Setting: Bangladesh national tuberculosis control programme districts supported by the Damien Foundation (population 18 million).
Objective: To determine whether the observed gender difference is epidemiologically true or whether it is due to accessibility barriers for women.
Design: Two-stage random sampling population survey. People were interviewed by teams composed of one male and one female volunteer about the presence of symptoms suggestive of tuberculosis. One early morning sputum specimen of all thus identified tuberculosis suspects was examined by Ziehl-Neelsen technique.
Results: A total of 266189 people were surveyed, among whom 223 936 (84.1%) could be interviewed. This led to the identification of 7001 tuberculosis suspects (2406 females and 4595 males) and 64 confirmed TB cases (16 females and 48 males). The female/male ratio (0.33:1) of cases found during the survey was not higher than that observed through routine diagnosis (0.42:1). The routine case-notification rate of sputum-positive cases during 2000 was about twice the rate observed during the survey (44.3 vs. 24.0 per 100 000 population).
Conclusions: The gender difference observed in routine tuberculosis diagnosis is real, and is not due to lesser accessibility of women to the health services. The routine notification rate compares favourably with the detection of previously unknown cases found during the survey.