Are Complete Blood Cell Counts Useful in the Evaluation of Asymptomatic Neonates Exposed to Suspected Chorioamnionitis?
Objective. Chorioamnionitis complicates 1% to 10% of pregnancies and increases the risk of neonatal infection. Women with chorioamnionitis receive intrapartum antibiotics, often resulting in inconclusive neonatal blood cultures. Peripheral neutrophil values are used frequently to assist in the diagnosis of neonatal infection and to determine duration of antibiotics; we sought to determine the utility of this approach.
Methods. A prospective observational study was performed in 856 near-term/term neonates who were exposed to suspected chorioamnionitis. Each received antibiotics for 48 hours unless clinical infection or positive blood cultures occurred. Peripheral neutrophils were measured serially and analyzed using the reference ranges of Manroe et al; an additional analysis of only the initial neutrophil values used the normal ranges of Schelonka et al. Results of neutrophil analyses were not used to determine duration of therapy. Fifty percent of asymptomatic neonates were seen postdischarge to ascertain recurrent infection. Local patient charges were examined.
Results. Ninety-six percent of neonates were asymptomatic and had negative cultures, and antibiotics were discontinued at 48 hours. A total of 2427 neutrophil counts were analyzed. Although abnormal neutrophil values were more frequent in infected or symptomatic neonates, 99% of asymptomatic neonates had ≥1 abnormal value. The specificity and negative predictive values for abnormal neutrophil values ranged between 0.12 and 0.95 and 0.91 and 0.97, respectively; sensitivity was 0.27 to 0.76. Significant differences in interpretation of the initial neutrophil values were noted, depending on the normal values used. Follow-up was performed for 373 asymptomatic neonates until 3 weeks’ postnatal age. Eight required rehospitalization; none had evidence of bacterial infection. If neutrophil values had been used to determine duration of antibiotics, then local costs would have increased by $76 000 to $425 000 per year.
Conclusions. Single or serial neutrophil values do not assist in the diagnosis of early-onset infection or determination of duration of antibiotic therapy in asymptomatic, culture-negative neonates who are ≥35 weeks’ gestation and are delivered of women with suspected chorioamnionitis.
- intraamniotic infection
- neutrophil values
- complete blood count
- early-onset infection
- antibiotic therapy
- length of stay
- resource utilization
Intrapartum maternal fever ≥38°C has been reported to occur in 1% to 10% of pregnancies, resulting in a diagnosis of suspected chorioamnionitis in 40 000 to 400 000 pregnant women a year in the United States.1,2 The term “chorioamnionitis” typically has a clinical connotation (maternal fever during labor, with or without uterine tenderness, leukocytosis, and/or fetal tachycardia), although histologic, microbiologic, and biochemical qualifiers may apply in selected studies.3,4 Intrapartum antibiotic therapy is given to these women to reduce the occurrence of neonatal pneumonia or sepsis,1,2,5–10 and postnatal antibiotic therapy may be initiated in these neonates.11,12 The 2002 Centers for Disease Control and Prevention (CDC) guidelines13 recommend the following: “If a woman receives intrapartum antibiotics for treatment of suspected chorioamnionitis, her newborn should have a full diagnostic evaluation and empiric therapy pending culture results, regardless of clinical condition at birth, duration of maternal antibiotic therapy before delivery, or gestational age at delivery.” These guidelines recognize that individual and institutional practices may vary, but the CDC does recommend that a complete blood cell count (CBC) be a part of the evaluation. Importantly, when signs of infection are absent and the blood culture is negative, there is a lack of consensus regarding the appropriate tests needed to establish a diagnosis of neonatal infection.14 Thus, the appropriate duration of antibiotic treatment in these asymptomatic neonates remains unclear.11,15,16
Although various acute-phase reactants have been used to assist in determining the presence or absence of neonatal infection,17–23 the CBC with differential white blood cell count is readily available and allows calculation of the absolute total neutrophil count (ATN), the total immature neutrophil count (ATI), and the ATI:ATN proportion (I:T), as determined by Manroe et al.24 In previous studies, the primary role of neutrophil values has been to assist the clinician in determining the absence of infection, especially in the presence of intrapartum antibiotic therapy when negative neonatal blood cultures may be misleading.24,25 Although some11,26,27 have questioned the applicability of these neutrophil parameters, they offer hour-specific reference ranges and remain widely referenced.28ndash;30
