M S Khan

Miss M S Khan, University of London, London, UK; mishalk@cantab.net

STUDY DESIGN

Design:

randomised controlled trial.

Allocation:

{concealed}*.†

Blinding:

blinded (laboratory staff).†

STUDY QUESTION

Setting:

specialised outpatient tuberculosis (TB) hospital in Rawalpindi, Pakistan.

Patients:

3055 patients 14–75 years of age (mean age 35 y, 51% men), who had cough for >3 weeks or fever for ⩾1 month, blood in sputum, night sweating, weight loss, and loss of appetite. Exclusion criteria were history of TB diagnosis and treatment or oral steroid use ⩽3 months before presentation.

Intervention:

1520 patients were allocated to instruction (2 min), given in a separate room by a female health worker, before getting a sputum container from the laboratory. The instruction group was told the importance of providing a sputum rather than saliva sample and the difference between the 2, shown how to expectorate good sputum (after 3 deep breaths, cough deeply to bring up …