• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Retrospective Review of Serious Bacterial Infections in Infants Who Are 0 to 36 Months of Age and Have Influenza A Infection

Posted on: Tuesday, 15 March 2005, 03:00 CST

ABSTRACT. Objective. Previous studies of febrile children who were 3 to 36 months of age and had clinically recognizable viral syndromes have shown low rates of concurrent bacteremia. We sought to determine the prevalence of serious bacterial infections (SBIs) among children with influenza A, a viral syndrome that can be established definitively by specific tests.

Methods. We performed a retrospective cross-sectional study of patients who were O to 36 months of age and presented with fever to the emergency department (ED) over 4 consecutive influenza seasons. Chest radiographs and urine and cerebrospinal fluid cultures also were reviewed.

Results. Of 705 included patients, 163 (23%) were influenza positive (IP) and 542 (77%) were influenza negative (IN). Only 1 IP patient was bacteremic (0.6%) versus 23 of the 542 IN control subjects (4.2%). Two (1.8%) of 110 IP cases had urinary tract infections versus 38 (9.9%) of the 382 IN control subjects. Thirteen (25.4%) of 51 IP patients had radiographic evidence of pneumonia versus 99 (41.9%) of 236 IN control subjects. There were no cases of meningitis in 41 cerebrospinal fluid samples obtained from IP patients versus 4 (2.2%) cases of culture-positive meningitis in 179 IN control subjects. A total of 16 (9.8%) SBIs were identified in the IP cases versus 153 (28.2%) in the IN control subjects.

Conclusions. Febrile children with influenza A had a lower prevalence of bacteremia, urinary tract infections, consolidative pneumonia, or any SBI compared with those without influenza A infection in this study. Pediatrics 2005;115:710-718; influenza A, fever, serious bacterial infections, recognizable viral syndrome, testable viral syndrome.

ABBREVIATIONS. Hib, Haemophilus influenzae type b; OB, occult pneumococcal bacteremia; ED, emergency department; SBI, serious bacterial infection; UTI, urinary tract infection; KAT, rapid antigen testing; TCH, Texas Children's Hospital; CSF, cerebrospinal fluid; VC, viral culture; CXR, chest x-rays; WBC, white blood cell count; IP, influenza-positive; IN, influenza-negative; CI, confidence interval; OR, odds ratio; PCV7, heptavalent conjugate pneumococcal vaccine.

Influenza attack rates in healthy children range from 10% to 40% each year.1 Among all age groups, influenza is associated with >20 000 deaths nationwide and >100 000 hospitalizations per year.2,3 In contrast to most other viral respiratory infections, influenza infection can cause a severe illness that may be difficult to distinguish from sepsis or shock. Despite the broad spectrum of disease presentation, the illness typically includes fever of sudden onset, respiratory symptoms (cough, sore throat, runny nose, and nasal congestion), headache, myalgias, and fatigue. In addition to fever and respiratory symptoms, young infants may present with feeding changes, irritability, lethargy, or even sepsis-like syndromes.4

The evaluation of children with fever comprises a significant portion of the daily practice of pediatric emergency medicine. Nationally, fever is the presenting complaint in 22% of children under the age of 3.5 In the era after vaccination for Haemophilus influenzae type b (Hib), the exact rates of bacteremia in febrile children are unknown, but estimates range from 1.8% to 4.2%.6 In well-appearing febrile children, occult pneumococcal bacteremia (OB) is estimated to occur in the emergency department (ED) setting in 1.6% to 3.0% of children who are aged 3 to 36 months, have a temperature ≥39C, and have no focus of infection on examination.6,7 Although OB and serious bacterial infection (SBIs) are of significant concern, most diagnostic evaluations of fever will yield no evidence of bacterial infection, leaving a viral infection as a diagnosis of exclusion.

The evaluation for SBIs and OB in febrile infants who are younger than 36 months consumes health care resources, causes transient discomfort to the patient, and contributes to antibiotic resistance. Previous studies of febrile children with noninfluenza febrile illnesses but with clinically recognizable viral syndromes such as uncomplicated croup, bronchiolitis, varicella, or stomatitis have demonstrated that highly febrile children who are 3 to 36 months of age have a very low rate of bacteremia and need not have blood drawn for cultures.8-12 Verboon-Maciolek et al,13 using enteroviral polymerase chain reaction, suggested limiting hospital admission for neonatal enteroviral infections because none of the enterovirus- positive patients experienced concomitant bacterial infections. Furthermore, studies that have evaluated the risk for SBI in young infants (0-60 days) with concurrent bronchiolitis and/or respiratory syncytial virus have demonstrated a decreased rate of bacteremia with an appreciable risk for urinary tract infection (UTI) in children with bronchiolitis versus children without bronchiolitis.14

The advent of rapid diagnostic testing for specific viral respiratory infections has provided pivotal tools in instituting appropriate isolation procedures and preventing nosocomial infections.15 One study suggested that detection of influenza A by enzyme immunoassay also has had an impact on medical management by decreasing antibiotic use among pediatric ED patients, by decreasing the duration of antibiotic use in hospitalized patients, and by encouraging antiviral therapy when appropriate.16 With the advent of rapid antigen testing (RAT) for influenza A, physicians now are able to identify this viral infection that is difficult to diagnose clinically, The utility of this technologic advancement in screening children who have fever and might otherwise undergo a substantial battery of test for evaluation of OB or SBI becomes apparent. However, necessary to understanding the potential for this application is the need to understand the concurrent rate of bacteremia in children with influenza. This study evaluates the prevalence of bacteremia and SBIs in febrile children who were aged 0 to 36 months of age and tested positive for influenza A.