In our institution, because of evidence supporting a delayed response in neutrophil aberrations,31 we obtained serial CBCs during the initial 24 hours of antibiotic therapy. These results were used in conjunction with the clinical course and blood culture results in suspected chorioamnionitis-exposed neonates to assist in determining the absence or presence of presumed infection and thus the duration of antibiotic therapy. Neonates without clinical signs of infection and with normal serial neutrophil values received 48 hours of antibiotic therapy. When neutrophil values were outside the reference ranges24 on 2 or more CBCs in asymptomatic chorioamnionitis-exposed neonates, antibiotic therapy generally was continued for 4 to 7 days. However, maternal fever is known to alter neonatal neutrophil values, and scant published data exist to support either the use of neutrophil values to predict infection in this specific population of neonates32 or the extension of antibiotic therapy on the basis of that prediction.33
The purpose of this study was to determine the utility of neutrophil values, using the Manroe et al reference ranges24 and the normal values reported by Schelonka et al,26 in the evaluation of asymptomatic and symptomatic near-term and term neonates who were born to women who received intrapartum antibiotics for chorioamnionitis. The secondary goals of the study were to determine the adequacy of 48 hours of antibiotic therapy and whether the use and interpretation of neutrophil values might affect the duration and cost of hospitalization.
This prospective observational study was conducted in the normal newborn nursery (NBN) at Parkland Memorial Hospital (Dallas, TX) between April 15, 1998, and January 31, 1999. The delivery population at Parkland is composed primarily of medically underserved Hispanic women who have access to prenatal care in affiliated outlying clinics. Neonates who are admitted to the NBN must have a birth weight ≥2100 g and gestational age ≥35 weeks. Gestational age was determined by the best obstetric estimate; however, when the difference between this estimate and the Ballard examination34 exceeded 2 weeks, the latter was used. Maternal chorioamnionitis was defined as an intrapartum temperature ≥38°C; other findings such as fetal tachycardia and uterine tenderness were observed in some cases but were not required to make the diagnosis.35 During the present study, only the diagnosis of suspected clinical chorioamnionitis was used. Neonates were considered eligible for the study when the obstetric diagnosis of suspected chorioamnionitis was made, intravenous ampicillin and gentamicin were initiated before delivery, and the neonate was admitted to the NBN (neonates who were admitted to the neonatal intensive care unit were excluded). During the study period, antenatal urine cultures were obtained routinely, and intrapartum group B streptococcus (GBS) prophylaxis with ampicillin was administered using maternal risk factors other than fever (gestational age <37 weeks, rupture of membranes >18 hours, maternal history of a previous GBS-infected neonate, and/or positive urine culture for GBS). When any of these variables were present before the diagnosis of chorioamnionitis was made, ampicillin therapy was initiated before the addition of gentamicin. In the event of suspected penicillin allergy, clindamycin was substituted for ampicillin.
After admission to the nursery, a blood culture and CBC were obtained by venipuncture within 1 hour of birth (1), and CBCs were repeated (usually by heel-stick) at ∼12 (2) and 24 (3) hours of age. The white blood cell count was measured using a Coulter STK-S machine (Beckman Coulter, Inc, Fullerton, CA) and corrected for nucleated red blood cells; differential white blood cell counts were performed manually by trained technicians who achieved College of American Pathologists’ quality control standards. Individual neutrophil values (ATN, ATI, and I:T) were analyzed using the hour-specific reference ranges described by Manroe et al,24 as well as the normal values reported by Schelonka et al26 for neonates at 4 hours of age; each value was designated as either normal or abnormal by each set of criteria. Because of the unproved utility of neutrophil values in this clinical setting, these values were not used to make therapeutic decisions in study neonates. Maternal and neonatal demographics were collected prospectively from the respective medical records.
