NIOSH charges wrong, say makers of N95s

NIOSH changes would be costly, they say

By Janice Bradley, CSP
Technical Director
International Safety Equipment Association
Arlington, VA


[Editor’s note: In the article below, the International Safety Equipment Association, a professional association of manufacturers of N95 respirators, responds to an article that appeared in the March 2001 issue of TB Monitor ("NIOSH cites poor fit of many current N95s, urges fit-test change"). In that article, researchers at the National Institute of Occupational Safety and Health (NIOSH) said that most N95 half-mask filtering facepiece respirators have been shown in tests by NIOSH to perform poorly. Researchers said that only 45% of subjects who were fit-tested could be adequately fitted, and that the respirators may pose a hazard to health care workers who count on the masks for protection against TB. In the article, NIOSH experts said they believed the onus to make sure masks fit should be placed back on the manufacturer. They also said they wanted to see fit testing changed from the current qualitative method to a quantitative method.]

The International Safety Equipment Association (ISEA) finds it necessary to respond to the article "NIOSH cites poor fit of many current N95s, urges fit-test change," published in the March 2001 issue of TB Monitor. ISEA is the trade association in the United States for companies that design and manufacture protective clothing and equipment used in factories, construction sites, health care facilities, and many other settings. A number of statements made by persons and agencies in the subject article are inaccurate and misleading. The comments and conclusions drawn by the NIOSH investigators are their own, based on opinions that do not represent a consensus of experts in the field.

NIOSH researchers identify two major concerns in their comments:

  • "Standard tests" show that N95 filtering facepiece respirators are ineffective.
  • Bitrex qualitative fit testing (QLFT) does not adequately identify poor-fitting respirators.

They propose resolving these concerns by:

  • requiring fit testing as a part of respirator certification;
  • requiring end users to perform quantitative fit tests (QNFT) on each respirator wearer. This is in contrast to current regulations, which allow either QNFT or QLFT to be used.1,2

This article will explain why NIOSH’s concerns are unfounded and their proposed resolutions unnecessary.

The laboratory study cited to support the claim that N95 respirators fit poorly was conducted with an unvalidated, non-standard method.3 Preliminary work performed by other investigators indicates that the NIOSH method is too variable to produce meaningful results. No one outside of NIOSH has used this method to draw any conclusion about respirator performance. It is also important to note that the manufacturer of the PortaCount Plus (TSI, St. Paul, MN) chose not to adopt the NIOSH method to allow N95 filtering facepiece respirators to be fit-tested with its instrument. TSI developed an alternative accessory known as the N95-Companion for this purpose. At best, it can be said that the results of the NIOSH study are questionable.

The NIOSH investigators further interpret their study results to be an indication of poor respirator performance and poor quality. It clearly indicates neither. Fit, quality, and performance (defined as how well a respirator protects the wearer in a workplace) are separate issues. Even if the study results are accepted at face value, the only valid conclusion that can be drawn is that some of the respirators NIOSH tested fit very few or none of the test subjects.

OSHA regs require fit tests

It is NIOSH’s belief that poor fit on its panel of test subjects puts workers at risk. Again, this is clearly incorrect. For more than 20 years, Occupational Safety and Health Administration (OSHA) regulations have required each user of a filtering facepiece respirator to be fit-tested by their employer.4 Workers who do not pass a fit test conducted by their employer are not to be permitted to wear the tested respirator. If NIOSH investigators know of specific employers who refuse to comply with this regulation, they should notify OSHA for appropriate enforcement action.

It should also be noted that NIOSH’s contention that requiring respirators to pass some sort of fit test during certification would do nothing to increase worker protection. No panel of test subjects (neither the 1972 Los Alamos panel nor the updated panel NIOSH proposes) can represent 100% of the working population. Individual fit tests on each wearer will still be necessary to determine if even a "good fitting" respirator fits a specific user. It can be logically argued that NIOSH’s plan to include fit testing in the certification process is counterproductive to the goal of reducing workers’ exposure risk. Employers who are given NIOSH assurance that they are purchasing "good fitting" respirators may be less likely to conduct the fit tests OSHA requires (Why bother? NIOSH says these respirators fit.). This would increase risk rather than reduce it.

