Historically, management of infants with fever without localizing signs (FWLS) has generated much controversy, with attempts to risk stratify based on several criteria. Advances in medical practice may have altered the epidemiology of serious bacterial infections (SBIs) in this population. We conducted this study to test the hypothesis that the rate of SBIs in this patient population has changed over time.
Patients and Methods
We performed a retrospective review of all infants meeting FWLS criteria at our institution from 1997–2006. We examined all clinical and outcome data and performed statistical analysis of SBI rates and ampicillin resistance rates.
668 infants met criteria for FWLS. The overall rate of SBIs was 10.8%, with a significant increase from 2002–2006 (52/361, 14.4%) compared to 1997–2001 (20/307, 6.5%) (p = 0.001). This increase was driven by an increase in E. coli urinary tract infections (UTI), particularly in older infants (31–90 days).
We observed a significant increase in E. coli UTI among FWLS infants with high rates of ampicillin resistance. The reasons are likely to be multifactorial, but the results themselves emphasize the need to examine urine in all febrile infants <90days and consider local resistance patterns when choosing empiric antibiotics.
Citation: Watt K, Waddle E, Jhaveri R (2010) Changing Epidemiology of Serious Bacterial Infections in Febrile Infants without Localizing Signs. PLoS ONE 5(8): e12448. doi:10.1371/journal.pone.0012448
Editor: Adam J. Ratner, Columbia University, United States of America
Received: April 12, 2010; Accepted: July 21, 2010; Published: August 27, 2010
Copyright: © 2010 Watt et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
The management of infants <90 days with fever without localizing source (FWLS) has been a source of much controversy and debate for the last 30 years. While the majority of these infants will only have a minor viral or bacterial infection, the literature reports that approximately 12% in those aged <30days and 9% in those 30–90 days will have a serious bacterial infection (SBI), such as bacteremia, meningitis, or urinary tract infection (UTI) , , , . In order to better predict those infants at risk for SBI, Dagan et al evaluated a combination of clinical and laboratory data (no focal exam, white blood cell count (WBC) between 5,000 and 15,000/mm3, band forms <1500/mm3, normal urinalysis (UA)) to identify low and high risk groups in what would become the Rochester Criteria . Modified criteria have followed that were effective at identifying low risk infants, but differed in the exact data included which resulted in inconsistent implementation , , , , , .
Several changes in the last 20 years have significantly altered the epidemiology of SBI in neonates. Group B Streptococci (GBS) and E coli have traditionally been the most important pathogens in this age group. Institution of culture-based screening and prophylaxis for GBS , ,  has significantly lowered the incidence of this pathogen. Additionally, several authors have noted possible increased ampicillin resistance rates among pathogens causing SBI in this age group , . In light of these changes, we conducted this study of all infants less than 3 months of age with FWLS over the last 10 years. The study questions were: What is the current frequency and distribution of SBIs in these infants and has this changed over time? What are the current rates of antibiotic resistance in pathogens identified in these patients? How do practitioners manage these patients?
The study was conducted at Duke University Hospital, a large, tertiary care hospital in Durham, NC. Physicians in the Duke Emergency Department (ED) see over 5500 children per year who are less than 3 years of age. This project was reviewed and approved by the Duke University Institutional Review Board. A waiver of informed consent was obtained for this study because this was a retrospective study examining hospital records containing data derived for the purposes of clinical care.
Using the Clinical Microbiology laboratory database, we identified all children less than 90 days of age seen in the emergent setting that had a blood culture performed from 1997–2006. After careful chart review, we identified those infants meeting criteria for FWLS and performed further analysis. Patients were considered febrile if they had a history or examination temperature of 38.0°C or higher. Temperatures were measured rectally in the ED, and we included a fever ≥38.0°C taken by the parents that was considered reliable by the ED provider. Exclusion criteria were significant underlying illness or past medical history (PMH), subjective reports of “feeling warm” without a temperature taken, ill appearance, localizing source of infection after a thorough physical examination, or incomplete medical records. PMH could include history of immunodeficiency, previous hospitalization, significant congenital anomaly, or current antibiotic use. Complete records included documented fever, presenting symptoms, physical findings, and culture results.
