Context.—

Blood culture contamination is a common problem faced by medical centers and leads to significant cost. A possible method to reduce contamination is to discard the initial aliquot of blood, which contains skin and bacteria.

Objective.—

To determine whether the rate of contaminant blood cultures could be reduced by changing the order of draw to divert the first 7 mL to a gold- or green-top tube.

Design.—

A preintervention and postintervention study was conducted. During the 18-month intervention phase (September 2015–February 2017), all nurses in the emergency department and inpatient floor phlebotomists collected blood cultures by drawing the first 7 mL of blood into a gold- or green-top tube followed by drawing blood for blood culture bottles. The 18 months immediately preceding the study period (February 2014–July 2015) were used for comparison.

Results.—

There was an overall statistically significant decrease in contamination rate from 2.46% in the prediversion protocol group to 1.70% in the postdiversion protocol group (P < .001). Emergency department drawn cultures and inpatient cultures showed significant decrease in contamination rates between the preprotocol and postprotocol groups, 2.92% versus 1.95% (P < .001) for emergency department, and 1.82% versus 1.31% (P = .03) for inpatient. We noted less month-to-month variation during the study period compared with the preintervention period.

Conclusions.—

By using this simple diversion method, we were able to improve blood culture contamination rates for our emergency department and inpatients while incurring no added cost to the procedure.

Blood culture contamination is a common and recurring problem for most medical centers, and the costs to the hospital operation are significant. According to independent, third-party peer-reviewed cost analyses performed at different institutions around the world, the cost of contaminated blood cultures ranges from $3000 to$4000, or higher, per patient. The added cost is based primarily on extended length of stay and incremental charges of administering unnecessary antibiotics.14

The American Society for Microbiology guidelines suggest blood culture contamination rates should be less than 3% of all blood cultures drawn.5  At AMITA Health Saint Francis Hospital in Evanston, Illinois, a 150-bed community hospital with a busy emergency department (ED), our nurses draw most of our blood cultures. We find it necessary to conduct frequent in-services and educational campaigns for our nurses and phlebotomists (as often as every 2–3 months) in order to keep the rate below the threshold. In addition, new employees are specially trained in blood culture collection as part of their orientation. We have found that, in general, phlebotomists at our hospital, whose jobs are focused on blood drawing alone, are better able to achieve lower contamination rates compared with our nurses in the ED setting. In the ED, multiple factors impact contamination rates, including rapid staff turnover, limited staff to handle high patient census, and multiple simultaneous demands on nursing time.

The main reasons for blood culture contamination have been identified by direct observation of our staff and include failure to adhere to the skin cleansing protocol (which is both exacting and time-consuming), and failure to draw the sample from the correct site. The steps that are frequently skipped in the rushed health care setting of the ED include cleansing the skin properly and waiting for the skin cleanser to dry completely. In addition, collecting from an in-dwelling hep-lock rather than from a venipuncture site can also cause contamination. Bacteria also can be introduced during the assembly of blood culture supplies or from the procedure itself, which may dislodge viable bacteria from the skin and introduce them into the needle.

By convention, blood culture bottles are typically drawn first after venipuncture, before any other blood tubes are filled, ostensibly to avoid introducing bacteria from any of the equipment used. However, given that most of the contaminants isolated are skin flora, reversing the order by collecting a tube of blood first, followed by blood culture draw, could serve as a way to “catch” any bacteria introduced into the needle.

The strategy of “flushing the line” has been well studied in blood banking, where strict sterility must be maintained while drawing blood for product preparation. The “diversion method” has been shown to decrease bacterial contamination of blood components by as much as 50%, with an 85% reduction of contamination by skin flora.6,7  The diversion method is a requirement for any collection resulting in a platelet product, and the process is listed in the Standards for Blood Banks and Transfusion Services.

We hypothesized that we could apply the diversion method to our blood culture protocol to reduce contamination rates simply by changing the order of blood draw, a modification that added nothing to the cost of the procedure.

