Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
406
arntip Sangsuwan, M.D.*, Rungtip Darayon, M.NS.**, Silom Jamulitrat, M.D.*
*Department of Family Medicine and Preventive Medicine, Faculty of Medicine, Prince of Songkla University, Hat Yai, Songkla 90110, ailand,
**Infection Control Unit, Songklanagarind Hospital, Hat Yai, Songkla 90110, ailand.
G-control Charts for Contamination Rates of Blood
Cultures in a University Hospital
ABSTRACT
Objective: To determine blood culture contamination rates, and display with a g-chart.
Materials and Methods: A retrospective cohort study was conducted. e medical records of patients, from whom
blood cultures were obtained in a university hospital, during January and December, 2019 were retrieved and
reviewed for contamination. e Centers for Disease Control and Prevention (CDC) criteria were used to classify
the blood culture results. e contamination rates were illustrated with a g-chart.
Results: We identied 331 false-positive blood cultures, among 32,961 cultured specimens; yielding a contamination
rate of 1.0% (95% CI = 0.9% – 1.1%). e highest contamination events occurred in the emergency department
(49.2%), pediatric ICU (5.2%) and neonatal ICU (4.8%), respectively. e most common contaminated commensal
bacterial genus were coagulase -negative Staphylococci (67.1%), Bacillus spp. (10.2%) and Corynebacterium spp.
(7.6%). e g-charts could identify 14 abnormal variations, in 41 locations.
Conclusion: e contamination rates found were within ranges of other reports. G-charts are simple to construct,
easy to interpret and sensitive for detection of real time epidemics.
Keywords: Hemoculture; blood culture; contamination; rate; geometric; SPC chart (Siriraj Med J 2021; 73: 406-412)
Corresponding author: arntip Sangsuwan
E-mail: be_med29@hotmail.com
Received 18 September 2020 Revised 1 April 2021 Accepted 2 April 2021
ORCID ID: http://orcid.org/0000-0003-3390-413X
http://dx.doi.org/10.33192/Smj.2021.54
INTRODUCTION
Blood cultures play an important role in the management
of bloodstream infections, due to it is a critical tool detecting
the dangerous presence of living organisms in the blood
stream. However, the merits of blood culture results are
jeopardized by false positives, resulting from contamination
during the taking or processing of blood specimens.
Blood culture contamination represents an ongoing
source of frustration for clinicians and microbiologists
alike. Ambiguous culture results oen lead to diagnostic
uncertainty in clinical management and are associated
with increased health care costs due to unnecessary
treatment and testing.
1
ere are several steps in the
process of taking blood cultures that may inuence the
contamination rate. Blood culture contamination has
been attributed to the transference of organisms from
the patient’s skin, the immediate environment of the
patients, supplies used to obtain or transfer the blood
samples or from the hands of the health care worker
performing the procedure.
2-3
In this era of strains on the resources and rising
cost of healthcare, it becomes increasingly apparent
that decisions must be made on facts, not just opinions.
Consequently, data must be gathered and analyzed. is
is where statistical process control (SPC) comes in. For
over decades, the healthcare setting has beneted from
the tools of SPC that have helped guide the decision-
making process.
4-5
Sangsuwan et al.
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
407
Original Article
SMJ
Control chart is the main tool in SPC and usually
used for monitoring and improving the ongoing process.
Geometric SPC chart (g-chart) is based on the geometric
distribution and was designed to monitor rare events.
Primary baseline data is an essential part of any quality
improvement project. Hence, the primary intention of
this study was to determine blood culture contamination
rates, and display with a g-chart. To document the rates,
and variations of blood culture contaminations needed
for a blood culture quality improvement project.
MATERIALS AND METHODS
Setting
e study was conducted in Songklanagarind Hospital,
a tertiary care, medical school, and training hospital in
Southern ailand. In our hospital, clinical blood culture
samples are usually collected at the bedside, from two
separated specimens; taken from dierent venipuncture
sites.
Studied samples
Blood culture specimens taken from patients
admitted to the hospital, from the 1
st
of January to the
31
st
of December, 2019.
Blood sample collection
Prior to venipuncture, the skin was disinfected with
a combination of 2% chlorhexidine gluconate and 70%
alcohol, for 30 seconds, then allowed to dry, except that
taken from infants <2 months, in which 70% alcohol
would be used instead. Aer antisepsis, the veins would
not be touch, without use of sterile gloves. en the vein is
pierced with a needle, and drawn into a syringe. Samples
were subsequently inoculated into blood culture bottles
without change of needles. Blood culture collection kits
are not used in this process.
