OBJECTIVES. The objectives of the study were (1) to study the test performance of procalcitonin for identifying serious bacterial infections in febrile infants ≤90 days of age without an identifiable bacterial source and (2) to determine an optimal cutoff value to identify infants at low risk for serious bacterial infections.
METHODS. A prospective observational study was performed with febrile infants ≤90 days of age presenting to an urban, pediatric, emergency department. Serum procalcitonin levels were measured by using an automated high-sensitivity assay. An optimal procalcitonin cutoff value was selected to maximize sensitivity and negative predictive value for the detection of serious bacterial infections. Infants were classified as having definite, possible, or no serious bacterial infections.
RESULTS. A total of 234 infants (median age: 51 days) were studied. Thirty infants (12.8%) had definite serious bacterial infections (bacteremia: n = 4; bacteremia with urinary tract infections: n = 2; urinary tract infections: n = 24), and 12 infants (5.1%) had possible serious bacterial infections (pneumonia: n = 5; urinary tract infections: n = 7). Mean procalcitonin levels for definite serious bacterial infections (2.21 ± 3.9 ng/mL) and definite plus possible serious bacterial infections (2.48 ± 4.6 ng/mL) were significantly higher than that for no serious bacterial infection (0.38 ± 1.0 ng/mL). The area under the receiver operating characteristic curve was 0.82 for definite serious bacterial infections and 0.76 for definite and possible serious bacterial infections. For identifying definite and possible serious bacterial infections, a cutoff value of 0.12 ng/mL had sensitivity of 95.2%, specificity of 25.5%, negative predictive value of 96.1%, and negative likelihood ratio of 0.19; all cases of bacteremia were identified accurately with this cutoff value.
CONCLUSIONS. Procalcitonin has favorable test characteristics for detecting serious bacterial infections in young febrile infants. Procalcitonin measurements performed especially well in detecting the most serious occult infections.
- serious bacterial infection
- urinary tract infection
- diagnostic tests
The appropriate diagnostic evaluation and treatment of infants with fever and no clear source of infection has been an ongoing challenge for clinicians. This challenge has been greatest for infants <3 months of age, who have the highest risk of serious bacterial infections (SBIs), such as bacteremia, meningitis, urinary tract infection (UTI), and pneumonia. For infants who present with fever without an identifiable source, the literature reports rates of SBIs of up to 12% in infants <28 days of age and up to 9% in infants 1 to 3 months of age.1–7 Previous publications demonstrated that clinical impression based on examination results alone is inadequate to exclude SBIs.2,8–11 With the inherent limitations of physical examination, guidelines using laboratory investigations to identify infants at high and low risk of SBIs have been suggested.2,7,12 Because of ongoing debate and lack of uniformity in practice, a set of recommendations appeared simultaneously in the pediatric and emergency medicine literature4,13; despite the effort, the approach to febrile infants remains a controversial topic.11,14–17
In an attempt to refine the care of young febrile infants, biomarkers have been considered to differentiate infants at higher risk for bacterial infections more accurately and objectively.18–23 Procalcitonin is one new biomarker that has been studied recently. Procalcitonin is almost undetectable under physiologic conditions but increases to high levels in pyelonephritis, bacteremia, sepsis, and bacterial meningitis.18–20,24–27 Procalcitonin seems to show a “dose-response” relationship, with higher levels correlating with increased severity of infection.28 Although 3 previous studies showed procalcitonin to have greater sensitivity and negative predictive value (NPV) than other, more-traditional markers of infection,19,29,30 such as leukocyte count and absolute neutrophil count (ANC), no study focused primarily on the performance of procalcitonin in young febrile infants <90 days of age presenting to an emergency department (ED). The objectives of this study were to study the test performance of a new automated sensitive assay for procalcitonin (PCT sensitive Kryptor; Brahms, Hennigsdorf, Germany) with febrile infants <90 days of age without a source of infection in physical examination and to determine an optimal cutoff value for procalcitonin to identify infants at low risk for SBIs.
Setting and Participants
The study was conducted at an urban, academic, pediatric hospital with 54000 ED visits per year. Infants ≤90 days of age with a documented body temperature of ≥38.0°C were studied.
The objectives of the study were to study the test performance of procalcitonin for identifying SBIs in febrile infants ≤90 days of age without an identifiable bacterial source (by examination) and to determine an optimal cutoff value for procalcitonin to identify infants at low risk for SBIs.