## MATERIALS AND METHODS

Our study design was to compare contamination rates during an 18-month preintervention period to an 18-month period after intervention. During the postintervention period, all nurses in the ED and inpatient floor phlebotomists were instructed to draw blood cultures by the following procedure: (1) sterilizing the skin; (2) drawing 7 mL of blood into a gold- or green-top tube; (3) drawing blood for a set of blood culture bottles (1 aerobic and 1 anaerobic); and (4) drawing blood for the remaining ordered tests as needed. In cases where no other tests were ordered, the green or gold top was sent to the lab and held in case testing was needed later.

We provided in-services to the staff during a 1-month transition period. We reminded the staff of the new protocol by providing colored paper inserts into prepackaged culture kits, which included all of the necessary equipment. We also included a photograph of the tubes arranged in the order they should be drawn. Finally, we required the staff to initial the blood culture bottles to help us target re-education as needed.

We looked at 18 months of data with the new protocol (14 046 patients) and compared this to 18 months of data with the usual protocol (13 350 patients), allowing for a 1-month interval transition period not included in the data analysis.

Our study period was September 2015–February 2017, and our control period was February 2014–July 2015. The month of August 2015 served as a transition period during which the staff received education about the new study protocol. Our hospital threshold was 3.0% contamination for blood cultures per the American Society of Microbiology guidelines. The step-by-step method procedure used is outlined in Table 1.

Table 1

Diversion Method Procedure

The automated blood culture system used for this study was the Bactec FX manufactured by Becton Dickinson (Franklin Lakes, New Jersey). The blood culture media employed were Bactec Plus Aerobic/F culture vials and Bactec Lytic/10 Anaerobic/F culture vials. Each vial was inoculated with 7 to 10 mL of patient blood sample. Short draws, defined as less than 7 mL of patient volume, were labeled “QNS” and not processed. Upon receipt in the hospital laboratory, the vials were placed in a 35°C incubator until the scheduled courier transfer to the Alverno Central Laboratory. The culture vials were transported in a 35°C incubated transport container. Upon arrival at the central laboratory the vials were transferred to the Bactec FX incubation instruments. When the instruments detected growth of bacteria and flagged the vials as positive, the vials were removed, they were subcultured to appropriate media, smears for Gram stain were prepared, and an aliquot of the vial was processed for direct matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) identification. The results of the Gram stain and direct identification were reported within an average of 2.5 hours of when the vials were flagged as positive. Final identification of the organism(s) present in the samples was confirmed by MALDI-TOF mass spectrometry from culture growth or, if necessary, phenotypic biochemical identification.

Patients who tested positive for sepsis by standard ICD-10 criteria received an order for a blood culture. Because blood cultures are only processed once an electronic order is placed in the electronic medical record, and we relied on electronic records for our study, we believe all blood cultures drawn between the prestudy and poststudy period were included in the study.

Blood culture contamination rates were assessed by reviewing all positive blood cultures and scoring a culture as a contaminant if common skin flora (coagulase-negative Staphylococcus spp, Micrococcus spp, viridans streptococci, or Cutibacterium spp) or other bacteria were present in only 1 of 2 sets. Before accepting a blood culture as an authentic positive, we performed clinical correlations for any cases in which a common skin flora was found in both sets.

The study outcome measured the proportion of total contaminated cultures in each study period. Pearson χ2 test was used to compare differences in contamination rates between the 2 periods under 1 study; statistical significance was defined as a P ≤ .05. Statistical analysis was conducted using Stata/SE, version 14 (StataCorp, College Station, Texas).

## RESULTS

During the February 2014 to July 2015 preintervention period there were a total of 13 350 samples collected, of which 329 were contaminated (2.46%). In the ED, there were a total of 7845 samples collected, of which 229 were contaminated (2.92%). For inpatients, there were a total of 5505 samples collected, of which 100 were contaminated (1.82%).

There was a statistically significant decrease in blood culture contamination rates with the diversion protocol in both inpatient and ED subgroups, and in the combined group (Figures 1 through 3).

