Assigned Protection Factors

DEPARTMENT OF LABOR

Occupational Safety and Health Administration

29 CFR Parts 1910, 1915, and 1926

[Docket No. H049C]

RIN 1218-AA05

Assigned Protection Factors

AGENCY:

Occupational Safety and Health Administration (OSHA), Department of Labor.

ACTION:

Proposed rule; request for comments and scheduling of informal public hearings.

SUMMARY:

OSHA is proposing to revise its existing Respiratory Protection Standard to add definitions and specific requirements for assigned protection factors (APFs) and maximum use concentrations (MUCs). The proposed revisions also would supersede the respirator selection provisions of existing substance-specific standards with these new APFs (except the APFs for the 1,3-Butadiene Standard).

The Agency developed the proposed APFs after thoroughly reviewing the available literature, including chamber simulation studies and workplace protection factor studies. The proposed APFs would provide employers with critical information to use when selecting respirators for employees exposed to atmospheric contaminants found in general industry, construction, shipyard, longshoring, and marine terminal workplaces. Proper respirator selection using APFs is an important component of an effective respirator protection program. Accordingly, OSHA has made a preliminary conclusion that the proposed APFs are necessary to protect employees who use respirators against atmospheric contaminants.

DATES:

Written comments.

The Agency invites interested parties to submit written comments regarding the proposed rule, including comments to the information-collection determination under the Supplementary Information section of this

Federal Register

notice, by mail, facsimile, or electronically. You must send all comments, whether submitted by mail, facsimile, or electronically through OSHA's Web site, by September 4, 2003.

Informal public hearings.

The Agency plans to hold an informal public hearing in Washington, DC in late summer or early fall of 2003. OSHA expects the DC hearing to last from 9:30 a.m. to 5:30 p.m. on the first day, and from 8:30 a.m. to 5:30 p.m. on subsequent days; however, the exact daily schedule is at the discretion of the presiding administrative law judge. If an additional hearing is held, the Agency will announce the date, time, and location of this hearing later in the subsequent

Federal Register

notice.

Notice of intention to appear to provide testimony at the informal public hearing.

Interested parties who intend to present testimony at the informal public hearing in Washington, DC, must notify OSHA of their intention to do so no later than September 4, 2003.

Hearing testimony and documentary evidence.

Interested parties who will be requesting more than 10 minutes to present their testimony, or who will be submitting documentary evidence at the hearing, must provide the Agency with copies of their full testimony and all documentary evidence they plan to present by September 4, 2003.

ADDRESSES:

Written comments.

You may submit three copies of written comments to the Docket Office, Docket No. H049C, Technical Data Center, Room N-2625, OSHA, U.S. Department of Labor, 200 Constitution Ave., NW., Washington, DC 20210; telephone (202) 693-2350. If your written comments are 10 pages or fewer, you may fax them to the OSHA Docket Office, telephone number (202) 693-1648. You do not have to send OSHA a hard copy of your faxed comments. You may submit comments electronically through OSHA's Home page at

http://ecomments.osha.gov/.

If you would like to submit additional studies or journal articles, you must submit three copies of them to the OSHA Docket Office at the address above. These materials must clearly identify your electronic comments by name, date, subject, and docket number so we can attach them to your comments.

Informal public hearings.

The informal public hearing to be held in Washington, DC will be located in the Auditorium on the plaza level of the Frances Perkins Building, U.S. Department of Labor, 200 Constitution Ave., NW., Washington, DC.

Notice of intention to appear to provide testimony at the informal public hearing.

Notices of intention to appear at the informal public hearing should be submitted in triplicate to the Docket Office, Docket No. H049C, Room N-2625, OSHA, U.S. Department of Labor, 200 Constitution Avenue, NW., Washington, DC 20210. Notices may also be faxed to the Docket Office at (202) 693-1648 or submitted electronically at

http://ecomments.osha.gov.

OSHA Docket Office and Department of Labor hours of operation are 8:15 a.m. to 4:45 p.m.

Hearing testimony and documentary evidence.

Interested parties who will be requesting more than 10 minutes to present their testimony, or who will be submitting documentary evidence at the informal public hearing must mail three copies of the testimony and the documentary evidence to the Docket Office, Docket No. H049C, Room N-2625, OSHA, U.S. Department of Labor, 200 Constitution Avenue, NW., Washington DC 20210. Additional information for submitting testimony and evidence is found under

SUPPLEMENTARY INFORMATION.

FOR FURTHER INFORMATION CONTACT:

For technical inquiries, contact Mr. John E. Steelnack, Directorate of Standards and Guidance, Room N-3718, OSHA, U.S. Department of Labor, 200 Constitution Ave., NW., Washington, DC 20210; telephone (202) 693-2289 or fax (202) 693-1678. For hearing information contact Ms. Veneta Chatmon, OSHA Office of Information, Docket No. H-49C, Room N-3649, U.S. Department of Labor, 200 Constitution Ave., NW., Washington, DC 20210 (telephone (202) 693-1999). For additional copies of this

Federal Register

notice, contact the Office of Publications, Room N-3103, OSHA, U.S. Department of Labor, 200 Constitution Ave., NW., Washington, DC 20210 (telephone (202) 693-1888). Electronic copies of this

Federal Register

notice, as well as news releases and other relevant documents, are available at OSHA's Home page at

http://www.osha.gov.

SUPPLEMENTARY INFORMATION:

OMB Review Under the Paperwork Reduction Act

After a thorough analysis of the proposed provisions, OSHA believes that these provisions would not add to the existing collection-of-information (

i.e.

, paperwork) requirements regarding respirator selection. OSHA determined that its existing Respiratory Protection Standard at 29 CFR 1910.134 has two provisions that involve APFs and also impose paperwork requirements on employers. These provisions require employers to: Include respirator selection in their written respiratory protection program (29 CFR 1910.134(c)(1)(i)); and inform employees regarding proper respirator selection (29 CFR 1910.(k)(ii)). The information on respirator selection addressed by these two provisions must include a brief discussion of the purpose of APFs, and how to use them in selecting a respirator that affords an employee protection from airborne contaminants. The burden imposed by this requirement remains the same

whether employers currently use the APFs published in the 1987 NIOSH RDL or the ANSI Z88.2-1992 Respiratory Protection Standard, or implement the APFs proposed in this rulemaking. Therefore, the proposed use of APFs in the context of these two existing respirator-selection provisions does not require an additional paperwork-burden determination because OSHA already accounted for this burden under its existing Respiratory Protection Standard (

see

63 FR 1152-1154; OMB Control Number 1218-0099).

Both OSHA's existing Respiratory Protection Standard and the proposed APF provisions require employers to use APFs as part of the respirator-selection process. This process includes obtaining information about the workplace exposure level to an airborne contaminant, identifying the exposure limit (

e.g.

, permissible exposure limit) for the contaminant, using this information to calculate the required level of protection (

i.e.

, the APF), and referring to an APF table to determine which respirator to select. Admittedly, this process involves the collection and use of information, but it does not require employers to inform others, either orally or in writing, about the process they use to select respirators for individual employees, or the outcomes of this process; by not requiring employers to communicate this information to others, OSHA removed this process from the ambit of the Paperwork Reduction Act of 1995 (PRA-95) (44 U.S.C. 3506(c)(2)(A)). In the alternative, even if PRA-95 applies, the proposal involves the same information-collection and -use requirements with regard to APFs as the existing standard (

see

paragraphs (d)(1) and (d)(3)(i) of 29 CFR 1910.134, and the rationale for the existing APF requirements in the preamble to the final Respiratory Protection Standard, 63 FR 1163 and 1203-1204); accordingly, the paperwork burden imposed by the proposal would be equivalent to the burden already imposed under the existing standard.

Interested parties who want to comment on OSHA's determination that the proposed provisions contain no additional paperwork burden compared to the existing paperwork requirements must send their written comments to the Office of Information and Regulatory Affairs, Attn: OMB Desk Officer for OSHA, Office of Management and Budget, Room 10235, 725 17th Street NW., Washington, DC 20503. The Agency also encourages commenters to submit their comments on this paperwork determination to OSHA along with their other comments.

Federalism

The Agency reviewed the proposed APF provisions according to the most recent Executive Order on Federalism (Executive Order 13132, 64 FR 43225, August 10, 1999). This Executive Order requires that federal agencies, to the extent possible, refrain from limiting state policy options, consult with states before taking actions that restrict their policy options, and take such actions only when clear constitutional authority exists and the problem is of national scope. The Executive Order allows federal agencies to preempt state law only with the expressed consent of Congress; in such cases, federal agencies must limit preemption of state law to the extent possible.

Under section 18 of the Occupational Safety and Health Act (the “Act” or “OSH Act”), Congress expressly provides OSHA with authority to preempt state occupational safety and health standards to the extent that the Agency promulgates a federal standard under section 6 of the Act. Accordingly, section 18 of the Act authorizes the Agency to preempt state promulgation and enforcement of requirements dealing with occupational safety and health issues covered by OSHA standards unless the state has an OSHA-approved occupational safety and health plan (

i.e.

, is a state-plan state) [

see Gade

v.

National Solid Wastes Management Association,

112 S. Ct. 2374 (1992)]. Therefore, with respect to states that do not have OSHA-approved plans, the Agency concludes that this proposal conforms to the preemption provisions of the Act. Additionally, section 18 of the Act prohibits states without approved plans from issuing citations for violations of OSHA standards; the Agency finds that the proposed rulemaking does not expand this limitation.

OSHA asserts that it has authority under Executive Order 13132 to propose APF requirements because the problems addressed by these requirements are national in scope. As noted in section VI (“Summary of the Preliminary Economic Analysis and Initial Regulatory Flexibility Analysis”) of this preamble, hundreds of thousands of employers must select appropriate respirators for millions of employees. These employees are exposed to many different types and levels of airborne contaminants found in general industry, construction, shipyard, longshoring, and marine terminal workplaces. Accordingly, the proposed requirements would provide employers in every state with critical information to use when selecting respirators to protect their employees from the risks of exposure to airborne contaminants. However, while OSHA drafted the proposed APF and MUC requirements to protect employees in every state, section 18(c)(2) of the Act permits state-plan states to develop their own requirements to deal with any special workplace problems or conditions, provided these requirements are at least as effective as the final requirements that result from this proposal.

State Plans

The 26 states and territories with their own OSHA-approved occupational safety and health plans must adopt comparable provisions within six months after the Agency publishes the final APF and MUC requirements. These states and territories are: Alaska, Arizona, California, Hawaii, Indiana, Iowa, Kentucky, Maryland, Michigan, Minnesota, Nevada, New Mexico, North Carolina, Oregon, Puerto Rico, South Carolina, Tennessee, Utah, Vermont, Virginia, Virgin Islands, Washington, and Wyoming. Connecticut, New Jersey and New York have OSHA approved State Plans that apply to state and local government employees only. Until a state-plan state promulgates its own comparable provisions, Federal OSHA will provide the state with interim enforcement assistance, as appropriate.

Unfunded Mandates

The Agency reviewed the proposed APF and MUC provisions according to the Unfunded Mandates Reform Act of 1995 (UMRA) (2 U.S.C. 1501

et seq.

) and Executive Order 12875. As discussed in section VI (“Summary of the Preliminary Economic Analysis and Initial Regulatory Flexibility Analysis”) of this preamble, OSHA estimates that compliance with this proposal would require private-sector employers to expend about $4.5 million each year. However, while this proposal establishes a federal mandate in the private sector, it is not a significant regulatory action within the meaning of section 202 of the UMRA (2 U.S.C. 1532).

OSHA standards do not apply to state and local governments, except in states that have voluntarily elected to adopt an OSHA-approved state occupational safety and health plan. Consequently, the proposed provisions do not meet the definition of a “Federal intergovernmental mandate” [

see

section 421(5) of the UMRA (2 U.S.C. 658(5)]. Therefore, based on a review of the rulemaking record to date, the Agency believes that few, if any, of the affected employers are state, local, and tribal governments. Therefore, the

proposed APF requirements do not impose unfunded mandates on state, local, and tribal governments.

Protecting Children From Environmental Health and Safety Risks

Executive Order 13045 requires that Federal agencies submitting covered regulatory actions to OMB's Office of Information and Regulatory Affairs (OIRA) for review pursuant to Executive Order 12866 must provide OIRA with (1) an evaluation of the environmental health or safety effects that the planned regulation may have on children, and (2) an explanation of why the planned regulation is preferable to other potentially effective and reasonably feasible alternatives considered by the agency. Executive Order 13045 defines “covered regulatory actions” as rules that may (1) be economically significant under Executive Order 12866 (

i.e.

, a rulemaking that has an annual affect on the economy of $100 million or more, or would adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs, the environment, public health or safety, or state, local, or tribal governments or communities), and (2) concern an environmental health risk or safety risk that an agency has reason to believe may disproportionately affect children. In this context, the term “environmental health risks and safety risks” means risks to health or safety that are attributable to products or substances that children are likely to come in contact with or ingest (

e.g.

, through air, food, water, soil, product use).

The proposed provisions are not economically significant under Executive Order 12866 (

see

section VI (“Summary of the Preliminary Economic Analysis and Initial Regulatory Flexibility Analysis”) of this preamble). In addition, after reviewing the proposed APF provisions, OSHA has determined that these provisions do not impose environmental health or safety risks to children as set forth in Executive Order 13045. The proposed provisions would require employers to use APFs in selecting proper respirators for employee use, with the objective of limiting employee exposures to airborne contaminants. To the best of OSHA's knowledge, no employees under 18 years of age work under conditions that require respirator use. However, if such conditions exist, children who use respirators selected according to these proposed provisions would receive adequate protection from the airborne contaminants. In this regard, the Agency is requesting public comment on whether employees under the age of 18 years use respirators, and, if they do, the extent to which the respirators provide them with adequate protection. Based on this discussion, OSHA believes that the APF and MUC requirements proposed in this rulemaking do not constitute a covered regulatory action as defined by Executive Order 13045.

Applicability of Existing Consensus Standards

Section 6(b)(8) of the OSH Act requires OSHA to explain “why a rule promulgated by the Secretary differs substantially from an existing national consensus standard,” by publishing “a statement of the reasons why the rule as adopted will better effectuate the purposes of the Act than the national consensus standard.” [

see

29 U.S.C. 655(b)(8)]. Accordingly, the Agency compared the proposed APF requirements with the APF provisions of ANSI Z88.2-1992 (“Respiratory Protection”). This consensus standard, published by the American National Standards Institute in 1992, is the only publicly available consensus standard that includes APFs. In most instances, the APFs being proposed by the Agency are identical to ANSI's APFs, however, some differences exist. Where OSHA has proposed an APF that differs from ANSI's, the Summary and Explanation provides the basis for that decision.

Environmental Impact Assessment

The Agency reviewed the proposed provisions according to the National Environmental Policy Act (NEPA) of 1969 (42 U.S.C. 4321

et seq.

), the regulations of the Council of Environmental Quality (40 CFR part 1500), and the Department of Labor's NEPA procedures (29 CFR part 11). OSHA estimates that this proposed rule would have a direct impact on a relatively small number of respirator users and, in so doing , merely alter the type of respirator they are using. The Agency does not anticipate that this will significantly alter solid waste patterns, water quality, or ambient air quality. As a result of this review, OSHA concludes that the proposed provisions would have no significant environmental impact.

I. General

Table of Contents

The following Table of Contents identifies the major preamble sections of this proposal and the order in which they are presented:

Introductory Material

Notice and Comment

Dates for Hearings

Supplementary Information

OMB Review Under the Paperwork Reduction Act

Federalism

State Plans

Unfunded Mandates

Protecting Children from Environmental Health and Safety Risks

Applicability of Existing Consensus Standards

Environmental Impact Assessment

I. General

Table of contents

Glossary

II. Pertinent Legal Authority

III. Events Leading to the Proposed Standard

A. Regulatory History

B. Need for Assigned Protection Factors

C. Review of the Proposed Standard by the Advisory Committee for Construction Safety and Health (ACCSH)

IV. Methodology for Developing Assigned Protection Factors

A. Dr. Nicas' Proposal and Response from Commenters

B. Analyses of WPF Studies

C. Analyses of SWPF Studies

D. OSHA's Overall Summary Conclusions

E. Summaries of Studies

V. Health Effects

VI. Summary of the Preliminary Economic Analysis and Initial Regulatory Flexibility Screening Analysis

VII. Summary and Explanation of the Proposed Standard

A. Revisions to the Respiratory Protection Standard

B. Superseding the Respirator Selection Provisions of Substance-Specific Standards in Parts 1910, 1915, and 1926

VIII. Issues

IX. Public Participation—Comments and Hearings

X. Proposed Amendments to Standards

Glossary

This glossary specifies the terms represented by acronyms, and provides definitions of other terms, used frequently in this proposal. This glossary does not change the legal requirements as proposed in this notice of proposed rulemaking, nor is it intended to propose new regulatory requirements or definitions. It is presented simply to assist the reader.