Records were reviewed using the electronic medical record including ED notes, discharge summaries, information on prior and subsequent visits, and all laboratory and radiologic tests. For each patient we recorded demographic information, pertinent history, physical exam, and test results.
Definition of SBIs
Blood culture isolates were considered pathogens if the organism is known to cause disease in healthy infants: Escherichia coli, Staphylococcus aureus, Streptococcus pneumoniae, Enterococcus sp., Streptococcus agalactiae (GBS), Enterobacter cloacae, Klebsiella sp., Pasteurella sp., Moraxella sp., and Citrobacter sp. Organisms that were considered contaminants included coagulase-negative Staphylococci, Bacillus sp., Diptheroids, and Streptococcus viridans. Organisms isolated from cerebrospinal fluid (CSF) were considered pathogens if the organism was known to cause meningitis. To separate colonization from infection, urine cultures were considered positive if a known pathogen was isolated from a catheterized specimen with >100,000 colony forming units (CFUs)/mL or 10,000–100,000 CFUs/mL with an abnormal urinalysis (UA) or in a clean catch specimen if there were >100,000 CFUs/mL and an abnormal UA. An abnormal UA was defined as positive leukocyte esterase, nitrite, or ≥10 white blood cells (WBC)/hpf and with less than 10 squamous epithelial cells/hpf. We analyzed all clinical outcomes, initial choice of antibiotics, as well as any subsequent changes made as a result of bacterial resistance or change in patient's clinical status.
Assignment of Low Risk/High Risk Status
Infants were classified as either low risk (LR) or high risk (HR) based on historical, clinical and laboratory data. We applied the practice guidelines of Baraff to our patients , . Infants were considered LR if they were nontoxic appearing, previously healthy, term infants with no focal source of infection upon examination. While the Baraff guidelines considered all infants less than 28d to be HR, we opted to only consider patients for LR stratification if they were older than 30 days. The laboratory criteria for LR patients included a WBC count of 5–15,000/mm3, <1500 bands/mm3, negative gram stain of unspun urine or <10 WBCs/hpf, and <5 WBCs/hpf in stool when diarrhea is present.
Statistical comparisons were performed using a two-tailed t-test and chi-square testing for all numerical and categorical values, respectively. Significance was set at p<0.05. We also performed a Bayesian analysis to calculate unconditional probabilities of SBI using a beta binomial with a (1,1) prior. This type of analysis generates an unconditional probability rather than a conditional one based on the null hypothesis being accepted or rejected . Bayesian analysis also allows the incorporation of past studies/experience into current data analysis to generate a probability distribution that may be more clinically relevant. We generated probability curves for SBIs in the early and later periods and compared them to 2 other published SBI rates, the original Rochester study and a more recent study from Utah , . We used SPSS Version 15.0, an open-source, web based program (www.openepi.com) and JMP7 (SAS) to perform the various statistical testing.
Between 1997–2006 1501 infants < = 90 days old had a blood culture drawn for any reason in our ED. The numbers of patients included in the final analysis are summarized in Table 1. Our screening process for FWLS looked first for significant PMH, then absence of fever, ill appearance, and focality of exam. We did include infants born as young as 36 weeks if there was documentation that they were routinely discharged and had no underlying medical problems. After review, 668 patients (44.5%) fit the criteria for FWLS and were included in the analysis. We divided the time period into two 5-year blocks (1997–2001 and 2002–2006) that roughly approximated the time before and after the transition from risk-based to culture-based screening for GBS by the Duke Department of Obstetrics and Gynecology. The earlier period included more infants in the youngest age group (≤30d), while the later period included more infants in the older age group (61–90 days). Otherwise the groups were not significantly different (Table 1). A total of 833 infants did not meet criteria for FWLS, 386 with significant PMH, 355 who were afebrile, 37 who were ill appearing, 29 with a focal exam, and 26 that had incomplete medical records. The majority of those with significant PMH were former premature infants with a previous NICU stay or infants born with a congenital anomaly (e.g. congenital heart disease, shunted hydrocephalus) or immunodeficiency (e.g. SCID, HIV exposure). Most of the remainder had either been hospitalized previously or were currently on antibiotics.
