Evaluation and remediation of bulk soap 
dispensers for biofilm
Authors: Lindsey A. Lorenz, Bradley D. Ramsay, 
Darla M. Goeres, Mathew W. Fields, Carrie A. 
Zapka, & David R. Macinga
NOTICE: This is an Accepted Manuscript of an article published in Biofouling on January 2012, 
available online: htp:/www.tandfonline.com/10.1080/08927014.2011.653637.
Lorenz LA, Ramsay BD, Goeres DM, Fields MW, Zapka CA, Macinga DR, "Evaluation and 
remediation of bulk soap dispensers for biofilm," Biofouling: The Journal of Bioadhesion and 
Biofilm Research, January 2012 28(1):99-109
Made available through Montana State University’s ScholarWorks 
scholarworks.montana.edu 
Recent studies evaluating bulk soap in public restroom soap dispensers have demonstrated up to 
25% of open refilable bulk-soap dispensers were contaminated with * 6log10(CFU ml71) heterotrophic bacteria. In this study, plastic counter-mounted, plastic wal-mounted and stainless 
steel wal-mounted dispensers were analyzed for suspended and biofilm bacteria using total cel and 
viable plate counts. Independent of dispenser type or construction material, the bulk soap was 
contaminated with 4–7 log10(CFU ml71) bacteria, while 4–6 log10(CFU cm72) biofilm bacteria were isolated from the inside surfaces of the dispensers (n ¼ 6). Dispenser remediation studies, 
including a 10 min soak with 5000 mg l71 sodium hypochlorite, were then conducted to determine 
the efficacy of cleaning and disinfectant procedures against established biofilms. The testing showed 
that contamination of the bulk soap returned to pre-test levels within 7–14 days. These results 
demonstrate biofilm is present in contaminated bulk-soap dispensers and remediation studies to 
clean and sanitize the dispensers are temporary.
Keywords: biofilm; bulk soap; soap dispensers; efficacy testing
Introduction
Hand washing has long been recognized to play an 
important role in public health (Garner and Favero 
1986), and is generaly accepted as an important practice 
to help prevent the spread of infectious microorganisms, 
which is especialy significant in the healthcare industry. 
Hand washing sinks and liquid soap are generaly 
provided to patrons of public restrooms to encourage 
good hand hygiene. Shared public bathrooms, however, 
can be a vector, con-tributing to the spread of 
pathogenic microorganisms (Mokhtari and Jaykus 
2009). As early as the 1960s, studies were published 
regarding significant surface contamination of bar soap 
(Bannan and Judge 1965; Kabara and Brady 1983). 
Liquid soap was eventualy recommended to be a more 
hygienic solution, and dispensers were developed to 
distribute liquid soaps (Graf et al. 1988; Chatman et al. 
2011).
Like bar soaps, liquid soap dispensers have been 
associated with microbial contamination issues. Reports 
dating back to the 1960s have linked bulk liquid hand 
soap and hand lotion contamination to nosocomial 
infections in hospital operating rooms and neonatal units 
(Morse et al. 1967; Archibald et al. 1997; Sartor et al. 
2000; Rabier et al. 2008; Buffet-Batailon et al. 2009). 
Washing with contaminated soap can leave more 
bacteria present on the hands after the washing event
than before, which undermines the effectiveness of hand 
washing (Sartor et al. 2000; Zapka et al. 2011). In 1986, 
the healthcare industry hand hygiene guidelines recog-
nized that ‘since liquid-soap containers can become 
contaminated and might serve as reservoirs of micro-
organisms, reusable liquid containers need to be cleaned 
when empty and refiled with fresh soap. Completely 
disposable containers obviate the need to empty and clean 
dispensers.’ (Garner and Favero 1986). In res-
ponsetothisguideline,the useof bulkhand 
soap dispensers is now rare in US healthcare setings. 
How-ever, these types of dispensers are stil common in 
public restrooms. Recent research has demonstrated that 
up to 25% of bulk hand soap dispensers from office 
buildings, health clubs, schools, food service centers, 
retail spaces and other locations are contaminated. 
Heterotrophic bacteria in contaminated soap averages 
6log10(CFU ml71), which is approximately 1000 times in excess of what industry guidelines recommend 
(Krowka and Bailey 2007; Chatman et al. 2011).
There are numerous unique dispenser designs but al 
include a reservoir area to store the soap, a mechanism to 
pump the soap out of the reservoir onto hands, and a way 
to refil the dispenser with new soap. Dispensers are 
constructed of metal or plastic and are typicaly semi-
permanently mounted to the wal or under the counter 
near the sink. Dispensers are
Evaluation and remediation of bulk soap dispensers for biofilm
Lindsey A. Lorenza, Bradley D. Ramsaya, Darla M. Goeresa*, Mathew W. Fieldsa, 
Carie A. Zapkab and David R. MacingabaCenter for Biofilm Enginering, Montana State University, Bozeman, Montana 59717, USA; bGOJO Industries, Inc., Akron, OH
44311, USA
designed to be refiled by one of two methods: bulk
refil and sealed soap refil. Bulk refil dispensers are
manualy refiled by pouring soap through an opening
in the top from a separate bulk soap refil botle,
commonly supplied in a 1 galon volume. These bulk
soap dispenser models typicaly have a built-in per-
manent nozzle through which soap is dispensed and is
not replaced under normal circumstances. Sealed soap
dispensing systems, in contrast, are typicaly refiled by
inserting a new bag or cartridge of soap that contains
a new built-in nozzle. As such, the nozzles in these
systems are replaced regularly and the soap does not
come into contact with the dispenser itself. Empty
cartridges are then either disposed or recycled.
