In short, certification fit testing of respirators depends upon two main factors: a well-conducted anthropometric survey of a representative sample of the respirator-wearing workforce, and a fit-test panel based on these anthropometric data that accurately represents the facial shapes and sizes of the millions of workers who use—or who should be using— respirators in their jobs.
For many years NIOSH has been using the fit-test panels developed in 1972 by researchers at Los Alamos National Laboratory (LANL) based upon anthropometric data available from a U.S. Air Force study. The population sample used by the LANL researchers was a group of men and women serving in the U.S. Air Force. It is unlikely that this sample was ever representative of the broader U.S. workforce—Air Force personnel are generally young and in good health, for instance, and the U.S. Air Force has height and weight requirements as well—but in the intervening years the U.S. workforce has become much more diverse, with more women workers and more minorities. Furthermore, the growing obesity problem in the United States means that workers are, on average, much heavier than they were two or three decades ago. In addition, the ethnic composition of the U.S. workforce had changed over the 30 years. So fit-test panels based on physical characteristics of Air Force personnel from the early 1970s are unlikely to accurately represent the broad U.S. workforce today.
Because of this situation, in 2001 NIOSH contracted with Anthrotech, Inc., to collect new anthropometric data that would be representative of today’s respirator-wearing workforce and to use those data to design new fit-test panels. After Anthrotech finished that task, NIOSH contracted with the Institute of Medicine (IOM) to establish an ad hoc committee to review the NIOSH-sponsored Anthrotech study. This report contains the findings, conclusions, and recommendations of that IOM committee (Box S-1).
The various elements of the NIOSH-sponsored Anthrotech study— anthropometric measurements, statistical sampling techniques, fit testing, and so on—are complicated, and this review must necessarily wade into them in some detail; but the basic message of this report can be summarized with the following three broad statements: