University of Michigan
Department of Pediatrics
Evidence-Based Pediatrics Web Site

Question

• Which components (if any) of the CBC with manual differential are predictive of whether a newborn infant has sepsis (bacteremia with systemic signs and symptoms)?

Clinical Bottom Lines

1. Neonatal sepsis occurs in 1 to 8 cases of all live births, but is 25 times more likely in low birthweight infants (<1500g) and 50-75 times more likely in very-low-birth-weight infants (500-1000g).¹
2. No one predictor of sepsis is both sensitive and specific. The most sensitive predictors are increased ratio of bands and either high or low total neutrophils. The most specific predictors are thrombocytopenia, neutrophilia or neutropenia or presence of >2 degenerative changes in the neutrophils on histology.²
3. Thrombocytopenia has by far the largest association with sepsis (by +LR) but is often a late marker.
4. Only the ratio of bands to segs (>0.3) has a significant positive and negative likelihood ratio.²
5. Other studies confirmed the finding of increased ratio of bands to segs (i.e. the left shift) as a positive and negative predictor of sepsis, but the association was weaker in these studies.^3,4
6. Earlier infections (at birth) tend to be due to maternally-transmitted organisms whereas later infections were nosocomial (usually due to coagulase-negative Staphyloccus ).^1,2

Summary of Key Evidence

1. A total of 298 evaluations were performed on 287 neonates. Most (243) were in the first 24 hours of life whereas the rest (55) were between 2-30 days. The distribution of term (130) and preterm (113) in the first group were roughly 50:50.
2. The "gold standard" was based on the "definite sepsis" (positive culture and symptoms/signs) group but not the "likely infection" (other evidence of sepsis) group.
3. Analysis was only performed with sensitivities and specificities given for the evaluations done on infants in the first 24 h as compared with the gold standard.
4. All the 7 criteria evaluated were combined to create a model for predicting sepsis based on additive scoring system (one point for each criteria). The vast majority of neonates however fell in the indeterminate 3-5 range.
5. The table shows sensitivities, specificities, and likelihood ratios for several tests:

Additional Comments

• The study did a poor job accounting for all the evaluations and patients. In addition, none of the sensitivity and specificity data for evaluations performed after one day of life were revealed.
• No single criteria or group of criteria had the net effect of ruling in or out a diagnosis.
• The organisms isolated from the blood cultures differed from the "usual offenders" for neonatal sepsis, casting some doubt on the gold standard.
• No long-term data (in terms of mortality outcomes) was reported.

Citation

1. Baley J, Goldfarb J. Neonatal Infections, Chapter 13 in Care of the High-Risk Neonate, 4th edition. MH Klaus and AA Fanaroff: editors. Philadelphia: WB Saunders Company, 1993.
2. Rodwell RL, Leslie AL, Tudehope DI. Early diagnosis of neonatal sepsis using a hematologic scoring system. Journal of Pediatrics 1988; 112: 761-777.
3. Kite P, Millar MR, Gorham P, Congdon P. Comparison of five tests used in diagnosis of neonatal bacteremia. Archives of Diseases of Children 1988; 63:639-643.
4. Berger C, Uehlinger J, Ghelfi D, Blau N, Fanconi S. Comparison of C-reactive protein and white blood cell count with differential in neonates at risk for septicaemia. Journal of Pediatrics 1995;154:138-144.

CAT Author: Cameron Dezfulian, MD

CAT Appraisers: Jonathan Fliegel , MD

Date appraised: January 10, 2001