Personal care and cosmetic products, such as soap,
are not expected to be sterile, but US manufacturers are
required by law to ensure that their products do not
present a hazard to consumers when they are used as
directed (Steinberg 2006). The Federal Food, Drug, and
Cosmetic Act ‘requires that successful preservation can
only be established if one considers al aspects of
development from concept and design through manu-
facturing to the last consumer use before disposal’ (Geis
2006). Industry guidelines suggest that to be safe, a
product should not contain any pathogens and that the
bacterial load should not exceed 1000 total bacteria per
gram or mililiter of product (Krowka and Bailey 2007).
In order to protect products from contamination during
use, soap manufacturers include preservatives in their
formulations and verify their performance by testing
that each newly-developed formulation effectively in-
hibits the growth of a range of microorganisms (Suton
2006). Liquid hand soaps, however, are perishable and
can become contaminated with microorganisms under
certain adverse circumstances, particularly when con-
sumers use or store the product in unintended ways that
are hostile to preservative efficacy (Geis 2006). Occa-
sionaly, products are sold that are either already
contaminated (intrinsic contamination) or that are
inherently susceptible to becoming contaminated be-
cause of poor formulation design. However, the
primary cause of failure of even a robust, wel-preserved
formulation is the introduction of contamination during
use of the product when a consumer intentionaly adds
water, mixes products, or stores the product in
inhospitable conditions, such as in warm or humid
places (extrinsic contamination) (Geis 2006). The design
of packaging and dispensing mechanisms used to store
and deliver products affects the probability that a
product wil become contaminated. Systems that have
an open design and that alow for increased opportunity
for consumers to manipulate the product inside are
inherently at greater risk of becoming contaminated as
compared to products with a closed design (Garner and
Favero 1986; Brannan and Dile 1990; Geis 2006).
Dispenser design and construction of soap packa-
ging is a critical factor to both the occurrence of
contamination and the chalenge of contamination
remediation. The likelihood of extrinsic contamination
is greatest when products are packaged, stored, or used
in a manner that alows for repeated introduction of
microorganisms from the consumer or the surrounding
environment (Brannan and Dile 1990; Geis 2006).
Dispenser designs, particularly those for wal-mounted
dispensers, do not take into consideration the potential
for microbial contamination, thus, cleaning is imprac-
tical because the dispensers are often securely bolted
into wals, making them difficult to remove. For this
reason, the same dispensers often remain in facilities
for many years. Some wal-mounted dispensers are
designed with a nozzle that is located centimeters
above the botom of the dispenser, rather than
dispensing the soap from the botom of the dispenser.
This design flaw ensures that the dispenser wil never
completely drain. Once the soap becomes contami-
nated, this serves to provide a reservoir of bacteria that
are uniquely adapted to survive in the soap environ-
ment. Also, some counter-mounted dispensers are sold
with one dispensing pump to be reused between botles
(Sartor et al. 2000). Once the pump becomes con-
taminated, it can transfer the bacteria between botles
(Graf et al. 1988).
Remediation of contaminated dispensers is one
option for reducing potential health risks to the general
public. There are no published research studies to date
that have determined if there is an effective way to
eliminate or reduce the contamination problem by
washing and/or sanitizing the dispenser. Furthermore,
even as far back as the late 1980s, biofilm was
suspected of being present in bulk soap dispensers
(Graf et al. 1988). Given that bacterial biofilm is
known to be more tolerant to disinfectants (Stewart
et al. 2000; Donlan and Costerton 2002; Smith and
Hunter 2008; Peeters et al. 2008), biofilms likely sur-
vive on internal surfaces in contact with soap. While
most published studies only tested the bulk soap com-
ing out of the dispenser for bacterial contamination,
the entire soap dispenser could be considered a micro-
bial habitat and should be examined. This examination
should include both the bulk soap for planktonic
contamination and the inner dispenser surfaces to test
for the presence of biofilm.
The objectives of this study were to test for the
presence of biofilm within dispensers colected from
public restrooms and to determine which organisms
were present, to understand the efficacy of cleaning and
disinfection procedures against established biofilm, and
to examine the recurrence of bulk soap contamination
folowing cleaning. Plastic counter-mounted, plastic
wal-mounted, and stainless steel (SS) wal-mounted
dispensers were analyzed for planktonic and biofilm
heterotrophic and coliform bacteria using viable plate
counts (VPC) and total cel counts (TCC). Isolated
bacterial colonies were identified using biochemical and
molecular profiling. Once the presence of biofilm within
dispensers was confirmed, several washing and sanitiz-
ing procedures were evaluated for their ability to
remediate contamination using both plastic and SS
wal-mounted dispensers.
Methods
Sampling dispensers for biofilm
Test dispenser information
Three counter-mounted plastic dispensers from a
shopping complex, two plastic wal-mounted dispensers
from an elementary school, and two SS wal-mounted
dispensers from a middle school and high school, al
located in Ohio, USA were evaluated. The dispensers
were sampled in the field and determined to be con-
taminated prior to being sent to the Center for Biofilm
Engineering (CBE) for analysis. The plastic dispensers
tested were designed with a top lid that completely lifted
open for refiling the dispenser with new soap. The SS
dispensers were designed with a smal, hinged lid that is
lifted to refil the dispenser with soap.
Experimental design
A schematic of the process used to sample the refilable
soap dispensers for viable and total cels is found in
Figure 1. Dispensers were visualy inspected and
imaged after arrival from the colection site. Three
samples were colected from each dispenser: bulk soap
to enumerate viable, planktonic bacteria (CFU ml71);
rinse water to enumerate loosely-atached, surface-
associated bacteria (CFU cm72); and inner surface
scrapings to determine the density of atached, biofilm
bacteria (CFU cm72). In addition, TCC were deter-
mined for each sample type colected, as described
below.
