ORIGINAL ARTICLE
Lack of Accuracy of Biomarkers and Physical Examination to Detect Bacterial Infection in Febrile Infants Marta Germán Díaz, MD, Rosa Pavo García, MD, Daniel Blázquez Gamero, MD, M. Isabel González-Tomé, PhD, Pilar Cedena Romero, MD, Milagros Marín Marín, MD, and Jesús Ruiz Contreras, PhD Objectives: The aim of this study was to analyze the usefulness of physical examination, C-reactive protein (CRP), procalcitonin (PCT), white blood cell (WBC) count, and absolute neutrophils counts (ANCs) for the diagnosis of invasive bacterial infections (IBIs) and potentially serious bacterial infections in infants younger than the age of 3 months presenting with fever without source (FWS) to the emergency department (ED). Methods: A descriptive retrospective study that includes all infants aged younger than 3 months who presented with FWS to the ED between July 2008 and January 2012. We evaluated diagnostic performance for each test by receiver operating characteristic curve analysis. Sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio were also calculated. Results: Three hundred eighteen patients met the inclusion criteria. Eleven bacteremia (3.5%) and 76 urinary tract infections (23.9%) were diagnosed. To detect IBI, the areas under the curve for the different tests were as follows: PCT, 0.77 (95% confidence interval [CI], 0.57–0.96); CRP, 0.54 (95% CI, 0.36–0.73); ANC, 0.53 (95% CI, 0.34–0.71); and WBC, 0.42 (0.24–0.61). To detect potentially serious bacterial infections, the areas under the curve were as follows: PCT, 0.66 (95% CI, 0.59–0.74); CRP, 0.68 (0.60–0.76); ANC, 0.64 (0.56–0.71); and WBC, 0.66 (0.58–0.72). Conclusions: Procalcitonin is better than CRP, WBC, and ANC to confirm or dismiss the presence of an IBI in infants aged younger than 3 months presenting with FWS to the ED. However, it could not identify almost 30% of infants with IBI. Most patients diagnosed with IBI (10 of 11) presented abnormal values in at least one of the analytical parameters and/or physical appearance. Four of 5 patients with IBI and well appearing presented abnormal results in at least one of the analytical parameters. Therefore, the development of tools combining different tests including the new biomarkers could increase the reliability of the tests for the diagnosis of IBI in these patients. Key Words: C-reactive protein, fever without source, infants younger than 3 months, invasive bacterial infection, potentially serious bacterial infections, procalcitonin (Pediatr Emer Care 2015;00: 00–00)
F
ever without source (FWS) is one of the most frequent reasons for pediatric emergency consultations,1,2 and it has been estimated that up to 10% to 25% of cases of FWS in infants aged younger than 3 months are caused by a potentially serious bacterial infection (PSBI).3–5 In some cases, infants in this age group and a PSBI may have a normal physical examination in the emergency department (ED).6–9 Therefore, it is necessary to have reliable analytical markers for early detection of PSBI in these patients. From the Pediatrics Department, 12 de Octubre University Hospital, Madrid, Spain. Disclosure: The authors declare no conflict of interest. Reprints: Marta Germán Díaz, MD, Pediatrics Department, 12 de Octubre University Hospital, Avenida de Córdoba s/n, 28041, Madrid, Spain (e‐mail: [email protected]). The authors declare no support from pharmaceutical industry or other organizations. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0749-5161
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Since the 1990s, several algorithms have been developed for the management of infants aged younger than 3 months with FWS.10–12 These algorithms classify infants according to the risk of having an invasive bacterial infection (IBI), which is estimated using analytical and clinical parameters. Over the past few years, new markers of bacterial infection, such as C-reactive protein (CRP) and procalcitonin (PCT), have been added to these algorithms.13–15 Several studies have evaluated the usefulness of CRP and PCT for the management of children with FWS. However, data regarding the group of infants aged younger than 3 months are scarce or are limited to well-appearing infants.16,17 The objective of our study was to analyze the usefulness of CRP, PCT, white blood cell (WBC) count, and absolute neutrophil count (ANC) for the diagnosis of IBI and PSBI in infants aged younger than 3 months presenting with FWS to the ED.
METHODS Characteristics of the Patients and Inclusion Criteria We conducted a descriptive retrospective study in the pediatric ED of a tertiary teaching hospital with about 60,000 visits annually. All infants aged younger than 90 days presenting with FWS to the ED between July 2008 and January 2012 were included in the study. Fever was defined as an axillary temperature of greater than or equal to 38°C (100.4°F) measured at home and/or in the ED. The following patients were excluded from the study: preterm neonates with a gestational age less than 37 weeks, infants with chronic diseases or congenital malformations, and infants who had been previously hospitalized or treated with antibiotics.
