved biomarkers might benefit clinicians in future, but they should always be seen in the context of the clinical presentation of a patient. This is particularly true for children at risk for sepsis. A large proportion of children at risk of sepsis will either have falsely reassuring vital parameters2 or conversely have falsely alarming abnormal vital parameters.70 Current biomarkers are unlikely to fully aid physicians in ruling in or ruling out the early presence of an SBI, which will largely remain a clinical diagnosis, based on a constellation of clinical signs and symptoms and fostered by clinical experience and exposure.
In conclusion, CRP and PCT are biomarkers that can improve clinical decision-making by altering the pre and post-test probabilities of having SBI significantly. When using biomarkers, one should keep the kinetics of the biomarker and the pre-test probability of infection in mind. Importantly, upcoming techniques looking at biomarkers in the field of pediatric infectious diseases, like machine learning, deep learning and –omics, will have to prove their value in clinical practice the coming years.
1. Van den Bruel A, Haj-Hassan T, Thompson M, et al; European Research Network on Recognising Serious Infection Investigators. Diagnostic value of clinical features at presentation to identify serious infection in children in developed countries: a systematic review. Lancet. 2010;375:834–845.
2. Thompson MJ, Ninis N, Perera R, et al. Clinical recognition of meningococcal disease in children and adolescents. Lancet. 2006;367:397–403.
3. Sankar V, Webster NR. Clinical application of sepsis biomarkers. J Anesth. 2013;27:269–283.
4. Coovadia HM, Rollins NC, Bland RM, et al. Mother-to-child transmission of HIV-1 infection during exclusive breastfeeding in the first 6 months of life: an intervention cohort study. Lancet. 2007;369:1107–1116.
5. Bodilsen J, Dalager-Pedersen M, Schønheyder HC, et al. Time to antibiotic therapy and outcome in bacterial meningitis: a Danish population-based cohort study. BMC Infect Dis. 2016;16:392.
6. Berman S. Otitis media in children. N Engl J Med. 1995;332:1560–1565.
7. Fernández Lopez A, Luaces Cubells C, García García JJ, et al; Spanish Society of Pediatric Emergencies. Procalcitonin
in pediatric emergency departments for the early diagnosis of invasive bacterial infections in febrile infants: results of a multicenter study and utility of a rapid qualitative test for this marker. Pediatr Infect Dis J. 2003;22:895–903.
8. Marnell L, Mold C, Du Clos TW. C-reactive protein
: ligands, receptors and role in inflammation. Clin Immunol. 2005;117:104–111.
9. Tillett WS, Francis T. Serological reactions in pneumonia with a non-protein somatic fraction of pneumococcus. J Exp Med. 1930;52:561–571.
10. Pratt A, Attia MW. Duration of fever and markers of serious bacterial infection in young febrile children. Pediatr Int. 2007;49:31–35.
11. Pepys MB, Hirschfield GM. C-reactive protein
: a critical update. J Clin Invest. 2003;111:1805–1812.
12. Hofer N, Zacharias E, Müller W, et al. An update on the use of C-reactive protein
in early-onset neonatal sepsis: current insights and new tasks. Neonatology. 2012;102:25–36.
13. Brown JVE, Meader N, Cleminson J, et al. C-reactive protein
for diagnosing late-onset infection in newborn infants. Cochrane Database Syst Rev. 2019;1:CD012126.
14. Osvald EC, Prentice P. NICE clinical guideline: antibiotics for the prevention and treatment of early-onset neonatal infection. Arch Dis Child Educ Pract Ed. 2014;99:98–100.
15. Mukherjee A, Davidson L, Anguvaa L, et al. NICE neonatal early onset sepsis guidance: greater consistency, but more investigations, and greater length of stay. Arch Dis Child Fetal Neonatal Ed. 2015;100:F248–F249.
16. Assicot M, Gendrel D, Carsin H, et al. High serum procalcitonin
concentrations in patients with sepsis and infection. Lancet. 1993;341:515–518.