Table 1. Patient Demographics.doi:10.1371/journal.pone.0012448.t001
Number and Rate of SBIs
The year by year results of our analysis are represented in Figure 1A. The overall rate of SBIs in our study over 10 years was 10.8%, but closer analysis of the trend line reveals a marked increase in SBIs during this ten year period (Figure 1B). When examined as a continuous variable, the rate of SBI in those infants ≤90d with a blood culture drawn increased over time (coef 0.43, p<0.008) and the goodness of fit of our trend line showed a modest association (R2 = 0.61). When examined as two distinct groups, infants in the later period had significantly more SBIs than those in the earlier period (14.4% vs. 6.5%, p = 0.001), with increased UTIs accounting almost entirely for the difference (45 cases of UTI in the later period vs. 13 in the earlier period, p = 0.002) (Table 2).
Figure 1. Analysis of Raw Data and Trends in SBIs among FWLS Infants, 1997–2006.
A) Raw numbers of FWLS infants as well as SBIs and % SBI. B) Data in Figure 1a is graphed by %SBI on the y axis and year on the x axis (blue line). Also presented: overall mean %SBI (green line) and linear regression (red line). Statistical data on linear regression line is included.doi:10.1371/journal.pone.0012448.g001
Table 2. Types of Serious Bacterial Infections in Early and Later periods.doi:10.1371/journal.pone.0012448.t002
Results from the Bayesian analysis comparing our data with that from the original Rochester study and a more recent study from Utah ,  are presented in Figure 2. The probability curves for the Rochester and Utah studies overlap with the SBI probability from the early period of our study. The SBI probability curve for the later period of our study has very little overlap with the tail of one of the other curves and no overlap with the others, reflecting a distinct epidemiology among the infants in this group.
Figure 2. Bayesian Analysis of SBIs among FWLS Infants, 1997–2001 and 2002–2006.
The graph represents the probability of SBI based on data from the early and later periods of this study compared with the Rochester data as well as data on FWLS infants from Utah , . The probability curve from the data from the later period centers at approximately 14.4% whereas the other three curves all overlap between 5–10% probabilities.doi:10.1371/journal.pone.0012448.g002
Breakdown of SBIs by type and age group
During the period of 1997–2001, 4.2% of the sample had positive urine cultures, the majority of which were E. coli (84.6%) (Table 3). From 2002–2006, 12.2% of the sample had positive urine cultures (p = 0.0002), almost all of which were E. coli (91.5%). We did see patients that had two or more species of bacteria isolated from their urine. Unless two distinct pathogens are isolated, our lab reports multiple speciation as “mixed flora” and these are considered contaminants and were not included as positive UTIs. We did, however, count as positive two patients with urine cultures that met our inclusion criteria who had two pathogenic species isolated by urine culture (Enterobacter and Proteus; E. coli and Proteus). In the early period, one of the patients with UTI had a concordant bacteremia, compared with five patients in the later period, all of which were concordant E. coli from each site.
Table 3. Bacterial Pathogens by Species and Site of Infection.doi:10.1371/journal.pone.0012448.t003
There were 19 patients with a positive urine culture but <10 WBCs on the UA, 5 in the early period and 14 in the late period. In the early period, one of these patients met criteria for colonization (clean catch specimen with >100,000 E. coli but negative UA), and five patients from the later period met criteria for colonization (3 with Enterococcus, one Pseudomonas, and one E. coli).
