Time to Blood Culture Positivity by Pathogen and Primary Service

This is a single-center retrospective cohort study performed at a freestanding, quaternary-care pediatric hospital. There are 444 licensed beds and ∼97 000 patient-days and 15 000 admissions annually. This freestanding pediatric hospital has a robust ASP that employs the Handshake Stewardship model,^10,11  including daytime reporting of positive blood culture results to providers by members of the stewardship team.¹²  Institutional review board approval for this study was obtained through our Multiple Institutional Review Board (19-1207), and the requirement of informed consent was waived.

We included all patients who had blood cultures drawn from January 2013 to December 2018. Demographic data (eg, date of birth, sex, immunocompromised status) and microbiologic data (eg, organism class, organism name, and culture collection date and time) were extracted from the Infection Control and Epidemiology data warehouse, which houses all patient-level data found in the electronic medical record (Epic Systems, Verona, WI). Only the first positive results on blood cultures drawn in hospitalized patients were included. A culture result was considered a “first” positive if it was a new organism or if 2 weeks had elapsed since the previous positive result. Polymicrobial cultures were excluded from analysis, as were blood cultures with an undefined source location (ie, not central line or peripheral) or held longer than the routine 5 days. Charts were reviewed within the electronic medical record for the culture date and time of sample collection, date and time that growth was first reported to providers, days of bacteremia, patient underlying medical conditions, and presence of a central venous catheter (CVC). Patients were considered immunocompromised if they had presence of either an immunocompromising condition (eg, cancer, bone marrow transplant, or solid organ transplant) or were receiving immunocompromising medications (eg, monoclonal antibodies, azathioprine, mycophenolate mofetil, calcineurin inhibitors) and were considered medically complex if they had one or more chronic medical condition with higher morbidity and mortality as well as increased health care use; patients could be categorized in >1 medical category (eg, medically complex and immunocompromised). To reflect true practice, we did not exclude patients with previous antimicrobial treatment.

Per hospital policy, our institution routinely draws ≥2 aerobic blood culture bottles before starting antibiotics, collecting ≥1 mL of blood with an additional 1 mL of blood for each year of age to a maximum of 10 mL. It is not policy (nor common practice) in children with central lines to also draw a peripheral blood culture. The laboratory used standardized blood culturing methods (plus Aerobic/F and PedsPlus/F [Becton, Dickinson and Co, Sparks, MD] bottles on a BacTec 9120/9240 automated continuous monitoring system [Becton, Dickinson and Co]). Routine blood cultures were incubated for 5 days. After detection of growth, a Gram-stain was performed and time stamped in the electronic medical record, and samples were processed by multiplex polymerase chain reaction blood culture identification panel (BioFire, Salt Lake City, UT). The culture was then subcultured to standard primary media plates, including Tryptic Soy Agar with 5% sheep blood, MacConkey agar, and chocolate agar (ThermoFisher Scientific, Waltham, MA). Depending on the Gram-stain results, selective and differential media were added for optimal organism isolation (eg, CHROMagar for the identification of methicillin-resistant Staphylococcus aureus [MRSA] and Columbia colistin nalidixic acid (CNA) agar for the isolation of Gram-positive organisms). Media plates were incubated at 35°C for 18 to 24 hours. Isolates that grew in subculture were identified by Matrix-Assisted Laser Desorption/Ionization-Time of Flight (Bruker Daltonics, Billerica, MA) and basic biochemical methods. Per laboratory protocol, repeat positive culture results within 4 days of first culture were only evaluated by Gram-stain to confirm previous identification unless a new morphology was present in the Gram-stain, which would prompt full identification.

Blood cultures were then categorized for analysis of the time to positivity (TTP). Microorganisms were categorized into Gram-positive-definite pathogens, Gram-negative-definite pathogens, Gram-positive-possible pathogens, yeast, common contaminants, and “other uncommon definite pathogens,” which comprises less frequently identified pathogens (eg, <30 blood cultures positive for that microorganism for the period of study).¹³  These groupings were determined a priori, and a list of organisms that made up each group is in Table 1. For commensal Gram-positives, possible pathogens were defined as those treated by the primary team with directed antimicrobial agents for ≥3 days, whereas the contaminants were defined as those that were either not treated or treated for <3 days. Blood cultures were also analyzed by subspecialties on the basis of the patient’s treatment team(s). Patients receiving primary care from multiple teams were attributed as such. For example, a gastroenterology patient in the PICU was attributed to both the gastroenterology subspecialty and the PICU; consequently, some blood cultures were counted twice in the treatment team analysis.

TTP was calculated on the basis of the date and time the culture was collected compared with the date and time the Gram-stain was first reported by microbiology to the provider. The number of patients who would need to be treated for 48 hours rather than 36 hours to prevent 1 case of antibiotic termination before positive result was calculated by subtracting the number of cultures with positive results at 48 hours by the number with positive results at 36 hours, dividing by the total number of cultures in the study period, and then calculating 1% of cultures with positive results between 36 and 48 hours. Box plots, t tests, and one-way analyses of variance were used to compare groups, and Kaplan-Meier curves were constructed (Minitab 19 Statistical Software, Minitab Inc: State College, PA).

