Time Trend Analysis of Tuberculosis Treatment While Using Digital Adherence Technologies—An Individual Patient Data Meta-Analysis of Eleven Projects across Ten High Tuberculosis-Burden Countries

Abstract

Worldwide, non-adherence to tuberculosis (TB) treatment is problematic. Digital adherence technologies (DATs) offer a person-centered approach to support and monitor treatment. We explored adherence over time while using DATs. We conducted a meta-analysis on anonymized longitudinal adherence data for drug-susceptible (DS) TB (n = 4515) and drug-resistant (DR) TB (n = 473) populations from 11 DAT projects. Using Tobit regression, we assessed adherence for six months of treatment across sex, age, project enrolment phase, DAT-type, health care facility (HCF), and project. We found that DATs recorded high levels of adherence throughout treatment: 80% to 71% of DS-TB patients had ≥90% adherence in month 1 and 6, respectively, and 73% to 75% for DR-TB patients. Adherence increased between month 1 and 2 (DS-TB and DR-TB populations), then decreased (DS-TB). Males displayed lower adherence and steeper decreases than females (DS-TB). DS-TB patients aged 15–34 years compared to those >50 years displayed steeper decreases. Adherence was correlated within HCFs and differed between projects. TB treatment adherence decreased over time and differed between subgroups, suggesting that over time, some patients are at risk for non-adherence. The real-time monitoring of medication adherence using DATs provides opportunities for health care workers to identify patients who need greater levels of adherence support.

Keywords

• tuberculosis
• digital adherence technologies
• meta-analyses
• implementation research
• multi-country
• medication adherence