Study Design The protocol for the management of infants aged younger than 3 months with FWS at our hospital includes a complete blood cell count, CRP and PCT levels, a blood culture, a urine dipstick, a urine sediment, and a urine culture collected by urinary catheterization or suprapubic puncture. A lumbar puncture was performed in not well-appearing infants (Yale Observation Scale [YIOS] score >7) and in those with irritability or analytical alterations (WBC count >15,000/mm3 or 5 WBCs/field in urine sediment], presence of nitrites or bacteriuria in urine sediment, CRP >3 mg/dL or PCT >0.5 ng/mL elevation). A lumbar puncture was performed before any antibiotic therapy. Invasive bacterial infection was defined as the isolation of pathogenic bacteria (without contaminants) in the 2 bottles of blood culture and/or in cerebrospinal fluid (CSF). Urinary tract infections (UTIs) and pneumonias were considered as PSBI. Diagnosis of UTI required the presence of at least 50,000 colony-forming units/mL of a uropathogen cultured from the urine specimen. Diagnosis of pneumonia required the presence of an infiltrate or consolidation on chest radiograph. www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
1
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Díaz et al
FIGURE 1. Flowchart for identification of infants to be included in the analysis.
To perform the final analysis, we only included patients in whom all the following diagnostic tests had been done at the ED: WBC count, CRP and PCT levels, complete urinalysis, blood culture, and urine culture. Children's overall appearance was assessed by a pediatric emergency physician on the initial physical examination upon arrival to the pediatric ED. The overall appearance value was determined by means of the YIOS, considering infants with YIOS score greater than 7 as not well appearing.7,10 The study was approved by the hospital research committee.
Data Collection Data were collected using the pediatric ED database. The following data were obtained from each patient: demographic data (age, sex), clinical history, maximum temperature at home and at hospital, duration of fever, overall appearance of the infant on arrival to hospital, results of the tests performed (WBC, ANC, CRP and PCT values, urine dipstick, urine sediment and results of blood, urine and CSF cultures), as well as the destination of the patient (discharge, admission to intensive care unit [ICU], or admission to hospital).
Statistical Analysis Statistical analyses were conducted using SPSS software 20.0 (IBM, SPSS, Chicago, Ill). Normally distributed continuous variables were expressed as mean (SD), whereas nonnormally distributed continuous variables were expressed as median and interquartile range. Categorical variables were reported as percentages. The diagnostic performance of the different blood tests was
2
www.pec-online.com
evaluated by receiver operating characteristic (ROC) curve analysis. We calculated the area under the curve (AUC), and we used the best cutoff points for each laboratory parameter. Sensitivity, specificity, positive likelihood ratio (LR), and negative LR were also calculated.
RESULTS Characteristics of Patients During the study period, 518 infants aged younger than 3 months presented with FWS to the ED. We excluded 200 infants TABLE 1. Main Characteristics of Patients Characteristics
n (%)
Sex, male/female Overall appearance Well appearing Not well appearing Duration of fever, h 3 PCT, ng/dL >0.5 >2 Not well appearing
18.2 (5.1–45.7) 72.7 (43.4–90.2) 45 (16–74.9) 54.54 (28–78.7)
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Specificity (95% CI) 84.11 (79.4–88) 85.7 (81.7–88.6) 94.1 (91.5–96.8) 85.3 (80.5–89)
+LR (95% CI)
−LR (95% CI)
1.24 (0.27–3.36)
0.97 (0.77–1.3)
5.14 (3.3–8.7) 7.8 (3.25–17) 3.7 (2–6.84)
0.32 (0.12–0.84) 0.58 (0.34–1) 0.53 (0.28–1.02)
www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
3
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Díaz et al
TABLE 4. Values of Sensitivity, Specificity, Positive LR, and Negative LR for CRP and PCT and Overall Appearance in the Detection of PSBI (Urine Infection) Sensitivity (95% CI) CRP, mg/dL >3 PCT, ng/dL >0.5 >2 Not well appearing
