Sepsis Workup Research Articles

Febrile Response After Knee and Hip Arthroplasty

Documentation of the normal fever response after total knee and hip replacement is important to avoid an unnecessary workup for sepsis, and to provide justification for early discharge (dictated by the current medical reimbursement climate) despite persistent postoperative fever. One hundred patients who underwent total knee arthroplasty and 100 patients who underwent total hip arthroplasty were reviewed, several of whom had extensive sepsis workups for evaluation of postoperative fever. No patient in this series had a documented joint infection. All patients were treated with warfarin for deep vein thombrosis prophylaxis. All patients used incentive spirometry and were started on ambulation training on postoperative Day 1. All were given antibiotic prophylaxis for 48 hours. The maximum daily postoperative temperature occurred in most patients on postoperative Day 1 and gradually leveled off toward normal by postoperative Day 5. Only one patient had a maximum temperature on postoperative Day 4 that was greater than that on postoperative Day 3. Patients undergoing revision procedures tended to have a more pronounced febrile response, but the differences were not statistically significant. No significant differences were seen between patients who had epidural anesthesia and patients who had general anesthesia. Seventeen patients had postoperative chest radiographs for evaluation of fever. None had significant atelectasis. The presence of a positive urine culture had no effect on the fever response, with most positive results being identified after the fever had returned toward normal. Postoperative fever after total joint arthroplasty is a normal inflammatory response. A workup for sepsis is not indicated in the perioperative period unless corroborating signs or symptoms are present. Early discharge is appropriate if the febrile response is decreasing progressively.

Controversies of Labor Epidural Analgesia

Controversies of Labor Epidural Analgesia

The Neonatal "Sepsis Work-Up": A Population-Based Study

The Neonatal "Sepsis Work-Up": A Population-Based Study

Maternal and neonatal conditions associated with sepsis work-up in the newborn

ANESTHESIOLOGY is the highest-impact, peer-reviewed medical journal that publishes trusted evidence that transforms the practice of perioperative, critical care, and pain medicine along with clinical reviews, editorials, podcasts, infographics, and videos.

Early detection of bacteremia in the neonatal intensive care unit using the new BACTEC system.

With the newer techniques of culture analysis such as the BACTEC 9240 fluorometric detection system, detecting bacteremia in the neonate may be possible in a significantly shorter time. We hypothesized that neonatal bacteremia can be detected in less than 48 hours by this method. Our study included a retrospective review of 613 blood cultures obtained during the period August 1, 1995 to March 18, 1996 taken from 325 infants who had cultures drawn with a sepsis work-up and/or repeat cultures who had initial positive cultures in our Neonatal Intensive Care Unit. Results of blood cultures were studied in conjunction with the variables of body weight, gestational age, organism grown, complete blood count (CBC), and timing of positive cultures. Statistical analyses were performed using Fisher's and two-tailed Student's t tests. The results showed that of 325 infant blood cultures 49 (15%) were positive. Of these, 64% were coagulase-negative staphylococci, 14% viridans streptococci, 8% Escherichia coli, 4% Enterococcus sp., 4% Pseudomonas sp., 2% Enterobacter sp., 2% Klebsiella sp., and 2% Candida albicans. Of the positive blood cultures taken from infants not on antibiotics at the time of culture, 54% were detected positive at 18 hours, 71% at 24 hours, and 100% by 30 hours. Detection time by organism type was as follows: coagulase-negative staphylococci, 21.7 hours; viridans streptococci, 15.6 hours; E. coli, 7.5 hours; Enterococcus sp., 12 hours; Enterobacter sp., 5 hours; Klebsiella sp., 10 hours; and Pseudomonas sp., 12 hours. Our results indicate that the BACTEC 9240 fluorometric detection system helps in early identification of neonatal bacteremia (in 24 to 30 hours), with Gram-negative organisms being detected earlier than Gram-positive organisms (p < 0.05) and having significantly higher immature neutrophils in a CBC (I:T ratio of > or = 0.2 (p < 0.001). Early detection of neonatal bacteremia using this method will allow earlier diagnosis and appropriate treatment of the potentially bacteremic and bacteremic infant.

The Neonatal “Sepsis Work-up”: Personal Reflections on the Development of an Evidence-Based Approach Toward Newborn Infections in a Managed Care Organization

"Rule out sepsis" may be the most common discharge diagnosis among infants admitted to the neonatal intensive care unit. Although the frequency of sepsis, meningitis, and other confirmed bacterial infections has remained constant (between 1 and 5/1000 live births) for many years, the number of infants evaluated and treated is much higher. Each year in the United States, as many as 600 000 infants experience at least one evaluation for suspected bacterial infection during the birth hospitalization. The number treated is estimated at 130 000 to 400 000 per year. Despite massive overtreatment, delayed diagnosis still occurs. The Kaiser Permanente Medical Care Program (KPMCP) considers developing and implementing an evidence-based approach to "rule out sepsis," a research and operational priority. To achieve these goals, it is essential to consider two key aspects of the problem. First, it is important to adopt a phenomenologic approach that takes clinicians' personal experience into account. This must include reflection on those aspects of experience often considered "irrational" or "subjective." Second, incorporation of a phenomenologic approach needs to be tempered with sound epidemiologic methods. If one considers these two aspects-physician experience and sound epidemiology-it is clear that much of the existing literature on "rule out sepsis" is of limited utility. Consequently, the KPMCP has conducted its own studies. These are aimed at characterizing the "sepsis work-up," developing electronic datasets that would permit clinicians to simulate various strategies, and developing techniques for ongoing electronic monitoring. This article summarizes the approach taken by the KPMCP Division of Research. It describes the results of a pilot study as well as the development and use of a dedicated neonatology outcomes database, the Kaiser Permanente Neonatal Minimum Data Set (NMDS). The NMDS database includes the Score for Neonatal Acute Physiology and permits ongoing monitoring of sepsis "work-ups" as well as confirmed cases of neonatal infection. The article also describes how the experience from the pilot as well as the NMDS was incorporated in the design of a much larger study on "rule out sepsis." Finally, the article describes some important theoretic issues affecting decision rule development and the use of computer simulations in neonatology. These issues are 1) how one handles possible overanalysis of a dataset; 2) how one handles data points that are unstable (eg, the absolute neutrophil count, which can vary considerably depending on age and sampling conditions); and 3) the limitations of decision rules based on computer simulations.

Immune alterations associated with epidural analgesia for labor and delivery.

Epidural analgesia is frequently used for the laboring woman and commonly regarded as safe. The association of epidural analgesia with fever, in the absence of infection, in this population can result in unnecessary sepsis workups in both neonates and mothers. Studies in other populations of patients have found that epidural anesthesia may be associated with alterations in white blood cell count parameters. This study was designed to determine if epidural analgesia is associated with alterations in immune profile or immune function in healthy afebrile postpartum women. This prospective quasi-experimental study examined a convenience sample of normal afebrile postpartum women in two groups for differences in immune profile and immune function based on whether they received epidural analgesia during labor and delivery. Mothers who agreed to participate in the study and met inclusion criteria had blood drawn for immune profile and immune function studies within 24 hours of their delivery. This study looked at immune phenotypic profile, lymphocyte proliferative response, and NK lysis assays. No differences in immune profile or immune function were found based on administration of epidural analgesia. Nurses can be confident that the white blood cell count and other phenotypic measures of leukocytes are not affected by epidural analgesia. Thus the epidural analgesic technique appears not to alter immune measures of infection in postpartum women. In evaluating fever in postpartum mothers and neonates, nurses should consider prior receipt of epidural analgesia and measures of immune profile. In addition to fever and white blood cell counts, other signs and symptoms of infection should also be evaluated in postpartum women when infection is suspected.