Specimen Processing
Blood samples were obtained in media bottles, and
kept at room temperature before being transferred, as
soon as possible, to the microbiology laboratory for
processing (within 2 hours). Blood culture specimens
were incubated in automated instruments for 5 days, or
until the automate alarm for positive blood culture.
e automated blood culture system used in the
hospital is BD BACTEC FX (BACTEC) by Becton Dickson
& Co., sparks, MD. It is used to process blood cultures with
isolates identied using MALDI-TOF and biochemical
methods, according to standard practices.
Microbiology lab identication
Once blood cultures become positive for growth,
either by manual subculture techniques (blood agar,
chocolate agar, and MacConkey agar) or signaling from
automated systems, a Gram stain is performed. A positive
Gram stain result is regarded as a critical value, and
the ordering clinician, or another responsible member
of the healthcare team providing care to the patient is
immediately informed. At this point, subcultures are
performed and these allow identication and, if indicated,
susceptibility testing is then performed; typically over the
next 24-48 hrs. Complete organism identication and
organism-specic susceptibility testing is performed on
all positive blood culture specimens.
Denitions of blood cultures
1. Positive Blood cultures: Any blood cultures which
microorganisms are found.
2. Blood Stream Infection: Positive Blood cultures which
the microorganisms are not included in CDC common
commensal lists
6
or two blood specimens found the same
microorganisms.
3. Secondary Blood Stream Infection: One or more positive
blood cultures which the microorganisms are included
in CDC common commensal lists6 and also found the
same microorganisms at another site of the body.
4. Contaminated Blood cultures: One positive blood
culture which the microorganisms are included in CDC
common commensal lists
6
with no more than one matching
organism identied in 2 separated blood specimens and
No primary infection source of the organism identied
Contaminated blood culture
Studied variables
e variables in this study include blood sample
collecting date, age and gender of patients, wards that
request blood culture and results of the blood cultures.
Statistical analysis
Continuous data were described in terms of
arithmetic or geometric mean, according to the types
of data distribution. Discrete data were presented as
percentage. Contamination rates were calculated by
dividing the number of contaminated specimens with
the total number of cultured specimens. Contamination
rates were reported in terms of percentage. e dispersion
of data was represented by variance or 95% condence
interval (95% CI). e 95% CI of proportion were estimated
based on exact binomial statistics.
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
408
Construction of the g-charts were done by line graph
plotting the numbers of non-contaminated specimens
between pairs of contaminated specimens (NBP), in
axis y against the consecutive contaminated specimen
numbers (CSN) in axis x. e y-axis is displayed in log
scale base 2. e chart then starts with the second CSN,
and the NBP between the rst and the second CSN.
7-8
We used median of total NBP to dene the center
line (CL), and used condence intervals to dene control
limits of the chart. e condence intervals were calculated
using equations proposed by Yang Z et al.
9
e equation
for lower limit is ln(1 -
α
/2)/ln(q) and the equation for
upper limit is ln(
α
/2)/ln(q)-1 ,where α is the cumulative
probability and q is the probability of a non-contaminated
specimens.
7
Denitions of the g-chart; lower and upper control
limits with calculation formula
Chart limits
1. Lower control limit (LCL): Lower bound of
95%CI and formula is ln(0.975)/ln(q)
2. Lower warning limit (LWL): Lower bound of
80%CI and formula is ln(0.9)/ln(q)
3. Upper warning limit (UWL): Lower bound of
80%CI and formula is ln(0.1)/ln(q)-1
4. Upper control limit (UCL): Lower bound of
95%CI and formula is ln(0.025)/ln(q)-1
ln = Natural logarithm or log
e
q = Probability of non-contaminated specimen
e outbreak of blood culture contamination can
be diagnosed by any of the following rules; 1) there is
one point of the graph that fell under LCL, 2) there are
two successive points falling under LWL 3) there are
ve successive points under CL, and 4) there are six
successive points decreasing.
Ethics in research
e study protocol was approved by the Ethics
Committees of the Faculty of Medicine, Prince of Songkla
University (EC: 62-451-9-1). Because of the observational
nature of the study, written informed consent was not
required.
RESULTS
Characteristics of studied samples
e study included 32,961 blood culture specimens,
from 8,841 hospital patients. e characteristics of the
patients are shown in Table 1.
Blood culture results
Using the Center of Disease Control and Prevention
(CDC) criteria
6
, we could identify 331 (1.0%) contaminated
blood specimens among 32,961 of the total blood specimens
requested (Fig 1). e Pareto diagram of the number of
contamination is illustrated in Fig 2. e contaminated
micro-organisms are listed in the appendix.
TABLE 1. Demographic data of the patient, for whom blood culture specimens were taken.
