Journal of Applied Microbiology ISSN 1364-5072
ORIGINAL ARTICLE
Application of rapid microbiological methods for the risk
assessment of controlled biopharmaceutical environments
T. Sandle, C. Leavy, H. Jindal and R. Rhodes
Bio Products Laboratory, Elstree, UK
Keywords
cleanrooms, environmental monitoring,
IMD-Aâ, particle counting, real-time
monitoring, spectrophotometric, viable but
nonculturable.
Correspondence
Tim Sandle, Bio Products Laboratory, Dagger
Lane, Elstree, Hertfordshire, UK WD6 3BX.
E-mail: timsandle@btinternet.com
2014/0020: received 5 January 2014, revised
16 February 2014 and accepted 20 February
2014
doi:10.1111/jam.12487
Introduction
Cleanrooms and controlled environments within pharma-
ceutical facilities need to have robust systems of environ-
mental control to minimize the generation of particulate
and microbiological contamination. Such control mea-
sures are also required to prevent the transfer of any
Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
Abstract
Aims: To assess the different operational states within a biopharmaceutical
grade clean room, using a rapid microbiological method. The method was a
novel system, based on spectrometry, designed for sampling, discriminating,
and enumerating airborne particles. Central to the study was the aim to
determine the microbiological levels as a clean room went from standard use
through maintenance and shutdown, disinfection, and then back to standard
use. The objective was to evaluate whether a rapid method could replace
conventional environmental monitoring using growth-based media.
Methods and Results: The instrument evacuated was a BioVigilant IMD-Aâ
System, which is a real-time and continuous monitoring technology based on
optical spectroscopy that can differentiate between biological particles and inert
ones (biological particles expressed as bio-counts based on the detection of
microbial metabolites). The results indicated that certain activities lead to a
high generation of biological particles and in showing an increase over the
baseline, would be regarded as presenting a microbiological risk to the
cleanroom. These activities include removing HEPA filter grilles, turning off an
air handing unit, and tasks which requires an active personnel presence, such
as cleaning and disinfection.
Conclusions: The optical instrument can be used to process sufficient
information, so that clean rooms can be returned to use following a period of
unexpected downtime or following maintenance without the need to wait for
the results from growth-based methods. As such, this type of rapid
microbiological method is worth exploring further for clean room air
monitoring.
Significance and Impact of the Study: Few studies have been undertaken
which examine air-monitoring devices that can both enumerate and
discriminate particulates, in a volume of air as ‘inert’ or ‘biological’. This study
extends this limited field. Furthermore, the data collected in relation to
cleanrooms is of interest in helping microbiologists understand that risks posed
by different activities in relation to clean air-handling systems and personnel
particle shedding.
contamination into a critical area and to protect pharma-
ceutical products. Whilst the emphasis should also be
with environmental control (through an appropriately
designed heating, ventilation and air conditioning system
(HVAC) and strict personnel behavioural disciplines)
(Sandle and Saghee 2013), understanding the extent
to which control is being maintained can only be
1
Application of rapid microbiological methods
determined through a robust environmental monitoring
programme.
Conventional environmental monitoring programmes
use a combination of nonviable particle counting and tra-
ditional microbiological sampling methods (including
volumetric samplers and settle plates; surface sampling
using swabs and contact plates; and, where appropriate,
samples of personnel suits and gloved hands). To address
these considerations, many sections of the industry have
moved towards the adoption of risk-based approaches
(Sandle 2012a).
Risk assessment and risk management have been
advanced through International Conference on Harmoni-
sation guidelines, most prominently ICH Q9 ‘Quality
Risk Management’ (ICH 2011). Different analytical tools
can be used for the identification and risk assessment of
hazards. The most widely used are failure modes and
effects analysis (FMEA) (Asefzadeh et al. 2013) and haz-
ard analysis and critical control points (HACCP) (Osi-
mani et al. 2013). The use of such risk guidance allows
for the identification of hazards within the cleanroom
(aspects that could potentially cause product harm) and
then to the evaluation of the risks posed by such hazards
through considering the severity of the hazard, should
contamination occur, and the probability of the hazard
occurring.
The primary hazards within cleanrooms are micro-
organisms. Most micro-organisms within cleanrooms are
found in the air and the risk is the extent to which those
present could contaminate a product, either through air
currents or through gravitational effects. Few micro-
organisms are found free-floating; most attached to
‘microbial carrying particles’ such as dust or skin detritus
(Whyte and Hejab 2007); notwithstanding that some
reports suggest that bacteria can occur in the airstream in
combination, forming larger particles (Tham and Zuraimi
2005).
Once a hazard has been identified and where it is con-
sidered to pose an unacceptable level of risk, the objective
should always be to eliminate the hazard. Where a hazard
cannot be completely eliminated, then some form of
detection system (in this case environmental monitoring)
should be in place to assess occurrence. Whilst environ-
mental monitoring can provide important information
relating to changing trends over time, the most widely
used methods are growth based and thus take some time
to produce a result. The incubation times required vary;
however, there is no time shorter than 2 days that will
give a meaningful assessment of the level of bacterial con-
tamination. Moreover, most regimes require far longer
incubation times, especially those which use a general
purpose agar and dual incubation for the assessment of
both bacteria and fungi (Arduino et al. 1991).
2 Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
T. Sandle et al.
Due to the variables of length of incubation and subse-
quent time-to-result, most cleanroom activities are car-
ried out with a degree of uncertainty given that the
environmental conditions are unknown at the time of
processing. Uncertainty increases when unusual or less