Childhood hypoglycemia in an urban emergency department: epidemiology and a diagnostic approach to the problem.

To 1) determine the prevalence of hypoglycemia in childhood in a pediatric emergency department (ED), 2) determine epidemiology of idiopathic ketotic hypoglycemia (IKH), 3) determine diagnostic yield of the workup of hypoglycemia, and 4) review a diagnostic approach to hypoglycemia. Urban pediatric ED of a tertiary level children's hospital. Retrospective review of all medical records with a primary or secondary diagnosis of hypoglycemia (ICD-9 code 251.2) seen at the ED between 1/92 and 8/95. Thirty-one patients were identified. Mean blood glucose was 34.2 mg/dl. Prevalence of hypoglycemia among population seeking care in our ED was 6.54/100,000 visits. Eighteen patients were diagnosed with IKH for a prevalence of 3.9/100,000. IKH demographics were: mean age 27.7 months; 12 males, 6 females; 8 white, 9 black, and 1 not available. The weights of five patients were < 25th percentile. Fourteen of the 18 IKH patients had hormone studies done insulin [cost $40], growth hormone [$69], cortisol [$54]. All 14 had appropriately suppressed insulin levels (< 5microU/ml) and high cortisol levels > 22 microg/ml. Thirteen of the 14 had normal or high growth hormone (GH) levels (0.7-6 ng/ml). Four IKH patients had urine drug screens ($280); all were negative. Although no IKH patient was febrile, six had sepsis workups ($380); all were negative. Urine ketones were positive in 15 of the 18 tested (> 3+ in eight patients). Mean anion gap was 20 (range: 16-30). Eight of the 18 IKH patients were discharged from the ED after return to normal status. IKH is the most common cause of hypoglycemia in children beyond the infancy period. In its typical presentation (previously healthy one- to five-year-old, with normal growth and development, who presents with a first episode of symptomatic fasting hypoglycemia and appropriate degree of ketonuria, without hepatomegaly, and with resolution of symptoms on administration of glucose), an extensive and overzealous workup for endocrinopathy or inborn error of metabolism is not necessary.

Rational Antibiotic Utilization in the Newborn Intensive Care Unit 1418

The nonspecifity of signs for infection in the newborn leads to frequent sepsis workups and the use of antibiotics. The wide range of potential pathogens makes it difficult to differentiate between those infants with true infection and those whose blood culture is contaminated. There are currently no microbiologic techniques or clinical criteria to differentiate true pathogens from contaminates. Skin colonization is common and increases with time in the neonatal intensive care unit but does not predict systemic illness. We reviewed 39 charts of infants with positive blood cultures. Each infant had a paired blood culture in addition to white blood cell count (WBC), leukocyte indices, platelet count and C-reactive protein (CRP). 24 were considered true pathogens; 15 were considered contaminants. Infants with proven sepsis were likely to have abnormal total white count (p value< 0.05) and elevated CRP (p value <0.0001). The odds ratio of not having an infection with a normal WBC and CRP was 1.6. The odds ratio of having a true infection with abnormal WBC and CRP was 1.2. No infant with true infection had a normal screen. Normal serial white counts and CRP make the diagnosis of true infection unlikely. A normal screen is useful in differentiating a contaminant from a true pathogen even in the premature population and will allow antibiotic therapy to be discontinued earlier safely, thereby reducing length of stay.

Sepsis evaluations in hospitalized infants with bronchiolitis.