A. Acronyms

ACGIH:

American Conference of Governmental Industrial Hygienists.

AIHA:

American Industrial Hygiene Association.

ANSI:

American National Standards Institute.

APF:

Assigned Protection Factor (

see

definition in proposed regulatory text).

DOP:

Dioctylphthalate (an aerosolized agent used for quantitative fit testing).

DFM:

Dust/Fume/Mist filter.

EPF:

Effective Protection Factor (

see

definition below under “Protection factor study”).

HEPA:

High efficiency particulate air [filter] (

see

definition below).

IDLH:

Immediately dangerous to life or health (

see

definition below).

LANL:

Los Alamos National Laboratory.

LLNL:

Lawrence Livermore National Laboratory.

MSHA:

Mine Safety and Health Administration.

MUC:

Maximum Use Concentration (

see

definition in proposed regulatory text).

NIOSH:

National Institute for Occupational Safety and Health.

NRC:

Nuclear Regulatory Commission.

OSHA:

Occupational Health and Safety Administration.

PAPR:

Powered air-purifying respirator (

see

definition below).

PEL:

Permissible Exposure Limit (an occupational exposure level specified by OSHA).

PPF:

Program Protection Factor (

see

definition below under “Protection factor study”).

QLFT:

Qualitative fit test (

see

definition below).

QNFT:

Quantitative fit test (

see

definition below).

RDL:

Respirator Decision Logic (respirator selection guidance developed by NIOSH that contains a set of respirator protection factors).

REL:

Recommended Exposure Limit (an occupational exposure level recommended by NIOSH).

SAR:

Supplied-air respirator (

see

definition below).

SCBA:

Self-contained breathing apparatus (

see

definition below).

WPF:

Workplace Protection Factor (

see

definition below under “Protection factor study”).

TLV:

Threshold Limit Value (an occupational exposure level recommended by ACGIH).

SWPF:

Simulated Workplace Protection Factor (

see

definition below under “Protection factor study”).

B. Definitions

Terms followed by an asterisk (*) refer to definitions that can be found in paragraph (b) (“Definitions”) of OSHA's Respiratory Protection Standard (29 CFR 1910.134).

Air-purifying respirator*:

A respirator with an air-purifying filter, cartridge, or canister that removes specific air contaminants by passing ambient air through the air-purifying element.

Atmosphere-supplying respirator*:

A respirator that supplies the respirator user with breathing air from a source independent of the ambient atmosphere, and includes SARs and SCBA units.

Canister or cartridge*:

A container with a filter, sorbent, or catalyst, or combination of these items, which removes specific contaminants from the air passed through the container.

Continuous flow respirator :

An atmosphere-supplying respirator that provides a continuous flow of breathable air to the respirator facepiece.

Demand respirator*:

An atmosphere-supplying respirator that admits breathing air to the facepiece only when a negative pressure is created inside the facepiece by inhalation.

Filter or air-purifying element*:

A component used in respirators to remove solid or liquid aerosols from the inspired air.

Filtering facepiece (or dust mask)*:

A negative pressure particulate respirator with a filter as an integral part of the facepiece or with the entire facepiece composed of the filtering medium.

Fit factor*:

A quantitative estimate of the fit of a particular respirator to a specific individual, and typically estimates the ratio of the concentration of a substance in ambient air to its concentration inside the respirator when worn.

Fit test*:

The use of a protocol to qualitatively or quantitatively evaluate the fit of a respirator on an individual.

Helmet*:

A rigid respiratory inlet covering that also provides head protection against impact and penetration.

High-efficiency particulate air filter*:

A filter that is at least 99.97% efficient in removing monodisperse particles of 0.3 micrometers in diameter. The equivalent NIOSH 42 CFR 84 particulate filters are the N100, R100, and P100 filters.

Hood*:

A respiratory inlet covering that completely covers the head and neck and may also cover portions of the shoulders and torso.

Immediately dangerous to life or health*:

An atmosphere that poses an immediate threat to life, would cause irreversible adverse health effects, or would impair an individual's ability to escape from a dangerous atmosphere.

Loose-fitting facepiece*:

A respiratory inlet covering that is designed to form a partial seal with the face.

Negative pressure respirator (tight-fitting)*:

A respirator in which the air pressure inside the facepiece is negative during inhalation with respect to the ambient air pressure outside the respirator.

Positive pressure respirator*:

A respirator in which the pressure inside the respiratory inlet covering exceeds the ambient air pressure outside the respirator.

Powered air-purifying respirator*:

An air-purifying respirator that uses a blower to force the ambient air through air-purifying elements to the inlet covering.

Pressure demand respirator*:

A positive pressure atmosphere-supplying respirator that admits breathing air to the facepiece when the positive pressure is reduced inside the facepiece by inhalation.

Protection factor study:

A study that determines the protection provided by a respirator during use. This determination is generally accomplished by measuring the ratio of the concentration of an agent (

e.g.

, hazardous substance) outside the respirator (Co) to the agent's concentration inside the respirator (Ci) (

i.e.

, Co/Ci). Therefore, as the ratio between Co and Ci increases, the protection factor increases, indicating an increase in the level of protection provided to employees by the respirator. Four types of protection factor studies are:

Effective Protection Factor (EPF) study

—a study, conducted in the workplace, that measures the protection provided by a properly selected, fit-tested, and functioning respirator when used intermittently for only some fraction of the total workplace exposure time (

i.e.

, sampling is conducted during periods when respirators are worn and not worn). EPFs are not directly comparable to WPF values because the determinations include both the time spent in contaminated atmospheres with and without respiratory protection; therefore, EPFs tend to understate the protection that would be obtained if the respirator were being worn at all times.

Program Protection Factor (PPF) study

—a study that estimates the protection provided by a respirator within a specific respirator program. Like the EPF, it is focused not only on the respirator's performance, but also the effectiveness of the complete respirator program. PPFs are affected by all factors of the program, including respirator selection and maintenance, user training and motivation, work activities, and program administration.

Workplace Protection Factor (WPF) study

—a study, conducted under actual conditions of use in the workplace, that measures the protection provided by a properly selected, fit-tested, and functioning respirator, when the respirator is correctly worn and used as part of a comprehensive respirator program. Measurements of Co and Ci are obtained only while the respirator is being worn during performance of normal work tasks (

i.e.

, samples are not collected when the respirator is not being worn). As the degree of protection afforded by the respirator increases, the WPF increases.

Simulated Workplace Protection Factor (SWPF) study

—a study, conducted in a controlled laboratory setting and in which Co and Ci

sampling is performed while the subject performs a series of set exercises. The laboratory setting is used to control many of the variables found in workplace studies, while the exercises simulate the work activities of respirator users. This type of study is designed to determine the optimum performance of respirators by reducing the impact of sources of variability through maintenance of tightly controlled study conditions.

Qualitative fit test*:

A pass/fail fit test to assess the adequacy of respirator fit that relies on the individual's response to the test agent.

Quantitative fit test*:

An assessment of the adequacy of respirator fit by numerically measuring the amount of leakage into the respirator.

Self-contained breathing apparatus*:

An atmosphere-supplying respirator for which the breathing air source is designed to be carried by the user.

Supplied-air respirator (or airline) respirator*:

An atmosphere-supplying respirator for which the source of breathing air is not designed to be carried by the user.

Tight-fitting facepiece*:

A respiratory inlet covering that forms a complete seal with the face.

II. Pertinent Legal Authority

The purpose of the Occupational Safety and Health Act, 29 U.S.C. 651

et seq.

(the “OSHA Act” or “Act”) is to “assure so far as possible every working man and woman in the Nation safe and healthful working conditions and to preserve our human resources.” [29 U.S.C. 651(b)]. To achieve this goal, Congress authorized the Secretary of Labor to promulgate and enforce occupational safety and health standards [

see

29 U.S.C. 654(b) (requiring employers to comply with OSHA standards), 29 U.S.C. 655(a) (authorizing summary adoption of existing consensus and federal standards within two years of the Act's enactment), and 29 U.S.C. 655(b) (authorizing promulgation of standards pursuant to notice and comment)].

A safety or health standard is a standard “which requires conditions, or the adoption or use of one or more practices, means, methods, operations, or processes, reasonably necessary or appropriate to provide safe or healthful employment or places of employment.” [29 U.S.C. 652(8)]. A standard is reasonably necessary or appropriate within the meaning of section 652(8) of the Act when it substantially reduces or eliminates significant risk, and is technologically and economically feasible, cost effective, consistent with prior Agency action or supported by a reasoned justification for departing from prior Agency action, and supported by substantial evidence; it must also effectuate the Act's purposes better than any national consensus standard it supersedes [

see International Union, UAW

v.

OSHA

(

LOTO II

), 37 F.3d 665 (DC Cir. 1994; and 58 FR 16612-16616 (March 30, 1993)].

OSHA has discussed the nature of adverse health effects caused by exposure to airborne chemical hazards many times in previous rulemaking activities [

see,

for example, the preambles to any of OSHA's substance-specific standards codified in 29 CFR 1910.1001 to 1910.1052]. As discussed in the Significance of Risk section of the Respiratory Protection Standard, the health risk presented to workers can be represented by the risk that a respirator will not be properly selected or used, which increases the possibility that the user will be overexposed to a harmful air contaminant. The risks that are addressed by the Respiratory Protection Standard are not characterized as illness-specific risks but, instead, relate to a more general probability that when a respirator provides insufficient protection, the wearer may be exposed to a level of air contaminant that is associated with material impairment of the worker's health.

The Agency believes that a standard is technologically feasible when the protective measures it requires already exist, can be brought into existence with available technology, or can be created with technology that can reasonably be expected to be developed [

see American Textile Mfrs. Institute

v.

OSHA

(

Cotton Dust

), 452 U.S. 490, 513 (1981);

American Iron and Steel Institute

v.

OSHA

(

Lead II

), 939 F.2d 975, 980 (DC Cir. 1991)]. A standard is economically feasible when industry can absorb or pass on the costs of compliance without threatening the industry's long-term profitability or competitive structure [

see Cotton Dust,

452 U.S. at 530 n. 55;

Lead II,

939 F.2d at 980], and a standard is cost effective when the protective measures it requires are the least costly of the available alternatives that achieve the same level of protection [

see Cotton Dust,

453 U.S. at 514 n. 32;

International Union, UAW

v.

OSHA

(

LOTO III

), 37 F.3d 665, 668 (DC Cir. 1994)].

All standards must be highly protective [see 58 FR 16612, 16614-15 (March 30, 1993);

LOTO III,

37 F.3d at 669]. Accordingly, section 8(g)(2) of the Act authorizes OSHA “to prescribe such rules and regulations as [it] may deem necessary to carry out its responsibilities under the Act” [

see

29 U.S.C. 657(g)(2)]. However, health standards must also meet the “feasibility mandate” of section 6(b)(5) of the OSH Act, 29 U.S.C. 655(b)(5). Section 6(b)(5) of the Act requires OSHA to select “the most protective standard consistent with feasibility” needed to reduce significant risk when regulating health hazards [

see Cotton Dust,

452 U.S. at 509]. Section 6(b)(5) also directs OSHA to base health standards on “the best available evidence,” including research, demonstrations, and experiments [

see

29 U.S.C. 655(b)(5)]. In this regard, OSHA must consider “in addition to the attainment of the highest degree of health and safety protection * * * the latest scientific data * * * feasibility and experience gained under this and other health and safety laws.” (Id.). Furthermore, section 6(b)(5) of the Act specifies that standards must “be expressed in terms of objective criteria and of the performance desired” [

see

29 U.S.C. 655(b)(7)].

The proposed APF and MUC provisions are integral components of an effective respiratory protection program. Respiratory protection is a supplemental method used by employers to protect employees against airborne contaminants in workplaces where feasible engineering controls and work practices are not available, have not yet been implemented, or are not in themselves sufficient to protect employee health. Employers also use respiratory protection under emergency conditions involving the accidental release of airborne contaminants. The proposed amendments to OSHA's Respiratory Protection Standard, and the Agency's substance-specific standards, would provide employers with critical information to use when selecting respirators for employees exposed to airborne contaminants found in general industry, construction, shipyard, longshoring, and marine terminal workplaces. Since it is generally recognized that different types of respiratory protective equipment provide different degrees of protection against hazardous exposures, proper respirator selection is of critical importance. The proposed APF and MUC provisions provide additional guidance on the point at which an increase in the level of respiratory protection is necessary. The APF and MUC provisions will greatly enhance an employer's ability to select a respirator that will adequately protect employees. OSHA believes that in the absence of these proposed provisions, employers will be less certain about which respirators to select for adequate employee protection.

The Agency also developed the proposed provisions to be feasible and cost effective, and is specifying them in terms of objective criteria and the level of performance desired. In this regard, section VI (“Summary of the Preliminary Economic Analysis and Initial Regulatory Flexibility Analysis”) of this preamble provides the benefits and costs of this proposal, and describes several other alternatives as required by section 205 of the UMRA (2 U.S.C. 1535). Based on this information, OSHA preliminarily concludes that the proposed APF and MUC provisions constitute the most cost-effective alternative for meeting its statutory objective of reducing risk of adverse health effects to the extent feasible.

III. Events Leading to the Proposed Standard

A. Regulatory History

Congress created the Occupational Safety and Health Administration (OSHA) in 1970, and gave it the responsibility for promulgating standards to protect the health and safety of American workers. As directed by the OSH Act, the Agency adopted existing Federal standards and national consensus standards developed by various organizations such as the American Conference of Governmental Industrial Hygienists (ACGIH), the National Fire Protection Association (NFPA), and the American National Standards Institute (ANSI). The ANSI standard Z88.2-1969, “Practices for Respiratory Protection,” was the basis of the first six sections (permissible practice, minimal respirator program, selection of respirators, air quality, use, maintenance and care) of OSHA's Respiratory Protection Standard (29 CFR 1910.134) adopted in 1971. The seventh section was a direct, complete incorporation of ANSI Standard K13.1-1969, “Identification of Gas Mask Canisters.”

The Agency promulgated an initial Respiratory Protection Standard for the construction industry (29 CFR 1926.103) in April 1971. On February 9, 1979, OSHA formally applied 29 CFR 1910.134 to the construction industry (44 FR 8577). Agencies that preceded OSHA developed the original maritime respiratory protection standards in the 1960s (

e.g.

, section 41 of the Longshore and Harbor Worker Compensation Act). The section designations adopted by OSHA for these standards, and their original promulgation dates, are: Shipyards—29 CFR 1915.82, February 20, 1960 (25 FR 1543); Marine Terminals—29 CFR 1917.82, March 27, 1964 (29 FR 4052); and Longshoring—29 CFR 1918.102, February 20, 1960 (25 FR 1565). OSHA incorporated 29 CFR 1910.134 by reference into its Marine Terminal standards (Part 1917) on July 5, 1983 (48 FR 30909). The Agency updated and strengthened its Longshoring and Marine Terminal standards in 1996 and 2000, and these standards now incorporate 29 CFR 1910.134 by reference.

Under the Respiratory Protection Standard that OSHA initially adopted, employers needed to follow the guidance of the Z88.2-1969 ANSI standard to ensure proper selection of respirators. Subsequently, OSHA published an Advance Notice of Proposed Rulemaking (“ANPR”) to revise the Respiratory Protection Standard on May 14, 1982 (47 FR 20803). Part of the impetus for this notice was the Agency's inclusion of new respirator requirements in the comprehensive substance-specific standards promulgated under Section (6)(b) of the OSH Act,

e.g.