Determination of planktonic bacteria
For the plastic counter- and wal-mounted dispensers,
the soap was drained through the nozzle into a sterile
beaker containing 220 g of 3 mm glass beads. For the
SS dispensers, the soap was drained into a sterile glass
beaker, and after vigorous mixing, a 10 ml aliquot was
added to a 50 ml conical vial containing 10 g of glass
beads.
Determination of loosely-atached bacteria
After the soap was removed from the dispenser, 100 ml
of sterile phosphate buffered water was added to the
dispenser and swirled around to remove any loosely-
atached bacteria. For the plastic dispensers, the rinse
water was drained into a sterile beaker containing 60 g
of glass beads. For the SS dispensers, the rinse water
was drained into a beaker and a 10 ml aliquot was
colected for culturing.
Determination of strongly-atached bacteria
For the plastic dispensers, the entire inside of the dis-
penser was scraped with a Teflon scraper and then rinsed
with 100 ml of DeyEngley (D/E) Neutralizing Broth.
Figure 1. Schematic of the experimental design used to analyze bacteria suspended in the soap and loosely- or strongly-atached
to the inside surfaces of contaminated bulk soap dispensers. Three samples were colected and analyzed for viable cels (VPC) and
total cels (TCC).D¼sample disaggregation steps.
The D/E broth was poured into a beaker containing 185
g of glass beads. The scrape and rinse procedure was
completed three times and al rinses were combined.
For the SS dispensers, 150 ml of cold phosphate
buffered water were added to the dispenser. The
dispenser was shaken vigorously for 5 min and the
inside surfaces of the dispenser that were accessible
were scraped with a sterile Teflon scraper.
Disaggregation and plating methods
Al samples were neutralized with D/E broth. Three
cycles of sonication and vortexing (1 min each)
folowed to disaggregate the biofilm. Sterile glass beads
were included to aid in biofilm disaggregation. The
efficiency of this method was confirmed microscopi-
caly. Samples were serialy diluted and 1 ml aliquots
were plated on both R2A and MacConkey agar. The
R2A plates were incubated at room temperature for 7
days and the MacConkey plates were incubated at
368C for a period of 24–72 h. In addition, 1 ml of the
disaggregated, undiluted soap was plated.
For TCC, an additional 1 ml aliquot from the
diluted sample was pipeted onto a 0.2mm membrane.
LIVE/DEADBacLight Bacterial Viability Kit stain
(Invitrogen #L7012, Carlsbad, CA) was added, in-
cubated for 15 min in the dark, and after rinsing, the
membrane was placed on a glass slide. The total cel
count slides were imaged on a Nikon Eclipse E800
microscope with a FITC cube (ex 480/15, DM 505, em
535/20) for the green and a TRITC cube (ex 546/5,
DM 575, em 590 LP) for the red. Images were analyzed
for total cels regardless of the color the cel stained. A
scan of 20 fields per slide was performed and this
information was processed for total counts per sample
dilution using Metamorph, v7.6.4 Software (MDS
Analytical Technologies, Sunnyvale, CA).
Identification of bacterial isolates
Colonies colected from the three sample types that
expressed a unique morphology were streaked for
isolation and sent to an outside laboratory (Medical
Laboratory Services, Inc., Bozeman, MT) for bacterial
identification based upon biochemical profiling. Iden-
tification of bacterial isolates was confirmed by
sequence determination of the V1–V3 region of the
SSU rRNA gene. The SSU rRNA gene was amplified
with previously described primers FD1 and 1540R and
sequenced with 529R via capilary Sanger sequencing
(Ye et al. 2004; Hwang et al. 2009). Sequences were
identified using the BLASTn algorithm through NCBI
(htp:/ncbi.nlm.nih.gov/blast).
For the plastic wal-mounted dispensers, the field
identifications (from historical data) and laboratory
identifications were determined using biochemical
profiling and molecular analysis. For the SS wal-
mounted dispensers, the laboratory identifications
were determined biochemicaly and the isolated colo-
nies used in the biochemical identifications were sent in
for molecular testing to provide direct comparisons
between the two methods.
Molecular analysis of whole biofilm community
Using separate dispensers from above, two plastic wal-
mounted and two SS wal-mounted dispensers were
sampled to determine microbial diversity of biofilm
within the dispensers. For each dispenser tested, bulk
soap was removed and 100 ml cold, sterile 1X PBS were
added. The inside surfaces of the dispenser were scraped
into the PBS and transferred to 50 ml conical centrifuge
tubes. Biomass was colectedviacentrifugation and
multiple pelets from the same dispenser were combined
until al biomass was in a single pelet for each sample.
Pelets were resuspended in 10 ml of PowerBead solution
and transferred into sterile mortars with sand. Samples
were flash frozen with liquid nitrogen and ground with
pestles three times. The whole sample was colected into
PowerBead tubes and nucleic acid extraction was done
according to the manufacturer’s instructions with the
PowerMax Soil DNA Extraction Kit (MO BIO, Inc.,
Carlsbad, CA). The extracted DNA was amplified as
above with primers FD1 and 1540R using PCR program
808C1:30,948C 2:00, 25 cycles of (948C0:30,588C1:00,
728C1:00),728C7:00folowedby48C hold. Appro-
priately-sized DNA was cloned into plasmid pCR2.1-
TOPO (plastic) or pCR4-TOPO (SS), transformed into
competentE.coliDH5aandplatedonLB-Kan50plates
as per the manufacturer’s instructions (Invitrogen, Inc.,
Carlsbad, CA). Transformants were screened for appro-
priately-sized inserts using primers M13F and M13R.
