EP-TBS-40 What can tuberculosis prevalence surveys tell us about the duration of asymptomatic bacteriologically-positive disease?

Background: Ratios of tuberculosis prevalence to notifi- cation rates are used to characterise the typical durations of tuberculosis disease. However, standard approaches ignore the spectrum of tuberculosis disease, and time with few or no symptoms prior to care-seeking. Methods: We developed novel analytical models to esti- mate progression from initial bacteriological-positivity including (depending on data): smear conversion, symp- tom onset and initial care-seeking. Case-detection ratios were also estimated by fitting the model to tuberculosis prevalence and notification data (1 subnational and 10 national datasets) within a Bayesian framework using Markov chain Monte Carlo methods.

Results: For Kenya and Blantyre, Malawi individual-lev- el data were available. The sex-specific durations of as- ymptomatic bacteriologically-positive tuberculosis were 0.31 years (95% Credible Interval, CrI: 0.21-0.45) and 0.41 years (95% CrI: 0.29-0.54) for females and males in Kenya; 0.28 years (95% CrI: 0.13-0.51) years a