There was no significant difference in episodes of bacteremia with 7 cases (2.3% or 23 cases per 1000 infants with FWLS) in the early period and 11 cases (3%) in the later period (p = 0.54) (Table 3). E. coli was responsible for 14.3% of cases in the early period and 45.5% in the late. During the period of 2002–2006, we saw 2 (0.6%) cases of meningitis. One of these was due to E. coli. The blood culture for this patient was negative, but multiple attempts to obtain a urine specimen were unsuccessful. The other case was caused by Enterobacter cloacae in a patient that also had a positive blood culture.
UTI rates were increased in all age groups, but the increase was not statistically significant in the youngest infants (≤30d) (Table 4). Between 1997–2001 and 2002–2006 the rate of UTI amongst infants ≤30d went from 3.8% to 7.2% (p = 0.3). Amongst those 31–60d it went from 3.7% to 11.9% (p = 0.01). In the infants age 61–90d the incidence of UTI increased from 5.8% to 17.7% (p = 0.02). There were no other significant differences between the early and late period by age group though there was a trend towards increasing bacteremia in the infants aged 0–60d.
Table 4. SBIs by Age Group.doi:10.1371/journal.pone.0012448.t004
Ampicillin resistance for all SBI was 41.7% over the 10 year period, 25% for the period 1997–2001 and 48.1% during the later period 2002–2006 (p = 0.08). Examining the UTI subset, the rate was 46.6% overall and, again, the difference between the early and late periods did not achieve statistical significance. Initial antibiotic choice was changed in a third of cases of SBI after antibiotic sensitivities were known. The six patients with ampicillin resistant gram negative rod bacteremia (1 from the early period, 5 from the later period) all had antibiotics switched when sensitivities were known and all had longer antibiotic courses compared to non-bacteremic infants. Four of those patients stayed at least 2 extra days in the hospital and the other two kept indwelling venous catheters for at least 2 extra days.
Management of FWLS Infants
Management differed both over time and between age groups (Table 5). Among the age groups over the 10 year period almost all infants 30d or less received a complete workup including collection of blood, urine, and CSF cultures. There was no statistical difference in urine studies performed between the younger (≤30d) and older (31–90d) infants (p = 0.54), but older infants (31–90d) had significantly fewer LPs performed and antibiotics given than younger infants (p<0.001). Among those older infants that were stratified as HR, only 66.9% had an LP performed which is highly significant compared with the infants less than 30d old (p<0.001). Comparing 1997–2001 with 2002–2006, there were significantly fewer LPs performed (p = 0.002) and antibiotics given (p = 0.0004) in the later period.
The purpose of this study was to investigate the rates of SBI, patterns of antibiotic resistance, and practitioner management, given the possible changing epidemiology. In our analysis, we identified several trends that may impact the care and outcomes of this group. The most significant was a major increase in the rate of SBI, arising predominantly from increased rates of E. coli UTIs which has not been previously reported. This increase is particularly pronounced when compared to previous studies. The original Rochester study had an SBI rate of 10%, but they included infants with a focal exam in their study which would have resulted in higher rates of SBI . A more recent study from Utah demonstrated an 8% SBI rate . Our 2002–2006 SBI rate of 14.4% is markedly higher. The reasons for this increase are not entirely clear, but we hypothesize several potential factors that may have contributed in some way to this increase.
The first is widespread intrapartum antibiotic prophylaxis (IAP) for GBS may be selecting for gram negative bacteria that have a propensity to cause disease in young infants. Many studies have examined IAP ampicillin use and found no increase neonatal E. coli sepsis , , , . In contrast, two studies found that infants with invasive E. coli were more likely to have been exposed to IAP , . Our hospital pharmacy data indicates that from 2002–2006, which are the earliest years available, approximately 20–30% of pregnant women received ampicillin. One would expect that if IAP was the primary reason for the increase, most of the cases would be among the youngest infants closest to the exposure during delivery. Our data shows the opposite, that the burden of UTI is spread disproportionately among the older infants.
