Predictive model for serious bacterial infections among infants younger than 3 months of age.

OBJECTIVE: To develop a data-derived model for predicting serious bacterial infection (SBI) among febrile infants <3 months old. METHODS: All infants /=38.0 degrees C seen in an urban emergency department (ED) were retrospectively identified. SBI was defined as a positive culture of urine, blood, or cerebrospinal fluid. Tree-structured analysis via recursive partitioning was used to develop the model. SBI or No-SBI was the dichotomous outcome variable, and age, temperature, urinalysis (UA), white blood cell (WBC) count, absolute neutrophil count, and cerebrospinal fluid WBC were entered as potential predictors. The model was tested by V-fold cross-validation. RESULTS: Of 5279 febrile infants studied, SBI was diagnosed in 373 patients (7%): 316 urinary tract infections (UTIs), 17 meningitis, and 59 bacteremia (8 with meningitis, 11 with UTIs). The model sequentially used 4 clinical parameters to define high-risk patients: positive UA, WBC count >/=20 000/mm(3) or /=39.6 degrees C, and age <13 days. The sensitivity of the model for SBI is 82% (95% confidence interval [CI]: 78%-86%) and the negative predictive value is 98.3% (95% CI: 97.8%-98.7%). The negative predictive value for bacteremia or meningitis is 99.6% (95% CI: 99.4%-99.8%). The relative risk between high- and low-risk groups is 12.1 (95% CI: 9.3-15.6). Sixty-six SBI patients (18%) were misclassified into the lower risk group: 51 UTIs, 14 with bacteremia, and 1 with meningitis. CONCLUSIONS: Decision-tree analysis using common clinical variables can reasonably predict febrile infants at high-risk for SBI. Sequential use of UA, WBC count, temperature, and age can identify infants who are at high risk of SBI with a relative risk of 12.1 compared with lower-risk infants.

Reliability of the urinalysis for predicting urinary tract infections in young febrile children.

BACKGROUND: Urinary tract infections (UTIs) are a common source of bacterial infection among young febrile children. Clinical variables affecting the sensitivity of the urinalysis (UA) as a screen for UTI have not been previously investigated. The limited sensitivity of the UA for detecting a UTI requires that a urine culture be obtained in some children regardless of the UA result; however, a proper urine culture requires an invasive procedure, so the criteria for its use should be optimized. OBJECTIVES: To determine how the sensitivity of the standard UA as a screening test for UTI varies with age, and to determine the clinical situation that necessitates the collection of a urine culture regardless of the UA result. METHODS: Retrospective medical record review of patients younger than 2 years with fever (>/=38 degrees C) seen in the emergency department during a period of 65 months. All urine cultures were reviewed for the collection method, isolates, and colony counts. A UA result was considered positive if the presence of 1 of the following was detected: leukocyte esterase, nitrite, or pyuria (>/=5 white blood cells per high power field). Patients who had a paired UA and urine culture were used to calculate the sensitivity, specificity, and likelihood ratios of the UA. The prevalence of UTIs was also subcategorized by age, race, sex, and fever. RESULTS: Medical records of 37 450 febrile children younger than 2 years were reviewed. Forty-four percent were girls. Median age and temperature were 10.6 months and 38.8 degrees C. A total of 11 089 patients (30%) had urine cultures obtained. The sensitivity of the UA was 82% (95% confidence interval [CI], 79%-84%) and did not vary by age subgroups. The specificity of UA was 92% (95% CI, 91%-92%). The likelihood ratios for a positive UA and negative UA were 10.6 (95% CI, 10.0-11.2) and 0.19 (95% CI, 0.18-0.20), respectively. Prevalence of UTI was 2.1% overall (2.9% for girls and 1.5% for boys, respectively). Among girls, the prevalence of UTI was 5.0% in white patients, 2.1% in Hispanic patients, and 1.0% in black patients. Among boys, the prevalence was 2.2% in Hispanic patients, 1.4% in white patients, and 0.8% in black patients. Higher prevalence was also seen among patients with a temperature at or above 39 degrees C compared with those whose temperature was between 38.0 degrees C and 38.9 degrees C. The greatest prevalence of UTI (13%) was found among white girls younger than 6 months with a temperature at or greater than 39 degrees C. The posttest probability of a UTI in the presence of a negative UA can be calculated using the negative likelihood ratio and the patient-specific prevalence of UTI. When the prevalence of UTI is 2%, 1 UA among 250 will produce a false-negative test result. CONCLUSIONS: The sensitivity of the standard UA is 82% (95% CI, 79%-84%) and does not vary with age in febrile children younger than 2 years. The prevalence of UTI varies by age, race, sex, and temperature. A negative likelihood ratio and estimates of prevalence can be used to calculate the risk of missing a UTI due to a false-negative UA result.

