Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Sepsis
Thomas B. Newman, Karen M. Puopolo, Soora Wi, David Draper and Gabriel J.
Escobar
Pediatrics 2010;126;903-909; originally published online Oct 25, 2010;
DOI: 10.1542/peds.2010-0935
The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://www.pediatrics.org/cgi/content/full/126/5/903
ABBREVIATIONS BACKGROUND: A complete blood count (CBC) with white blood cell
CBC—complete blood count
GBS—group B Streptococcus
differential is commonly ordered to evaluate newborns at risk for
WBC—white blood cell sepsis.
ANC—absolute neutrophil count OBJECTIVES: To quantify how well components of the CBC predict sep-
I/T—proportion of neutrophils that are immature
KPMCP—Northern California Kaiser Permanente Medical Care sis in the first 72 hours after birth.
Program METHODS: For this retrospective cross-sectional study we identified
BWH—Brigham and Women’s Hospital
LR—likelihood ratio
67 623 term and late-preterm (ⱖ34 weeks gestation) newborns from
ROC—receiver operating characteristic 12 northern California Kaiser hospitals and 1 Boston, Massachusetts
This work was presented at the annual meeting of the Pediatric hospital who had a CBC and blood culture within 1 hour of each other at
Academic Societies; May 2, 2009; Baltimore, MD. ⬍72 hours of age. We compared CBC results among newborns whose
www.pediatrics.org/cgi/doi/10.1542/peds.2010-0935 blood cultures were and were not positive and quantified discrimina-
doi:10.1542/peds.2010-0935 tion by using receiver operating characteristic curves and likelihood
Accepted for publication Jul 20, 2010 ratios.
Address correspondence to Thomas B. Newman, MD, MPH, RESULTS: Blood cultures of 245 infants (3.6 of 1000 tested newborns)
Department of Epidemiology and Biostatistics, University of were positive. Mean white blood cell (WBC) counts and mean absolute
California, San Francisco, Box 0560, San Francisco, CA 94143.
E-mail: newman@epi.ucsf.edu neutrophil counts (ANCs) were lower, and mean proportions of imma-
PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
ture neutrophils were higher in newborns with infection; platelet
counts did not differ. Discrimination improved with age in the first few
Copyright © 2010 by the American Academy of Pediatrics
hours, especially for WBC counts and ANCs (eg, the area under the
FINANCIAL DISCLOSURE: The authors have indicated they have
no financial relationships relevant to this article to disclose. receiver operating characteristic curve for WBC counts was 0.52 at ⬍1
hour and 0.87 at ⱖ4 hours). Both WBC counts and ANCs were most
informative when very low (eg, the likelihood ratio for ANC ⬍ 1000 was
115 at ⱖ4 hours). No test was very sensitive; the lowest likelihood ratio
(for WBC count ⱖ 20 000 at ⱖ4 hours) was 0.16.
CONCLUSION: Optimal interpretation of the CBC requires using inter-
val likelihood ratios for the newborn’s age in hours. Pediatrics 2010;
126:903–909
904 NEWMAN et al
Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on December 2, 2010
ARTICLES
WBC count ⫻ (% segmented neutro- areas under ROC curves in 3 time born by cesarean delivery, born in the
phils ⫹ % bands)/100. I/T was calcu- periods. earlier years of the study, had
lated as the total number of immature We estimated percentiles according to 5-minute Apgar scores of ⬍7, or had
neutrophils (promyelocytes, myelo- age for the WBC count, ANC, and I/T us- their CBCs measured at ⬍1 hour of
cytes, metamyelocytes, and bands) di- ing quantile regression (the qreg pro- age.
vided by the total number of cells in the cedure in Stata [Stata Corp, College The organism most commonly identi-
neutrophilic cell line (immature plus Station, TX]) for tests performed at fied in the positive cultures (Table 2)
segmented neutrophils). ⬍24 hours after birth, with terms for was GBS (56%), followed by Esche-
age, age squared, and linear splines richia coli (22%) and Enterococcus
Outcome Variables with knots at 3, 6, 9, and 12 hours. faecalis (4%). The proportion with GBS
We classified all blood cultures as pos- To estimate LRs for the WBC count, ANC, infections was lower at BWH (49% vs
itive or not without regard for CBC re- and I/T, we selected cutoffs that were 59%) and declined over the years of
sults by using an algorithm based pri- round numbers and that generated 5 the study, from 67% of infections to
marily on the organism risk category strata with approximately equal num- 37%. As a result, the proportion of in-
and the time to culture positivity.18 bers of cases. We chose this method fections caused by E coli increased
Three infants whose blood cultures rather than identifying inflection slightly (from 19% to 28%), although
only grew coagulase-negative Staphy- points in observed ROC curves to avoid the rate of E coli infections in each time
lococcus were included because they overfitting. We similarly stratified ac- period was similar.