Figure 1

Blood culture contamination rates for emergency department (ED) and inpatient at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. Blood culture contamination rates of less than 3% were consistently met for our hospital upon analysis of the combined data of both emergency room and inpatient, with less month-to-month variation using the diversion protocol.

Figure 2 Blood culture contamination rates for emergency department at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. In the emergency department (ED) subgroup, the average contamination rate was 2.92% in the preintervention period compared with a 1.95% contamination rate in the postintervention period.

Figure 1

Blood culture contamination rates for emergency department (ED) and inpatient at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. Blood culture contamination rates of less than 3% were consistently met for our hospital upon analysis of the combined data of both emergency room and inpatient, with less month-to-month variation using the diversion protocol.

Figure 2 Blood culture contamination rates for emergency department at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. In the emergency department (ED) subgroup, the average contamination rate was 2.92% in the preintervention period compared with a 1.95% contamination rate in the postintervention period.

Figure 3

Blood culture contamination rates for inpatients at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. For the inpatient subgroup, the average contamination rate was 1.82% in the preintervention period compared with a 1.31% contamination rate in the postintervention period.

Figure 4 Isolates by organism. The most commonly identified organism was Staphylococcus epidermidis, followed by Staphylococcus hominis, as well as other coagulase-negative Staphylococcus species. All contaminant types decreased following the diversion protocol. Abbreviations: CoNS, coagulase-negative Staphylococcus species; S capitis, Staphylococcus capitis; S epidermidis, Staphylococcus epidermidis; S hominis, Staphylococcus hominis; spp, species.

Figure 3

Blood culture contamination rates for inpatients at AMITA Health Saint Francis Hospital Evanston, February 2014–February 2017. For the inpatient subgroup, the average contamination rate was 1.82% in the preintervention period compared with a 1.31% contamination rate in the postintervention period.

Figure 4 Isolates by organism. The most commonly identified organism was Staphylococcus epidermidis, followed by Staphylococcus hominis, as well as other coagulase-negative Staphylococcus species. All contaminant types decreased following the diversion protocol. Abbreviations: CoNS, coagulase-negative Staphylococcus species; S capitis, Staphylococcus capitis; S epidermidis, Staphylococcus epidermidis; S hominis, Staphylococcus hominis; spp, species.

During the September 2015–February 2017 postintervention period, 14 046 samples were collected, of which 239 were contaminated (1.70%). In the ED, 8611 samples were collected, of which 168 were contaminated (1.95%). For inpatients there were a total of 5435 samples taken, of which 71 were contaminated (1.31%).

We noted relative decreases of approximately the same magnitude in both groups (approximately 30%), and less month-to-month variation during the postintervention study period compared with the preintervention period.

The contamination rates remained below 3% throughout the entire study period but drifted above the 3% threshold on average 1 out of 6 months during the prestudy control period.

In the ED subgroup (Figure 2) the average contamination rate in the preintervention group was 2.92% compared with the inpatient subgroup average at 1.82%. The preintervention period had a pattern of wide variation in the contamination rates, ranging between 2.07% and 5.16%. In the postintervention period, there was a continued decrease in the contamination rate (range, 0.65%–4.08%) as well as a decrease in the variation in the contamination rates (2.26% postintervention versus 3.09% preintervention). During the months of September–November 2016, there was an unusually high contamination rate of 2.4%. This time period coincides with the highest patient volumes both in the form of ED visits as well as hospital admissions.

For the inpatient subgroup (Figure 3), contamination rates had been consistently lower than the 3% threshold (1.82% average for the period). A slight decrease in contamination rates was seen in this group after the diversion protocol (1.31%), and this difference was found to be statistically significant (P = .03).

The overall contamination rate, including both ED and inpatient, was 2.46% in the prediversion protocol period and 1.70% in the postdiversion protocol group. When the blood cultures were further subdivided into ED drawn cultures and inpatient cultures, the contamination rates still showed decreases between the preprotocol and postprotocol groups (2.92% versus 1.95% for ED and 1.82% versus 1.31% for inpatient cultures; Table 2).

