Health-social Partnership Intervention Programme for Community-dwelling Older Adults: a Randomized Controlled Trial

A staff of community center in service area who will not take part in the study will be involved in drawing a list of potential participants from the accessible population. Simple random sampling will be adopted to recruit participants. Those who meet the inclusion and exclusion criteria and agree to participate in the research study will be assigned to intervention group or customary care group according to the result of randomization.

Sample size calculation was based on power analysis. Power analysis adopts a hypothesis-testing method to determine sample size according to several parameters which include pre-specified significance level, desired power level and expected effect size. Assuming a two-tailed alpha of 0.05, a probability of 0.2 for beta error (80% power), and an effect size of 0.28 after calculating with respect to the same primary outcome measure (self-efficacy) from results of two previous researches which provide home visit and telephone follow up to older adults, 200 subjects per group are required. With reference to the 10% to 15% attrition reported in previous community-dwelling elderly program, assume there is a 20% drop out rate in this study, the total sample size needed is 240 subjects per group i.e. a total of 480 subjects.

In order to assure that the numbers of both groups are equal, a block randomization method will be used. Before randomization, a project team member who will not be involved in the subject recruitment and data collection will prepare a block randomization list with 240 sets of numbers which is either 0 (customary care group) or 1 (intervention group) by using a computer software Research Randomizer (http://www.randomizer.org/). These 240 sets of numbers will be printed out separately and sealed in each envelope. Another project team member, who successfully recruits a subject, will unfold an envelope by sequence after the enrolled participants finished all the baseline assessment and ready for interventions. The number written in the envelope will represent the group of that particular subject.

The group assignments will keep blinded to participants during the whole study period. However, the nurse case manager who will provide care to the participants will know the group allocation. Therefore, this study is a single-blinded design. In order to minimize bias, the nurse case manager will not participate in the data collection procedure.

Descriptive analysis will be calculated using mean and standard deviation for continuous variables, percentage and frequency for categorical variables, and median and quartile range when the continuous variables are not normally distributed. Independent t-tests, Mann-Whitney U test, chi-square test or Fisher's exact test will be used to compare any differences in socio-demographic data or outcome variables between groups. Significant results are indicated if p-value of all statistical analysis is less than 0.05 for two-tailed test. To determine the differences or changes between study group and control group (between group effects), within group effects (time), and the interaction effects (group x time), two way repeated measures Analysis of variances (ANOVA) or co-variances (ANCOVA) will be applied when assumptions for parametric test are fulfilled. When the data cannot fulfill the assumptions of parametric test, Mann-Whitney U test will be applied to compare the differences between groups at each time interval while Friedman test will be used to determine the time effects within each group.