METHODS

Setting

The Texas Children's Hospital (TCH) ED is part of a tertiary care pediatric hospital in Houston, Texas, with ~80 000 ED visits annually. This ED is staffed by attending physicians 24 hours per day. For the evaluation of febrile children who are 0 to 36 months of age and do not have a focus of infection, the acquisition of blood cultures, urine cultures, chest radiographs, and cerebrospinal fluid (CSF) is done at the discretion of the attending physician; however, most physicians follow guidelines set forth in the existing literature.

Patient Selection

We performed a retrospective cross-sectional observational study of children who were 0 to 36 months of age, were evaluated in the TCH ED during 5 consecutive influenza seasons (October through March 1997-2001), presented with a chief complaint of fever, and underwent viral testing and blood cultures. Children were included when they had documented influenza screening by RAT and/or by viral culture (VC) as identified through the diagnostic virology laboratory's computerized database. This database was cross-referenced with the database of children who were younger than 36 months and presented to TCH ED with a chief complaint of fever using an electronic patient tracking system, EmStat (A4 Health Systems). Patients were included in the chart review process when a blood culture also was obtained. Charts were requested for review twice and when unavailable were no longer considered for evaluation.

Patients were excluded when they had any preexisting medical condition that would affect their probability of having a culturable bacterial infection. Exclusion criteria were as follows: antibiotic use within the preceding 48 hours, including prophylaxis for a chronic condition (eg, vesicoureteral reflux); an immunocompromised host, including malignancy, chronic steroid use, human immunodeficiency virus infection, or asplenia; increased risk for infection secondary to indwelling foreign bodies such as central lines, ventriculoperitoneal shunts, or prosthetic valves; and conditions with increased risk for bacteremia (eg, sickle cell disease), UTIs (eg, vesicoureteral reflux, neurogenic bladder), or pneumonia (eg, cystic fibrosis, tracheostomy).

Clinical and Laboratory Assessment

Medical records from eligible patients were reviewed by a single investigator (H.F.S.), and standard forms (Teleform) were used to record the following: demographic information; general examination findings; vital signs; the home use of antipyretics or other medications; and results of any additional diagnostic studies, such as complete blood counts, urinalysis, urine culture, CSF analysis, CSF culture, and/or chest x-rays (CXR). White blood cell counts (WBCs) were categorized as follows: <5000, 5000 to 15 000, and >15 000. Three age groups were created as well: 0 to 28 days, 29 to 89 days (1-3 months), and 90 to 1056 days (3-36 months).

For this study, fever was defined by chief complaint only. Parents, with no strict temperature guidelines, merely reported "fever" as their chief presenting complaint at triage. The word then was used as a search tool in the ED patient tracking system EMSTA\T (A4 Health Systems).

Influenza-positive (IP) cases were considered patients with documented positive Directigen Flu A RAT test or positive VC for influenza A, considered the gold standard; influenza-negative (IN) control subjects were patients whose viral studies were negative for influenza A. Viral studies were obtained by trained respiratory therapists via nasopharyngeal washes and aspirates, the specimens of choice, as they have been shown to be superior to nasopharyngeal and pharyngeal swabs.17 Directigen Flu A antigen detection test (RAT) uses an enzyme immunomembrane filter assay to detect influenza A antigen extracted from suitable specimens from symptomatic patients. Total testing time is <15 minutes with reactivity determined by visual color development, Antigenic drift is not an issue with Directigen Flu A test because the target antigen is the nucleoprotein, which is type specific and highly preserved. The overall sensitivity of the Directigen Flu A Test is referenced at 91% with a specificity of 95%.17

Blood cultures were obtained via standard laboratory protocol. The following organisms were considered pathogens: Streptococcus pneumonias, Hib, Neisseria meningitidis, Salmonella species, Streptococcus pyogenes (group A streptococcus), and Streptococcus agalactiae (group B streptococcus). Other organisms from blood cultures were determined to be pathogens or contaminants on the basis of those accepted in a general textbook of infectious diseases,18 past literature,19 and current accepted clinical practices and by actions taken by the treating physician as ascertained during the chart review process. When the blood culture grew a presumed contaminant such as coagulase-negative staphylococci, micrococci, α-hemolytic streptococci, nonhemolytic streptococci, or diphtheroids,20 the chart was reviewed to confirm the treatment of these organisms as contaminants (ie, chart documented "likely contaminant").

UTIs were defined as growth of a single pathogen of ≥10^sup 4^ colony-forming units on a catheter specimen or a single organism of >10^sup 5^ colony-forming units on a clean-catch specimen. Urine obtained via transurethral bladder catheterization is reported to have a false-positive rate of <2%.21 No suprapubic aspirations were performed, and hag urine collection, performed only for screening purposes, was excluded from analysis. Determining urine culture contaminants was based on the current literature and expert opinion.21,22 Organisms such as Lactobacillus species, coagulase- negative staphylococci, α-hemolytic streptococci, and Corynebacterium species are not considered clinically relevant urine isolates in the otherwise healthy 2-month- to 2-year-old.22 Multiple pathogens cultured from a urine culture were considered contaminated specimens.

Pneumonia was diagnosed when the attending radiologists read "possible,""probable," or "definite" focal parenchymal density on CXR (including the phrase "bronchopneumonia") on routine dictation. The radiologists were generally blinded to the ED physician's interpretation, the patient's clinical status, the patient's influenza status, and the treatment plans. Any SBI was defined as the growth of a bacterial pathogen from blood, urine, or CSF or pneumonia on CXR as defined above.

