Consequences of Implementing a “Better” Blood Culture System

Abstract

Background:

Blood cultures (BCx) are the gold standard for diagnosing blood stream infections. However, contamination remains a challenge and can increase cost, hospital days, and unnecessary antibiotic use. National goals are to keep overall BCx contamination rates to ≤3%. Our healthcare system recently moved to a BCx system with better organism recovery, especially for gram-negative, fastidious, and anaerobic bacteria. The study objectives were to determine the benefits/consequences of implementing a more sensitive blood culture system, specifically on contamination rates.

Methods:

The electronic health record was queried for all BCx obtained within our tertiary-care health system from April 2015 to October 2016. Cultures were divided into those obtained 12 months before and six months after the new system was introduced. A positive BCx was defined as one with any growth. Contaminated BCx were defined as those showing coagulase-negative Staphylococcus, Corynebacterium, Bacillus, Micrococcus, or Propionibacterium acnes. Cultures with Staphylococcus aureus, Klebsiella pneumoniae, or Escherichia coli were said to contain a true pathogen. Results based on hospital location of blood drawing also were determined.

Results:

A total of 20,978 blood cultures were included, 13,292 before and 7,686 after the new system was introduced. With the new system, positive BCx rates increased from 7.5% to 15.7% (p < 0.001). Contaminants increased from 2.3% to 5.4% (p < 0.001), and pathogens increased from 2.5% to 5.8% (p < 0.001). Contaminated BCx increased significantly in the surgical/trauma intensive care unit (STICU), emergency department (ED), and medical ICU (MICU), while pathogen BCx increased on the surgical floor, ED, and MICU.

Conclusions:

A new blood culture system resulted in significant increases in the rates of positive, contaminated, and pathogen BCx. After the new system, multiple hospital units had contamination rates >3%. These data suggest that a “better” BCx system may not be superior regarding overall infection rates. More research is needed to determine the impact of identifying more contaminants and pathogens with the new system.

Blood cultures remain the gold standard for diagnosing blood stream infections and continue to be an important diagnostic modality. Positive cultures allow identification of causative microorganisms and appropriate narrowing of antibiotic activity, in addition to monitoring infection rates at a systems level [1]. However, contamination of blood cultures remains a significant challenge. Contaminated cultures can result in greater lengths of stay (LOS), higher healthcare costs, need for repeat blood draws, and additional laboratory tests, needless antibiotic use, and unnecessary invasive procedures [2-6]. Contamination also results in a greater workload for nurses, phlebotomists, and laboratory technologists. Many studies have demonstrated the financial burden of contaminated cultures to be, on average, more than several thousand dollars per contamination episode [2-4,7,8]. Additionally, there is evidence that, given high rates of contamination, blood cultures obtained in the emergency department (ED) often are of little clinical value and rarely change management, yet falsely positive cultures can result in additional visits to the ED and unnecessary hospitalization [3,9-13].

National goals for rates of blood culture contamination are ≤3%, with rates in the literature ranging from <1% to more than 6% [1,7,14-16]. Contaminants can represent up to 50% of all positive cultures [4,6,12,14,16,17]. Emergency departments and academic centers have been reported to have rates of contamination >3%, usually as a result of specific issues inherent to these locations [2,7,8,14,15]. Additionally, contamination rates appear to be increasing, most likely because of changes such as new culture collection techniques, increased use of both central venous catheters and indwelling vascular devices (both in general and for obtaining cultures), and better blood culture identification systems [1,7,18]. Many studies have documented improvements in contamination rates by employing various interventions, such as retraining employees, using trained phlebotomists, opting for venipuncture over catheter blood harvest, ensuring an adequate volume of blood is drawn, changing skin antiseptic solution, standardizing or sterilizing collection kits, and implementing laboratory-based protocols to limit work-up of potentially contaminated cultures [1-3,5,8,15,18,19]. With advances in technology, newer automated culture systems can detect smaller numbers of organisms that may have been missed previously, and many systems demonstrate better staphylococci identification, including the most frequent contaminant, coagulase-negative staphylococci (CoNS) [7,8,16,18,19].