had significant symptoms. Three blood cording to age at the time of sampling. Results for components of the CBC are
cultures were positive for more than 1 We estimated LRs as the proportion of shown in Fig 1. Because 95% of the in-
organism. bacteremia cases, with each result di- fants with infections had their CBCs
vided by the proportion of noncases measured at ⬍24 hours, we restricted
Statistical Analysis with that result.16 We performed all the figures to that time period. The fig-
We compared demographic and clini- analyses by using Stata/SE 11, supple- ures show a bimodal distribution in
cal characteristics of infants with and mented by a Stata algorithm that we timing of CBCs, with the first peak
without infection by using t tests or 2 developed to calculate interval LRs shortly after birth, and a second peak
tests as appropriate. To determine if (available from the authors). at ⬃4 hours. The median WBC count
adjusting the WBC count, ANC, and I/T and ANC values rise after birth, peak at
for other variables would improve dis- RESULTS 6 to 8 hours, and then decline slightly
crimination, we estimated predicted Of the 550 367 infants eligible for the during the next 18 hours. In contrast,
values of these tests using multiple lin- study on the basis of their hospital, the median I/T declines slightly and ap-
ear regression models on the infants year of birth, and gestational age, we proximately linearly during the first
who did not have infection. Models in- identified 311 (0.57 of 1000 live births) day. Among newborns with infection,
cluded age (using linear splines with with positive blood cultures as defined mean WBC counts were 29% lower,
knots at 3, 6, 9, 12, and 24 hours), age above. We included in this study the mean ANCs were 39% lower, and I/Ts
squared, birth weight (in 5 catego- subset of 67 623 infants (12.3% of the were 133% higher than in newborns
ries), gender, birth facility, year of 550 367 eligible newborns) who had a without infection; platelet counts did
birth, maternal preeclampsia, cesar- CBC performed within 1 hour of a blood not differ significantly.
ean delivery, and 5-minute Apgar score culture, including 245 of the 311 whose Because the WBC count, ANC, and I/T
(dichotomized at ⬍7). We subtracted blood culture was positive (3.6 of 1000 may be influenced by many factors
these predicted values from observed infants who received CBCs). other than sepsis, we created new
values and visually inspected receiver Characteristics of the newborns in- variables for each test that corre-
operating characteristic (ROC) curves cluded in the study and the propor- sponded to the difference between
for these residuals and raw test re- tions with documented infections are that test result and what would be pre-
sults to ensure that their slopes were shown in Table 1. The majority of in- dicted on the basis of other factors
monotonically decreasing. We then fants had white mothers, were born at available in this data set that were
compared the ability of the residuals Kaiser, were ⱖ39 weeks’ gestation, known or observed to influence test re-
with the raw test results to discrimi- and had their CBCs measured in the sults (see “Methods”). Resulting areas
nate between newborns who did and first 8 hours after birth. Infants were at under ROC curves, stratified according
did not have infections by comparing higher risk of infection if they were to age, are shown in Table 3. Overall
906 NEWMAN et al
Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on December 2, 2010
ARTICLES
908 NEWMAN et al
Downloaded from www.pediatrics.org. Provided by Indonesia:AAP Sponsored on December 2, 2010
ARTICLES
results of the CBC can provide more in- Among infants for whom a decision to ACKNOWLEDGMENTS
formation about the risk of sepsis after start antibiotics can be deferred until af- This work was funded by National Insti-
the first few hours after birth. If the in- ter 4 hours of age, the CBC is more likely tute of General Medical Sciences grant
tent of drawing a CBC is to use that infor- to be helpful. However, most of these in- R01-GM-80180-3.
mation to make clinical decisions re- fants will have CBC results with modest We thank Juan Carlos LaGuardia for cre-
garding the likelihood of infection, for LRs, in the range of 0.2 to 5. Thus, even ating a Stata ado file to calculate LRs,
example, whether to initiate empiric an- when the CBC is optimally interpreted, Sherian Xu Li for SAS programming,
tibiotic therapy, then delaying the CBC decisions about antibiotic treatment Amy Zolit for chart review and data ab-
for a few hours may be advisable. On should remain highly dependent on ma- straction, Manuel Chinchilla and Issa
the other hand, if an infant’s risk factors ternal risk factors and newborn symp- Alaweel for database construction and
or symptoms are sufficiently worrisome toms of infection.5 Future approaches to management, Paul Hughes, MBA, and
to draw a CBC and blood culture before the sepsis evaluation will require models Gregory Tomilonus for providing hospi-
⬃4 hours of age, it may be prudent to that are explicit about how multiple in- tal demographic information, and Stella
start antibiotics at the same time, rather formation sources (maternal risk fac- Kourembanas, MD, Steven A. Ringer, MD,
than waiting for the results of the initial tors, newborn clinical examinations, and PhD, and Dennis L. Kasper, MD, for their
CBC. laboratory test results) are integrated. ongoing support of our work.