We conducted a prospective cohort study with subject recruitment over an 18-month period, from October 2005 to March 2007. Infants with age of ≤90 days and measured temperature of ≥38.0°C who were seen in the ED were eligible for enrollment. Infants with a previously identified immunodeficiency or chronic disease, focal bacterial infection (other than otitis media) on physical examination, vesicoureteral reflux requiring antibiotic prophylaxis, surgery in the previous 7 days (excluding neonatal circumcision), immunizations in the 48 hours preceding the visit, or antibiotic treatment within the previous 48 hours were excluded. Infants of all gestational ages were included. All subjects received clinical care as determined by the treating pediatric emergency medicine physician. Institutional guidelines for the care of febrile infants ≤90 days of age included a complete blood count with differential, blood culture, urinalysis and urine culture with samples collected through bladder catheterization, cerebrospinal fluid (CSF) cell count, protein level, and glucose level analyses, Gram-staining, and culture, chest radiograph if pneumonia was suggested by physical examination, and stool fecal leukocyte count and culture if clinical history or physical examination suggested possible bacterial gastroenteritis (eg, presence of bloody or heme-positive diarrhea). As determined in previous investigations at this institution, ∼78% of febrile infants <90 days of age undergo CSF examinations and 88% of those patients undergo complete blood count and blood culture analyses.1 During the study period, parents or caregivers of infants who were having blood drawn for clinical evaluation were approached to participate in the study, and informed consent was obtained by an attending physician for use of blood remaining after clinical tests (if ordered by the clinical team).
To ensure identification of all potentially eligible febrile infants and to assess a capture rate for the study, an electronic log of ED visits was reviewed daily. The medical record was reviewed for all infants ≤90 days of age, regardless of chief complaint, to identify potentially missed cases. Infants' caregivers who had not been approached for consent during the ED visit were called by the treating ED physician and offered enrollment in the study (if blood had been drawn at the ED visit). In those cases, verbal informed consent was obtained.
Definitions and Outcome Measures
Patients were classified as having a definite SBI, possible SBI, or no SBI. Definite SBIs were (1) bacteremia, as a positive blood culture result with a pathogen; (2) UTI, as a urine culture (from catheterization) with ≥50000 colony-forming units (CFUs) per mL of a single pathogen or 10000 to 49000 CFUs per mL with positive urinalysis results (defined as leukocyte esterase and nitrite positive in dipstick testing or ≥5 white blood cells [WBCs] per high-power field on microscopic examination of a centrifuged urine specimen); (3) bacterial meningitis, as a positive CSF culture result with a pathogen or bacteremia with CSF pleocytosis (>10 WBCs per μL); (4) bacterial pneumonia, as a positive pleural fluid culture result with a pathogen or a chest radiograph interpreted by an attending radiologist as indicating pneumonia with a positive blood or sputum culture result with a respiratory pathogen; or (5) bacterial gastroenteritis, as a bacterial pathogen in stool culture. Possible SBIs were defined as (1) UTI, as a urine culture with 10000 to 49000 CFUs per mL of a single pathogen with a negative urinalysis result (as defined above); or (2) bacterial pneumonia, as a chest radiograph interpreted by an attending radiologist as indicating pneumonia or possible pneumonia in the absence of a positive pleural fluid, sputum, or blood culture result. In the uncommon circumstance of multiple pathogens in a urine culture, the culture result was classified as a definite UTI if there was a single dominant pathogen (≥50000 CFUs per mL) or as a possible UTI if there were multiple pathogens at ≥50000 CFUs per mL, regardless of the urinalysis result. All other patients were considered not to have a SBI. The final classification (definite SBI, possible SBI, or no SBI) was determined through consensus review by the 4 authors based at the primary study site, before knowledge of the procalcitonin results.
At the time of enrollment, the attending physician responsible for the care of the patient completed a questionnaire to assess the overall appearance of the infant on a 5-point scale (with anchors of 1 = moribund, toxic, ill-appearing, unresponsive, and 5 = perfectly healthy, interactive infant). The electronic ED medical record was reviewed for inclusion criteria such as age, history, presence of fever without a focal bacterial source on examination, triage temperature (rectal), and laboratory and radiographic results.