Ninety-six M13 amplicons were submited from each
dispenser for Sanger sequencing using primer 529R.
Sequence libraries were checked for chimeras and
identified as described above.
Dispenser imaging
Prior to any sampling steps, the dispensers were
visualy inspected and various outside and/or remo-
vable dispenser pieces were imaged using a Nikon
SMZ1500 stereo zoom microscope.
Experimental design of remediation study
Washing studies were conducted on plastic and SS
wal-mounted dispensers. Five plastic wal-mounted
dispensers from an elementary school in Ohio were
used in the first set of experiments. Some of these
dispensers were previously used to investigate hand
transfer of contaminants in a different study (Zapka
et al. 2011). Eight SS wal-mounted dispensers from a
school district in New Jersey were used in the second
set of experiments. Each experiment included a posi-
tive control (randomly chosen dispenser that had
tested positive for contamination in the bulk soap)
and a negative control (a new dispenser that had never
tested positive for bacteria in the bulk soap). The
experiments were performed in triplicate and control
dispensers remained the same for each of three
experimental repeats. The remaining dispensers used
in the studies had al tested positive for viable bacteria
(at least 3 log10(CFU ml71) in the bulk soap prior to
commencing each washing experiment. The washing
procedure tested on each dispenser was randomly
assigned before every experiment.
The washing procedures were designed to vary in
difficulty and to utilize products that would be readily
available to any cleaning personnel, including the use of
tap water. Just prior to washing the dispenser, a sample
of the bulk soap was colected and analyzed for
heterotrophic bacteria. Samples from plastic dispensers
were neutralized with D/E Neutralizing Broth and
disaggregated and plated on R2A, while samples from
SS dispensers were neutralized with a modified Buter-
field’s phosphate buffer solution containing lecithin,
polysorbate 80, KH2PO4, K2HPO4, Na2S2O3 5H2O,
Tamol SN, and Triton X-100 (BPBþ Neutralizer)
(Beausoleil 1999), folowed by disaggregation and
plating on TSA. The control dispensers were then
drained and refiled with an antibacterial soap labeled
to contain triclosan (percent triclosan not listed on the
label) for the plastic dispensers or a bland (non-
antimicrobial) soap for the SS dispensers. The soap
formulation used to fil each dispenser was consistent
with the formulation used to fil that dispenser in the
field. The test dispensers were washed with sodium
hypochlorite (5000 mg l71), a quaternary ammonium
compound-containing disinfectant (Ecolab Oasis 146
Multi-Quat Sanitizer, 8 ml l71), or a mildew remover
(Tilex Mildew Root Penetrator & Remover, 24,000 mg
l71sodium hypochlorite, active ingredient), as depicted
in Figure 2. They were then filed with the appropriate
soaps as described above. The bulk soap from al the
dispensers was then sampled immediately after filing
and for up to 2 weeks or until the population reached
pre-test levels. Both the fresh soap and tap water were
platedandtestedoneachexperimentdayforviablecels.
Results
Planktonic and biofilm contamination
Bulk soap from contaminated dispensers harbored
between 3.7 to 6.7 log10(CFU ml71) of viable coliform
and heterotrophic bacteria and between 6.9 to 8.0
log10(CFU ml71) total cels (Figure 3). Soap from
plastic wal-mount dispensers had the highest density
of viable planktonic bacteria (5.4 to 6.7 log10(CFU
ml71) while plastic counter-mounted dispensers con-
tained the lowest density (3.7 to 4.9 log10(CFU ml71)
and SS wal-mounted dispensers contained an inter-
mediate density (4.9 to 5.2 log10(CFU ml71). The
TCC in the soap were* 1 to 3 logs greater than the
viable counts for al of the dispensers.
Loosely- and strongly-adhered viable coliform or
heterotrophic cels were present at densities between
3.3 to 6.4 log10(CFU cm72) in al dispensers (Figure 4).
The SS wal-mounted dispensers had the highest
density of surface-associated viable bacteria (5.1 to
6.4 log10(CFU cm72) as compared to the plastic
counter-mounted and wal-mounted (3.3-5.8 log10(C-
FU cm72) dispensers. The TCC from the loosely- and
strongly adhered bacteria were generaly greater than
the loosely- and strongly-adhered viable bacteria,
except for the strongly-adhered bacteria from the
plastic wal-mounted dispenser. For the majority of
dispensers, slightly more strongly-adhered and total
cel count bacteria were recovered than loosely-
adhered bacteria, except for the plastic counter-
mounted dispensers, in which much higher densities
of loosely-atached bacteria and TCC were recovered
(6.3 to 6.9 as compared to the strongly-associated
bacteria at 4.6 to 5.1).
