Dr. Blinkhorn makes several comments in response to our article. Although some of the intensified tuberculosis control measures that we described were in place already, there was substantial room for improvement, even in a city with an excellent control program. Regarding the impact of HIV-tuberculosis co-infection on clustering, HIV infection was not the only variable affecting clustering in our population. In fact, the rate of clustered cases decreased significantly among persons not infected with HIV or those with unknown HIV status. In addition, rates of clustering were found to decrease among persons born in the United States (a group at high risk for clustering).

We disagree that knowledge of DNA fingerprinting results will not directly affect the clustering phenomenon. For example, our experiences in San Francisco demonstrating marked clustering among HIV-infected persons and others (1, 2) were the impetus for creating several of the intensified tuberculosis control measures described in our study, including improved contact investigations, expanded use of directly observed therapy, development of an HIV-related tuberculosis prevention program, and screening for tuberculosis among persons in residential care facilities. These measures probably accounted for the marked decreases in clustered cases that we observed. Furthermore, as technology improves, the possibility of "real-time" DNA fingerprinting will become a reality (3) so that the results can directly influence tuberculosis control.

Dr. Blinkhorn also hypothesizes that clustering occurs when the source case-patient is immunocompromised or has far-advanced disease before seeking medical attention and that a tuberculosis control program cannot influence either scenario. He states that in Cleveland, secondary cases among contacts to the above persons were common because medical attention was not sought by "disenfranchised members of the community" until active tuberculosis was present. We disagree. In fact, one of our main accomplishments was in expanding the scope of our contact investigations so that we identified more contacts per case; far fewer cases had no contacts than in previous years. This was especially true among case-patients born in the United States, who frequently had a history of substance abuse, homelessness, or HIV infection. This was largely the result of intensive efforts to improve communication with these disenfranchised persons. We agree that public education regarding the symptoms of tuberculosis is important, but DNA fingerprinting is especially useful in both suggesting and evaluating methods to decrease transmission of tuberculosis. By showing only the cases and not infections, DNA fingerprinting reveals the missed opportunities for prevention of tuberculosis.