The NIOSH investigators’ contention that qualitative fit tests are less effective than quantitative tests in identifying poorly fitting respirators is not supported by available data. Analysis of respirator performance in the workplace demonstrates that both QLFT and QNFT are effective in screening out poor-fitting respirators.5 If NIOSH is aware of data that refute this conclusion, they should publish it. In addition, the study NIOSH cites to support their belief that the Bitrex QLFT produces false negatives (which are conservative, in that workers with acceptable fit fail the test) was performed with unvalidated procedures.6 Accepted procedures for evaluating fit tests have been in use for nearly 10 years and have recently been formalized.7 A previously published article demonstrated that the Bitrex fit test meets the criteria for an acceptable fit test.8 NIOSH does not explain why they did not follow the recognized procedures to determine if the original work is reproducible.

One key element in evaluating the efficacy of a fit test requires comparison of its failure results with the failure results of a reference fit test without disturbing the fit of the respirator. This is necessary so the "same fit" is evaluated by both tests. Interestingly, the conclusions to the NIOSH study state, "One possible explanation for the high error (false negatives and false positives) seen in this study could be an inherent variation in facepiece fit between different donnings" (emphasis added). It is puzzling that NIOSH chose an experimental method that essentially compares apples to oranges if they were truly interested in assessing the performance of the Bitrex fit test.

It must be recognized that the NIOSH computer simulation discussed in the subject article is a purely theoretical exercise.9 NIOSH believes its results indicate that the risk of assigning an individual a poor-fitting respirator compounds as successive fit tests are failed. However, the model is based on flawed assumptions that render its results meaningless. Oversimplified assumptions were used to model a complex process. For example, "fit" is defined as a single point estimate; in reality it is a distribution of values. The model assigns people who fail an initial fit test the same fit characteristics as the original population. However, because they have failed the first fit test, they should be defined as a new group with different fit characteristics from the original group. This error leads to the incorrect assumption that the new group will continue to be assigned a poorly fitting respirator at the same rate as the original group.

A second flawed assumption of the NIOSH model mischaracterizes the performance of a fit test. Any fit test has an associated beta error, which is the chance that a subject will pass the fit test with respirator performance less than the desired level. The model assumes the beta error of a fit test is uniform over the entire range of fit factors below the passing value. In reality, the beta error decreases with decreasing fit factors. This means that poorer fits are more likely to be rejected than are fits close to the passing value. In other words, the assignment error rate will be much lower than the NIOSH model indicates.

In conclusion, there is no need for the health care industry to develop its own respirator certification as NIOSH suggests. Such a system would create confusion and would likely increase the cost of respirators for health care workers. Current NIOSH certification procedures assure that all filtering facepiece respirators effectively remove aerosol contaminants. If existing OSHA regulations for fit testing are followed, the probability that poor-fitting respirators will be assigned to users is low. Economic forces will eliminate respirators that fit few people from the market: Employers with sound respiratory protection programs will not buy respirators that don’t fit and therefore can’t be used by their employees. As in any other industry, health care facilities that administer comprehensive respiratory protection programs are currently providing effective protection to their respirator wearers.


1. "Respiratory Protection." Code of Federal Regulations Title 29, Section 1910.134. 1999, pp. 402-427.

2. "Respiratory Protection for M. Tuberculosis." Code of Federal Regulations Title 29, Section 1910.139. 1999, pp. 431-435.

3. Centers for Disease Control and Prevention. Laboratory performance evaluation of N95 filtering facepiece respirators, 1996. Morbidity and Mortality Weekly Report 1998; 48:1045-1049.

4. Occupational Safety and Health Administration. Respirator Fit Testing. (CPL 2-2.29). 1980.

5. Nelson TJ. The assigned protection factor of 10 for half-mask respirators. Am Ind Hyg Assoc J 1995; 56:717-724.

6. Coffey CC, Zhuang Z, Campbell DL. Evaluation of the Bitrex qualitative fit test method using N95 filtering facepiece respirators. J Int Soc Respir Prot 1998; 16:48-55.

7. American National Standards Institute. American National Standard Respirator Fit Testing Methods (ANSI Z 88.10). Fairfax: American Industrial Hygiene Association, 2001.

8. Mullins HE, Danisch SG, Johnston AR. Development of a new qualitative test for fit testing respirators. Am Ind Hyg Assoc J 1995; 11:1068-1073.

9. Campbell DL, Coffey CC, Lenhart SW. Respiratory protection as a function of respirator fitting characteristics and fit-test accuracy. AIHAJ 2001; 62:36-44.