low, but well-structured programs involving education and directly observed therapy can improve rates of completion of treatment for both conditions (Camins et al., 1996). Recent guidelines from the CDC and the American Thoracic Society strongly recommend treatment for latent infection (ACT/CDC, 2000b).

In sum, the committee believes that the 1997 estimates of cases of active tuberculosis that a rule will avert are overstated on three grounds. First, the estimate is inconsistent with reported data on tuberculosis cases by occupation. Second, the rate of progression from infection to active disease is likely lower than traditionally cited. Third, the estimate of infection levels to which the progression rate is applied is too high. One committee member disagreed with this general assessment. That member argued that the validity and reliability of CDC’s own data on tuberculosis case rates by occupation are questionable and that the 10 percent progression figure is reasonable since it continues to be cited by CDC.

In the 1997 proposed rule, OSHA estimated that the proposed rule would prevent between 115 and 136 tuberculosis-related deaths among covered workers each year. (It also estimated that the rule would also avert 23 to 54 additional deaths among family and other contacts of workers.) The committee concludes that the mortality estimates are overstated. First, as discussed above, the committee believes that the estimates of number of tuberculosis cases that would be averted by a standard are too high. Second, the estimated mortality rate used in the assessment does not take into account demographic factors or the effects of treatment.

In the 1997 proposed rule, OSHA estimated that 7.8 percent of all active tuberculosis cases among workers would end in death. It based the estimate on the 3-year average of mortality data reported by CDC for 1989 to 1991 (62 FR 201 at 54207). (More recent CDC surveillance reports apparently include revised numbers for tuberculosis cases and deaths for these years. Based on these numbers, the average case death rate for 1989– 1991 is 7.3 percent.) Case mortality rates reported by CDC for recent years are lower: 6.0 percent in 1998, 5.9 percent in 1997, and 5.6 percent in 1996 (CDC, 2000b). (Population mortality rates dropped from 0.8 per 100,000 population in 1989 to 0.6 in 1994 to 0.4 in 1998.) Thus, use of revised and recent tuberculosis case mortality data would reduce the OSHA estimates.

More important, estimates of deaths among health care and other workers should take into account the effects of treatment. The majority of deaths due to tuberculosis occur in individuals in whom the disease is first recognized after death, meaning that their disease was not being treated (Rieder et al., 1991). In addition, the majority of cases of tuberculosis occur among unemployed individuals (CDC, 2000b). Such individuals