In April 2016, our healthcare organization changed to a new blood culture system, given its better overall performance in organism recovery, especially with gram-negative, fastidious, and anaerobic organisms. The new system also demonstrates a decreased time-to-detection, as determined by an in-house performance evaluation, as well as improved organism recovery even if cultures were obtained after the initiation of antibiotics. Compared with the old system, which required 5 mL of blood, the new system requires 10 mL for the sample and entailed an upgrade to a different set of blood culture bottles. Otherwise, there were no changes in collection technique or materials. There also were no other system-wide changes in practices surrounding blood cultures during this time.

After implementation of the new system, there was a concern about the potential increase in the number of contaminated blood cultures observed in the STICU. There also was concern about a higher number of duplicate cultures being obtained. Thus, we sought to determine the unintended consequences of implementing this new blood culture system in our hospital, especially regarding the rates of contaminated cultures and the number of repeat blood cultures obtained. We hypothesized that with the augmented sensitivity of the new system, rates of contamination and recognition of pathogens would both increase.

Patients and Methods

This study was approved by the MetroHealth Medical Center Institutional Review Board. It was an analysis of all blood cultures obtained from patients presenting to any of our health system locations between April 2015 and October 2016. Our health system has a variety of health care sites, ranging from community-based urgent-care facilities to our academic, tertiary-care main hospital. Blood cultures were identified by performing an electronic medical record (EMR)-based report, which included all blood cultures obtained, final culture results, and hospital location in which the blood was drawn. Any patient who had a blood culture during the study period was included in the final analysis. As the new blood culture system was implemented in April 2016, this month was considered the transition period, and blood cultures obtained during this month were excluded. Cultures were then divided into those obtained twelve months prior to and six months after the implementation of the new system.

Within our hospital system, blood cultures are collected by nurses and clinical care partners on the regular nursing floors and by nurses and paramedics in the emergency department. Chlorhexidine is used to prepare the skin. Aerobic bottles are filled first, followed by the anaerobic bottles. The desired fill volume for each bottle is 10 mL. The bottles are then inverted approximately five times and sent to the microbiology laboratory for follow-up.

A positive blood culture was defined as one with any growth on final culture. For the purposes of the study, a contaminated culture was one that grew CoNS, Corynebacterium, Bacillus spp., Micrococcus, or Propionibacterium acnes. These bacteria were designated contaminates because they are commensal organisms that are rarely pathogenic and because their presence in a blood culture usually is considered to represent contamination [1,6,14,16,18]. Cultures with Staphylococcus aureus, Klebsiella pneumoniae, or Escherichia coli were said to contain a true pathogen because the presence of these organisms in a blood culture almost always represents a true bacteremia [1,16,18]. The rates of blood cultures containing the eight micro-organisms of interest were evaluated before and after the new system was introduced.

Rates of positive, contaminated, and pathogen-containing blood cultures were examined across the healthcare system overall, as well as by location of blood draw. For our purposes, we evaluated cultures obtained in the STICU, MICU, burn ICU (BICU), surgical floor, and ED. For patients with a contaminated culture, the number of total cultures per patient, number of contaminated cultures per patient, and number of repeat cultures per patient were determined. This tally also was performed for patients with pathogen-containing cultures. Any additional culture obtained on a different day after a contaminated culture result was deemed to be a repeat or duplicate culture.

Statistical analysis was performed using IBM SPSS© version 24 (IBM, Armonk, NY). Categorical data were analyzed using either the χ² or Fischer exact test as appropriate. Continuous variables were compared using the Mann-Whitney U test given non-normal distribution. Data are reported as median (25^th–75^th percentile). A p-value of 0.05 was used to determine statistical significance.

Results

A total of 20,978 blood cultures were identified in 7,132 patients across the study period. There were 13,292 cultures in the 12 months prior to the new system and 7,686 in the six months after the new system was implemented.