Nonresponders: prolonged fever among infants with urinary tract infections.

BACKGROUND: The majority of young children with fever and urinary tract infections (UTIs) have evidence of pyelonephritis based on renal scans. Resolution of fever during treatment is 1 clinical marker of adequate treatment. Theoretically, prolonged fever may be a clue to complications, such as urinary obstruction or renal abscess. OBJECTIVE: Describe the pattern of fever in febrile children undergoing treatment of a UTI. Compare the clinical characteristics of those patients with prolonged fever to those who respond faster to therapy. SETTING: An urban pediatric hospital. DESIGN: Medical record review. METHODS: All children /=38 degrees C and those who met standard culture criteria were studied. Temperatures are not recorded hourly on the inpatient unit; therefore, they were assigned to blocks of time. Nonresponders were defined as those above the 90th percentile for the time to defervesce. Nonresponders were then compared with the balance of the study patients, termed responders. RESULTS: Of 288 patients studied, the median age was 5.6 months (interquartile range: 1.3-7.9 months old). Median admission temperature was 39.3 degrees C (interquartile range: 38.5 degrees C-40.1 degrees C). Median time to defervesce ranged in the time block 13 to 16 hours. Sixty-eight percent were afebrile by 24 hours and 89% by 48 hours. Thirty-one patients had fever >48 hours (nonresponders). Nonresponders were older than responders (9.4 vs 4.1 months old) but had similar initial temperatures (39.8 vs 39.2 degrees C), white blood cell counts (18.4 vs 17.1 x 1000/mm(3)), and band counts (1.4 vs 1.2 x 1000/mm(3)). Nonresponders had similar urinalyses with regard to leukocyte esterase positive (23/29 vs 211/246), nitrite-positive (8/28 vs 88/221], and the number of patients with "too numerous to count" white blood cell counts per high power field (12/28 vs 77/220). Nonresponders were as likely as responders to have bacteremia (3/31 vs 21/256), hydronephrosis by renal ultrasound (1/31 vs 12/232), and significant vesicoureteral reflux (more than or equal to grade 3; 5/26 vs 30/219). Eschericia coli was the pathogen in cultures of 28 of 31 (nonresponders) and 225 of 257 (responders) cultures. The number of cultures with >/=100 colony-forming units/mL was similar (25/31 nonresponders vs 206/257 responders). Repeat urine cultures were performed in 93% of patients during the admission; all culture results were negative. No renal abscesses or pyo-hydronephrosis was diagnosed. CONCLUSIONS: Eighty-nine percent of young children with febrile UTIs were afebrile within 48 hours of initiating parenteral antibiotics. The patients who took longer than 48 hours to defervesce were clinically similar to those whose fevers responded faster to therapy. If antibiotic sensitivities are known, additional diagnostic studies or prolonged hospitalizations may not be justified solely based on persistent fever beyond 48 hours of therapy.

Reevaluation of outpatients with Streptococcus pneumoniae bacteremia.

BACKGROUND: The reevaluation process for outpatients recalled for Streptococcus pneumoniae bacteremia has not been standardized. Children who return ill or with new serious focal infections require admission and parenteral antibiotic therapy. Limited data exist to guide the follow-up management of those patients identified as having occult pneumococcal bacteremia. OBJECTIVES: Characterize the outcomes of outpatients with pneumococcal bacteremia based on their evaluation at follow-up. For patients who are well-appearing without serious focal infection, propose a management scheme for reevaluation. METHODS: Retrospective review of outpatients with pneumococcal bacteremia. Patients with immunocompromise, those identified with focal bacterial infection at the initial visit, or those admitted at the initial visit were excluded. Data were collected from the initial visit (when blood culture drawn) and follow-up visit with regard to clinical parameters, laboratory data, diagnoses, and any antibiotic treatment. Decision tree analysis was used to generate a model to predict children at high risk for persistent bacteremia (PB). RESULTS: A total of 548 episodes of pneumococcal bacteremia were studied. Seventy-three children received no antibiotic, 239 oral antibiotic, and 236 parenteral antibiotic at the initial visit. Median age, temperature, and white blood cell (WBC) count were 13.5 months, 40.0 degrees C, and 20 400/mm(3). Forty-one patients had PB or new focal infections (15 with PB alone, 4 had focal infection and PB). Eight patients had meningitis at follow-up. Ninety-two percent returned because of notification of the positive blood culture result. A repeat blood culture was obtained in 92%, 23% had a lumbar puncture, 33% had a chest radiograph, and 12% were admitted. PB was associated with the antibiotic treatment group, elevation of temperature, and WBC count at follow-up. A simple management scheme using 2 sequential decision nodes of antibiotic treatment (none vs any) and then temperature at follow-up (>38.8 degrees C) would have predicted 16/19 patients with PB (sensitivity =.84 and specificity =.86). CONCLUSIONS: All patients with pneumococcal bacteremia need prompt reevaluation. For well-appearing patients without new focal infection, the utility of diagnostic testing (specifically repeat blood cultures) and the need for admission may be determined by the use of antibiotics at the initial evaluation and the presence of fever at follow-up. The majority of patients can be managed as outpatients entirely. Patients who did not receive antibiotics at the initial evaluation and those treated with oral antibiotics but remain febrile are at the highest risk for persistent bacteremia.