Characteristics 95%CI
Age (year) Mean = 50.53 49.95 - 51.11
Gender (%)
Male 51.75 50.71 - 52.79
Female 48.25 47.21 - 49.29
Service (%)
Emergency 31.22 30.72 - 31.72
Medicine 25.93 25.45 - 26.40
Surgery 18.05 17.64 - 18.47
Pediatric 13.58 13.21 - 13.95
Obstetric & Gynecology 4.04 3.83 - 4.26
Others 7.18 6.90 - 7.45
Sangsuwan et al.
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
409
Original Article
SMJ
Fig 1. Blood culture results for the year 2019.
Fig 2. Pareto diagram for the number of contaminated blood culture specimens (presented with diagram) and corresponding cumulative
percentage of contamination (presented with line diagram)
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
410
Descriptive data of number between contamination
The average as well as variance of numbers
between contaminated blood specimens was 98 and
9,127, respectively. e median was 71. e histogram
is demonstrated in Fig 3.
g-Control chart
e g-Chart of blood culture contamination, in
PSU hospital for the year 2019, is illustrated in Fig 4.
Outbreak of blood culture contamination
Outbreak of blood culture contamination in
Songklanagarind Hospital is shown in Table 2. We could
identify 14 outbreaks in the year 2019. e average run
length (average of number between outbreaks) was 19.
DISCUSSION
e study design of this research was a cross-sectional
descriptive analysis, which can only study a point in
time, and lacks the ability to identify the cause-eect
relationship. erefore, the results can only represent
the magnitude of the problem.
Some microorganisms such as Burkholderia pickettii
10
are not enrolled in the common commensal organism’s list
of the CDC
6
; nonetheless, microorganisms can be causative
agents for blood culture contamination. erefore, this
may be the reason for the occurrence of false negative, in
the other words, the contamination rate may be possibly
lower than the actual result.
Although, Songklanagarind Hospital has no
phlebotomy team available the blood sample collection
method is practiced via standard protocol.
Fig 4. g-Chart of blood culture contamination in Songklanagarind Hospital for the year 2019.
Fig 3. Histogram of the number between contaminations.
Sangsuwan et al.
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
411
Original Article
SMJ
TABLE 2. Outbreak of blood culture contamination in Songklanagarind Hospital.
Ward Number Criteria Date of outbreak
Month Day
All wards 1 Five points under February 8-9
2 Two points under 23-24
3 One point under May 1
4 One point under 20
5 One point under June 16
6 One point under 18
7 One point under July 17
8 Two points under August 10
9 Two points under 15
10 Five points under 28-30
11 One point under October 17
12 Five points under November 5-7
13 Five points under December 2-3
14 Five points under 22-23
Average run length = 19 (95%CI = 13-25)
e results show that there is a huge dierence
between contamination in the Emergency Department
and other services. It has been suggested that urgent
care, lack of ongoing training, workload and nature of
present patients may contribute to this. From the literature
reviews show that Zahra Hashemizadeh had the highest
contamination rate (8.47%) in Neonatal Care Units in
Shiraz, Southwest-Central Iran.
11
In contrast, Chang CJ,
et al. and Washer LL had the lowest contamination rate
(0.2%) in discharged patients from Emergency Department,
National Cheng Kung University Hospital, Taiwan, and
patients using povidone-iodine and iodine as antiseptics
in University of Michigan Health System respectively.
12-13
e contamination rate in a single Emergency Department
at a university-aliated, tertiary care adult hospital in
the United States was maintained below 3% during each
biweekly interval throughout the intervention period in
the study of Self HW et al.
14
ey developed the sterile
blood culture intervention to convert blood culture
collection from a clean to a sterile procedure.
More than 50 % of contaminated microorganism are
coagulase-negative staphylococci including Stahylococcus
epidermidis (37.18%), Staphylococcus hominis (8.93%),
Staphylococcus capitis (7.49%).
e Pareto chart is one of the seven basic tools of
quality control. It is a type of chart that contains both bars
and a line graph, where individual values are represented
in descending order by bars, and the cumulative total
is represented by the line. e le vertical axis is the
frequency of occurrence and the right vertical axis is the
cumulative percentage of the total number of occurrences.
e purpose of the Pareto chart is to highlight the most
common sources of defects. We used general 80/20 rule
to identify the 20% of wards that created 80% of overall
contamination
Statistic process control (SPC) techniques have played
an eective part in monitoring hospital performance. e
Geometric SPC chart (g-chart) is appropriately used in
this study, because the contamination data has an over-
dispersion problem, which is shown in histogram of
Volume 73, No.6: 2021 Siriraj Medical Journal
https://he02.tci-thaijo.org/index.php/sirirajmedj/index
412
number between contaminations (Fig 3). G-chart analysis
is based on inverse sampling to either detect process
changes, or verify improvements faster. Prospective
g-chart analysis is able to trigger specic awareness
when relevant increases or decreases of rare events are
detected. Such alarms enable timely root cause analysis,
so as to secure early clinical process.