17. Chiesa C, Panero A, Rossi N, et al. Reliability of procalcitonin
concentrations for the diagnosis of sepsis in critically ill neonates. Clin Infect Dis. 1998;26:664–672.
18. Lapillonne A, Basson E, Monneret G, et al. Lack of specificity of procalcitonin
for sepsis diagnosis in premature infants. Lancet. 1998;351:1211–1212.
19. Kordek A, Giedrys-Kalemba S, Pawlus B, et al. Umbilical cord blood serum procalcitonin
concentration in the diagnosis of early neonatal infection. J Perinatol. 2003;23:148–153.
20. Enguix A, Rey C, Concha A, et al. Comparison of procalcitonin
with C-reactive protein
and serum amyloid for the early diagnosis of bacterial sepsis in critically ill neonates and children. Intensive Care Med. 2001;27:211–215.
21. Yo CH, Hsieh PS, Lee SH, et al. Comparison of the test characteristics of procalcitonin
to C-reactive protein
and leukocytosis for the detection of serious bacterial infections in children presenting with fever without source: a systematic review and meta-analysis. Ann Emerg Med. 2012;60:591–600.
22. Galetto-Lacour A, Zamora SA, Gervaix A. Bedside procalcitonin
and C-reactive protein
tests in children with fever without localizing signs of infection seen in a referral center. Pediatrics. 2003;112:1054–1060.
23. Van den Bruel A, Thompson MJ, Haj-Hassan T, et al. Diagnostic value of laboratory tests in identifying serious infections in febrile children: systematic review. BMJ. 2011;342:d3082.
24. Ramsay ES, Lerman MA. How to use the erythrocyte sedimentation rate in paediatrics. Arch Dis Child Educ Pract Ed. 2015;100:30–36.
25. Kocher MS, Zurakowski D, Kasser JR. Differentiating between septic arthritis and transient synovitis of the differentiating between septic arthritis and transient synovitis of the hip in children: an evidence-based clinical prediction algorithm *. J Bone Jt Surg Am. 1999;81:1662–1670.
26. Olaciregui I, Hernández U, Muñoz JA, et al. Markers that predict serious bacterial infection in infants under 3 months of age presenting with fever of unknown origin. Arch Dis Child. 2009;94:501–505.
27. Luaces-Cubells C, Mintegi S, García-García JJ, et al. Procalcitonin
to detect invasive bacterial infection in non-toxic-appearing infants with fever without apparent source in the emergency department. Pediatr Infect Dis J. 2012;31:645–647.
28. Segal I, Ehrlichman M, Urbach J, et al. Use of time from fever onset improves the diagnostic accuracy of C-reactive protein
in identifying bacterial infections. Arch Dis Child. 2014;99:974–978.
29. Nijman RG, Moll HA, Smit FJ, et al. C-reactive protein
and the lab-score for detecting serious bacterial infections in febrile children at the emergency department: a prospective observational study. Pediatr Infect Dis J. 2014;33:e273–e279.
30. Kapasi AJ, Dittrich S, González IJ, et al. Host biomarkers for distinguishing bacterial from non-bacterial causes of acute febrile illness: a comprehensive review. PLoS One. 2016;11:e0160278.
31. Hoffmann JJML. Neutrophil CD64: a diagnostic marker for infection and sepsis. Clin Chem Lab Med. 2009;47:903–916.
32. Ng PC, Li K, Wong RP, et al. Neutrophil CD64 expression: a sensitive diagnostic marker for late-onset nosocomial infection in very low birthweight infants. Pediatr Res. 2002;51:296–303.
33. Ng PC, Li G, Chui KM, et al. Neutrophil CD64 is a sensitive diagnostic marker for early-onset neonatal infection. Pediatr Res. 2004;56:796–803.
34. Bhandari V, Wang C, Rinder C, et al. Hematologic profile of sepsis in neonates: neutrophil CD64 as a diagnostic marker. Pediatrics. 2008;121:129–134.