System-wide, the rate of positive blood cultures increased significantly, from 7.5% of all blood cultures before the new system to 15.7% after the new system (p < 0.001). The rate of contaminated cultures increased from 2.3% to 5.4%, and similarly, the rate of pathogen-containing cultures rose from 2.5% to 5.8% (p < 0.001) (Table 1). Rates of CoNS increased from 1.5% to 3.4% and that of Bacillus species from 0.3% to 1.5% (p < 0.001). Rates of cultures containing Corynebacterium, Micrococcus, and P. acnes did not differ significantly after the implementation of the new system. Rates of recovery of all three true pathogens increased significantly after the new system was installed. Rates for E. coli increased from 0.7% to 1.4%, S. aureus from 1.5% to 3.8%, and K. pneumoniae from 0.3% to 0.7% (p < 0.001) (Table 2).

Table 1.

Positive, Contaminated, and Pathogen-Containing Blood Cultures in Healthcare System Overall before and after Implementation of New Blood Culture System

	Before (n = 13,292 [%])	After (n = 7,686 [%])	p
Positive culture	1,003 (7.5)	1,204 (15.7)	<0.001
Any contaminant	300 (2.3)	412 (5.4)	<0.001
Any pathogen	330 (2.5)	447 (5.8)	<0.001

Table 2.

Rates of Contaminant and Pathogenic Micro-Organisms in Healthcare System Overall before and after Implementation of New Blood Culture System

	Before (n = 13,292 [%])	After (n = 7,686 [%])	p
Contaminant organisms
Coagulase-negative Staphylococcus	194 (1.5)	264 (3.4)	<0.001
Corynebacterium	47 (0.4)	34 (0.4)	0.318
Bacillus	34 (0.3)	116 (1.5)	<0.001
Micrococcus	28 (0.2)	18 (0.2)	0.725
Propionibacterium acnes	5 (0.04)	4 (0.1)	0.733
Pathogenic organisms
Escherichia coli	98 (0.7)	107 (1.4)	<0.001
Staphylococcus aureus	193 (1.5)	294 (3.8)	<0.001
Klebsiella pneumoniae	40 (0.3)	50 (0.7)	<0.001

Cultures were then evaluated according to the site of the blood draw, focusing on the STICU, surgical floor, BICU, MICU, and ED. The percentage of total hospital blood cultures that were obtained by each unit did not differ significantly after the new system was utilized, except in the STICU, which sent 6.8% of all samples obtained in the hospital prior to the new system but only 5.8% of the samples afterward (p = 0.002) (Table 3). After the new system was installed, the rates of positive blood cultures increased significantly in all five units (Table 4). The MICU experienced the largest increase in the positive blood culture rate, from 5.5% to 18.0% of cultures (p < 0.001). Rates of contaminated blood cultures increased significantly in the STICU (1.8% to 6.3%), ED (4.8% to 9.1%), and MICU (1.1% to 4.8%) with the new system (p < 0.001) (Table 5). Rates of cultures with true pathogens increased on the surgical floor (1.3% to 4.9%), ED (4.4% to 7.1%), and MICU (2.4% to 10.7%) (p < 0.001) (Table 5).

Table 3.

Total Cultures Sent by Hospital Location before and after Implementation of New Blood Culture System

	Before (n = 13,292 [%])	After (n = 7,686 [%])	p
STICU	910 (6.8)	443 (5.8)	0.002
Surgical floor	607 (4.6)	364 (4.7)	0.574
BICU	129 (1)	93 (1.2)	0.102
ED	3,569 (26.9)	2,146 (27.9)	0.093
MICU	1,169 (8.8)	722 (9.4)	0.144

BICU = burn intensive care unit; ED = emergency department; MICU = medical intensive care unit; STICU = surgical/trauma intensive care unit.

Table 4.

