Content uploaded by Bernard Ho
Author content
All content in this area was uploaded by Bernard Ho on Dec 07, 2020
Content may be subject to copyright.
Pediatric Pulmonology 34:336–341 (2002)
Do In-Line Respiratory Filters Protect Patients?
Comparing Bacterial Removal Efficiency of Six Filters
Anne-Marie Canakis, BSc,MDCM,FRCPC,FAAP,
1
Bernard Ho, BSc,RCPT (P),
1
Sharon Ho, BSc,RT,
1
Danuta Kovach, BA,
2
Anne Matlow, MD,FRCPC,
2
and Allan L. Coates, BEng,MDCM
1
*
Summary. With all pulmonary function diagnostic and respiratory therapy equipment, cross-
infection has always been a concern, especially in the cystic fibrosis population, in whom
pulmonary function tests are done routinely. The aim of this study was to identify and compare the
bacterial removal efficiency (BRE, ability of a filter to remove microorganisms) of six different filters
used in hospital settings: Microgard (MG), Spirobac (SB), PALL (PL), and KOKO (KK), used in the
pulmonary function laboratory; and Clear-Guard (CG) and Respigard (RG), used in ventilator
circuits. Filters were tested in both saturated and nonsaturated conditions. A Pseudomonas
aeruginosa suspension of 1 10
4
to 1 10
8
CFU/mL was nebulized onto each filter. A blood agar
plate was held immediately downstream from the filter. Colony-forming units (CFU) were then
counted after 24 hr of incubation. A peak flow was applied across the spirometry filters. Bacterial
thresholds of the filters were also identified (concentration of bacteria at which a filter no longer has
100% BRE).
There was a significant difference in BRE among the six filters in saturated states when
challenged with 1 10
4
CFU/mL (MG, KK, CG, and RG, 100%; SB, 98.8%; PL, 42.7%; P¼0.003).
There was no significant difference between saturated and nonsaturated states, or after application
of a peak flow. Filter thresholds were significantly different (KK 1 10
8
,MG110
7
,CG110
6
,
RG 1 10
5
, and SB and PL <110
4
CFU/mL).
In conclusion, when all filters are exposed to the same extreme challenges, significant dif-
ferences exist in their ability to remove bacteria. Pediatr Pulmonol. 2002; 34:336– 341.
ß2002 Wiley-Liss, Inc.
Key words: spirometry filters; cross-infection; hygiene; disease transmission; Pseudo-
monas aeruginosa; cystic fibrosis; pulmonary function testing; mechanical
ventilation.
INTRODUCTION
Cross-infection of respiratory therapy and pulmonary
function (PF) laboratory equipment has long been of
concern. When a subject performs a forced expiratory
maneuver, droplets of oropharyngeal secretions which
may contain bacteria and viruses are aerosolized. A case of
tuberculin skin test conversion after use of contaminated
PF equipment was reported.
1
In 1991, Rutala et al.
2
demonstrated that with spirometry in the PF laboratory,
14% of the associated tubing contains the respiratory
pathogen after patient use. This is of particular concern for
cystic fibrosis (CF) patients who are colonized with
respiratory pathogens and in whom pulmonary function
tests (PFTs) are done routinely at every clinic visit. Cross-
infection was previously documented in winter camps.
3
Questions have been raised as to whether PFTs contribute
to the spread of Burkholderia cepacia among the CF
population.
4
In PF equipment, the highest risk of cross-
infection occurs during maneuvers that require repetitive
forceful inhalation and exhalation, as in forced vital
capacity and maximum voluntary ventilation. High-risk
maneuvers are also used during measurements of static
lung volumes, such as gas dilution techniques and body
plethysmography for functional residual capacity and dif-
fusing capacity.
5
When these tests are repeated several
times as per American Thoracic Society criteria,
6
the
potential for cross-infection increases, especially if the
maneuvers are accompanied by coughing. Outbreaks of
1
Division of Respiratory Medicine, Department of Pediatrics, Hospital for
Sick Children, University of Toronto, Toronto, Ontario, Canada.
2
Division of Infectious Diseases, Department of Pediatrics, Hospital for
Sick Children, University of Toronto, Toronto, Ontario, Canada.
Grant sponsor: Divisions of Respiratory Medicine and Infectious Diseases,
Hospital for Sick Children-University of Toronto; Grant sponsor: Canadian
Cystic Fibrosis Foundation.