Statistical Analysis

Statistical analysis was performed using SPSS for Windows. The total sample size of 1035 was anticipated to be achieved over a 5- year period presuming that ~210 patients per year presented with a chief complaint of fever, were tested for influenza, had a blood culture obtained, and whose charts were available for review. Secondary outcome included differences in the prevalence of pneumonia, UTIs, and SBIs (with and without the inclusion of pneumonia). Exploratory outcomes included differences in prevalence of patients with meningitis in the IP and IN groups. Patient characteristics of age, gender, race, and WBCs were described using frequencies with 95% confidence intervals (CIs) and χ^sup 2^ tests. The prevalences of bacteremia, UTI, pneumonia, and meningitis were compared between IN and IP patients using frequencies with 95% CIs and χ^sup 2^ test. Because of small numbers, the Fisher exact test was used to compare blood and urine results between the influenza groups. Because of the absence of meningitis in the IP group, the Yates correction for continuity was used. P < .05 was considered statistically significant.

Prevalence odds ratios (ORs) and 95% CIs were calculated for the association between the presence of influenza infection and bacteremia, UTI, pneumonia, and any SBI. ORs then were stratified by age and WBC categories for any SBI including and then excluding pneumonia. There were insufficient numbers to conduct stratified analysis for bacteremia and UTI. There were also insufficient numbers to stratify by age and WBC categories concurrently. After the stratified analysis, logistic regression was not used to adjust for age and WBC because there was no evidence that they confounded the relationship between influenza and any SBI. In addition, there was no evidence that age confounded the relationship between influenza and pneumonia, whereas there was evidence that the association was modified by WBC categories. The Baylor College of Medicine Institutional Review Board approved this study.

RESULTS

During the study period of 4 consecutive influenza seasons (20 winter months), 21 393 children who were younger than 3 years presented to the TCH ED with fever as part of their chief complaint. Of these, 1029 charts were eligible for review (children who were younger than 36 months, had fever without a source, and subsequently had blood cultures and viral studies performed). A total of 562 were male, and 443 were female. A total of 324 of these patients were excluded: 154 because they had received antibiotic treatment within the previous 48 hours (of note, 17 of these 154 patients on antibiotics incidentally tested positive for influenza); 74 were considered immunocompromised; 26 secondary to indwelling foreign material (eg, ventriculo-peritoneal shunt, central line); 19 because of sickle cell disease; and 51 for "other" reasons, including tracheostomies, chromosomal abnormalities, or other anomalies that would increase the risk for a SBI.

TABLE 1. Patient Characteristics

Of the 705 included patients, 390 (55.3%) were male, 184 (26.1%) were white, 295 (41.8%) were Hispanic, 172 (24.4%) were black, and 55 (7.8%) were categorized as other. There were 62 (8.8%) patients in the 0- to 28-day category, 230 (32.6%) patients in the 28-day to 3-month range, and 413 (58.6%) patients in the 3- to 36-month range. The underlying demographics were similar between the IP and the IN control subjects regarding age distribution (P = .220), gender (P = .454), and race (P = .429; Table 1). In contrast, the IN group had a higher percentage of patients with WBC counts >15 000 (27.8%; 95% CI: 24.0-31.6) compared with the IP patients (18.0%; 95% CI: 12.0- 23.9).

Figure 1 presents the influenza results by RAT and VC. The sensitivity of the Directigen RAT in this study was 87% (95% CI: 81%- 92%), the specificity was 99% (95% CI: 98%-99%), the positive predictive value was 95% (95% CI: 93%-99%), and the negative predictive value was 96% (95% CI: 94%-98%).

Bacteremia

All 705 patients included had blood cultures drawn. The prevalence of bacteremia was lower (P = .024) in the IP group (0.6%; 95%: CI 0.0%-1.8%; Table 2) compared with the IN group. As shown in Table 3, the odds of bacteremia were 86% less in those in the IP group than in the IN group (OR: 0.14; 95% CI: 0.02-1.04). There were insufficient numbers to stratify by age and WBC count in the association between influenza and bacteremia. In the IN group, the pathogens that led to positive blood cultures consisted of 18 S pneumonias, 2 Escherichia coli, and 3 pathogens classified as "other" (Table 4).

The 1 patient with influenza A and bacteremia was a 3-month-old Latin American male, previously healthy, term infant who presented with the chief complaint of not feeding well and increased sleepiness. There was a 2-day history of yellow/green eye discharge, 2 bouts of emesis, and 2 loose stools. There were multiple school- aged ill contacts at home. On presentation, the child was ill- appearing and fussy but consolable and had a temperature of 40.9C (105.6F), a heart rate of 184 beats per minute, respirations of 46 per minute, a blood pressure of 81/50 mm Hg, and oxygen saturations of 99% on room air. Physical examination was significant for conjunctivitis and erythematous tympanic membranes with decreased mobility. A normal saline bolus (20 mL/kg) was given for a capillary refill of 3 seconds. This child's WBC count was 9.36 (10^sup 3^/UL) with 57% polymorphonuclear cells, 7% bands, 26% lymphocytes, and 10% mononuclear cells. The CSF contained 1 WBC (mononuclear cell), no red blood cells, a protein of 16 and a glucose of 59, and no organisms seen on Gram stain. The child was acidotic with a bicarbonate of 16 (mmol/L), sodium of 139 (mmol/ L), potassium of 4.2 (mmol/L), chloride of 107 (mmol/L), blood urea nitrogen of 4 (mg/ dL), and a creatinine of 0.3 (mg/dL). RAT and eventually VC were positive. The child was admitted on empiric cefotaxime. The blood culture became positive within 48 hours with Gram-positive cocci in pairs and chains that were eventually identified as group A β- hemolytic streptococcus. The child was discharged from the hospital on hospital day 5 in good condition.

Fig 1. Influenza A RAT and VC results.