well-practiced events happen, including those which
require a high level of personnel input or where air han-
dling is not maintained. An example, and one upon
which this paper is based, is when a cleanroom shuts
down. The closure of a cleanroom can occur due to an
unplanned event, such as power cut; or due to a planned
event, where a cleanroom is shutdown to allow for repair
work. With these events, a cleanroom is restored for nor-
mal processing through re-activation of the air system
and by personnel performing cleaning and disinfection.
At the point when the cleanroom is ready to be handed
back to operations staff, it will typically be subject to
environmental monitoring. At this point, the cleanroom
manager has a choice: whether to wait for the results of
the growth-based environmental monitoring results or to
use the cleanroom at risk.
To complement growth-based microbiological meth-
ods, particle counting can be performed. A concern with
particle counting is that only a small proportion of parti-
cles detected within a cleanroom will be indicative of
micro-organisms and there is no means to differentiate
between ‘viable’ and ‘non-viable’ particulates (some
authors have attempted to show a correlation between
the two (Raval et al. 2012); however, correlative relation-
ships are dependent upon too many variables for any
universal ratio to be provided) (Oxborrow et al. 1975; Li
and Hou 2003). To address this, rapid microbiological
air-monitoring methods have been developed (Miller
2012).
One recent type of technology is optical spectrophoto-
metric technologies, designed to simultaneously detect
the number and size of particles from a volume of air
and to additionally detect micro-organisms within the air
sample (Sandle 2012b). The advantage of these systems is
that by counting the microbial content of an aerosol sam-
ple on a particle-by-particle basis, they provide a continu-
ous, real-time analysis of the microbial content of a given
volume of air.
A second advantage is with improved detection and
accuracy. Because the detection of micro-organisms is
not reliant upon conventional culture-based microbiolog-
ical methods, the data relating to microbiological activity
are more representative than conventional growth-based
methods that are capable of detecting. This relates to the
phenomena of the so-termed ‘viable but nonculturable’
micro-organisms (Weichart 1999), which is a reference to
micro-organisms that will not grow on culture media, yet
retain metabolic activity and viability. Moreover, there
T. Sandle et al.
are organisms that might grow on media under one set
of conditions but will not do so because they have
become stressed or because they are sublethally damaged.
The adoption of rapid microbiological methods within
the pharmaceutical sector has gained regulatory and
industry acceptance (Miller 2008). Examples include the
2013 revision of the Parenteral Drug Association Techni-
cal Report 33, which provides revised guidance on equip-
ment and method validation and regulatory expectations
(PDA 2013). In keeping with regulatory concerns, the
instruments, by collecting data in ‘real time’ are in keep-
ing with the driver initiated by the US Food and Drug
Administration for pharmaceutical manufacturers to uti-
lize the concept of process analytical technology (PAT),
where measurements are made during processing (FDA
2004).
The dilemma discussed in the opening section of this
paper—the unknown risks associated with using clean-
rooms for processing without an assessment of the
microbial levels—can be addressed with rapid microbio-
logical methods. An example of a ‘real-time’ air counter
is the Azbil BioVigilant IMD-Aâ system. The instrument
is detects ‘bio-counts’. These are biological counts, based
on the detection of microbial metabolites and chemical
compounds, enabling the counting of micro-organisms
(Miller et al. 2009a).
This paper describes a study that was undertaken to
assess whether the IMD-Aâ system could be used to com-
pare the microbiological levels in the air of a cleanroom
after it had been reinstated following a period of shutdown
and maintenance with the microbiological levels prior to
the shutdown occurring. The aim was to assess whether
the reconditioned cleanroom was suitable for biopharma-
ceutical processing by drawing this comparison.
Materials and methods
The instrument used for the study was an IMD-Aâ 350
System. The IMD-Aâ system simultaneously detects the
size, number and biologic fluorescence of airborne partic-
ulates within a 0
5–15 lm range (Miller et al. 2013).
With the counting of nonviable particles, the counter
functions in a similar way to conventional particle coun-
ters, utilizing Mie scatter (Bohren and Huffmann 2010).
However, the basis of the technology pertaining to the
detection of micro-organisms is quite different. Microbial
detection is via a laser light source (405-nm) where the
presence or absence of intrinsic fluorescence determines
whether a particle is biologic or inert (Miller et al.
2009b). This is based on micro-organisms containing cer-
tain metabolites and chemical compounds, such as nico-
tinamide adenine dinucleotide (NADH), riboflavin
(vitamin B2), and dipicolinic acid (DPA). When excited
Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
Application of rapid microbiological methods
by a certain wavelength of light, these compounds emit
an intrinsic fluorescence signal that serves as a marker for
biological content of the sample. The assessment of
NADH and riboflavin is necessary for the detection of
vegetative bacteria and fungi, whereas the detection of
DPA is required when assessing the presence of endosp-
ores. Fluorescence detection is suitable for the detection
of microbial contamination as it has a high sensitivity
and requires a relatively short collection time (Bjerner
et al. 2012). With the generated bio-counts, because more
than one micro-organism passing through the instrument
would be counted as ‘one count’, one bio-count does not
necessarily correlate with one micro-organism.
For the study, an EU GMP Grade D area (14644 class 8
in operation) was selected. The cleanroom had a surface
area of approximately 200 square metres and it was used
for the processing and preparation of equipment. The
occupancy level was relatively high (up to ten staff) and
the area included a wash-bay. Samples were taken during
the summer of 2013. The cleanroom was located in a phar-
maceutical facility based in the south-east of England.
With the operation of the IMD-Aâ, the instrument
was located two metres from the area where vessels are
