Adherence trajectories and predictors of digital balance exercise for fall prevention in community-living older people

Long-term adherence to exercise delivered through digital health interventions is critical but remains a major challenge, particularly among older people. We pooled intervention-arm data from two randomised controlled trials in 511 community-living older people who participated in a home-based digital balance exercise programme for 52 weeks. Weekly adherence was automatically recorded and expressed as percentage of prescribed dose. Group-based trajectory modelling identified distinct trajectories of adherence and multinomial adaptive lasso regression examined cognitive, physical, demographic and psychological predictors of trajectory group membership. Eight adherence trajectories emerged: overachievers (3.9%), consistent adherers (8.2%), fluctuating adherers (23.1%), intermittent adherers (15.9%), late attrition (11.4%), steady decliners (18.4%), early attrition (14.9%), and non-adherers (4.3%). On average, participants achieved 56% of their prescribed dose. Executive function and standing balance predicted adherence trajectory group membership. Poor executive function was associated with high initial adherence followed by a steep decline, while poorer balance was associated with more sustained adherence. Demographic or general health variables did not significantly predict adherence trajectories. Adherence to digital exercise is a dynamic behaviour, shaped by behavioural capability rather than by age or comorbidities. These findings inform the design of adaptive digital interventions to promote long-term adherence to digital programmes among older people.