Non-typhi Salmonella bacteremia in children.

BACKGROUND: Non-typhi Salmonella (NTS) infections are a frequent cause of self-limited diarrheal illness in healthy children. Bacteremia is a known complication of NTS infection, but the management of children with bacteremia has been based on limited data. OBJECTIVE: To study the outcomes of pediatric patients with NTS bacteremia. METHODS: Retrospective review of patients with NTS bacteremia covering a 16-year period at an urban pediatric hospital. Clinical data from the initial visits and any follow-up visits or hospitalizations were abstracted from the medical record. RESULTS: We studied 144 patients. Median age was 10.5 months. Fifty-four patients were hospitalized at the initial visit including all the patients with immunodeficiency (n = 12). Of the 90 patients initially managed as outpatients, 79 were subsequently admitted; only 1 of these patients developed a focal complication. Persistent bacteremia was found in 51 (41%) patients. Among nonimmunocompromised patients, persistent bacteremia was noted in 34% [95% confidence interval (CI), 20 to 52%] of those initially treated with oral antibiotics, 52% (CI 30 to 74%) of those initially treated with a parenteral dose of antibiotics and in 31% (CI 22 to 43%) of those who were not initially given antibiotics. No laboratory or clinical factors predicted persistent bacteremia. Twelve patients developed focal infections: 3 of 119 previously healthy children (2.5%, CI 0.5 to 7%); and 9 of 25 children with underlying medical conditions (36%, CI 19 to 57%). Focal infections included meningitis (3), osteomyelitis (4), septic arthritis (2), pneumonia (2) and cholangitis (1). CONCLUSIONS: NTS bacteremia occurs in otherwise healthy children, although the risk of focal infections is small. Patients with NTS bacteremia frequently have persistent bacteremia at follow-up regardless of initial antibiotic treatment.

Bacteremia-associated pneumococcal pneumonia and the benefit of initial parenteral antimicrobial therapy.

OBJECTIVES: To describe clinical characteristics of patients with bacteremia-associated pneumococcal pneumonia (BAPP) and evaluate features that may distinguish these patients from those with uncomplicated pneumococcal bacteremia. To determine the impact of the route of initial antibiotic therapy on the clinical course of patients with BAPP. DESIGN/METHODS: Retrospective review of children with pneumococcal bacteremia comparing those with pneumonia to those without focal infections. RESULTS: We identified 110 patients with BAPP and 112 patients with pneumococcal bacteremia alone. Patients with pneumonia were significantly older (mean age, 34 vs. 19 months; P = 0.002) and more likely to present with cough/congestion (28% vs. 14%; P = 0.01) or difficulty breathing (12% vs. 4%; P = 0.047). There was no difference in mean temperature (39.5 vs. 39.7 degrees C; P = 0.3), mean white blood cell count WBC (21.9 vs. 22.6 x 1000/mm,3 P = 0.5) or presence of tachypnea (23% vs. 22%, P = 0.8). Sixty-one patients (55%) with pneumonia were discharged home from the initial visit in the emergency department. Those who received a parenteral antibiotic before discharge, when compared with the group who received an oral antibiotic alone, were more likely to have an improved condition (95% vs. 67%, P = 0.03) and were less likely to be admitted to the hospital (0% vs. 24%; P = 0.007) at follow-up. CONCLUSIONS: Children with bacteremia-associated pneumococcal pneumonia are older and more likely to complain of cough/congestion or difficulty breathing than those with uncomplicated pneumococcal bacteremia. The use of a parenteral antibiotic at the initial visit for children with bacteremia-associated pneumococcal pneumonia resulted in a lower admission rate and more likely parental report of improved condition at follow-up than those for children treated only with an oral antibiotic.