, fit testing protocols, respirator selection tables, use of PAPRs, changing filter elements whenever an employee detected an increase in breathing resistance, and requirements referring employees with breathing difficulties to a physician trained in pulmonary medicine, either at fit testing or during routine respirator use [

see, e.g,

29 CFR 1910.1025 (OSHA's Lead Standard)]. The respirator provisions in these substance-specific standards took into account advances in respirator technology and changes in related guidance documents that were state-of-the-art when OSHA published these substance specific standards and, in particular, recognized that effective respirator use depends on a comprehensive respiratory protection program that includes use of APFs.

OSHA's 1982 ANPR sought information on the effectiveness of its current Respiratory Protection Standard, the need to revise this standard, and suggestions on the nature of the revisions. The 1982 ANPR referenced the ANSI Z88.2-1980 standard on respiratory protection with its table of protection factors, the 1976 report by Dr. Ed Hyatt from the LASL titled “Respiratory Protection Factors” (Ex. 2), and the RDL developed jointly by OSHA and NIOSH, as revised in 1978 (Ex. 9, Docket No. H049). Questions #2, #3, and #4 in the 1982 ANPR asked for comments on how OSHA should use protection factors. The Agency received responses from 81 interested parties. The commenters generally supported revising OSHA's Respiratory Protection Standard, and provided recommendations regarding approaches for including a table of protection factors (Ex. 15).

On September 17, 1985, OSHA announced the availability of a preliminary draft of the proposed Respiratory Protection Standard. This preproposal draft standard included the public comments received in response to 1982 ANPR, and OSHA's own analysis of revisions needed in the Respiratory Protection Standard to account for state-of-the-art respiratory protection. The Agency received 56 responses from interested parties (Ex. 36) which OSHA carefully reviewed in developing the proposal.

On November 15, 1994, OSHA published the proposed rule to revise 29 CFR 1910.134, and provided public notice of an informal public hearing on the proposal (59 FR 58884). The Agency convened the informal public hearing on June 6, 1995. On June 15, 1995, as part of the public hearing, OSHA held a one-day panel discussion by respirator experts of APFs. Areas discussed included difficulties in measuring performance of respiratory protection in WPF and SWPF studies, statistical uncertainties regarding the distribution of data from these studies, and the problems associated with setting APFs for all respirators that protect all potential respirator users across a wide variety of workplaces and exposure conditions.

OSHA reopened the rulemaking record for the revised Respiratory Protection Standard on November 7, 1995 (60 FR 56127), requesting comments on a study performed for OSHA by Dr. Mark Nicas titled “The Analysis of Workplace Protection Factor Data and Derivation of Assigned Protection Factors' (Ex. 1-156). That study, which the Agency placed in the rulemaking docket on September 20, 1995, addressed the use of statistical modeling for determining respirator APFs. OSHA received 12 comments on the Nicas report. This report, and the comments received in response to it, convinced OSHA that more information would be necessary before it could resolve the complex issues regarding how to establish APFs, including what methodology to use in analyzing existing protection factor studies (

see

Section IV below for a more detailed explanation of the Nicas report and the comments made on it).

OSHA published the final, revised Respiratory Protection Standard, 29 CFR 1910.134, on January 8, 1998 (63 FR 1152). The standard contains worksite-specific requirements for program administration, procedures for respirator selection, employee training, fit testing, medical evaluation, respirator

use, and other provisions. However, OSHA reserved the sections of the final standard related to APFs and maximum use concentration (MUC) pending further rulemaking (see 63 FR 1182 and 1203). The Agency stated that, until a future rulemaking on APFs is completed:

[Employers must] take the best available information into account in selecting respirators. As it did under the previous [Respiratory Protection] standard, OSHA itself will continue to refer to the [APFs in the 1987 NIOSH RDL] in cases where it has not made a different determination in a substance specific standard. (

see

63 FR 1163)

The Agency subsequently established a separate docket (

i.e.

, H049C) for the APF rulemaking. This docket includes copies of material related to APFs that it previously placed in the docket (H049) for the revised Respiratory Protection Standard. The APF rulemaking docket also contains other APF-related materials, studies, and data that OSHA obtained after it promulgated the final Respiratory Protection Standard in 1998.

History of Assigned Protection Factors

In 1965, the Bureau of Mines published “Respirator Approval Schedule 21B,” which contained the term “protection factor” as part of its approval process for half-mask respirators (for protection up to 10 times the TLV) and full facepiece respirators (for protection up to 100 times the TLV). The Bureau of Mines based these protection factors on quantitative fit tests, using dioctyl pthalate (DOP), that were conducted on six male test subjects performing simulated work exercises.

The Atomic Energy Commission (AEC) published proposed protection factors for respirators in 1967, but later withdrew them because quantitative fit testing studies were available for some, but not all, types of respirators. To address this shortcoming, the AEC subsequently sponsored respirator studies at LASL, starting in 1969.

ANSI standard Z88.2-1969, which OSHA adopted by reference in 1971, did not contain APFs for respirator selection. Nevertheless, this ANSI standard recommended that “due consideration be given to potential inward leakage in selecting devices,” and contained a list of the various respirators grouped according to the quantity of leakage into the facepiece expected during routine use.

In 1972, NIOSH and the Bureau of Mines published new approval schedules for respiratory protection under 30 CFR Part 11. However, these new approval schedules did not include fit testing provisions as part of the respirator certification process.

NIOSH sponsored additional respirator studies at LASL, beginning in 1971, that used quantitative test systems to measure the overall performance of respirators. Dr. Edwin C. Hyatt of LASL included a table of protection factors for, single-use dust respirators; quarter-mask, half-mask, and full facepiece air-purifying respirators; and SCBAs in a 1976 report titled “Respirator Protection Factors” (Ex. 2). The protection factors were based on data from DOP and sodium chloride quantitative fit test studies performed on these respirators at LASL between 1970 and 1973. The table also contained recommended protection factors for respirators that had no performance test data. Dr. Hyatt based these recommended protection factors on the judgment and experience of LASL researchers, as well as extrapolations from available facepiece leakage data for similar respirators. For example, he assumed that performance data for SCBAs operated in the pressure demand mode could be used to represent other (non-tested) respirators that maintain positive pressure in the facepiece, hood, helmet, or suit during inhalation. In addition, he recommended in his report that NIOSH continue testing the performance of respirators that lacked adequate fit test data. Relative to this, staff members at LASL (from 1974 to 1978) used a representative 35-person test panel to conduct quantitative fit tests on all air-purifying particulate respirators approved by the Bureau of Mines and NIOSH.

In August 1975, the Joint NIOSH-OSHA Standards Completion Program published the RDL (Ex. 25-4, Appendix F, Docket No. H049). The RDL contained a table of protection factors that were based on quantitative fit testing performed at LASL and elsewhere, as well as the expert judgment of the RDL authors. The 1978 NIOSH update of the RDL contained the following protection factors:

5 for single-use respirators;

10 for half-mask respirators with DFM or HEPA filters;

50 for full facepiece air-purifying respirators with HEPA filters or chemical cartridges;

1,000 for PAPRs with HEPA filters;

1,000 for half-mask SARs operated in the pressure demand mode;

2,000 for full facepiece SARs operated in the pressure demand mode; and

10,000 for full facepiece SCBAs operated in the pressure demand mode.

ANSI's respiratory protection Subcommittee decided to revise Z88.2-1969 in the late 1970s. During its deliberations, the Subcommittee conducted an extensive discussion regarding the role of respirator protection factors in an effective respiratory protection program. As a result, the Subcommittee decided to add an APF table to the revised standard. In May 1980, ANSI published the revision as Z88.2-1980 (Ex. 10, Docket No. H049) and it contained the first ANSI Z88.2 respiratory protection factor table. The ANSI Subcommittee based the table on Hyatt's protection factors, which it updated using results from fit testing studies performed at LANL and elsewhere since 1973. For example, the protection factor for full facepiece air-purifying particulate respirators was 100 when qualitatively fit tested, or 1,000 when equipped with high efficiency filters and quantitatively fit tested. The table consistently gave higher protection factors to tight-fitting facepiece respirators when employers performed quantitative fit testing rather than qualitative fit testing. The ANSI Subcommittee concluded that PAPRs (with any respiratory inlet covering), atmosphere-supplied respirators (in continuous flow or pressure demand mode), and pressure demand SCBAs required no fit testing because they operated in a positive pressure mode. Accordingly, it gave these respirators high protection factors, limited only by IDLH values. The Subcommittee assigned protection factors of 10,000 and over to respirators used in IDLH atmospheres.

In response to a complaint to NIOSH that the PAPRs used in a plant did not appear to provide the expected protection factor of 1,000, Myers and Peach of NIOSH conducted a WPF study during silica bagging operations. Myers and Peach tested half-mask and full facepiece PAPRs and found protection factors that ranged from 16 to 215. They published the results of the study in 1983 (Ex. 1-64-46). The results of this study led NIOSH and other researchers, as well as respirator manufacturers, to perform additional WPF studies on PAPRs and other respirators.

NIOSH revised its RDL in 1987 (Ex. 1-54-437Q). While the revision retained many of the provisions of the 1978 RDL, it recognized the problems involved in developing APFs. The 1987 RDL also revised the APFs for some respirators, based on NIOSH's WPF studies. For example, the APFs were lowered for the following respirator classes: PAPRs with a loose-fitting hood or helmet to 25; PAPRs with a tight-fitting facepiece and a HEPA filter to 50; supplied-air continuous flow hoods or helmets to 25; and supplied-air continuous flow tight-fitting facepiece respirators to 50.

NIOSH stated that it may revise the 1987 RDL if warranted by subsequent WPF studies.

In August 1992, ANSI again revised its Z88.2 Respiratory Protection Standard (Ex. 1-50). The ANSI Z88.2-1992 standard contained a revised APF table, based on the Z88.2 Subcommittee's review of the available protection factor studies. In a report describing the revised standard (Ex. 1-64-423), Nelson, Wilmes, and daRoza described the rationale used by the ANSI Subcommittee in setting APFs:

If WPF studies were available, they formed the basis for the [APF] number assigned. If no such studies were available, then laboratory studies, design analogies, and other information was used to decide what value to place in the table. In all cases where the assigned protection factor changed when compared to the 1980 standard, the assigned number is lower in the 1992 standard.

In addition, the 1992 ANSI Z88.2 standard abandoned the 1980 standard's practice of giving increased protection factors to some respirators if quantitative fit testing was performed.

Tom Nelson, the co-chair of the ANSI Z88.2-1992 Subcommittee, published a second report, entitled “The Assigned Protection Factor According to ANSI” (Ex. 135), four years after the Z88.2 Subcommittee completed the revised 1992 standard. In the report, he reviewed the reasoning used by the ANSI Subcommittee in setting the 1992 ANSI APFs. He noted that the Z88.2 Subcommittee gave an APF of 10 to all half-mask air-purifying respirators, including quarter-mask, elastomeric, and disposable respirators. The Subcommittee also recommended that full facepiece air-purifying respirators retain an APF of 100 (from the 1980 ANSI standard) because no new data were available to justify another value. The Z88.2 Subcommittee also reviewed the 1987 NIOSH RDL values, particularly the RDL's reduction of loose-fitting facepiece and PAPRs with helmets or hoods to an APF of 25 based on their performance in WPF studies. For half-mask PAPRs, the ANSI Subcommittee set an APF of 50 based on a WPF study by Lenhart (Ex. 1-64-42). The ANSI Subcommittee had no WPF data available for full facepiece PAPRs, so it decided to select an APF of 1,000 to be consistent with the APF for PAPRs with helmets or hoods. The Subcommittee, in turn, based its APF of 1,000 for PAPRs with helmets or hoods on design analogies (

i.e.

, same facepiece designs, operation at the same airflow rates) between these respirators and airline respirators. Nelson noted that a subsequent WPF report by Keys (Ex. 1-64-40) on PAPRs with helmets or hoods was consistent with an APF of 1,000. According to Nelson, the Subcommittee used WPF studies by Myers (Ex. 1-64-48), Gosselink (Ex. 1-64-23), Myers (Ex. 1-64-47), and Que Hee and Lawrence (Ex. 1-64-60) to set an APF of 25 for PAPRs with loose-fitting facepieces. Nelson stated that two WPF studies, conducted by Gaboury and Burd (Ex. 1-64-24) and Stokes (Ex. 1-64-66) subsequent to publication of ANSI Z88.2-1992, supported the APF of 25 selected by the Subcommittee for PAPRs with loose-fitting facepieces.

Tom Nelson stated in his report that the ANSI Subcommittee had no new information on atmosphere-supplying respirators. Therefore, the APFs for these respirators were based on analogies with other similarly designed respirators (Ex. 135). The ANSI Subcommittee based the APF of 50 for half-mask continuous flow atmosphere-supplying respirators, and the APF of 25 for loose-fitting facepiece continuous flow atmosphere-supplying respirators, on the similarities between these respirators and PAPRs with the same airflow rates. Nelson noted that the ANSI Subcommittee set the APF of 1,000 for full facepiece continuous flow atmosphere-supplying respirators to be consistent with the APF for SARs with helmets or hoods found in two earlier studies—a WPF study by Johnson (Ex. 1-64-36) and a SWPF study by Skaggs (Ex. 1-3803). The Subcommittee used the analogy between PAPRs and continuous flow supplied-air respirators to select the APF of 50 for half-mask pressure demand SARs and 1,000 for full facepiece pressure demand SARs. Nelson stated: “The committee believed that setting a higher APF because of the pressure demand feature was not warranted, but rather that the total airflow was critical.”

Nelson noted in the report that the Subcommittee selected no APF for SCBAs. In explaining the committee's decision, he stated that “the performance of this type of respirator may not be as good as previously measured in quantitative fit test chambers.” Nelson also observed that the ANSI 88.2-1992 standard justified this approach in a footnote to the APF table. The footnote states:

A limited number of recent simulated workplace studies concluded that all users may not achieve protection factors of 10,000. Based on [these] limited data, a definitive assigned protection factor could not be listed for positive pressure SCBAs. For emergency planning purposes where hazardous concentrations can be estimated, an assigned protection factor of no higher than 10,000 should be used.

A new ANSI Z88.2 Subcommittee currently is reviewing the ANSI Z88.2-1992 standard, in accordance with ANSI policy specifying that each standard receive a periodic review. This review likely will result in revisions to the Z88.2 APF table based on WPF and SWPF respirator performance studies conducted since publication of the current standard in 1992.

B. Need for APFs

The proposed APF definition and regulatory text are important additions to, and an integral part of, OSHA's Respiratory Protection Standard because employers need this information to select appropriate respirators for employee use when engineering and work-practice controls are insufficient to maintain hazardous substances at safe levels in the workplace. Employers need the consistent and valid information contained in the proposed APF provisions to select respirators for employee protection, based on the type of hazardous substance and the level of employee exposure to that substance.

As noted in Table I of the proposed regulatory text, the proposed APFs differ for each class of respirator. In this regard, the proposed APF for a class of respirators specifies the workplace level of protection that class of respirator should provide under an effective respiratory protection program. Therefore, when the concentration of a hazardous substance in the workplace is less than 10 times the PEL, the employer must select a respirator from a respirator class with an APF of at least 10 for use by employees exposed to that substance. However, when the concentration of the hazardous substance is greater than 10 times the PEL, the employer must select a respirator that has an APF greater than 10 for this purpose. In addition, employers would derive MUCs from the APFs proposed for the different respirator classes. These MUCs determine the maximum atmospheric concentration of toxic gasses and vapors at which respirators equipped with cartridges and canisters can be used to protect employees.

In summary, when used in conjunction with the existing provisions of the Respiratory Protection Standard, especially the respirator selection requirements specified in paragraph (d), the proposed APF definition and regulatory text would provide employers with the information they need to select the appropriate respirators for reducing employee exposures to hazardous substances to safe levels. Accordingly, integrating the proposed APF provisions into the Respiratory Protection Standard will

ensure that employees receive the optimum level of protection afforded by that standard.

C. Review of the Proposed Standard by the Advisory Committee for Construction Safety and Health (ACCSH)

The proposed provisions would replace the existing respirator-selection requirements specified by the Respiratory Protection Standard for the construction industry (29 CFR 1926.103). Accordingly, OSHA's regulation governing the Advisory Committee on Construction Safety and Health (ACCSH) at 29 CFR 1912.3 requires OSHA to consult with the ACCSH whenever the Agency proposes a rulemaking that involves the occupational safety and health of construction employees. On December 5, 2002, OSHA briefed the ACCSH membership on the proposed provisions and responded to their questions. On March 27, 2003, the APF proposal was distributed to the ACCSH membership for their review prior to their next regular meeting on May 22, 2003. OSHA staff discussed the APF proposal and answered questions from the ACCSH members during their meeting on May 22, 2003. The ACCSH then recommended that OSHA proceed with publishing the proposal.

