National Personal Protective
Technology Laboratory
Science Webinar
Part 2: Variability of Respirator Fit Test
Panels NIOSH Study Results
Christopher Coffey, PhD
James T. Wassell, PhD
Ziqing Zhuang, PhD
Colleen Miller, AAS, BS
August 20, 2013
NIOSH PPT / NPPTL
Vision & Mission
The VISION is to be the leading provider of quality, relevant, and timely
PPT research, training, and evaluation.
The MISSION of the PPT program is to prevent work-related injury,
illness and death by advancing the state of knowledge and application
of personal protective technologies (PPT).
PPT in this context is defined as the technical methods, processes, techniques, tools, and materials
that support the development and use of personal protective equipment worn by individuals to
reduce the effects of their exposure to a hazard.
Purpose
 Scientific meeting
 Discuss additional analyses and new preliminary results
 Discuss possible panel implementation strategies
 Continue and encourage dialog
Webinar Summary
 Overview of Last Webinar – Chris Coffey
 Pass Rate Data Analyses – Terry Wassell
 Alternate Analyses – Ziqing Zhuang
 Panel Implementation – Colleen Miller
July Webinar Summary
 Purpose - Discuss objectives and preliminary results
 Methodology
 Ten N95 respirators (five elastomeric/five FFR)
 Measured ambient inward leakage
 Panels

Evenly distributed

Population based
 CV analysis
 Results
 Smaller the panel size, the larger CV
 For n=30, population-based smaller CV than evenly dist.
July Webinar Summary (Cont’d)
 Stakeholder input solicited
 Is CV an appropriate approach for this analysis? If
not, what do you suggest?
 What other ways can the data be used?
 What further research and data analysis is needed
to answer the question of inter-panel variability?
 No comments received
QUESTIONS?
Pass Rate Data Analyses
James T. Wassell, Ph.D.
Pass Rate Data Analysis Methodology
 Nonparametric Bootstrapping
 data-based resampling of subjects
 10,000 panels generated
 If first test value is below the pass criteria, then test
a second time and use the second value
 2880 data points used in this analysis
 144 subjects x 2 test measurements (or replicates)
x 10 respirator models = 2880
 Outcome for each subject is pass / fail
 Outcome for each panel is pass / fail
Subject Pass Rate
 Subject Pass definitions
 Inward leakage values
 < 0.01 or Fit factor > 100 (most conservative)
 < 0.02 or Fit factor > 50
 < 0.05 or Fit factor > 20 (least conservative)
 “Second Chance”
 If the IL measurement on the first try is above the
pass criterion (i.e. it fails), then take a second
measurement to give a second chance to pass.
Panel Pass Rate
 The number of subjects in the panel must meet
the subject pass criteria (> 74%)
 For panels of size n=20, 15 (75%) or more subjects
must meet pass criterion
 For panels of size n=25, 19 (76%) or more subjects
must meet pass criterion
 For panels of size n=30, 23 (76.7%) or more
subjects must meet pass criterion
 For panels of size n=35, 26 (74.3%) or more
subjects must meet pass criterion
 For panels of size n=40, 30 (75%) or more subjects
must meet pass criterion
Bootstrap – Data-Based Resampling
35-member population-based Bivariate Panel
Cell # Subj
sample
1
12
2
12
3
12
4
27
5
12
6
12
7
21
8
12
9
12
10
12
total 144
2
2
4
9
2
2
7
3
2
2
35
Panel
Observations
n=35
Keep test 2
value.
NO
YES
Is test 1 value a pass?
Sample subjects with
replacement.
Keep test 1
value.
Bootstrap – Data-Based Resampling
40-member Equal Distribution Panel
Cell # Subj
sample
1
12
2
12
3
12
4
27
5
12
6
12
7
21
8
12
9
12
10
12
total 144
4
4
4
4
4
4
4
4
4
4
40
Panel
Observations
n=40
Keep test 2
value.
NO
YES
Is test 1 value a pass?
Sample subjects with
replacement.
Keep test 1
value.
Panels Investigated
key to table abbreviations
 equ 40: Equal cell distribution of 4, n=40
 equ 30: Equal cell distribution of 3, n=30
 equ 20: Equal cell distribution of 2, n=20
 pop 35: Population Based Panel of n=35
 pop 30: Population Based Panel of n=30
 pop 25: Population Based Panel of n=25
Panel pass rate tables
 The next four slides present the panel pass
rates tables to compare the 6 panels
 Tables 1 and 2 are for the equal distribution
panels (n=40, n=30, n=20)
 Table 1 shows five FFR models’ rates.