Clinical implications of penicillin and ceftriaxone resistance among children with pneumococcal bacteremia.

OBJECTIVES: To determine whether reduced penicillin or ceftriaxone susceptibility affects clinical presentation and outcome in children with pneumococcal bacteremia. DESIGN: Retrospective review of patients with Streptococcus pneumoniae bacteremia. RESULTS: We reviewed 922 cases of pneumococcal bacteremia. Of 744 isolates with known penicillin (PCN) susceptibilities 56 were PCN-nonsusceptible. The majority displayed intermediate resistance; 14 of 730 isolates with known ceftriaxone (CTX) susceptibilities were CTX-nonsusceptible. Neither the PCN- nor the CTX-nonsusceptible cohort displayed a difference from its susceptible counterpart in temperature, respiratory rate or white blood cell count on initial patient evaluation, although trend suggested they were more often admitted at the initial visit. At follow-up only children treated initially with antibiotic were evaluated. Children with PCN-nonsusceptible isolates were no more likely to be febrile than those with PCN-susceptible isolates (28% vs. 25%, P = 0.61) and were no more likely to have a positive repeat blood culture (0% vs. 1%, P = 0.59) or a new focal infection (10% vs. 6%, P = 0.79). Data concerning CTX-nonsusceptible organisms were limited by the low number of such isolates. Although patients with CTX-nonsusceptible pneumococci were more likely to be febrile at follow-up than those with CTX-susceptible organisms (67% vs. 24%, P = 0.04), we were unable to demonstrate a significant difference for other endpoints. CONCLUSIONS: Reduced antibiotic susceptibility does not alter the clinical presentation of pneumococcal bacteremia. With current practice intermediate resistance to PCN is of little clinical significance in nonmeningitic systemic pneumococcal infections.

Occult pneumonias: empiric chest radiographs in febrile children with leukocytosis.

STUDY OBJECTIVE: We sought to determine the incidence of radiographic findings of pneumonia in highly febrile children with leukocytosis and no clinical evidence of pneumonia or other major infectious source. METHODS: We conducted a prospective cohort study at a large urban pediatric hospital. Clinical practice guidelines for the use of chest radiography in febrile children were established by the emergency medicine attending staff. All records of emergency department patients with leukocytosis (WBC count >/= 20, 000/mm3), triage temperature 39.0 degreesC or higher, age 5 years or less were reviewed daily for 12 months. Physicians completed a questionnaire to note the diagnosis, the presence of respiratory symptoms and signs, and the reason for the chest radiograph (if one was obtained). Patients were excluded for immunodeficiency, chronic lung disease, or major bacterial sources of infection other than pneumonia. Pneumonia was defined by an attending radiologist's reading of the radiograph. RESULTS: We studied 278 patients. Chest radiographs were obtained in 225 for the following reasons: 79 because of respiratory findings suggestive of pneumonia and 146 because of leukocytosis and no identifiable major source of infection. Fifty-three patients did not undergo radiography. Pneumonia was found in 32 of 79 (40%; 95% confidence interval, 20% to 52%) of those with findings suggestive of pneumonia and in 38 of 146 (26%; 95% confidence interval, 19% to 34%) of those without clinical evidence of pneumonia. If patients who did not have a radiograph are assumed to not have pneumonia, the minimum estimate of occult pneumonia was 38 of 199 patients (19%; 95% confidence interval, 14% to 25%). CONCLUSION: Empiric chest radiographs in highly febrile children with leukocytosis and no findings of pneumonia frequently reveal occult pneumonias. Chest radiography should be considered a routine diagnostic test in children with a temperature of 39 degreesC or greater and WBC count of 20,000/mm3 or greater without an alternative major source of infection.

Parenteral vs oral antibiotics in the prevention of serious bacterial infections in children with Streptococcus pneumoniae occult bacteremia: a meta-analysis.

