Time to Grow Positive Blood Cultures and Its Impact on Clinical Outcomes in Patients with Bacteremia Admitted to Intensive Care Unit

Bloodstream infections (BSIs) are one of the leading causes of mortality and morbidity due to infection in the USA [1]. Bacteremia may result from pneumonia, soft tissue or skin infection, urinary tract infection, urinary catheter related infection, or catheter related BSI. Blood cultures are cost effective measures for optimal management of septic shock [1, 2]. Time to positive blood culture in specimens taken concomitantly from catheters and peripheral blood help define if the catheter is the source of bacteremia [3]. It is plausible that early positivity (>2 h early) in specimens taken from a catheter suggests catheter is the source. Similarly, earlier growth of organisms in the specimen may suggest a bigger load of organisms in blood, which may be reflective of the severity of infection. Early detection of positive blood culture leads to early initiation of antimicrobial therapy, targeted therapy, or escalation or de-escalation of antimicrobial treatment. Therefore, these timings are important clinical parameters which may affect clinical outcomes. Giving early and correct antibiotics to patients with sepsis possibly within 1–3 h improves clinical outcomes [4]. Therefore, time to positivity (TTP) of culture in bacteremia is an important clinical variable which may impact outcome. Most studies are done on hospital population without focus on ICU. Most such studies also do not record and adjust many important parameters determining clinical outcomes as mostly conducted by infectious disease researchers or microbiologists. Clinical outcomes like mortality and length of stay in ICU (LOSICU) and length of stay in hospital (LOSH) are determined by multiple factors; age, gender, BMI, comorbid conditions [5], immune status of the patient, severity of the illness, virulence of the organism, resistance of the organism to antimicrobials, coinfection by multiple organisms, fluid therapy, corticosteroid use [6], glucose control, and adherence to protocols, including early goal-directed therapy and infection-control measures. Therefore, any study evaluating the relationship of TTP in bacteremia and clinical outcome must measure and adjust for all important clinical variables which is lacking in most clinical studies on this issue. We aim to study the impact of TTP of blood culture upon clinical outcomes while recording and adjusting for these important clinical confounding parameters by conducting a retrospective chart review, single-center clinical study in the ICU population.

Records of all positive blood cultures for 17 months duration August 1, 2017 to December 31, 2018 were reviewed. Primary outcome measures included in-hospital mortality, LOSICU, and LOSH.

The TTP were determined from the time interval between the start of incubation and the detection of microbial growth, as documented using an automated monitoring system (BD BACTEC^TM FX Blood Culture System – Becton Dickinson). When multiple cultures from the same time were positive, only the shortest TTP was selected for analysis. When a patient had multiple blood cultures at different times only the last incidence of bacteremia was included for analysis. Demographics (age, gender, BMI, and nationality), APACHE score for severity of illness, comorbid conditions (diabetes, hypertension, and COPD) type of blood culture (central catheter and cutaneous peripheral), results of respiratory culture, urine culture, catheter tip culture, antibiotic susceptibility pattern, start time of antimicrobial, use of mechanical ventilation (MV), presence of infiltrate on chest X-ray, usage of vasopressors, presence of renal failure, thrombocytopenia, and results of follow-up blood culture, were also recorded.

C-reactive protein (CRP), procalcitonin, and WBC count on the day of blood culture draw were recorded. Presence of immunosuppression from medication review (i.e., steroids) or from a pathology (leukemia) if present were also recorded. Other confounding variables include cancer, chemotherapy, prior use of antibiotics within 90 days of admission, need for MV, and need for renal replacement therapy.

LOSICU and hospital was measured from the day the blood culture was taken till the discharge of the patient. Ethical approval was approved by Dubai scientific Research Ethics Committee (DSREC-12/2018, 13 approved on January 13, 2019).

Dependent variables include, in hospital mortality, LOSICU, and Hospital (LOSH). Primary independent variable of interest was TTP of blood cultures. All other variables were used as confounding variables with possible impact on clinical outcomes.

χ² tests were performed to compare all categorical variable between the two groups (died vs. survived) (Table 1 upper panel). Data were not normally distributed. Median test was performed for all continuous variables to compare medians between groups (Table 1 lower panel). For the outcome of mortality, logistic regression was performed. For the outcomes of LOSICU and LOSH (numerical), we normalized the LOSICU and LOSH data by log transformation and then ran the least square regression on the transformed data.