Study neonates were divided into 2 groups on the basis of the presence or absence of clinical signs of infection during their hospitalization (Fig 1). These included respiratory distress (tachypnea or dyspnea), cyanotic or apneic episodes not associated temporally with a feeding, persistently abnormal rectal temperature (<36.5°C or >37.5°C), hypotonia, poor feeding, or persistent unexplained hypoglycemia. Neonates with these problems are treated routinely within the NBN, unless the neonate’s condition warrants continuous monitoring and transfer to the neonatal intensive care unit.
Symptomatic neonates were presumed to have infection. These neonates and those with a positive blood culture were treated with intravenous antibiotics for 4 to 7 days regardless of neutrophil values and were evaluated by their primary care provider within 1 to 2 weeks of age, according to our customary care practice. Transient hypoglycemia in the absence of other signs, transient tachypnea of the newborn lasting <6 hours, and/or temperature instability as a result of environmental alterations occurred in 42 neonates; they were grouped with the asymptomatic neonates. When the blood culture remained negative, asymptomatic neonates were treated for 48 hours, regardless of neutrophil values, with intramuscular ampicillin (150 mg/kg/day divided every 8 hours) and gentamicin (5 mg/kg/day divided every 12 hours).
For determining whether asymptomatic chorioamnionitis-exposed neonates developed subsequent complications after 48 hours of antibiotics, they were grouped according to I:T values as shown in Fig 1. All asymptomatic neonates with an abnormal I:T value on either the second or third CBC were examined within 10 days of discharge by 1 of the investigators, and the parents were contacted by telephone within 3 weeks. To evaluate asymptomatic neonates with normal I:T values on the second and third CBCs, we contacted a convenience subset of this group (those with an even medical record number, 49%) by telephone within 1 to 2 weeks of discharge. Furthermore, all asymptomatic neonates were given appointments for routine follow-up care within 1 to 2 weeks of hospital discharge.
We compared the distribution of serial neutrophil values in symptomatic and asymptomatic neonates using the reference ranges of Manroe et al.24 In addition, we compared the percentage of abnormal initial neutrophil values using the data of Manroe et al24 and Schelonka et al26 for symptomatic and asymptomatic neonates.
To estimate local savings that would be realized if CBCs were not performed in asymptomatic neonates (assuming that the incidence of chorioamnionitis was 8.2%), we used daily charges for hospitalization and laboratory charges for 3 CBCs at Parkland Memorial Hospital based on the fee schedule for 2001, the most recent year for which complete data are available. The patient charge for a manual CBC was $40.28; the charge for the observation nursery in the NBN was $316.59 per day; additional charges (intravenous fluid, antibiotic use, etc) were not included in the calculations. Need for additional hospitalization of 2 to 5 days was assumed when any of the following was observed: ≥1 abnormal I:T values, ≥2 abnormal I:T values, or ≥4 of 9 abnormal neutrophil values; these cutoff levels were chosen arbitrarily to offer varied degrees of abnormality to calculate a range of savings estimates. In addition, using the distribution of initial I:T values in asymptomatic neonates, we estimated incremental hospital charges based on an I:T value ≥0.20.
Symptomatic and asymptomatic groups of neonates were compared using t test or Mann-Whitney rank sum test for continuous data and χ2 analysis for categorical variables. The Kolmogorov-Smirnov test was used to test for normality; statistics that were determined to be normal are displayed as mean ± standard deviation; those that were determined to be nonparametric are shown as median (range). All tests of statistical significance were 2-sided, with significance defined as α < .05. Sensitivity, specificity, and positive and negative predictive values for neutrophils relative to the development of symptoms were calculated according to standard formulas.36 This study was approved by the Institutional Review Board of the University of Texas Southwestern Medical Center; informed consent was not required.