TABLE 3. Association Between Influenza A and SBIs

TABLE 2. Influenza A and SBIs

UTIs

A total of 452 patients had urine cultures performed. The prevalence of UTI was lower in the IP than in the IN group (P = .005; Table 2). Two (1.8%) of 110 IP cases that had urine cultures done had UTIs (95% CI: 0.7%-4.3%) versus 38 (9.9%) of the 382 IN control subjects (9\1% CI: 6.9%-13.1%). As shown in Table 3, the odds of UTI were 83% less in those in the IP group than in the IN group (OR: 0.17; 95% CI: 0.04-0.71). There were insufficient numbers to stratify by age and WBC count in the association between influenza and UTI. Of the IN patients who had a diagnosis of UTI, 25 had E coli, 5 had Enterococcus species, 1 had Klebsiella species, and 7 were classified as "other" (Table 4).

TABLE 4. Culture Results

Of the 2 IP patients with UTIs, 1 was a 3-month-old healthy girl with cough, congestion, fever to 40C (104F), and decreased urine output. On physical examination, she appeared well, with no focus for her fever. Her WBC count was 10.3 (10^sup 3^/UL) with 35% polymorphonuclear cells, 4% bands, 49% lymphocytes, and 10% mononuclear cells. The urinalysis was unremarkable, and her RAT and VC were positive for influenza A. The urine culture was positive within 48 hours for 10^sup 5^ E coli, and the patient was called back to the ED. She was doing well clinically with no additional fever and was treated for a UTI. Renal ultrasound and voiding cystourethrogram subsequently were negative.

The second infant, an 8-month-old boy, presented similarly with 1 day of cough, fever, loose stools, and "irritability" at night. The initial temperature was 40.3C (104.6F), with a heart rate of 160 beats per minute, 32 respirations per minute, and a blood pressure of 108/70 mm Hg. On physical examination, he was noted to be well nourished, in no apparent distress, and "very happy." The WBC count was 8.6 (10^sup 3^/UL) with 39% polymorphonuclear cells, 7% bands, 43% lymphocytes, and 11% mononuclear cells. The urinalysis, again, was unremarkable, and his RAT and VC were positive for influenza A. He was discharged from the hospital with acetaminophen for fever. His urine culture grew 10^sup 5^ E coli.

Pneumonia

A total of 287 patients had CXRs performed. Pneumonia was diagnosed less frequently in the IP than in the IN patients (P = .029; Table 2). Thirteen (25.5%) of 51 of the IP patients had radiographic evidence of pneumonia (95% CI: 13.5%-37.5%) versus 99 (41.9%) of 236 IN control subjects (95% CI: 35.6%-48.2%). No ORs were calculated in the association between influenza A and pneumonia.

One death in the IP group was attributed to pneumonia. A 4-month- old white boy with no known medical history presented to the ED with 48 hours of increasing respiratory rate, cough, difficulty feeding, and diarrhea. Vital signs at presentation were temperature 38.6C (101.4F), pulse 187, respirations 70 to 80, and blood pressure 105/ 77 mm Hg. Lung auscultation revealed moderate aeration of the lungs with coarse breath sounds bilaterally. Cardiac examination was significant for the following: a right ventricular tap, a palpable P2, an accentuated S2, an S3 gallop, a grade 2/6, long low-pitched systolic ejection murmur along the lower sternal border without radiation, and a 2/6 diastolic rumble at the left lower sternal border. The liver was 5 cm below the midright costal margin. The patient was intubated secondary to progressive respiratory distress and shock. An echocardiogram revealed total anomalous pulmonary venous return of mixed type with a restrictive atrial septal defect. The patient died after 10 days of aggressive medical management. The RAT was positive for influenza A. Influenza A and adenovirus were cultured from the lungs at autopsy. Incidentally, Pseudomonas aeruginosa also was cultured from the lungs (after 10 days of intubation), but blood cultures remained negative. The final autopsy report sited interstitial pneumonia, influenza A, cardiorespiratory failure, disseminated intravascular coagulation, and renal failure as the causes of death.

Meningitis

A total of 215 patients had spinal fluid cultured. Of 40 CSF samples obtained from IP patients, there were no cases of meningitis compared with 4 cases of culture-positive meningitis in the 179 (2%) CSF samples obtained from IN control subjects (Tables 2 and 4). Using a Yates correction for continuity, there was no evidence for a significant difference in the prevalence of meningitis between the IN and IP groups (P = .335).

SBIs

There were significantly fewer SBIs in the IP cases compared with the IN control subjects (P = .000; Table 2). Sixteen (9.8%) of 163 IP patients had SBIs (95% CI: 5.2-14.4) versus 153 (28.2%) of 542 IN control subjects (95% CI: 24.4-32.0). Of note, none of the 16 IP patients had >1 SBI; however, 5 IN patients had >1 SBI (E coli in the urine and CSF). Also of note, the prevalence of SBI in the IN group seemed elevated at 28%, but when pneumonia was excluded from the definition of SBI, the prevalence dropped to 11.6% (95% CI: 8.9- 14.2; Table 2), a number more consistent with previously published figures. The prevalence of SBI in the IP group, excluding pneumonia, dropped to 3 (1.8%; 95% CI: 2.4-3.8; P = .00).

As shown in Table 3, the odds of any SBI (including pneumonia) were 72% less in the IP group than in the IN group (OR: 0.28; 95% CI: 0.16-0.48). There were insufficient numbers in the youngest age group to examine that group, but the 2 older age categories had similar ORs to each other. The ORs were also similar across WBC categories. Given the high prevalence of pneumonias in the IN and IP groups and the authors concerns over the vague and broad definition of pneumonia in this retrospective chart review, the SBI category was subdivided into any SBI including pneumonia (as discussed above) and SBI excluding pneumonia considering only bacteremia, UTI, and meningitis. Therefore, the odds of any SBI (excluding pneumonia) in the IN group were 86% less in than those in the IP group (OR: 0.14; 95% CI: 0.04-0.46). Again, there were insufficient numbers in the youngest age group (0-28 days) and lowest WBC categories (<5000) to stratify. However, in the older age categories, 29 to 90 days and 91 to 1056 days, the ORs were very similar, 0.19 and 0.17, respectively; however, the CIs are wide and even cross 1 in the former age group. Likewise, in the WBC category of 5000 to 14999 and ≥15 000, the ORs were very similar, 0.18 and 0.15, respectively. Again, however, the CIs are wide and crossed 1 in the latter WBC category.