Blood samples that remained after clinical tests were stored for up to 1 week in a 4°C refrigerator before processing. Procalcitonin is stable at room temperature and at lower temperatures typically used for specimen storage.31 Within the 1-week time frame, the blood specimens were centrifuged at 4000 rpm for 10 minutes, and the plasma was then stored at −80°C. Procalcitonin was measured at a reference laboratory by using an immunometric assay with time-resolved amplified cryptate emission technology (PCT sensitive Kryptor kit; Brahms).32 This method relies on nonradiative energy transfer between 2 fluorescent tracers (europium cryptate and XL665) during antibody-antigen complex formation. Antiprocalcitonin antibodies are added with the 2 tracers to 250 μL of serum or plasma; the resultant amplification of fluorescence is a measure of the stimulation of XL665 by the europium cryptate at a wavelength ratio of 665/620 nm. The intensity of the signal is proportional to the amount of procalcitonin. The concentration of procalcitonin is calculated from an internal procalcitonin standard curve. For certain samples, purified procalcitonin-free plasma was used to reach a minimal assay volume of 250 μL, with appropriate adjustment in the procalcitonin result. Once the Kryptor system has been calibrated, specimen processing takes ∼20 minutes. The functional sensitivity of the assay is 60 pg/mL. The laboratory investigators were blinded to the identity of all clinical information about the subjects.
Statistical analysis was performed by using SPSS 14.0 (SPSS, Chicago, IL). Demographic characteristics of and laboratory values for subjects with and without SBIs were compared. For continuous variables, independent-sample t tests and the nonparametric Wilcoxon rank test were used; for categorical data, Fisher's exact test or χ2 analysis was used. Differences with a P value of ≤.05 were considered statistically significant. Analyses were performed between definite SBIs and no SBI and between definite plus possible SBIs and no SBI.
The overall test performance of procalcitonin in identifying low-risk (for SBI) patients was reviewed by using receiver operating characteristic (ROC) curve analysis. Given the objective of identifying low-risk patients, cutoff points that maximized sensitivity and NPV were identified. The sensitivity, specificity, positive and NPVs, and positive and negative likelihood ratios were calculated for specific procalcitonin cutoff values; this analysis was performed for both the definite SBI and combined definite and possible SBI outcomes. The 95% confidence intervals (CIs) for test characteristics were calculated by using Stata 6 (Stata, Dallas, TX).
The institutional review board of the hospital approved the study and the informed consent process. This study was compliant with the Health Insurance Portability and Accountability Act of 1996.
A total of 874 patients, ≤90 days of age, with documented temperatures of ≥38°C were evaluated in the ED during the 18-month study period. One hundred sixty (18.3%) were excluded for ≥1 of the following reasons: chronic disease (n = 10), immunizations within the previous 48 hours (n = 112), antibiotic use in the previous 48 hours (n = 36), surgery in the previous 7 days (n = 1), focal bacterial infection (n = 4), and known vesicoureteral reflux (n = 4). Seven of the subjects met >1 exclusion criteria. Of the remaining 714 eligible infants, informed consent was obtained for 435 infants (61%) to participate in the study, of whom 234 had adequate blood samples for procalcitonin determination. Procalcitonin levels could not be determined for the other 201 subjects because a blood sample was not sent from the ED (n = 5), the remaining blood sample could not be located in the laboratory storage refrigerator (n = 25), or the remaining sample was either too hemolyzed or of insufficient quantity for accurate measurement of procalcitonin (n = 171).
Demographic and laboratory parameters for the 234 subjects with informed consent and available procalcitonin measurements were compared with those for the 480 infants for whom consent to measure procalcitonin could not be obtained or for whom consent was obtained but procalcitonin levels could not be determined (Table 1). With the exceptions of gender (53% male subjects among eligible subjects with procalcitonin measurements, compared with 62% male subjects among eligible subjects without procalcitonin measurements; P = .03) and rates of lumbar puncture performed (84% for eligible subjects with procalcitonin measurements, compared with 75% for eligible subjects without procalcitonin measurements; P = .01), the 2 groups did not differ. The rates of definite SBIs did not differ statistically between patients with measured procalcitonin levels and patients without measured procalcitonin levels (12.8% vs 9.7%; P = .21), but the difference in the rates of definite and possible SBIs did reach statistical significance (17.9% vs 12.5%; P = .05).