handled and personnel activity was greatest; this was
within the theoretical sampling range of the instrument
and was at the most representative area within the clean-
room. For sampling, a continuous monitoring mode was
used and the sample size was set at one cubic foot of air.
The particle sizes measured were ≥0
5 and ≥5
0 lm,
which met GMP recommendations for particle monitor-
ing in pharmaceutical cleanrooms.
Over the course of the study, the selected cleanroom
was assessed over different periods of time and under dif-
ferent conditions, for particle counts and bio-counts.
These conditions were as follows:
• Cleanroom assessed during normal operations (pre-
shutdown).
• Cleanroom during shutdown and maintenance.
• Cleanroom during cleaning and disinfection.
• Cleanroom returned to normal operations (postshut-
down).
With the conventional particle counting, EU GMP
Grade C/ISO 14644 class 8 (in operation) limits were
applied (although the cleanroom was graded D, there are
no specified ‘in operation’ limits within EU GMP).
The particle levels were (Table 1) as follows:
In terms of setting limits, given that one of the objec-
tives of the exercise was to assess the bio-counts for the
preshutdown period and to compare these to the post-
shutdown period, to assess whether the norm had been
restored, it was decided that the preshutdown period
would provide the limit for acceptability. To determine
3
Application of rapid microbiological methods
Table 1 Table showing particle count limits for the studied clean-
room
Maximum permitted number of
particles per cubic metre
EU GMP Grade 0
5 lm
C (in operation) 3 520 000
whether the limit had been exceeded when the postshut-
down period was examined, provided that the bio-counts
between the two periods were within one logarithm (base
10) of each other, then equivalence would be taken as
having been met.
Over the course of the study, conventional microbiolog-
ical monitoring was undertaken at periodic intervals.
However, due to the limitations with conventional micro-
biological air-sampling methods, a direct comparison with
such methods was not made. Air samplers have design
limitations which affects their ability to capture all of the
micro-organisms within the air, and, even when such
micro-organisms are captured, damage or desiccation can
occur which can prevent them from being recovered via
the agar medium (Jensen et al. 1994). On this basis, there
was little value in seeking a direct comparison between via-
ble air samplers and the IMD-Aâ method.
Results
The data collected from the different monitoring condi-
tions were examined, and the analysis is presented below
in tabular and graphical formats.
Cleanroom during normal operations
To assess ‘normal operations’, the cleanroom was exam-
ined more than a 4-day period to achieve a baseline read-
ing. During this period, the cleanroom was used more
than a 24-h period, with a medium-level occupancy
(averaging at six members of staff). The activities taking
place primarily consisted of the processing of equipment
vessels (cleaning, drying and storage) and initial formula-
tion of drug products. During this time, equipment was
moved in and out of the room and wash bays were used.
The particle counts and bio-counts produced from this
period of the study have been summarized in Table 2.
The table provides the mean counts per minutes, the
standard deviation (as an indicator of variance) and the
highest recorded counts in relation to both particle
counts and bio-counts. To aid the review, the data have
been subdivided into periods of days.
Table 2, which summarizes data collected more than a
96-h period, indicates that the nonviable particle counts
4 Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
T. Sandle et al.
remained considerably below the maximum recom-
mended values as indicated by GMP guidance (in relation
to both mean counts and the highest recorded counts);
moreover, the spread of the particle counts was relatively
small (in relation to the standard deviations).
5
0 lm
With the bio-counts, the typical value was 27 bio-
29 000 counts per minute and the highest recorded bio-count
was 220. These bio-counts had a relatively narrow spread,
as indicated by the standard deviation of 30. Although
the exercise did not seek to measure any direct correla-
tion between bio-counts and the numbers or incidents of
colony-forming units recovered from conventional envi-
ronmental monitoring methods, the data from standard
microbiological monitoring remained in control for the
assessment period, with the active air sampler counts
averaging at 20 CFU m 3.
The assessment of the IMD-Aâ instrument represented
the use of novel technology and the cleanroom had not
been previously examined using any comparable device;
hence, there was no direct reference point for the bio-
counts. Nonetheless, the bio-counts obtained can be
regarded as relatively low. Moreover, the collected data
provided a suitable benchmark for the subsequent com-
parative analysis, as discussed below.
Exploring the basis of this benchmark further, with
both particle counts and bio-counts, there was some vari-
ation according to different activities taking place within
the cleanroom at different time points. For some of the
different activities, the data were trended. Figures 1 to 2
display selected data plots for the preshutdown period of
normal operations.
The above chart displays some data collected on Day 1
for a selected period of time (15:10 h to 18:46 h). The per-
iod chosen is intended to be illustrative of the fluctuation
that can occur with both particle counts and bio-counts.
The chart shows that four peaks in relation to 0
5 lm par-
ticle counts occurred. One of these coincided with a rela-
tively high peak for bio-counts (at around 16:40 h).
These peaks coincided with equipment being moved in
and out of the area; in addition, at around 16:40 h, there
was considerable movement of personnel with six people
being present in the area. From this, the data suggest that
the personnel-related activities have some association
with bio-counts and that with the equipment, there was a
‘nonviable’ particle count association but not a bio-count
association. These events were affirmed through the use
of a video camera.
Figure 2 illustrates a second period of the preshutdown
operations from Day 2 of monitoring (time period B). The
chart has the same scale as per Fig. 1, for comparative
purposes.
Figure 2 indicates that during the period of time
assessed, the particle counts and bio-counts remained
T. Sandle et al. Application of rapid microbiological methods