Bacterial identification
The colonies recovered from the plastic counter-
mounted dispensers were identified through bio-
chemical profiles asKlebsiela oxytocaandKluyvera
ascorbata, both of which are Gram-negative opportu-
nistic pathogens. The bacteria identified in the
plastic wal-mounted dispensers were commonly
Gram-negative, presumptive opportunistic pathogens
(egProvidencia, Citrobacter, Klebsiela, Serratia, and
Pseudomonas) (Table 1). Bacterial populations were
also identifiedviaclone libraries of SSU rRNA gene
sequences. The isolates identified with both biochem-
ical and molecular techniques revealed similar identi-
fications at the genus level, although not surprisingly
the clone library data identified potential organisms
that were not cultivated. The bacteria identified in the
SS wal-mounted dispensers were consistent with that
observed for the other dispensers (Table 2). In total,
the SS dispensers contained bacteria from five unique
genera that includedPseudomonas, Providencia, Serra-
tia, StenotrophomonasandAcinetobacter. Interestingly,
the molecular data did not reveal additional sequences
that were not cultivated from the SS dispensers. In
previous unpublished work, historical data indicated
that the dominant colony types in each dispenser were
Pseudomonas aeruginosaandSerratia liquefaciens. This
research confirmed that these genera were present in
the respective dispensers but did not confirm that they
were the dominant colony types. Bacterial isolates
obtained from SS dispensers were also identified using
both SSU rRNA gene sequencing biochemical profil-
ing to compare the two techniques. The results from
the comparison revealed equivalent identities at the
genus level for al but one of 14 isolates.
Effectiveness of dispenser remediation techniques
The heterotrophic plate count results of the dispenser
washing experiments are shown in Figures 5 and 6 for
the plastic wal-mounted and SS wal-mounted dis-
pensers, respectively. In Figure 5, the standard error of
the mean (SEM) for the hot water rinse procedure was
0.10 and 0.47 on day 0 and day 4, as averaged over the
three experiments. The SEM for the hot water rinse
and scrub procedure on day 0 was 0.28 and ranged
from 0.07 to 0.53 on days 0, 4, and 7 for the scrub and
Figure 2. Schematic of the experimental design used to evaluate the effectiveness of dispenser remediation procedures.
Procedures 1, 2, and 3 were folowed for plastic wal-mounted dispensers (solid lines). Procedures 1–6 were tested for the SS
dispensers (dashed lines). The doted line denotes the control dispenser protocol.
sodium hypochlorite rinse washing procedure, over
the three experiments. The triplicate experiments for
these dispensers were not conducted consistently with
respect to the frequency of plating. In experiment 1,
the dispensers were only plated on day 0 and 14,
whereas in experiment 2, they were plated on days 0, 2,
4, 7, and 10. The dispensers were plated on days 0, 1, 4,
and 7 for experiment 3 and for al experiments, plating
was discontinued once the bacterial counts returned to
pre-test contamination levels. For these reasons, the
SEM could not be calculated for al dispensers and al
experiments for each day.
In Figure 6, the SEM for the hot water rinse
procedure over the three experiments was 0.30, 0.14,
and 0.20 for days 0, 2, and 4. The SEM could not be
calculated for day 7 because some of the dispensers
had reached their pre-test contamination levels and
plating was discontinued. The SEM for the hot water
rinse and scrub procedure was 0.27, 0.25, 0.15, and
0.15 for days 0, 2, 4, and 7 averaged over three
experiment replicates. For the scrub and sodium
hypochlorite rinse procedure, the SEM was 0.25,
0.52, 1.01, and 0.34 for days 0, 2, 4, and 7. The SEM
could not be calculated on day 10 because some of the
dispensers had already reached their pre-test contam-
ination levels. For the 10 min sodium hypochlorite
soak procedure and for the 10 min quat soak
procedure, the SEM was 0.26, 0.31, 0.71, and 0.22,
and 0.28, 0.15, 0.87, and 0.16 on days 0, 2, 4, and 7,
respectively. For the 10 min mildew remover soak
washing procedure, the SEM was 0.34, 0.63, 1.55, and
1.24 for days 0, 2, 4, and 7. The SEM could not be
calculated for days 10 and 14 because some of the
dispensers had already reached their pre-test contam-
ination levels and plating was discontinued.
The dispensers initialy contained 4.3 to 6.0
log10(CFU ml71) in the bulk soap dispensed after
cleaning. Rinsing the dispenser with hot water, with or
without scrubbing, did litle to reduce the contamina-
tion levels in the soap. Based upon industry guidelines
that suggest a microbial load limit of 1000 CFU ml71,
these soaps would be considered contaminated within
1–2 days after performing the remediation procedures.
Figure 3. Coliform (COL), heterotrophic (HPC), and total
cel count (TOTAL) results from the bulk soap for plastic
counter-mount, plastic wal-mount, and SS wal-mount
dispensers (n¼2 of each). The black solid line connects
the mean log10(CFU ml71) of the data points.
Figure 4. Coliform (COL), heterotrophic (HPC), and total
cel count (TOTAL) results from the loosely-atached (Panel
A) and strongly-adhered (Panel B) sampling steps for plastic
counter-mounted, plastic wal-mounted, and SS wal-
mounted dispensers (n¼2 of each). The black solid line
connects the mean log10(CFU cm72) of the data points.
The most effective remediation treatments were the
sodium hypochlorite soak, sodium hypochlorite rinse
and scrub, and the mildew remover soak, which were
al able to reduce the bacterial contamination densities
to below the 1000 CFU ml71threshold for* 4to5
days after treatment. However, the levels in the soap
continued to increase and returned to pre-remediation
levels after only 7 to 14 days post-remediation. The
quat soak did litle to decrease contamination levels,
and on average, only decreased levels below the 3
log10(CFU ml71) microbial load limit for 2 days, post-
treatment.
When considering the individual data points, the
10 min mildew remover soap procedure in experiment
3 took 10 days to recover beyond a 3 log10(CFU ml71)
level. Interestingly, it had reached that level after 4
days in the first two experiments.
Table 1. Bacteria identified in wal-mounted plastic dis-
pensers.
Organisms identified
Field
identified
Lab
identified
Clone
library
analysis
Providencia retgeri þ þ þ
Pseudomonassp.
