The aim of this thesis was to inform the development of a digital precision intervention to promote physical activity among older adults (aged 60+ years) of low socioeconomic status (SES) in the United Kingdom (UK). The first empirical chapter, Chapter Two showed that pre-existing socioeconomic inequalities in older adults’ physical activity levels based on education, occupational class, and wealth, persisted during the coronavirus disease (COVID-19) pandemic. Despite the social gradient in physical activity, a prospective cohort study of 3,720 older adults who participated in wave 9 (2018/2019) of the main English Longitudinal Study of Ageing (ELSA) survey and wave 2 of the ELSA COVID-19 sub-study (November/December 2020) found inconclusive evidence regarding any socioeconomic differences in physical activity change between the two survey periods. Subsequently, using latent class analysis and latent transition analysis on data from wave 9 (2018/2019) of the main ELSA survey and wave 1 of the ELSA COVID-19 sub-study (June/July 2020), Chapter Three identified distinct clusters of older adults, characterised by increasing breadth of internet use (a proxy for digital engagement), before and during the COVID-19 pandemic. Although few people transitioned between latent classes over time, education, occupational class, and wealth were positively associated with membership in clusters reflecting greater diversity of online activities. Chapter Four is the first systematic review with meta-analysis to explore statistical associations of modifiable correlates with physical activity, separately for UK community-dwelling older adults of low versus high SES. Of the seventy-seven eligible studies, fifty-one were included in at least one random effects meta-analysis, with results structured according to the Capability, Opportunity, Motivation, and Behaviour (COM-B) model. There was no evidence of socioeconomic disparities in the associations of the assessed correlates with physical activity, apart from the presence of built facilities, walking and cycling infrastructure, and less smoking, which were positively associated with physical activity only in low-SES older adults. Fewer low-SES individuals reported good social participation and perceived general health, suggesting that inequalities in the prevalence of these correlates (i.e., low-SES older adults tended to report poorer health, which may limit physical activity) rather than the ways in which they influence behaviour may explain gaps in physical activity. To provide a more in-depth understanding of how these factors influence older adults’ behaviour, Chapter Five synthesised qualitative literature on the modifiable enablers of, and barriers to, physical activity based on low-SES older adults’ perspectives. Thirty studies were included, identifying a wide range of interacting factors motivating or hindering physical activity engagement across socioecological and COM-B domains. The five studies examining socioeconomic inequalities in physical activity enablers and barriers pointed predominantly to disparities in the physical or built environment. Turning our attention to innovative mobile health technologies, Chapter Six reported the feasibility and acceptability of a seven-day, smartphone-based, activity-triggered Ecological Momentary Assessment (EMA) study among low-income older adults (N = 39, 76% earning below £25,000/year). Participants were compliant with the EMA protocol on 84% of occasions. Responses to open-ended questions revealed that the most enjoyable aspects of taking part were making a worthwhile contribution to science, becoming more self-aware of their feelings and/or activities, and meeting the research team. The least enjoyable aspects included the frequency or timing of EMA prompts, the interruption of activities to complete an EMA survey, and limited or inadequate response options for certain EMA items. The final empirical study, Chapter Seven (a separate analysis of the sample described in Chapter Six), explored acute modifiable correlates of physical activity. In multilevel models restricted to inactivity-triggered and timeout prompts (i.e., EMA prompts triggered when participants fell below an inactivity threshold or when neither the activity or inactivity threshold was surpassed but two hours had elapsed between prompts), reporting higher positive affect or lower negative affect than one’s usual level was associated with fewer steps in the subsequent 15 minutes, whereas reporting higher intentions than one’s usual level was associated with more steps in the subsequent 30 minutes. Models restricted to activity-triggered prompts found a positive within-person association of self-efficacy with physical activity over the ensuing 15 and 30 minutes. Prior physical activity was unrelated to current affective and physical feeling states or behavioural cognitions. Chapter Eight provided a summary of the study findings, discussed implications for academia, practice, and policy, outlined novel contributions to the field, and appraised the strengths and limitations of this thesis. This programme of research enhanced our understanding of the relationships between SES, physical activity, and digital engagement among older adults, as well as potential underlying mechanisms, using the socioecological and COM-B models. Collectively, these findings could help to inform the content and delivery of a personalised, digital just-in-time adaptive intervention. Moreover, this thesis has identified useful strategies for recruiting, engaging, and retaining low-SES older adults in digital health research.
| Date of Award | 27 Jan 2025 |
|---|
| Original language | English |
|---|
| Awarding Institution | |
|---|
| Supervisor | Max Western (Supervisor), Nicholas Townsend (Supervisor), Mark Kelson (Supervisor) & Charlie Foster (Supervisor) |
|---|