 Table 2 shows five Elastomeric models’ rates.
 Tables 3 and 4 are for the population based
panels (n=35, n=30, n=25)
 Table 3 shows five FFR models’ rates.
 Table 4 shows five Elastomeric models’ rates
Table 1. Panel Pass Rates
Equal Distribution Panels
FFR Respirator Models
for N=40 > 30, for N=30 > 23, for N=20 > 15
based on 10,000 bootstrap samples
Model
Subject Pass
Level*
N=40
N=30
N=20
B
IL< 0.05
0.4%
0.8%
4.5%
E
IL< 0.05
0.6%
1.0%
5.3%
C
IL< 0.05
15.2%
14.8%
28.2%
D
IL< 0.02
22.5%
21.8%
35.5%
A
IL< 0.02
67.5%
59.6%
67.8%
*IL= Inward Leakage
Table 2. Panel Pass Rates
Equal Distribution Panels
Elastomeric Respirator Models
for N=40 > 30, for N=30 > 23, for N=20 > 15
based on 10,000 bootstrap samples
Model
Subject Pass
Level*
N=40
N=30
N=20
H
IL< 0.02
23.4%
21.8%
36.2%
G
IL< 0.02
48.9%
43.1%
55.3%
I
IL< 0.02
72.0%
63.2%
71.8%
J
IL< 0.02
91.8%
85.6%
87.8%
F
IL< 0.01
99.5%
98.0%
97.9%
*IL= Inward Leakage
Table 3. Panel Pass Rates
Population Based Panels
FFR Respirator Models
for N=35 > 26, for N=30 > 23, for N=25 > 19
based on 10,000 bootstrap samples
Model
Subject Pass
Level*
N=35
N=30
N=25
B
IL< 0.05
0.7%
0.4%
1.6%
E
IL< 0.05
3.8%
2.1%
5.4%
C
IL< 0.05
16.7%
12.7%
16.8%
D
IL< 0.02
15.8%
12.0%
19.2%
A
IL< 0.02
52.0%
41.1%
44.2%
*IL= Inward Leakage
Table 4. Panel Pass Rates
Population Based Panels
Elastomeric Respirator Models
for N=35 > 26, for N=30 > 23, for N=25 > 19
based on 10,000 bootstrap samples
Model
Subject Pass
Level*
N=35
N=30
N=25
H
IL< 0.02
16.1%
13.0%
17.3%
G
IL< 0.02
39.1%
33.0%
43.9%
I
IL< 0.02
79.2%
69.0%
72.3%
J
IL< 0.02
93.4%
88.5%
87.6%
F
IL< 0.01
98.0%
94.3%
94.7%
*IL= Inward Leakage
Summary of Panel Pass Rates
 In the equal distribution panels:
 the highest pass rates are in the n=20 panels for all
of the FFR models and 2 of the elastomeric models.
 the highest pass rates are in the n=40 panels for 3
of the 5 elastomeric models.
 In the population based panels:
 the highest pass rates are in the n=25 panels for 4
of the 5 FFR models and 2 of the elastomeric
models.
 the highest pass rates are in the n=35 panels for 1
FFR model and for 3 of the 5 elastomeric models.
QUESTIONS?
Analysis of the number of subjects
in a panel who pass
 Another way to compare the 6 different panels
is to look at how many subjects in a panel
meet the subject pass criterion.
 For each panel the number of subjects that
meet the subject pass criterion was
determined.
 The number of subjects that pass for a
selected respirator models was plotted as
histograms to compare panels.
Histograms of the number of subjects
who pass
 The next four graphics show how the six
different panels compare.
 Graph 1 shows the six panels for the FFR with
the lowest pass rates
 Graph 2 shows the six panels for the FFR with
the highest pass rates
 Graph 3 shows the six panels for the
elastomeric with the lowest pass rates
 Graph 4 shows the six panels for the
elastomeric with the highest pass rates
equ20
0.20
0.15
0.10
0.05
0.00
4.5% passing
rate (%>15)
pop25
0.20
0.15
0.10
0.05
0.00
1.7% passing
rate (%>19)
pop30
0.20
0.15
0.10
0.05
0.00
0.4% passing
rate (%>23)
Density
equ30
0.20
0.15
0.10
0.05
0.00
0.7% passing
rate (%>23)
pop35
0.20
0.15
0.10
0.05
0.00
0.7% passing
rate (%>26)
equ40
0.20
0.15
0.10
0.05
0.00
0.4% passing
rate (%>30)
5
10
15
20
25
No. of Subjects who Pass in the Panel (IL<0.02)
30
equ20
36.2% passing rate
(% >15)
pop25
0.20
0.15
0.10
0.05
0.00
43.9% passing rate
(% >19)
pop30
13.0% passing rate
(%>23)
Density
equ30
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
21.8% passing rate
(%>23)
pop35
16.1% passing
rate (%>26)
0.20
0.15
0.10
0.05
0.00
0.20
0.15
0.10
0.05
0.00
equ40
23.4% passing
rate (%>30)
10
20
30
No. of Subjects who Pass in the Panel (IL<0.02))
0.20
0.15
0.10
0.05
0.00
Summary
 This comparison of histograms reveals more
information than the tables of pass rates.
