Contamination rates in serially sampled sputum specimens obtained during tuberculosis treatment to capture culture conversion.

Sputum cultures are the gold standard for tuberculosis (TB) diagnosis and treatment monitoring. However, cultures in MGIT liquid media are susceptible to microbial contamination, often rendering them uninterpretable. Research has shown that maintaining strict cold chains and supervised sample collection can reduce contamination rates, but few longitudinal studies with weekly sampling have explored this. Here we evaluated whether (1) the time between specimen collection and laboratory processing and (2) unsupervised specimen collection are associated with contamination rates. Additionally, we estimated contamination rates over the first 12 weeks of treatment and assessed clinical and behavioral predictors of contamination. We collected 3155 sputum specimens from 301 participants undergoing TB treatment. Contamination was lowest (12.3%) at treatment initiation, increased over the first few weeks, and stabilized around 30% from week 8 onwards. Samples collected without supervision were more likely to be contaminated at treatment initiation (p=0.048) and over the 12 weeks (p=0.028). We observed an inverse relationship between smear grade and contamination risk throughout the sampling period. These findings underscore the importance of supervised sputum collection to reduce contamination and provide ways to enhance the clinical and research value of weekly cultures, particularly those collected later in treatment. This is especially relevant for community-collected specimens used in monitoring treatment response.