To define variation in the decision to perform a sepsis evaluation in hospitalized infants with bronchiolitis, to define predictors of the decision and to measure the clinical and cost outcomes. Retrospective chart review of all nonintensive care unit infants < or = 60 days with any discharge diagnosis of bronchiolitis (n = 282 from 1993 to 1995 in a 232-bed pediatric hospital. Process variables included temperature at sepsis work-up or Tmax if no sepsis workup. Outcome variables were charges, length of stay, sepsis workup and serious bacterial infection. There was no difference in mean temperature between groups with or without sepsis evaluation (38.1 degrees C, P = 0.75). Of 282 infants 140 had a sepsis workup; 5 (1.8%) had serious bacterial infection. Infants with sepsis workup had an average total charge of $4507 and length of stay of 3.4 days compared with $2998 and 2.8 days for those without (P = 0.0001 and P = 0.002, respectively). A multivariate logistic regression model was constructed with sepsis workup as the dichotomous dependent variable. Significant (P < or = 0.05) predictor variables with a positive coefficient were: higher bronchiolitis score and normal chest roentgenogram. Significant variables with a negative coefficient were: admission diagnosis of bronchiolitis, chest roentgenogram typical for bronchiolitis and age > 28 days. Temperature was not a predictor of sepsis evaluation. Infants with respiratory distress and normal chest roentgenogram were more likely to receive sepsis evaluations; those with recognized typical bronchiolitis and those > 28 days of age were less likely. Risk of serious bacterial infection is low; the costs of a sepsis evaluation outweigh the benefits in infants with obvious bronchiolitis.

Neonatal Deaths in the Asir Region of Saudi Arabia: Experience in a Referral Neonatal Intensive Care Unit

We reviewed a total of 169 deaths among 969 infants referred from various health institutions in the Asir Region to the neonatal intensive care unit of Asir Central Hospital, Abha, Saudi Arabia, during a four-year period (January 1992 to December 1995). The results showed that the major causes of neonatal mortality in the region were low birth weight (LBW) (45%), congenital malformations (CM) (30.8%), infection (13.6%), and birth asphyxia (7.7%). The majority of the LBW infants died from respiratory insufficiency, sepsis, and necrotizing enterocolitis. The alimentary, central nervous and cardiovascular systems were the parts most commonly involved in CM, with cardiac defect accounting for the highest fatality rate due to lack of facilities for cardiac surgery in the region, and delay in effecting a transfer. Klebsiella, Serratia and E. coli are the predominant organisms causing neonatal sepsis, in contrast with the group B streptococcus dominating in America and the United Kingdom. Generally, factors causing neonatal deaths in the Asir region are similar to those reported in other tertiary hospitals in Saudi Arabia, and tally closely with observations in America and Europe, where the standard of health care is excellent. Nevertheless, the study indicates the need for upgrading of facilities, as well as for more trained personnel for obstetric and neonatal care in Asir region. It may be necessary to consider the need for the establishment of a regional cardiac surgery center in the region.

Diagnostic Accuracy of Csf β2M Compared To Total Proteins in Neonates

The diagnosis of CNS infections may be difficult in neonates. CSF shunt infections may be particularly difficult to assess .Objectives: (1) to compare the diagnostic accuracy ofβ2-microglobulin (β2m) vs total proteins in the CSF, in neonates with suspected CNS infections; (2) to determine the value of CSFβ2m in the diagnosis of CSF shunt infections .Methods: Neonates were included prospectively if a lumbar puncture was performed as part of a sepsis or meningo-encephalitis work-up. β2m was measured by enzyme immunoassay; total proteins by nephelometry. Results: 105 neonates without CSF shunts were studied: 30 had CNS infection (13 bacterial meningitis, 11 viral, 6 congenital meningo-encephalitis), and 75 did not. Areas under the ROC curve were: β2m 0.95±0.02 (cut-off value 2.5 mg/L), and total proteins 0.84±0.05 (cut-off value 100 mg/dL). β2m showed higher sensitivity and positive predictive value than proteins (97 vs 78, and 92 vs 54, respectively, p<0.001). In addition, 10 infants with CSF shunt were studied; 4 had infection. β2m was higher in infants with infection around the time of clinical suspicion (6.1±1.75 vs 3.1±1.04 mg/L, p<0.01), and β2m levels tended to increase 2 to 5 days after (10.2±0.97 vs 6.1±1.75,p=0.06). Conclusion: β2m showed a high diagnostic accuracy, with higher specificity and positive predictive value than total proteins. Total proteins had a lower area under the ROC curve, although the difference was not statistically significant.β2m in the CSF might be a useful ancillary tool in the diagnosis of neonatal CNS infection, including infants with CSF shunts.

