Digital adherence technologies to improve tuberculosis treatment outcomes in China: a cluster-randomised superiority trial

BACKGROUND: Drug-sensitive tuberculosis treatment requires 6 months of therapy, so adherence problems are common. Digital adherence technologies might improve tuberculosis treatment outcomes. We aimed to evaluate the effect of a daily reminder medication monitor, monthly review of adherence data by the health-care provider, and differentiated care for patients with adherence issues, on tuberculosis treatment adherence and outcomes. METHODS: We did a cluster-randomised superiority trial across four prefectures in China. 24 counties or districts (clusters) were randomly assigned (1:1) to intervention or control groups. We enrolled patients aged 18 years or older with GeneXpert-positive, rifampicin-sensitive pulmonary tuberculosis, who were receiving daily fixed-dose combination treatment. Patients in the intervention group received a medication monitor for daily drug-dosing reminders, monthly review of adherence data by health-care provider, and management of poor adherence; and patients in the control group received routine care (silent-mode monitor-measured adherence). Only the independent endpoints review committee who assessed endpoint data for some participants were masked to study group assignment. Patients were followed up (with sputum solid culture) at 12 and 18 months. The primary outcome was a composite of death, loss to follow-up, treatment failure, switch to multidrug-resistant tuberculosis treatment, or tuberculosis recurrence by 18 months from treatment start, analysed in the intention-to-treat population. Analysis accounted for study design with multiple imputation for the primary outcome. This trial is now complete and is registered with ISRCTN, 35812455. FINDINGS: Between Jan 26, 2017, and April 3, 2019, 15 257 patients were assessed for eligibility and 3074 were enrolled, 2686 (87%) of whom were included in the intention-to-treat population. 1909 (71%) of 2686 patients were male, 777 (29%) were female, and the median age was 44 years (IQR 29–58). By 18 months from treatment start, using multiple imputation for missing outcomes, 239 (16% [geometric mean of cluster-level proportion]) of 1388 patients in the control group and 224 (16%) of 1298 in the intervention group had a primary composite outcome event (289 [62%] of 463 events were loss to follow-up during treatment and 42 [9%] were tuberculosis recurrence). The intervention had no effect on risk of the primary composite outcome (adjusted risk ratio 1·01, 95% CI 0·73–1·40). INTERPRETATION: Our digital medication monitor intervention had no effect on unfavourable outcomes, which included loss to follow-up during treatment, tuberculosis recurrence, death, and treatment failure. There was a failure to change patient management following identification of treatment non-adherence at monthly reviews. A better understanding of adherence patterns and how they relate to poor outcomes, coupled with a more timely review of adherence data and improved implementation of differentiated care, may be required. FUNDING: Bill & Melinda Gates Foundation.