Sample characteristics are presented in Table 1. Sample included 101 patients with 346 positive blood cultures (many patients had multiple occurrences of bacteremia) with a median age of 65 years and a median APACHE-2 score of 21 with an overall observed mortality of 61%. Median TTP was 20.2 h with quartiles cutoff; Q1 = 15.3, Q2 = 20.2, Q3 = 28, range 8–104 h. Patients who died were older, have significantly higher evidence for infiltrate on X-ray, MV, shock, use of antibiotics within 90 days of admission to ICU, and presence of repeated orders of blood cultures. Platelet count was significantly lower in those died.

For the outcome of mortality, logistic regression analysis was performed. We determined TTP does not predict mortality, only procalcitonin, CRP, WBC, shock, and MV predicted mortality (Table 2).

For the outcome of LOSICU, least square regression analysis on log transformed data showed TTP does not predict LOSICU. No other variable predicts LOSH (Table 3). Similar analysis for LOSH determines that TTP does not predict LOSH and no other variable predicts LOSH (Table 4).

Our analyses suggest that TTP of a blood culture for the single incidence of bacteremia is not a significant predictor for mortality, LOSICU, and hospital. The issue is complex though, as multiple blood cultures in a patient at different time frames may have a cumulative effect which complicates the model to assess this relationship. Cillóniz et al. [7] showed that TTP on blood cultures in pneumococcal pneumonia is a significant predictor of clinical outcomes. Martín-Gutiérrez et al. [8] showed higher mortality in patient who either grow blood organism <12 h or >27 h comparing the patient whose blood culture grow organism in 12–27 h. Another study showed that BSI with Pseudomonas with TTP ≤18 h is independently associated with mortality [9]. Kim et al. [10] studied TTP for Staphylococcus aureus and found that patients with TTP ≤12 and >48 h suffered the higher case-fatality rate than those with TTP (12–48 h). The reasons for the difference in results between our study and these studies are few; population studied as they studied hospital patients (ICU and non-ICU), while we studied bacteremia as a group regardless of organism in ICU population. Moreover, our patients have more than one episode of bacteremia and more likely have cumulative effects of these events.

Most studies addressing the impact of TTP on mortality were focused on specific organisms and sites of infection with less focus on clinical coordinates and other variables affecting mortality, that is, comorbidities like cancer, chemotherapeutic agents, immunosuppression, presence of MV, shock, or pressers which are significant determinants of mortality and LOS. For example, Khatib et al. [11] showed that patients with TTP for Staphylococcus aureus within 14 h are more likely to have endovascular infection sources and complications, although they did not control for the severity of the underlying illnesses. Our sample was an ICU population with assessment of clinical variables without single organism focus, a real-life phenomenon in ICU. To our knowledge, there is no study on ICU population addressing impact of TTP on clinical outcomes. The difference could be from different organisms in the sample. In our sample, Gram-negative bacteria were >50%, Gram positives were 31%, while 13% were fungi, prevalence of resistance to antimicrobials in our sample was 45% for GN, 38% for Gram positive, and 15% for fungi. More than 50% of blood cultures in our sample comprised following 4 organisms in decreasing order of frequencies; Klebsiella pneumoniae (20%) with a fatality rate of 55%, Candida species (14%) with fatality a rate of 92%, Escherichia coli (13%) with fatality rate of 69%, and Pseudomonas aeruginosa (9%) with fatality rate of 66.6% (Table 5).

We identify the following weakness of our study. It is a single-center retrospective chart review study with relatively small sample size, which may not be enough to exclude this association. This study was without specific organism focus, although it is the real-life scenario in ICU that our study specifically addressed, where bacteremia is a clinical finding regardless of the organism, and its impact on clinical outcome is addressed with control of significant coexisting clinical factors affecting clinical outcomes which are not addressed before in any study on ICU population with bacteremia.

TTP in bacteremia is an important clinical variable. TTP of bacteremia does not predict mortality. Larger studies are required to evaluate impact of TTP upon clinical outcome.

Ethical approval was provided by DSREC, DSREC-12/2018_03/approved on January 13, 2019.

There are no conflicts of interest.

The authors did not receive any funding.

R.N.: research idea conception, proposal writing, data collection, data analysis, and manuscript writing. A.H.: idea conception, proposal writing, and review of final manuscript. L.S.: idea conception and data collection. M.M.: idea conception and data collection. I.B.; data collection. S.K.: idea conception and data collection. A.S.: idea conception and data collection. R.A.: idea conception and data collection. Z.O.: idea conception and data collection. M.E.: idea conception and data collection.