 Histograms give a visual display of the mode
and the number of subjects that meet the pass
criterion for each of the six panels.
 Pass rates do not appear to give a definitive
determination of which of the six panels is
“best”.
 Other considerations may be more important
in determining which panel might be “best”.
QUESTIONS?
Inward Leakage Variability between
Respirator Fit Test Panels - Alternate
Analyses
Ziqing Zhuang, Ph.D.
Definition of the Problem
•
•
•
•
Inter- and Intra-subject variability
Inter-respirator variability
Inter-respirator sample variability (Zhuang et al., 2011)
Inter-panel variability
‒ Variability between two geographical locations
‒ Variability at the same location
Objectives
•
•
To determine if passing rates are significantly different
among different respirator fit test panels
To determine if passing/failing a respirator (one size fits all
or more than one size system or family) in inward leakage
test is consistent among different respirator fit test panels
Methods
•
From the 144 subjects, assign subjects randomly to one of
three panels (35 subjects)
Table 1. Randomized Assignment of Three 35-Member Panels
Bivariate
Panel Cell
No. of Subjects
No. of Subjects for Each Panel
A
B
C
1
12
2
2
2
2
12
2
2
2
3
12
4
4
4
4
27
9
9
9
5
12
2
2
2
6
12
2
2
2
7
21
7
7
7
8
12
3
3
3
9
12
2
2
2
10
12
2
2
2
All
144
35
35
35
Methods (Continued)
•
•
•
•
A subject is considered to pass the inward leakage (IL) test
if one of the two IL values is less than or equal to the
pass/fail level
Calculate passing rate for all 144 subjects and 10
respirators combined using three passing levels (IL ≤ 0.01,
0.02 or 0.05)
Calculate passing rate for each panel and each respirator
for the three passing levels (IL ≤ 0.01, 0.02 or 0.05)
Test if passing rates are significantly different among the
three respirator fit test panels
Methods (Continued)
•
•
A respirator is considered to pass the IL test if 26 or more
subjects pass the IL test
Determine if passing/failing a respirator in inward leakage
test is consistent among different respirator fit test panels
Results
Figure 1. Probability of Inward Leakage for all Respirators by Panel
1.00
0.90
0.80
Probability
0.70
0.60
Panel A
0.50
Panel B
0.40
Panel C
0.30
0.20
0.10
0.00
0.00
0.10
0.20
0.30
Inward Leakage
0.40
0.50
Figure 2. Probability of Inward Leakage by Panel for the
Respirator with the Lowest Mean IL
Respirator F
1.00
0.90
0.80
Probability
0.70
0.60
0.50
Panel A
0.40
Panel B
Panel C
0.30
0.20
0.10
0.00
0.00
0.10
0.20
0.30
Inward Leakage
0.40
0.50
Figure 3. Probability of Inward Leakage by Panel for the
Respirator with the Highest Mean IL
Respirator B
1.00
0.90
0.80
Probability
0.70
0.60
0.50
Panel A
0.40
Panel B
Panel C
0.30
0.20
0.10
0.00
0.00
0.10
0.20
0.30
Inward Leakage
0.40
0.50
Table 2. Passing Rate for All 144 subjects for Three Different
Pass/Fail Levels
IL ≤ 0.01
Pass/Fail Level
IL ≤ 0.02
IL ≤ 0.05
A (FFR)
66 (46)
109 (76)
130 (90)
B (FFR)
8 (6)
39 (27)
79 (55)
C (FFR)
19 (13)
49 (34)
100 (69)
D (FFR)
45 (31)
97 (67)
127 (88)
E (FFR)
17 (12)
43 (30)
86 (60)
F (Elastomeric)
125 (87)
140 (97)
142 (99)
G (Elastomeric)
50 (34)
106 (74)
137 (95)
H (Elastomeric)
61 (42)
96 (67)
136 (94)
I (Elastomeric)
53 (37)
112 (78)
140 (97)
J (Elastomeric)
88 (61)
121 (84)
136 (94)
All
532 (37)
912 (63)
1213 (84)
Respirator
Data were expressed as frequency (%).