Another potential reason is the emergence of a new clone that coincided with the switch in GBS prophylaxis at our institution. Clonal group A is known as a multidrug resistant uropathogenic group known to cause UTIs in adults . Its epidemiology in infants is unknown. The other possibility is E. coli expressing the Dr adhesins that are usually associated with pregnancy related UTIs may have crossed into the infant population . We are currently initiating studies of stored E. coli isolates to investigate this clonal possibility.
Another potential contributor is the lower rate of circumcisions over the past several years, which would increase a patient pool at higher risk of UTI , , . NC Medicaid stopped paying for circumcision permanently starting November, 2002 with a resulting drop in the circumcision rate from 74% in 2000 to 40% in 2004 . While this remains a possible explanation, it is noteworthy that male to female distribution of SBIs did not change significantly over the 10-year period.
Another possible explanation is sampling error and/or selection bias. We selected our patient population from those infants seen by physicians in the ED who had a blood culture drawn for any reason. It is possible that there were infants who presented to our ED with FWLS who did not have a blood culture drawn and if these infants did not subsequently develop an SBI that would mean our rates are artificially high. We also noted that there were more infants excluded in the early period compared to the later period for insufficient data. This is likely a reflection of the transitional years from a combined paper and electronic record to the current electronic one, but that difference could not explain the differences between the two periods.
Regardless of the reason, our results emphasize the need to sample urine on every infant less than 90 days of age. While practitioners are doing a better job at collecting urine studies, even at 93.4% of patients with a urine evaluation, with a 12.2% UTI rate in the 2002–2006 period out of the 361 FWLS infants, 2–3 UTIs were potentially missed.
The second important issue our results raise relates to antibiotic resistance. Ampicillin, in combination with gentamicin, has been traditionally used in the younger infants for its efficacy against Listeria, Enterococcus, and GBS. In 1999, Sadow et al found that the majority of pathogens were gram negative bacteria with high rates of ampicillin resistance . They observed no cases of Listeria over a 4 year period. In 2003, Byington et al found similar results over a 3-year period . In our study, 80% of disease was caused by E. coli with high rates of ampicillin resistance. We would argue that empiric antibiotics should cover the most likely invasive pathogen and appropriately account for the risk of antibiotic resistance. The vast majority of isolates causing UTI would no doubt respond to a regimen of ampicillin and gentamicin given the latter's utility in treating UTI. However, 8–9% of the SBIs in these infants would be a case of bacteremia or meningitis with an ampicillin resistant E. coli (5/52 from 2002–2006, 6/70 overall), a situation where most practitioners would not consider what effectively would be gentamicin monotherapy as appropriate. As a comparison, vancomycin plus cefotaxime/ceftriaxone is recommended in S. pneumoniae meningitis where the risk of cefotaxime/ceftriaxone non-susceptibility is approximately 7% . We would suggest that careful attention be paid to local resistance rates in this infant population, and perhaps in a setting with resistance rates as high as ours, an empiric regimen of ampicillin and cefotaxime would appropriately cover virtually all likely pathogens in this patient population.
Our study also shows that practitioners perform much less aggressive evaluations on older infants. At first glance this may seem appropriate since these patients are old enough to be stratified as HR or LR and, if LR, receive a more limited evaluation. However, when the older infants were analyzed by risk according to the Baraff guidelines , we found that only 66.9% of HR patients received an LP, suggesting that guidelines are only inconsistently followed. Given the concern that UTI in this age group puts infants at risk for meningitis and our own observation that one of our infants was diagnosed with concomitant UTI and meningitis, it still seems reasonable that all high-risk patients should receive an LP.
