JMIR Publications

ABSTRACT

Meaningful engagement in mobile health (mHealth) means using a tool enough to achieve its intended health outcome, not simply using it more. Yet studies of medication-related mHealth often assess adherence through self-report or dose confirmations entered in the very app being evaluated. In the latter case, engagement and adherence are partly inferred from the same record. We propose a therapeutic drug monitoring (TDM) framework that links engagement at 3 clinical touchpoints to proximal pharmacokinetic indicators and distal outcomes, with 3 tiers of data requirements. To examine what minimal clinical records can reveal, we analyzed routine tacrolimus monitoring data from a tertiary hospital’s kidney transplant outpatient clinic in China (2017-2021; 606 patients and 14,198 stable-phase measurements). Concentrations and sampling dates allowed us to characterize longitudinal variability: 22.6% (137/606) of patients had a Medication Level Variability Index (MLVI) above 2.0. Of patient-years classified as highly variable, 62.5% were followed by a low-variability year, compared with 69.0% expected under independence. Individual transient excursions, however, were difficult to distinguish from random fluctuation; confirming the validity of a particular trough requires information about dosing and sampling times. Although interpreted results circulate electronically among professionals at this hospital, patients’ digital access is limited to the unannotated value, alongside existing paper reports and face-to-face communication. We identify engagement-sensitive gaps before blood draws, after results are released, and between visits. On this basis, we question 4 common assumptions: that more engagement is always better, that engagement can follow the calendar rather than the clinical cycle, that the patient is the only one engaging, and that professionals are mere background. We argue that engagement should be assessed around clinical events and benchmarked against independently measured drug levels, and we outline implications for measurement, prospective testing, and implementation.