Evaluating a biochemical TB medication adherence test: Insights from image-based reporting through a patient companion app.

Tuberculosis (TB) treatment requires strict adherence over the long course of treatment. To support adherence, we developed the TB Treatment Support Tools (TB-TST), which combines a urine-based lateral flow assay (LFA) with a mobile application for patient education, support, and communication with treatment supporters. In a pragmatic randomized controlled trial participants uploaded images of the LFA. A blue-purple color change indicates detection of an isoniazid metabolite consistent with recent medication ingestion. This study assessed real-world interpretability and usability of the TB-TST LFA by analyzing the quality and characteristics of user-submitted images. We conducted structured qualitative coding (visual content analysis) of 1,655 submitted test images to identify recurring issues affecting interpretation. Eligible participants were aged 16 or older, newly diagnosed with TB, and had access to a mobile phone. An iteratively developed codebook identified three primary issue categories: 1) color variations in the test or control strip, 2) photo quality issues, and 3) test cartridge/test-strip artifacts. Most (72%) images were consistent with a clear positive (blue-purple) color change or clear negative. Variations that were not clear positives included no color change (white), yellow/orange, and green/teal hues. Blurriness was the most common image quality issue. The most common test artifact was visual evidence consistent with dye bleed/possible cross-strip contamination between the control and the test strips. Despite challenges in image submission and test appearance, relatively few images were difficult to interpret, supporting the feasibility of user-submitted LFA images for monitoring recent isoniazid ingestion. Opportunities to improve performance and usability include enhanced user guidance for image capture, refinement to cartridge design to reduce artifacts, clearer interpretation protocols for atypical color outcomes, and automated image-based reading. Integrating qualitative image analysis with engineering design principles can inform user-centered improvements to home-based LFAs connected to digital adherence technologies.