Specificity (95% CI)
39.1 (28.9–49.3)
93 (89.8–96.4)
31 (21.3–40.8) 17.2 (9.3–25.2) 14.12 (6.7–21.5)
89.1 (85.1–93.2) 96.5 (94.2–98.9) 83.4 (79–88.6)
+ LR (95% CI)
−LR (95% CI)
5.6 (3.3–9.7)
0.65 (0.5–0.78)
2.9 (1.8–4.7) 4.9 (2.2–11.3) 0.88 (0.48–1.6)
0.75 (0.64–0.89) 0.86 (0.78–0.95) 1.02 (0.9–1.1)
rate (16%) of false-positive cases, making this test useless in this setting. Our data contrast with previously published studies in which CRP has shown a better diagnostic performance.15,16 The differences obtained may be due to several reasons. First, our study was limited to infants aged younger than 90 days, and the magnitude of response to acute phase reactants in invasive infections may be different in younger infants. Although there are several published studies on the usefulness of CRP and PCT for diagnosis of infants with FWS, few of them are focused on this age group. Andreola et al14 demonstrated the superior discriminative capability of CRP and PCT over WBC or ANC for the detection of PSBI in infants with FWS. However, they could not find a statistically significant difference between CRP and PCT, with an area under the ROC curve (AUC) of 0.85 and 0.82, respectively, values that are much higher than those obtained in our study ( Fig. 4). Nevertheless, the patients included in the study of Andreola et al14 were infants aged up to 36 months, which could explain the better performance of the tests. In 2008, Maniaci et al19 published the first study on PCT as a marker of bacterial infection in infants aged younger than 90 days, finding an AUC of 0.82 that was clearly superior to that of WBC and ANC. However, in this study, CRP values were not analyzed. A year later, Olaciregui et al16 found similar predictive values for CRP and PCT for the diagnosis of PSBI. Nevertheless, when they only took into account IBI, such as sepsis, bacteremias, or meningitis, the results of PCT were clearly superior to those of CRP, with positive LR of 2.15 for CRP and 12.3 for PCT. This may be also observed in our study,
as in the IBI group, the AUC for PCT (0.77) was higher than AUC for CRP (0.54). However, in urinary infections, ROC curves showed a similar diagnostic performance for CRP and PCT. The study published by Gómez et al17 in 2012 assessed PCT and CRP values in well-appearing infants aged younger than 90 days with FWS. In this study, CRP shows a better diagnostic performance than in our study, although the cutoff point used to dismiss IBI is slightly lower (2 mg/dL). It should be highlighted that among the 23 infants diagnosed with IBI, 6 had a CRP less than 20 mg/L (4 of them had PCT >2 ng/mL), and 4 had PCT less than 0.5 ng/mL (none with a CRP >40 mg/L). In our study, even if we used a lower cutoff point for CRP (2 mg/dL), only 4 of the 11 IBIs would be diagnosed, which shows the low diagnostic performance of this test, with a positive LR of 1.69 (0.41–2.41) and a negative LR of 0.81 (0.7–1.42). Another possible explanation for the poor results of CRP in our study could be the short duration of fever in infants with an IBI, as 3 of them had less than 4 hours of fever, 2 of them had 4 to 6 hours, and only 4 of them had 6 to 12 hours. Only 1 patient with IBI had more than 12 hours of fever when tests were performed. It is known that PCT levels in blood rise more rapidly than CRP levels, reaching a peak of 6 hours after the infectious insult.14,20,21 In the study by Gómez et al,17 the young febrile infants who had a shorter duration of fever with normal urine sediment were analyzed separately, and PCT presented better results than CRP (AUC values 0.819 vs 0.563), similar to the data found in our study. Olaciregui et al16 had previously observed that in febrile infants with less than 12 hours of fever, there was a trend
FIGURE 2. ROC curve of WBC, ANC, CRP, and PCT for the detection of IBI.
FIGURE 3. ROC curve of WBC, ANC, CRP, and PCT for the detection of PSBI.
4
www.pec-online.com
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Detecting Bacterial Infection in Febrile Infants
FIGURE 4. Areas under ROC curves for the diagnosis of IBI and PSBI.
