Effect of digital adherence technologies on treatment outcomes in people with drug-susceptible tuberculosis: four pragmatic, cluster-randomised trials.

Background: The impact of digital adherence technologies on tuberculosis treatment outcomes remains poorly understood. We investigated whether smart pillboxes and medication labels can reduce poor treatment outcomes in patients with tuberculosis.

Methods: We did independent pragmatic, cluster-randomised trials in the Philippines, South Africa, Tanzania, and Ukraine. 110 clusters were randomly assigned (1:1) to standard of care versus intervention arms, which were further randomly assigned (1:1; except in Ukraine) to a smart pillbox or medication labels. We enrolled adult patients receiving treatment for drug-susceptible tuberculosis. The pillbox gave an audio-visual reminder to take medication, and when the box was opened, a signal was transmitted to the adherence platform. Those in the labels arm received medications with label attached, showing a code, which they messaged when a dose was taken; otherwise, a reminder was sent. The primary outcome was a composite poor end of treatment outcome, defined as having documented treatment failure, loss to follow-up (treatment interruption for ≥2 consecutive months), switched to a multidrug-resistant regimen more than 28 days after treatment start, or death. The trials are complete and registered with ISRCTN, 17706019.

Findings: Between June 21, 2021, and July 8, 2022, we enrolled 25 606 individuals (12 626 on standard of care and 12 980 on intervention) across 220 clusters in the four trials, of whom 23 483 (91·7%; 11 313 on standard of care and 12 170 on intervention) were included in the intention-to-treat population. 8208 (35·0%) of 23 483 individuals were female. 9717 (85·9%) of 11 313 individuals in the standard of care arm and 10 540 (86·6%) of 12 170 individuals in the intervention arm were analysed for the primary outcome. The risk of the primary outcome did not differ by intervention arm for all countries (Philippines adjusted odds ratio 1·13, 95% CI 0·72-1·78, p=0·59; Tanzania 1·49, 0·99-2·23; p=0·056; South Africa 1·19, 0·88-1·60; p=0·25; Ukraine adjusted risk ratio 1·15, 95% CI 0·83-1·59; p=0·38). Two incidents of social harm were reported due to inadvertent disclosure of treatment status in the pillbox arm, resulting in withdrawal of the participants.

Interpretation: Digital adherence technologies did not reduce poor treatment outcomes in the four countries investigated. The use of digital adherence technologies should be based on careful review of additional data on economic evaluation, patient and stakeholder preferences, and the effect on other important patient outcomes beyond programmatic treatment outcomes.

Funding: Unitaid.