Table 2 Data summary for the assessment of the cleanroom during normal operations, preshutdown

Mean counts/minute Standard deviation Maximum count/minute

≥0
5 lm Bio- ≥5
0 lm ≥0
5 lm Bio- ≥5
0 lm ≥0
5 lm Bio- ≥5
0 lm

Day Room condition count count count count count count count count count

1 Normal operations, preshutdown 143 38 5 252 46 4 1372 268 22

2 Normal operations, preshutdown 63 22 3 78 22 4 556 157 23
3 Normal operations, preshutdown 119 34 5 222 33 4 2151 280 30
4 Normal operations, preshutdown 37 12 2 75 19 4 604 175 29
Mean counts for days 1–4 91 27 4 157 30 4 1171 220 26

Day 1 preshutdown, selected time period A

1400

1200

1000
Count

800

600

400
Figure 1 Chart displaying a selected time
period (A) from the preshutdown period, Day 200
1. The graph shows counts per minute 0
(equivalent to counts per cubic foot). ( )
15
15 0
15 0
15 0
15 0
16 0
16 0
16 0
16 0
16 0
16 0
17 0
17 0
17 0
17 0
17 0
17 40
18 0
18 0
18 0
18 0
18 0
:1
:2
:3
:4
:5
:0
:1
:2
:3
:4
:5
:0
:1
:2
:3
:
:5
:0
:1
:2
:3
:4
≥0
5 lm; ( ) Bio and ( )≥5
0 lm.

0
Day 2 preshutdown, selected time period B
1400

1200

1000

800
Count

600

400
Figure 2 Chart displaying a selected time
200
period (B) from the preshutdown period, Day
2. The graph shows counts per minute 0
14:11
14:21
14:31
14:41
14:51
15:01
15:11
15:21
15:31
15:41
15:51
16:01
16:11
16:21
16:31
16:41
16:51
17:01
17:11
17:21
17:31
17:41
17:51
18:01
18:11
18:21
18:31
(equivalent to counts per cubic foot). ( )
≥0
5 lm; ( ) Bio and ( )≥5
0 lm.

relatively low. The main bio-count peak, at around
15:55 h, coincided with operators moving a large vessel.
Thus, as with the Fig. 1 data, the primary trigger for the
bio-counts was the degree of operator movement.
The overall assessment from the preshutdown period
was that the impact of the standard activities was mini-
mal in terms of both particle and bio-count generation
and that the cleanroom remained in a state of control.
Cleanroom during shutdown and maintenance
On Day 5 of the study, the cleanroom was formally shut-
down and handed over to the maintenance department
for the testing and replacement of the high-efficiency
particulate air (HEPA) filters which made up the air-han-
dling units (AHU). The maintenance works included the
following:
• The removal of HEPA filter grilles.
• The testing of HEPA filters using an aerosol smoke test.
• The deactivation of the AHU and the replacement of
a HEPA filters.
• Low level of activity following HEPA filter replace-
ment (rooms not clean, although with minimal per-
sonnel activity).
The removal, testing and replacement of the HEPA fil-
ters took 3 days to complete (this was not a continuous
activity). This was followed by a period of low level of

Journal of Applied Microbiology © 2014 The Society for Applied Microbiology 5

Application of rapid microbiological methods T. Sandle et al.

activity, which lasted for fourteen days (this long period
of time was due to other works being undertaken
throughout the facility which resulted in the cleanroom
not being needed and hence cleaning and disinfection
was not performed until the seventeenth day of the shut-
down).
The particle and bio-counts relating to the different
activities across the shutdown have been summarized in
Table 3.
The data displayed in Table 3 indicate that relatively
higher particle and bio-counts were obtained compared
with the preceding period of normal operations. During
the period of normal operations, the level of bio-counts
had averaged at 27 counts per minute. With the first type
of maintenance activity (the removal of the outer cover-
ing grille for the HEPA filters), the mean bio-counts
more than doubled to a value of 58 and peaked at 2555
during the main part of the activity. This high count is
shown in Fig. 3, below.
The main peak on the chart, for the selected time per-
iod (C), relates to the removal of the grille to the HEPA
filter and its subsequent replacement. During this time,
two maintenance staff were present in the area. The activ-
ity was of a relatively short duration (under 6 min); nev-
ertheless, it led to both the highest recorded bio-counts
and the highest recorded 0
5 lm size particle counts.
For the Day 6 activity, the period immediately follow-
ing smoke testing, the bio-counts and particle counts
remained relatively low and, in terms of mean counts,
close to the levels recorded during standard operations.
Here, the air-handling unit would be running at normal
capacity, and any residual smoke used for HEPA filter
leak testing would have been removed. Monitoring was
not undertaken during the use of the smoke because this
could have caused damage to the instrument.
The Day 7 activity, where the AHU was deactivated and
the HEPA filter replaced, led to a large rise in both particle
and bio-counts. This is unsurprising because the clean-
room was no longer supplied with ‘clean’ (low particulate)
air. Furthermore, the air was not circulating turbulently
neither was the air being removed from the cleanroom at
a rapid rate. Under these conditions, the bio-counts aver-
aged at their highest level (at 730) and peaked at 2665.
Figure 4, below, illustrates the time point when the
AHU was deactivated.
With Fig. 4, the AHU was deactivated at 13:31 h. The
chart has the same scale as that of Fig. 3 to illustrate the
increase in bio-counts (and for this reason, the particle
counts rise above the chart scale). The bio-counts showed
an initial increase to the mid-2000s and then reduced
slightly to the 1000-level and then remained around this
level whilst the AHU was offline. The spike with the bio-
counts related to an increase in traffic from process oper-
ators who entered the room at 14:15 h and remained in
the room until 15:07 h. Again, bio-counts tallied with
human activity.
With the bio-counts recorded during this part of the
shutdown period, when maintenance work was occurring,
these counts were considerably higher than those
recorded during the period of normal operations. Con-
sidering the ‘normal’ baseline, the maintenance period
could, therefore, be considered as presenting a greater
risk to the cleanroom and its operations.
Following the re-activation of the AHU, the cleanroom
was closed and no personnel entered to room. This
period produced the lowest level of particle and bio-
counts yet recorded. Therefore, despite the room surfaces
being ‘unclean’, the restoration of HEPA filtered air along
with other cleanroom parameters, such as adequate room
air changes and air flow, the levels of particles and