toward improvement in the predictive value of PCT compared with CRP. Despite this, most patients diagnosed with IBI (10 of 11) presented abnormal values in at least one of the analytical parameters and/or physical appearance. Both physical examination and all analytical parameters were normal in only 1 patient with occult bacteremia by Enterococcus faecalis, who had had fever only for 6 hours before being seen at the ED. Regarding patients diagnosed with IBI who were well appearing on the initial physical examination (5 of 11 patients), most of them (4) presented abnormal results in at least one of the analytical parameters. Therefore, although the performance of the analytical parameters when used in isolation is quite limited, the development of tools combining different tests including the new biomarkers, such as PCT, could increase the reliability of the tests for the diagnosis of IBI in these patients. Several authors suggest managing these patients “stepby-step,”18 combining physical examination, PCT, CRP, and other biomarkers to improve the detection of children at risk for IBI. The study is subject to inherent limitations of retrospective studies. In a significant number (150) of infants aged younger than 3 months with FWS, CRP and/or PCT values had not been determined, and therefore, these infants could not be included in the study. This fact could have biased some results. However, CRP and PCT missing values were uniformly distributed throughout the study period, making biases less likely. In this group of patients, 4 bacteremias were diagnosed. Moreover, because a lumbar puncture was not performed in every patient, some bacterial meningitis could have been left undiagnosed. Nevertheless, clinical records of every infant without lumbar puncture were examined, and none of the infants who were discharged, admitted to hospital, or admitted to pediatric ICU were diagnosed with bacterial meningitis afterwards. In summary, neither physical examination nor laboratory tests accurately identify all the febrile infants aged younger than 3 months with IBI. Although PCT has the best diagnostic performance, it could not identify almost 30% of the infants with IBI. Therefore, the management of infants aged younger than 3 months should be performed with caution, even if physical examination and laboratory tests results are normal, and we suggest a careful observation period in infants aged younger than 3 months with FWS, especially in those patients with a few hours of fever. ACKNOWLEDGMENT The authors thank Alicia Germán for her help in writing the article. REFERENCES 1. Nelson DS, Walsh K, Fleisher GR. Spectrum and frequency of pediatric illness presenting to a general community hospital emergency department. Pediatrics. 1992;90:5–10. 2. Massin MM, Montesanti J, Gerard P, et al. Spectrum and frequency of illness presenting to a pediatric emergency department. Acta Clin Belg. 2006;61:161–165. 3. Baraff LJ. Management of fever without source in infants and children. Ann Emerg Med. 2000;36:602–614.
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
4. Bleeker SE, Moons KGM, Lubsen GD, et al. Predicting serious bacterial infection in children with fever without apparent source. Acta Paediatr. 2001;90:1226–1232. 5. Maheshwari N. How useful is C-reactive protein in detecting occult bacterial infection in young children with fever without apparent focus? Arch Dis Child. 2006;91:533–535. 6. Bonadio WA, McElroy K, Jacopy PL, et al. Relationship of fever magnitude to rate of serious bacterial infections in infants aged 4-8 weeks. Clin Pediatr (Phila). 1991;30:478–480. 7. McCarthy PL, Scarpe MR, Spiesel SZ, et al. Observation scale to identify serious illness in febrile children. Pediatrics. 1982;70:802–809. 8. McCarthy PL, Lembo RM, Baron MA, et al. Predictive value of abnormal physical examination findings in ill-appearing and well-appearing febrile children. Pediatrics. 1985;76:167–171. 9. Baker MD, Avner JR, Bell LM. Failure of infant observation scale in detecting serious illness in febrile, 4 to 8 week-old infants. Pediatrics. 1990; 85:1040–1043. 10. Baker MD, Bell LM, Avner JR. Outpatient management without antibiotics of fever in selected infants. N Engl J Med. 1993;329:1437–1441. 11. Baskin MN, O'Rourke EJ, Fleisher GR. Outpatient treatment of febrile infants 28 to 89 days of age with intramuscular administration of ceftriaxone. J Pediatr. 1992;120:22–27. 12. Jaskiewicz JA, McCarthy CA, Richardson AC, et al. Febrile infants at low risk for serious bacterial infection—an appraisal of the Rochester criteria and implications for management. Febrile Infant Collaborative Study Group. Pediatrics. 1994;94:390–396. 13. Pulliam PN, Attia MW, Cronan KM. C-reactive protein in febrile children 1 to 36 months of age with clinically undetectable serious bacterial infection. Pediatrics. 2001;108:275–1279. 14. Andreola B, Bressan S, Callegaro S, et al. Procalcitonin and C-reactive protein as diagnostic markers of serious bacterial infections in febrile infants and children in the emergency department. Pediatr Infect Dis J. 2007;26:672–677. 15. Gomez B, Mintegi S, Benito J, et al. Blood culture and bacteremia predictors in infants less than three months of age with fever without source. Pediatr Infect Dis J. 2010;29:43–47. 16. 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. 17. Gomez B, Bressan B, Mintegi S, et al. Diagnostic value of procalcitonin in well-appearing young febrile infants. Pediatrics. 2012;130:815–822. 18. Mintegi S, Bressan D, Gomez B, et al. Accuracy of a sequential approach to identify young febrile infants at low risk for invasive bacterial infection. Emerg Med J. 2013;0:1–6. 19. Maniaci V, Dauber A, Weiss S, et al. Procalcitonin in young febrile infants for the detection of serious bacterial infections. Pediatrics. 2008;122: 701–710. 20. Dandona P, Nix D, Wilson MF, et al. Procalcitonin increase after endotoxin injection in normal subjects. J Clin Endocrinol Metab. 1994;79: 1605–1608. 21. Assicot M, Gendrel D, Carsin H, et al. High serum procalcitonin concentrations in patiens with sepsis and infection. Lancet. 1993;341: 515–518.