Our findings should be analyzed in light of certain limitations. This is a retrospective study from a single center, and the volume and patient population seen in our ED may not be generalizable to larger practice settings. This study should be repeated at other institutions to see if the increase in UTIs is present. Our inability to determine the number of febrile children that did not receive a blood culture in the ED means that our frequencies of FWLS may be falsely high and our total rates of SBI also may be falsely high. This, in turn, limits our ability to draw comparisons with other studies, such as the Utah study, where data was collected prospectively on all infants with fever . There are limitations with applying criteria retrospectively since some classification data may not have been recorded in the chart.
In this study, we have observed a significant increase in the number of SBI in infants with FWLS. These were primarily UTIs and the reasons for this are not yet clear. These results reinforce the need to examine the urine in all of these infants regardless of risk stratification and age. Our results also further suggest a move away from the combination of ampicillin and gentamicin as primary empiric therapy. The data from this study also may assist to streamline the management of these infants by helping to delineate risk differences in older versus younger infants.
The authors would like to thank the following individuals for their assistance in the planning and execution of this project: Dalene Stangl PhD, Stanley Mirret MS, Lizzie Harrell PhD, Pat Seed MD, PhD, Kammy McGann MD, Michael Cohen-Wolkowiez MD and Elizabeth Dodds PharmD.
Conceived and designed the experiments: RJ. Performed the experiments: KMW EW. Analyzed the data: KMW EW RJ. Wrote the paper: KMW RJ.
- 1. Crain EF, Shelov SP (1982) Febrile infants: predictors of bacteremia. J Pediatr 101: 686–689.
- 2. Roberts KB, Charney E, Sweren RJ, Ahonkhai VI, Bergman DA, et al. (1983) Urinary tract infection in infants with unexplained fever: a collaborative study. J Pediatr 103: 864–867.
- 3. Marshall R, Teele DW, Klein JO (1979) Unsuspected bacteremia due to Haemophilus influenzae: outcome in children not initially admitted to hospital. J Pediatr 95: 690–695.
- 4. McGowan JE Jr, Bratton L, Klein JO, Finland M (1973) Bacteremia in febrile children seen in a “walk-in” pediatric clinic. N Engl J Med 288: 1309–1312.
- 5. Dagan R, Powell KR, Hall CB, Menegus MA (1985) Identification of infants unlikely to have serious bacterial infection although hospitalized for suspected sepsis. J Pediatr 107: 855–860.
- 6. Baskin MN, O'Rourke EJ, Fleisher GR (1992) Outpatient treatment of febrile infants 28 to 89 days of age with intramuscular administration of ceftriaxone. J Pediatr 120: 22–27.
- 7. Baker MD, Bell LM, Avner JR (1993) Outpatient management without antibiotics of fever in selected infants. N Engl J Med 329: 1437–1441.
- 8. Belfer RA, Gittelman MA, Muniz AE (2001) Management of febrile infants and children by pediatric emergency medicine and emergency medicine: comparison with practice guidelines. Pediatr Emerg Care 17: 83–87.
- 9. Pantell RH, Newman TB, Bernzweig J, Bergman DA, Takayama JI, et al. (2004) Management and outcomes of care of fever in early infancy. Jama 291: 1203–1212.
- 10. Young PC (1995) The management of febrile infants by primary-care pediatricians in Utah: comparison with published practice guidelines. Pediatrics 95: 623–627.
- 11. Jaskiewicz JA, McCarthy CA, Richardson AC, White KC, Fisher DJ, et al. (1994) Febrile infants at low risk for serious bacterial infection–an appraisal of the Rochester criteria and implications for management. Febrile Infant Collaborative Study Group. Pediatrics 94: 390–396.
- 12. (2005) Early-onset and late-onset neonatal group B streptococcal disease–United States, 1996–2004. MMWR Morb Mortal Wkly Rep 54: 1205–1208.
- 13. (2007) Perinatal group B streptococcal disease after universal screening recommendations–United States, 2003–2005. MMWR Morb Mortal Wkly Rep 56: 701–705.
- 14. Zangwill KM, Schuchat A, Wenger JD (1992) Group B streptococcal disease in the United States, 1990: report from a multistate active surveillance system. MMWR CDC Surveill Summ 41: 25–32.
