Background. Neonatal sepsis is a low incidence, high-risk disease with many sepsis work-ups performed to detect a single case. Seventy-two hours of antibiotic therapy have been traditionally recommended pending negative culture results. Improved culture media and new technology integrated into blood culture systems could shorten incubation time required to detect positive culture results. This would then change the length of antibiotic therapy in the management of the newborn infant with suspected sepsis. In addition, previous data supporting the 72-hour recommendation were retrospectively acquired, utilized nonautomated systems, and reported in an era with a different population of microorganisms cultured in special care nurseries.
Objective. Evaluate the time of incubation to detect positive blood cultures from newborn infants with suspected sepsis using a computer-assisted, automated blood culture system, ESP (Trek Diagnostic Systems, Inc, Westlake, OH).
Design. Prospective, observational study.
Patients and Setting. All positive blood culture results that were obtained from term and preterm newborn infants born from November 1993 through June 1997 at a publicly funded hospital with over 6000 live births per year.
Methods. As positive blood culture results were identified, data were prospectively obtained from the patient's medical record. The computer algorithm in the automated blood culture system determined the time to positivity. Time to positivity was determined for blood cultures obtained before the initiation antimicrobial therapy and compared with those cultures obtained after beginning therapy. Time to positivity was also evaluated for clinically important Gram-positive and Gram-negative bacteria and yeast.
Results. Four hundred fifty-five positive blood culture results were obtained from 222 patients. Gram-positive organisms accounted for 80% (366/455) of the positive culture results, Gram-negative organisms accounted for 11% (48/455), and yeast for 9% (41/455). Virtually all cultures growing clinically significant Gram-positive and Gram-negative organisms were positive by 24 to 36 hours of incubation. Cultures growing Staphylococcus epidermidis were virtually all positive after 36 to 48 hours of incubation. Of cultures growing yeast, 88% (36/41) were positive by 48 hours of incubation. There was no difference in time to positivity in pretherapy or posttherpay obtained positive blood cultures. Prenatally administered antibiotics did not affect time to positivity in positive cultures drawn on the first day of life. In a selected group of microorganisms that are the frequent cause of bacteremia in term infants, 97% and 99% of cultures were positive by 24 to 36 hours of incubation when only pretherapy cultures are evaluated.
Conclusions. The ESP blood culture system identified 77%, 89% and 94% of all microorganisms at 24, 36, and 48 hours of incubation in aerobic cultures obtained from both term and preterm infants. Introduction of antimicrobial therapy did not affect time to positivity. Reducing duration of antibiotic therapy to 24 to 36 hours should be considered in term, asymptomatic newborn infants undergoing evaluation for suspected sepsis for maternal indications. Confirmation of similar rapidity of detection using other blood culture systems should be undertaken.
The early identification of septic infants is difficult before obvious symptoms are evident and clinical deterioration occurs. For this reason, when maternal risk factors for neonatal sepsis are present or the infant manifests symptoms suggestive of infection, cultures are obtained and antibiotic therapy begun.1–5However, because neonatal sepsis is a low incidence, high-risk disease,1 many work-ups are performed to detect a single case of septicemia. This results in additional days of hospitalization and antibiotic treatment for many newborn infants.
Because most infants with suspected sepsis, especially term infants without symptoms, are not infected and have negative blood culture results,6 48 to 72 hours of antibiotic treatment is recommended pending negative culture results.7 ,8 This length of time is based on retrospectively acquired data8–10 regarding time to positivity of blood cultures in newborn infants. These data were generated by using noninstrument-based manual methods (assessing blood culture turbidity)8 to determine positivity as well as methods which use carbon dioxide detection (Bactec 4609 and Bactec NR 66010).
Newer technology and improved culture media have been integrated into blood culture systems that reduce the time to detection of positive blood culture results. One computer-assisted, automated blood culture system, ESP (Trek Diagnostic Systems, Inc, Westlake, OH) is reported to more rapidly detect positive blood culture results in pediatric patients than other systems.11 ,12 Should these improvements shorten the time required for detection of positive cultures in newborn infants, changes in management of the newborn infant with suspected sepsis might be considered. The present study evaluated time to detect positivity in the blood cultures from newborn infants with suspected sepsis by the ESP blood culture system.