P. aeruginosa þ þ þ
P. fluorescens þ
P. luteola þ
P. stuzeri þ
Citrobactersp. þ
C. koseri þ
C. freundi þ
Serratiasp.
S. oderifera þ
S. liquefaciens þ
S. rubidae þ þ
Stenotrophomonassp. þ
S. maltophilia þ
Klebsiela pneumoniae þ
Aeromonas hydrophilia þ
Burkholderia cepacia þ
Enterobactersp. þ
E. cloacae þ
Achromobacter xylosoxidans þ
Alcaligenes xylosoxidans þ
Curvibactersp. þ
Leptothrixsp. þ
Pelomonassp. þ
Delftia acidovorans þ
Rubribacter xylanophilus þ
Table 2. Bacteria identified in SS wal-mounted dispensers.
Organisms identified
16S ID
of isolates
Biochemical
ID of
isolates
Clone
library
analysis
Pseudomonassp. þ
P. aeruginosa þ þ
P. fluorescens/putida þ
Providenciasp. þ þ
P. vericola þ
P. retgeri þ þ
Serratiasp. þ þ
S. marcescens þ
S. liquefaciens þ
Stenotrophomonassp. þ þ
S. maltophilia þ
Acinetobacter lwoffii þ
Alcaligenes/
Achromobactersp.
þ
Figure 5. HPC results for plastic wal-mounted dispenser
washing studies, averaged over three experiments..¼hot
water rinse procedure; & ¼hot water rinse and scrub
procedure; '¼scrub and sodium hypochlorite rinse
washing procedure. The solid horizontal line at 3
log10(CFU ml71) depicts the cosmetic industry guidelinerecommendation.
Figure 6. HPC results for SS wal-mounted dispenser
washing studies, averaged over three experiments..¼hot
water rinse procedure; & ¼hot water rinse and scrub
procedure; '¼scrub and sodium hypochlorite rinse
washing procedure, ¼10 min sodium hypochlorite soak
procedure;¤¼10 min quat soak procedure;}¼10 min
mildew remover soak washing procedure. The dispenser bulk
soap was sampled until the populations reached pre-test
contamination levels. The solid horizontal line at 3
log10(CFU ml71) depicts the cosmetic industry guidelinerecommendation.
The effectiveness of the three remediation methods
performed on both the plastic and SS dispensers (hot
water rinse, hot water rinse and scrub, and sodium
hypochlorite rinse and scrub) was not significantly
different depending on dispenser type. Positive control
dispensers, which were simply drained of soap and
refiled with fresh soap, maintained their contamination
levels at approximately 5 log10(CFU ml71), and no
bacteria were detected from the negative control dis-
pensers throughout the experiments (data not shown).
Discussion
The results of this study demonstrate that open, bulk-
refilable soap dispensers found to contain contami-
nated soap also contained bacterial biofilms. Three
samples were colected from each dispenser to assess
the bacterial contamination, viz. bulk soap, loosely-
atached cels, and biofilm. Analyzing the soap for
bacteria in addition to the surface samples alowed for
comparisons between historical findings, field data,
and the present laboratory evaluation.
The density of surface-associated bacteria in SS wal-
mounted dispensers was up to ten-fold greater than that
seen for the other two dispenser types. This is interesting
because the bacterial density in the soap was slightly
greater than that recovered in the plastic counter-
mounted dispensers and slightly less than the bacteria
recovered from the soap in the plastic wal-mounted
dispensers. This result suggests that there is no direct
correlation between biofilm density in a dispenser and
the level of contamination in the bulk soap.
Previous reports suggest that bulk liquid samples
are not necessarily predictive of the microbial health of
the system (Goeres 2010). In general, if the bulk soap is
contaminated, then biofilm is also most likely present
in the dispenser. Perhaps the most interesting case
would be to determine whether dispensers containing
no bulk soap contamination stil contain biofilm.
Additional factors that would be interesting to include
in a correlation study are the type of soap, the location
of dispenser, and the use patern.
For this study, the type of dispenser (plastic wal-
mounted, plastic counter-mounted and SS wal-
mounted) did not appear to be a significant factor,
although a slightly greater diversity of organisms was
detected in the plastic dispensers. This is an interesting
result given the design of the SS dispensers, which does
not alow for the dispenser to ever completely empty.
Bacterial isolates from the soap were almost exclu-
sively Gram-negative. While isolates were identified to at
least the genus level, the identifications provided a
qualitative description of organisms contaminating the
dispensers but did not serve to quantify each species. In
most cases, molecular typing of the isolates provided
similar results to the biochemical typing. Identifications
from both methods are limited to matching the bio-
chemical profile or the sequence to an organism already
in the database. Biochemical profiling of environmental
isolates is particularly limited due to the extremely great
diversity of organisms which have not yet been
characterized as wel as those multiple species which
are similar, if not identical, in the limited size of the array
used for profiling. While the bacterial diversity was
relatively low compared to other environments, 16S
rRNA gene sequencing demonstrated the presence of
organisms not detectedviacultivation-based techniques
in plastic dispensers. The same was not true for the SS
dispensers. Identified isolates are consistent with organ-
isms previously reported to have been isolated from
liquid soap (Chatman et al. 2011; Zapka et al. 2011).
The molecular data can be used to further direct
cultivation methods in order to isolate a broader
diversity of the present microbiota, which could be
useful information when crafting new formulations of
soap. Intentional incubation of isolates already known to
be wel-suited for survival in soaps during the formula-
tion phase would give insight into the ability of the new
formulation to resist bacterial growth. Future work
could include molecular techniques that differentiate
bacterial populations in the bulk soapvsbiofilm
populations.