IV. Methodology for Developing Assigned Protection Factors

This section contains an overview of the analyses performed for OSHA and summaries of the studies used in these analyses. OSHA entered the complete analyses and studies into Docket H049C as Exhibits 3, 4, and 5 and Exhibit 1-156 (Dr. Nicas' report). Studies and information supporting the APF for each class of respirator are discussed in Section VII of this document. The analyses discussed below assisted OSHA in determining its proposed approach to deriving APFs. Commenters expressed appreciation for the approach suggested by Dr. Nicas, but nearly all did not support implementation of his methods. However, his recommendations provided guidance to the Agency regarding the types of studies and data needed for determining APFs. Dr. Brown's complex statistical analyses demonstrated the widespread variability inherent in current workplace protection factor studies. However, he found in his final analysis that the performance of filtering facepiece and elastomeric half-mask respirators could not be differentiated, thereby supporting grouping of these two types of respirator under one APF.

A. Dr. Nicas' Proposal and Response From Commenters

During the June 1995 APF hearings, OSHA devoted a full day to a panel discussion on the uncertainties associated with sample statistics and their use for deriving APFs. Based on this discussion, OSHA contracted with Dr. Mark Nicas to develop a statistical method for deriving APFs. Nicas used two approaches to account for within-wearer and between-wearer variabilities. For penetration data collected from a specific cohort of respirator wearers, he used a one-factor lognormal analysis of variance. He used a two-factor lognormal analysis of variance to perform a meta-analysis of the data from studies of different cohorts of respirator wearers. Using these approaches, Nicas proposed assigning two different protection factors; he recommended one for chronic toxicants (

i.e.

, substances regulated by an 8-hour PEL), and the other for acute toxicants (

i.e.

, substances regulated by a STEL). Nicas also made recommendations regarding sampling data management and inclusion of studies in statistical analyses of respirator performance.

OSHA reopened the rulemaking record on November 7, 1995 (60 FR 56127) to request comment on Dr. Nicas' report titled “The Analysis of Workplace Protection Factor Data and Derivation of Assigned Protection Factors” (Ex. 1-156). OSHA received 12 comments on the report. While some commenters expressed general support for Nicas' approach (

e.g.

, Ex. 1-182-4, American College of Occupational and Environmental Medicine), others had serious reservations about establishing APFs using this approach. The issues raised by these commenters are described below.

1. Lack of Valid and Reliable WPF Data

Two commenters stated that the available WPF data were of insufficient quality to permit a sophisticated statistical analysis. The 3M Company (3M) commended OSHA for “attempting to use science to evaluate workplace studies for determining Assigned Protection Factors,” but stated that insufficient valid data were available for such an evaluation, and that the data that were available were too variable (Ex. 1-182-5). In addition, Organization Resource Counselors, Inc. (ORC) stated: “The use of existing, often flawed, workplace protection factor studies, is not a solution to the problem. * * * A reliance on sophisticated statistics in an attempt to compensate for a lack of reliable scientific data on respirator performance is both bad science and bad policy” (Ex. 1-182-10).

2. Inappropriate Use of ANOVA Model

Three commenters believed that using Nicas' lognormal ANOVA model to analyze existing data was inappropriate (Exs. 1-174, 1-182-5, 1-182-1). Two of these commenters advocated using a simple analysis of the aggregate data instead (Exs. 1-174, 1-182-5). Thomas Nelson (Ex. 1-174) and 3M (1-182-5) expressed concern that the ANOVA model focuses primarily on within-wearer and between-wearer variability, while ignoring the potential variability contributed by other sources such as work site, respirator model, filter, and contaminant. Nelson stated: “A simple analysis of the entire data (

i.e.

, geometric mean, estimates of percentiles and confidence intervals) includes these and other possible sources of variation and the within-person variability in the model.” Two other commenters, Drs. Rappaport and Kupper [contractors for the Industrial Safety Equipment Association (ISEA)] believed that using an ANOVA model provided some benefits; however, they had concerns regarding the assumption of log-normality of penetration values, the lack of validation of the model, and errors that appeared in some of the equations. Therefore, they regarded “implementation of Dr. Nicas' ideas as being problematic at this time,” and encouraged the industry to develop improved methods and data for deriving APFs (Ex. 1-182-1).

3. ANOVA Model Fails To Account for Differences Between WPF Studies

Five commenters stated that the proposed analysis fails to account for important differences between studies that could affect WPF values. Thomas Nelson and 3M believed that the ANOVA model does not account for other sources of variability (Exs 1-174, 1-182-5). NIOSH stated that Nicas' report did not address the effect of the test subjects' work rates and other activities on a respirator's performance (Ex. 1-182-3), and did not account for employee training and program surveillance (Ex. 1-182-9). The Chemical Manufacturers Association (CMA) also commented on factors not considered in the Nicas report, “including differences in training, experience, work site, work rate and sample collection” (Ex. 1-182-7). ORC noted: “ The results of a WPF study are based on at least the following components: quality of the respirator chosen; quality of the training program; quality of the fit testing and selection program; nature of the work and ability

to challenge the fit of a respirator (sedentary versus high exercise work)” (Ex. 1-182-10).

4. Using a Conservative Criterion for Setting APFs

Five commenters stated that Nicas' criterion for setting APF values was overly conservative. The Dow Chemical Company (Dow) stated that the Nicas approach “would result in protection factors which are very conservative” (Ex. 1-182-2), while 3M believed that OSHA's use of Nicas's recommendation would result in a major change in the pattern of respirator use (Ex. 1-182-5). NIOSH commented that the approach may result in very low APF estimates because of high WPF variability, and that while the approach would derive more conservative (

i.e.

, more protective) APFs, its use for “WPF studies with small sample sizes * * * could result in APF estimates less than or equal to 1.0 (APF values less than 1.0 are meaningless)” (Ex. 1-182-3). Drs. Rappaport and Kupper stated that only weak precedence existed for Nicas' use of 95th percentiles to define APFs, and suggested that other percentiles (

e.g.

, the 90th percentile) would be more practical to implement (ISEA, Ex. 1-182-1). Finally, CMA believed that the proposed criterion rated “all respirators on the lowest protection achieved by the lowest performing person” (Ex. 1-182-7).

5. APFs Based on a Contaminant's Toxicity (Acute Versus Chronic Toxicants)

Dr. Nicas proposed that two APFs be assigned to a respirator, depending on its use against either a chronic toxicant or an acute toxicant. Four commenters remarked on the feasibility and effects of this approach. NIOSH commented that “defining acceptable protection against short-term exposures is very complex * * *.” (Ex. 1-182-3). 3M commented that dual APFs would be confusing to the user community and workers, and would make program management difficult (Ex. 1-182-5). CMA provided similar comments, and noted that many materials have both chronic and acute effects (Ex. 1-182-7). ORC believed that:

* * * different APFs for different contaminants or types of exposure is not appropriate. Occupational exposure standards should have adequate safety factors which are based on the health outcome (

e.g.

, irritation, systemic toxicity, carcinogenicity) of exposure. (Ex. 1-182-10)

While Drs. Rappaport and Kupper stated that Nicas' argument about respiratory protection for substances with chronic effects was logical, they regarded the question of how to deal with acutely toxic substances as unresolved (Ex. 1-182-1).

6. Distribution of Contaminant Concentrations

Two participants believed that it was necessary to incorporate information on the variability of ambient exposure concentrations, as well as the maximum anticipated concentration, when discussing respirator selection. CMA stated that since an employee's exposures will vary from day to day, employers should select respirators with maximum use limits well above the mean exposure levels to ensure “that there is less than 5% probability of exposures above the maximum use limit of the respirator” (Ex. 1-182-7). In a related comment, ORC stated that many industrial applications typically have exposures only 2-3 times the acceptable exposure limit; therefore, “selecting a respirator with an APF of 10 may mean there is only a remote chance of overexposure to a contaminant due to fit/wear variability” (Ex. 1-182-10).

7. Other Concerns With Nicas' Method

The commenters raised several other issues with Dr. Nicas' methodology. For example, 3M (Ex. 1-182-5) and CMA (Ex. 1-182-7) believed that the relationship between outside concentration and WPF (

i.e.

, WPF increases with increasing Co) was poorly understood; therefore, a sophisticated analysis of the data is questionable. Other commenters noted errors in the equations of the proposed model (

e.g.

, Ex. 1-182-1) and with the distribution of the respirator penetration values (Ex. 1-182-1).

8. Miscellaneous Comments (

e.g.

, ANSI APFs)

In addition to responding to the Nicas report, a number of commenters supported using the APFs recommended in the ANSI Z88.2-1992 respiratory protection standard (Exs. 1-182-1, 1-182-2, 1-182-5, 1-182-7, 1-182-10). These commenters stated that the members of the ANSI Z88.2 committee were “respected industrial hygiene and respirator experts” (Ex. 1-182-5), that the ANSI Z88.2-1992 APFs were “the appropriate values” (Ex. 1-182-7), and that the ANSI APFs “have been through the ANSI peer review process” (Ex. 1-182-5). In advocating use of the ANSI APFs, none of the commenters described the process by which the ANSI Z88.2 committee derived its APFs, or identified the studies and other information on which that committee relied. Furthermore, several commenters (Exs. 1-182-7, 1-182-5, 1-182-10, 1-182-6, 1-182-8) noted that the ANSI Z88.2-1992 standard does not explicitly account for several factors in assigning APF values to different respirator classes, or the use of a respirator in different situations, which they indicated were necessary considerations. Moreover, some commenters (Exs. 1-182-11,1-182-12) recommended APFs that differ from those published by the ANSI Z88.2 Committee. Other commenters believed that it was OSHA's responsibility to show that the commonly used ANSI Z88.2 1992 APFs were erroneous (Ex. 1-182-2), and that the Agency should not use SWPF studies to derive APFs (Ex. 1-182-5). Several participants at the hearing for the final Respiratory Protection Standard stated that OSHA should issue a second NPRM to address the development of APFs (Exs. 1-182-1, 1-182-5, 1-182-10).

After carefully considering Dr. Nicas' model and the comments received in response to his report of the model, the Agency concluded that other possible approaches to deriving APFs should be investigated. Accordingly, the Agency identified and collected available data for this purpose. Of particular interest were data that OSHA could use to discriminate between the performance of different respirator classes. The Agency gathered information from both published and non-published papers and reports, and included WPF, SWPF, PPF, and EPF studies; Health Hazard Evaluations conducted by NIOSH; respirator performance data from manufacturers, such as SWPF data submitted to OSHA by Bullard (Ex. 3-8); and other material related to assessing respirator performance. This information is in Docket H049 as Exhibits 2, 3, and 4.

To assist in evaluating the data, OSHA employed Dr. Kenneth Brown (a statistician) and several respirator authorities: Mr. Harry Ettinger, Dr. Gerry Wood of LANL, and Drs. James Johnson, Kenneth Foote, and Arthur Bierman of LLNL. After the Agency reviewed all of the studies and information, it decided to attempt to analyze only WPF and SWPF studies since they address respirator performance exclusively. OSHA discusses the work and findings of these individuals below.

B. Analyses of WPF Studies

OSHA contracted with Dr. Brown to investigate possible approaches, other than those approaches proposed by Nicas, to evaluate respirator performance data from WPF studies. The following discussion is a general description of the analyses performed by Brown, as well as his overall

conclusions. For a detailed explanation of the methodology and rationale used in the analyses, refer to Brown's reports in the docket (Exs. 5-1, 5-2).

OSHA reviewed the available WPF studies for possible inclusion in Brown's analyses. Early in this review process, the Agency decided to exclude WPF studies with a gas or vapor workplace challenge agent because: The preponderance of studies were conducted in workplaces with particulate challenges; gas/vapor studies did not provide any further insight or clarification regarding sources of variability in WPF studies (most likely, gas/vapor studies add variability to the data such as the effects of humidity on sampling media collection and desorption efficiencies); and pulmonary elimination differs between gases/vapors and particulates. Therefore, OSHA decided to analyze only WPF studies using particulate challenge agents. The Agency evaluated those studies initially selected for further analysis for compliance with the requirements of OSHA's Respiratory Protection Standard (29 CFR 1910.134), as well as completeness of the data. The Agency compiled a list of review items to use in evaluating each study (Ex. 5-5).

OSHA then divided the remaining studies into two categories: Half-mask negative-pressure air-purifying respirators (APRs) and atmosphere-supplying respirators (PAPRs and SARs). This procedure resulted in 22 APR studies and 16 PAPR/SAR studies for analysis. OSHA placed a list of these studies, and their respective respirators, in the docket (Ex. 7-4). Brown subsequently identified 14 APR studies and 13 PAPR/SAR studies for further analysis (see Exs. 5-1 and 5-2 for more information on the evaluation criteria).

Brown's analyses divided the respirators used in these studies into separate respirator classes. The analyses divided APRs into 5 classes, listed below in Table 1. As this table shows, Brown's analyses separated filtering facepieces into four classes based on the characteristics listed under the Description column heading, with the fifth class comprised of elastomeric facepiece APRs.

Table 1.—Half-Mask APR Classes

Class

Type

Description

Adjustable head straps

Exhalation valve

Double shell construction

Foam ring liner

1

Filtering facepiece

2

Filtering facepiece

X

X

3

Filtering facepiece

X

X

X

4

Filtering facepiece

X

X

X

X

5

Elastomeric facepiece

In addition, Brown's analyses divided PAPRs into five classes and SARs into two classes, as shown in Table 2.

TABLE 2.—PAPR and SAR Classes

Class

Type

Description

1

PAPR

Loose-fitting facepiece.

2

PAPR

Loose-fitting facepiece with hood and/or helmet.

3

PAPR

Hood and/or helmets—not loose-fitting.

4

PAPR

Tight-fitting half-mask facepiece.

5

PAPR

Tight-fitting full facepiece.

6

SAR

Loose-fitting.

7

SAR

Hood or helmet.

Later in the analyses, Brown further divided these classes according to class of respirator, study, and challenge agent (CLSA). This division resulted in 26 CLSAs for the APRs and 14 CLSAs for the PAPRs/SARs.

The data from the WPF studies consisted of simultaneous measurements of the challenge agent concentration inside the respirator facepiece (

i.e.

, concentration inside or Ci) and outside the respirator facepiece (

i.e.

, concentration outside or Co) in the ambient workplace atmosphere. Corresponding Co and Ci measurements can be used to calculate the workplace protection factor (WPF = Co/Ci) or penetration of the contaminant into the respirator (PEN = Ci/Co = 1/WPF). The APR studies had a total of 917 data pairs, while the PAPR/SAR studies provided 443 data pairs.

1. Half-Mask APRs

In the first phase of his analysis, Brown statistically analyzed the data for half-mask negative pressure APRs, both filtering facepiece and elastomeric APRs, using the following three approaches: (1) Pooled the data within classes, corrected the data for the positive relationship found between WPF values and increasing Co, and compared the differences in WPF statistics between classes; (2) conducted an intra-study analysis of the performance of two different classes of respirator used against the same contaminant under similar workplace conditions; and (3) divided the data into class-study-agent combinations, and evaluated WPF as a function of Co. The following sections discuss these approaches in detail.

Approach 1.

Brown's initial approach was to determine if he could pool the data within each respirator class and estimate the fifth percentile WPF for that respirator class; he then tested for differences in WPFs between the respirator classes. He divided and analyzed the data by study, treating the data from each study as a homogeneous sample arising from the same parent distribution. Then he examined the data in each study for a Co effect, and constructed a scatterplot of ln(WPF) versus ln(Co) for each respirator class. In doing so, he treated extreme or poorly fitting data as outliers and removed them from the analysis. He subsequently derived a linear regression of ln(WPF) on ln(Co) for each study, and extrapolated from the observed range to the entire range of Co values in all of the data. The positive slopes, which he found for most classes, showed that ln(WPF) increased as ln(Co) increased. In addition, the regression lines were well mixed, indicating that studies within the same respirator class varied more than anticipated. This result indicated that variability occurring within respirator classes could obscure differences between respirator classes.