35. Groselj-Grenc M, Ihan A, Derganc M. Neutrophil and monocyte CD64 and CD163 expression in critically ill neonates and children with sepsis: comparison of fluorescence intensities and calculated indexes. Mediators Inflamm. 2008;2008:202646.
36. van Veen M, Nijman RG, Zijlstra M, et al. Neutrophil CD64 expression is not a useful biomarker
for detecting serious bacterial infections in febrile children at the emergency department. Infect Dis (Lond). 2016;48:331–337.
37. Rudensky B, Sirota G, Erlichman M, et al. Neutrophil CD64 expression as a diagnostic marker of bacterial infection in febrile children presenting to a hospital emergency department. Pediatr Emerg Care. 2008;24:745–748.
38. Galetto-Lacour A, Zamora SA, Andreola B, et al. Validation of a laboratory risk index score for the identification of severe bacterial infection in children with fever without source. Arch Dis Child. 2010;95:968–973.
39. Lacour AG, Zamora SA, Gervaix A. A score identifying serious bacterial infections in children with fever without source. Pediatr Infect Dis J. 2008;27:654–656.
40. De S, Williams GJ, Hayen A, et al. Accuracy of the “traffic light” clinical decision rule for serious bacterial infections in young children with fever: a retrospective cohort study. BMJ. 2013;346:f866.
41. Lacroix S, Vrignaud B, Avril E, et al. Impact of rapid influenza diagnostic test on physician estimation of viral infection probability in paediatric emergency department during epidemic period. J Clin Virol. 2015;72:141–145.
42. Doan Q, Enarson P, Kissoon N, et al. Rapid viral diagnosis for acute febrile respiratory illness in children in the Emergency Department (Review). 2014. doi: 10.1002/14651858.CD006452.pub4.
43. Srugo I, Klein A, Stein M, et al. Validation of a novel assay to distinguish bacterial and viral infections. Pediatrics. 2017;140:e20163453.
44. van Houten CB, de Groot JAH, Klein A, et al. A host-protein based assay to differentiate between bacterial and viral infections in preschool children (OPPORTUNITY): a double-blind, multicentre, validation study. Lancet Infect Dis. 2017;17:431–440.
45. Shapiro NI, Self WH, Rosen J, et al. A prospective, multi-centre US clinical trial to determine accuracy of FebriDx point-of-care testing for acute upper respiratory infections with and without a confirmed fever. Ann Med. 2018;50:420–429.
46. Van den Bruel A, Aertgeerts B, Bruyninckx R, et al. Signs and symptoms for diagnosis of serious infections in children: a prospective study in primary care. Br J Gen Pract. 2007;57:538–546.
47. Craig JC, Williams GJ, Jones M, et al. The accuracy of clinical symptoms and signs for the diagnosis of serious bacterial infection in young febrile children: prospective cohort study of 15 781 febrile illnesses. BMJ. 2010;340:c1594.
48. Nijman RG, Vergouwe Y, Thompson M, et al. Clinical prediction model to aid emergency doctors managing febrile children at risk of serious bacterial infections: diagnostic study. BMJ. 2013;346:f1706.
49. Nijman RG, Vergouwe Y, Moll HA, et al. Validation of the Feverkidstool and procalcitonin
for detecting serious bacterial infections in febrile children. Pediatr Res. 2018;83:466–476.
50. Irwin AD, Grant A, Williams R, et al. Predicting risk of serious bacterial infections in febrile children in the emergency department. Pediatrics. 2017;140:e20162853.
51. Gomez B, Mintegi S, Bressan S, et al; European Group for Validation of the Step-by-Step Approach. Validation of the “step-by-step” approach in the management of young febrile infants. Pediatrics. 2016;138:e20154381.
52. Verbakel JY, Lemiengre MB, De Burghgraeve T, et al. Point-of-care C reactive protein to identify serious infection in acutely ill children presenting to hospital: prospective cohort study. Arch Dis Child. 2018;103:420–426.
53. Moorman JR, Carlo WA, Kattwinkel J, et al. Mortality reduction by heart rate characteristic monitoring in very low birth weight neonates: a randomized trial. J Pediatr. 2011;159:900.e1–906.e1.