Radiological Pulmonary Changes During Gram Negative Bacillary Nosocomial Bloodstream Infection (BSI) In Premature Infants. † 1143

We designed a retrospective study to characterize changes in chest radiographs that occur at the time of Gram negative bacilli (GNB) nosocomial BSI and to determine the radiological findings of neonatal nosocomial pneumonia. Forty-one infants who had BSI (study group) and 50 GNB airway colonized infants who had sepsis workup with negative blood cultures (control group) were studied. We correlated clinical data, 95 blood and 305 tracheal aspirate (TA) cultures, with radiographic findings from the day before, the day of, and the day after blood cultures were taken. Two radiologists independently graded 258 films using a modified score (AJDC, 1984;138:581) for bronchopulmonary dysplasia (BPD) and a semi-quantitative system (probable, possible or unlikely) for pneumonia. Of the study population, 54% were male, 21% black, 89% received surfactant, 79% dexamethasone and 88% survived. Mean birth weight was 1057g and 1044g for the study and control groups, respectively. Average age at the time of blood cultures was 23 days for both groups. Most common isolates from blood and TA were Klebsiella pneumoniae, Enterobacter cloacae, Escherichia coli and Pseudomonas aeruginosa. BSI infants (mainly P.aeruginosa and K.pneumonia) died twice as often as controls. Radiological findings: BPD scores did not change in relation to the day of the blood culture and were similar between study and control groups. Twenty-one BSI patients presented with newly positive TA at the time, or within 24 hours of blood culture. Probable or possible pneumonia was diagnosed in 18 of them. Their BPD scores were higher than those of the remaining BSI patients, of whom 7 were already airway colonized, 9 had negative TA's, and 4 were not intubated. Only one of these 20 patients had a possible pneumonia score. CONCLUSION: Radiographic signs of pneumonia in infants with BPD are of significance only when associated with positive blood culture with a respiratory pathogen. Thus, the CDC definition of nosocomial pneumonia is correct and is further supported by the recovery of the same GNB from the airway of an infant with previously negative tracheal aspirate cultures.

Managing fever in infants and toddlers

PREVIEWWhen does a child's high temperature call for aggressive treatment? In what circumstances is a febrile infant at such low risk for serious infection that outpatient management is permissible? The authors answer these questions and more as they discuss fundamental considerations (fever, toxicity, age-group differences, the sepsis workup), describe use of the Rochester criteria for assessing infection risk, recommend treatment strategies by age-group, and note acceptable variations, based on physician and parent preferences, from the recommended strategies.

Identification of febrile neonates unlikely to have bacterial infections.