www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
5
Lack of Accuracy of Biomarkers and Physical Examination to Detect Bacterial Infection in Febrile Infants Marta Germán Díaz, MD, Rosa Pavo García, MD, Daniel Blázquez Gamero, MD, M. Isabel González-Tomé, PhD, Pilar Cedena Romero, MD, Milagros Marín Marín, MD, and Jesús Ruiz Contreras, PhD Objectives: The aim of this study was to analyze the usefulness of physical examination, C-reactive protein (CRP), procalcitonin (PCT), white blood cell (WBC) count, and absolute neutrophils counts (ANCs) for the diagnosis of invasive bacterial infections (IBIs) and potentially serious bacterial infections in infants younger than the age of 3 months presenting with fever without source (FWS) to the emergency department (ED). Methods: A descriptive retrospective study that includes all infants aged younger than 3 months who presented with FWS to the ED between July 2008 and January 2012. We evaluated diagnostic performance for each test by receiver operating characteristic curve analysis. Sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio were also calculated. Results: Three hundred eighteen patients met the inclusion criteria. Eleven bacteremia (3.5%) and 76 urinary tract infections (23.9%) were diagnosed. To detect IBI, the areas under the curve for the different tests were as follows: PCT, 0.77 (95% confidence interval [CI], 0.57–0.96); CRP, 0.54 (95% CI, 0.36–0.73); ANC, 0.53 (95% CI, 0.34–0.71); and WBC, 0.42 (0.24–0.61). To detect potentially serious bacterial infections, the areas under the curve were as follows: PCT, 0.66 (95% CI, 0.59–0.74); CRP, 0.68 (0.60–0.76); ANC, 0.64 (0.56–0.71); and WBC, 0.66 (0.58–0.72). Conclusions: Procalcitonin is better than CRP, WBC, and ANC to confirm or dismiss the presence of an IBI in infants aged younger than 3 months presenting with FWS to the ED. However, it could not identify almost 30% of infants with IBI. Most patients diagnosed with IBI (10 of 11) presented abnormal values in at least one of the analytical parameters and/or physical appearance. Four of 5 patients with IBI and well appearing presented abnormal results in at least one of the analytical parameters. Therefore, the development of tools combining different tests including the new biomarkers could increase the reliability of the tests for the diagnosis of IBI in these patients. Key Words: C-reactive protein, fever without source, infants younger than 3 months, invasive bacterial infection, potentially serious bacterial infections, procalcitonin (Pediatr Emer Care 2015;00: 00–00)
F
ever without source (FWS) is one of the most frequent reasons for pediatric emergency consultations,1,2 and it has been estimated that up to 10% to 25% of cases of FWS in infants aged younger than 3 months are caused by a potentially serious bacterial infection (PSBI).3–5 In some cases, infants in this age group and a PSBI may have a normal physical examination in the emergency department (ED).6–9 Therefore, it is necessary to have reliable analytical markers for early detection of PSBI in these patients. From the Pediatrics Department, 12 de Octubre University Hospital, Madrid, Spain. Disclosure: The authors declare no conflict of interest. Reprints: Marta Germán Díaz, MD, Pediatrics Department, 12 de Octubre University Hospital, Avenida de Córdoba s/n, 28041, Madrid, Spain (e‐mail: [email protected]). The authors declare no support from pharmaceutical industry or other organizations. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0749-5161
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Since the 1990s, several algorithms have been developed for the management of infants aged younger than 3 months with FWS.10–12 These algorithms classify infants according to the risk of having an invasive bacterial infection (IBI), which is estimated using analytical and clinical parameters. Over the past few years, new markers of bacterial infection, such as C-reactive protein (CRP) and procalcitonin (PCT), have been added to these algorithms.13–15 Several studies have evaluated the usefulness of CRP and PCT for the management of children with FWS. However, data regarding the group of infants aged younger than 3 months are scarce or are limited to well-appearing infants.16,17 The objective of our study was to analyze the usefulness of CRP, PCT, white blood cell (WBC) count, and absolute neutrophil count (ANC) for the diagnosis of IBI and PSBI in infants aged younger than 3 months presenting with FWS to the ED.
METHODS Characteristics of the Patients and Inclusion Criteria We conducted a descriptive retrospective study in the pediatric ED of a tertiary teaching hospital with about 60,000 visits annually. All infants aged younger than 90 days presenting with FWS to the ED between July 2008 and January 2012 were included in the study. Fever was defined as an axillary temperature of greater than or equal to 38°C (100.4°F) measured at home and/or in the ED. The following patients were excluded from the study: preterm neonates with a gestational age less than 37 weeks, infants with chronic diseases or congenital malformations, and infants who had been previously hospitalized or treated with antibiotics.