Inclusion of microscopy in these experiments
proved to be useful for two reasons. First, the TCC
demonstrated that only a fraction of the bacteria were
recovered by the VPC. On average, the TCC were 1 to 2
log10(CFU ml71) higher than the VPC, indicating the
presence of a population that was either non-viable or
non-culturable by the plating techniques used in this
study. Second, microscopy demonstrated whether or
not the disaggregation method was adequate (Figure 7).
Figure 7. Total cel count image displaying biofilm
clumping when disaggregation was inadequate. X100.
Bar¼10 um.
The physical properties of soap makes it chalenging to
disaggregate cel clusters; it foams when homogenized
and is difficult to vortex vigorously, so microscopy was
an important means by which to assess the disaggrega-
tion method used. Improper disaggregation wil result in
an underestimate of the viable cels present in a sample
(Hamilton et al. 2009). Previously published results
obtained without using disaggregation techniques
showed that contaminated bulk soap in public re-
strooms contains an average of 6 log10(CFU ml71)of
heterotrophic bacteria, therefore, they may have under-
estimated the true levels of viable bacteria in the soaps
(Chatman et al. 2011). Another interesting use of the
imaging from the dispensers was to visualy record that
the dispensers often contained substances that presum-
ably did not originate from the soap (Figure 8).
Once a biofilm has established on a surface,
cleaning and eradicating the biofilm from that surface
becomes a chalenge, as the dispenser remediation
experiments demonstrated. The ineffectiveness of
washing soap bottles dates back to the 1960s, so these
findings are not surprising (Burdon and Whitby 1967).
The present study showed that even soaking the
dispensers with sodium hypochlorite, a quat, or with
a ful strength mildew remover for 10 min before
adding new soap, was ineffective at eradicating biofilm.
Because the soap used to refil each dispenser
contained no detectable bacteria, the results demon-
strated that the recovery of bacterial populations in the
bulk soap resulted from dispersal of bacteria from
biofilms present inside the dispensers. The rate of
recolonization was inconsistent between replicates and
likely represents a host of different factors including
density of the biofilm, age of the biofilm, species
composition of the biofilm, and quality of disruption
of the biofilm during disinfection. The slowest recovery
took 14 days to reach pre-test contamination levels.
This particular dispenser received the mildew remover
treatment in experiment 3, where the recovery was 14
days, but this dispenser also received that treatment in
experiment 1 and received the sodium hypochlorite
rinse treatment in experiment 2. The mildew remover
contains 24,000 mg l71sodium hypochlorite. It is
conceivable that the two mildew sodium hypochlorite
treatments, coupled with an approximate 5,000 mg l71
sodium hypochlorite rinse treatment, al occurring
within just under 2 months, were able to decrease the
biofilm counts and delay regrowth and contamination,
but stil failed to completely eradicate the biofilm.
The soap dispenser remediation procedures eval-
uated in this study were very time and labor intensive
and would not realisticaly be utilized by a custodial
staff, especialy in a facility with multiple dispensers to
maintain. Furthermore, the trials conducted in tripli-
cate were completed in rather quick succession, some-
times with just a week between replicate experiments.
A custodian would be very unlikely to add an every-
other-week soap dispenser cleaning regimen to an
already long list of cleaning duties. Finaly, the design
of the dispenser systems contributes to the chalenges
of keeping them clean. They are composed of intricate
pieces that are difficult to reach with a scrubbing
brush. For instance, some of the top openings are quite
smal, making it difficult to use a scrubbing brush or to
get into them at al. Bulk soap dispensers are
constructed of many materials including plastics, SS,
and rubber (gaskets). SS and rubber are incompatible
with high level concentrations of sodium hypochlorite,
which makes continuous cleaning of these materials
with such disinfectants impractical, as the dispenser
components wil begin to corrode or deteriorate.
It is possible that dispenser design guidelines could
be writen to facilitate easier cleaning and disinfecting
protocols for bulk soap dispensers. The SS wal-
mounted dispensers, for example, had an inefficient
valve placement on the front of the dispenser, about
2.5 cm above the botom, leaving a constant reservoir
of soap. Valve systems that are both easily replaceable
and not economicaly prohibitive would eliminate the
need to clean intricate and delicate valve components.
It is important to consider both the potential for
contamination and the ease of cleaning a system as
design parameters for a dispenser. As with any
environment where microbial contamination could be
a concern, including dispenser systems, these consid-
erations must be evaluated in the engineering design.
Conclusions
Bulk soap dispensers were shown to be highly
contaminated, both by bacteria in the soap, and also
Figure 8. Stereoscope image of inner dispensing tube of a
plastic counter-mounted dispenser coated with unknown
brown substance. 7.5X.
by biofilm bacteria atached to the inner dispenser
surfaces. The bacteria identified were consistent with
those typicaly found in cosmetics/soap environments,
as determined by both culture- and molecular-based
identification analyses. The remediation effectiveness
experiments demonstrated that, due to biofilm at-
tached to the dispenser surfaces, even cleaning with
highly concentrated disinfectants does not eliminate
the bacterial populations that are adapted to live in the
soap environment.
Acknowledgement
This project was funded by GOJO Industries, Inc.
References
Archibald L, Corl A, Shah B, Schulte M, Arduino M,
Aguero S, Fisher D, Stechenberg B, Banerjee S, Jarvis W.
1997.Serratia marcescens outbreak associated with
extrinsic contamination of 1% chlorxylenol soap. Infect
Cont Hosp Ep 18:704–709.