A total of 856 neonates (8.2% of admissions to the NBN) were delivered of women with chorioamnionitis who received intrapartum antibiotics and were enrolled during the study period. Maternal and neonatal characteristics for symptomatic and asymptomatic neonates are presented in Tables 1 and 2, respectively. The median duration of rupture of membranes was significantly longer in the symptomatic group. The interval between initiation of intrapartum antibiotics and delivery and the percentage of women who received ≥2 doses did not differ between the 2 groups. The symptomatic neonates were more likely to require a brief period of bag-mask ventilation at delivery and, as expected, had longer hospitalizations; the median age of onset of symptoms was 16 hours.
A total of 2427 differential white blood cell counts were performed and analyzed at postnatal ages of 0.9 hours (median), 12.1 hours, and 25.9 hours. Of the 856 study neonates, 725 (85%) had 3 CBCs, 121 (14%) had only 2, and 10 (1%) had only 1. The absolute numbers and percentages of neutrophil values outside the reference ranges of Manroe et al24 for symptomatic and asymptomatic neonates are shown in Table 3. Although no statistical differences were observed for ATN1 and ATI1, more symptomatic neonates had an abnormal I:T1. ATI2, I:T2, ATI3, and I:T3 also assisted in distinguishing the symptomatic from asymptomatic neonates. Of note, 58% of the symptomatic neonates had a normal I:T1.
Of the 725 neonates in whom all 3 CBCs (9 neutrophil values) were obtained, at least 1 abnormal neutrophil value was observed in 97% (32 of 33) of symptomatic neonates and 99% (686 of 692) of asymptomatic neonates (P = .74). When extended to ≥4 abnormal values, 64% (21 of 33) of symptomatic neonates were identified versus 42% (290 of 692) of asymptomatic neonates (P = .02). Furthermore, at least 2 of 3 I:T values were abnormal in 36% (12 of 33) of the symptomatic neonates versus 9% of asymptomatic neonates (P < .0001). However, when only CBC #2 and #3 were analyzed, 89% of the symptomatic and 79% of the asymptomatic neonates had at least 1 abnormal neutrophil value (P = .20). Using the outcome of development of symptoms, we analyzed various combinations of neutrophil values (data available on request); no combination achieved a positive predictive value >42%.
Because some have proposed the use of a single CBC13,26 obtained at the time of initial evaluation, we compared the number of abnormal neutrophil values in symptomatic and asymptomatic neonates using the ranges of Schelonka et al26 and Manroe et al24 (Table 4). In the symptomatic neonates, significantly fewer abnormal ATI values were observed using the criteria of Schelonka et al,26 and there was a trend toward fewer abnormal I:T values. Conversely, in the asymptomatic neonates, significantly more abnormalities for all 3 neutrophil values were observed with the reference ranges of Manroe et al.24
Four (0.5%) of 856 neonates had a positive blood culture; their clinical characteristics and pathogens are described in the Appendix. Three were clinically asymptomatic after birth. When analyzed using the reference ranges of Manroe et al,24 multiple abnormal neutrophil values were observed on serial CBCs, particularly CBCs 1 (10 of 12 abnormal) and 2 (8 of 12 abnormal); overall, 75% (18 of 24) of values were outside the reference ranges. Of the I:T values, only 33% (4 of 12) were abnormal. When CBC 1 was analyzed using the criteria of Schelonka et al,26 only 3 of 12 neutrophil values (1 ATN and 2 I:T values) were outside the normal range (P = .01 compared with CBC 1 using the criteria of Manroe et al24). In 2 of these 4 neonates with positive blood cultures, all neutrophil values were normal when the criteria of Schelonka et al26 were used.