These studies collected data over different ranges of Co. Therefore, to compare the WPFs observed in the studies, Brown corrected the WPF values for all studies, using a common Co adjustment factor. He pooled the adjusted WPFs by class, and then plotted the cumulative distributions to determine if he could identify differences between respirator classes, despite intra- and inter-study differences. Finding no differences

between respirator classes using the Co adjustment factor, he concluded that:

Observed 5th percentiles for WPFs, and their lower confidence intervals when adjusted for the Co effect, showed no clear evidence that any class was preferable to another. In particular, there was no indication that Class 5 (elastomerics) performed better than four disposable classes. (Ex. 5-1, p. 8)

The results of these analyses prompted a more detailed examination of the data. To control for study-related and agent-related factors that may contribute to variability, Brown performed an intra-study analysis on two different respirator classes used against the same workplace challenge agent under similar workplace conditions (Approach 2).

Approach 2.

The second approach attempted to determine respirator performance after controlling for study-to-study and agent-to-agent sources of variability. Among the half-mask APRs, the chance of detecting performance differences appeared to be greatest for comparisons between elastomeric and filtering facepiece respirators. In implementing this approach, Brown assumed that controlling for study and agent sources of variability would result in WPF differences attributable, in large part, to variability in respirator performance.

Four of the studies compared the performance of elastomeric and filtering facepiece respirators against the same challenge agent in the same workplace. After reviewing these studies, a study by Meyers and Zhuang (Ex. 1-64-51) was selected for further analysis because it was recent, followed a protocol patterned after other published WPF study protocols, and was well documented. Brown's statistical analyses of this study (see Ex. 5-1, Appendix C) indicated large sources of variability within the study, making comparison of the two respirator classes difficult and tenuous. Based on plots of the data and the occurrence of several outliers, it appeared that even data on the same agent, obtained under similar workplace conditions, may not have come from the same parent distribution. In addition, the variability of WPFs within the study (regardless of adjustment for the Co effect) was large. Therefore, the results of this second approach led Brown to state that, at least in this analysis, “workplace studies may have too much intra-study variability for reasonably valid/accurate/reliable assessments and comparisons of respirator effectiveness.” (Ex. 5-1, p. C-17)

Approach 3.

Brown began the third statistical approach by dividing the data into units smaller than respirator class,

i.e.

, units based on class of respirator, study, and workplace challenge agent (class-study-agent or CLSA). This procedure resulted in 42 CLSA combinations. After removing deficient data (

e.g.

, no data on Co), he narrowed the data set to 26 combinations. Again, he tested the data for each CLSA to determine if WPF increases with Co and, if so, whether the effect held for all respirator classes. Data analyses of the 26 CLSAs indicated that WPF increased with Co; Brown then derived a common estimate (across all CLSAs) of the Co effect. He subsequently estimated the means for the CLSAs within each class of respirator, both with and without adjustment for Co effect. Brown compared the means of these CLSAs within and between respirator classes. For each respirator class, he grouped the CLSAs that had no significant difference between their means into common subclasses, and plotted both the adjusted and non-adjusted means [

i.e.

, mean of ln(PEN)] of the subclasses, as well as their associated confidence intervals. The results of the comparisons showed that: the estimated means of CLSAs vary so much within a class that the mean of one CLSA is likely to be a poor predictor of the mean of another CLSA within the same class; and it was not visually apparent from the plots that one class of respirator performed better than another class. In general, the comparison indicated that study outcomes, even within the same class of respirator, are highly heterogeneous.

Final analysis.

Since the three approaches discussed above could not distinguish between respirator effectiveness within or across classes, the data were viewed, as a whole, from the relationship of Ci and Co. Brown pooled the data for all 26 CLSAs and derived several functional relationships from the pooled data. This approach showed that the majority of the observed data pairs achieved a WPF of 10. (See Ex. 5-1 for more details.)

After performing the above analyses, Brown made a number of observations and conclusions. He noted that the range of WPF values within a CLSA was typically wide, and that the observations were highly variable. In addition, he believed that variability in WPF studies can affect the accuracy, validity, and reliability of study results, as well as the ability to compare study results. Brown noted several possible sources of variability in WPF studies, including: (1) Study characteristics related to study design, execution, sample analysis, and data management and reporting; (2) measurements of Ci at different outside concentrations (Co effect), taken in conjunction with other poorly described factors (

e.g.

, particle size, temperature, humidity) that may affect the relationship of Ci and Co; (3) characteristics of the ambient agent itself (

e.g.

, possible effects of the agent occurring in a mixture with other agents); and (4) variations in data among studies related to using different study procedures (

e.g.

, repeated measurements on the same worker in some studies versus single measurements on each worker in other studies, random versus non-random selection of study participants). He also commented that the analyses assumed that the data were representative of workplace conditions; however, the data may not represent either current or future workplaces in which employees use respirators. Finally, Brown observed that studies with high Ci values, relative to Co, may have influenced his findings. He believed that these studies should be closely reviewed because some study weakness, unrelated to respirator performance, could be the reason for the high Ci values.

Brown also made some general observations about WPF studies. First, he believed that the role of WPF studies in assessing and comparing respirator effectiveness, and influencing APFs, should be reevaluated. He believed that a more refined instrument that is amenable to experimental design and control, such as chamber studies, is better suited for providing information during determination of assigned protection factors. Brown noted that the use of high concentrations of a challenge agent in chamber studies may minimize the uncertainty of extrapolating test results obtained at low outside concentrations to levels well above the observed range. Therefore, WPF studies would serve as a counterpart to chamber studies,

i.e.

, WPF studies would provide data on the respirator during actual use in the workplace, and identify workplace conditions in which a respirator may perform poorly. To improve comparability of results, he advocated using uniform procedures to: select the challenge agent; collect samples; record the data; and measure and interpret Ci and Co (Ex. 5-1, pp. 42-44).

Overall, the analyses led Brown to several conclusions. First, workplace studies have limitations for comparing respirator performance because of uncontrolled sources of variability. Support for this conclusion comes from the wide confidence intervals for the means of the CLSAs, and the wide range of those confidence intervals within the same respirator class. Second, Brown believed that the WPF has limits as a

measure of respirator effectiveness because, in general, it tends to increase as Co increases. This relationship complicates comparisons of WPF values measured at different Co levels. Third, he found no clear evidence that one class of respirator is better than any other class, particularly between elastomeric half-mask and filtering facepiece respirators. In addition, the differing results between CLSAs within the same class of respirators indicated that the outcome of one CLSA may be a poor predictor for another CLSA in the same class.

2. PAPRs and SARs

Dr. Brown analyzed 13 studies to evaluate and compare the effectiveness of PAPRs and SARs. Ten of the studies were conducted with PAPRs, and three with SARs. Brown's analyses divided these “high-performance” respirators into seven classes (

i.e.

, five types of PAPR and two types of SAR) based on their design features (

see

Table 2), with subsequent separation of these respirator classes into 14 CLSAs.

Brown used the CLSAs to determine whether any differences in respirator effectiveness existed among the respirator classes. He analyzed the data for trends of WPFs, either upward or downward, as Co increases, and for homogeneity. Brown plotted all of the data, fitted lines to these plots, made comparisons of study results within each respirator class, and developed functions from the fitted lines. (For additional details on these statistical analyses and the data plots,

see

Ex. 5-2.)

On reviewing the data plots, Brown concluded that the data were consistent with a linear relationship between ln(Ci) and ln(Co). Also, the presence of outliers and/or an imbalanced distribution of the observations influenced the results. He recommended further investigation of the outliers, particularly those with unusually high Ci values, to determine if they resulted from characteristics of the respirator or other variables. He also recommended studying the imbalanced distributions to determine if they represented individual study biases caused, for example, by collecting data at different work sites or on different work shifts. Finally, Brown noted that the robust least trimmed squares line may be useful for estimating the relationship between ln(Ci) and ln(Co).

Fifth percentiles are commonly used as a benchmark for respirator performance. Brown's analyses showed that fifth percentile estimates differed considerably within respirator classes that contained more than one CLSA. The range of the fifth percentile estimates was 28-389 for the five CLSAs in Class 2, 17-107 for the two CLSAs in Class 4, 29-1779 for two CLSAs in Class 5, and 74-188 for the two CLSAs in Class 7. The fifth percentile estimates in Classes 3 and 6 were large, while the fifth percentile estimates were small in Classes 1, 4, and 7. Brown believed that, while some of these differences may be attributed to a real difference in respirator performance between classes, the sample sizes were too small and/or the sampling variability too large to obtain reliable estimates at low percentile levels. He noted that the fifth percentile estimates were variable, and were not predictable from one CLSA to another CLSA within the same respirator class. Thus, he concluded that the fifth percentile estimates of WPFs have limited utility for setting assigned protection factors. Table 3 lists the descriptive statistics for WPFs, for each class-study-agent combination.

Table 3.—Descriptive Statistics for WPF, by Class, Study Agent

CL1.26.Cd

CL2.22.Pb

CL2.23.Pb

CL2.24.Si

CL2.3.BAP

CL2.5.Asb

CL3.27.EBZ

Curve Label

1

2a

2b

2c

2d

No curves

3

Median

2,972.97

127.88

155.29

3,553.72

1,788.32

156.00

11,935.87

Range

25,186.05

1,040.75

6,131.76

95,518.07

8,203.89

537.00

4,746,673.83

Minimum

53.70

22.58

28.24

36.31

371.49

66.00

1,152.26

Maximum

25,239.75

1,063.33

6,160.00

95,554.38

8,575.38

603.00

4,747,826.09

No. Observations (N)

33

46

43

59

20

7

58

5th Percentile

280.25

27.82

35.03

92.07

388.70

70.50

1,797.79

10th Percentile

581.87

53.04

43.08

267.60

407.51

75.00

2,365.29

Reject Lognormality?

No

No

No

No

No

No

Yes

Geometric Mean

2,523.49

126.85

184.69

2,765.75

1,408.10

151.95

15,623.81

Geometric Stan. Dev

3.56

2.28

3.21

6.33

2.50

2.54

5.56

CL4.21.Si

CL4.6.Pb

CL5.18.Pb

CL5.21.Si

CL6.19.Si

CL7.25.Sr

CL7.28.Si

Curve Label

4a

4b

5

No curves

6

7a

7b

Median

48.67

438.60

7,948.14

85.44

9,178.81

3,827.16

2,480.55

Range

176.27

2,310.33

73,081.90

189.92

34,735.48

87,137.82

33,384.67

Minimum

16.40

23.00

579.04

24.75

668.34

41.67

43.33

Maximum

192.67

2,333.33

73,660.94

214.67

35,403.82

87,179.49

33,428.00

No. Observations (N)

7

25

53

4

15

21

52

5th Percentile

17.20

107.06

1,779.12

29.10

1,407.60

74.07

188.14

10th Percentile

18.00

160.95

2,300.18

33.50

2,229.66

79.37

383.47

Reject Lognormality?

No

No

No

N too small

No

No

No

Geometric Mean

49.20

400.34

8,319.09

76.10

7,389.62

2,315.04

2,066.00

Geometric Stan. Dev

23.60

2.81

3.03

25.60

2.92

9.99

4.02

The objective of the review of these 13 WPF studies was to see what can be learned about the performance of each respirator class, and its relative effectiveness, based on the data for Co and Ci. He also attempted to determine how Ci changes as Co changes, and what factors affected this relationship.

Brown found too much unexplained variability between study outcomes, even within the same respirator class and within similar ranges of Co, to make valid and reliable comparisons. He noted that study outcomes for the same class of respirator may differ significantly, which raised concerns about interpreting the outcome for a class from a single study. More specifically, he questioned whether the results from one study would be similar

to another study. He concluded that it is not possible to know to what extent the outcome of a study is attributable to characteristics of the respirator used.

Brown believed that the variability identified in this analysis was probably due to uncontrolled parameters in the workplace test situations, such as aerosol particle size distributions and densities, and work activities. Based on the data from these studies, he found that WPF tends to increase as Co increases (equivalently, penetration, or PEN., tends to decrease). He believed that the probability of a Co dependence for WPFs seemed to be established by his analyses.

C. Analyses of SWPF Studies

1. Bullard Models 77 and 88, Clemco Apollo Models 20 and 60, and 3M Whitecap II

In the mid-1980s, SWPF studies provided OSHA with information on the effects of temperature, relative humidity, airflow, and facial hair on respirator performance (LANL, 1988; Ex. 1-64-101, LLNL, 1986; Ex. 1-64-94). More recent SWPF studies provided additional information on the performance of the following abrasive blasting respirators: the Bullard Models 77 and 88 (Ex. 3-8-3), the Clemco Apollo Models 20 and 60 (Ex. 3-7-3), and the 3M Whitecap II (Ex. 3-9-2).

OSHA contracted with Mr. Harry Ettinger to review and comment on the study principles and protocols described in the five reports (Bullard, Clemco, 3M Whitecap, the LLNL study, and the LANL study). His report (Ex. 3-3) contained the following observations and conclusions.

Mr. Ettinger noted that while the reports do not satisfy the typical criteria for defining peer-reviewed publications, this was not a serious problem because the studies were conducted in national laboratories by knowledgeable and experienced investigators. Furthermore, the review procedures generally used by these national laboratories most likely provide a sufficient peer-review process. He noted that none of the reports provided sufficient detail to permit a statistical re-analysis of the data by OSHA. In addition, he observed that the studies of the Bullard, Clemco, and 3M respirators reported considerably higher fit factors than the 1986 and 1988 national laboratory studies. However, he believed that it was not appropriate to compare the results of recent studies with the older studies, but he noted that older respirators may not perform as well as newer designs.

Mr. Ettinger also noted that the tests of the Bullard, Clemco, and 3M respirators satisfied the established criteria of fit factors that exhibited only brief negative pressure spikes. He believed these results indicated that if these devices are used and maintained properly, they appear to have fit factors of at least 20,000. He believed that, using a safety factor of 20, a protection factor of 1,000 is attainable, assuming that the testing protocol is adequate.

Ettinger stated that he could not define clearly a relationship between the older and more recent study results. For example, he suggested that the additional exercises in the more recent study (ORC, 2001; Ex. 3-4-2) did not adequately represent normal or extreme work situations. Ettinger cautioned against assuming that all blasting helmets would achieve the high fit factors measured in the recent studies because performance is device specific, and indicated that older respirator designs may need to be reevaluated. Furthermore, he believed that quality control, human factors, minimum flow rate, and the sturdiness of respirator construction are important variables that should be evaluated in the testing protocol.

2. NIOSH N95 Study

In 1999, NIOSH conducted a chamber study of 21 N95 respirators (20 filtering facepiece, and 1 elastomeric, respirators) and statistically analyzed the respirators' performance (Ex. 4-14). At the request of OSHA, Drs. Johnson, Foote, and Bierman of LLNL undertook a review of this study to assist the Agency in evaluating APFs of half-mask respirators (Ex. 3-2). OSHA provided the raw data files from the study to LLNL for independent evaluation.

The NIOSH investigators used ambient (

i.e.

, room) aerosol as the challenge agent, and a PortaCount to measure respirator penetration. Use of ambient aerosol does not require aerosol generation equipment, thereby circumventing use of a possibly hazardous chemical. However, if this technique generates a low ambient particle concentration it is difficult to detect the reduced number of particles that penetrate the respirator; this effect results in an artificially low protection factor. In addition, an ambient aerosol that is varying in concentration during testing can cause error in the penetration measurements. Study participants can also produce aerosols ranging from 0.1 to 3 particles/cc through their breathing (

i.e.

, “breathing” background). Whenever the amount of challenge agent that penetrates the respirator is low (

i.e.

, on the order of particles/cc or less), the PortaCount cannot distinguish between particles in the breathing background and the challenge aerosol penetrating the respirator. The LLNL researchers believed that the breathing background can limit fit factor measurements to 1,000 and less when the challenge concentration is below 2,000 particles/cc (Ex. 4-15). They concluded that challenge aerosol concentrations can be better controlled in chamber studies than under this protocol.

When calculating faceseal leakage, the NIOSH authors assumed that all study participants have the same constant volumetric flow rate through the respirator. Using a filtration model developed by Rubow (Ex. 3-7-3), the LLNL reviewers determined media penetration that was approximately 5% less than the media penetration calculated by the NIOSH authors using the constant flow rate assumption. Since the method used by the NIOSH authors results in only a 5% error, and gives a conservative estimate of the filter penetration, the LLNL reviewers believed that the constant flow rate assumption is reasonable. The LLNL reviewers also discussed other considerations, including fluctuations in peak flows under various exercise conditions, and the correction factor for filter media penetration used by the NIOSH authors.