A prospective study was undertaken to evaluate the usefulness of low risk criteria for identifying febrile neonates unlikely to have bacterial infections, particularly bacteremia and meningitis. Using these criteria we wished to check whether these low risk neonates can be safely managed as inpatients under close observation but without receiving empiric antibiotic therapy. We conducted a prospective study of 250 consecutive infants 28 days of age or less (range, 4 to 28 days) who had rectal temperatures of at least 38 degrees C. After a complete history, physical examination and sepsis workup, the 131 febrile neonates with clinical and laboratory findings indicating low risk for bacterial infections were hospitalized, closely observed and not given antibiotics. The low risk criteria used to identify these neonates included well appearance; absence of physical signs of an ear, eye or soft tissue infection; a white blood cell count of 5000 to 15,000/mm3, a neutrophil band form count of < 1500/mm3, a spun urine specimen that had < 10 white blood cells per high power field on microscopy and a C-reactive protein value of < 20 mg/l. The remaining 119 febrile neonates who did not meet the low risk criteria were hospitalized and given empiric antibiotics pending culture results. The overall incidence of bacterial infections in the 250 febrile neonates was 16.4% (41 neonates) with bacteremia and/or meningitis occurring in 4.4% (11 neonates). Of the 131 low risk neonates only one (0.8%) had a bacterial infection (urinary tract infection) compared with 40 (33.6%) among the 119 who did not meet the criteria (P < 0.05). The negative predictive value of these criteria for excluding bacterial infections was 99.2% (95% confidence interval, 97.7% to 100%). For bacteremia and meningitis this figure became 100%. However, the positive predictive value and specificity of the criteria were too low to identify all febrile neonates with bacterial infections. Among the low risk group 58 (44.3%) were reclassified on the second or third hospital day, because of continued fever (35), poor activity (14) and white blood cell count > 15000/mm3 on a repeat blood test (9). They were given antibiotics, but only 1 had a urinary tract infection. This patient was then treated with a 7-day course of antibiotics. All low risk neonates recovered uneventfully with no relapses observed during their hospital stay or at the time of minimum 1-week outpatient follow-up visits. By the low risk criteria a substantial number of febrile neonates at low risk for bacterial infections can be identified and managed safely as inpatients without antibiotic therapy.

Detection of enteroviruses by polymerase chain reaction in urine samples of patients with aseptic meningitis.

Detection of enteroviruses by polymerase chain reaction in urine samples of patients with aseptic meningitis.

Hospital Readmission Due to Neonatal Hyperbilirubinemia

Severe neonatal hyperbilirubinemia can occur without apparent reason in term healthy breast-fed infants and some develop kernicterus. The aim of our study was to assess the incidence of severe hyperbilirubinemia in term healthy newborns discharged from the hospital. From January 1 through December 31, 1994, 6705 infants were delivered at Bikur-Cholim and Misgav-Ladach Community Hospitals. All 1448 newborns discharged with a serum bilirubin level > 10.0 mg/dL were instructed to return to the hospital within 3 days for follow-up, as well as bilirubin determination. Twenty-one newborns with a bilirubin level > 18.0 mg/dL were identified and readmitted at mean +/- standard deviation (SD) 5.5 +/- 1.8 (range, 5 to 10 days of life). This represents 1.7% of the 1220 infants who returned for follow-up examination. Mean +/- SD serum bilirubin levels at readmission were 19.6 +/- 2.5 mg/dL. All but one of the infants were breast-fed. No cases of ABO incompatibility were found and two newborns were glucose-6-phosphate dehydrogenase (G6PD)-deficient. Sepsis work-up and direct Coomb's tests were negative in all cases. None had hemolysis or were found to have any cause for hyperbilirubinemia other than breast-feeding. Phototherapy was provided in all but two cases, and an exchange transfusion was performed in one case. Three additional infants, with bilirubin levels < 10 mg/dL at discharge, were readmitted due to hyperbilirubinemia. One was diagnosed with neonatal hepatitis. We conclude that, based on our study population, 0.36% of term infants may subsequently develop severe neonatal hyperbilirubinemia in the first postnatal week.(ABSTRACT TRUNCATED AT 250 WORDS)

Cerebrospinal fluid 309-1309-1309-1in neonates with central nervous system infections

Beta2-microglobulin (β2m) determination in CSF of 72 neonates who underwent a spinal tap as part of a sepsis or meningo-encephalitis workup was performed to evaluate the usefulness of this test in the diagnosis of CNS infections. β2m was measured by enzyme immunoassay. Sixty neonates had sterile culture and normal neurological status at discharge. Twelve infants had CNS infections: 8 bacterial meningitis, 3 TORCH infections (T=toxoplasmosis, O=others, R=rubella, C=cytomegalovirus and H=herpes simplex) and 1 viral meningitis. Neonates with CNS infection exhibited significantly higher CSF β2m levels compared to neonates with sterile culture (6.24±2.66 vs 1.74±0.5 mg/l;P<0.0001). CSF β2m levels did not correlate with the white cell count, total protein concentration or glucose level in CSF. When serum and CSF levels were measured simultaneously, the CSF β2m level was significantly higher than the corresponding serum level in patients with CNS infection (6.98±2.5 vs 3.2±0.25 mg/l;P<0.01). Sensitivity, specificity, and predictive values were estimated for different cut-off points. The best operational diagnostic cut-off value was 2.25 mg/l. Receiver operating characteristic curve analysis showed an appropriate trade-off between specificity and sensitivity and indicated that CSF β2m was accurate in distinguishing between neonates with and without CNS infection.