The median age and temperature of the 234 subjects included in the study population were 51 days (interquartile range: 31–70 days) and 38.6°C (interquartile range: 38.3–38.9°C), respectively. Forty-eight infants (20.5%) were ≤28 days of age. Blood cultures were performed for 100% of the subjects, urine cultures for 97%, CSF cultures for 84%, and chest radiographs for 37%. The median value for clinical impression (assigned by the treating physician) was 4 (interquartile range: 4–5) for all infants. Fifty-one percent of subjects were admitted to the hospital, including 38% of the patients >28 days of age.
Definite and Possible SBIs
Thirty subjects (12.8%) had definite SBIs, 12 (5.1%) had possible SBIs, and 192 had no SBI. Among those with definite SBIs, there were 24 infants with UTIs, 2 infants with concurrent UTIs and bacteremia, and 4 infants with bacteremia. There were no cases of bacterial meningitis, definite bacterial pneumonia, or bacterial gastroenteritis. Of those with possible SBIs, 7 subjects had UTIs and 5 subjects had pneumonia. Among the 7 possible UTIs, 6 patients had low colony counts (10000–49000 CFUS per mL) of a single pathogen with negative urinalysis results and 1 had ≥50000 CFUs per mL of multiple pathogens with negative urinalysis results. Among infants ≤28 days of age, there were 9 definite SBIs (8 UTIs and 1 case of Group B streptococcal bacteremia) and 5 possible SBIs (3 UTIs and 2 cases of pneumonia). Table 2 shows the distribution of all SBIs. The characteristics of subjects with definite SBIs, definite and possible SBIs, and no SBIs are shown in Table 3.
Group B streptococcus (Streptococcus agalactiae) was the causative organism for the 4 subjects with bacteremia (1 subject ≤28 days of age and 3 subjects >28 days of age), and 2 subjects had Escherichia coli bacteremia with concurrent UTI (both >28 days of age). Urinary tract pathogens included 23 cases of E coli (21 definite and 2 possible SBIs), 2 cases of Klebsiella pneumoniae (1 definite and 1 possible SBI), 4 cases of Enterococcus (2 definite and 2 possible SBIs), 1 case of Staphylococcus aureus (possible SBI), and 1 urine culture with ≥50000 CFUs per mL of both E coli and Enterococcus (possible SBI).
The mean ± SD procalcitonin levels were 2.21 ± 3.89 ng/mL for patients with definite SBIs, 2.48 ± 4.59 ng/mL for patients with definite and possible SBIs, and 0.38 ± 1.04 ng/mL for patients without SBIs (Table 3). The mean procalcitonin levels were 2.52 ± 2.72 ng/mL for patients with bacteremia and 2.20 ± 4.3 ng/mL for patients with UTIs. With regard to specific bacterial isolates, the mean procalcitonin levels were 0.44 ± 0.27 ng/mL for E coli bacteremia (n = 2) and 3.56 ± 2.81 ng/mL for Group B streptococcal bacteremia (n = 4). The distributions of procalcitonin values for patients with definite UTIs and bacteremia and those with no SBI are presented in Fig 1. The mean WBC count was significantly higher for patients with definite or definite plus possible SBIs, compared with patients without SBI (Table 3).
We further subdivided procalcitonin levels among the various groups according to age. For patients ≤28 days of age, the mean procalcitonin level for infants with definite SBIs was 3.41 ± 6.46 ng/mL, definite plus possible SBIs, 4.53 ± 7.21 ng/mL, and no SBIs, 0.40 ± 0.44 ng/mL. For patients >28 days of age, the mean procalcitonin level for infants with definite SBIs was 1.69 ± 2.05 ng/mL, definite plus possible SBIs, 1.45 ± 1.91 ng/mL, and no SBIs, 0.38 ± 1.13 ng/mL (Table 4). The mean procalcitonin level was significantly higher for patients with definite or definite plus possible SBIs, compared with patients without SBI, for patients ≤28 days of age and those >28 days of age (Table 4).