Investigating the possible effect of breathing background on the PortaCount fit factor measurement, the LLNL reviewers applied an estimated worst-case scenario to the data. The scenario consisted of the following two assumptions: (1) A challenge aerosol concentration of 3,000 particles/cc, and (2) a breathing background of 5 particles/cc. Applying these assumptions to the NIOSH data, the LLNL reviewers recalculated total penetrations, and adjusted the results for breathing background. They found that, when compared to the NIOSH results, 14 of the 21 respirators had more tests passing the 0.01 penetration criteria than before. The LLNL reviewers also calculated the 50th and 95th percentiles for the penetration data, both with and without applying the breathing background assumption. In view of their results, they believed that the original NIOSH analysis and findings result in a conservative estimate of the respirators' performance.

The LLNL reviewers also used the NIOSH raw data to reproduce values, geometric standard deviations, and the 95th percentile for total penetration, filter penetration, and face seal leakage. They then compared these results to total penetration and face seal leakage penetrations summarized in the NIOSH study (Exs. 4-1, Table 2; 4-14, Table I).

The few discrepancies were small, and could be attributed, for example, to rounding off values. The 95th percentiles in the NIOSH study were based on a formula using the geometric mean and geometric standard deviation, and assumed that the distribution was log normal. For comparison, the reviewers calculated the 50th and 95th percentiles based on the raw data alone (

i.e.

, assuming no distribution). Using this approach, the LLNL reviewers noted that, for many respirator models, the 50th percentile differed markedly from the geometric mean. They also saw differences between the 95th percentile calculated using a log normal distribution and the corresponding percentile determined directly from the data. LLNL reviewers stated that the NIOSH study demonstrated the advantages of SWPF studies for half-mask respirators. Their results confirm the quality of this important SWPF study of filtering facepiece and elastomeric half-mask respirators.

3. ORC Study of PAPRs and SARs

In 1997, ORC and a group of its member companies sponsored a study of 11 powered air-purifying and supplied-air respirators (PAPRs and SARs) to evaluate the protection that these respirators afforded to workers in the pharmaceutical industry. The study, “Simulated Workplace Protection Factor Study of Powered Air Purifying and Supplied Air Respirators' (Ex. 3-4-1) was completed in 1998 by researchers at LLNL. OSHA requested Dr. Gerry Wood of LANL to evaluate ORC's LLNL study. He evaluated the study using the data received from ORC, as well as information on the study published in the American Industrial Hygiene Association Journal (Exs. 3-1, 3-4-2).

The raw data files from the study consisted of instantaneous (0.1 second) photometer aerosol measurements obtained before, during, and after 12 exercise periods (including four periods of normal breathing) performed by each study participant. The instantaneous penetration results for the 144 tests were plotted against time. Wood examined patterns of aerosol penetration into the respirator that occurred throughout testing, noting that certain exercises often exhibited penetration spikes. He found that running in place produced the most penetration spikes. However, he also noted other respirator/subject combinations result in spikes. Wood indicated that such non-random distributions of readings was not surprising, as different movements during an exercise should affect instantaneous penetrations differently.

Wood calculated 95% confidence limits for the average and maximum penetration values during each exercise. In doing so, he assumed that pre-test and post-test background, and chamber aerosol measurements were distributed normally, since no movement variables were present. He then calculated aerosol penetration. Wood found that the photometer reading averages and standard deviations that he analyzed for all 144 data sets were in agreement with the LLNL figures, and that rounding off figures accounted for any minor differences in average penetrations that he calculated.

In summary, Dr. Wood believed that the quality of the data, experimental protocol, measurements and data, and calculations applied to the data in the ORC-LLNL study were excellent. He agreed with the authors' conclusions that SWPF studies are useful for comparing respirators, and that the study protocol was reproducible.

D. OSHA's Overall Summary Conclusions

Prior to this current rulemaking, OSHA explored several procedures to evaluate and compare respirator performance across models, studies, agents, and testing protocols. The Agency thoroughly reviewed the available data on respirator performance to determine the current concepts, and possible methodologies, for deriving APFs. To evaluate the data, OSHA had to make several decisions.

For example, while OSHA was aware that particle size can affect concentration values, the Agency was unable to quantify this factor based on available information. Consequently, OSHA did not attempt to adjust for differences in particle size in the analyses. Furthermore, the Agency had to decide how to address sampling results that were below the limit of detection (LOD). Accordingly, whenever sampling results were below the limit of detection, OSHA set the Ci at a percentage of the LOD reported in the study. When the study reported extremely low Ci results as a percentage of the LOD, the Agency used the values provided by the authors.

OSHA was concerned that the analyses be those best able to account for parameter uncertainty, and be a measure of respirator effectiveness that is valid over a plausible range of concentrations for each of the agents against which the respirator is to be used. As discussed above, the Agency contracted with Drs. Nicas and Brown to independently evaluate the raw WPF data. As a result of these analyses, OSHA preliminarily agrees with Drs. Rappaport and Kupper, who indicated that, while some modeling may be useful, concerns remain regarding the lack of model validation (Ex. 1-182-1). Furthermore, OSHA finds merit in Thomas Nelson's comment that a simple analysis of the entire data may sufficiently cover the relevant sources of variation in these data (Ex. 1-174). Databases of the information used by the Agency in its analyses have been placed in the docket for review by interested parties (Exs. 5-3, 5-4, 5-5).

The Agency also recognizes that WPF and SWPF studies have their strengths and weaknesses. SWPF studies can control for a number of variables, thus providing less variable results across respirators classes than WPF studies. Also, SWPF studies can test respirators safely at the limits of their effectiveness. However, WPF studies evaluate respirators during use in the workplace. Therefore, the Agency believes that WPF or SWPF studies provide complementary information.

OSHA developed the proposed APFs using a multi-faceted approach. The Agency reviewed the various analyses of respirator authorities, available WPF and SWPF studies, and other APF literature. For example, OSHA reviewed Brown's analyses and noted no difference in performance between filtering facepiece and elastomeric half-mask APRs, and that few data pairs from the combined data sets analysis failed to achieve a WPF of 10. In addition, the data from WPF and SWPF studies, as well as a qualitative review of the available APF literature, supported an APF of 10 for all half-mask APRs. Therefore, OSHA is proposing an APF of 10 for half-mask APRs. The Agency used a similar approach in developing the remaining proposed APFs.

In conclusion, the APFs proposed by OSHA in this rulemaking represent the Agency's evaluation of all the available data and research literature;

i.e.

, a composite evaluation of all the relevant quantitative and qualitative information. The Agency seeks comment on this approach, as well as the proposed APFs developed using this approach.

E. Summaries of Studies

Researchers often determine the protection afforded by a respirator by conducting Workplace Protection Factor (WPF) studies and Simulated Workplace Protection Factor (SWPF) studies. A WPF study measures the effectiveness of respirators under workplace conditions. Workers participating in a WPF study wear respirators while performing their usual job tasks. The WPF is a measure of the reduction in exposure achieved while using respiratory protection and

is the ratio of the concentration of the contaminant found in the workplace air to the concentration found inside the respirator facepiece. Similarly, a SWPF study measures the ratio of a contaminant's concentration both outside and inside the facepiece. However, researchers obtain these measurements in test chambers, which allows them to control some important variables (

e.g.

, outside concentration of the challenge agent). Rather than performing the actual job tasks found in a particular work setting, the study participants perform a series of exercises in the test chamber that simulate the actions of workers in general.

In developing the proposed APFs listed in Table 1 of the proposed amendments to the standards (Section XII). OSHA reviewed data from properly conducted WPF studies and SWPF studies. In addition, the Agency reviewed published APF tables. These data formed the basis for OSHA's proposed APFs. OSHA also reviewed other types of studies, such as Effective Protection Factors (EPF) and Program Protection Factor (PPF) studies, along with respirator performance studies that lacked raw data. A review of those studies can be found in the Docket (Exs. 3-10, 3-11). However, EPF and PPF studies account for aspects of respirator use other than effectiveness of the respirator while it is being worn, while studies that lack raw data give little information for in-depth statistical analysis. Therefore, OSHA relied on WPF and SWPF studies, since they attempt to account for actual use conditions and focus on the performance characteristics of the respirator only.

1. WPF Studies—Filtering Facepiece and Elastomeric Half-Mask Respirators

Study 1B.

C.E. Coulton, H.E. Mullins, and J.O. Bidwell gave a presentation at the May 1994 American Industrial Hygiene Conference and Exposition (AIHCE) on worker protection afforded by the same respirator in two different environments and against two different contaminants (Ex. 1-64-13). At the first site, the authors determined exposure to cadmium dust for 18 workers in a plastic colorant manufacturing facility. They determined exposure to lead fume for 18 workers during ship breaking and recycling at the second site. At the colorant facility, cadmium-containing pigments were weighed, mixed with plastic resin, and fed into extruders for production of concentrated colorant. Samples were obtained from workers in the weighing, mixing, and extruding areas. Workers at the ship breaking facility used torches to cut an aircraft carrier into large sections that were then cut into smaller pieces on shore. Burners and firemen, on the ship and on shore, were sampled for lead. Work rate at the colorant facility was judged to be low, while the work rate of the ship breaking workers was assessed as being moderate. The respirator used in the study was a 3M 6000 series elastomeric half-mask equipped with either 3M 2040 or 3M 2047 HEPA filters (the 2047 HEPA filter has some activated charcoal for removal of nuisance levels of organic vapors). Employees normally wore the study respirator and were provided with training in its proper donning, fitting, and operation. In addition, the employees had to pass a saccharin qualitative fit test prior to study participation; they also had to be clean-shaven. The study was explained to the participants and they were observed on a one-on-one basis throughout the sampling periods.

The inside-the-facepiece sampling train consisted of a 25 mm three-piece cassette with a 0.8 micron pore size mixed cellulose ester filter. Respirators were probed with a Liu probe inserted opposite the mouth and projecting one cm into the facepiece. The sampling cassette was attached directly to the probe, and a cassette heater was utilized to prevent condensation of moisture from exhaled breath. Outside-the-facepiece samples used a 25 mm three-piece cassette with a 0.8 micron pore size mixed cellulose ester filter. The outside sample cassette was also connected to a Liu probe, and this combination was attached in the worker's breathing zone. Inside samples and outside samples were collected at a flow rate of 2 Lpm. Respirators were donned and doffed, and sampling trains started and stopped, in a clean area. Field blanks were used for contamination evaluation. Particle size distribution was ascertained with a six-stage single-jet cascade impactor that sampled all day at 1 Lpm.

Samples were analyzed by inductively coupled plasma (ICP) spectroscopy. For both cadmium and lead, the authors presented the range of outside concentrations, inside concentrations, and the associated geometric means and standard deviations. Three sets of WPFs were determined for cadmium and lead, based on three different methods for reporting inside samples that were below the limit of detection (LOD) (

i.e.

, calculating WPF using 70% of the LOD; calculating WPF using the LOD; or eliminating these samples from the WPF calculation database). No field blank adjustments were made (

i.e.

, no cadmium or lead detected), and no mention is made of adjusting the data for pulmonary retention of particles. In addition, samples were invalidated as a result of equipment and procedural problems, and if the outside filter weights were less than 100 times the limit of detection (or 101 times the field blank value). The authors reported a mean WPF of 353, with a fifth percentile of 34, for the cadmium samples, and a mean WPF of 135, with a fifth percentile of 15, for the lead fume samples. The authors noted a sizable difference in WPFs for cadmium and lead (using the same respirator), and discussed a number of possible reasons for the difference (

e.g.

, differences in particle size, work environment, work rate). The authors concluded that the ANSI Z88.2-1992 recommended APF of 10 for half-facepieces was appropriate.

Study 1C.

In a poster presentation at the 1992 AIHCE, C.E. Coulton and H.E. Mullins provided results of a study of several contaminants (Ex. 1-146). Exposure to iron (Fe), manganese (Mn), titanium(Ti), and zinc (Zn) were determined for shipyard workers involved with welding and grinding. The respirators studied were 3M 9920 and 3M 9925 dust/fume/mist disposable respirators.

At the Agency's request, 3M provided the raw data from the study, but the information provided had no discussion of sampling or analytical methodologies. However, in a brief abstract, the authors mention using blank samples and observing participants during sampling (in the context of discarding particular sample sets). Outside- and inside-the-facepiece concentrations, and associated WPFs, were provided for the four analytes: Fe (31 data sets), Mn (32 data sets), Ti (28 data sets), and Zn (32 data sets). Calculated WPFs ranged as follows: 24 to 1010 for Fe, 10.21 to 715 for Mn, 50.38 to 2545 for Ti, and 27.41 to 854.89 for Zn. Tom Nelson (Ex. 135) calculated a geometric mean (GM) of 147, a geometric standard deviation (GSD) of 2.5, and a best estimate fifth percentile of 33 for the 32 sample sets he used in evaluating this study. The information he provided contained no additional discussion of the results or study conclusions.

Study 1D.

Workplace performance of an elastomeric half-mask against exposure to lead was reported in 1984 by S.W. Dixon and T.J. Nelson for 11 workers in an unidentified work environment (Ex. 1-64-19). The participants' work rate was judged to be moderate to heavy. Workers viewed a training program and selected from three mask sizes of a Survivair 2000 elastomeric half-mask respirator,

equipped with organic vapor/high-efficiency particulate filters. Participants were qualitatively fit tested with isoamyl acetate. Prior to participation, employees were quantitatively fit tested with a Dynatec/Frontier FE250A portable unit while wearing the Survivair with high-efficiency filters and performing six ANSI-recommended exercises. In addition, paired (before and after) quantitative fit tests were performed for about half of the WPF determinations to ascertain if quantitative fit tests can predict WPFs. Participants were instructed not to break the faceseal during sampling, and were observed throughout the sampling period.

Samples were collected on 25 mm 0.8 micron pore size polycarbonate filters, for 30 to 120 minutes (a complete job cycle) at a flow rate of 2 Lpm. Sampling trains were calibrated before and after each day's sampling, and respirators were disassembled, cleaned, and reassembled at the end of each day. The authors do not provide a more detailed discussion of the inside or outside sampling trains (

e.g.

, type of respirator probe, placement of outside sampling apparatus). Particle size analysis was performed using light microscopy and scanning electron microscopy.

Proton induced x-ray emission analysis (PIXEA) was used to analyze the samples. This method's limit of detection was 2 nanograms per sample. The authors provide an approximate particle aerodynamic diameter based on the particle size analyses. Inside-the-facepiece results were corrected for losses caused by the sample probe but were not corrected for lung deposition (which the authors believed caused only a small bias). Thirty-seven WPFs were determined; however, the individual data sets (

i.e.

, inside concentration, outside concentration, and associated WPF) were not provided. During the study, some participants were observed to break the faceseal to talk. The authors provide an overall range of WPFs achieved, GM, and GSD, for undisturbed facepiece samples and pooled disturbed and undisturbed facepiece samples. The authors reported a GM WPF of 3,400, and a best estimate of the fifth percentile of 390 when the facepiece was not disturbed, and a GM WPF of 2,400, and a best estimate of the fifth percentile of 160 when the facepiece was disturbed. The authors also found no correlation (at the 5% level) between WPF and outside concentration, or the relationship between WPF and quantitative fit factors for predicting workplace protection. The authors also estimated the program protection factor based on historical measures of air lead concentrations versus blood lead levels (a table and graph of this data was provided). They concluded that the half-mask respirator they tested provided WPFs that exceeded an APF of 10, and provided program protection factors (PPFs) that exceeded 10.

Study 2.

Workplace protection against exposure to asbestos fibers (chrysotile and amosite) was reported at the 1985 AIHCE by T.J. Nelson and S.W. Dixon for 17 workers who removed asbestos-containing materials at two sites (Ex. 1-64-54). Six of these workers were removing asbestos fireproofing from a ceiling at the first site, while eleven workers at the second site were removing asbestos-containing pipe insulation. The participants' work rate was judged to be moderate, site temperatures ranged from 65-85 degrees Fahrenheit, and humidity was very high.