*Correspondence to: Dr. Allan L. Coates, Division of Respiratory
Medicine, Hospital for Sick Children, 555 University Ave., Toronto,
Ontario M5G 1X8, Canada. E-mail: allan.coates@sickkids.ca
Received 30 October 2001; Accepted 24 January 2002.
DOI 10.1002/ppul.10171
Published online in Wiley InterScience (www.interscience.wiley.com).
ß2002 Wiley-Liss, Inc.
nosocomial respiratory infections due to contaminated
respiratory equipment in intensive care units (ICUs) in
ventilated patients have been reported.
7,8
Specific to this
setting is the fact that many of the filters used in ventilator
equipment become saturated with the high humidity used
in these circuits, providing a good environment for
microbial growth.
In order to avoid the time-consuming decontamination
procedures and complex aseptic techniques previously
described,
9
in-line bacterial filters were introduced in the
1980s to act as barriers, with the goal of preventing cross-
contamination. Very few studies in the literature (exclud-
ing technical reports from manufacturing companies)
have looked at filter efficiency, and usually these have been
in healthy subjects.
10
Few studies have compared the
bacterial removal efficiency (BRE) of different filters
under extreme conditions,
11– 13
and none has looked at the
filtering of Pseudomonas species.
The primary aim of this study was to identify and
compare the BRE of two different groups of filters, using
aerosolized Pseudomonas aeruginosa (PA) to challenge
these filters. The secondary aim was to determine the
bacterial threshold of each filter (i.e., the concentration of
aerosolized bacteria at which a filter will no longer have
100% BRE). Four of the filters tested are used in PF
laboratory equipment, and the other two are used in
mechanically ventilated patients in the ICU. The filters
were studied in both the nonsaturated and saturated states
(i.e., saturated with moisture).
MATERIALS AND METHODS
Two separate groups of filters were tested. One group
compared four spirometry filters, and the second group
compared two ventilator circuit filters to each other. The
four spirometry filters included the Pall PRO-TEC Barrier
Filter
1
(PL, PF 30S, PALL Biomedical Products Co.,
Glen Cove, NY), the SensorMedics MicroGard Filter
1
(MG, SensorMedics Corp., Yorba Linda, CA), the Spiro-
bac filter
1
(SB, 500P30001, Dar, Italy), and the KOKO
1
filter (KK, Pulmonary Data Service Instrumentation,
Inc., Louisville, CO). The two ventilator filters includ-
ed an Intersurgical Clear-Guard II Breathing Filter
1
(Intersurgical Inc., Liverpool, NY), and a Respigard
Filter
1
(Vital Signs, Inc., Totowa, NJ). Each filter was
set up in the system outlined in Figure 1. The first step
was to determine the saturation time of each test filter. This
was done by nebulizing 5 cc of methylene blue through the
studied test filter, and onto a second filter paper held down-
stream. The nebulizer (Hudson 1730, Hudson Respiratory
Care, Inc., Temecula, CA) was powered by a compressor
(Pari ProNeb, Pari Respiratory Equipment, Mississauga,
Ontario, Canada) at flow rates of 4.5 liters per minute
(LPM). The time at which methylene blue was detected
downstream of the system (i.e., by visualization of a
stream of mist downstream and blue discoloration of the
second filter paper) was defined as the saturation time for
that filter. This technique was found to be reproducible.
The study was then done in two parts.
Part 1: BRE When Filters Challenged With 1 10
4
Colony-Forming Units (CFU)/mL
Each filter was challenged with a Pseudomonas aeru-
ginosa (PA) bacterial concentration of 1 10
4
CFU/mL,
prepared by our microbiology department. PAwas a labo-
ratory isolate of a nonmucoid strain. The PA was grown
and all dilutions were made with a brain-heart infusion
(BHI) broth (BBL/Becton-Dickinson), reconstituted from
powder form by adding distilled water. A 5-mL aliquot
was nebulized onto each filter, passing first from the
nebulizer, through a plastic T-piece closed off on one end,
and into the mouthpiece containing the filter. A blood agar
plate (BAP) (5% sheep blood in Columbia agar base, PML
ABBREVIATIONS
BAP Blood agar plate
BHI Brain-heart infusion
BLT Number of bacteria let through
BRE Bacterial removal Efficiency
CF Cystic fibrosis
CFU Colony-forming unit
CG Clear-Guard filter
KK KOKO filter
MG Microgard filter
PA Pseudomonas aeruginosa
PF Pulmonary function
PFT Pulmonary function tests
PL PALL filter
RG Respigard filter
RT Respiratory therapy
SB Spirobac filter
SEM Standard error of means
TNTC Too numerous to count Fig. 1. Test filter setup, simulating continuous tidal breathing
through test filter.