ROC curves for procalcitonin level, WBC count, and ANC were constructed by comparing patients with definite SBIs and those with no SBI, as well as those with definite and possible SBIs and those with no SBI (Fig 2). These curves were used to identify various cutoff values with differing sensitivities and specificities. In comparing definite SBIs and no SBI, the ROC area under the curve (AUC) was 0.82; a cutoff value of 0.13 ng/mL yielded sensitivity of 96.7% (95% CI: 81.0%–99.8%), specificity of 30.3% (95% CI: 24.0%–37.5%), NPV of 98.3% (95% CI: 89.7%–99.9%), and negative likelihood ratio of 0.11 (95% CI: 0.02–0.76). In discriminating definite plus possible SBIs and no SBI, the ROC curve had an AUC of 0.76. For this analysis, a cutoff value of 0.12 ng/mL had sensitivity of 95.2% (95% CI: 83%–99%), specificity of 25.5% (95% CI: 20%–32%), NPV of 96.1% (95% CI: 85.4%–99.3%), and negative likelihood ratio of 0.19 (95% CI: 0.05–0.74). With either of these cutoff values, 2 patients with low-colony count UTIs with negative urinalysis results would have been misclassified as being at low risk for SBIs. All 6 patients with bacteremia would have been correctly identified as being at high risk for SBIs. The overall performance of procalcitonin levels in ROC curve analysis had an AUC of 0.85 for patients >28 days of age, compared with an AUC of 0.73 for patients ≤28 days of age (analysis with definite plus possible SBIs).
The evaluation and care of febrile infants has been a subject for investigation and debate for decades; in a broad Medline search for “fever” and “infants” conducted on September 1, 2007, it was found that 5090 publications were published in the past decade alone. In an attempt to balance risk and benefit, clinicians must consider aggressive invasive evaluations to identify a minority of infants with serious illnesses; a universal approach that facilitates risk stratification has been complicated by imperfect and often unvalidated strategies described in the medical literature, the varying acceptance of risk by individual physicians and physician groups, and the inherent medicolegal implications related to establishing standards of care.
Many strategies and management guidelines have attempted to use both clinical and laboratory data to identify febrile infants at low risk for SBIs.2,7,8Many of the low-risk criteria have included nontoxic clinical appearance, previously healthy term infant, no focal bacterial infection on examination, and normal laboratory screening test results. The exact laboratory studies and the criteria for normal values vary between guidelines.2,7,8 Baker et al2 reported sensitivity of 98% and NPV of 99.7% with application of the Philadelphia criteria in identifying infants at low risk for SBIs. The Rochester criteria had similar performance, with sensitivity of 92% and NPV of 98.9%.7 In a more-recent repeat evaluation of the Philadelphia and Rochester criteria, NPVs were 97.1% (95% CI: 85.1%–99.8%) and 97.3% (95% CI: 90.5%–99.2%), respectively.14 However, the 25% SBI rate reported in this study exceeds all previous estimates of SBIs in this age group, and both protocols misclassified cases of bacteremia into low-risk groups, emphasizing the difficulty of studying this patient population and interpreting the literature.
The inability to predict accurately the presence of SBIs on the basis of clinical impression and routine laboratory tests has led to the investigation of various biomarkers of SBIs.23,25 To date, no marker, either alone or in combination, has demonstrated adequate sensitivity or specificity for the diagnosis of SBIs in well-appearing febrile infants. Many algorithms have attempted to use peripheral WBC counts in the evaluation of febrile infants, mostly because of the availability of testing in both office- and hospital-based settings and the proven value in the evaluation of older infants.33 In this population, however, WBC counts and ANCs have poor test performance.34,35 In a study of young febrile infants by Bonsu et al,35 the utility of a WBC count cutoff value of 15000 cells per μL in detecting bacteremia yielded sensitivity of 45% and specificity of 78%. Using the same WBC count cutoff value, Galetto-Lecour et al19 reported sensitivity of 52% and specificity of 79% for detecting SBIs in 99 infants 7 days to 36 months of age. The search for a marker that increases quickly with infection has led to numerous reports of the utility of C-reactive protein (CRP) as a potential marker of SBIs in febrile infants.22,36–39 Galetto-Lecour et al19 showed that CRP levels were superior to WBC counts in detecting SBIs, with reported sensitivity and specificity of 79%. Pulliam et al22 demonstrated CRP levels to be a more useful screening tool than WBC counts in febrile infants 1 to 36 months of age, with sensitivity of 79% and specificity of 91%. Despite the encouraging work with CRP, its use has not been incorporated into febrile infant guidelines except in Europe.40