Of 404 asymptomatic neonates who met criteria for telephone follow-up, 373 (92%) were contacted 22 ± 14 days (mean ± standard deviation) after discharge. Eight (2%; all from the normal I:T group) required rehospitalization within 3 weeks of discharge for jaundice requiring phototherapy (n = 3), for hypocalcemic seizures (n = 1), or for fever with a negative sepsis work-up and antibiotic treatment for <3 days (n = 4). No other complications were reported or observed, and none had either presumed or proven bacterial sepsis.
Calculations performed to determine potential annual savings if serial neutrophil values were not used in the evaluation of asymptomatic neonates revealed that, in our institution, $311 000 to $425 000 would be saved each year (data available on request). If one were to examine the distribution of only initial I:T values (Table 5) among asymptomatic neonates and assume that values ≥0.20 indicate a need for treatment, then 17 to 113 neonates would have received 2 or 5 additional days of hospitalization. This would result in additional yearly charges of $11 449 to $183 425.
The approach to the treatment of neonates who are delivered of women with suspected chorioamnionitis is confounded by the routine administration of antibiotics at variable times before delivery, which could modify the neonate’s clinical presentation and/or incidence of positive blood cultures. The utility of neutrophil values in assisting to establish the diagnosis of infection and thus determining the duration of antibiotic therapy in this potentially large population of neonates is unclear.11,27,33,37 We have shown in asymptomatic, culture-negative near-term and term neonates who were exposed to maternal chorioamnionitis and intrapartum antibiotics that serial neutrophil values do not furnish useful ancillary information, provided that antibiotic therapy is initiated promptly after birth and continued for 48 hours in the absence of positive blood cultures. Ottolini et al14 drew similar conclusions in a population of asymptomatic term and near-term neonates with a variety of risk factors for neonatal sepsis. Furthermore, because our results demonstrate the frequent finding of abnormal neutrophil values in asymptomatic chorioamnionitis-exposed neonates, they suggest that reliance on neutrophil values alone may result in unnecessary prolongation of antibiotic therapy and hospitalization and a substantial increase in health care expenditures.
It should be noted that this study was conducted in a relatively high-risk population, and our findings may not be applicable to all settings. Furthermore, the diagnosis of suspected chorioamnionitis was based on clinical findings, although a previous study in our institution found that 80% of women with fever during labor had histologic evidence of chorioamnionitis.38 As clinicians, often we are limited to the single sign of intrapartum maternal fever, and we agree with the sentiments of Chen et al39 and others,40,41 who have noted the need for a better marker for clinically significant chorioamnionitis.
The Manroe et al reference ranges24 were generated in this institution; these neutrophil ranges are used in our nurseries, have the advantage of providing hour-specific ranges, are widely referenced, and are relevant for the examination of serial neutrophil values that were obtained in this study. Of the 3 neutrophil parameters, the I:T value is reported to have the highest sensitivity for determining the presence of early-onset neonatal infection24,25,42,43 and is used frequently to the exclusion of other parameters to support this diagnosis and direct the duration of antibiotic therapy. Although we found that an abnormal I:T value occurred more frequently in symptomatic neonates at each time period, elevated values on CBC 1 and CBC 2 were observed in 20% and 13%, respectively, of asymptomatic neonates (Table 3). To provide an alternative approach and to address the value of obtaining a single CBC, we analyzed the initial neutrophil values using the ranges of Schelonka et al,26 which were obtained in normal neonates at 4 hours of age. When considering the initial I:T value alone in asymptomatic neonates (Table 4), significantly more abnormal values were observed using the Manroe et al24 ranges, whereas neutrophil values analyzed by the Schelonka et al26 criteria were more likely to be normal in symptomatic neonates. Furthermore, when all 3 neutrophil values from CBC 1 were analyzed using the latter criteria, only 3 of 12 were abnormal in those with positive blood cultures, versus 10 of 12 using the Manroe et al24 reference ranges, raising important questions as to the sensitivity of the data reported by Schelonka et al26 in a chorioamnionitis-exposed population.
