50 | Chapter 3 project assistant will verify the data entries and check the case report forms in the data management system after all trainings and assessments are finished. If data are missing, the assistant will check the logbook. Inertial sensor data will be uploaded to the server of the manufacturer of the accelerometers (McRoberts B.V.) for processing of activity classification of physical activity (walking, stairs walking, standing up, shuffling, cycling) and sedentary behaviour (lying, sitting). This classification will be downloaded from the manufacturer’s server and will be saved together with the raw data on a password protected external hard disk. Sample size To obtain a reduction of 50% in the number of falls between the intervention and control group, a minimum of 106 persons are required per group, at a power of 0.80, beta of 0.02 and alpha of 0.05. Taking into account a dropout rate of 20%, the total required sample size is 256 participants. Hence, we expect that about 16 In Balance intervention groups of 8 participants each, are needed to include the required sample size of 128 participants in both the intervention and control group. Statistical analysis Data will be analysed using SPSS (SPSS Inc., Chicago, IL, USA), RStudio (Version 1.3.1073) and MATLAB (version R2021a; MathWorks Inc., Natick, MA, USA). All analyses will be performed according to the intention-to-treat principle. Numbers and reasons for drop-out and for missing data on the primary outcome will be provided. Demographic characteristics Data at baseline and post-intervention will be described using means and standard deviations for normally distributed continuous variables, medians and interquartile ranges for non-normally distributed continuous variables, and numbers and percentages for non-continuous variables. Primary and secondary outcomes The effectiveness of the In Balance intervention in comparison with a control group will be analysed with multilevel mixed model regression analyses for both primary and secondary outcomes. Three hierarchical levels will be included in the mixed models; therapist, participant and time. If necessary, analyses will be adjusted for confounders and stratified for the presence of potential effect modifiers. The primary effect in these analyses is described by the coefficient of the time treatment interaction term. To identify possible differences in intervention effects between non-frail and pre-frail respondents, an a priori subgroup analysis will be performed, stratified for frailty level (non- and pre-frail).
RkJQdWJsaXNoZXIy MjY0ODMw