Table 3. Comparison of Passing Rates (Inward Leakage ≤ 0.01)
among the Three 35-Member Panels
A
Panel
B
C
A (FFR)
B (FFR)
17 (49)
2 (6)
15 (43)
1 (3)
13 (37)
1 (3)
0.63
0.77
C (FFR)
2 (6)
5 (14)
4 (11)
0.49
D (FFR)
E (FFR)
F (Elastomeric)
G (Elastomeric)
10 (29)
2 (6)
29 (83)
11 (31)
7 (20)
5 (14)
30 (86)
11 (31)
14 (40)
5 (14)
29 (83)
12 (34)
0.18
0.43
0.93
0.96
H (Elastomeric)
13 (37)
16 (46)
17 (49)
0.60
I (Elastomeric)
13 (37)
14 (40)
12 (34)
0.88
J (Elastomeric)
20 (57)
20 (57)
24 (69)
0.53
All
119 (34)
124 (35)
131 (37)
0.64
Respirator
P value
Data were expressed as frequency (%). P value was estimated by ChiSquare test.
Table 4. Comparison of Passing Rates (Inward Leakage ≤ 0.02)
among the Three 35-Member Panels
A
Panel
B
C
A (FFR)
B (FFR)
28 (80)
13 (37)
26 (74)
5 (14)
20 (57)
8 (23)
0.09
0.09
C (FFR)
9 (26)
12 (34)
13 (37)
0.57
D (FFR)
E (FFR)
F (Elastomeric)
G (Elastomeric)
23 (66)
12 (34)
33 (94)
23 (66)
21 (60)
9 (26)
33 (94)
24 (69)
22 (63)
8 (23)
35 (100)
26 (74)
0.88
0.54
0.35
0.73
H (Elastomeric)
21 (60)
20 (57)
28 (80)
0.09
I (Elastomeric)
27 (77)
24 (69)
29 (83)
0.37
J (Elastomeric)
31 (89)
29 (83)
29 (83)
0.74
All
220 (63)
203 (58)
218 (62)
0.35
Respirator
P value
Data were expressed as frequency (%). P value was estimated by ChiSquare test.
Table 5. Comparison of Passing Rates (Inward Leakage ≤ 0.05)
among the Three 35-Member Panels
A
Panel
B
C
A (FFR)
B (FFR)
32 (91)
22 (63)
32 (91)
21 (60)
29 (83)
14 (40)
0.58
0.11
C (FFR)
23 (66)
26 (74)
24 (69)
0.73
D (FFR)
E (FFR)
F (Elastomeric)
G (Elastomeric)
30 (86)
20 (57)
35 (100)
35 (100)
31 (89)
23 (66)
33 (94)
33 (94)
30 (86)
21 (60)
35 (100)
33 (94)
0.92
0.76
0.33
0.54
H (Elastomeric)
31 (89)
32 (91)
34 (97)
0.53
I (Elastomeric)
34 (97)
33 (94)
35 (100)
0.77
J (Elastomeric)
34 (97)
33 (94)
35 (100)
0.77
All
296 (85)
297 (85)
290 (83)
0.74
Respirator
P value
Data were expressed as frequency (%). P value was estimated by Chi-Square
test.
Summary of Findings
•
•
•
•
•
Passing rate was not significantly different among the three
panels for all respirators combined or by each model
This was true for all pass/fail levels of 0.01, 0.02 and 0.05
Using 26 or more subjects to pass the IL test, all three
panels had consistent passing/failing results for pass/fail
levels of 0.01 and 0.05
Some disagreement was observed for 0.02 level
Pass/fail level is an important factor affecting inter-panel
variability
Discussion
•
•
Results in these studies are consistent with the findings by
Landsittel et al. (2013 In press) and Zhuang (2012
presentation at ISRP conference in Boston)
Further simulation analyses can be conducted to estimate
the probability that two panels will both pass or fail a
respirator and to investigate the effect of pass/fail level and
sample size on the probability
Rejection Probabilities for 25 Subjects
100%
Percentage of Time a Given Respirator Will Fail
90%
Beta Error
80%
70%
50%
60%
60%
50%
70%
40%
80%
30%
90%
Alpha Error
20%
10%
0%
13
14
15
16 17 18 19 20 21 22
Number of Subjects Required to Pass
23
24
25
Effect of Sample Size on Error Rates
Alpha Error
(20%)
Beta Error
(5%)
Number of
Sample Subjects
Size
Required
90%
80%
60%
50%
to Pass Effective Effective Effective Effective
10
8 (80%)
7.0%
32.2%
16.3%
5.5%
16
12 (75%)
1.7%
20.2%
16.7%
3.8%
25
19 (76%)
0.9%
22.0%
7.4%
0.7%
35
26 (74%)
0.2%
14.6%
5.8%
0.3%
40
30 (75%)
0.1%
16.1%
3.5%
0.1%
50
37 (74%)
<0.1%
11.1%
2.8%
<0.1%
Conclusions
•
•
Inter-panel variability exists, but it is small
The concern about inter-panel variability and other types of
variability can be alleviated by properly selecting:
‒ Pass/fail level (IL 0.01 to 0.05)
‒ Panel size (e.g., 25 or 35)
‒ Minimum number of subjects required to pass (e.g., 26
of 35 or 23 of 35)
QUESTIONS?