Table 2

Contamination Rates by Location Prediversion and Postdiversion Protocol

The decrease in the overall blood culture contamination rates (ED and inpatient combined) between the preprotocol and postprotocol periods was statistically significant (P < .001) and resulted in a significant reduction in total contaminants during the postintervention 18-month study period.

The most commonly identified organism was Staphylococcus epidermidis, followed by Staphylococcus hominis, as well as other coagulase-negative Staphylococcus spp. Other common contaminants were Corynebacterium sp and viridans streptococci (Figure 4).

The type of contaminant cultured remained unchanged in the preprotocol and postprotocol group. However, there was a dramatic decrease in the number of organisms isolated in the postintervention group compared with the preintervention group. Decreases in numbers of cultures of contaminant bacteria were as high as 98% (Table 3).

Table 3

Prediversion and Postdiversion Isolates

## DISCUSSION

Various strategies have been reported to reduce blood culture contamination rates. These include interventions such as type of skin preparation/antiseptic used,8  culture bottle preparation, single-needle versus double-needle use,9  use of a specialized phlebotomy team,2,4,10  and costly commercial blood culture collection kits. So far none of the interventions have been shown to be superior over the others, and some result in significant up-front costs. Commercially available diversion kits range in price from $20 to$28 per device, a cost that detracts from the savings produced.

The diversion method we used in this study has the advantage of wide acceptance in blood banking product collection and no significant increase in cost or change in basic equipment used. The protocol is simple to understand and, in most cases, results in no extra phlebotomy for the patient.

As others have stressed,2,5  the cornerstone of a program to reduce blood culture contamination rates is good education of the staff coupled with frequent reminders and targeted re-education as necessary. Nurses and phlebotomists must pay careful attention to all the steps in the traditional blood culture protocol, which will provide good-quality results whether or not the diversion principle is used. In our study, all nursing staff and phlebotomists were re-educated on the intended protocol for blood culture collection every 2 to 3 months as needed.

We found that adding the diversion method in addition to a sound re-education program will improve blood culture contamination rates significantly compared with education and the traditional protocol alone. Moreover, we found that even when the possible gain was modest, statistically significant improvement was seen.

The study measured and compared performance in nursing staff and phlebotomists. All cultures drawn in the ED were by nursing staff only, in contrast to the cultures drawn by phlebotomists on the inpatient floor. Previous studies have demonstrated that a dedicated phlebotomist team in the ED can be successful in decreasing contamination rates. Our study showed a 30% decrease in contamination rates in blood cultures drawn by both the phlebotomists (inpatient) and the nursing staff in the ED.

Another distinctive feature of the diversion protocol was the decrease in variation in the contamination rates noted in the postintervention group compared with the preintervention period. The diversion protocol maintained a low contamination rate consistently and uniformly throughout the 18-month time period.

We did not observe any untoward consequences as a result of using the diversion method. True-positive blood culture rates were similar in both periods (7.33% prediversion versus 6.87% postdiversion). The protocol does not involve switching needles or connecting new devices, so it was well accepted by staff. No increase in the number of needle stick injuries in association with blood cultures drawn in the ED or inpatient setting was noted. There were 6 needle stick injuries reported in the preintervention period and 4 reported in the postintervention period.

To improve accountability during the intervention period, monthly feedback and opportunities for additional training were given to nurses who drew cultures contaminated by skin flora. When this was done on a monthly basis, it was noticed that retrained nurses did not reappear on the list. To address the issue of compliance, we modified our protocol to include a mandatory 3-mL “waste tube,” which is incorporated into preassembled blood culture collection kits to return with the culture bottles. The waste tube serves 2 important purposes. Including the waste tube as part of the kit serves as a visual reminder to staff before drawing the cultures, and, second, its return with the cultures serves as a means for us to verify that the protocol was followed correctly.