DISCUSSION

Influenza Assays and VCs

This study is based on the premise that influenza A, although not a clinically recognizable viral syndrome, is now a viral syndrome that can be diagnosed via a rapid test. However, for influenza screening to be a practical tool for evaluating infant and children with fever in the ED, the results must be accurate and available in a timely manner. Although the gold standard is the VC, the RAT results in our institution are available within 1 to 2 hours, thereby affecting patient care and medical decision making.

Although the sensitivity in this study is less than that referenced in the packet insert,17 the physicians should be more concerned about false-positive results that may incorrectly reassure against a bacterial cause or lead them away from performing an SBI evaluation. Many factors affect this rate of positivity: the method of collection, handling/transport systems, detection method used, age of the patient, geographic location, the time of year, and, most important, local disease prevalence all may have had an impact on the results reported. This study period, typical influenza epidemic months (October through March), was chosen specifically when disease prevalence would be high and the risks associated with false- positive influenza results would be lower.

Bacteremia

Only 1 IP patient was bacteremic, and he did well overall. On presentation, this infant was clearly ill-appearing, required intravenous fluid resuscitation and empiric antibiotic therapy, and would not have been considered "low risk" by any criteria.23-25 This idea of general clinical appearance is pivotal in clinical decision making as the increased risk for bacteremia in toxic-appearing infants is well documented.26 The vast majority of physicians would embark on a sepsis evaluation in any ill-appearing infant regardless of influenza status.

However, this study is geared toward the well-appearing febrile patient who is influenza positive yet for whom the physician may feel in a quandary regarding the necessity of evaluation for a bacterial cause. Unfortunately, we were unable to define reliably "well-appearing" or "toxic" in a retrospective manner given the variety of terminology chosen by our faculty. Including all patients, regardless of clinical appearance, represents a worst- case scenario and likely inflates the OR (conversely suggesting a decreased OR in the well-appearing subpopulation). Although nothing in the enrollment process biased either group to ill-appearing children, the retrospective design does not allow for this distinction. Future prospective studies will need to standardize the terminology to capture the appropriate population.

National background prevalence data of bacteremia in febrile children have been reported to be in the 1.2% to 2% range.6,7 Since the early 1990s, vaccination against Hib has had a major impact in the epidemiology of bacteremia in febrile children, virtually eliminating Hib as a bacterial pathogen in febrile children.7,27,28 In addition, with the introduction of the heptavalent conjugate pneumococcal vaccine (PCV7) in June 2000,29 there has been a significant decline in invasive pneumococcal disease.30 There was a 50% decline in the number of cases of invasive pneumococcal disease in 2001 and an 82% decline in 2002 (compared with rates between 1994 and 2000) in children who were younger than 2 years.30 When compared with the mean of the years 1994-2000, the annual number of invasive pneumococcal infections for children who were younger than 24 months declined 58% in 2001 and 66% in 2002 (vaccine-serogroup isolates) in 8 US children's hospitals (including TCH). When considering the bacterial epidemiology, the time frame \of this study captures a population of patients that straddles the introduction of the PCV7 vaccine, thus providing 2 potentially different populations. Table 4 shows that 78% of positive blood cultures grew S pneumoniae. Although there were not enough patients in this study to make statistical comparisons between preand post-PCV7 populations, assuming this decrease in prevalence of invasive pneumococcal disease, the ORs presented here would be conservative and the rate of SBI from pneumococcus in IP patients could be postulated to be overstated in those who have been immunized adequately.

Unfortunately, immunization status, like clinical appearance, is another piece of information that is difficult to extract reliably and accurately from a retrospective chart review. Even if it were appropriately documented that immunization was up-to-date, most charts did not indicate whether this information was verified or based solely on parents' recall, which may be inaccurate. Studies that are done in a fully immunized population may yield even lower odds of bacteremia, assuming that no other disease entity takes over the pathogenic role of pneumococcus. Although beyond the scope of this retrospective study, immunization status is an important issue that needs to be addressed in the future to clarify and to capture this crucial post-PCV7 population.

Unknown is the implications that this decline in the frequency of bacteremia after the introduction of the Hib and PCV7 vaccines may have on the strategies for evaluating fever without localizing signs or the guidelines concerning the evaluation of children who may be "low risk" for bacteremia. Given this changing bacterial epidemiology, some are recommending a change in management strategy of well-appearing febrile young children,31 suggesting that a complete blood count and blood culture are not necessary in children who are older than 6 months and have received at least 3 doses of PCV7 vaccine. The use of influenza A screening may encourage the further development of a recommendation to place well-appearing influenza-positive patients into a low-risk population.

Many investigators have reported that children are more susceptible to secondary bacterial infection while concurrently infected with influenza A virus.18,33 Indeed, viral-bacterial co- infections or bacterial superinfections secondary to viral infections are well documented.34 Suggested reasons for superinfection include virus-induced immunosuppression, virus destruction of respiratory epithelium increasing bacterial adhesion, and viral infection upregulating expression of molecule that bacteria use as receptors.35 For example, influenza and parainfluenza contain neuroaminidase activity, which seems to increase bacterial adherence after viral preincubation.35 Okamoto et al,36 using a mouse model, showed that previous infection with influenza A virus resulted in hemagglutinin expression in the surface of alveolar epithelial cells and promoted internalization of S pyogenes, inducing a lethal synergism resulting in >90% of their dying. A synergism between influenza virus and S pneumoniae also may be present, whereby the neuroaminidase in influenza actually may be a primer for pneumococcal infections.37 We found no published evidence on S pyogenes, the actual pathogen cultured from the 1 IP bacteremic patient, demonstrating similar synergistic relationships with influenza.