Panel Implementation
Colleen Miller, AAS, BS
NIOSH Implementation Options - Panels
 NIOSH bivariate panel
 Principal Component Analysis (PCA) Panel
 Continue to use LANL Panels
 Develop a NIOSH Bivariate half-facepiece
panel using face length and lip length
Use of NIOSH Bivariate Panel
 Stakeholders have expressed concern about
arbitrarily assigning sizing to specific boxes or
cells within a panel
 May be design restrictive
 No correlation between panel cell designations and
respirator fit
 Use of panels to “fit to size” or “size to fit”?
Panel ISO 17420-1 Respiratory Protective
Devices – Supplied Breathable Gas RPD
 25 test subjects (for RPD designed to fit whole population) shall be
tested.
 Subjects will be distributed at a minimum of 3 per cell in 8 cell of
the Principle Component Analysis (PCA) with the remaining subject
randomly distributed.
 Note: The PCA-distribution is specified in ISO 16900-1.
 For RPD designated by the manufacturer to fit a select population,
identified by specific PCA cells the following test panels apply:
(minimum of 3 subjects per cell)





a. up to 3 PCA cells - 10 subjects
b. up to 4 cells - 13 subjects
c. up to 5 cells - 16 subjects
d. up to 6 cells - 19 subjects
e. up to 7 cells - 22 subjects
Panel ISO 17420-1 Respiratory Protective
Devices – Filtering RPD
 25 test subjects (for RPD designed to fit whole population) shall be
tested.
 Subjects will be distributed at a minimum of 3 per cell in 8 cell of
the Principle Component Analysis (PCA) with the remaining subject
randomly distributed.
 Note: The PCA-distribution is specified in ISO 16900-1.
 For RPD designated by the manufacturer to fit a select population,
identified by specific PCA cells the following test panels apply:
(minimum of 3 subjects per cell)





a. up to 3 PCA cells - 10 subjects
b. up to 4 cells - 13 subjects
c. up to 5 cells - 16 subjects
d. up to 6 cells - 19 subjects
e. up to 7 cells - 22 subjects
Principal Component Analysis Panel
Long/Narrow Nose
Long Face
Shape
Short Face
Short/Wide Nose
Small
Overall Size
Large
Overview – NIOSH Standard Testing
Procedure (STP) Panel Structure
Determination of Qualitative Isoamyl
Acetate (IAA) Facepiece Fit, Air-Purifying
Respirators Standard Testing Procedure
(STP) Purpose
 This test establishes the procedure for ensuring that
the level of protection provided by the isoamyl
acetate facepiece fit test requirements on airpurifying respirators for Approval, Extension of
Approval, or examined during Certified Product
Audits, meet the minimum certification standards set
forth in 42 CFR Part 84, Subpart G, Section
84.63(a)(c)(d), Subpart I, Section 84.124, Subpart L,
Section 84.205, and Subpart KK, Section 84.1141 and
84.1142; Volume 69, Number 110, June 8, 1995.
