Exploring habit and context stability in the maintenance of adherence to medication for people living with long-term conditions

Adherence to prescribed medications is a global challenge, with average adherence to long-term medications estimated at 50%. The active ingredients of successful interventions to improve medication adherence remain elusive. Behavioural science research has identified habit as a mechanism of interest for the maintenance of health behaviours, including medication adherence. A habit is defined as a mental process which is automatically elicited by a stable contextual cue. Existing methods to measure habits are methodologically limited: many require participants to self-report on implicit mental associations and are subject to recall bias; others require time- and resource-intensive procedures to capture data and most fail to capture contextual stability of the behaviour. This is important because elicitation of behaviour in response to a stable contextual cue is a key feature in distinguishing habits from other forms of automatic behaviour. The overarching aim of this thesis is to explore the role, conceptualisation and measurement of habit and context stability in adherence to medication for people with long-term conditions. The thesis presents a series of critiques and research studies which explore habit, habitual behaviour, context stability and methods of measuring context stability in people taking medications for long-term conditions. A machine learning assisted review found that habit has not been uniformly conceptualised in line with contemporary habit theory in the medication adherence intervention literature, that behaviour change techniques used to form habits are multiple and heterogeneous and that habit formation is seldom measured. The review also highlighted that medical devices that collect time- and date-stamped adherence data are widely available in medication adherence research and this data could be used to objectively measure habit using the ‘behaviour frequency x context stability’ approach. This measure has been in use since 2005 but in self-report form. A subsequent critique of the ‘behaviour frequency x context stability’ measure of habit describes the need to develop an objective metric of context-stability which is demonstrably delineated from frequency of behaviour. The following research studies developed and evaluated the utility of an objective metric of context stability to meet this need. This was conducted using a sample of 607 people with Cystic Fibrosis whose data described the time of day at which they took prescribed nebulised medications over a 12-month follow-up period. This data was captured on electronic medical devices. The chosen metric of context stability was derived from a combination of simple summary statistics and non-linear methods which capture complex patterns in longitudinal data. Albeit with small effect sizes, the metric explained statistically significant variance in ongoing, future and the degree of maintained behaviour. These findings are consistent with performance of self-report metrics of habit and context stability used elsewhere in the literature. Future work can benefit from the review’s explicit clarification of common misconceptions about what a habit is and how it can be targeted, as well as theoretical critiques of common practice in habit research. The remaining chapters provide a starting point for future research to continue the development of the objective metric of context stability by examining the scope of its generalisability in other samples, population groups, study designs and behaviours.