OBJECTIVE: To determine whether parenteral antibiotics are superior to oral antibiotics in preventing serious bacterial infections in children with Streptococcus pneumoniae occult bacteremia. METHODS: Using the MEDLINE database, the English language literature was searched for all publications concerning bacteremia, fever, or Streptococcus pneumoniae from 1966 to January 1, 1997. All nonduplicative studies with a series of children with S. pneumoniae occult bacteremia having both orally treated and parenterally treated groups were reviewed. Children were excluded from individual studies if at the time of their initial evaluation they were immunocompromised, had a serious bacterial infection, underwent a lumbar puncture, or did not receive antibiotics. RESULTS: Only 4 studies met study criteria. From these studies, 511 total cases of S. pneumoniae occult bacteremia were identified. Ten of 290 (3.4%) in the oral group and 5 of 221 (2.3%) in the parenteral antibiotic group developed serious bacterial infections (pooled p-value = 0.467, pooled OR = 1.48; 95% CI, 0.5-4.3). Two patients in the oral group (0.7%) and 2 patients in the parenteral group (0.9%) developed meningitis (pooled p-value = 0.699, pooled OR = 0.67; 95% CI, 0.1-5.1). CONCLUSION: The rates of serious bacterial infections and meningitis did not differ between children who were treated with oral and parenteral antibiotics. The extremely low rate of complications observed in both groups suggests no clinically significant difference between therapies. A study with >7,500 bacteremic children (or >300,000 febrile children) would be needed to have 80% power to prove parenteral antibiotics are superior to oral antibiotics in preventing serious bacterial infections.

Comparison of self-inflating bags with anesthesia bags for bag-mask ventilation in the pediatric emergency department.

OBJECTIVE: To compare bag-mask ventilation performed by emergency department (ED) personnel using anesthesia bags (AB) and self-inflating bags (SIB). SETTING: ED in a teaching children's hospital where the AB is the device used during resuscitations. DESIGN: Experimental study. Bag-mask ventilation was evaluated with an infant resuscitation mannequin equipped to measure airway volumes and pressures. Pediatric residents, ED nurses, and pediatric emergency medicine fellows performed bag-mask ventilation with AB and SIB and rated their confidence using each device. MAIN OUTCOME MEASURE: Ventilation failure rates. RESULTS: Seventy subjects participated (17 interns, 16 junior residents, 13 senior residents, 10 fellows, and 14 nurses). There were 13 failures with the AB (18.6%) versus 1 (1.4%) with the SIB (P < 0.01) [95% confidence interval: 5-29%], with a significant difference even after excluding the least experienced subjects. There was no difference in high pressure breaths delivered (SIB 19% vs AB 15%, P = 0.4) and a higher incidence of hyperventilation with the SIB (67 vs 25%, P < 0.01). While using the SIB, 19 (27%) of the subjects did not turn on the O2 flow. There was no difference in pretest confidence rating, but the posttest confidence rating was higher for the SIB (P < 0.05). CONCLUSIONS: Compared to SIB use for bag-mask ventilation in an ED, AB use resulted in more ventilation failures, no advantage in preventing excessive airway pressures, and less confidence among operators. The SIB should be the first choice for bag-mask ventilation in the ED, with attention to maximize oxygen delivery.

Do oral antibiotics prevent meningitis and serious bacterial infections in children with Streptococcus pneumoniae occult bacteremia? A meta-analysis.

OBJECTIVE: To determine whether oral antibiotics prevent meningitis and serious bacterial infections in children with Streptococcus pneumoniae occult bacteremia. DATA SOURCES: Using the Medline database, the English-language literature was searched for all publications concerning bacteremia, fever, or S pneumoniae from 1966 to April 1996. STUDY SELECTION: All studies that included a series of children with S pneumoniae occult bacteremia containing orally treated and untreated groups. Children were excluded from individual studies if they were immunocompromised, had a serious bacterial infection, underwent a lumbar puncture, or received parenteral antibiotics. DATA EXTRACTION: Three authors independently reviewed each article to determine the number of eligible children and the outcome of children meeting entry criteria. DATA SYNTHESIS: Eleven of 21 studies were excluded, leaving 10 evaluable studies with 656 total cases of S pneumoniae occult bacteremia identified. Patients who received oral antibiotics had fewer serious bacterial infections than untreated patients (3.3% vs 9.7%; pooled odds ratio, 0.35; 95% confidence interval, 0.17 to 0.73). Meningitis developed in 3 (0.8%) of 399 children in the oral antibiotic group and 7 (2.7%) of 257 untreated children (pooled odds ratio, 0.51; 95% confidence interval, 0.12 to 2.09). CONCLUSION: Although oral antibiotics modestly decreased the risk of serious bacterial infections in children with S pneumoniae occult bacteremia, there was insufficient evidence to conclude that oral antibiotics prevent meningitis. Published recommendations that oral antibiotics be administered to prevent serious bacterial infections in children with possible S pneumoniae occult bacteremia should be reevaluated in light of the lower risk of sequelae from S pneumoniae occult bacteremia and newer data concerning side effects from treatment.