Determination of Laboratory Respirator Protection
Level (LRPL) Values for CBRN Self-Contained
Breathing Apparatus (SCBA) Facepieces or CBRN
Air-Purifying Respirator (APR), Standard Testing
Procedure
 Respiratory Protection factors provided by SCBA
facepiece converted to a SCBA facepiece negative
pressure configuration or APR, full facepiece
configurations submitted for new approval, extension
of approval or examined during certified product
audits, meet the minimum certification standards
prescribed in 42 CFR Part 84, Subpart G, Sections
84.63(a)(c)(d), the Statement of Standard for Chemical,
Biological, Radiological, and Nuclear CBRN OpenCircuit SCBA and APR
Air-Purifying Escape Respirators (APER)
 Laboratory Respirator Protection Level (LRPL)
and Human Subject Breathing Gas
Concentration (HSBG) Requirements
 Uses LANL full facepiece panel
 Head circumference
 Neck circumference
Current NIOSH Certification Panel for
Elastomeric Half-Facepiece Respirators
(LANL 10-member Panel number of subjects evaluated per cell)
LANL Full Facepiece Panel
Face Width (mm)
117.5
126.5
133.5
135.5
2 subjects
Cell 9
Medium/Large
123.5
2 subjects
cell 6
Small/Medium
5 subjects
Cell 7
Medium/Large
2 subjects
Cell 3
Small/Medium
4 subjects
Cell 4
Small/Medium
2 subjects
Cell 5
Medium/Large
2 subjects
Cell 1
Small/Medium
2 subjects
Cell 2
Small/Medium
113.5
103.5
93.5
144.5 153.5
2 subjects
Cell 10
Medium/Large
2 subjects
Cell 8
Medium/Large
LANL Full Facepiece Panel – Two sizes
Face Width (mm)
117.5
126.5
133.5
135.5
2 subjects
Cell 9
Medium/Large
123.5
4 subjects
Cell 6
Small/Medium
5 subjects
Cell 7
Medium/Large
2 subjects
Cell 3
Small/Medium
4 subjects
Cell 4
Small/Medium
4 subjects
Cell 5
Medium/Large
2 subjects
Cell 1
Small/Medium
2 subjects
Cell 2
Small/Medium
113.5
103.5
93.5
144.5 153.5
2 subjects
Cell 10
Medium/Large
2 subjects
Cell 8
Medium/Large
LANL Full Facepiece Panel – Three sizes or more
Face Width (mm)
117.5
126.5
135.5
144.5 153.5
133.5
3 subjects
Cell 9
Large
3 subjects
Cell 10
Large
123.5
3 subjects
Cell 6
Medium
6 subjects
Cell 7
3Medium 3Large
113.5
103.5
93.5
4 subjects
Cell 3
Small or Medium
6 subjects
Cell 4
Small or Medium
2 subjects
Cell 1
Small
3 subjects
Cell 2
Small
3 subjects
Cell 5
Medium or
Large
5 subjects
Cell 8
2Medium 3Large
Facepiece Size
Panel Cells
included
Total Subjects
LANL HALF
Total Subjects
LANL FULL
Small
1,2,3,4
6–1or 2 each cell
10-2 or 3 each cell
Medium
3,4,5,6,7,8
6–1 each cell
17-2 or 3 each cell
Large
7,8,9,10
6–1or 2 each cell
11-2 or 3 each cell
Small/Medium
1,2,3,4,5,6
6–1 each cell
14-2 or 3 each cell
Medium/Large
5,6,7,8,9,10
6–1 each cell
15-2 or 3 each cell
XSmall or XLarge
1 or 10
Varies/review
One size fits all
1 through 10
10–1 each cell
Hoods/Helmets
25-2 or 3 each cell
Non CBRN same as above, CBRN includes neck/head
Determination of Man Test for Gases and
Vapors-Gas Tightness Test-Isoamyl
Acetate, Type C and CE, Supplied Air
Respirators–Standard Testing Procedure
 This test establishes the procedure for ensuring
that the level of protection provided by the man
test for vapors and gases requirement on type C
and CE Supplied-Air Respirators for Approval,
Extension of Approval, or examined during
Certified Product Audits, meet the minimum
certification standards set forth in 42 CFR Part
84, Subpart G, Section 84.63(a)(c)(d), Subpart J,
Sections 84.159, 84.162 and 84.163; Volume 69,
Number 110, June 8, 1995.
 Requires only 2 test subjects
Panel Variability Study Data
Provide data to facilitate comparisons of cell
placement and current derived “sizes” of the 144
subjects included in the study using the NIOSH
Bivariate, the LANL half-facepiece panel and the
LANL full-facepiece panel.
