Objective: To determine the minimum volume of blood and the absolute n
umber of organisms required for detection of bacteremia and fungemia b
y an automated colorimetric blood culture system (BacT/Alert, Organon
Teknika). Design: Common neonatal pathogens, Escherichia coil, Strepto
coccus agalactiae (group B streptococcus (GBS): one American Type Cult
ure Collection (ATCC) strain and one clinical isolate), Staphylococcus
epidermidis, and Candida albicans, were seeded into blood to produce
bacteremia or fungemia with low colony counts (1 to 3 colony-forming u
nits (CFU) per milliliter) and ultra-low colony counts (<1 CFU/ml). Fo
r each organism, 96 culture bottles were inoculated with either 0.25,
0.5, 1.0, or 4.0 mi of the two seeded blood concentrations. Blood cult
ure bottles were incubated in the BacT/Alert device for 5 days, and ti
me to positivity was noted when applicable. All bottles were subcultur
ed on plated media. Data analysis: The Poisson statistic was used to c
alculate the probability of finding at least one viable CFU per inocul
ated culture bottle. The fraction of culture bottles with positive fin
dings per group was divided by the probability of one or more organism
s present to give the positivity index. Results: Plated subculture ide
ntified no growth of organisms not detected by the colorimetric detect
ion system. The false-positive rate for the automated device was less
than 1%. The positivity index for the GBS clinical isolate was 1.13, f
or the GBS ATCC isolate 0.96, for S. epidermidis 0.94, for C. albicans
0.97, and for E. coil 0.95. There was a statistically significant dif
ference with time to positivity and inocula volume (p < 0.01), but the
difference was not clinically important. Conclusions: If one or two v
iable colony-forming units are in the blood inoculated into culture me
dia, the BacT/Alert system will detect growth rapidly, Because there a
ppears to be a sizable subset of neonates who are at risk of sepsis wi
th a colony count less than 4 CFU/ml, then a 0.5 mi inoculum of blood
into the culture media is inadequate for sensitive and timely detectio
n of bacteremia. One to two milliliters of blood should increase micro
organism recovery in the face of low-colony-count sepsis.