Since the investigation of procalcitonin levels in children by Assicot et al28 in 1993, numerous investigations have reported procalcitonin as a diagnostic marker of bacterial infection, as well as a prognostic indicator.41 The majority of pediatric clinical investigations have focused on identifying bacterial infections in critically ill infants and children,42–48 discriminating bacterial from viral causes of meningitis and pneumonia,26,49–58 distinguishing pyelonephritis from cystitis,59–63 predicting bacteremia in febrile neutropenic patients,64,65 and identifying serious infections among pediatric outpatients.19,20,24,25,29,30,66,67
Because many of the previous investigations studied critically ill children or hospitalized patients, the test performance would likely vary from the less ill patients typically seen in the ED. In the few previous studies that investigated procalcitonin levels in the ED, a much broader range of ages was studied. Many of the studies included infants <3 months of age but did not perform separate analyses for that subgroup. Lacour et al25 performed an observational study at a single pediatric ED in Switzerland, comparing procalcitonin, interleukin 6, interleukin 8, interleukin 1 receptor antagonist, and CRP as markers of SBIs in children 7 days to 36 months of age. The analysis included only 31 infants <3 months of age, of whom 8 had SBIs. For the entire study group, a procalcitonin cutoff value of 0.9 ng/mL offered the best sensitivity (93%) and specificity (78%), compared with CRP (sensitivity: 89%; specificity: 75%) and interleukin 6 (sensitivity: 79%; specificity: 66%). A larger, multicenter, ED study evaluating procalcitonin and CRP in 352 febrile infants between the ages of 1 and 36 months (mean age: 12.3 months) was conducted by Fernández Lopez et al24 in 2003. Infants were divided into 3 groups, that is, viral infection (n = 122), localized bacterial infection (n = 80, including lower tract UTI and bacterial diarrhea), and invasive bacterial infection (n = 150, including meningitis, bacteremia, pyelonephritis determined with renal scans, bone or joint infection, and lobar pneumonia). In that study, the diagnostic performance for detecting bacterial infection was similar for procalcitonin and CRP levels, with ROC AUC values of 0.82 and 0.78, respectively. When only the diagnosis of invasive bacterial infections was considered, the AUC for procalcitonin (AUC: 0.95) was better than and statistically different from that for CRP (AUC: 0.81; P < .001). The optimal cutoff values suggested by the authors for detecting any bacterial infection and for detecting invasive bacterial infection were 0.53 ng/mL (sensitivity: 65%; specificity: 94%; positive predictive value: 96%; NPV: 59%) and 0.59 ng/mL (sensitivity: 91.3%; specificity: 75%; positive predictive value: 69%; NPV: 81%), respectively. In the same study, the authors noted an earlier increase of procalcitonin levels, compared with CRP levels, which confirmed previous observations about procalcitonin's early increase41,68 and therefore its potential utility in the ED setting, where strategies for early detection of illness are needed.
In the only other recent, prospective, observational ED study of procalcitonin, Andreola et al30 enrolled 408 febrile infants, 7 days to 36 months of age, who presented to an ED at a pediatric referral hospital in Italy. Unlike the previous studies, an age subgroup analysis was performed to elucidate the role of procalcitonin in infants <3 months of age. In that study, the SBI rate was 23.1% overall, including 107 infants <3 months of age with a SBI rate of 26.2%. The authors compared the diagnostic value of WBC counts, ANCs, CRP levels, and procalcitonin levels. Both procalcitonin and CRP levels outperformed WBC counts and ANCs, with AUCs of 0.82, 0.85, 0.71, and 0.74, respectively. Procalcitonin had a similar AUC for patients <3 months of age, compared with infants >3 months of age (reported, but no actual values provided). Among all subjects, procalcitonin had the following test characteristics: sensitivity of 73.4%, specificity of 76.4%, positive likelihood ratio of 3.1, and negative likelihood ratio of 0.35 for a procalcitonin cutoff value of 0.5 ng/mL. Comparatively, CRP had a higher sensitivity (88.3%) but a lower specificity (60.8%). Of the measured markers, only procalcitonin had different values for the most invasive infections (P < .0001) and among infants presenting in the first 8 hours of fever (AUC: procalcitonin: 0.92; CRP: 0.75).