Over a 6-year period, a total of 6184 blood cultures with positive results were identified of 89 663 total cultures (6.9%). Of the 6184 blood cultures with positive results, 4036 were excluded because of repeat growth (n = 3629), polymicrobial growth (n = 261), unidentified source (n = 164), and extended incubations (n = 9) (Fig 1). Patients with positive blood culture results were, on average, 6.2 years old, 1224 of 2121 (58%) were boys, 514 of 2121 (24%) were considered immunocompromised, and 1360 of 2121 (64%) were considered medically complex.

The blood cultures with positive results included 1454 (69%) definite or possible pathogens and 667 common contaminants (31%). Overall, the number and percentage of positive blood culture results at 24, 36, and 48 hours were 1441 of 2121 (68%), 1845 of 2121 (87%) and 1970 of 2121 (93%), respectively. Of all blood cultures obtained (89 663), only 125 cultures (0.14%, 66 pathogens, and 59 contaminants) yielded positive results between 36 and 48 hours, indicating that 719 patients would need to be treated for 48 hours rather than 36 hours to prevent 1 case of antibiotic termination before culture positivity. Specifically for Gram-positive-definite pathogens, the numbers and percentages of positive blood culture results at 24, 36, and 48 hours were 520 of 592 (88%), 567 of 592 (96%), and 580 of 592 (99%), respectively, and for Gram-negative-definite pathogens, the numbers and percentages of positive blood culture results at 24, 36, and 48 hours were 299 of 341 (88%), 317 of 341 (93%), and 323 of 341 (95%), respectively (Fig 2).

The median TTP (interquartile range [IQR]) for all included blood cultures (Table 1) positive for bacteria included in this study was 19.83 (14.4–26.0) hours, with a range of 1 to 118.4 hours. More specific analysis by pathogen category revealed the following TTPs: Gram-positive-definite pathogens (27.8%, n = 592) with a median (IQR) TTP of 15.93 (13.06–19.41) hours, Gram-negative-definite pathogens (16%, n = 341) with a median (IQR) TTP of 14.33 (11.60–18.35) hours, Gram-positive-possible pathogens (15.5%, n = 330) with a median (IQR) TTP of 20.74 (14.36–25.80) hours, common contaminants (31.7%, n = 667) with a median (IQR) TTP of 24.48 (21.19–32.27) hours, and yeast (2.58%, n = 55) with a median (IQR) TTP of 32.50 (20.08–44.54) hours. Microorganisms that were classified as “other uncommon definite pathogens” and treated as true pathogens grew in 6.57% (n = 136) of blood cultures with a median (IQR) TTP of 24.90 (17.67–39.07) hours (Fig 3).

Gram-positive-definite pathogens, such as methicillin-susceptible S aureus (MSSA) and MRSA, composed 12.6% (n = 269) and 2.6% (n = 55) of all blood cultures with first positive results, respectively. The median (IQR) TTP for these specific pathogens were 17.57 (15.10–21.95) and 17.98 (14.02–23.41) hours, respectively. Other Gram-positive-definite pathogens, such as Streptococcus pyogenes (β Streptococcus group A), Streptococcus agalactiae (β Streptococcus group B), Streptococcus pneumoniae, and Enterococcus species had median (IQR) TTPs of 13.37 (12.11–16.20), 12.28 (10.42–13.45), 14.33 (12.95–16.08) and 14.5 (12.20–17.02) hours, respectively (Fig 3). Gram-negative-definite pathogens, such as Klebsiella species, Pseudomonas species, Enterobacter species, Salmonella species, and Escherichia species had median (IQR) TTP blood cultures of 11.68 (9.76–14.40), 20.55 (17.40–22.79), 12.95 (11.24–14.40), 18.30 (16.51–22.50), and 13.38 (11.85–16.43) hours, respectively (Fig 3).

Of the 2121 patients with first positive culture results, 442 (21%) were hematology, oncology, or bone marrow transplant patients, 204 (10%) were gastroenterology patients, 328 (15%) were PICU patients, 296 (14%) were NICU patients, 184 (9%) were cardiac ICU or cardiac progressive care unit patients, and the remainder were patients admitted to the hospital medicine service. When comparing the TTPs for all treated pathogens between the NICU, PICU, and cardiac ICU or cardiac progressive care unit, there was no difference between groups (P = .30). Longer TTPs (medians) were observed in the NICU with MSSA, MRSA, and Streptococcus spp. TTPs were 22.60, 27.68, and 22.12 hours, respectively, and, for Pseudomonas aeruginosa in gastroenterology, the median TTP was 46.43 hours, although these increases in TTP were not statistically significant when subspecialty pathogen was compared with overall TTP for the pathogen. Additional analysis for select pathogens by subspecialty is presented in Fig 4.