Bacterial Removal by Respiratory Filters 337
Microbiologicals, Mississauga, Ontario, Canada) was held
immediately 5 mm downstream from the filter (Fig. 1).
The BAP was then incubated for 24 hr, after which time
colonies were counted.
To assess results with a nonsaturated filter, the PA
suspension was nebulized onto the filter for a period of
2 min less than the filter saturation time. To assess results
with a saturated filter, PA was nebulized onto the filter for a
period of 2 min longer than the filter saturation time. Each
procedure was repeated in triplicate.
Part 2: Identification of Filter Thresholds
Each filter was challenged with five different bacterial
concentrations (10
4
,10
5
,10
6
,10
7
, and 10
8
CFU/mL).
Bacterial concentrations were confirmed by viability
counts. Two trials were done at each concentration. Filters
were studied only in saturated states.
For the four filters used primarily for spirometry in the
PF laboratory (Pall PRO-TEC, SensorMedics Microgard,
Spirobac, and KOKO spirometry filters), an additional
step was done to simulate a forced expiratory maneuver
in both parts 1 and 2 of the experiment. After the Pseu-
domonas suspension was nebulized onto the filter, the
BAP was removed and a clean new one was placed down-
stream. The contaminated filter remained in place. An
Ambu-Bag was hooked up to the system via a Peak Flow
Meter (Asses Peak Flow Meter, Health Scan Products,
Inc., Cedar Grove, NJ) (Fig. 2). For each trial, a peak flow
above 500 LPM was manually applied three consecutive
times to the contaminated filter with an existing bacterial
load. BAPs were then incubated for 24 hr, and colonies
were counted. Each trial therefore had two BAPs, one for
the collection of bacteria during the nebulization period,
and one for the collection of extra bacteria that crossed
through the contaminated filter after three peak flows were
applied across it.
Previous studies looked at sputum bacterial density
in CF patients.Smith et al.
14
showed that during an acute
CF exacerbation, the density of PA is in the range of
7.8 1.1 log
10
colony forming units (CFU)/g of sputum.
One of the definitions of ‘‘infection’’ used by the authors
included the presence of organisms in densities greater
than 10
5
CFU/mL in fluids or tissues. Based on these data,
to simulate in vivo conditions, we chose to challenge the
filters with concentrations of PA ranging from 10
4
–
10
8
CFU/mL.
All nebulization of bacteria was done in a laminar flow
biological safety cabinet, which was disinfected and
sterilized between experiments. All reusable equipment
was sterilized between experiments.
Calculation of Bacterial Removal Efficiency
% bacterial removal efficiency was calculated by the
equation:
%BRE ¼mean#CFU without filter mean#CFU with filter
mean#CFU without filter
100;
where the mean represents the mean of three trials.
The mean number of CFU grown on the BAP in the
absence of a filter is representative of the bacterial
challenge to the filter. This was calculated by nebulizing
a known concentration of PA suspension ata constant flow
through the system, onto a BAP in the absence of a filter.
This was done for 1 min. Mean #CFU without filter was
adjusted for the actual duration of nebulization, using the
equation:
Mean CFU without filtern minutes ¼
ð#CFU without filter1 minuteÞðn minutes nebulizationÞ
where n ¼number of minutes PA organisms were
nebulized.
When CFUs were too numerous to count (TNTC) or
confluent in growth, an arbitrary value of 500 CFUs was
assigned in order to be able to perform statistical analyses.
In cases where no colonies were present on the BAP, BRE
was reported as >99.99%. The number of bacteria let
through the filter (BLT) (represented by the number of
CFU grown on the BAP) was reported for experiments
measuring filter thresholds, because when suspensions
with high bacterial concentrations were nebulized directly
onto the BAP without a filter, colonies were either TNTC
or confluent.
The data obtained were analyzed using ANOVA and the
Tukey-Kramer multiple comparisons test. For each filter,
to determine whether a significant difference in efficiency
existed between pre- and postsaturation, a paired t-test was
used. A P-value of 0.05 was considered significant. There
was no adjustment for multiple comparisons.
Fig. 2. Test filter setup, simulating forced expiratory peak flow
across test filter.
338 Canakis et al.
RESULTS
The filter saturation times for CG, RG, and MG filters
were 12 min, 6 min 40 sec, and 7 min 30 sec, respectively.
