Twenty-four hour face mask sampling in pulmonary tuberculosis reveals three distinct patterns of bacterial aerosol production dissociated from conventional markers of transmission risk.

Rationale: Although tuberculosis (TB) is transmitted by Mycobacterium tuberculosis (Mtb) in aerosols, little is known of the dynamic characteristics of spontaneous output of bacilli in this form. We have developed and implemented a mask aerosol sampling system (MASS) for longitudinal capture and study of spontaneous aerosol.\n\nObjective: To determine patterns of Mtb output in aerosols, captured using the MASS over 24 hours and their association with existing criteria used to assess transmission risk in patients with pulmonary TB.\n\nMethods: Twenty-four hospitalised patients with newly diagnosed pulmonary TB recruited in Pretoria, South Africa, wore FFP1 masks for one hour out of every three for 24 hours. Aerosol was captured in a gelatine filter processed for Mtb quantitation by PCR. Serial sputum was collected and objective cough frequency monitoring performed over the same period.\n\nMeasurements and Main Results: Mtb was detected in 86.5% of 192 mask samples and 20.7% of 38 assessable sputum samples obtained from the cohort. Mtb was detected by MASS in all but two patients. Three dynamic patterns of expression were identifiable in Mtb aerosol producers: i. variable high; ii. consistent; and iii. variable low. No diurnal variation was apparent and there was no correlation between mask Mtb and either sputum Mtb levels or cough frequency. Sputum smear status, culture time to positivity and chest radiographic characteristics also failed to associate with MASS bacillary output.\n\nConclusions: Conventional markers of tuberculosis case infectivity do not predict bacillary aerosols detected by the MASS. The MASS provides a novel, non-invasive tool for tuberculosis diagnosis and control.