UTIs

The prevalence of UTI in febrile infants and young children who are 2 months to 2 years of age and have no apparent source is high (~5%).21 The clinical presentation of UTI usually is nonspecific (often limited only to fever), and the diagnosis frequently is challenging as a result of the difficulty in obtaining valid urine specimens. In Shaw and Gorelick's article,22 it was believed that it often is difficult to determine which children have UTIs even in febrile children with an equivocal source of fever, such as upper respiratory infection or otitis media. However, in the same article, they reported that "infants with unequivocal sources for fever on examination, such as pneumonia, meningitis, varicella infection, or bronchiolitis, have the lowest prevalence of UTI (<2%)."22 In this study, only 2 children had influenza A and simultaneous UTI. Neither was ill-appearing, and both had unremarkable urinalyses. In these patients, there is uncertainty whether a UTI was present or these were cases of asymptomatic bacteriuria with influenza A infection

Pneumonia

There was a high prevalence of pneumonia in the IN control group. Although some may argue that this can be explained by the high acuity of patients seen (TCH ED is a tertiary referral center, and a great number of the patients who are seen in the ED have already been seen by their primary doctor), the authors believe that the problem lay in the actual diagnosis of pneumonia. Viral pneumonia or pneumonitis is common, especially with influenza A infections, but it is very different from a bacterial pneumonia. In a retrospective chart review, it was very difficult to tease out these differences. For this study, when wording such as "cannot exclude pneumonia" or "bronchopneumonia" were used in the dictations of the attending radiologist, the patient was counted as having pneumonia, thereby creating a relatively high prevalence rate of pneumonia. If the radiologists were truly blinded to the patient's influenza status, then the "overreading" phenomenon should have occurred equally in both the IN and IP cohorts, and one could conclude that there were no significantly greater numbers of pneumonia diagnosed in the IN control group as compared with the IP cases versus 99 of 236 IN control subjects.

Because of the high prevalence of pneumonia using the a priori definition of pneumonia, a post hoc sensitivity analysis was performed after pneumonias were recategorized using a more stringent definition. Equivocal pneumonia, "bronchopneumonia," readings of atelectasis versus infiltrate, and "cannot rule out" were considered negative. The prevalence of pneumonia in the IN group dropped to 17.0% (95% CI: 12.2%-21.8%) from 41.9% (95% CI: 35.6%-48.2%). The prevalence of pneumonia in the IP group dropped from 25.5% (95% CI: 13.5%-37.5%) with the a priori all-inclusive definition to 9.8% (95% CI: 1.6%-18.0%) using the strict (consolidation, focal infiltrate) definition. Although these numbers are more realistic, the rate of pneumonia in the IP group remains substantial and CXRs are still prudent in IP patients when clinically indicated.

Indeed, the clinician who treated the patient was not blinded to the influenza status; however, the gold standard for determining the diagnosis of pneumonia was the radiologist's interpretations. Although we cannot guarantee that the radiologist was blinded to the patient's influenza status, the CXR requests were reviewed and there was no indication that influenza status was conveyed to diagnostic imaging via these orders. Furthermore, the radiologist's office is in a remote area from the ED, and the radiologists have limited interactions with the clinicians and routinely are unaware of clinical findings or laboratory results.

The 1 death in the IP cohort attributed to pneumonia was in a patient who, similar to the IP bacteremic patient, was very ill- appearing on presentation. An ill-appearing patient generally will be evaluated and treated aggressively by physicians regardless of influenza status, and this patient's previously undiagnosed cardiac condition likely contributed to the decompensation.

Meningitis

There were no cases of bacterial meningitis in the IP group. The pathogens for the 4 cases of meningitis in the IN group (Table 4) consisted of E coli (n = 1), group B Streptococcus (n = 1), and S pneumoniae (n = 2).

SBIs

The SBI data were interpreted including pneumonia in the definition of SBI and then with the pneumonia data excluded because of the concern regarding the broad definition of pneumonia and the seemingly elevated numbers of pneumonia. Regardless, the odds of having an SBI are lower when 1 is IP. Furthermore, there were no differences in odds across the age or WBC categories.

Study Limitations

The study population from 1 tertiary care ED may affect the generalizability of the study findings. Also, given the timing of data collection, these data are generalizable only to the winter flu epidemic months (October through March), when influenza A prevalence is high. Furthermore, the study spanned only a 4-year period, but influenza epidemics are on an ~7-year cycle. Anecdotal evidence suggests that the winter of 2003-2004, which was not captured in this study, was a particularly hard-hitting influenza season with a circulating strain of influenza known for its disease severity.38 Future studies should consider capturing these epidemic years incorporating influenza subtyping.

The treating physician, unblinded to the influenza status, may have halted additional laboratory evaluation for SBI and have biased the results away from the null. As the reviewer was not blinded to the influenza status or the study question, it is possible that those who were IP were less likely to be identified with an SBI than those who were IN. This was particularly problematic with the assignment of contaminant status. However, the use of a priori contamination definition limited the possibility of this type of bias.

Some of these patients may have been misclassified according to presenting symptoms. By using EmStat to identify patients who presented with a chief complaint of "fever," itself a vague and interpretable term with no strict definition in this study, we missed an entire population of patients who may have presented with other complaints that lend themselves to a diagnosis of influenza and/or SBI (eg, chief complaints of "lethargic,""irritable,""respiratory distress").

