The 12-week progressive home-based exercise and Tai-chi snacking programme had significant positive effects on physical function, with improvements in the SPPB S&B domains, 5 reps STS, 60s STS, and ability to maintain standing balance on one leg. These improvements were observed in week 4 and sustained until the 12-week intervention period was over.

Significant differences in the SPPB scores between intervention and control groups were observed at all follow-up timepoints in participants who underwent laboratory testing and completed the full battery. For context, previous research has indicated that even small changes in SPPB scores of 0.4–1.5 are clinically meaningful in pre-frail older adults who initially scored less than 9 on the SPPB (25, 26). We also found significant improvements on 5 reps STS, mobility (TUG), and standing on one leg balance tests and in week 4 and week 8 amongst these participants. Most participants reached the highest level of exercise difficulty by week 8, potentially leading to training plateau.

These findings build upon our previous pilot work, indicating that engagement with and impact of a home-based exercise snacking routine extends beyond a 4-week period, and can lead to potentially meaningful changes in physical function (16–18). Nevertheless, while participants in the intervention group had better performance in standing on right leg at week 4, no differences were found afterwards. Qualitative interviews and logbook notes may explain this consequence; a few participants reported that they habitually put their weight on the same side during exercises. Future instructions should highlight weight balancing and shifting.

In this study, we observed that progressive home-based exercise and Tai-chi snacking improved lower extremity strength, balance, and mobility over time in pre-frail older adults. Prior studies have consistently demonstrated the positive impact of home-based strength and balance exercises on muscle function, balance, and mobility in older adults (27–29). The results align with recent reviews highlighting balance and muscle functions benefits among healthy older adults (30) and improved leg muscle strength and mobility for individuals over 60 (31) through home-based exercises. Comparatively, our study suggests that 10-minutes of daily exercise and tai-chi snacking can be as or more effective in the short term as more complex behavioural interventions (32, 33). While encouraging, it should be noted that the present study had some withdrawals, and its duration was shorter than these aforementioned complex interventions. Verifying the sustained benefit of exercise snacking would be required to determine long-term equivalency.

Our qualitative findings indicate that the acceptability and ease of fitting exercise snacks into participant’s daily routines were suitable for older populations. Nevertheless, experiences varied among participants, with both exercise and Tai-chi snacking being physically challenging, particularly single-leg weight-bearing and air squat movements. Exercise snacking was seen as enhancing strength and stamina, while Tai-chi snacking was perceived as beneficial for balance, flexibility, and ankle mobility. These findings align with prior qualitative studies in which older participants believed that strength-based exercise snacking could improve muscle functions and alleviate joint pain (34–36), whereas Tai-chi training could enhance their balance, mobility, and full body relaxation (17, 37–39). Regarding acceptability, Tai-chi snacking was perceived as more time-consuming and cognitively demanding yet was seen as gentler, more mindful and elegant which participants enjoyed, while exercise snacking was straightforward and repetitive, allowing for more automatic performance. Similarly, participants in previous studies reported Tai-chi to be beautiful and refreshing, in contrast to repetitive exercises (37, 38). However, we could not isolate the effectiveness of either approach in this study, as participants did both. Exploring different variations of this programme may determine which exercise approach is more effective for specific functional outcomes or promoting better engagement, creating opportunity for further tailoring of the intervention to meet individual needs.

As we explored the functional outcomes, valuable insights into the timing of our intervention’s benefits were uncovered. The strength and balance scores of SPPB consistently favoured the intervention group from week 4 to week 12, reaffirming the sustained effects of continued, 10-minutes daily practice. Additionally, the SPPB strength item (5 reps STS) showed consistently improvement in the intervention group, albeit with a slight decrease over time. Single leg standing balance initially improved but later stabilised. In the lab subset participants, SPPB scores were consistently higher in the intervention group across all time points after baseline, along with positive effects on balance, gait speed, and 5 reps STS. The varying improvement rates may be due to participants progressing through exercise levels at different paces. While previous studies have reported post-intervention benefits (40–42), the specific timing of adaptations and progression remains less explored. While our measurements extended to 12 weeks, representing progression from our previous work, the question of longer-term (i.e., years) effects remains unanswered. As participants age, further investigation into these sustained changes is important for understanding the lasting effects. Moreover, we observed little impact of the intervention on exercise cognition process variables or health and being outcomes, which would be worth further scrutinising in larger and longer studies.

