Progressive task-oriented extended reality in a cognitive stimulation programme for older adults with mild or early cognitive impairment: a pragmatic clinical trial of patient adaptation, usability and cognitive-functional outcomes

Objective Extended reality (XR) can incorporate cognitive training into sensory, cognitive-motor and functionally oriented tasks, but delivery depends on matching task and interaction demands to participants. We examined how a progressively introduced XR component could be integrated and adapted within routine cognitive stimulation for older adults with mild or early cognitive impairment, with exploratory comparison of cognitive and functional change against computerised cognitive training alone. Methods This single-centre, non-randomised controlled trial included 70 rehabilitation courses from 59 individuals. Both groups received supervised computerised cognitive training for 12 weeks; 45 courses allocated 15–20 min of scheduled sessions to XR, while 25 received computerised cognitive training alone. XR progressed from familiarisation and guided 360° environments to mixed-reality cognitive-motor tasks, immersive everyday-task virtual reality and controller-based cognitive exercises. Exposure, task and interface difficulties, adaptations, adverse symptoms and usability were recorded. Exploratory comparisons of the trial’s prespecified primary cognitive and functional outcomes used generalised estimating equations (GEE) with participant-level clustering and Holm adjustment. Results Across 45 XR-supported courses, 729 participant-exposures were recorded and 715 (98.1%) were fully delivered. Task understanding or execution difficulties affected 25/43 controller-based XR courses, compared with 5/40 mixed-reality and 6/43 immersive everyday-task virtual reality courses. Interaction or interface difficulties occurred only with controller-based XR (4/43, 9.3%). Patient-perceived usability was 68.50 ± 16.57/100 and headset comfort and tolerability 81.99 ± 14.54/100. Professional-perceived usability was 85/100 (n = 4). Adverse symptoms occurred in 30/729 exposures (4.1%), affecting 14/45 XR participants (31.1%), with no serious adverse events or symptom-related discontinuations. In 38 XR-supported and 23 non-XR pre-post courses, exploratory adjusted between-group differences remained after Holm adjustment for the Mini-Mental State Examination (+1.62, p = 0.013) and Montreal Cognitive Assessment (+1.88, p = 0.030), but not Lawton-Brody or other cognitive outcomes. Conclusion Progressively introduced, clinician-supervised XR was repeatedly deliverable within routine cognitive stimulation, with adverse symptoms uncommon at the exposure level and no serious events. Rather than treating XR as a uniform intervention, these findings support selecting and adapting configurations according to their immersion, interaction and task demands, participant capabilities and practical delivery conditions. Exploratory MMSE and MoCA between-group differences warrant further controlled study but do not establish an XR-specific treatment effect. Clinical Trial Registration NCT06977815, https://clinicaltrials.gov/study/NCT06977815 .