The median (IQR) TTP for blood cultures based on source of blood was 18.0 (13.01–24.75) hours for CVCs and 21.14 (15.68–27.49) hours for peripheral cultures; this was not statistically different (P = .27). Median (IQR) for TTP when analyzed by those patients aged <30 days and those aged >30 days was 20.18 (14.47–27.1) and 20.68 (14.28–26.0), P = .86, respectively.

This study reveals that 36 hours is likely sufficient for observation of blood cultures in infants and children started on empirical antimicrobial agents for concern for bacteremia. We demonstrate that 87% of all positive blood culture results were positive by 36 hours (Fig 2) and that only an additional 125 cultures yielded positive results between 36 and 48 hours, highlighting that 719 patients would need to be treated for 48 hours rather than 36 hours to prevent 1 case of antibiotic termination before culture positivity. These data indicate that early de-escalation is appropriate in a clinically stable patient. Recognizing that early discontinuation can be associated with risks, including readmission, increasing severity of illness, and increased morbidity and mortality, safety netting with robust education, and strict return precautions is important in cases in which early discontinuation is employed.

Taken together, the median (IQR) TTP of all bacterial cultures during the 6-year period was 19.83 (14.4–26.0) hours; the median TTPs for Gram-positive-definite (15.93 hours) and Gram-negative-definite (14.33 hours) pathogens are <24 hours, and, although common contaminants were slightly longer (median TTP of 24.48 hours), this was not statistically significant. The longest median TTP in this analysis was 32.50 hours for yeast. This study also reveals that there is no statistical difference in TTP when assessing the source from which the blood was drawn (CVC versus peripheral, P = .27), or the treatment team (P = .30 for NICU, PICU, CICU or cardiac progressive care unit comparison) (Fig 4).

TTP has been studied before in several distinct populations, including adults, those with febrile neutropenia, neonates, and those with presence of CVCs. In all studies, the authors conclude that the typical 48- to 72-hour observation period for empirical antimicrobial agents can, in most cases, be shortened to 24 to 36 hours,^3,8,14,15  a finding that is consistent with the results of this study. These previous studies included comparatively small numbers of positive culture results. With this study, we wanted to address the common statement heard on stewardship rounds of “that study was not done in patients similar to mine.” Thus, we endeavored to include enough positive culture results across the whole hospital setting to evaluate by pathogen type and primary treatment team, an evaluation otherwise lacking in the literature. In clinical practice, these data by pathogen and primary treatment team will help providers deescalate or stop antimicrobial agents sooner in patients, thus limiting toxicity and other consequences of antimicrobial use.

Although the primary treatment team (eg, NICU, oncology, gastroenterology), or presence of a CVC are anecdotally thought to impact TTP and thus clinical decisions, this study does not endorse these beliefs. Furthermore, age less than 30 days was not associated with a longer TTP. Blood volume collected is also thought to impact the TTP,^16,17  but we were unable to assess this because we lacked confidence in the recorded blood volume data within our electronic medical record.

There are several limitations of this study. First, it is well recognized that TTPs of blood cultures are dependent on many factors, including the amount of blood in each blood culture bottle, previous antimicrobial administration, and time from specimen collection to receipt by the microbiology laboratory test result, as well as the concentration of microorganism within the blood.^13,18–20  We did not account for these factors, and, in particular, we did not exclude patients for previous receipt of antimicrobial agents. However, we believe including all patients is more reflective of the true clinical context and thus more applicable data. Second, repeat positive culture results within a 2-week period for the same organism were excluded. Although it is possible that the TTP may vary between repeat positive culture results because of differences in antimicrobial exposure or specimen volume, clinical decisions about antimicrobial coverage are generally made around the first positive result. Third, patients who had polymicrobial cultures were excluded, so these TTPs do not reflect patients who were infected with multiple pathogens; these cultures are being analyzed in a separate study. Fourth, there was no control for blood culture volume; although this information was originally collected and analyzed, the data were not presented because the validity of recorded volumes could not be confirmed. Fifth, standard blood culturing techniques were used, so these data may not be applicable to centers using alternative platforms. Sixth, TTP historically has been one factor considered when deciding if an organism should be called a contaminant or not, thus biasing the difference in TTP between organisms treated as true pathogens and contaminants.²¹  Last, these data are representative of only a single center; thus, they may not be applicable to other centers with other processes in place (eg, no on-site microbiology).

Despite these limitations, the TTP data from this large study suggests that the probability of true bacteremia after 36 hours is low enough that the number of children needed to be treated beyond that time window to prevent 1 case of premature termination is likely excessive. Future directions include analysis of these data in the context of polymicrobial cultures, addressing the impact of previous receipt of antimicrobial agents, and assessing the impact of practice change within our institution.

In this 6-year retrospective review, we examine TTP of blood cultures at a quaternary care pediatric hospital. Concern for infection often leads to the initiation of broad-spectrum empirical antibiotic coverage for a 48- or 72-hour observation period while awaiting negative culture results. Our study indicates a practice change to 36 hours is warranted in most circumstances, and, when implemented hospital wide, will save thousands of days of unnecessary antimicrobial therapy,²²  thus preventing related untoward effects.