Programme adherence was high for those who completed the intervention. Studies have identified self-efficacy and outcome expectancy as key factors for exercise participation (43, 44) although we observed no significant self-reported change in these constructs via questionnaire. However, interview with completers indicated increased self-efficacy in physical activities and exercise, boosted by a sense of accomplishment and increased confidence in performing daily activities. These findings align with research showing that enjoyment, satisfaction, and perceived effectiveness have strong correlations to exercise participation (45, 46). However, a few participants found the programme too simple and desired more variations, although this presumably depended on participants physical activity levels since some participants found the movements in level 2 and 3 to be too difficult and complicated to perform based on their health conditions, and it also depended on the specific movements that participants liked. It is important to consider the programme’s intensity and variability to maintain the adherence (17, 47).

Regarding intervention dropout, most participants withdrew during the first month, reporting busyness and lack of time as primary reasons, consistent with recent studies (48, 49). As the exercise snacking concept and protocol is explicitly designed to overcome participatory barriers of this very nature, one might assume that the provision of more motivational and habit-forming support could be warranted to maximise initial engagement in the intervention. Future research could, for example, explore the use of cognitive strategies such as ‘Reframing’ exercise (50), or behaviour change techniques like motivational interviewing, setting graded tasks, instructing on when and where to perform the behaviour and explicit prompting of self-monitoring of outcome (i.e., physical function) that seem pertinent for increasing exercise self-efficacy in older adults (51).

Strengths of the study include the quantitative and qualitative mixed-method study design, the randomised design, the novel, safe and implementable remote delivered home-based exercise programme (without serious adverse events), and the thorough self-reported adherence and progression on the exercise logs. This study also discovered the feasibility of unsupervised home-based exercise programme compared with previous home-based exercise intervention studies which were mainly with supervision (31, 52).

Nevertheless, this study has several limitations. Firstly, many participants were female and highly educated, limiting the generalisability of our findings to populations in lower socioeconomic conditions. It is imperative to find strategies that encourage active participation and retention in male older adults who would benefit from muscle strengthening exercise. Digital literacy is also an issue for many older adults, which the recruitment and digital study procedures did not account for. Secondly, the intervention group had fewer individuals living alone and more married participants which may be the potential impact of this discrepancy on the study outcomes. While our primary focus did not specifically aim to explore differences between those living alone and those married, we acknowledge the potential impact of this discrepancy on the study outcomes. It is worth noting that couples assigned to the intervention group may have unique motivations, and the crucial role of social support for older adult physical activity participation, particularly from family members, is well documented (53). That said, our sensitivity analysis revealed little impact of living status on the physical function results.

Our qualitative interviews revealed that not all couples executed the programmes together, indicating variability in adherence, but larger trials of exercise snacking are warranted to verify the influence of living status. The validity of remote assessments of physical function remains unknown, although conducting functional assessments remotely has been deemed feasible and safe (16). Future studies can explore the validity of these remote assessments. Moreover, due to limitations in remote assessment and concerns about accuracy, we did not examine the SPPB gait speed domain via online video calling sessions. However, a study demonstrated that chair stand and balance test could be substitutes if mobility/gait speed performance assessments are challenging (54). Nonetheless, it would be ideal to find solutions and assess the full battery of physical function remotely in future research. Thirdly, the high dropout rate is an important factor to consider when interpreting our results. A baseline characteristic comparison between withdrawers and completers showed no significant differences (Supplementary File A1), suggesting that potential biases related to baseline characteristics were minimised. Given that we did not achieve the desired sample size, conducting further research in larger studies implementing strategies to reduce attrition is essential to verify the findings and ensure the presently reported effects and acceptability are not an artefact of a survivor bias.

A further consideration on the function outcome is the order of the in-person lab-based and remote assessment sessions. All lab-based participants completed remote sessions first, potentially leading to improved lab-based testing performance due to familiarity. This is particularly relevant to control group participants in the lab-based subset, possibly diluting intervention effects. Similarly, we acknowledged that monthly functional tests for the control group might impact functional outcomes via familiarisation, potentially leading to small improvementsbeyond their usual care. In addition, our exercise programme includes both upper and lower body movements owing to PPI feedback and previous qualitative work (17), yet we did not examine upper body functions. Furthermore, we did not investigate the maintenance of physical function improvements after the intervention ceased. Future studies should include upper body functional tests and follow-up assessments beyond programme’s end to explore any maintenance effect of physical function, exercise cognitions and health.