- 15. Byington CL, Rittichier KK, Bassett KE, Castillo H, Glasgow TS, et al. (2003) Serious bacterial infections in febrile infants younger than 90 days of age: the importance of ampicillin-resistant pathogens. Pediatrics 111: 964–968.
- 16. Sadow KB, Derr R, Teach SJ (1999) Bacterial infections in infants 60 days and younger: epidemiology, resistance, and implications for treatment. Arch Pediatr Adolesc Med 153: 611–614.
- 17. Baraff LJ (2000) Management of fever without source in infants and children. Ann Emerg Med 36: 602–614.
- 18. Brophy JM, Joseph L (1995) Placing trials in context using Bayesian analysis. GUSTO revisited by Reverend Bayes. JAMA 273: 871–875.
- 19. Rentz AC, Samore MH, Stoddard GJ, Faix RG, Byington CL (2004) Risk factors associated with ampicillin-resistant infection in newborns in the era of group B streptococcal prophylaxis. Arch Pediatr Adolesc Med 158: 556–560.
- 20. Alarcon A, Pena P, Salas S, Sancha M, Omenaca F (2004) Neonatal early onset Escherichia coli sepsis: trends in incidence and antimicrobial resistance in the era of intrapartum antimicrobial prophylaxis. Pediatr Infect Dis J 23: 295–299.
- 21. Schrag SJ, Hadler JL, Arnold KE, Martell-Cleary P, Reingold A, et al. (2006) Risk factors for invasive, early-onset Escherichia coli infections in the era of widespread intrapartum antibiotic use. Pediatrics 118: 570–576.
- 22. Hyde TB, Hilger TM, Reingold A, Farley MM, O'Brien KL, et al. (2002) Trends in incidence and antimicrobial resistance of early-onset sepsis: population-based surveillance in San Francisco and Atlanta. Pediatrics 110: 690–695.
- 23. Glasgow TS, Young PC, Wallin J, Kwok C, Stoddard G, et al. (2005) Association of intrapartum antibiotic exposure and late-onset serious bacterial infections in infants. Pediatrics 116: 696–702.
- 24. Bizzarro MJ, Dembry LM, Baltimore RS, Gallagher PG (2008) Changing patterns in neonatal Escherichia coli sepsis and ampicillin resistance in the era of intrapartum antibiotic prophylaxis. Pediatrics 121: 689–696.
- 25. Johnson JR, Russo TA (2002) Uropathogenic Escherichia coli as agents of diverse non-urinary tract extraintestinal infections. J Infect Dis 186: 859–864.
- 26. Watt S, Lanotte P, Mereghetti L, Moulin-Schouleur M, Picard B, et al. (2003) Escherichia coli strains from pregnant women and neonates: intraspecies genetic distribution and prevalence of virulence factors. J Clin Microbiol 41: 1929–1935.
- 27. Zorc JJ, Levine DA, Platt SL, Dayan PS, Macias CG, et al. (2005) Clinical and demographic factors associated with urinary tract infection in young febrile infants. Pediatrics 116: 644–648.
- 28. Wiswell TE (1992) John K. Lattimer Lecture. Prepuce presence portends prevalence of potentially perilous periurethral pathogens. J Urol 148: 739–742.
- 29. Singh-Grewal D, Macdessi J, Craig J (2005) Circumcision for the prevention of urinary tract infection in boys: a systematic review of randomised trials and observational studies. Arch Dis Child 90: 853–858.
- 30. Circumcision Reference Library (2004) Incidence of Circumcision by State. http://www.cirp.org/library/statistics/USA/state-by-state/.
- 31. Centers for Disease Control (2008) Active Bacterial Core Suveillance Report, Emerging Infections Program Network, Streptococcus pneumoniae, 2007. http://www.cdc.gov/abcs/reports-findings/survreports/spneu07.html.