Invasive pneumococcal infections in human immunodeficiency virus-infected children.

Invasive pneumococcal infection (IPI) is the most common serious bacterial infection in human immunodeficiency virus (HIV)-infected children. Data from a population-based pediatric HIV surveillance project were used to determine the incidence of IPI in HIV-infected children and to conduct a case-control study assessing potential risk factors for IPI in HIV-infected children. There were 50 episodes of IPI and a cumulative incidence of 6.1 cases/100 patient-years through age 7 years. Children with IPI were more likely to have a prior AIDS diagnosis (odds ratio, 4.2; 95% confidence interval, 1.2-15.1) and higher levels of IgG and IgM (P=.01) than were controls. In a separate case-control study, the manifestations of IPI in HIV-infected children were compared with those in HIV-negative controls. Focal complication rates in the 2 groups did not differ; however, HIV-infected children were less likely than controls to have leukocytosis (P<.001) and more likely to have isolates with penicillin resistance (P=.03).

Bacteremia and meningitis among infants with urinary tract infections.

A retrospective analysis of 354 patients < or = 2 years of age with urinary tract infections (UTIs) was performed to characterize patients with bacteremia or meningitis and to identify any objective predictors of these complications. Thirty-three patients with bacteremia were identified. Blood culture isolates included Escherichia coli (25), Staphylococcus aureus (4), enterococcus (1), group B Streptococcus (2), and Enterobacter (1). Besides one patient with group B Streptococcus bacteremia at 1.5 months of age, all bacteremias after one month of age were with E. coli. Bacteremia was limited to those < 6 months old and inversely related to age (R = 0.24, P = 0.0008). Grouped by age, the incidence of bacteremia was 21% for 0 < or = 1 month, 13% for 1.1-2.0 months, 4% for 2.1-3.0 months, and 8% for 3.1-6.0 months. Mean white blood cell count, initial temperature, initial serum bicarbonate, and erythrocyte sedimentation rate were not statistically significant between bacteremic (B) and nonbacteremic (NB) patients. Statistically significant differences were noted for percentage of bands (6.2% [NB] vs. 12.3% [B] P < 0.001), total band count (1048 [NB] vs. 2252 [B] P < 0.001), and band-neutrophil ratio (0.16 [NB] vs. 0.36 [B] P = 0.01); however, no practical value for any of these measures would reliably discriminate between bacteremic and nonbacteremic patients. Four patients, all neonates, had meningitis; too few patients with meningitis were identified for analysis. In summary, bacteremia with UTIs was observed to be inversely related to age and limited to patients less than six months of age. No objective parameters were identified to distinguish patients with bacteremia at the time of presentation.

Effect of antibiotic therapy on the outcome of outpatients with unsuspected bacteremia.

The records of 559 consecutive outpatient children with unsuspected bacteremia (467 Streptococcus pneumoniae) were reviewed. When compared with patients receiving oral or parenteral antibiotics, those patients who received no antibiotics at the initial visit were in follow-up: (1) less likely to be improved (32% vs. 86%, P < 0.01); (2) more likely to be febrile (75% vs. 28%, P < 0.01); (3) more likely to be hospitalized (67% vs. 22%, P < 0.01); (4) more likely to have persistent bacteremia (28% vs. 3%, P < 0.01); and (5) more likely to have new focal infections (13% vs. 5%, P < 0.01). Compared with patients receiving parenteral antibiotics at the initial visit, patients receiving oral antibiotics were in follow-up: (1) less likely to be improved (81% vs. 89%, P < 0.05); and (2) more likely to have persistent bacteremia (5% vs. 0%, P < 0.05). There was no statistical difference between patients receiving parenteral or oral therapy in the development of focal infections, although children with new focal infections receiving oral antibiotics more often had persistent or new positive cultures. No patients receiving parenteral antibiotics at the initial visit had positive blood or spinal fluid cultures at the follow-up visit. Analyses of the subgroups with (1) occult bacteremia with all organisms, (2) unsuspected bacteremia S. pneumoniae and (3) occult bacteremia with S. pneumoniae show results similar to those for the entire group.(ABSTRACT TRUNCATED AT 250 WORDS)