SETTING: Gweru Provincial Hospital, Zimbabwe.

OBJECTIVES: To evaluate the accuracy of the diagnosis of pulmonary tuberculosis (PTB) in routine circumstances.

DESIGN: Prospective cohort study of routinely diagnosed PTB patients between September 2000 and September 2001.

RESULTS: Of 300 patients started on treatment, 161 (53.7%) were positive on direct microscopy. Of the 139 sputum-negative patients, 51 (36.7%) were positive after concentration of specimens, an additional 30 (21.6%) were positive on culture only and 58 (19.3% of the total) were negative for all laboratory investigations. There was no difference in sex, human immunodeficiency virus (HIV) serostatus or treatment outcome between the culture-positive and culture-negative presumed PTB pa-tients. Sputum-negative patients had an increased risk of dying during treatment (RR 2.39, P = 0.015).

CONCLUSIONS: The laboratory findings reveal that PTB could be confirmed in more than 80% of patients put on treatment in this setting. The treatment outcomes of the remainder did not differ from those in patients with microbiologically confirmed PTB. In a high HIV-prevalent area, clinicians rightly consider the results of the sputum microscopy test as only one element in the decision making process to put a patient on TB treatment.