Table 3 Data summary for the assessment of the cleanroom during the shutdown period

Mean Standard Maximum

counts/minute deviation count/minute

≥0
5 lm Bio- ≥5
0 lm ≥0
5 lm ≥5
0 lm ≥0
5 lm ≥5
0 lm

Day (s) Room condition count count count count Bio-count count count Bio-count count

5 Works included the 164 58 9 702 252 35 7098 2555 350

removal of HEPA filter grilles
6 Works included testing of 51 21 5 96 25 7 695 169 46
HEPA filters using an aerosol
smoke test
7 Operations included the 1720 730 116 2202 647 88 12 682 2665 403
deactivation of the AHU and
the replacement of a HEPA filters
8–17 Shutdown period 9 6 1 16 10 3 96 55 14
continued, minimal room use
Mean for shutdown period 486 204 33 754 234 33 5143 1361 203

6 Journal of Applied Microbiology © 2014 The Society for Applied Microbiology

T. Sandle et al. Application of rapid microbiological methods

Day 5 - shutdown - AHU grille removal (time period C)

7000
6000
5000

Counts
4000
3000
Figure 3 Chart displaying the first work 2000
undertaken during the shutdown: removal of
the HEPA filter grills. The graph shows counts 1000
per minute (equivalent to counts per cubic 0
foot). ( ) ≥0
5 lm; ( ) Bio and

12
12 9
12 5
12 1
12 7
12 3
12 9
13 5
13 1
13 7
13 3
13 9
13 5
13 1
13 7
13 3
13 9
14 5
14 1
14 7
:1
:2
:3
:3
:4
:4
:5
:0
:0
:1
:1
:2
:3
:3
:4
:4
:5
:0
:0
:1
( )≥5
0 lm.

3
Day 7 - shudown AHU deactivated (time period D)
7000
6000
5000
Counts

4000
3000

Figure 4 Chart displaying the time prior to 2000

and at the point of the deactivation of the 1000
cleanroom AHU. The graph shows counts per 0
minute (equivalent to counts per cubic foot).
13

13

13

13

13

14

14

14

14

14

14

14

14

14

14

15
:3

:3

:4

:5

:5

:0

:0

:1

:2

:2

:3

:3

:4

:5

:5

:0
( ) ≥0
5 lm; ( ) Bio and ( )≥5
0 lm.
2

2
micro-organisms remained low. These data arguably sup-
port the general assumption in the literature that people
are the main sources of microbial carrying particles within
cleanrooms; moreover, personnel movement would also
stir up some settled particles (McIntosh et al. 1978).
Cleanroom during cleaning and disinfection
The third part of the exercise was an assessment of the
particle and bio-count levels generated during the clean-
ing (use of detergent) and disinfection of the cleanroom.
This was a standard exercise undertaken following a per-
iod of production downtime to restore a cleanroom for
production use. The cleaning exercise was undertaken
more than a 2-day period. The data collected during this
activity are summarized in Table 4.
The data displayed in Table 4 indicate that the bio-
counts were around twice the value seen when the clean-
room was running for standard operations (as shown in
Table 2). However, the bio-counts were at a much lower
level than those seen for the maintenance works. The
inference from these data is that, whilst the cleaning
activity itself causes a relatively high level of particle gen-
eration, the operation of the AHU and clean air supply
provide a strong degree of protection for the cleanroom
and AHU operation is relatively effective at removing
microbial carrying particles.
Cleanroom returned to normal operations
The final stage of the evaluation was to assess the clean-
room postcleaning, following the handover of the clean-
room back to the process area staff. The period following
handover was monitored for 4 days with the objective of
assessing whether the counts were similar to those of the
preshutdown period. These data from the postshutdown
period are illustrated in Table 5, below.
The data from the postshutdown period compared rela-
tively favourably to the preshutdown period, although the
bio-counts were notably higher at the outset. This suggests
that the area required a ‘rest’ period following the comple-
tion of the cleaning and disinfection prior to the com-
mencement of standard operations. By the second day of
the postshutdown period, the counts had lowered and by
the end of the assessment, the data were within the normal
pattern established prior to the shutdown.
The changing bio-counts for each of the main events
are summarized in Fig. 5:
Figure 5 displays the changing bio-counts for the dif-
ferent periods of activity; however, the data are relatively
distorted by the AHU going offline. When this event is
removed from the data set, the relative values appear as
presented in Fig. 6.
The presentation of the data in Fig. 6 allows a clearer
comparison to be made of the relative bio-counts. Here,

Journal of Applied Microbiology © 2014 The Society for Applied Microbiology 7

Application of rapid microbiological methods T. Sandle et al.