Unlike previous ED studies that incorporated a wide age range of febrile children, our study is the largest and the first to evaluate exclusively the performance of procalcitonin to identify SBIs in infants <90 days of age. Furthermore, it is the first study to use a procalcitonin assay with a functional sensitivity in the range of 60 pg/mL, which proved valuable for evaluation in EDs, where patients often present early in their illness. In overall diagnostic performance, procalcitonin had a ROC AUC of 0.82 (for definite bacterial infection), which is equal to the findings by Andreola et al30 in older febrile infants. As noted previously, the overall performance of procalcitonin levels was shown to be much better than that of WBC counts or ANCs. The optimal procalcitonin cutoff value for identifying infants at low risk for SBIs was 0.12 ng/mL, providing a sensitivity of 95.2% and NPV of 96.5%. Although the numbers are small, the most important value for clinicians would be for detection of invasive SBIs such as bacteremia; in this study, a cutoff value of 0.12 ng/mL identified all 6 cases of bacteremia (procalcitonin level range: 0.25–7.3 ng/mL). Importantly, this procalcitonin cutoff value underscores the importance of the previously unstudied, high-sensitivity procalcitonin assay (the common, commercially available, immunoluminetric LUMItest [Brahms] has a functional sensitivity of only ∼0.5 ng/mL47,69). In our population, with a cutoff value of 0.12 ng/mL, 27% of subjects would be identified as low risk. In a population with a SBI prevalence of 7% to 9%, a procalcitonin level of <0.12 ng/mL would decrease the post-test probability of SBIs to 1.45% to 1.8%. Translating the value to the patient, ideally these low-risk febrile infants who appear well and do not have a bacterial focus of infection could be treated with a more-limited diagnostic evaluation and observation, without empiric antibiotic treatment or hospitalization.
The strengths of the study include this being the first study to focus the evaluation of procalcitonin on young febrile infants <90 days of age presenting to the ED. Use of the high-sensitivity assay allowed us to identify a cutoff value that would adequately distinguish infants at low risk for SBIs. The 2 patients with SBIs who were misclassified as being at low risk both had low-colony count UTIs with negative urinalysis results. The major limitations of the study are the small sample size and the concomitant small number of patients with SBIs. The results indicate that procalcitonin assessments worked well for the youngest infants, but the relatively low procalcitonin values for the 2 patients with E coli bacteremia (0.246 and 0.637 ng/mL) should be carefully noted. Although procalcitonin is stable in stored specimens, we did not anticipate the loss of many study specimens because of hemolysis resulting from storage; this would obviously not be an issue if the blood was processed immediately. Without a larger study, many of the clinically logical subgroups cannot be analyzed. In addition, we were unable to capture adequately the duration of fever, which might influence the accuracy of procalcitonin assessments in this population. These infants often present within hours after the onset of fever; however, it would be beneficial in future studies to consider serial measurements or to quantify the duration of fever before “snapshot” measurements. Finally, we had a higher rate of definite SBIs (12%), compared with previous prevalence studies. Although this should have no effect on sensitivity, the actual NPV would be higher in a more-representative population.
With a new, automated, sensitive assay, procalcitonin measurements had favorable test characteristics for detecting SBIs in well-appearing febrile infants <90 days of age without a focus of infection. Furthermore, procalcitonin measurements performed especially well in detecting the most serious occult infections, such as bacteremia. The performance of procalcitonin as a single clinical marker of infection approaches that of popular strategies that incorporate various laboratory studies and clinical impression scores. However, the future utility of procalcitonin likely depends on its combination with other clinical data; better discrimination of infants with bacterial and viral infections could potentially lead to more-focused evaluations of febrile infants.
We are thankful for the financial support of the Frederick H. Lovejoy, Jr, MD, Resident Research Fund and an American Academy of Pediatrics resident research grant. In addition, we acknowledge the technical support related to specimen processing by the General Clinical Research Center at Children's Hospital Boston (National Center for Research Resources, General Clinical Research Centers Program, National Institutes of Health grant M01RR02172).
We acknowledge our physician colleagues in the ED who facilitated enrollment for the study. We gratefully acknowledge the help and expertise of Richard Snider, PhD, and Robyn Neches, BS, for their work on the procalcitonin assays.
- Accepted January 8, 2008.
- Address correspondence to Richard Bachur, MD, Division of Emergency Medicine, 300 Longwood Ave, Boston, MA 02115. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
This study was presented in part at the Pediatric Academic Societies meeting, May 5–8, 2007, Toronto, Ontario, Canada.
What's Known on This Subject
Procalcitonin has been shown to be an accurate discriminator between viral and bacterial infections for older children and adults.