The following six brands of half-mask respirators were studied: 3M 8710 disposable dust/mist respirator; 3M 9910 disposable dust/mist respirator; American Optical R1050 disposable dust/mist respirator; Survivair 2000 elastomeric respirator with high-efficiency filters or DFM filters; MSA Comfo II elastomeric respirator with high-efficiency filters or DFM filters; and a North 7000 elastomeric respirator with high-efficiency filters. Participants were trained in respirator use by the investigators and were qualitatively fit tested using the saccharin fit test. Supplemental data indicate that participants wore one or more respirator brands. No mention is made of respirator donning and doffing procedures, or starting sampling trains in a clean area; however, the sampling procedures state pumps were stopped and cassettes removed in a dust-free area. Participants were observed by the researchers throughout the sampling period.

The inside-the-facepiece sampling train was a 25 mm closed-face three-piece cassette with a

1/2

-inch extender, containing a 0.8 micron pore size mixed cellulose ester filter. The cassette was attached directly to a tapered probe inserted into the respirator midway between the nose and mouth. In-mask samples were collected at a flow rate of 2.0 Lpm. The outside-the-facepiece sampling cassettes and probes were identical to the inside-the-facepiece sampling train and were fastened to the lapel of the subject. Outside samples were gathered at 0.5 to 1.0 Lpm. Sampling times ranged from 30 to 120 minutes, and the pumps were calibrated before and after each sampling period. The authors investigated uniform deposition of asbestos fibers across the filters; they noticed a slight trend for heavier deposition at the filter center using both methods. They also computed the precision of sample gathering using open- versus closed-face cassettes and found no difference between the methods.

Asbestos analysis was based on NIOSH method P&CAM 239 and NIOSH method 7400 (

i.e.

, the filter mounting and “A” counting rules). To increase analytical sensitivity, the methodology was modified by counting fibers in a minimum of 500 fields per inside-the-facepiece filter when less than 100 fibers were counted. The actual number of fibers counted in each sample was used to compute the airborne concentration. In addition, one microscopist performed all fiber counting. The distributions of fiber length and diameter were determined by transmission electron microscopy using lapel sample filters. The GM and GSD values for the fiber length, fiber diameter, and equivalent aerodynamic diameter at each worksite and the combined data from both sites were reported, but the values for fiber density and the length-diameter correlation coefficient were not provided. A total of 84 pairs of inside and outside fiber concentrations, and corresponding WPFs, were provided by participant, respirator brand, and sampling period in supplemental data tables. However, the authors considered seven WPF values measured for the American Optical respirator as suspect because the inside-the-facepiece filter samples contained glass fibers, originating from the respirator's filter matrix. These glass fibers have the same appearance as asbestos fibers under light microscopy. The authors did not adjust measured values for field blank values (

i.e.

, blanks were below the limit of quantification) or fiber retention in the respiratory tract (

i.e.

, the authors believed that pulmonary fiber retention resulted in only a slight change in concentration inside the facepiece).

The 3M 8710 results showed a GM WPF of 310, a GSD of 5.3, and a best estimate of the fifth percentile of 20. The 3M 9910 had a GM WPF of 580, a GSD of 4.2, and a best estimate of the fifth percentile of 55. The AO R1050 had a GM WPF of 52, a GSD of 4.2, and a best estimate of the fifth percentile of 5. The Survivair 2000 or MSA Comfo II equipped with DFM filters had a GM WPF of 240, a GSD of 6.3, and a best estimate of the fifth percentile of 12. With high-efficiency filters, the GM WPF was 94, the GSD was 3, and the best estimate of the fifth percentile was 16. For the North 7700 equipped with high-efficiency filters, the GM WPF was

250, the GSD was 6.9, and the best estimate of the fifth percentile was 11.

Since the WPFs for respirators equipped with DFM and high-efficiency filters were similar, and were well below the protection expected if filter efficiency alone was the determining performance factor, the authors concluded that “* * * filter efficiency was not as significant a factor in determining the relative workplace performance against asbestos as the face fit”. The authors also noted comparable performance between disposable and elastomeric respirators. With regard to this, the authors noted that perspiration and wetting solutions led to the elastomeric facepieces slipping on the participants' faces, something that was not noted with the fibrous disposable respirators. The authors postulate that the effect of this slippage could be a reason why the two types of respirators had similar performance.

Study 3.

In 1993, A. Gaboury and D.H. Burd performed a WPF study by measuring exposure to benzo(a)pyrene [B(a)P] on particles among 22 workers in a primary aluminum smelter (Ex. 1-64-24). The participants were rack raisers, stud pullers, and rod raisers on anode crews. The following three brands of elastomeric half-mask respirator devices were studied: Willson, Survivair, and American Optical. (

Note:

Respirator model numbers were not provided) The respirators were equipped with combination organic vapor/acid gas cartridges and DFM pre-filters, with the exception that dust/mist pre-filters were used on the American Optical respirator. The study also examined the performance of a powered air-purifying respirator (PAPR), but only the negative-pressure, air-purifying half-mask respirator data are presented here (the PAPR results are discussed below). The participants had used respirators for several years, had been previously trained in the use of the particular respirator under study, and had used it for more than six months. All participants in half-mask respirators were clean-shaven and were quantitatively fit tested using the TSI Portacount. The minimum acceptable fit factor was 100. Industrial hygiene technologists assisted participants with donning and doffing respirators, cleaned and maintained the respirators at the end of each work cycle, and observed participants on a one-to-one basis throughout the sampling period. Participants were directed not to tamper with the respirator or sampling equipment. Due to the high heat in the work area, the employer required that employees rest in a cool environment for one-half hour during each hour.

The inside-the-facepiece sampling train consisted of a closed-face three-piece cassette with a 25 mm organic binder free glass fiber filter, backed with a cellulose ester pad. The sampling cassettes were connected to a tapered Liu probe inserted into the respirator between the nose and mouth. The outside-the-facepiece sampling train was identical to the inside-the-facepiece sampling train; however, no mention is made of connecting the cassette to a Liu probe. All filters were pre-calcined at 400 degrees Centigrade for 24 hours. Both inside and outside samples were collected at a flow rate of 2 Lpm for approximately 300 minutes, or one-half of the 10-hour work shift. Respirators and sampling trains were worn and operated until the employee entered the rest area; they were donned and started prior to leaving the rest area for the next work cycle. Sampling cassettes were plugged when not in use and the respirators were cleaned after each work cycle. Field blanks were used to identify possible contamination due to handling. Sampling train airflow rates were checked at the beginning, middle (

i.e.

, after lunch), and end of the work day; on changing the cassettes; and when a problem was suspected. Sampling occurred over a five-day period. Only stud pullers and rod raisers used the elastomeric half-mask respirators.

B(a)P analysis followed the Alcan Method #1223-84. The ambient B(a)P particle size distribution was determined by collecting four samples, as close as possible to the workers, using an 8-stage Anderson cascade impactor (Model 296). Impactor samples were collected for two to five hours at a flow rate of 2 Lpm. The average percent of B(a)P mass (across four samples) per impactor stage (defined by an aerodynamic diameter cut point, in micrometers) was reported. About 93% of the B(a)P mass was associated with particles having diameters of less than 9.8 micrometers. A total of 18 pairs of inside and outside sample concentrations, with associated WPFs, were provided by brand of respirator and job category, but were not linked to specific participants. Overall GM, GSD, and 95% confidence interval on the mean were also provided for the inside and outside concentrations and WPF, along with an overall fifth percentile WPF. The authors stated that some employees participated more than once during the study. No mention is made of adjusting inside-the-facepiece concentrations for particle retention in the respiratory tract. The half-masks had WPF ranging from 13 to 410, with a GM of 47. The two-sided 95% confidence intervals were 30 and 74 for the dual cartridge respirators. The fifth percentile was 9. The authors found no significant relationship between B(a)P concentrations inside and outside the facepiece. Also, while the data were limited, the authors believed no correlation existed between WPF and quantitative fit factor. The authors concluded that the fifth percentile for the half-masks they tested were in agreement with the APF of 10 recommended by the NIOSH RDL.

Study 6.

S.W. Lenhart and D.L. Campbell reported in 1984 on a WPF study in which they measured protection against exposure to particulate lead (Pb) for 25 primary lead smelter workers; seven of whom worked in the sinter plant and eighteen of whom were in the blast furnace area (Ex. 1-64-42). The predominant aerosol forms of lead were dust in the sinter plant and fume in the blast furnace. In both areas, lead comprised about 50% of the total aerosol particulate with composition of the remaining 50% being unknown. All participants wore an MSA elastomeric half-mask with high-efficiency filters. (

Note:

No respirator model number was provided) The study also examined the performance of an MSA PAPR, but only data for the negative-pressure, air-purifying half-mask respirator are presented here (the PAPR results are discussed below). The employees routinely used respirators; however, no mention is made of them with respirator training. Participants were quantitatively fit tested using an unspecified method, and had to achieve the employer's required fit factor of 250. Workers were instructed not to remove or manipulate the respirator during sampling, and were observed by the researchers throughout the sampling period.

The inside-the-facepiece sampler consisted of a closed-face 37 mm cassette containing an AA filter and AP10 support pad. This cassette was connected to a tapered Liu probe that was inserted into the respirator between the nose and upper lip. In-mask samples were collected at 2 Lpm. The outside-the-facepiece sampling train was a closed-face 37 mm cassette containing an AA filter and AP 10 support pad; no tapered Liu probe was used. The outside sample cassette was attached to the worker's lapel. Outside samples were gathered at 2 Lpm. The authors collected samples for as much of each 8-hr work shift as possible. Respirators and sampling trains were donned and doffed, and samplers were started and stopped, in a lead-free area. Respirator facepieces were wiped clean inside

prior to donning after each break and cleaned and sanitized after each shift. One WPF was measured for each employee. The ambient particle size distribution was determined using 19 Marple cascade impactor samples (11 in the sinter plant; 8 in the blast furnace area).

Lead analysis was by flame atomic absorption spectroscopy according to NIOSH Method S-341. Inside-the-facepiece samples that contained less than l0ug of lead were reanalyzed by graphite furnace atomic absorption (limit of detection = 0.2 μg). The ranges for the mass median aerodynamic diameters (in micrometers) and for the GSD values were reported. A total of 25 pairs of inside and outside half-mask values, and the corresponding WPFs, were provided by employee, job title, and job location. An overall GM and GSD of the WPFs, and various percentile WPFs, were provided. When samples contained lead below the level of detection, the authors reported concentration values “* * * determined from the least amount of lead detectable by the analytical method and the sampled volume of air.”

In-mask values were not adjusted for particle retention in the respiratory tract (the authors imply retention probably had a non-significant effect on results, but could result in overestimated WPFs). No mention is made of the investigators using field blanks. They reported that approximately 98% of the WPFs would be expected to be at or above 10, 90% above 30, and 75% would be expected to be above 100. They concluded that an APF of 10 was appropriate for the half-mask negative pressure air-purifying respirator evaluated in this study. The authors also discussed two proportional methods of defining an APF.

Study 7.

W.R. Meyers and Z. Zhuang conducted a 3-part workplace protection factor study in three different work environments. In addition to presenting the study findings, the authors also discuss their rationale for selecting exposure agents, study facilities, and workers; study procedures followed at the sites; and analytical methods. W.R. Meyers and Z. Zhuang in January, 1993 (Ex. 1-64-51) and W.R. Meyers, Z. Zhuang, and T.J. Nelson in 1996 (Ex. 3-12) reported on the first part of the study in which the authors determined protection against exposure to particulate lead (Pb), zinc (Zn), and total airborne mass (TAM) for 25 workers, on day and evening shifts, in three brass foundries (3, 9, and 13 participants, respectively). (

Note:

The reports mention 26 participants, but data were presented for only 25 participants.) Four brands of half-mask devices were studied: 3M 9920 disposable DFM respirator; American Optical 5-Star elastomeric respirator with DFM filters (R56A); MSA Comfo II elastomeric respirator with DFM filters (Type S); and Scott Model 65 elastomeric respirator with DFM filters (642-F).

Participants were selected from volunteers who normally wore respirators, were clean-shaven, and passed a fit test. Their work rate was subjectively determined by observing their work activities. Respirators were worn for the usual period. For the elastomeric half-mask respirators, the participants were quantitatively fit tested using a TSI Portacount; a fit factor of 100 or more constituted a pass. Disposable respirators were fit tested using the saccharin qualitative fit test. The investigators trained the participants in the proper donning and adjustment of the respirators, and instructed them not to remove or lift the respirator from their face in the work area. Readjustment of the respirator had to be accomplished by sliding the facepiece on their face. Workers were observed throughout the sampling period. Each participant wore two or more respirator brands, and one WPF was measured per employee for each brand worn.

The inside-the-facepiece sampling train was a 25 mm closed-face cassette attached directly to a flared mouth probe, inserted into the respirator opposite the mouth. The cassette contained a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A 4.5 mm ring under the filter restricted airflow to an 18 mm circle in the center of the filter to keep deposition in an area that could be entirely covered by the proton beam used for sample analysis. A heating bonnet was slid over the outside of the cassette to minimize condensation of moisture from exhaled breath. Sampled air was then drawn through a moisture trap using a personal sampling pump operating at 2 Lpm. The outside-the-facepiece sampling train was a 10 mm nylon cyclone attached to 25 mm closed-face cassette (the cassette was not connected to a flared mouth probe). The cassette contained a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A 4.5 mm ring under the filter restricted airflow to an 18 mm circle in the center of the filter. This sampling train was attached in the lapel area and samples were collected at a flow rate of 1.7 Lpm.

Two separate samples were gathered during the shift, one during the first half and another during the second half. Individual WPFs were based on monitoring times of approximately one to four hours. Respirators were donned and doffed, and sampling trains were started and stopped, in a clean area. Elastomeric facepieces were cleaned and inspected at the end of each shift, but were not wiped out during the shift unless such wiping was a standard practice before the study (the authors noted that most of the time workers did not wipe out facepieces). Air-purifying filters (cartridges) and disposable respirators were changed at the end of each shift unless the employer's policy dictated more frequent changing. In addition, the mouth of the in-mask probe was plugged whenever the respirator was not being worn. Working (field) blanks and manufacturer's (media) blanks were used to determine possible contamination of filters due to handling or manufacturing. The investigators also washed the interior of the sampling cassettes to ascertain retention of sample particles on the cassette wall. The ambient particle size distribution was determined by PIXE 8-stage cascade impactor samples at several work locations in each foundry. These area samples were collected at roughly mid-chest to shoulder level of workers for approximately 1 hour, to prevent impactor overloading.

All samples were analyzed by proton induced X-ray emission analysis (PIXEA). The mass distribution of Pb, Zn, and TAM by particle aerodynamic diameter was graphically presented for all cascade impactor samples. Across the three foundries, 66 pairs of inside-the-facepiece and outside-the-facepiece concentrations, and the corresponding WPFs, were provided by job task, employee, brand of respirator, and analyte (Pb, Zn, and TAM). The authors did not adjust measured values for particle retention on sampling cassette walls since these losses appeared to be random, independent of collected mass, and of a negligible amount. No mention is made of correcting measured in-mask values for pulmonary particle retention. A foundry-specific average of the field blank loadings was used as a correction factor for estimating background and handling contamination for each foundry. Outside-the-facepiece samples were collected as respirable particulate, thereby providing respirable mass levels, while in-mask samples were collected as total particulate mass. The authors initially assumed that particles larger than 10 microns did not penetrate respirator faceseals; however, this was found to be incorrect after analyzing in-mask particle size. Therefore, to avoid comparison of dissimilar measurements, the investigators used particle size data

obtained by ambient sampling to convert the respirable mass levels to total mass levels (using Chimera/TSI Disfit software). The reported levels represent these total mass values, and form the basis of the reported WPF values. The authors also provide data and discussion on a number of sampling analyses, including GM concentration of analyte by job task, GM concentration of analyte for in-mask and ambient concentrations, particle size distribution by job category, GM WPF estimates by job category, GM WPF by respirator type, within shift sampling variation, and variation between foundries. For the pooled data from the three foundries, the 3M 9920 filtering facepiece had a 50% WPF of 108, a GSD of 5.2, and a fifth percentile estimate of 7. The AO half-mask had a 50% WPF estimate of 98, a geometric standard deviation (GSD) of 5.8, and a fifth percentile WPF of 5. The MSA Comfo II half-mask had a 50% WPF of 163, a GSD of 3.1, and a fifth percentile WPF of 26. The Scott half-mask had a 50% WPF of 94, a GSD of 4.8, and a fifth percentile WPF of 7. For all respirators a 50% WPF of 114, a GSD of 4.6, and a fifth percentile estimate of 9 was reported. The authors concluded that “* * * dust-fume-mist (DFM) half-facepiece respirators, when conscientiously used, worn, and maintained, provided effective worker protection.”

