Tuberculosis transmission in Botswana.

Lockman and colleagues recently described a study on conventional and molecular epidemiology of tuberculosis in Botswana (4) and found a rate of clustering cases of pulmonary tuberculosis equal to 42%. The authors were surprised by the result, since they expected a higher rate of clustering in a country such as Botswana, where the incidence of tuberculosis is extremely high (over 500 per 100,000). The authors discussed factors that might have biased the estimation of the extent of recent transmission of tuberculosis and indicated that high population mobility and a rather incomplete epidemiological evaluation of the subjects enrolled likely accounted for the level of transmission recorded. In our opinion, a crucial issue was not taken into consideration to explain the unexpected rate. In their paper (Materials and Methods), the authors stated that only patients who had both acid-fast bacillus (AFB)-positive and culture-positive sputum were recruited and epidemiologically evaluated. This was probably due to obvious organizational and logistic problems in such a setting, but sputum smears are known to be AFBpositive only in 50 to 70% of culture-positive pulmonary tuberculosis (2). This rate may be affected by specific clinical and epidemiological conditions. HIV-infected tuberculous subjects, for example, especially those who are seriously immunosuppressed and develope cavitary lesions less frequently (3), are prone to show a lesser degree of AFB-positive sputum smears. These considerations and the high incidence of human immunodeficiency virus-tuberculosis coinfection reported in the study area (65% of the eligible patients) suggest that a sizeable proportion of pulmonary tuberculosis was not included in the epidemiological analysis of Lockman et al. This may have led to their underestimating the clustering rate in two different ways. First, a relevant portion of pulmonary tuberculosis due to recent transmission might have not been evaluated in the clustering analysis, because the secondary pulmonary cases generated in a cluster would not necessarily have yielded an AFB-positive sputum smear. Second, the infectiousness of smear-negative but culture-positive tuberculosis was recently revalued by Behr et al. (1), who showed that such cases were responsible for about 17% of tuberculosis transmission in San Francisco, Calif., yet the potential sources of transmission with smear-negative but culture-positive pulmonary tuberculosis were not identified by Lockman and coworkers. In conclusion, we think that fingerprinting limited to tuberculosis cases where the sputum smear was AFB positive should be considered the main confounding factor of the clustering analysis of this study.