difficult in the absence of any recent systematic data on infection levels on a national or state-by-state basis. The last national survey of infection was undertaken in the early 1970s. As discussed in Chapters 5 and 6 and Appendixes C and D, all studies used to estimate the occupational risk of infection and the effects of tuberculosis control measures have their limitations.
The committee recognizes OSHA’s efforts to take some criticisms of its estimating strategy into account. For example, after reviewers criticized the use of 1982 and 1984 Washington State data, in part, because the data were more than 10 years old, OSHA staff substituted data from a 1994 survey. Similarly, after data from a 1984–1985 North Carolina survey were criticized as likely being confounded by cross-reactions to atypical mycobacteria in the central and eastern parts of the state, the analysts used only data from hospitals in western North Carolina. The analysts also adjusted the 1984 data to reflect subsequent decreases in active tuberculosis in the state.
Nonetheless, the committee concludes that OSHA’s original estimate that the proposed regulations would reduce yearly work-related tuberculosis infections by 90 percent from 1994 levels (thereby averting 21,400 to 25,800 infections) is overstated.5 As discussed above, tuberculosis cases and case rates have declined substantially since 1993. Further, the committee is concerned that OSHA’s analysis did not adequately recognize the contributions to worker infections of (1) unsuspected and undiagnosed cases of active tuberculosis in the workplace or (2) exposure in the community. One concern involves the choice of the definition for internal control and exposed groups for Washington State data (definition 1 as discussed in 62 FR 201 at Table V-3). Another concern is the use of North Carolina data flawed by very low hospital response rates and inconsistent skin testing procedures.
The committee also has some concerns about OSHA’s use of 1991 data from Jackson Memorial hospital, which experienced a 1989 to 1990 outbreak of tuberculosis among patients on an HIV ward. Although the data used were for the year after the conclusion of the outbreak, skin test conver-
The agency’s state-by-state estimates of the “annual excess risk of tuberculosis infection due to occupational exposure” were defined as “a multiplicative function of the background rate of infection” (62 FR 201 at 54192). The agency then derived its estimates of the background rate of infection on the basis of a mathematical model that assumes that the rate of infection in an area can be “expressed as a numerical function of active tuberculosis cases reported in the same area” (62 FR 201 at 54197). Given the limited time and resources available to it, the committee did not evaluate this mathematical model. In the 1997 proposed rule, OSHA estimates the occupational risk of tuberculosis infection over a 45-year working lifetime to range from 4 to 723 per 1,000 population for hospital workers (with the lowest estimates based on the Washington State data and the highest based on the Jackson Memorial Hospital data).