REFERENCES
1. Schrag S, Gorwitz R, Fultz-Butts K, Schuchat born infants. Am J Clin Pathol. 1979;72(4): positive blood cultures: prospective valida-
A. Prevention of perinatal group B strepto- 608 – 611 tion of a multivariate algorithm. JAMA. 1992;
coccal disease. Revised guidelines from 10. Kayiran SM, Ozbek N, Turan M, Gurakan B. 267(14):1962–1966
CDC. MMWR Recomm Rep. 2002;51(RR-11): Significant differences between capillary 19. Greenberg DN, Yoder BA. Changes in the dif-
1–22 and venous complete blood counts in the ferential white blood cell count in screening
2. American Academy of Pediatrics. Practice neonatal period. Clin Lab Haematol. 2003; for group B streptococcal sepsis. Pediatr
guideline endorsement. Prevention of peri- 25(1):9 –16 Infect Dis J. 1990;9(12):886 – 889
natal group B streptococcal disease: re- 11. Chirico G, Gasparoni A, Ciardelli L, Marti- 20. Rozycki HJ, Stahl GE, Baumgart S. Impaired
vised guidelines from CDC 2002. Available notti L, Rondini G. Leukocyte counts in rela- sensitivity of a single early leukocyte count
at: http://aappolicy.aappublications.org/ tion to the method of delivery during the in screening for neonatal sepsis. Pediatr In-
misc/Prevention_of_Perinatal_Group_B. first five days of life. Biol Neonate. 1999; fect Dis J. 1987;6(5):440 – 442
dtl. Accessed December 7, 2009 75(5):294 –299
21. Ottolini MC, Lundgren K, Mirkinson LJ, Ca-
3. Schelonka RL, Yoder BA, desJardins SE, Hall 12. Schmutz N, Henry E, Jopling J, Christensen
son S, Ottolini MG. Utility of complete blood
RB, Butler J. Peripheral leukocyte count and RD. Expected ranges for blood neutrophil
leukocyte indexes in healthy newborn term count and blood culture screening to diag-
concentrations of neonates: the Manroe
infants. J Pediatr. 1994;125(4):603– 606 nose neonatal sepsis in the asymptomatic
and Mouzinho charts revisited. J Perinatol.
at risk newborn. Pediatr Infect Dis J. 2003;
4. Manroe BL, Weinberg AG, Rosenfeld CR, 2008;28(4):275–281
22(5):430 – 434
Browne R. The neonatal blood count in 13. Hasan R, Inoue S, Banerjee A. Higher white
health and disease. I. Reference values for blood cell counts and band forms in new- 22. Adair CE, Kowalsky L, Quon H, et al. Risk fac-
neutrophilic cells. J Pediatr. 1979;95(1): borns delivered vaginally compared with tors for early-onset group B streptococcal
89 –98 those delivered by cesarean section. Am J disease in neonates: a population-based
5. Escobar GJ, Li DK, Armstrong MA, et al. Neo- Clin Pathol. 1993;100(2):116 –118 case-control study. CMAJ. 2003;169(3):
natal sepsis workups in infants ⱖ2000 14. Weinberg AG, Rosenfeld CR, Manroe BL, 198 –203
grams at birth: a population-based study. Browne R. Neonatal blood cell count in 23. Benitz WE, Gould JB, Druzin ML. Risk factors
Pediatrics. 2000;106(2 pt 1):256 –263 health and disease. II. Values for lympho- for early-onset group B streptococcal
6. Christensen RD, Henry E, Jopling J, Wied- cytes, monocytes, and eosinophils. J Pedi- sepsis: estimation of odds ratios by critical
meier SE. The CBC: reference ranges for ne- atr. 1985;106(3):462– 466 literature review. Pediatrics. 1999;103(6).
onates. Semin Perinatol. 2009;33(1):3–11 15. Engle WD, Rosenfeld CR. Neutropenia in Available at: www.pediatrics.org/cgi/
7. Jackson GL, Engle WD, Sendelbach DM, et al. high-risk neonates. J Pediatr. 1984;105(6): content/full/103/6/e77
Are complete blood cell counts useful in the 982–986 24. Oddie S, Embleton ND. Risk factors for early-
evaluation of asymptomatic neonates ex- 16. Newman T, Kohn M. Evidence-Based Diagno- onset neonatal group B streptococcal
posed to suspected chorioamnionitis? Pedi- sis. New York, NY: Cambridge University sepsis: case-control study. BMJ. 2002;
atrics. 2004;113(5):1173–1180 Press; 2009:68 – 85 325(7359):308
8. Peevy KJ, Grant PH, Hoff CJ. Capillary venous 17. Lukacs SL, Schoendorf KC, Schuchat A. 25. Schuchat A, Zywicki SS, Dinsmoor MJ, et al.
differences in neonatal neutrophil values. Trends in sepsis-related neonatal mortality Risk factors and opportunities for preven-
Am J Dis Child. 1982;136(4):357–358 in the United States, 1985–1998. Pediatr In- tion of early-onset neonatal sepsis: a multi-
9. Christensen RD, Rothstein G. Pitfalls in the fect Dis J. 2004;23(7):599 – 603 center case-control study. Pediatrics. 2000;
interpretation of leukocyte counts of new- 18. Bates DW, Lee TH. Rapid classification of 105(1 pt 1):21–26