For the SB, PALL, and KK filters, the saturation time was
0 min, as methylene blue was detected downstream of the
system immediately upon starting nebulization. There-
fore, BRE could only measured in the saturated state for
these filters.
There was a significant difference in BRE among
the four spirometry filters in their post-saturated states,
when challenged with a bacterial concentration of
110
4
CFU/mL (P¼0.0127). The MG and KK filters
all had >99.99% BRE, the SB 98.8%, and the PL filter
42.7% (Table 1). The PL filter was significantly less
efficient than all other spirometry filters (P<0.05 for all
comparisons). There was no difference in BRE between
saturated and nonsaturated states for the MG filter. There
was no difference in BRE after a peak flow was applied
across the contaminated spirometry filters. Mean peak
flow applied ranged from 513–580 LPM. The two
ventilator filters (RG, CG) both had >99.99% BRE in
the saturated and nonsaturated states.
The filter threshold of the SB and PL filters was
significantly lower than that of the other filters (Table 2).
The BLT of the SB filter was significantly different from
each of the other spirometry filters at all concentrations.
The BLT of the PL filter was significantly higher than that
of the other filters at concentrations of 10
6
–10
8
CFU/mL.
The four spirometry filters underwent the additional
step of being subjected to a peak expiratory flow across the
contaminated filters during the filter threshold experi-
ments. Mean peak flows ranged from 500– 650 LPM.
There were no further colonies grown on the BAP
postpeak flow for the MG and KK filters. For the SB filter
at 10
8
CFU/mL, two extra colonies grew postpeak flow.
For the PL filter at 10
5
and 10
6
CFU/mL, 6 and 10 extra
colonies grew, respectively.
DISCUSSION
Cross-transmission of infectious agents has always been
of utmost importance in PF equipment and ventilators,
particularly if these filters become saturated. Our study
helped identify whether filters currently on the market
have a risk of cross-contamination in both the saturated
and nonsaturated state. This study also helped identify
whether application of a peak flow of 500– 650 LPM to a
spirometry filter, as is done in a PF laboratory, affects BRE
of a filter.
Respiratory filters must prevent bacterial cross-
contamination by removing organisms (bacterial or viral)
and keeping them trapped within the filter. They must
ensure that if a flow is applied across it (either as a large
peak flow in the PF laboratory, or as normal tidal volume
flows in a ventilated patient), the organisms will not be
forced through the filter and deposited downstream.
Several mechanisms have been described for this trapping
system. Direct interception or filtration physically removes
large particles (>1mm) whose diameter is larger than the
pores of the filter membrane. Inertial impaction results in
the removal of smaller particles (0.5– 1.0 mm in diameter)
by collision within the filter material. Diffusional inter-
ception removes very small particles (<0.5 mm) due to
their Brownian motion, which increases the like-
lihood that they will collide with the filter material.
15
In
addition to the three mechanisms discussed, other mecha-
nisms include hydrophobic properties,
15,16
electrostatic
properties,
16,17
and coalescence of small drops into large
drops.
18
P. aeruginosa is approximately 0.5 mm in diam-
eter. In this study, bacteria were suspended in droplets
created by nebulization, using the Hudson 1730 nebulizer.
Previous studies described the particle size distribution
seen with the Hudson 1730 nebulizer.
19,20
With this
nebulizer in our laboratory, the majority of the aerosoli-
zed droplets were large (3.5–5 mm), and should therefore
be stopped by a filter via direct interception. However,
there was a minor proportion of droplets that were smaller
(0.5–1 mm), in which case the other mechanisms of
droplet removal described above must come into play.
In 1946, Duguid
21
looked at the size of respiratory
droplets with sneezes, coughs, and loud speech. They rang-
ed from 1–2,000 mm, with 95% being between 2 –100 mm,
and the most common between 4–8 mm. In the case of
activities with higher flows (e.g., sneezing), the overall
number of droplets were more numerous (1,000,000 vs.
250 droplets for sneezing and counting loudly, respec-
tively). In addition, the proportion of smaller droplets was
more numerous for the more violent activities. Our study
simulated activities of low flow such as talking, and
activities of high flow such as coughing and sneezing.
With the majority of its particles in the 3– 5-mm range, the
Hudson nebulizer is representative of what is seen during
exhalation in vivo.