54. Sawyer AM, Deal EN, Labelle AJ, et al. Implementation of a real-time computerized sepsis alert in nonintensive care unit patients. Crit Care Med. 2011;39:469–473.
55. Cruz AT, Williams EA, Graf JM, et al. Test characteristics of an automated age- and temperature-adjusted tachycardia alert in pediatric septic shock. Pediatr Emerg Care. 2012;28:889–894.
56. Williams JB, Ghosh D, Wetzel RC. Applying machine learning to pediatric critical care data. Pediatr Crit Care Med. 2018;19:599–608.
57. Herberg JA, Kaforou M, Wright VJ, et al; IRIS Consortium. Diagnostic test accuracy of a 2-transcript host RNA signature for discriminating bacterial vs viral infection in febrile children. JAMA. 2016;316:835–845.
58. Mahajan P, Kuppermann N, Mejias A, et al; Pediatric Emergency Care Applied Research Network (PECARN). Association of RNA biosignatures with bacterial infections in febrile infants aged 60 days or younger. JAMA. 2016;316:846–857.
59. Rodriguez-Fernandez R, Tapia LI, Yang CF, et al. Respiratory syncytial virus genotypes, host immune profiles, and disease severity in young children hospitalized with bronchiolitis. J Infect Dis. 2017;217:24–34.
60. de Steenhuijsen Piters WA, Heinonen S, Hasrat R, et al. Nasopharyngeal microbiota, host transcriptome, and disease severity in children with respiratory syncytial virus infection. Am J Respir Crit Care Med. 2016;194:1104–1115.
61. Anderson ST, Kaforou M, et al. Europe PMC funders group diagnosis of childhood tuberculosis and host RNA expression in Africa. 2014;370:1712–1723.
62. McHugh L, Seldon TA, Brandon RA, et al. A molecular host response assay to discriminate between sepsis and infection-negative systemic inflammation in critically ill patients: discovery and validation in independent cohorts. PLoS Med. 2015;12:e1001916.
63. Miller RR III, Lopansri BK, Burke JP, et al. Validation of a host response assay, septiCyte LAB, for discriminating sepsis from systemic inflammatory response syndrome in the ICU. Am J Respir Crit Care Med. 2018;198:903–913.
64. Zimmerman JJ, Sullivan E, Yager TD, et al. Diagnostic accuracy of a host gene expression signature that discriminates clinical severe sepsis syndrome and infection-negative systemic inflammation among critically ill children. Crit Care Med. 2017;45:e418–e425.
65. Verbakel JY, Turner PJ, Thompson MJ, et al. Common evidence gaps in point-of-care diagnostic test evaluation: a review of horizon scan reports. BMJ Open. 2017;7:e015760.
66. Baer G, Baumann P, Buettcher M, et al. Procalcitonin
guidance to reduce antibiotic treatment of lower respiratory tract infection in children and adolescents (ProPAED): a randomized controlled trial. PLoS One. 2013;8:e68419.
67. Lacroix L, Manzano S, Vandertuin L, et al. Impact of the lab-score on antibiotic prescription rate in children with fever without source: a randomized controlled trial. PLoS One. 2014;9:e115061.
68. de Vos-Kerkhof E, Nijman RG, Vergouwe Y, et al. Impact of a clinical decision model for febrile children at risk for serious bacterial infections at the emergency department: a randomized controlled trial. PLoS One. 2015;10:e0127620.
69. Stocker M, van Herk W, El Helou S, et al; NeoPInS Study Group. Procalcitonin
-guided decision making for duration of antibiotic therapy in neonates with suspected early-onset sepsis: a multicentre, randomised controlled trial (NeoPIns). Lancet. 2017;390:871–881.
70. Scott HF, Deakyne SJ, Woods JM, et al. The prevalence and diagnostic utility of systemic inflammatory response syndrome vital signs in a pediatric emergency department. Acad Emerg Med. 2015;22:381–389.