New approaches to studying adherence

We know poor adherence leads to poor disease control and increased healthcare costs – but are we doing enough to help patients optimise their therapy? Better device design is one obvious solution – but it’s not enough on its own. A recent study looked at the impact of multiple-dose versus single-dose inhaler devices on treatment via persistence. The study indicated that, in fact, inhaler type seems to make no difference to adherence. So where are we going wrong? The answer could lie much closer to home – perhaps it’s time to reassess how much we really understand the patient. And particularly those patient groups who typically don’t shout loud enough about their concerns – e.g. children and the elderly.

Focus on specific demographics

Asking children or the elderly what they look for in a drug delivery device rarely leads to success. Similarly, monitoring them in user trials equipped with the usual myriad of cameras, microphones and one-way glass hiding review teams often causes the Hawthorne effect – namely, that our behaviour changes when we know we’re being watched.

What if we could leverage development partners and access the latest remote observation technologies, so that we can observe users in their homes without affecting their behaviour? What if we could also quantify their exact behaviour and usage interactions? With this quantitative truth, surely we’d be best placed to make informed product development decisions, where we could aim innovations squarely at those user sequences that a particular demographic most struggles with?

User trial results

Results from our recent at-home monitoring study of the use of a product over a one-month period have shown a significant drop in use after just two weeks. While the logging technology was hidden from view, users were informed of the capabilities of this approach – and yet still we saw clear drop-off in adherence during the study. We used miniature bolt-on logging ‘pucks’ which are able to capture a range of user behaviours, without impacting the very use they’re assessing. This trial confirmed two things – firstly, even though the respondents insisted they had continued to use the device each day, their own usage data disputed this. Secondly, after an initial increase in usage, soon after week two many respondents clearly found the immediate gains to be insufficient – and usage quickly dropped off.

This clearly shows there is a defined point in time when users are about to reduce their adherence. If this inflexion point could be known, relative to when a patient started to take a treatment, additional measures could be taken – e.g. personalised incentives – to attempt to bridge the gap between initial interest and longer-term verifiable gains/results.
If we use technology to understand the behaviour of patients – not just in the market but during the development of a product – then we may gain access to previously unknown information that can be used to help patients achieve the clinical effect of their medication.

A certain amount of success can be achieved purely through patient engagement software to create the optimal carrot/stick incentivisation regime. It remains a difficult path to tread, however, as the aspirations of the sector continue to mature beyond monitoring to include assisting in treatment or diagnosis.

What is clear is that the base technology already exists to enable designers to better understand the needs of patients – and, in doing so, it empowers them to make the required design changes to a drug delivery device to provide a beneficial user experience. Translating raw usage data into quantifiable insights is hard, requiring large high-quality datasets, a sound grasp of the fundamentals of classic data science and (then, and only then) vast amounts of computer power to drive artificial intelligence and machine learning.

Uri Baruch is head of drug delivery at technology and product design firm Cambridge Design Partnership