Study Design The protocol for the management of infants aged younger than 3 months with FWS at our hospital includes a complete blood cell count, CRP and PCT levels, a blood culture, a urine dipstick, a urine sediment, and a urine culture collected by urinary catheterization or suprapubic puncture. A lumbar puncture was performed in not well-appearing infants (Yale Observation Scale [YIOS] score >7) and in those with irritability or analytical alterations (WBC count >15,000/mm3 or 5 WBCs/field in urine sediment], presence of nitrites or bacteriuria in urine sediment, CRP >3 mg/dL or PCT >0.5 ng/mL elevation). A lumbar puncture was performed before any antibiotic therapy. Invasive bacterial infection was defined as the isolation of pathogenic bacteria (without contaminants) in the 2 bottles of blood culture and/or in cerebrospinal fluid (CSF). Urinary tract infections (UTIs) and pneumonias were considered as PSBI. Diagnosis of UTI required the presence of at least 50,000 colony-forming units/mL of a uropathogen cultured from the urine specimen. Diagnosis of pneumonia required the presence of an infiltrate or consolidation on chest radiograph. www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
1
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Díaz et al
FIGURE 1. Flowchart for identification of infants to be included in the analysis.
To perform the final analysis, we only included patients in whom all the following diagnostic tests had been done at the ED: WBC count, CRP and PCT levels, complete urinalysis, blood culture, and urine culture. Children's overall appearance was assessed by a pediatric emergency physician on the initial physical examination upon arrival to the pediatric ED. The overall appearance value was determined by means of the YIOS, considering infants with YIOS score greater than 7 as not well appearing.7,10 The study was approved by the hospital research committee.
Data Collection Data were collected using the pediatric ED database. The following data were obtained from each patient: demographic data (age, sex), clinical history, maximum temperature at home and at hospital, duration of fever, overall appearance of the infant on arrival to hospital, results of the tests performed (WBC, ANC, CRP and PCT values, urine dipstick, urine sediment and results of blood, urine and CSF cultures), as well as the destination of the patient (discharge, admission to intensive care unit [ICU], or admission to hospital).
Statistical Analysis Statistical analyses were conducted using SPSS software 20.0 (IBM, SPSS, Chicago, Ill). Normally distributed continuous variables were expressed as mean (SD), whereas nonnormally distributed continuous variables were expressed as median and interquartile range. Categorical variables were reported as percentages. The diagnostic performance of the different blood tests was
2
www.pec-online.com
evaluated by receiver operating characteristic (ROC) curve analysis. We calculated the area under the curve (AUC), and we used the best cutoff points for each laboratory parameter. Sensitivity, specificity, positive likelihood ratio (LR), and negative LR were also calculated.
RESULTS Characteristics of Patients During the study period, 518 infants aged younger than 3 months presented with FWS to the ED. We excluded 200 infants TABLE 1. Main Characteristics of Patients Characteristics
n (%)
Sex, male/female Overall appearance Well appearing Not well appearing Duration of fever, h 3 PCT, ng/dL >0.5 >2 Not well appearing
18.2 (5.1–45.7) 72.7 (43.4–90.2) 45 (16–74.9) 54.54 (28–78.7)
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Specificity (95% CI) 84.11 (79.4–88) 85.7 (81.7–88.6) 94.1 (91.5–96.8) 85.3 (80.5–89)
+LR (95% CI)
−LR (95% CI)
1.24 (0.27–3.36)
0.97 (0.77–1.3)
5.14 (3.3–8.7) 7.8 (3.25–17) 3.7 (2–6.84)
0.32 (0.12–0.84) 0.58 (0.34–1) 0.53 (0.28–1.02)
www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
3
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Díaz et al
TABLE 4. Values of Sensitivity, Specificity, Positive LR, and Negative LR for CRP and PCT and Overall Appearance in the Detection of PSBI (Urine Infection) Sensitivity (95% CI) CRP, mg/dL >3 PCT, ng/dL >0.5 >2 Not well appearing