Study 8.

W.R. Meyers and Z. Zhuang in January, 1993 (Ex. 1-64-51) and W.R. Myers, Z. Zhuang, and T.J. Nelson in 1996 (Ex. 3-12) reported on the second part of the three-part study, which evaluated protection against exposure to particulate iron (Fe) for 16 workers in the sinter plant and basic oxygen process (BOP) facility of a steel manufacturing plant. In addition, exposure to particulate calcium (Ca) in the BOP facility was determined for one worker. The five brands of half-mask respirators studied were: 3M 8710 disposable dust/mist respirator; Gerson 1710 disposable dust/mist respirator; American Optical 5-Star elastomeric respirator with dust/mist filters (R30); MSA Comfo II elastomeric respirator with dust/mist filters (Type F); and Scott, Model 65 elastomeric respirator with dust/mist filters (642-D).

In general, each participant wore two or more brands, and one WPF was measured per employee per brand worn. One employee had one WPF determined for only one respirator brand. For the elastomeric half-mask respirators, the participants were quantitatively fit tested. A fit factor of 100 or more constituted a pass. Disposable respirators were fit tested using the saccharin qualitative fit test. The overall study and sampling protocols were discussed by the authors in the foundry portion of the investigation (

see

Study 7 discussion above). While not specifically discussed, it is assumed that the same sampling parameters used in the foundry study were in place during this particular study, unless the authors stated otherwise. These assumptions include: composition of the sampling trains was unchanged; individual WPFs were based on monitoring times of one to four hours; elastomeric facepieces were cleaned and inspected at the end of each shift but the insides were not wiped during the shift such wiping was the employer's standard practice before the study; air-purifying filter cartridges and disposable respirators were changed at the end of each shift unless the employer's policy dictated more frequent changing; and the in-mask probe mouth was plugged whenever the respirator was not being worn. In addition, it is assumed that the participants were clean shaven, normally used respirators, were trained in the proper donning and adjustment of the respirators, were instructed not to remove or lift the respirator from their face in the work area, and were observed throughout the sampling period.

The inside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A reducing ring under the filter restricted airflow to an 18 mm circle in the center of the filter to aid in PIXE analysis. A heating bonnet was slid over the outside of the cassette to minimize condensation of moisture from exhaled breath. This cassette was attached directly to a flared mouth probe, inserted into the respirator opposite the mouth. Sampled air was drawn through a moisture trap using a personal sampling pump operating at 1.5 Lpm. The outside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A reducing ring under the filter restricted airflow to an 18 mm circle in the center of the filter. The cassette was not connected to a flared mouth probe. This sampling train was attached in the lapel area and samples were collected at a flow rate of 1.5 Lpm. (Note: Unlike the foundry portion of the study, outside samples were collected as total mass rather than respirable mass samples.) Sampling pump flows were calibrated before and after each sampling period and pumps were monitored at approximately 15-20 minute intervals. Respirators were donned and doffed, and sampling trains were started and stopped, in a clean area. New cassettes were used for each sampling period. Working (

i.e.

, field) blanks and manufacturer's (media) blanks were used to determine possible contamination of filters due to handling or manufacturing. The investigators also washed the interior of the sampling cassettes to determine retention of sample particles on the cassette wall. The ambient particle size distribution was determined by PIXE cascade impactor samples. Personal impactor samples, rather than area samples, were collected at the steel mill sites (

see

foundry sampling procedures discussed above in Study 7).

Analysis for Fe and Ca on inside-the-facepiece filters was by proton induced X-ray emission analysis (PIXEA). Due to filter overloading, analysis for Fe and Ca on outside-the-facepiece filters was by atomic absorption spectroscopy. The mass distribution of Fe by particle aerodynamic diameter was tabulated for all cascade impactor samples. A total of 54 individual pairs of inside- and outside-the-facepiece concentrations, and the corresponding WPFs, were provided by shift and date, job category, employee, and brand of respirator. For 16 workers, the WPFs reported were based on the Fe data, while Ca data were used to calculate the WPF for one worker (flux unloader) in the BOP facility. Based on analytical information, the authors did not adjust measured values for particle retention on the walls of the sampling cassette. No mention is made of adjusting inside-the-facepiece values for particle retention in the respiratory tract. The average field blank mass loading was used as a correction factor for estimating background contamination. The 3M 8710 had a reported GM WPF of 377, a GSD of 3.7, and a fifth percentile WPF of 44. The Gerson 1710 had a reported GM WPF of 123, a GSD of 2.7, and a fifth percentile WPF of 24. The American Optical elastomeric half-mask had a reported GM WPF of 280, a GSD of 2.7, and a fifth percentile WPF of 56. The MSA Comfo II had a reported GM WPF of 427, a GSD of 4.3, and a fifth percentile WPF of 39. The Scott elastomeric half-mask had a reported GM WPF of 252, a GSD of 2.9, and a fifth percentile WPF of 45. The authors concluded that “The 5th percentiles for the WPF distributions for each respirator or pooled data were greater than 20.”

The authors also provided data and discussion on a number of sampling analyses, including GM concentration of analyte and GM WPF by job task, GM concentration of Fe inside the facepiece

and ambient and GM WPF by respirator brand, and particle size distribution by job category. The authors stated that “* * * half-facepiece respirators (maximum use concentration 10 times the PEL) were a suitable selection for the tasks included in this study.”

Study 9.

In January 1993, W.R. Meyers and Z. Zhuang reported on the third part of their investigation, in which they determined protection against exposure to particulate titanium (Ti), chromium (Cr), strontium (Sr) and total ambient mass (TAM) for 22 workers who spray painted aircraft on day, evening, and night shifts (Ex. 1-64-52). The three brands of half-mask elastomeric respirators studied were the: American Optical 5-Star, MSA Comfo II, and Scott Model 65. All respirators were equipped with combination high-efficiency filter/organic vapor cartridges.

Twelve participants each wore two brands of respirator with a WPF determined for each brand worn; nine participants wore one brand of respirator and had one WPF determined; and one employee had one WPF determined for one respirator brand and two WPFs determined for another brand. The participants were quantitatively fit tested and a fit factor of 100 or more constituted a pass. The overall study and sampling protocol was discussed by the authors in the foundry portion of the studies, summarized in Study 7 above (Ex. 1-64-51). While not specifically discussed, it is assumed that the same sampling parameters were in place during this particular study as in the foundry study, unless the authors stated otherwise. These assumptions include: composition of the sampling trains was unchanged; individual WPFs were based on monitoring times of one to four hours; elastomeric facepieces were cleaned and inspected at the end of each shift but were not the inside was not wiped during the shift, unless such wiping was the employer's standard practice before the study; filters and disposable respirators were changed at the end of each shift unless the employer's policy dictated more frequent changing; and the mouth of the in-mask probe was plugged whenever the respirator was not being worn. In addition, it is assumed that the participants were clean-shaven, normally used respirators, were trained in the proper donning and adjustment of the respirators, were instructed not to remove or lift the respirator from their face in the work area, and were observed by the researchers throughout the sampling period.

The inside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A reducing ring under the filter restricted airflow to an 18 mm circle in the center of the filter to aid in sample analysis. A heating bonnet was slid over the outside of the cassette to minimize condensation of moisture from exhaled breath. This cassette was attached directly to a flared mouth probe, inserted into the respirator opposite the mouth. Sampled air was then drawn through a moisture trap using a personal sampling pump operating at approximately 2 Lpm. The outside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.5 micron pore size polyethylene filter and polypropylene backup pad. A reducing ring under the filter restricted airflow to an 18 mm circle in the center of the filter. The cassette was not connected to a flared mouth probe. This sampling train was attached in the lapel area, and samples were collected at a flow rate of 1 Lpm. (

Note:

Unlike the foundry portion of the study, outside samples were collected as total mass rather than respirable mass samples.) Sampling pump flows were calibrated before and after each sampling period and pumps were monitored at approximately 15-20 minute intervals. Respirators were donned and doffed, and sampling trains were started and stopped, in a clean area. New cassettes were used for each sampling period. Working (

i.e.

, field) blanks and manufacturer's (media) blanks were used to determine possible contamination of filters due to handling or manufacturing. The investigators did not wash the interior of the sampling cassettes to determine retention of particles on the cassette wall, since a simple alcohol wash would not have removed dried paint spray. Ambient particle size distributions were not characterized.

Analysis of all filters was by proton induced X-ray emission analysis (PIXEA). The average field blank mass loading was used as a correction factor for estimating background contamination. The authors did not mention adjusting inside-the-facepiece measured values for particle retention in the respiratory tract. A total of 36 individual pairs of inside-the-facepiece and outside-the-facepiece concentrations of each analyte (total airborne mass, titanium, chromium, strontium) were provided by shift and date, painting location on the plane (

i.e.

, top, side, or underside of the aircraft), employee, brand of respirator, and paint type (

i.e.

, top coat, primer). A total of 36 WPFs were reported by shift, task location on the plane, employee, and respirator brand; of the original 38 data sets, two sets were eliminated as outliers. For primer spraying, the reported WPFs were based on Cr data, while WPFs for spraying topcoat were based on Ti data. WPFs were not calculated for total airborne mass. The authors also provided data and discussion on a number of sampling analyses, including GM concentration of analyte (TAM, Ti, Cr) for both in-mask and ambient measurements by task location on the plane; GM WPF as a function of painting location on plane and paint type, and respirator brand; and GM WPF by respirator brand. The fifth percentile estimates for all WPF data were reported to be much greater than 10. The authors concluded that these half-facepiece elastomeric respirators, when properly worn and used in conjunction with existing controls provided effective worker protection.

Study 13.

G. Wallis, R. Menke, and C. Chelton reported in 1993 on a WPF study in which they evaluated exposure to manganese dioxide dust for an unknown number of participants in several alkaline battery manufacturing plants (number of plants not provided) (Ex. 1-64-70). All participants wore the disposable 3M 8710 dust/mist respirator and performed their normal work activities. The participants were not trained by the investigators, but had been previously trained and routinely used respirators. It was not stated whether the participants had ever been fit tested for the 3M 8710 respirators. Prior to sampling, the participants washed their faces and were taken to a clean area, where the study was explained. The participants were observed throughout the sampling period.

The inside-the-facepiece sampling train was a closed-face 37 mm cassette containing a 0.8 micron pore size mixed cellulose ester filter. The cassette was connected to a tapered Liu probe (made of nylon) which was inserted into the respirator midway between the nose and mouth. The outside-the-facepiece sampling train was a closed-face 37 mm cassette containing a 0.8 micron pore size mixed cellulose ester filter. The outside sampling cassette was attached to the employee's lapel. No mention is made of connection of the outside cassette to a tapered Liu probe. Inside- and outside-the-facepiece samples were collected at an airflow rate of 1.5 Lpm for 30 to 40 minutes. The authors chose a short sampling interval to prevent resistance across the inside-the-facepiece sampling filter due to a buildup of moisture from exhaled breath. Sampling pump flows were

calibrated before, and rechecked after, each sampling period. Respirators were donned and doffed, and the sampling trains started (and assumed stopped), in the clean area. Field blanks were used to identify possible contamination of filters due to handling. The number of sample pairs collected per subject was not specified. The ambient manganese particle size distribution was determined by 6-stage Marple Cascade impactor equipped with an inlet cowl to prevent debris from entering the impactor. Samples were collected for several hours at a flow rate of 2 Lpm, and flows were calibrated before and after each sampling interval. Four samples were gathered: One in the powder drop area (Plant A) and three at the bag slitting operations (one in Plant A, two in Plant B).

Samples were analyzed for Mn by atomic absorption (AA) spectroscopy according to NIOSH Method 7300. The mass distribution of Mn by particle aerodynamic diameter was tabulated for all cascade impactor samples. Less than 30% of the mass was associated with respirable particles. A total of 70 individual pairs of inside-the-facepiece and outside-the-facepiece concentrations, and the corresponding WPFs, were provided by job activity (but not by employee or plant). No mention is made of adjusting measured values for particle retention in the respiratory tract or results of field blank analysis. A GM of 50 and a GSD of 3.5 was reported for all the WPF values measured. A calculated fifth percentile protection factor of 7.5 was also reported. The authors reported that their data indicated a systematic dependence of WPF on the concentration outside the respirator. In their discussion of this observation, the investigators refer to three possible causes presented by authors of other studies: Program protection factors tend to be low in low exposure settings since the workers, aware of the low exposure, exercise less care; low outside concentrations result in inside-the-facepiece concentrations so small that reliable quantification is difficult; and filter efficiency increases with loading, and low concentrations do not adequately load the filter. The authors discuss these causes relative to their study results, and postulate that another cause may be particle size selectivity (

i.e.

, smaller particles have a higher probability of entering the respirator). They conclude that it is important to characterize respirator performance in the environment where the respirator will be used.

Study 14.

At the 1990 AIHCE, C.E. Colton, A.R. Johnston, H.E. Mullins, C.R. Rhoe, and W.R. Meyers presented a WPF study in which they measured protection against exposure to aluminum dust for five participants working as carbon changers in an aluminum smelter (Ex. 1-64-15). All participants wore the disposable 3M 9906 dust/mist respirator. The investigators trained the participants in donning the respirator and the participants were qualitatively fit tested, although the fit test method was not described. The total number of samples collected per employee was not specified, although it is stated that the five employees were sampled daily for five days. Participants were observed throughout the sampling period.

The inside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.8 micron pore size polycarbonate filter. The cassette was connected to a tapered Liu probe, inserted into the facepiece in an unspecified location. In-mask samples were collected at an airflow rate of 2.0 Lpm. The outside-the-facepiece sampling train was a closed-face 25 mm cassette containing a 0.8 micron pore size polycarbonate filter. Outside samples were gathered as respirable dust samples with the cassette being connected downstream from a cyclone apparatus. Sampling airflow rate was 1.7 Lpm. Sampler airflow rates were calibrated before and after each sample period. No mention is made of donning and doffing procedures. Field blanks were used to identify possible filter contamination caused by handling. The ambient aluminum particle size distribution was determined through 12 area samples (unspecified locations) collected by Marple personal cascade impactors. In addition, particulates that passed a cyclone selector were sized by optical microscopy.

Aluminum was determined by proton induced x-ray emission analysis (PIXEA). The mass distribution of aluminum by particle diameter and percent penetration to the collector was graphically presented. Final calculations used only those outside filter weights that were greater that 11 times the detection limit. A total of 24 time-weighted-average (TWA) inside-the-facepiece and outside-the-facepiece concentrations, with corresponding TWA WPFs, are provided in supplemental data (Ex. 1-146). The sample pairs are not linked to specific participants. No mention is made of adjusting sample results for particle retention in the respiratory tract. The mean blank value was zero, so no adjustment to measured values was made. The authors reported a GM of 27, a GSD of 1.5, and a fifth percentile of 13 for the 23 sample sets used. The report concluded that the respirator provided reliable WPFs of 10. Cumulative probability of achieving a particular WPF, and the effect of filter weight on WPF, were also graphically presented. The authors stated that the WPFs represented conservative estimates of protection since outside concentrations were measured as respirable dust. In the summary of this study (Ex. 1-146), submitted to OSHA along with the raw sampling data, the authors recommended that the study not be used to assess the ultimate APF for this class of respirator since they felt that the real WPF of the respirator was significantly underestimated.

Study 15.

C.E. Colton, H.E. Mullins, and C.R. Rhoe presented a WPF study at the 1990 AIHCE in which they determined exposure to particulate Pb and Zn for 17 participants working in core making, mold making, pouring, and cleaning areas of a brass foundry (Ex. 1-64-16). All participants wore the disposable 3M 9970 high-efficiency respirator. The investigators trained the participants in the proper donning and fitting of the respirator, and participants were fit tested using the saccharin qualitative fit test method described in Appendix D of OSHA's Lead Standard (29 CFR 1910.1025). Sampling took place over five days.

The inside-the-facepiece sampling train was a 25 mm three-piece ca

This text is long and has been trimmed here. Open the source document for the complete record.