15
Also g-Chart is
appropriate for very low incident event for its take less
eort to collect data and can provide real time outbreak
detection”.
Previously we actually had no formal blood culture
monitoring system. is study provides information
needed to priority setting, and establishing baseline data
for the hospital’s quality improvement, which has never
been done before. Quality improvement of blood cultures
can reduce additional costs, overuse of antibiotics and
drug-resistant bacteria in the hospital.
CONCLUSION
We identied 331 false-positive blood cultures, among
32,961 cultured specimens; yielding a contamination
rate of 1.0% (95%CI = 0.9 - 1.1). is blood culture
contamination rate is very low when compared to other
reports. e g-control chart is a very eective tool that
can detect 14 abnormal variations in 41 locations, by a
3 outbreak criteria comprising of: 1 point under LCL,
2 points under LWL and 5 points under CL.
ACKNOWLEDGEMENTS
We would like to thank Mr.Andrew Jonathan Tait
who assisted by editing the English language of the
manuscript.
Conict of interest: All authors of this article certied
that there were no nancial nor non-nancial conicts
of interest.
REFERENCES
1. Hall KK, Lyman JA. Updated Review of Blood culture
contamination. Clin Microbiol Infect 2006;19:788-802.
2. Alahmadi YM, Aldeyab MA, McElnay JC, Scott MG, Darwish
Elhajji FW, Magee FA, et al. Clinical and economic impact
of contaminated blood cultures within the hospital setting. J
Hosp Infect 2011;77:233-6.
3. Goto M, Al-Hasan MN. Overall burden of bloodstream infection
and nosocomial bloodstream infection in North America and
Europe. Clin Microbiol Infect 2013;19: 501-9.
4. Lawless, J.F. Negative binomial and mixed Poisson regression.
Can J Stat 1992;15: 209-25.
5. Miller JM, Binnicker MJ, Campbell S, Carroll KC, Chapin KC,
Gilligan PH, et al. A guide to utilization of the microbiology
laboratory for diagnosis of infectious diseases: 2018 updated
by the Infectious Diseases Society of America and the American
Society for Microbiology. Clin Infect Dis 2018;67:813-6.
6. CDC.gov [Internet]. Atlanta : NHSN Patient Safety Component
Manual 2021; [cited 2021 March 1]. Available from: https://
www.cdc.gov/ncidod/dh
7. Benneyan JC, Lloyd RC, Plsek PE. Statistical process control
as a tool for research and healthcare improvement. Qual Saf
Health Care 2003;12:458-64.
8. Lalezari A, Cohen MJ, Svinik O, Tel-Zer O, Sinvani S, Abed
Al-Dayem Y, et al. A simplied blood culture sampling protocol
for reducing contamination and costs: a randomized controlled
trial. Clin Microbiol Infect 2020;26:470-74.
9. Yang Z, Xie M, Kuralmani V, Tsui KL. On the performance
of geometric charts with estimated control limits. J Qual
Technol 2002;34:448-58.
10. W K Luk. An outbreak of pseudobacteraemia caused by
Burkholderia pickettii: the critical role of an epidemiological
link. J Hosp Infect 1996;34:59-69.
11. Hashemizadeh Z, Bazargani A, Davarpanah MA. Blood Culture
Contamination in a Neonatal Intensive Care Unit in Shiraz,
Southwest-Central Iran. Med Princ Pract 2011; 20:133-6.
12. Chang C-J, Wu C-J, Hsu H-C, Wu C-H, Shih F-Y, Wang S-W,
et al. Factors Associated with Blood Culture Contamination
in the Emergency Department: Critical Illness, End-Stage Renal
Disease, and Old Age. PLoS ONE [Internet]. 2015 Oct 8 [cited
2020 Jan 14];10(10). Available from: https://www.ncbi.nlm.
nih.gov/pmc/articles/PMC4598129/
13. Washer LL, Chenoweth C, Kim H-W, Rogers MAM, Malani
AN, Riddell J, et al. Blood culture contamination: a randomized
trial evaluating the comparative eectiveness of 3 skin antiseptic
interventions. Infect Control Hosp Epidemiol 2013; 34:15-21.
14. Self WH, Spero T, Grijalva CG, McNaughton CD, Ashburn J,
Liu D . Reducing Blood Culture Contamination in the Emergency
Department: An Interrupted Time Series Quality Improvement
Study. Acad Emerg Med 2013;20:89-97.
15. Jame C, Bennayan. Performance of Number-Between g- Type
Statistical Control Charts for Monitoring Adverse Events.
Health Care Management 2001;4:319336.
Sangsuwan et al.