TABLE 1— Comparison of Bacterial Removal Efficiency
(BRE) of Filters in Pre- and Postsaturation States, When
Challenged With 1 10
4
CFU/mL of Pseudomonas
aeruginosa
Filter
% BRE (SD)
Nonsaturated Saturated
Spirobac (SB) 98.8 1.1
Microgard (MG) >99.99 0>99.99 0
Pall (PL) 42.7 37.6
Koko (KK) >99.99 0
Clear-Guard (CG) >99.99 0>99.99 0
Respigard (RG) >99.99 0>99.99 0
Bacterial Removal by Respiratory Filters 339
A filter must have maximal efficiency in trapping and
removing bacteria, but also a low resistance to airflow.
22
The spirometry filters used in this study are known to have
resistances that meet the American Thoracic Society
criteria for pulmonary function testing (<1.5 cmH
2
O/L/
sec for flows of up to 14 L/sec).
6
The ventilator circuit
filters, which are not intended for use in the PF laboratory,
had much higher resistances. It is interesting to note that
the spirometry filters with the lowest resistances to flow
were also those with poorer BRE, suggesting that perhaps
BRE is sacrificed for low resistance.
One may argue that when comparing filters, all should
be exposed to the same bacterial challenge. In order to
mimic a patient with CF coughing or bringing up sputum
during the performance of PF tests, we felt that filters
should be tested under fully saturated conditions. For this
reason, each filter had to be subjected to a different length
of nebulization time in order to achieve saturated condi-
tions. The filters with the poorest performance were those
with the shortest time of nebulization (i.e., shortest satu-
ration time). In other words, if these filters were subjected
to longer nebulization times and therefore larger bacterial
challenges similar to the other filters tested, the differ-
ences seen would have been of even greater significance.
In retrospect, given the different times required for satura-
tion, a preferable method would have been to presaturate
all filters with sterile saline, followed by exposing all
filters to the same bacterial challenge and nebulization
time. This would have ensured that all filters truly had
equal challenges.
The concentration of the PA suspension changed during
nebulization. Nebulizer output became more concentrated
with time (i.e., more PA per mL challenging the filter).
A system with a constant output of bacteria would have
been ideal; however, once again, the filters with longer
nebulization times and therefore greater bacterial chal-
lenge were those which performed the best. Therefore, this
weakness did not bias us towards significant results.
One of the limitations of this study was seen in
the threshold experiments. When higher concentrations
were used, for some filters the number of CFUs grown
were TNTC or confluent in growth (SB and PL filters). The
value of 500 CFUs was assigned in order to be able to
make calculations. This value is a gross underestimation
of the actual number of CFUs, and underestimates the
significance of the results.
In conclusion, when in-line filters were subjected to
identical extreme conditions, our study showed that there
are significant differences in bacterial removal efficiency
among these six filters currently on the market. These
findings may assist physicians in choosing a filter for use in
their PF laboratory. Our study was designed as an in vitro
study, and further studies will need to compare the BRE
of these filters when used in patients with CF who are
colonized with Pseudomonas aeruginosa. In vivo studies
will help determine whether the spread of bacteria across
filters does occur in these settings, and more impor-
tantly whether sufficient quantities of pathogens will be let
through the filters to cause clinical infection or neocolo-
nization of the respiratory tract with P. aeruginosa in
CF patients.
ACKNOWLEDGMENTS
Special thanks go to Dr. Ian MacLusky, Ms. Jennifer
Chay, and Ms. Victoria Snell for all their technical
assistance and help with diagrams, tables, and schematic
representations.
REFERENCES
1. Hazaleus RE, Cole J, Berdischewsky M. Tuberculin skin test
conversion from exposure to contaminated pulmonary function
testing apparatus. Respir Care 1981;26:53–55.
2. Rutala DR, Rutala WA, Weber DJ, Thomann CA. Infection risks
associated with spirometry. Infect Control Hosp Epidemiol 1991;
12:89– 92.
3. Ojeniyi B, Frederiksen B, Hoiby N. Pseudomaonas aeruginosa
cross-infection among patients with cystic fibrosis during winter
camp. Pediatr Pulmonol 2000;29:177–181.
4. Isles A, MacLusky IB, Corey M, Gold R, Prober C, Fleming P,
Levison H. Pseudomonas cepacia infection in cystic fibrosis:
an emerging problem. J Pediatr 1984;206:206–210.
5. Clausen JL. Lung volume equipment and infection control. ERS/
ATS Workshop report series. European Respiratory Society/
American Thoracic Society. Eur Respir J 1997;10:1928– 1932.
6. American Thoracic Society. Standardization of spirometry 1994
update. Am J Respir Crit Care Med 1995;152:1107–1136.