Rates of Positive Blood Cultures by Hospital Location before and after Implementation of New Blood Culture System

	Before (%)	After (%)	p
STICU	59 (6.5)	73 (16.5)	<0.001
Surgical floor	32 (5.3)	37 (10.2)	0.004
BICU	14 (10.9)	20 (21.5)	0.030
ED	491 (13.8)	469 (21.9)	<0.001
MICU	64 (5.5)	130 (18)	<0.001

For abbreviations, see Table 3.

Table 5.

Rates of Contaminated and Pathogen-Containing Cultures by Hospital Location before and after Implementation of New Blood Culture System

	Contaminants			Pathogens
	Before (%)	After (%)	P Value	Before (%)	After (%)	p
STICU	16 (1.8)	28 (6.3)	<0.001	17 (1.9)	14 (3.2)	0.136
Surgical floor	8 (1.3)	11 (3)	0.063	8 (1.3)	18 (4.9)	0.001
BICU	1 (0.8)	2 (2.2)	0.573	2 (1.6)	0 (0)	0.511
ED	171 (4.8)	195 (9.1)	<0.001	156 (4.4)	153 (7.1)	<0.001
MICU	13 (1.1)	35 (4.8)	<0.001	28 (2.4)	77 (10.7)	<0.001

For abbreviations, see Table 3.

The total number of individual cultures per patient was then evaluated for both patients with contaminants and those with pathogens. For patients with contaminated cultures, there was no significant difference in the number of total cultures obtained from each patient before and after the new system was introduced. The median total number of cultures obtained per patient was 4 (3–7) in the old system and 5 (3–8) in the new system (p = 0.092). For patients with pathogen-containing cultures, median number of total cultures per patient was 6 (4–9) and 7 (4–11) in the old and new systems, respectively (p = 0.158). The median number of duplicate cultures ordered per patient after a contaminated culture result was obtained was 2 (1–3.25) with the old system and 2 (0–4) after the implementation of the new system (p = 0.279). However, 79% of the patients with contaminated cultures had at least one repeat culture with the old system, whereas 69.4% of patients had a repeat culture with the new system, representing a significant reduction (p = 0.008).

Discussion

In April 2016, our hospital changed to a new blood culture system, one with higher sensitivity and shorter time to detection. With the old system, only 7.5% of all ordered blood cultures were positive, a relatively low yield similar to what has been reported in the literature [6,11-13]. Our overall hospital contamination rate was 2.3%, in accordance with the guidelines set forth by the Clinical and Laboratory Standards Institute [1]. Additionally, all evaluated locations had individual contamination rates <2%, except for the ED, which had a contamination rate of 4.8%. This is in agreement with previous findings that contamination rates in the ED tend to be higher, most likely because of a variety of ED-specific factors, including rapid staff turn-over, fast work pace, variable techniques for sample collection, limited staff, and rush to collect samples prior to initiating antibiotics [2,8].

After implementation of the new blood culture system, the rates of overall positive, contaminated, and pathogen-containing cultures increased significantly. Specifically, significant increases in the rates of cultures containing CoNS, Bacillus spp., E. coli, S. aureus, and K. pneumoniae were seen with the new system. Importantly, the new system resulted in contamination rates greater than the national goal of 3% in the STICU, ED, and MICU, as well as the hospital overall. The new system did not lead to any change in the number of repeat cultures obtained for patients with contaminated cultures, and in fact, fewer patients had repeat cultures.