We prospectively evaluated all blood cultures in both term and preterm newborn infants from November 23, 1993 through June 30, 1997, in a large, publicly funded hospital with over 6000 deliveries per year. All blood cultures were drawn by physicians via venipuncture after iodine preparation of the skin. Blood volumes collected ranged from .5 to 1.0 mL and were inoculated into ESP 80A aerobic vials.
Blood Culture System
The ESP blood culture system is a computer-assisted, automated system based on the near continuous measurement (every 12–24 minutes) of the headspace pressure in each sealed blood culture bottle. The production or consumption of gases (oxygen, nitrogen, or carbon dioxide) by growing microorganisms is detected as a pressure change in the headspace of each bottle. The computer generates a curve of the headspace pressure plotted over time. If microorganisms are present, an increase or decrease in slope of this curve will be observed (Fig 1). This change is identified as positive for microorganisms by a computer algorithm that employs several preset criteria based on the rate of change in the headspace pressure. Both positive and negative rates of change are considered. Once the computer identifies a culture as positive, an audible and visual signal device located on the incubator cabinet alerts laboratory personnel. The computer-generated graph and time to positivity may be printed.
A representative from the microbiology section of the clinical laboratory provided the blood culture graph to one of the authors (J.G.P., V.S., and T.C.) of all positive blood culture results on patients on the neonatology service. Data were then collected from the patient's hospital chart which included: hospital number, gestational age by maternal dates, birth weight, race, gender, date of birth, location of work-up, maternal pretreatment, patient's age when the culture was obtained, and whether the culture had been obtained before or after the initiation of antimicrobial therapy. The computer-assisted blood culture system determined the incubation time to positivity.
In addition, each positive blood culture result was stratified as either drawn ≤72 hours of life (early-onset) or >72 hours of life (late-onset).
Organisms isolated from each positive blood culture were identified by the microbiology section of the hospital's clinical laboratory. Microorganisms were grouped according to the number of hours of incubation required for detection by the computer algorithm as well as by Gram-staining characteristics. Blood cultures yielding yeast were grouped separately.
Analysis of Data
To compare positive blood culture rates, the time to positivity of pretreatment and posttreatment blood cultures was used to determine a curve with 95% confidence intervals. To construct the curve for each group of cultures, the cumulative number of positive blood culture results at each 12-minute period of incubation was divided by the total number of positive cultures. This information was expressed as percent of cultures becoming positive versus time. For example, at zero time of incubation, 0% of cultures were positive (0/455), whereas at 148 hours 100% were positive (455/455). Confidence limits (95%) for proportions were determined for each time interval.14
The study was approved by the Baylor Affiliates Review Board for Human Subject Research.
During the 43-month study period, there were 23 078 live births of which 81% were term and 19% were preterm (<37 completed weeks). Approximately 8% of all newborn infants were evaluated for sepsis during this period. Four hundred and fifty-five blood cultures were reported as positive in 222 patients. Sixty-five percent of these newborn infants resided in the neonatal intensive care unit and 35% were located in the intermediate care nursery when their positive culture was drawn. The clinical characteristics of the infants with positive blood culture results are listed in Table 1.
The list of the microorganisms identified and the range of incubation times required for detection is shown in Table 2. At 24, 36, 48, and 72 hours of incubation, 77% (350/455), 89% (407/455), 93% (424/455), and 96% (437/455) of all cultures, respectively, were positive. This included cultures obtained after antimicrobial therapy had been initiated, cultures growing yeast, and cultures yielding microorganisms that are considered contaminants (eg, Corynebacterium sp,Propionibacterium sp, etc).
There was no difference noted in time to positivity whether the blood culture was obtained before or after initiating antimicrobial therapy (Fig 2). The 95% confidence limits for each curve varied by no more than 3% to 7% at the curves' points of greatest difference. It is for this reason that all cultures were included in the analysis regardless of whether they had been drawn preinitiation or postinitiation of antimicrobial therapy.
Virtually all cultures (97% or 118/122) growingStaphylococcus aureus, Enterococcus, Streptococcus agalactiae, Listeria monocytogenes, andStreptococcus pyogenes were positive by 24 to 36 hours of incubation. The 4 cultures that became positive after 36 hours of incubation were all obtained after antibiotics had been started (Table 2).
The most frequently cultured microorganism was Staphylococcus epidermidis, especially in very low birth weight infants (154 of 232 cultures growing S epidermidis). This microorganism is known to grow more slowly in culture than other Gram-positive organisms. Nonetheless by 36 to 48 hours of incubation, virtually all (95% or 220/232) cultures of S epidermidis had been detected.