Pediatric providers frequently give antibiotics to neonates with sepsis risk factors because of the potentially devastating consequences of missing a neonate with sepsis, the conviction that postnatal antibiotic therapy has added to the reduction of neonatal sepsis already begun in the era of intrapartum antibiotics for GBS prophylaxis, and the inability of available laboratory tests to distinguish accurately the truly septic newborn. In the context of chorioamnionitis-exposed neonates, as well as those with other risk factors for GBS, the CDC has published an algorithm advocating the use of a single CBC to assist in decisions regarding the use of postnatal antibiotics.13 Our data and those of Ottolini et al14 suggest that the use of a single set of neutrophil values soon after birth has poor sensitivity and specificity. Furthermore, our data suggest that the clinician may be misled in the decision to implement antibiotic therapy for the chorioamnionitis-exposed neonate based primarily on a single set of neutrophil values shortly after birth.
In the 4 neonates with culture-proven sepsis, only 33% (4 of 12) of I:T values were abnormal within 24 hours of initiation of antibiotics. This suggests that 1) neutrophil responses to infection in these neonates may be corrected rapidly in the presence of postnatal antibiotic exposure and/or 2) the combination of intrapartum and early postnatal antibiotics ameliorates the disease course and neutrophil response. This combination may account for the infrequency of clinical signs of acute-onset infection in this small cohort of septic neonates, as well as the paucity of neonates with positive blood cultures. Similarly, in asymptomatic neonates in whom blood cultures were negative, maternal intrapartum antibiotics may have been sufficient to inhibit bacterial dissemination and, in combination with 48 hours of postnatal antibiotic therapy, seems to have provided adequate treatment.
It is unclear how long after birth the clinician should treat asymptomatic chorioamnionitis-exposed neonates whose mothers have received intrapartum antibiotics.12 In our study, these neonates received 48 hours of antibiotic therapy. Of the 373 neonates who were followed after discharge (46% of all asymptomatic neonates), only 8 (2%) were subsequently hospitalized during the ensuing 3 weeks. This readmission rate is consistent with published rates for a term uncomplicated newborn population,44 and none of these neonates had a documented bacterial infection. Because of resource limitations, we were unable to follow directly all asymptomatic neonates; however, it seems likely that this large number (n = 373) provides meaningful information regarding the entire population.
We then examined the local costs of the serial CBCs and additional hospital days that would have been incurred if we had used neutrophil values to extend the duration of antibiotic therapy to 4 to 7 days; ours was a conservative estimate of patient charges, because it did not include additional testing, such as examination of spinal fluid, or the cost of the antibiotics and intravenous fluids. At Parkland Memorial Hospital, the estimated annual savings ranged from $76 000 to $425 000, depending on one’s clinical approach. A savings analysis on a national level would be difficult, because the number of practitioners who follow the CDC guidelines and/or who use neutrophil reference ranges different from those of Manroe et al24 is unknown.
On the basis of this study, the current recommendations from the CDC, and consideration of our high-risk population, our approach to the care of a neonate who is delivered of a woman who is given intrapartum antibiotics for the treatment of suspected chorioamnionitis is to obtain a blood culture within the first hour after birth and promptly initiate antibiotic therapy without obtaining a CBC. This approach emphasizes the importance of clinical judgment as suggested by Escobar et al11,27; if the neonate remains asymptomatic for 48 hours and the blood culture is negative,45 then he or she can be discharged with a follow-up examination at the routine time interval as recommended by the American Academy of Pediatrics. Extended antibiotic therapy for 4 to 7 days is reserved for neonates with clinical signs of infection (eg, respiratory distress, feeding disorders, apnea, temperature instability) and/or those who have a positive blood culture within 48 hours. Assuming that practitioners comply with the CDC recommendation regarding empiric therapy, this approach to the chorioamnionitis-exposed neonate will result in the reduction of prolonged hospitalization based solely on an abnormal CBC result.