Specificity (95% CI)
39.1 (28.9–49.3)
93 (89.8–96.4)
31 (21.3–40.8) 17.2 (9.3–25.2) 14.12 (6.7–21.5)
89.1 (85.1–93.2) 96.5 (94.2–98.9) 83.4 (79–88.6)
+ LR (95% CI)
−LR (95% CI)
5.6 (3.3–9.7)
0.65 (0.5–0.78)
2.9 (1.8–4.7) 4.9 (2.2–11.3) 0.88 (0.48–1.6)
0.75 (0.64–0.89) 0.86 (0.78–0.95) 1.02 (0.9–1.1)
rate (16%) of false-positive cases, making this test useless in this setting. Our data contrast with previously published studies in which CRP has shown a better diagnostic performance.15,16 The differences obtained may be due to several reasons. First, our study was limited to infants aged younger than 90 days, and the magnitude of response to acute phase reactants in invasive infections may be different in younger infants. Although there are several published studies on the usefulness of CRP and PCT for diagnosis of infants with FWS, few of them are focused on this age group. Andreola et al14 demonstrated the superior discriminative capability of CRP and PCT over WBC or ANC for the detection of PSBI in infants with FWS. However, they could not find a statistically significant difference between CRP and PCT, with an area under the ROC curve (AUC) of 0.85 and 0.82, respectively, values that are much higher than those obtained in our study ( Fig. 4). Nevertheless, the patients included in the study of Andreola et al14 were infants aged up to 36 months, which could explain the better performance of the tests. In 2008, Maniaci et al19 published the first study on PCT as a marker of bacterial infection in infants aged younger than 90 days, finding an AUC of 0.82 that was clearly superior to that of WBC and ANC. However, in this study, CRP values were not analyzed. A year later, Olaciregui et al16 found similar predictive values for CRP and PCT for the diagnosis of PSBI. Nevertheless, when they only took into account IBI, such as sepsis, bacteremias, or meningitis, the results of PCT were clearly superior to those of CRP, with positive LR of 2.15 for CRP and 12.3 for PCT. This may be also observed in our study,
as in the IBI group, the AUC for PCT (0.77) was higher than AUC for CRP (0.54). However, in urinary infections, ROC curves showed a similar diagnostic performance for CRP and PCT. The study published by Gómez et al17 in 2012 assessed PCT and CRP values in well-appearing infants aged younger than 90 days with FWS. In this study, CRP shows a better diagnostic performance than in our study, although the cutoff point used to dismiss IBI is slightly lower (2 mg/dL). It should be highlighted that among the 23 infants diagnosed with IBI, 6 had a CRP less than 20 mg/L (4 of them had PCT >2 ng/mL), and 4 had PCT less than 0.5 ng/mL (none with a CRP >40 mg/L). In our study, even if we used a lower cutoff point for CRP (2 mg/dL), only 4 of the 11 IBIs would be diagnosed, which shows the low diagnostic performance of this test, with a positive LR of 1.69 (0.41–2.41) and a negative LR of 0.81 (0.7–1.42). Another possible explanation for the poor results of CRP in our study could be the short duration of fever in infants with an IBI, as 3 of them had less than 4 hours of fever, 2 of them had 4 to 6 hours, and only 4 of them had 6 to 12 hours. Only 1 patient with IBI had more than 12 hours of fever when tests were performed. It is known that PCT levels in blood rise more rapidly than CRP levels, reaching a peak of 6 hours after the infectious insult.14,20,21 In the study by Gómez et al,17 the young febrile infants who had a shorter duration of fever with normal urine sediment were analyzed separately, and PCT presented better results than CRP (AUC values 0.819 vs 0.563), similar to the data found in our study. Olaciregui et al16 had previously observed that in febrile infants with less than 12 hours of fever, there was a trend
FIGURE 2. ROC curve of WBC, ANC, CRP, and PCT for the detection of IBI.
FIGURE 3. ROC curve of WBC, ANC, CRP, and PCT for the detection of PSBI.
4
www.pec-online.com
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care • Volume 00, Number 00, Month 2015
Detecting Bacterial Infection in Febrile Infants
FIGURE 4. Areas under ROC curves for the diagnosis of IBI and PSBI.