This is the first study to our knowledge that evaluated hospital-wide blood culture contamination rates before and after the implementation of a new, automated culture system. As such, this study raises important questions regarding the clinical impact of employing a more sensitive blood culture system. First, how does identifying more contaminants and pathogens change clinical practice? Although the implementation of the new system did not result in more duplicate blood samples being sent on each patient with a contaminated culture, there was a significant decrease in the percentage of patients having repeat cultures. Although the reason for this is not clear, it may represent a “positive blood culture fatigue;” i.e., that physicians are seeing more contaminants with the new system and therefore may be less likely to pursue work-up of cultures believed to be contaminated. This is a potential area of concern, for although the typical organisms that represent contaminants are often just that, they may represent a true bacteremia in certain populations. Specifically, CoNS are one commensal organism that has variable rates of clinical significance and increasingly may represent a true pathogen, especially in vascular catheter-associated bacteremia and immunocompromised patients [1,6-8,14,19]. In fact, some studies suggest that in as many as 25% of cultures, CoNS are a true pathogen [1,6,8]. Further evaluation and implementation of methods to improve blood culture yields and to help discriminate between contaminants and true pathogens would lessen this risk [1,6]. Utilizing clinical decision tools and sending blood samples selectively also can help improve the diagnostic yield of blood cultures [6,10-13]. Finally, beyond changes in daily clinical practice, the effects of more positive blood cultures on outcomes such as ICU and hospital LOS, antibiotic usage, and cost deserves additional investigation and quantification.

Additionally, the higher rates of pathogen-containing cultures observed with the new system suggest that some true pathogens may have been missed with the old system. However, we cannot know retrospectively what this means from a clinical standpoint. Did these patients go untreated, and if so, did they have worse outcomes or a higher mortality rate? Conversely, if the new system detects smaller numbers of organisms, we may end up treating episodes of transient or clinically insignificant bacteremia that would not have been detected by the older system and may not have necessitated treatment. However, as automated blood culture systems continue to evolve, and sensitivity rates and time-to-detection continue to improve, there is an opportunity for better antibiotic stewardship. With improved detection of true pathogens, providers may be able to tailor antibiotic regimens more appropriately and reduce the use of empiric, broad-spectrum anti-microbial drugs.

Furthermore, it is important to gauge the real clinical impact of newer, more sensitive tests, not just for the individual patients, but also for healthcare systems and hospitals as a whole. In the age of greater focus on quality measures and pay-for-performance, higher rates of contamination and overall positive blood cultures attributable solely to using a “better” blood culture system (without actual changes in practice) is not better for the hospital. In 2008, Medicare ceased payments for certain hospital-acquired conditions, including vascular catheter-associated infections [20]. The use of a more sensitive blood culture system, leading to higher rates of positive blood cultures (which may or may not be clinically significant) could affect hospital reimbursements significantly. As payments decrease despite rising healthcare costs, and staffing ratios decrease in the face of higher inpatient censuses, determining the impact of higher rates of contaminated and potentially clinically insignificant positive blood cultures becomes not just about infection control and antibiotic stewardship, but about system sustainability.

The limitations of this study include its retrospective nature, as well as the fact that it is solely descriptive regarding changes in rates of blood culture positivity and contamination with a new automated blood culture system. Our EMR-based report did not include data such as age, clinical and laboratory data, source of blood specimen (peripheral vs. central line), volume of blood sample, if central–line-associated bloodstream infection was a concern, antibiotic usage (both type of drug and duration of treatment), ICU days, LOS, healthcare costs, disposition from hospital, re-admission rates, or outcomes such as a need for additional procedures or death. Additionally, our definitions of contaminants and pathogens were limited, in that we evaluated cultures only for the eight micro-organisms of greatest interest. Cultures having organisms such as fungi, other typical pathogens, or other contaminants were not evaluated, and therefore, our total numbers will underrepresent the total burden of both contaminants and pathogens. Additionally, the presence of one of the five contaminant micro-organisms was assumed to represent a truly contaminated culture, regardless of the number of positive bottles. That is, there was no distinction made between patients who had one versus multiple bottles positive for contaminants. Also, there was no further investigation to determine if a typical commensal organism, such as CoNS, appeared to be pathogenic. Finally, it is unclear why the new system is more sensitive, although it is possible that the larger blood volume in the samples plays a contributing role. However, the scope of the study was to define the degree of unintended consequences of employing a more sensitive instrument, which has not been demonstrated in the literature.