By 24 hours of incubation, virtually all (34/39) cultures growingEscherichia coli, Morganella morganii,Klebsiella pneumoniae, Serratia marcescens,Enterobacter cloacae, and Pseudomonas aeruginosawere positive. Of the 5 cultures growing after 24 hours, 4 had been obtained after antibiotic therapy had been begun. The single preantibiotic culture that became positive after 72 hours grew K pneumoniae. We believe that this culture was not processed properly because a simultaneously drawn anaerobic culture grew K pneumoniae after 15.4 hours of incubation and repeat aerobic and anaerobic cultures drawn 2 days later were positive at 8 hours and 10 hours of incubation, respectively.
In those cultures growing yeast, 51% (21/41) were positive by 24 hours of incubation, 76% (31/41) by 36 hours, and 88% (36/41) by 48 hours of incubation. All the patients who grew yeast in their blood culture were <31 weeks' gestational age at birth.
Prenatally Administered Antibiotics
In 13 neonates with positive blood cultures drawn on the first day of life, predelivery antibiotics were administered to their mothers. Incubation time to positivity was <24 hours in 12 of the 13 positive cultures and all were positive by 36 hours of incubation (range: 7–34 hours; median: 12.6 hours).
We report the only prospectively designed study in a population of newborn infants dealing with incubation times required for blood cultures to become positive. In addition, this study assessed the largest number of blood cultures of any previous reports in a population of exclusively newborn infants8–10 using the technology available today—an automated, computer-assisted blood culture system. The spectrum of microorganisms included in our report more accurately reflects the current pathogens and nonpathogens present in newborn intensive care units and special care nurseries15–17 compared with previous reports. Past reports addressing time to positivity have been retrospective in design and have used culture media and methods for detection of microorganisms that have been supplanted.
The improvement in culture media and in the technology used for detection in these culturing systems is reflected in the fact that even cultures in our study that were obtained after the initiation of therapy (both prenatal and postnatal antimicrobial use) were detected just as quickly as those drawn before antimicrobials were started. On comparing the rates of positivity over time, both curves were nearly identical (Fig 2). Schelonka et al18 demonstrated that rapidity of detection in an automated blood culture system is directly correlated with the number of microbes in the sample cultured. However, the differences noted in detection time to positivity between high and low colony counts are not clinically significant. We speculate that if organisms are still present in the patient's blood stream and are part of the blood sample cultured, the current technology and culture media will rapidly identify them as positive even if the colony count is low. The rapid detection of positive blood cultures obtained after therapy has been initiated will alert the physician to the persistence of bacteremia or fungemia and allow for more rapid adjustments in dose, interval, or choice of antimicrobial therapy.
The list of microorganisms cultured in this population of patients more accurately reflects the current spectrum of bacteria and yeast causing infections,15–17 both early-onset and nosocomial acquired, in newborn infants than do earlier studies. In the reports of Pichichero and Todd8 in 1979 and Rowley and Wald9 in 1986, only 61/201 and 41/175 cultures grewS epidermidis and no yeast were cultured in either report. In 1989, Kurlat et al10 reported 27/98 cultures of S epidermidis and 9/98 cultures of yeast in their population of term and preterm infants. That study reflects the trend of the survival of more immature infants and their increased risk for infections with yeast and S epidermidis. Our nursery service yielded 41/455 cultures of yeast and 232/455 cultures of S epidermidis. Yet our rate of positivity at 24 and 48 hours is notably better (77% and 93%) compared with that in the report by Kurlat et al10(48% and 79%) when all cultures are considered (cultures growing pathogens, possible-pathogens, yeast, and contaminants). Our data reflect that the newer technology with improved media is faster in identifying a positive culture even with the inclusion of a high proportion of slow growing microorganisms (S epidermidis and yeast) as well as cultures obtained after antimicrobials had been started.
