Music-visual Guided Physical Activity Promotion for Adults With Intellectual Disabilities

The specific objectives of this project are to (1) evaluate the effectiveness of this intervention in improving the PA level of these adults, enhancing their self-efficacy and interest in performing PA and improving their functional exercise capacity by conducting a cluster-randomized controlled trial, and (2) examine the acceptability of this intervention among the participants and the staff of residential care facilities.

Participant recruitment The person-in-charge of residential care facilities and the non-governmental organizations that serve adults with ID, including the Hong Chi Association who had collaborated with us in our pilot project, will be approached by the research team to seek their collaboration and support in recruiting participants and implementing the project. The research team will organize promotion talks and distribute promotion leaflets of the project to potential participants and their families or legal guardians. To ensure safety of the participants, PA pre-participation health screening will be performed in accordance with the guidelines of American College Sport Medicine. Medical clearance/approval to participate MVgPA will be sought if they have known cardiovascular, metabolic and/or renal disease, or signs and symptoms suggesting these diseases (as detailed in Appendix 1). An information sheet related to the MVgPA, including the heart rates during the PA and type of exercise, will be provided to the physicians.

Sample size calculation A similar study for adults with ID in community residences showed an effect size of 0.35 on PA level change. Assuming a similar effect size on PA level change in our proposed trial, a sample size of 129 participants per each of the control and intervention groups will provide the study with 80% power at 2-sided 5% level of significance, as estimated using the power analysis software PASS 13.0 (NCSS, Kaysville, USA). Furthermore, to account for potential reduction in statistical efficiency due to randomization by residential care facilities (clusters) instead of individual participants, a variance inflation factor, called design effect, will be applied to impose to the estimated sample size. The design effect is given by 1 + (m-1)*ICC, where m=average cluster size and ICC=intracluster correlation coefficient of the underlying outcome. We anticipate that the outcomes of the participants recruited from the same residential care facility are unlikely correlated with one another, particularly for the PA outcome. In this regard, the ICC would be small. A synthesis study revealed that ICC tends to be small in primary care research with a median of 0.005 and inter-quartile range of 0.000 to 0.021. Allowing for a conservative ICC of 0.01 in our proposed study, at least 16 participants per each of the 20 residential care facility are therefore required. Furthermore, to allow up to 20% attrition rate, at least 20 participants per facility will be recruited.

Cluster randomization Randomization will be performed at the cluster level (residential care facility) instead of individual participants to avoid contamination between the intervention and control groups. The facilities will be randomized to either intervention or control group in a 1:1 ratio. Eligible participants recruited from the same facility will be allocated to either intervention or control groups accordingly. Group allocation will be concealed from outcome assessors and made according to the random group allocation sequence generated by an independent statistician and the sequence of entry into the study.

Data analysis By considering the potential design effect of the randomization conducted in cluster level instead of individual participants, the outcome analysis will be performed on the basis of a three-level mixed effects model accounting for inter-correlation among individuals within the same cluster as well as intra-correlation over time within an individual and variations between individuals. This type of model can account for intra-correlated clustered and repeated measures data and produce unbiased estimates even in the presence of missing data, provided that the data are missing at random. Mixed-effects model will be used to compare the differential changes in each outcome variables (PA level, PA self-efficacy and interest, and exercise capacity) across time (T0, T1, T2) between the two groups. Positive significant group-by-T1 and group-by-T2 interaction-terms in the mixed-effects models indicate significantly greater increment of intervention group in the underlying outcome than the control group at the respective time-points with respect to T0.