The a priori analysis plan did not allocate α and β value\s among the various age strata or WBC categories; thus, the strata-specific results must be interpreted with caution to avoid any confusion about the use of the term "risk." This study lacks an economic analysis, which should be embedded in future studies. If the cost-benefit analysis proves favorable, then more RAT kits may be developed to aid in diagnosing these nonclinically recognizable viruses to limit full sepsis evaluations in well-appearing febrile infants.

Influenza Immunizations

Influenza immunization status was not taken into account. A vaccine targeted at children, the major reservoir for influenza infection, should have a major impact on epidemics and preventing the serious consequences of influenza.39 "Current inactivated influenza vaccines have shown efficacy and effectiveness in preventing influenza like illness, hospitalization for pneumonia, and death and in reducing health care costs."40 A live attenuated intranasal influenza virus for children was approved recently in the United States, and in a study that involved children who were 1 to 5 years of age, the vaccination was well tolerated and conferred protection against influenza in ~90% of recipients.41 The 2002 Morbidity and Mortality Weekly Report states that the "influenza vaccine is recommended annually for children greater than and equal to 6 months with certain risk factors . . . and to all others wishing to obtain immunity."42 Recently, the American Academy of Pediatrics extended these guidelines in a policy statement recommending vaccination of "healthy children age 6 to 24 months, for household contacts and out-of-home caregivers of all children younger than 24 months of age, and for health care professionals."43 Two doses are recommended for children who are younger than 9 years unless they have been vaccinated previously.44

CONCLUSION

Influenza is a common cause of morbidity, mortality, and evaluation of febrile children in the ED. Influenza RATs may be a cost-effective tool for screening febrile infants. Although not clinically recognizable, this study suggests that influenza is identifiable via a test and could be treated as a recognizable viral syndrome. Just as recognizable viral syndromes in febrile children obviate the need for fever workups,8 soon testable viral syndromes such as influenza A also may eliminate the evaluations of infants with fever. This study indicates that bacteremia, UTIs, pneumonia, and SBIs occur less frequently in infants with concurrent influenza A infection compared with influenza A-negative control subjects. With great caution, in well-appearing, vaccinated febrile children, blood cultures may no longer be routine. However, prospective studies to address clinical appearance, strict definitions of pneumonia (bacterial vs viral "bronchopneumonia"), and immunization status (number of PCV7 doses and influenza vaccination) are needed to clarify these points. An economic evaluation should be embedded in future studies as well. In the meantime, ED physicians need to heed the importance of influenza as a fever source and work on ED isolation techniques, encourage vaccination, and use antiviral medication for treatment and prophylaxis of influenza.

ACKNOWLEDGMENTS

This study was funded in part by an unrestricted grant from ZymeTx (ZstatFlu).

We thank Jennifer Jones for technical support, Wendy Smitherman for assistance in data collection, and Dr Erin Endom for critical review of an earlier version of the manuscript.

REFERENCES

1. American Academy of Pediatrics. Report of the Committee on Infectious Diseases. Redbook. Elk Grove Village, IL: American Academy of Pediatrics; 1997:309-315

2. Arias E, Smith BL. Deaths: data for 2001. Natl Vital Stat Rep. 2003;51: 1-44

3. Centers for Disease Control and Prevention. Influenza Summary Update (week ending February 24, 2001-week 08). Report prepared March 1, 2001. Available at: www.cdc.gov/nicidod/diseases/flu/ weeklyarchives/ week08.htm

4. Centers for Disease Control and Prevention. Clinical Features of Influenza. Available at: www.cdc.gov/ncidad/diseases/flu/ fluinfo.htm

5. Nelson DS, Walsh K, Fleisher GR. Spectrum and frequency of pediatric illness presenting to a general emergency department. Pediatr Emerg Care. 1991;7:67-71

6. Greenes DS, Harper MB. Low risk of bacteremia in febrile children with recognizable viral syndromes. Pediatr Infect Dis J. 1999;18:258-261

7. Liebelt EL. Diagnostic testing for serious bacterial infections in infants aged 90 days or younger with bronchiolitis. Arch Pediatr Adolesc Med. 1999;153:525-530

8. Antonow JA. Use of respiratory syncytial virus testing could safely eliminate many sepsis evaluations. Arch Pediatr Adolesc Med. 1999;153: 1310-1311

9. Kupperman N, Bank DE, Walton EA, Senac MO Jr, McCaslin I. Risks for bacteremia and urinary tract infections in young febrile children with bronchiolitis. Arch Pediatr Adolesc Med. 1997;151:1207- 1214

10. Melendez E, Harper MB. Utility of sepsis evaluation in infants 90 days of age or younger with fever and clinical bronchiolitis. Pediatr Infect Dis J. 2003;22:1053-1056

11. Lee GM, Harper MB. Risk of bacteremia for febrile children in the post-Haemophilus influenzae type B era. Arch Pediatr Adolesc Med. 1998; 152:624-628

12. Kuppermann N, Fleisher GR, Jaffe DM. Predictors of occult bacteremia in young febrile children. Ann Emerg Med. 1998;31:679- 687

13. Verbon-Maciolek MA, Nijhuis M, van Loon AM, et al. Diagnosis of enterovirus infection in the first 2 months of life by real-time polymerase chain reaction. Clin Infect Dis. 2003;37:1-6

14. Levine DA, Platt S, Dayan PS, et al. Risks of serious bacterial infection in young febrile infants with respiratory syncytial virus infections. Pediatrics. 2004;113:1728-1734

15. Serwint JR, Miller RM. Why diagnose influenza in hospitalized pediatric patients? Pediatr Infect Dis J. 1993;12:200-204