Bivariate Panel - Three Facepiece Sizes
Small
Medium
Large
Bivariate Panel - Two Facepiece Sizes
Smaller size
Larger size
Scatter Plot against NIOSH & LANL Full
Facepiece Panels (2003 Survey US Workers)
NIOSH
Panel
LANL
Panel
“Size”
LANL
FullFacepiece
Cell
“Size”
Subject
Bivariate
Cell
LANL HalfFacepiece
Cell
4
1
4
S/M
4
S/M
14
1
2
S
1
S
15
1
2
S
1
S
16
1
4
S/M
4
S/M
57
1
2 or 4
S/M
2 or 4
S/M
59
1
1 or 3
S/M
2 or 4
S/M
85
1
3
S/M
4
S/M
88
1
4
S/M
4
S/M
96
1
3
S/M
129
1
4
S/M
4
S/M
143
1
3
S/M
3
S/M
147
1
3
S/M
4
S/M
NIOSH Bivariate Cell 1 Implementation
 The subjects included in Bivariate Cell 1, all female, are
considered S/M by the current sizing designations we use
for certification testing today.
 Are manufacturers designing for XSmall faces?
 What classes of respirators?
 How can NIOSH address the certification fit testing of
facepieces designed to fit XSmall faces?
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
“Size”
8
2
4
S/M
5
M
17
2
4
S/M
4
S/M
34
2
4
S/M
4
S/M
39
2
5
M
69
2
4
S/M
4
S/M
76
2
4
S/M
5
M
79
2
2
S
out
80
2
3
S/M
4
81
2
2
S
out
82
2
91
2
2
95
2
4
S/M
5
M
S
2
S
S/M
4
S/M
NIOSH Bivariate Cell 2 Implementation
 The subjects included in Bivariate Cell 2, 11
female and 1 male, are designated by the LANL
half-facepiece panel as more small to medium
 The LANL full-facepiece panel designates the
same subjects as relatively medium
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
7
3
4
S/M
4
S/M
18
3
4
S/M
4
S/M
19
3
4
S/M
4
S/M
21
3
4
S/M
40
3
4 or 6
S/M
47
3
6
M
64
3
4
S/M
73
3
4
S/M
87
3
4
S/M
89
3
97
3
3
S/M
4
105
3
7
M/L
out
4 or 7
4
4
“Size”
LANL
Fullfacepiece
Cell
“Size”
S/M/L
S/M
S/M
out
S/M
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
“Size”
2
4
3
S/M
4
S/M
10
4
4
S/M
5
M
13
4
7
M/L
7
M/L
23
4
7
M/L
7
M/L
24
4
4
S/M
5
M
25
4
4
S/M
5
M
33
4
6
M/L
7
M/L
46
4
4
S/M
4
S/M
54
4
7
M/L
6
M
55
4
4
S/M
5
M
65
4
4
S/M
5
M
75
4
7
M/L
7
M/L
115
4
4 or 7
S/M/L
5 or 7
M/L
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
125
4
3
S/M
5
M
135
4
4
S/M
4
S/M
136
4
4
S/M
5
M
144
4
6
M/L
7
M/L
157
4
7
M/L
7
M/L
162
4
3
S/M
5
M
165
4
4
S/M
5
M
170
4
3
S/M
5
M
171
4
4
S/M
5
M
172
4
6
M/L
7
M/L
175
4
7
M/L
7
M/L
131
4
7
M/L
7
M/L
148
4
4 or 7
S/M/L
5,7 or 8
M/L
153
4
5
S/M
out
“Size”
LANL
Fullfacepiece
Cell
“Size”
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
1
5
3 or 6
S/M
8
M/L
9
5
4 or 7
S/M/L
8
M/L
12
5
5
M
out
44
5
8
M/L
8
M/L
53
5
8
M/L
8
M/L
61
5
8
M/L
8
M/L
72
5
out
74
5
out
100
5
8
M/L
103
5
8
M/L
104
5
out
124
5
7
“Size”
LANL
Fullfacepiece
Cell
“Size”
M/L
8
M/L
NIOSH Bivariate Cell 5 Implementation
 The subjects included in the Bivariate panel cell 5
tend to be considered larger by the current size
designations.
 Subjects currently testing as Cell 8 LANL Full fit into
Cell 5 Bivariate
 Cell 5 no longer “small/medium”
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
28
6
9
L
out
31
6
7
M/L
6
32
6
9
L
out
66
6
7
M/L
6
90
6
out
92
6
6
94
6
out
117
6
9 or 10
L
out
133
6
6
M
6
M
150
6
7
M/L
6
M
158
6
9
L
out
159
6
7
M/L
6
“Size”
M
M
M
M
Bivariate Cell 6 Implementation
 The subjects included in Bivariate panel cell 6 are
considered medium to large by the current size
designations of the LANL half-facepiece panel and
medium by the LANL full-facepiece panel –
reality?