Bannan E, Judge L. 1965. Bacteriological studies relating to
handwashing. Am J Public Health Nations Health
55:915–922.
Beausoleil C. 1999. A guide for validation of neutralizer
systems used in topical antimicrobial efficacy evaluations.
In: Paulson DS, editor. Handbook of topical antimicro-
bials. New York (NY): Marcel Dekker, Inc. p. 365–376.
Brannan D, Dile J. 1990. Type of closure prevents microbial
contamination of cosmetics during consumer use. Appl
Environ Microb 56:1476–1479.
Buffet-Batailon S, Rabier V, Be´tre´mieux P, Beuche´e A,
Bauer M, Pladys P, Le Gal E, Cormier M, Jolivet-
Gougeon A. 2009. Outbreak ofSerratia marcescensin a
neonatal intensive care unit: contaminated unmedicated
liquid soap and risk factors. J Hosp Infect 72:17–22.
Burdon D, Whitby J. 1967. Contamination of hospital
disinfectants withPseudomonas species. Brit Med J
2:153–155.
Chatman M, Maxwel S, Gerba C. 2011. Occurrence of
heterotrophic and coliform bacteria in liquid hand soaps
from bulk refilable dispensers in public facilities. J
Environ Health 73:26–29.
Donlan R, Costerton JW. 2002. Biofilms: survival mechan-
isms of clinicaly relevant microorganisms. Clin Micro-
biol Rev 15:167–193.
Garner J, Favero M. 1986. CDC guideline for handwashing
and hospital environmental control. Infect Control
7:231–243.
Geis P. 2006. Preservation strategies. In: Geis P, editor.
Cosmetic microbiology. Boca Raton (FL): Taylor &
Francis Group. p. 163–175.
Goeres DM. 2010. Understanding the importance of biofilm
growth in hot tubs. In: Paulson D, editor. Applied
biomedical microbiology. Boca Raton (FL): Taylor &
Francis Group. p. 133–148.
Graf W, Kersch D, Scherzer G. 1988. Microbial contamina-
tion of wal-mounted liquid soaps dispensers with single-
use botles. [Translated from German]. Zbl Bakt Hyg B
186:166–179.
Hamilton M, Buckingham-Meyer K, Goeres D. 2009.
Checking the validity of the harvesting and disaggregat-
ing steps in laboratory tests of surface disinfectants.
J AOAC Int 92:1755–1762.
Hwang C, Wu W, Gentry T, Carley J, Corbin G, Carrol S,
Watson D, Jardine P, Zhou J, Criddle C, et al. 2009.
Bacterial community succession duringin situuranium
bioremediation: spatial similarities along controled flow
paths. ISME J 3:47–64.
Kabara J, Brady M. 1983. Contamination of bar soaps under
‘in-use’ conditions. J Environ Pathol Tox 5:1–14.
Krowka J, Bailey J. 2007. Establishing microbial quality of
cosmetic products. CTFA microbiology guidelines 2007
[internet]. [Edition unknown]. Washington (DC): Cos-
metic, Toiletry, and Fragrance Association; [cited 2011
Nov 3]. Available from: eservices.personalcarecouncil.
org/site/vision_Bookstore_SplashPage2.htm
Mokhtari A, Jaykus L. 2009. Quantitative exposure model
for the transmission of norovirus in retail food prepara-
tion. Int J Food Microbiol 133:38–47.
Morse L, Wiliams H, Grenn F, Eldridge E, Rotia J. 1967.
Septicemia due toKlebsiela pneumoniaoriginating from
a hand-cream dispenser. N Engl J Med 277:472–473.
Peeters E, Nelis HJ, Coenye T. 2008. Evaluation of the
efficacy of disinfection procedures againstBurkholderia
cenocepaciabiofilms. J Hosp Infect 70:361–368.
Rabier V, Batailon S, Jolivet-Gougeon A, Chapplain J-M,
Beuche´eA,Be´tre´mieux P. 2008. Hand washing soap as a
source of neonatalSerratia marcescensoutbreak. Acta
Paediatr 97:1381–1385.
Sartor C, Jacomo V, Duvivier C, Tissot-Dupont H, Sambuc
R, Drancourt M. 2000. NosocomialSerratia marcescens
infections associated with extrinsic contamination of a
liquid nonmedicated soap. Infect Cont Hosp Ep 21:196–
199.
Smith K, Hunter I. 2008. Efficacy of common hospital
biocides with biofilms of multi-drug resistant clinical
isolates. J Med Microbiol 57:966–973.
Steinberg DC. 2006. Preservatives for cosmetics. 2nd ed.
Carol Stream (IL): Alured Publishing Corporation. 137
pp.
Stewart P, McFeters G, Huang C. 2000. Biofilm control by
antimicrobial agents. In: Bryers J, editor. Biofilms II:
process analysis and applications. New York (NY):
Wiley-Liss. p. 373–405.
Suton S. 2006. Antimicrobial preservative efficacy and
microbial content testing. In: Geis P, editor. Cosmetic
microbiology. Boca Raton (FL): Taylor & Francis
Group. p. 112–139.
Ye Q, Roh Y, Carrol S, Blair B, Zhou J, Zhang C, Fields M.
2004. Alkaline anaerobic respiration: isolation and
characterizationof a novel alkaliphilic and metal-
reducing bacterium. Appl Environ Microb 70:5595–5602.
Zapka C, Campbel E, Maxwel S, Gerba C, Dolan M,
Arbogast J, Macinga D. 2011. Bacterial hand contam-
ination and transfer after use of contaminated bulk-soap-
refilable dispensers. Appl Environ Microbiol 77:2898–
2904.