Table 4 Data summary for the assessment of the cleanroom during cleaning and disinfection

Mean counts/minute Standard deviation Maximum count/minute

≥0
5 lm ≥5
0 lm ≥0
5 lm ≥5
0 lm ≥0
5 lm ≥5
0 lm

Day Room condition count Bio-count count count Bio-count count count Bio-count count

18 Detergent cleaning 390 43 6 656 63 8 3011 364 43

19 Disinfection cleaning 315 50 8 630 109 10 3770 1281 59
Mean for cleaning period 352 47 7 643 86 8 3391 823 51

the largest rise in bio-counts is between the end of clean-
ing and the point of room handover. A day later, the val-
ues returned to the pattern seen prior to the shutdown
period. Given that the rudimentary criterion to compare
the period prior to and after the shutdown was a one log
difference, the bio-counts for these two periods easily fall
within this criterion. Thus, as a crude measure, there is
no noticeable difference and the cleanroom—postshut-
down—can be considered to have similar environmental
conditions compared with the preshutdown period.
Discussion
The study showed that the IMD-Aâ system was capable
of detecting, sizing and enumerating particles from con-
tinuous air sampling and that the instrument was addi-
tionally able to determine whether these collected
particles were biological or nonbiological in origin and
thus able to obtain a measure of the free-floating micro-
organisms and micro-organisms attached to microbial
carrying particles in the air.
In terms of the exercise, the instrument was able to
provide an indication of cleanroom conditions during a
period of normal operation and allow a comparison to
be made of conditions following extensive maintenance
works and reinstatement activity. With this exercise, the
bio-counts indicated that there was a similar state of
microbiological control postshutdown as compared with
its condition prior to the shutdown. On this basis, a
cleanroom user could use the IMD-Aâ instrument to
assess the risk of an area following maintenance and have
a degree of confidence that the likelihood of contamina-
tion would be low, provided that the data from the two
periods were comparable.
The exercise also provided interesting data about a
cleanroom as it is taken through different phases of use.
Events such as removing HEPA filter grilles and turning
off an air handing unit are significant activities which
lead to a high generation of biological particles and, in
showing an increase over the baseline, would be regarded
as presenting a microbiological risk to the cleanroom. As
a signal of confidence, the restoration of the correctly
functioning air-handling unit returns a cleanroom to a
level of low biological activity relatively quickly (at least
for the distributed air, surfaces will probably remain con-
taminated until cleaning and disinfection has been under-
taken).
Another interesting observation related to the activity
of cleaning and disinfection of a cleanroom by opera-
tors using the classic mop and bucket methods. Whilst
this activity is more likely to generate nonviable parti-
cles than bio-counts, the time taken for the area to
settle down and return to the norm following cleaning
took longer than the time required to return the area
to a ‘safe’ level following the reactivation of the air-
handling system. This adds further weight to the dis-
cussions that people present the biggest risk factor to
cleanrooms and also that the activity of cleaning
appears to re-deposit biologic and nonbiologic particles
into the airstream.
An important consideration for the study was the zone
within the cleanroom where the IMD-Aâ instrument was

Table 5 Data summary for the assessment of the cleanroom during normal operations, postshutdown

Mean counts/minute Standard deviation Maximum count/minute

≥0
5 lm ≥5
0 lm ≥0
5 lm ≥5
0 lm ≥0
5 lm ≥5
0 lm

Day Room condition count Bio-count count count Bio-count count count Bio-count count

20 Normal operations, postshutdown 619 80 11 1097 162 15 5505 2005 110

21 Normal operations, post 122 28 7 153 29 7 1311 218 70
22 Normal operations, post 187 46 7 324 50 6 1879 277 33
23 Normal operations, post 22 10 3 36 14 5 371 138 49
Mean for days 20–23 238 41 7 402 64 8 2267 660 66

8 Journal of Applied Microbiology © 2014 The Society for Applied Microbiology

T. Sandle et al. Application of rapid microbiological methods

Bio count mean values/day

800
700
600

Bio count
500
400
300
200
100
0

in nt c wn
ng

et t-sh line
1

Po n c ing

Po hut ing

Po hut n 1

td 2
hu n 3

4
d
ok ove
y
ay

do ay

hu n

n
AH etsi

o
Pr utd da

da

w
ow

ow
-
td
f
d

ea

ea
of

do

do
Sm em

u
t
sh wn

sh wn

sh wn

td
l

l
e

U
w

r
o

s
Pr utd

td

io

-s

-s

-s

-s
s
ge
le
ut

po

ct
u

st

st

st

st
ril
Figure 5 Chart showing changes for mean sh

er

fe

Po
rg

ng
e-

e-

e-

e-
bio-counts during the study (selected periods

te

is
ni
Pr

Pr

fil

D
ai
em
of operation). The graph shows counts per

A
EP

R
minute (equivalent to counts per cubic foot).

H
( ) Bio-count. Cleanroom event

Bio-count/mean values per day (excluding AHU)

90
80
70
60
Bio count

50
40
30
20
10
0
n
g

ng
1

1
2

hu n 2

hu n 3

4
ng oke ed

w
in

in
y

n
y

n
ni
o
v

ts

n
da

da

da

da

ow

ow

ow

ow
d
Sm mo

ea

ea
te

ut

td
n

td

td

td
n

cl

Po n cl
h
re
w

hu

hu
-s

nt
do

do

do

do

io
st

-s

-s

-s

-s
e
le
ut

ut

ut

ut

ct
po

Figure 6 Chart showing changes for mean

st

st

st

st
ril

er
sh

sh

sh

sh

fe

Po

Po

Po
rg

et

in
e-

e-

e-

e-

bio-counts during the study (selected period

te

is
ni
Pr

Pr

Pr

Pr

fil

D
ai

of operation). The graph shows counts per

em
A
EP

minute (equivalent to counts per cubic foot).