Presented in part at the Annual Meeting of the Pediatric Academic Societies; Boston, MA; May 15, 2000.
We thank Pam Ford, RN, Unit Manager; Susan Egwuagu, RN, Associate Unit Manager; Grace Joseph, RN, Associate Unit Manager; and the Parkland Memorial Hospital Newborn Nursery nursing staff for efforts and cooperation in collection of laboratory samples and data.
- Received June 19, 2003.
- Accepted January 5, 2004.
- Reprint requests to (G.L.J.) Department of Pediatrics, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390-9063. E-mail:
- ↵Benitz WE, Gould JB, Druzin ML. Antimicrobial prevention of early-onset group B streptococcal sepsis: estimates of risk reduction based on a critical literature review. Pediatrics. 1999;103(6) . Available at: pediatrics.org/cgi/content/full/103/6/e78
- ↵Escobar GJ, Li D, Armstrong MA, et al. Neonatal sepsis workups in infants ≥2000 grams at birth: a population-based study. Pediatrics. 2000;106 :256– 263
- ↵Benitz WE, Han MY, Madan A, Ramachandra P. Serial serum C-reactive protein levels in the diagnosis of neonatal infection. Pediatrics. 1998;102(4) . Available at: pediatrics.org/cgi/content/full/102/4/e41
- Martin H, Olander B, Norman M. Reactive hyperemia and interleukin 6, interleukin 8, and tumor necrosis factor-α in the diagnosis of early-onset neonatal sepsis. Pediatrics. 2001;108(4). Available at: pediatrics.org/cgi/content/full/108/4/e61
- Fowlie PW, Schmidt B. Diagnostic tests for bacterial infection from birth to 90 days—a systematic review. Arch Dis Child Fetal Neonatal Ed. 1998;78 :F92– F98
- ↵Galetto-Lacour A, Zamora SA, Gervaix A. Bedside procalcitonin and C-reactive protein tests in children with fever without localizing signs of infection seen in a referral center. Pediatrics. 2003;112 :1054– 1060
- ↵Avery GB, Fletcher MA, MacDonald MG. Neonatology: Pathophysiology and Management of the Newborn. Philadelphia, PA: JB Lippincott; 1999:1511
- Lilleyman J, Hann I, Blanchette V. Pediatric Hematology. Philadelphia, PA: Churchill Livingstone (Harcourt Brace); 1999:12
- ↵Remington JS, Klein JO. Infectious Diseases of the Fetus, Newborn, and Infant. Philadelphia, PA: WB Saunders; 1995:57
- ↵Browner WS, Newman TB, Cummings SR. Designing a new study: III. Diagnostic tests. In: Hulley SB, Cummings SR, eds. Designing Clinical Research: An Epidemiologic Approach. Baltimore, MD: Williams & Wilkins; 1988:87–96
- ↵Dashe JS, Rogers BB, McIntire DD, Leveno KJ. Time course of clinical and histologic intrauterine infection at term. Am J Obstet Gynecol. 1998;178 :S204
- ↵Engle WD, Rosenfeld CR, Mouzinho A, Risser RC, Zeray F, Sanchez PJ. Circulating neutrophils in septic preterm neonates: comparison of two reference ranges. Pediatrics. 1997;99 (3). Available at: pediatrics.org/cgi/content/full/99/3/e10
- ↵Braveman P, Egerter S, Pearl M, Marchi K, Miller C. Problems associated with early discharge of newborn infants. Early discharge of newborns and mothers: a critical review of the literature. Pediatrics. 1995;96 :716– 726
- ↵Mohamed WA, Cushion N, Siegel JD, et al. Discontinuation of antibiotics in neonates: when does the blood culture become positive? Pediatr Res. 1999;45 :271A
- Copyright © 2004 by the American Academy of Pediatrics