In conclusion, the implementation of a more sensitive automated blood culture system at our hospital was associated with significant increases in overall positive blood cultures, as well as higher contaminated and pathogenic blood culture rates. Contamination rates increased over the national goal of 3% in the STICU, ED, MICU, and the hospital overall. More research is needed to determine the clinical significance of these findings, as well as to quantify the impact on both patients and the healthcare system as a whole.

Footnotes

Author Disclosure Statement

The authors of this article have no disclosures.

References

Hall

, Lyman

. Updated review of blood culture contamination. Clin Microbiol Rev, 2006; 19:788–802.

Self

, Talbot

, Paul

, et al. Cost analysis of strategies to reduce blood culture contamination in the emergency department: Sterile collection kits and phlebotomy teams. Infect Control Hosp Epidemiol, 2014; 35:1021–1028.

Self

, Speroff

, Grijalva

, et al. Reducing blood culture contamination in the emergency department: An interrupted time series quality improvement study. Acad Emerg Med, 2013; 20:89–97.

Bates

, Goldman

, Lee

. Contaminant blood cultures and resource utilization: The true consequences of false-positive results. JAMA, 1991; 265:365–369.

Chang

, Wu

, Hsu

, et al. Factors associated with blood culture contamination in the emergency department: Critical illness, end-stage renal disease, and old age. PloS One, 2015; 10:e0137653.

Shafazand

, Weinacker

. Blood cultures in the critical care unit: Improving utilization and yield. Chest, 2002; 122:1727–1736.

Roth

, Wiklund

, Palsson

, et al. Reducing blood culture contamination by a simple informational intervention. J Clin Microbiol, 2010; 48:4552–4558.

Gander

, Byrd

, DeCrescenzo

, et al. Impact of blood cultures drawn by phlebotomy on contamination rates and health care costs in a hospital emergency department. J Clin Microbiol, 2009; 47:1021–1024.

Lin

, Boehm

. Positive predictive value of blood cultures utilized by community emergency physicians. ISRN Infect Dis, 2013; 2013.

10.

Shapiro

, Wolfe

, Wright

, et al. Who needs a blood culture? A prospectively derived and validated prediction rule. J Emerg Med, 2008; 35:255–264.

11.

Stalnikowicz

, Block

. The yield of blood cultures in a department of emergency medicine. Eur J Emerg Med, 2001; 8:93–97.

12.

Kelly

. Clinical impact of blood cultures taken in the emergency department. J Accid Emerg Med, 1998; 15:254–256.

13.

Mountain

, Bailey

, O'Brien

, Jelinek

. Blood cultures ordered in the adult emergency department are rarely useful. Eur J Emerg Med, 2006; 13:76–79.

14.

Fairfax

MRS

. Beware of unusual organisms masquerading as skin contaminants. In: Azevedo

, ed. Sepsis: An Ongoing and Significant Challenge. InTech; 2012:275–286.

15.

Self

, Mickanin

, Grijalva

, et al. Reducing blood culture contamination in community hospital emergency departments: A multicenter evaluation of a quality improvement intervention. Acad Emerg Med, 2014; 21:274–282.

16.

Culshaw

, Glover

, Whiteley

, et al. Healthcare-associated bloodstream infections in critically ill patients: Descriptive cross-sectional database study evaluating concordance with clinical site isolates. Ann Intensive Care, 2014; 4:34.

17.

Waltzman

, Harper

. Financial and clinical impact of false-positive blood culture results. Clin Infect Dis, 2001; 33:296–269.

18.

Weinstein

. Blood culture contamination: Persisting problems and partial progress. J Clin Microbiol, 2003; 41:2275–2278.

19.

Mermel

, Allon

, Bouza

, et al. Clinical practice guidelines for the diagnosis and management of intravascular catheter-related infection: 2009 Update by the Infectious Diseases Society of America. Clin Infect Dis, 2009; 49:1–45.

20.

Peasah

, McKay

, Harman

, et al. Medicare non-payment of hospital-acquired infections: Infection rates three years post implementation. Medicare Medicaid Res Rev, 2013; 3.