H

( ) Bio-count. Cleanroom event

placed. In a nonunidirectional flow environments, local
airflow dynamics, particle settling velocity and sampler
flow rate can each affect the size of the ‘zone’ and how
quickly, or indeed whether, a particle originating some
distance away from the sampling instrument may be
detected. Thus, the area selected should be representative
of the cleanroom and ideally has some corresponding
basis to the primary risk to the product or process; this
could be, for example, near open processing or close to
where the highest level of personnel activity is. The risk is
with selecting an area that is unrepresentative of the main
personnel or equipment activities, thus leading to an
underestimation of the particles generated. There is a
degree of uncertainty as to the size of the sampling zone
that an instrument such as the IMD-Aâ can sample (as
would be the case with a conventional particle counter);
on this basis, the counter was sited within a few metres
of the intended target area.
Whilst the data gathered were of great interest and there
were clearly identifiable trends, the study had some limita-
tions. Firstly, although the study ran for several weeks, the
cleanroom was studied on just one occasion. Further stud-
ies would be ideally required to demonstrate the repeatabil-
ity of the results. Nonetheless, the results obtained were
consistent across the time window evaluated.
Secondly, with any study using novel technology, there
will always be a degree of uncertainty with the measure-
ments obtained. With the case of bio-counts, the correla-
tion with ‘colony-forming units’ and microbial carrying
particles could not be shown directly, and microbial levels