What This Study Adds
This is the first prospective study to evaluate the performance of a high-sensitivity procalcitonin assay for febrile infants <3 months of age.
- ↵Bachur RG, Harper MB. Predictive model for serious bacterial infections among infants younger than 3 months of age. Pediatrics.2001;108 (2):311– 316
- Byington CL, Enriquez FR, Hoff C, et al. Serious bacterial infections in febrile infants 1 to 90 days old with and without viral infections. Pediatrics.2004;113 (6):1662– 1666
- ↵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. Pediatrics.1994;94 (3):390– 396
- Baker MD, Avner JR, Bell LM. Failure of infant observation scales in detecting serious illness in febrile, 4- to 8-week-old infants. Pediatrics.1990;85 (6):1040– 1043
- ↵Baraff L, Schriger D, Bass J, et al. Practice guideline for the management of infants and children 0 to 36 months of age with fever without source. Pediatrics.1993;92 (1):1– 12
- Kadish HA, Loveridge B, Tobey J, Bolte RG, Corneli HM. Applying outpatient protocols in febrile infants 1–28 days of age: can the threshold be lowered? Clin Pediatr (Phila).2000;39 (2):81– 88
- Bauchner H, Pelton SI. Management of the young febrile child: a continuing controversy. Pediatrics.1997;100 (1):137– 138
- ↵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 (5):1054– 1060
- Hatherill M, Tibby SM, Sykes K, Turner C, Murdoch IA. Diagnostic markers of infection: comparison of procalcitonin with C reactive protein and leucocyte count. Arch Dis Child.1999;81 (5):417– 421
- ↵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 (6):1275– 1279
- ↵Fernández Lopez A, Luaces Cubells C, García García JJ, Fernández Pou J. 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 (10):895– 903
- ↵Gendrel D, Raymond J, Assicot M, et al. Measurement of procalcitonin levels in children with bacterial or viral meningitis. Clin Infect Dis.1997;24 (6):1240– 1242
- ↵BRAHMS. Welcome to PCT. Available at: www.procalcitonin.com. Accessed September 1, 2007
- ↵Bonadio WA, Smith D, Carmody J. Correlating CBC profile and infectious outcome: a study of febrile infants evaluated for sepsis. Clin Pediatr (Phila).1992;31 (10):578– 582
- 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 (6):533– 535
- ↵Putto A, Ruuskanen O, Meurman O, et al. C reactive protein in the evaluation of febrile illness. Arch Dis Child.1986;61 (1):24– 29
- ↵Becker KL, Nylen ES, White JC, Muller B, Snider RH Jr. Clinical review 167: procalcitonin and the calcitonin gene family of peptides in inflammation, infection, and sepsis: a journey from calcitonin back to its precursors. J Clin Endocrinol Metab.2004;89 (4):1512– 1525
- Athhan F, Akagunduz B, Genel F, Bak M, Can D. Procalcitonin: a marker of neonatal sepsis. J Trop Pediatr.2002;48 (1):10– 14
- ↵Nylen E, Muller B, Becker KL, Snider R. The future diagnostic role of procalcitonin levels: the need for improved sensitivity. Clin Infect Dis.2003;36 (6):823– 824
- ↵Moulin F, Raymond J, Lorrot M, et al. Procalcitonin in children admitted to hospital with community acquired pneumonia. Arch Dis Child.2001;84 (4):332– 336
- Michelow IC, Olsen K, Lozano J, et al. Epidemiology and clinical characteristics of community-acquired pneumonia in hospitalized children. Pediatrics.2004;113 (4):701– 707
- Leroy S, Adamsbaum C, Marc E, et al. Procalcitonin as a predictor of vesicoureteral reflux in children with a first febrile urinary tract infection. Pediatrics.2005;115 (6). Available at: www.pediatrics.org/cgi/content/full/115/6/e706
- ↵Pecile P, Miorin E, Romanello C, et al. Procalcitonin: a marker of severity of acute pyelonephritis among children. Pediatrics.2004;114 (2). Available at:www.pediatrics.org/cgi/content/full/114/2/e249
- ↵Carrol ED, Newland P, Riordan FA, Thomson AP, Curtis N, Hart CA. Procalcitonin as a diagnostic marker of meningococcal disease in children presenting with fever and a rash. Arch Dis Child.2002;86 (4):282– 285
- Copyright © 2008 by the American Academy of Pediatrics