Pichichero et al8 reported that 96% and 98% of cultured pathogens were positive at 48 and 72 hours when they evaluated positive cultures obtained before antibiotics where begun. Current recommendations7 regarding length of antibiotic therapy pending negative cultures are based on this study. If we evaluate only pretherapy obtained positive cultures (Table 2) in the following group of microorganisms (S aureus, Enterococcus,S agalactiae, L monocytogenes, S pyogenes, E coli, M morganii, K pneumoniae, S marcescens, E cloacae, andP aeruginosa), 97% and 99% of these cultures are positive by 24 and 36 hours, respectively. It seems that virtually all the microorganisms of concern to pediatricians dealing with the term newborn at risk for early-onset sepsis are identified between 24 and 36 hours of incubation with this automated blood culture system using aerobic media. Because the risk of sepsis is quite low (.5%) and no cases of meningitis were reported in asymptomatic, term infants evaluated for sepsis for maternal indications,6 it may be reasonable to shorten the length of antibiotic therapy to 24 to 36 hours when cultures are processed using the ESP computerized, automated blood culture system. This shortened antibiotic course is especially practical when good follow-up is assured. This approach potentially saves a significant number of hospital days and doses of antimicrobial therapy administered particularly in locations where the rate of evaluation for suspected sepsis are as high as 20% of newborns.6 This approach reduces length of stay and the number of doses of antibiotics given and minimizes the time that a mother and her newborn infant are separated.
The newborn infants who remain in the hospital for over 72 hours are also at risk for infections because of slower growing microorganisms (ie, yeast and S epidermidis). Those positive blood cultures are usually detected at 36 to 48 hours of incubation. Consideration might be given to shorten the length of antimicrobial therapy to 48 hours because 95% of cultures identified with S epidermidisare positive by that time. Although this group of patients is also at risk for infection with yeast, antibiotic therapy will not treat a septicemia caused by this microorganism.
Length of treatment becomes an important issue when one considers the effects of antibiotic pressure on hospital flora in special care nurseries. Shortening the length of antimicrobial therapy could have a beneficial effect in this area.
We conclude that the ESP blood culture system identifies 77% of microoorganisms at 24 hours of incubation, 89% by 36 hours, and 94% by 48 hours of incubation in aerobic blood cultures obtained from term and preterm infants. Positive culture results obtained after initiation of antimicrobial therapy were identified as rapidly as cultures obtained before initiating therapy. When only common bacterial pathogens in early-onset infection in term infants are considered, all cultures but one (thought not to have processed properly) were identified as positive by 24 to 36 hours of incubation. Reducing duration of therapy to 24 to 36 hours should be considered in term, asymptomatic newborn infants undergoing evaluations for suspected infection for maternal indications. Reducing the duration of treatment of in-patients with suspected late-onset infection could also be considered, although there are a few cultures (5%) of S epidermidis not identified until after 48 hours of incubation. We cannot be certain that this rapid identification of positive blood culture results can be extrapolated to other automated blood culture systems and recommend that similar data be obtained for those systems.
This rapidly improving technology for quick and accurate detection of positive blood culture results continues to reduce the time required to identify positive blood culture results. These systems may serve as an impetus to reevaluate the recommendations for the length of antibiotic therapy for suspected sepsis with negative culture results.
We acknowledge and appreciate the advice and assistance of Leonard E. Weisman, MD, in the analysis of the data and review of this manuscript.
- Received January 21, 1999.
- Accepted June 25, 1999.
Reprint requests to (J.A.G-P.) Department of Pediatrics, Baylor College of Medicine, 1 Baylor Plaza, Houston, TX 77030. E-mail:
- Nyham WL,
- Fousek MD
- ↵Johnson CE, Witwell JK, Pethe K, Saxena K, Super DM. Term newborns who are at risk for sepsis: are lumbar punctures really necessary. Pediatrics. 1997;99(4). URL: http://www.pediatrics.org/cgi/content/full/4/e10
- ↵Mustafa MM, McCracken GH Jr. Perinatal Bacterial Diseases. In: Feigin RD, Cherry JD, eds. Textbook of Pediatric Infectious Diseases. Philadelphia, PA: WB Saunders; 1992:902
- Kurlat I,
- Stoll BJ,
- McGowen JE Jr
- Welby PL,
- Keller DS,
- Storch GA
- Welby PL,
- Keller DS,
- Ferrett RJ,
- Storch GA
- Klein JO, Marcy SM. Bacterial sepsis and meningitis. In: Remington JS, Klein JO, eds. Infectious Diseases of the Fetus and Newborn Infant. 2nd ed. Philadelphia, PA: WB Saunders; 1983:679–735
- ↵Zar JH. Biostatistical Analysis. Englewood Cliffs, NJ: Prentice Hall; 1984:378–379
- Philip AGS
- ↵St Geme JW, Harris MC. Coagulase-negative staphylococcal infection in the neonate. 1991;18:281–302
- Copyright © 2000 American Academy of Pediatrics