toward improvement in the predictive value of PCT compared with CRP. Despite this, most patients diagnosed with IBI (10 of 11) presented abnormal values in at least one of the analytical parameters and/or physical appearance. Both physical examination and all analytical parameters were normal in only 1 patient with occult bacteremia by Enterococcus faecalis, who had had fever only for 6 hours before being seen at the ED. Regarding patients diagnosed with IBI who were well appearing on the initial physical examination (5 of 11 patients), most of them (4) presented abnormal results in at least one of the analytical parameters. Therefore, although the performance of the analytical parameters when used in isolation is quite limited, the development of tools combining different tests including the new biomarkers, such as PCT, could increase the reliability of the tests for the diagnosis of IBI in these patients. Several authors suggest managing these patients “stepby-step,”18 combining physical examination, PCT, CRP, and other biomarkers to improve the detection of children at risk for IBI. The study is subject to inherent limitations of retrospective studies. In a significant number (150) of infants aged younger than 3 months with FWS, CRP and/or PCT values had not been determined, and therefore, these infants could not be included in the study. This fact could have biased some results. However, CRP and PCT missing values were uniformly distributed throughout the study period, making biases less likely. In this group of patients, 4 bacteremias were diagnosed. Moreover, because a lumbar puncture was not performed in every patient, some bacterial meningitis could have been left undiagnosed. Nevertheless, clinical records of every infant without lumbar puncture were examined, and none of the infants who were discharged, admitted to hospital, or admitted to pediatric ICU were diagnosed with bacterial meningitis afterwards. In summary, neither physical examination nor laboratory tests accurately identify all the febrile infants aged younger than 3 months with IBI. Although PCT has the best diagnostic performance, it could not identify almost 30% of the infants with IBI. Therefore, the management of infants aged younger than 3 months should be performed with caution, even if physical examination and laboratory tests results are normal, and we suggest a careful observation period in infants aged younger than 3 months with FWS, especially in those patients with a few hours of fever. ACKNOWLEDGMENT The authors thank Alicia Germán for her help in writing the article. REFERENCES 1. Nelson DS, Walsh K, Fleisher GR. Spectrum and frequency of pediatric illness presenting to a general community hospital emergency department. Pediatrics. 1992;90:5–10. 2. Massin MM, Montesanti J, Gerard P, et al. Spectrum and frequency of illness presenting to a pediatric emergency department. Acta Clin Belg. 2006;61:161–165. 3. Baraff LJ. Management of fever without source in infants and children. Ann Emerg Med. 2000;36:602–614.
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
4. Bleeker SE, Moons KGM, Lubsen GD, et al. Predicting serious bacterial infection in children with fever without apparent source. Acta Paediatr. 2001;90:1226–1232. 5. Maheshwari N. How useful is C-reactive protein in detecting occult bacterial infection in young children with fever without apparent focus? Arch Dis Child. 2006;91:533–535. 6. Bonadio WA, McElroy K, Jacopy PL, et al. Relationship of fever magnitude to rate of serious bacterial infections in infants aged 4-8 weeks. Clin Pediatr (Phila). 1991;30:478–480. 7. McCarthy PL, Scarpe MR, Spiesel SZ, et al. Observation scale to identify serious illness in febrile children. Pediatrics. 1982;70:802–809. 8. McCarthy PL, Lembo RM, Baron MA, et al. Predictive value of abnormal physical examination findings in ill-appearing and well-appearing febrile children. Pediatrics. 1985;76:167–171. 9. Baker MD, Avner JR, Bell LM. Failure of infant observation scale in detecting serious illness in febrile, 4 to 8 week-old infants. Pediatrics. 1990; 85:1040–1043. 10. Baker MD, Bell LM, Avner JR. Outpatient management without antibiotics of fever in selected infants. N Engl J Med. 1993;329:1437–1441. 11. Baskin MN, O'Rourke EJ, Fleisher GR. Outpatient treatment of febrile infants 28 to 89 days of age with intramuscular administration of ceftriaxone. J Pediatr. 1992;120:22–27. 12. Jaskiewicz JA, McCarthy CA, Richardson AC, et al. Febrile infants at low risk for serious bacterial infection—an appraisal of the Rochester criteria and implications for management. Febrile Infant Collaborative Study Group. Pediatrics. 1994;94:390–396. 13. Pulliam PN, Attia MW, Cronan KM. C-reactive protein in febrile children 1 to 36 months of age with clinically undetectable serious bacterial infection. Pediatrics. 2001;108:275–1279. 14. Andreola B, Bressan S, Callegaro S, et al. Procalcitonin and C-reactive protein as diagnostic markers of serious bacterial infections in febrile infants and children in the emergency department. Pediatr Infect Dis J. 2007;26:672–677. 15. Gomez B, Mintegi S, Benito J, et al. Blood culture and bacteremia predictors in infants less than three months of age with fever without source. Pediatr Infect Dis J. 2010;29:43–47. 16. 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. 17. Gomez B, Bressan B, Mintegi S, et al. Diagnostic value of procalcitonin in well-appearing young febrile infants. Pediatrics. 2012;130:815–822. 18. Mintegi S, Bressan D, Gomez B, et al. Accuracy of a sequential approach to identify young febrile infants at low risk for invasive bacterial infection. Emerg Med J. 2013;0:1–6. 19. Maniaci V, Dauber A, Weiss S, et al. Procalcitonin in young febrile infants for the detection of serious bacterial infections. Pediatrics. 2008;122: 701–710. 20. Dandona P, Nix D, Wilson MF, et al. Procalcitonin increase after endotoxin injection in normal subjects. J Clin Endocrinol Metab. 1994;79: 1605–1608. 21. Assicot M, Gendrel D, Carsin H, et al. High serum procalcitonin concentrations in patiens with sepsis and infection. Lancet. 1993;341: 515–518.
www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
5