16. Noyola DE, Demmler GJ. Effect of rapid diagnosis on management of Influenza A infections. Pediatr Infect Dis J. 2000;19:303-307

17. Directigen Flu A [package insert]. Sparks, MD: Becton Dickinson; 2002

18. Feigin RD, Cherry JD, Fletcher J. Textbook of Pediatric Infections Diseases. Philadelphia, PA: WB Saunders; 1998

19. Segal GS, Chamberlain JM. Resource utilization and contaminated blood cultures in children at risk for occult bacteremia. Arch Pediatr Adolesc Med. 2000;154:469-473

20. Kornberg AE, Jain N, Dannenhoffer R. Evaluation of false- positive blood cultures: guidelines for early detection of contaminated cultures in febrile children. Pediatr Emerg Care. 1994;10:20-22

21. Practice parameter: the diagnosis, treatment, and evaluation of the initial urinary tract infection in febrile infants and young children. American Academy of Pediatrics. Committee on Quality Improvement. Subcommittee on Urinary Tract Infection. Pediatrics. 103:843-852

22. Shaw KN, Gorelick MH. Urinary tract infection in the pediatric patient. Pediatr Clin North Am. 1999;46:1111-1124

23. Dagan R, Powel KR, Hall CB. Identification of infants unlikely to have SBI although hospitalized for suspected sepsis. J Pediatr. 1985;107: 855-860

24. Alpern WR, Alessandrini EA, Bell LM, Shaw KN, Mcgown KL. Occult bacteremia from a pediatric emergency department: current prevalence, time to detection and outcome. Pediatrics. 2000;106:505- 511

25. Bachur G, Harper MB. Predictive model for serious bacterial infections among infants younger than 3 months of age. Pediatrics. 2001;108: 311-316

26. Bonadio WA. The history and physical assessments of the febrile infant. Pedatr Clin North Am. 1998;45:65-77

27. Haemophilus influenzae invasive disease among children 2-5 years-California 1990-1996. MMWR Morb Mortal Wkly Rep. 1998;47:737- 740

28. Adams WG, Deaver KA, Cochi SL, et al. Decline in childhood Haemophilus influenza type B (Hib) disease in the Hib vaccine era. JAMA. 1993; 269:221-226

29. American Academy of Pediatrics, Committee of Infectious Disease. Recommendations for the prevention of pneumococcal infections, including the use of pneumococcal conjugate vaccine (Prevnar), pneumococcal polysaccharide vaccine, and antibody prophylaxis. Pediatrics. 2000;106:363-366

30. Kaplan SL, Mason EO, Wald ER, et al. Decrease of invasive pneumococcal infections in children among 8 children's hospitals in the united states after introduction of the 7-valent pneumococcal conjugate vaccine. Pediatrics. 2004;113:443-449

31. Nigrovic LE, Malley R. Evaluation of the febrile child 3-36 months old in the era of pneumococcal conjugate vaccine: focus on occult bacteremia. Clin Pediatr Emerg Med. 5:13-19

32. Korppi M, Leinonen M, Koskela M, Makela PH, Launiala K. Bacterial co-infections in children hospitalized with respiratory syncytial virus infections. Pediatr Infect Dis J. 1989;8:687-692

33. Hament JM, Kimpen JL, Fleer A, Wolfs TF. Respiratory viral infection predisposing for bacterial disease: a concise review. FEMS Immunol Med Microbiol 1990;26:189-195

34. Peltola VT, McCullers JA. Respiratory viruses predisposing to bacterial infections: role of neuraminidase. Pediatr Infect Dis J. 2004;23(suppl): S87-S97

35. Okamoto S, Kawabala S, Nakagawa I, et al. Influenza A virus- infected hosts boost an invasive type of Streptococcus pyogenes infection in mice. J Virol. 2003;77:4104-4112

36. McCullers JA, Bartmess KC. Role of neuroaminidase in lethal synergism between Influenza viruses and Streptococcus pneumoniae. J Infect Dis. 2003;187:1000-1009

37. Meadows M. A look at the 2003-2004 flu season. FDA Consum. 2001;38: 9-11

38. Couch R. Prevention and treatment of Influenza. N Engl J Med. 2000; 343:1778-1787

39. Stamboulian D, Bonvehi PE, Nacinovich FM, Cox N. Influenza. Infect Dis Clin North Am. 2000;14:141-166

40. Heikkinen T, Ruuskanen O, Waris M, Ziegler T, Arola M, Halonen P. Influenza vaccination in the prevention of acute otitis media in children. Am J Dis Child. 1991;145:445-448

41. Update: influenza activity-United States, 2002-2003 season. MMWR Morb Mortal Wkly Rep. 2002;51:1095-109\1

42. American Academy of Pediatrics Committee on Infectious Diseases. Recommendations for Influenza immunization of children. Pediatrics. 2004;113:1441-1447

43. American Academy of Pediatrics, Committee on Infectious Diseases. Recommended childhood and adolescent immunization schedule- United States, January-June 2004. Pediatrics. 2004;113:142-147

Hannah F. Smitherman, MD; A. Chantal Caviness, MD, MPH; and Charles G. Macias, MD, MPH

From the Department of Pediatrics, Section of Emergency Medicine, Baylor College of Medicine, Houston, Texas.

Accepted for publication Sep 1, 2004.

doi:10.1542/peds.2004-1112

No conflict of interest declared.

Reprint requests to (H.S.) Department of Pediatrics, Section of Emergency Medicine, Baylor College of Medicine, 6621 Fannin St, MC 1- 1481, Houston, TX 77030. E-mail: hfsmithe@texaschildrenshospital.org

PEDIATRICS (ISSN 0031 4005). Copyright 2005 by the American Academy of Pediatrics.

Copyright American Academy of Pediatrics Mar 2005

Source: Pediatrics

More News in this Category