## STUDY LIMITATIONS

This study was performed in a single small community hospital, and the results may not be generalizable to other types of hospitals. Another limitation of the study was not being able to account for compliance with the protocol in every case. Moreover, frequent nurse staff turnover in the ED challenged compliance with the protocol. However, all testing personnel were required to attend an educational session, and any outlier personnel were re-educated. Interviews with nurse managers and phlebotomists were conducted periodically during the study to reinforce compliance with the protocol.

Targeted education programs in November 2015 and March 2016 preceded our best month ever, with record low contamination rates of only 0.9% and 0.7%, respectively, in the ED following those programs. This reinforces the conclusions of others that staff education is critical to the success of a program to reduce blood culture contamination.8,11,12

Another limitation to note is that our available opportunity for improvement was small because the hospital was already meeting the overall goal of less than 3%, on average, when the study began. However, we were still able to see a statistically significant decrease in blood culture contamination rates in the pre and post study period in both the ED and inpatient settings. A hospital with a greater opportunity for improvement would likely experience more significant gains.

## References

1
Bates
DW
,
Goldman
L
,
Lee
TH
.
Contaminant blood cultures and resource utilization: the true consequences of false positive results
.
JAMA
.
1991
;
265
(
3
):
365
369
.
2
Gander
RM
,
Byrd
L
,
DeCrescenzo
M
,
Hirany
,
Bowen
M
,
Baughman
J
.
Impact of blood cultures drawn by phlebotomy on contamination rates and health care costs in a hospital emergency department
.
J Clin Microbiol
.
2009
;
47
(
4
):
1021
1024
.
3
Dunagan
WC
,
Woodward
RS
,
Medoff
G
, et al.
Antimicrobial misuse in patients with positive blood cultures
.
Am J Med
.
1989
;
87
(
3
):
253
259
.
4
Surdulescu
S
,
Utamsingh
D
,
Shekar
R.
Phlebotomy teams reduce blood-culture contamination rate and save money
.
Clin Perform Qual Health Care
.
1998
;
6
(
2
):
60
62
.
5
Ruge
DG
,
Sandin
RL
,
Siegelski
SA
,
Greene
JN
,
Johnson
N.
Reduction in blood culture contamination rates by establishment of policy for central intravenous catheters
.
Lab Med
.
2002
;
33
(
10
):
797
800
.
6
Lee
CK
,
Wong
HK
,
Ho
PL
, et al.
Significant bacterial contamination risk reduction with the use of diversion pouch
.
Transfus Med
.
2012
;
22
(
6
):
404
408
.
7
McDonald
CP
,
Roy
A
,
Mahajan
P
,
Smith
R
,
Charlett
A
,
Barbara
JA
.
Relative values of the interventions of diversion and improved donor-arm disinfection to reduce the bacterial risk from blood transfusion
.
Vox Sang
.
2004
;
86
(
3
):
178
181
.
8
Schifran
RB
,
Pindur
A.
The effect of skin disinfection materials on reducing blood culture contamination
.
Am J Clin
.
1993
;
99
(
5
):
536
538
.
9
Bekeris
LG
,
Tworek
JA
,
Walsh
MK
,
Valenstein
PN
.
Trends in blood culture contamination: a College of American Pathologists Q-Tracks study of 356 institutions
.
Arch Pathol Lab Med
.
2005
;
129
(
10
):
1222
1225
.
10
Weinbaum
FI
,
Lavie
S
,
Danek
M
,
Sixsmith
D
,
Heinrich
GF
,
Mills
SS
.
Doing it right the first time: quality improvement and the contaminant blood culture
.
J Clin Microbiol
.
1997
;
35
(
3
):
563
565
.
11
Eskira
S
,
J
,
Schlaeffer
P
, et al.
Reduction of blood culture contamination rate by an educational intervention
.
Clin Microbiol Infect
.
2006
;
12
(
8
):
818
821
.
12
M
,
Jackson
T
,
Williams
C.
Simple measures to reduce the rate of contamination of blood cultures in accident and emergency
.
Emerg Med J
.
2005
;
22
(
11
):
810
811
.

## Author notes

The authors have no relevant financial interest in the products or companies described in this article.