TABLE 2— Filter Thresholds
1
Concentration
(CFU/mL)
Clear-Guard
CFU (SD)
Respigard CFU
(SD)
Microgard CFU
(SD)
Spirobac CFU
(SD)
Pall CFU
(SD)
Koko CFU
(SD)
10
4
00000032.1 185 38.8* 0 0
10
5
0010.7 0 05710.6 540 42.4** 0 0
10
6
64.2 0 000 284 53.7* TNTC** 0 0
10
7
18 7.1 5 3.5 2 0 TNTC** TNTC** 0 0
10
8
68 43.8 25 14.8 94 60.8 TNTC** TNTC** 1 0
1
Total number of bacteria let through the filters (CFU) over two trials, when challenged with increasing bacterial concentrations of Pseudomonas
aeruginosa. TNTC, too numerous to count; CFU, colony-forming units.
*P<0.05.
**P<0.001.
340 Canakis et al.
7. Craven DE, Steger KA. Nosocomial pneumonia in the intubated
patient. Infect Dis Clin North Am 1989;3:843 –866.
8. Cross AS, Roup B. Role of respiratory assistance devices in
endemic nosocomial pneumonia. Am J Med 1981;70:681 –685.
9. Depledge MH, Barrett A. Aseptic techniques for lung function
testing. J Hosp Infect 1981;2:369–372.
10. Kirk YL, Kendall K, Ashworth HA, Hunter PR. Laboratory
evaluation of a filter for the control of cross-infection during
pulmonary function testing. J Hosp Infect 1992;20:193–198.
11. Castel O, Planchon C, Denjean A, Soyer S, Barriere M, Merle C,
Fauchere JL. Assessment of three filters for respiratory function
tests [in French]. Rev Mal Respir 1998;15:759– 764.
12. Perissin R. The Microgard
1
filter, an advance in patient and
spirometric protection. Cardiolpulmonary review: current appli-
cations and economics. Part no. 770404. Yorba Linda, CA;
Sensormedics Continuing Medical Education Group; 1994
(available on request).
13. Becquemin MH, Bouchikhi A, Croix N, Malarbet JL, Bertholon
JF, Roy M. Experimental measurements of particle retention
efficiency of filters used to prevent contamination in respiratory
devices. Intensive Care Med 1998;24:81–85.
14. Smith AL, Redding G, Doershuk C, Goldmann D, Gore E, Hilman
B, Marks M, Moss R, Ramsey B, Rubio T. Sputum changes
associated with therapy for endobronchial exacerbation in cystic
fibrosis. J Pediatr 1988;112:547–554.
15. Ball PR, Saunders D. Viral removal efficiency of the Pall
1
Ultipor
1
Breathing System filter. PALL Company communica-
tion. Bm 2115a. Glen Cove, NY; Scientific and Laboratory
Services, Pall Europe Ltd.; 1987 (available on request).
16. Miorini T. Assessment of the hygienic/microbiologic properties
of the Spirobac spirometry filter. Test report GU 16/93. Graz;
Institute of Applied Hygiene. 1994 (available on request).
17. Benbough JE, Bennet A. A test protocol and evaluation of
Intersurgical’s Filta-Therm and Filta-Guard as a viral filter.
Biosafety Testing Section, Biologics Division, Center for Applied
Microbiology and Research. Intersurgical company communica-
tion. Proton Down, Salisbury, Wiltshire: Biosafety Testing
Section, Biologics Division, Center for Applied Microbiology
and Research; 1990 (available on request).
18. Spielberg R. Cross contamination reduction efficiency of the Pall
PRO-TEC Barrier Filter (PF30) for pulmonary function testing.
PALL technical report. BPF-30. Glen Cove, NY; Pall Biomedical
Products Corporation. 1988 (available on request).
19. Coates AL, MacNeish CF, Lands LC, Meisner D, Kelemen S,
Vadas EB. A comparison of the availability of tobramycin for
inhalation from vented versus unvented nebulizers. Chest 1998;
113:951– 956.
20. Ho SL, Coates AL. The effect of dead volume on the efficiency
and the cost to deliver medications in cystic fibrosis with four
disposable nebulizers. Can Respir J 1999;6:253–260.
21. Duguid JP. The size and duration of air carriage of respiratory
droplet nuclei. J Hygiene 1946;44:471–479.
22. Guimond VJ, Gibson NN. Effect of in-line filters on spirometry.
Can J Respir Ther 1990;26:9–12.
Bacterial Removal by Respiratory Filters 341