Medication Adherence as a Digital Biomarker

Medication adherence is the extent to which a person takes medication as prescribed. A widely adopted taxonomy separates it into three components: initiation, whether the first dose is taken; implementation, how closely the dosing regimen is followed thereafter; and persistence, how long treatment continues before it stops.

Whether it belongs in a digital biomarker library is a fair question, and this page takes the position that it belongs with a caveat. A biomarker is an indicator of a biological process or response. Adherence is a behaviour, and measuring it is closer to process measurement than to physiological measurement.

It is included because digitally captured adherence behaves like the measures around it: it is collected continuously by a device, it is objective where the alternative is recall, and its variation explains variation in the outcomes other measures track. A trial that reports an effect size without measuring adherence has left the largest source of that effect size unobserved.

Several digital methods exist and they measure different things. Electronic packaging records the time of each opening, which is the closest to implementation but records opening rather than ingestion. Smart inhalers and injectors record actuation. App based systems prompt and log self report, which reintroduces the reporting problem they were meant to solve. Ingestible sensors confirm ingestion directly and carry their own acceptability and cost constraints.

Systematic review of electronic medication packaging devices has examined whether they actually improve adherence as well as measure it, and overviews of adherence measurement set out how the available methods compare, including pill counts, pharmacy refill records and drug level assays.

The unit of measurement is contested. Percentage of doses taken, percentage of days with correct dosing and time to discontinuation answer different questions and are not interchangeable.

In trials, adherence data supports per protocol analysis, explains apparent non response and protects against concluding that a treatment failed when it was not taken. In pragmatic and decentralised designs, where oversight is lower than at a site, this matters more rather than less.

In care delivery it supports targeted intervention. Knowing that someone is not taking a medication is only useful with the reason, which is why measurement is paired with the Beliefs about Medicines Questionnaire, the direct counterpart in this library. Non adherence driven by concern about harm requires a different response from non adherence driven by forgetting, and the questionnaire distinguishes them where a count cannot.

Self efficacy and condition specific self management instruments are reported alongside for the same reason.

Every method measures a proxy for ingestion rather than ingestion itself, except ingestible sensors. Packaging records opening, apps record reporting, refills record collection.

Measurement also changes behaviour. People who know their adherence is monitored adhere better, which is useful clinically and a problem scientifically, because trial adherence then overstates what will happen in practice.

Digital adherence systems are also unevenly accessible, requiring devices, connectivity and confidence with technology, and the populations with the lowest adherence often have the least access. A monitoring approach that works best for people who were already adhering measures the problem where it is smallest.

• Vrijens B, et al. A new taxonomy for describing and defining adherence to medications. Br J Clin Pharmacol. 2012. pubmed.ncbi.nlm.nih.gov
• Horne R, Weinman J. Patients' beliefs about prescribed medicines and their role in adherence to treatment in chronic physical illness. J Psychosom Res. 1999. pubmed.ncbi.nlm.nih.gov
• Checchi KD, et al. Electronic medication packaging devices and medication adherence: a systematic review. JAMA. 2014. pubmed.ncbi.nlm.nih.gov
• Anghel LA, et al. An overview of the common methods used to measure treatment adherence. Med Pharm Rep. 2019. pubmed.ncbi.nlm.nih.gov