Cerebrospinal fluid ?2-microglobulin in neonates with central nervous system infections

Beta 2-microglobulin (beta 2m) determination in CSF of 72 neonates who underwent a spinal tap as part of a sepsis or meningo-encephalitis workup was performed to evaluate the usefulness of this test in the diagnosis of CNS infections. Beta 2m was measured by enzyme immunoassay. Sixty neonates had sterile culture and normal neurological status at discharge. Twelve infants had CNS infections: 8 bacterial meningitis, 3 TORCH infections (T = toxoplasmosis, O = others, R = rubella, C = cytomegalovirus and H = herpes simplex) and 1 viral meningitis. Neonates with CNS infection exhibited significantly higher CSF beta 2m levels compared to neonates with sterile culture (6.24 +/- 2.66 vs 1.74 +/- 0.5 mg/l; P < 0.0001). CSF beta 2m levels did not correlate with the white cell count, total protein concentration or glucose level in CSF. When serum and CSF levels were measured simultaneously, the CSF beta 2m level was significantly higher than the corresponding serum level in patients with CNS infection (6.98 +/- 2.5 vs 3.2 +/- 0.25 mg/l; P < 0.01). Sensitivity, specificity, and predictive values were estimated for different cut-off points. The best operational diagnostic cut-off value was 2.25 mg/l. Receiver operating characteristic curve analysis showed an appropriate trade-off between specificity and sensitivity and indicated that CSF beta 2m was accurate in distinguishing between neonates with and without CNS infection. Conclusion. CSF beta 2m may be a useful ancillary tool in neonates when CNS infection is suspected.

Outpatient Management without Antibiotics of Fever in Selected Infants

In many academic centers it is standard practice to hospitalize all febrile infants younger than two months of age, whereas in community settings such infants are often cared for as outpatients. We conducted a controlled study of 747 consecutive infants 29 through 56 days of age who had temperatures of at least 38.2 degrees C. After a complete history taking, physical examination, and sepsis workup, the 460 infants with laboratory or clinical findings suggestive of serious bacterial illness were hospitalized and treated with antibiotics. The screening criteria for serious bacterial illness included a white-cell count of at least 15,000 per cubic millimeter, a spun urine specimen that had 10 or more white cells per high-power field or that was positive on bright-field microscopy, cerebrospinal fluid with a white-cell count of 8 or more per cubic millimeter or a positive Gram's stain, or a chest film showing an infiltrate. The 287 infants who had unremarkable examinations and normal laboratory results were assigned to either inpatient observation without antibiotics (n = 148) or outpatient care without antibiotics but with reexaminations after 24 and 48 hours (n = 139). Serious bacterial illness was diagnosed in 65 infants (8.7 percent). Of these 65 infants, 64 were identified by our screening criteria for inpatient care and antibiotic treatment (sensitivity = 98 percent; 95 percent confidence interval, 92 to 100). Of the 287 infants assigned to observation and no antibiotics, 286 (99.7 percent) did not have serious bacterial illness. Only two infants assigned to outpatient observation were subsequently admitted to the hospital; neither was found to have a serious illness. Outpatient care without antibiotics of the febrile infants at low risk for serious illness resulted in a savings of about $3,100 per patient. With the use of strict screening criteria, a substantial number of febrile one-to-two-month-old infants can be cared for safely as outpatients and without antibiotics.