Design for adherence: a model for predicting and improving medication adherence

Medication adherence is often studied after a drug or delivery device has been developed, but what if we were able to predict levels of adherence for a particular drug delivery device during development?

Last year, I was in the US interviewing people with severe asthma and COPD about their experiences with inhalers. When asked which of their inhalers they liked best, almost all patients said that they preferred their rescue device (which is used to relieve symptoms during an asthma attack) to their preventer inhaler (used daily to reduce inflammation and prevent asthma attacks). In fact, many said they didn’t really see the point of the preventer. As one user, who had had severe asthma since childhood, said: “I’ve been feeling fine lately, so I don’t need it”.

In the light of evidence about the effectiveness of preventer inhalers this response is surprising. The preventer device reduces reliance on rescue inhalers and the frequency of trips to A&E for severe asthmatics. Yet, unlike the rescue device, the benefit of the preventer is not immediately obvious in terms of symptom relief. In fact, as with many preventative measures, the user has no way of knowing what would have happened if he or she hadn’t used it.

This is one of a number of problems relating to medication adherence, which can have a serious impact on both patient outcomes and the bottom line for the pharmaceuticals industry. Studies suggest that the proportion of prescriptions that are actually filled ranges from 45-85% [1]. If a drug is associated with an annual revenue of $5 billion, this implies lost revenue between $1 billion and $6 billion. While payers may appear to benefit from these savings, this is a short-term benefit. In the long term the costs they incur from non-adherence will outweigh any short-term savings [2].

In terms of patient outcomes, studies suggest that poor adherence to asthma and COPD medication is associated with increased mortality, with relative risk increasing up to 2.5 times at three years [3]. In addition to this, with increasing pressures on healthcare costs, payers are now increasingly pushing back on expensive drugs and asking questions about evidence of real-world effectiveness – in this context, poor adherence is increasingly seen as a failure of the drug or delivery device.

Understandably, efforts to improve adherence are ongoing and wide-ranging, yet they present their own challenges. While a typical usability study can be completed in a few weeks, it takes much longer to get results about adherence. For example, it can take twelve months to establish whether a device design feature is having a positive impact on adherence.

Having conducted an extensive review of adherence research papers and been involved in numerous design development projects where adherence was a concern, our view is that adherence interventions too often are haphazard and piecemeal. Rarely is a full appraisal made of what could be done. Instead, a few potential ideas are developed up and tested with users over a prolonged period of time – but this risks wasting time and money on an idea that is not optimal. The principle of our model is that a structured and informed view of adherence factors is the best way to prioritise interventions and maximise the chances of success.

For these reasons, we set ourselves the challenge of coming up with a way to predict adherence, and we’ve now developed a deterministic model for this purpose. The value of such a model includes:

The Design for Adherence (DfA) model we have developed is based on an analysis of factors relating to user capability (can the patient perform the task), opportunity (does the patient have the chance to perform the task) and motivation (does the patient want to perform the task) – which are three components of behaviour into which all aspects of adherence can be classified [4].

We also considered the impact on adherence of disease (e.g. does the disease cause loss of capability), patient population (e.g. are patients likely to come from a particular social group that might limit their opportunities), treatment factors (e.g. how are side effects likely to impact patient motivation), and device design (e.g. how easy is the device to use, i.e. what level of mental capability is required). Following a review of evidence from research in social and cognitive psychology, healthcare education, pharmacy as well as medicine, we came up with a question set that considers factors affecting capability, opportunity and motivation as they relate to disease factors, patient factors, treatment factors and device design factors.

Our model enables informed assessments of drug delivery systems. To date, we have tested the model with data from published research that compares adherence for different devices or device variants, and results so far are promising. Figure 1 shows a comparison of predictions with actual adherence data for one of the studies that we used to test our model. 

In summary, adherence is a serious and difficult problem. Economically, it affects payers in the long term as patient conditions worsen and more drastic and expensive interventions such as surgery and round-the-clock care become necessary; it affects industry as it causes the market for their products to shrink; and it affects patients as they are more likely to require time in hospital, have exacerbations that put additional demand on A&E departments and must spend more on health care premiums and any additional treatments or care that are not covered. Our aim is to come up with a fast and cost-effective way to design or select technologies that have the best chance of improving adherence. Achieving this goal will benefit not just patients and pharma but society as a whole.

Digital Health | Medical Devices | Drug Delivery | Healthcare Blogs | Healthcare Case Studies