 Cell 6 subject measurements could vary from small
to large
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
“Size”
3
7
7
M/L
7
M/L
22
7
9
L
9
L
36
7
7
M/L
7
M/L
41
7
7
M/L
8
M/L
48
7
7
M/L
7
M/L
50
7
9
L
9
L
51
7
7
M/L
6
M
52
7
8
M/L
6
M
56
7
7 or 8
M/L
7
M/L
62
7
7
M/L
7 or 8
M/L
71
7
7
M/L
7
M/L
93
7
6
M
7
M/L
LANL
Halffacepiece
Cell
LANL
Fullfacepiece
Cell
“Size”
7
M/L
Subject
Bivariate
Cell
106
7
110
7
8
M/L
8
M/L
111
7
7
M/L
7
M/L
112
7
7
M/L
6 or 7
M/L
123
7
7
M/L
6 or 7
M/L
126
7
8
M/L
8
M/L
140
7
7
M/L
9
L
146
7
7
M/L
7
M/L
169
7
9
L
9 OR 10
L
“Size”
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
11
8
9
L
out
20
8
9
L
10
L
29
8
9
L
10
L
35
8
7
M/L
8
L
45
8
9
L
10
L
60
8
7
M/L
out
63
8
9
L
out
83
8
8
L
8
L
86
8
8
L
8
L
98
8
out
99
8
8
L
102
8
8
L
“Size”
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
6
9
out
26
9
10
L
9
27
9
9
L
out
30
9
out
38
9
10
43
9
out
58
9
9
L
9
L
67
9
9
L
9
L
70
9
out
114
9
10
L
10
L
122
9
9
L
9
L
164
9
out
“Size”
LANL
Fullfacepiece
Cell
“Size”
10
L
L
out
L
9
L
out
out
out
Subject
Bivariate
Cell
LANL
Halffacepiece
Cell
“Size”
LANL
Fullfacepiece
Cell
5
10
10
L
out
37
10
9
L
10
L
49
10
9
L
10
L
68
10
9
L
10
L
77
10
L
out
78
10
out
84
10
10
101
10
out
out
108
10
out
out
120
10
9 or 10
L
10
L
132
10
9 or 10
L
10
L
138
10
10
L
out
“Size”
out
L
10
L
NIOSH Bivariate Cells 7-10
Implementation
 Cell 7 subjects are “medium/large”
 Cell 8, 9, and 10 subjects are “large”
 Must there be limits to face width?
Creating Standard Testing Procedures
using the New NIOSH Bivariate Panel
and panel variability data analysis
 For respirators available in 3 sizes, what “clouds” or
groups of cells should be used to further evaluate the
panel variability data or for standard testing procedures?
 “Small” includes cells 1,2,3 or 1,2,3,4 or 1,2,3,4 and 6
 “Medium” includes cells 3,4,6,7 or 4,5,6,7
 “Large” includes cells 5,7,8,9,10 or 5,8,9,10 or 7,8,9,10 or only
8,9,10
 Same clouds for OSFA respirators and facepieces available in
3 or more sizes?
*Zhuang
Z, Bradtmiller B, and Shaffer, RE. New Respirator Fit Test Panels Representing the Current U.S.
Cilivian Work Force. JOEH 4:647-659, 2007
3-Size Bivariate LANL
Creating Standard Testing Procedures
using the New NIOSH Bivariate Panel
and panel variability data continued
 For respirators available in 2 sizes, what “clouds” or
groups of cells should be used for certification
testing?
 1,2,3,4 or 1,2,3,6, or 1,2,3,4,6 or cells 1-6
 4,5,7,8,9,10 or cells 5-10
*Zhuang

Include cell 5 in the “small/medium”

Include cell 6 in the “medium/large”
Z, Bradtmiller B, and Shaffer, RE. New Respirator Fit Test Panels Representing
the Current U.S. Cilivian Work Force. JOEH 4:647-659, 2007.
2-Size Bivariate LANL
25-member NPPTL Lip Length Panel
Summary – Panel Discussion
 NIOSH is seeking stakeholder feedback
concerning today’s presentation and
discussion
 NIOSH is aware of the complexity of the panel
decision given our current procedures, ISO
proposed standards and a changing workforce
Posting of Materials
 Materials from both webinars will be available
at:
http://www.cdc.gov/niosh/docket
 Docket 250 will be opened and webinar
materials will be posted
Quality Partnerships Enhance Worker
Safety & Health
Visit Us at: http//www.cdc.gov/niosh/npptl
Disclaimer:
The findings and conclusions in this presentation have not been
formally disseminated by the National Institute for Occupational
Safety and Health and should not be construed to represent any
agency determination or policy.
Thank you