Journal of Applied Microbiology © 2014 The Society for Applied Microbiology 9

Application of rapid microbiological methods
could instead only be inferred as ‘low’ or ‘high’ depending
upon the degree of biological activity. Nevertheless, a level
of uncertainty exists with conventional environmental
monitoring methods and given that the IMD-Aâ is capable
of continuous monitoring, unlike an active air sampler, this
level of uncertainty can be assumed to be less with the
IMD-Aâ system (despite the uncertainty not being quanti-
fiable). In time, further studies will build upon the study
outlined in this paper and this will further understanding
of the significance of different levels of bio-counts.
It follows that the most difficult aspect when evaluating a
system such as the IMD-Aâ is in understanding the rela-
tionship between bio-counts and micro-organisms. This is
especially so because it cannot be quantified how many
micro-organisms are present in cleanroom air but not
detected by conventional methods, so direct parallels can-
not be made. Comparative analysis can be undertaken using
a bioaerosol chamber and a mixed population of micro-
organisms; however, this is somewhat artificial, given that
such chambers cannot mimic cleanroom air patterns, and
for most cleanroom users, the interest will be upon using
the instrument in situ (as this study attempted).
A way around this, and an area for future study, would
be by building up voluminous data and showing that very
high or numerous excursions with conventional microbio-
logical methods are reflected in the recording of high bio-
counts. More exact causality is not possible, given that
comparing any two instruments—such as comparing the
IMD-Aâ system with an active air sampler—is difficult.
This is because the two instruments would sample different
volumes of air (within which micro-organisms are not nor-
mally distributed). Furthermore, the instruments would
have different ranges, be calibrated with different d50 (col-
lection efficiency) values; they would ‘register’ microbial
activity through different means (Whyte et al. 2007). This
latter point, about different means of detection, is likely to
be the reason for the largest variation in results should the
IMD-Aâ be compared with an active air sampler.
Another limitation is with not knowing what species of
micro-organisms the bio-counts relate to. For this study,
this was not a direct concern. Nonetheless, in most situa-
tions, the microbiologist needs to know this; therefore, at
present, it is unlikely that real-time technology will wholly
replace growth-based method. The technology remains a
destructive method in that the instruments will indicate
whether micro-organism is present, but they cannot not
determine the species.
A third limitation, remaining with the concept of
uncertainty of measurement, is that some of the bio-
counts may not relate to micro-organisms. Various mate-
rials, such as clothing fibres and potentially some alcohol
sprays, used within the cleanroom can potentially
disperse unknown numbers of nonmicrobial fluorescing
10
T. Sandle et al.
particles. These could falsely increase the number of
assumed viable micro-organisms (Eaton et al. 2012). This
potential concern can be overcome, to an extent, through
the collection of a large number of samples and by set-
ting a baseline. With the study described in this paper,
the materials that might cause this effect will have been
present within the cleanroom before it was shut down for
maintenance and when it was restarted; thus, by under-
standing the baseline, the effect of nonmicrobial fluoresc-
ing particles can be captured within the normal operating
bio-levels within the cleanroom. Thus, the presence of
any nonmicrobial fluorescing particles does not detract
from the finding that the pre- and postshutdown periods
were similar.
These limitations, whilst noteworthy, did not detract
from the study’s findings. Based on the exercise, real-time
monitoring of airborne microbial concentrations within
cleanrooms and controlled environments provides con-
siderable advantages. This is primarily as a means to
assess whether an area is ‘safe’ or ‘clean’, thereby avoiding
the issues arising from the delay in obtaining results
caused by the slow growth of micro-organisms measured
by conventional sampling methods. For assessing such
activities, the device will prove useful for microbiologists.
The instrument can be run continuously and provided
sufficient data can be gathered, so that operational limits
can be set, and the instrument offers advantages over tra-
ditional microbiological methods.
As with describing any novel technology, the results of
this paper add to a developing database of studies. Fur-
ther work will be required to enhance this, including a
replication of the study outlined in this paper as well as
other studies, such as an analysis of different types of
cleanrooms, including unidirectional air flow worksta-
tions used for the aseptic filling of products.
Acknowledgements
The authors would like to thank Biovigilant and Techno-
path for the use of the IMD-Aâ instrument and for
advice during the study. This paper has been written
independently of these companies.
Conflict of Interest
No conflict of interest declared.
References
Arduino, M.J., Bland, L.A., Aguero, S.M. and Favero, M.S.
(1991) Effects of incubation time and temperature on
microbiologic sampling procedures for hemodialysis fluids.
J Clin Microbiol 29, 1462–1465.
Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
T. Sandle et al.
Asefzadeh, S., Yarmohammadian, M.H., Nikpey, A. and
Atighechian, G. (2013) Clinical risk assessment in
intensive care unit. Int J Prev Med 4, 592–598.
Bjerner, G., Kene, V., Akerlund, E. and Andersson, K. (2012)
Real-time microbiological air monitoring. Eur J Parenter
Pharm Sci 17, 52–55.
Bohren, C.F. and Huffmann, D.R. (2010) Absorption and
Scattering of Light by Small Particles. New York: Wiley-
Interscience.
Eaton, T., Davenport, C. and Whyte, W. (2012) Airborne
microbial monitoring in an operational cleanroom using
an instantaneous detection system and high efficiency
microbial samplers. Eur J Parenter Pharm Sci 10, 339–351.
FDA (2004) Guidance for Industry: PAT - A Framework for
Innovative Pharmaceutical Development, Manufacturing and
Quality Assurance. Food and Drug Administration,
Rockville: MD, USA.
International Conference on Harmonisation (ICH). (2011)
Quality Risk Management (ICH Q9), EMA/INS/GMP/
79766/2011.
Jensen, P.A., Lighthart, B., Mohr, A.J. and Shaffer, B.T. (1994).
Instrumentation used with microbial bioaerosol. In
Atmospheric Microbial Aerosols: Theory and Applications
eds Lighthart, B. and Mohr, A.J. pp. 226–284. New York,
NY: Chapman & Hall.
Li, C.-S. and Hou, P.-A. (2003) Bioaerosol characteristics in
hospital clean rooms. Sci Total Environ 305, 169–176.
McIntosh, C., Lidwell, O.M., Towers, A.G. and Marples, R.
(1978) The dimensions of skin fragments dispersed into
the air during activity. J Hyg 81, 471–479.
Miller, M.J. (2008). Rapid Microbiological Methods. In
Microbiology Pharmaceutical Manufacturing ed. Prince, R.
pp. 171–221. Vol. 2, Bethesda, MD: DHI/PDA.
Miller, M.J. (2012) New year, old challenges. Eur Pharm Rev
17, 2–5.
Miller, M.J., Walsh, M.R., Shrake, J.L., Dukes, R.E. and Hill,
D.B. (2009a) Evaluation of the BioVigilant IMD-Aâ, a
novel optical spectroscopy technology for the continuous
and real-time environmental monitoring of viable and
nonviable particles. Part II: case studies in environmental
monitoring during aseptic filling, intervention assessments
and glove integrity testing in manufacturing isolators. PDA
J Pharm Sci Technol 63, 258–282.
Miller, M.J., Lindsay, H., Valverde-Ventura, R. and O’Conner,
M.J. (2009b) Evaluation of the BioVigilant IMD-Aâ, a
novel optical spectroscopy technology for the continuous
and real-time environmental monitoring of viable and
nonviable particles. Part I. Review of the technology and
comparative studies with conventional methods. PDA J
Pharm Sci Technol 63, 244–257.
Miller, M.J., Lindsay, H., Valverde-Ventura, R. and O’Connor,
M.J. (2013) Evaluation of the BioVigilant IMD-A, a novel
Journal of Applied Microbiology © 2014 The Society for Applied Microbiology
Application of rapid microbiological methods
optical spectroscopy technology for the continuous and
real-time environmental monitoring of viable and
nonviable particles. Part I: review of the technology and
comparative studies with conventional methods. In
Encyclopedia of Rapid Microbiological Methods ed. Miller,
M.J. pp. 213–234. Volume 4. USA: PDA and Davis
Healthcare International Publishing.
Osimani, A., Aquilanti, L., Tavoletti, S. and Clementi, F.
(2013) Microbiological monitoring of air quality in a
university canteen: an 11-year report. Environ Monit Assess
185, 4765–4774.
Oxborrow, G.S., Fields, N.D., Puleo, J.R. and Herring,
C.M. (1975) Quantitative relationship between
airborne viable and total particles. Health Lab Sci 12,
47–51.
PDATechnical Report No. 33, Revised (2013) Revised 2013
(TR 33) Evaluation, Validation and Implementation of
Alternative and Rapid Microbiological Methods. Bethesda,
MD, USA: Parenteral Drug Association.
Raval, J.S., Koch, E. and Donnenberg, A.D. (2012)
Real-time monitoring of non-viable airborne particles
correlates with airborne colonies and represents an
acceptable surrogate for daily assessment of cell-
processing cleanroom performance. Cytotherapy 14,
1144–1150.
Sandle, T. (2012a). Environmental Monitoring: a practical
approach. In Environmental Monitoring: A Comprehensive
Handbook ed. Moldenhauer, J. Volume 6, pp. 29–54. River
Grove, USA: PDA/DHI.
Sandle, T. (2012b) Real-time counting of airborne particles
and microorganisms: a new technological wave? Clean Air
Containment Rev 9, 4–6.
Sandle, T. and Saghee, M.R. (2013). Cleanroom certification
and ongoing compliance. In Cleanroom Management in
Pharmaceuticals and Healthcare eds Sandle, T. and Saghee,
M.R. pp. 169–184. Passfield, UK: Euromed
Communications.
Tham, K.W. and Zuraimi, M.S. (2005) Size relationship
between airborne viable bacteria and particles in a
controlled indoor environment study. Indoor Air 15(Suppl
9), 48–57.
Weichart, D. (1999). Stability and survival of VBNC cells -
conceptual and practical implications. In Proceedings of the
8th Symposium on Microbial Ecology ed. Bell, C.B.-G.
Halifax, Canada: Atlantic Canada Society for Microbial
Ecology.
Whyte, W. and Hejab, M. (2007) Dispersion of particles and
microorganisms from people. Eur J Parenter Pharm Sci 12,
39–46.
Whyte, W., Green, G. and Albisu, A. (2007) Collection
efficiency and design of microbial air samplers. J Aerosol
Sci 38, 101–114.
11