status of the participants in this trial was not determined as part of the study and was generally unknown. One of the nine persons who died was known to be receiving treatment for AIDS at the time of death, however. Of the other eight, at least six were noncompliant or did not complete prescribed therapy because of drug toxicity. Moreover, the causes of death reported in this trial were taken from death certificates, which may not have reflected the true cause (L.Geiter, personal communication, December 2000).
A large portion of individuals dying of tuberculosis have the diagnosis made at the time of death and hence do not receive therapy. Rieder and his colleagues examined this aspect of tuberculosis mortality in the United States for the years 1985 through 1988, a time period preceding the major explosion of both AIDS and multidrug-resistant tuberculosis in America (99).
Overall, 5.1 percent of tuberculosis diagnosed nationwide in those years was recognized at the time of death. During those years, there were a total of 7,210 tuberculosis deaths in the United States (1). Rieder and coworkers identified 4,373 diagnoses made at death. This represents 60.7 percent of the total deaths due to tuberculosis for those years.
Tuberculosis is not evenly distributed among Americans (1). About 23 or 24 percent occurs in individuals over the age of 65; presumably most of them are no longer in the work force. Nearly 60 percent occurs among the unemployed. Six percent of patients are inmates of correctional facilities; 6 percent are homeless; and 3 percent are residents of long-term-care facilities. Reduced access to health care among the homeless and the unemployed can be presumed to increase their risk of being diagnosed and treated late or not at all and, in turn, their risk of death from untreated tuberculosis.
HIV infection and drug resistance increase the mortality risk. During the decade prior to the HIV epidemic, Goble and colleagues at a national referral hospital noted a rate of mortality of 20.1 percent (27/134) among patients with tuberculosis due to organisms resistant to both isoniazid and rifampin (100). Data from 466 patients with a culture positive for M. tuberculosis in New York City in April 1991 were assembled by Frieden and collaborators (62). Follow-up of these patients was achieved until death or for 14 months. The case fatality rate for patients with multidrug-resistant organisms who were HIV infected was 80 percent; for HIV-uninfected individuals it was 47 percent.
In summary, the risk for immunocompetent individuals in the United States of dying from appropriately treated tuberculosis due to drug-susceptible organisms is vanishingly small. For health care workers it should be smaller than for the general population because they should have the advantages of more rapid diagnosis and institution of appropriate therapy