Effectiveness of Tai Ji Quan vs Multimodal and Stretching Exercise Interventions for Reducing Injurious Falls in Older Adults at High Risk of Falling

This follow-up analysis of a randomized clinical trial compares the effectiveness of tai ji quan, multimodal exercises, and stretching exercise interventions for older adults at high risk of falling.

Post-Intervention  As part of the original research protocol, upon completing the active intervention, 27 participants were followed up for a duration of 6 months. During the follow-up, participants were 28 given no specific instructions on exercise but they were encouraged to continue to exercise in 29 any form. Participants' levels of weekly physical activity or exercise behaviors were monitored 30 via a home-exercise log in which they were asked to record information on type, frequency, and

45
In addition, information on physical activity (exercise behaviors) was collected at 3-month 46 intervals by research assistants who were blinded to intervention group allocation.

48
Data analysis 49 The original study was powered, with a sample size of 666 (including a planned 15% 50 attrition rate), to detect a 35% reduction in fall incidence rate across a 6-month active 51 intervention period (a respective IRR of 0.65) between either of the two interventions (TJQMBB, 52 multimodal exercise) relative to stretching exercise. No a priori sample size calculations for the 53 6-month post-intervention follow-up study were conducted. However, on the basis of 54 preliminary estimates obtained from a prior trial, conservative estimates of a 30% reduction in 55 moderate injurious falls and a 20% reduction in serious injurious falls for TJQMBB and 56 multimodal exercise relative to stretching exercise were established. No specific predictions on 57 reductions in injurious falls between TJQMBB and multimodal exercise interventions were 58 made.

59
Analysis of variance was planned to examine between-group differences in mean change 60 from baseline to 12 months in primary and secondary outcomes. We also present the rate of falls 61 over 12 months, calculated using the number of falls observed from each participant as the 62 numerator and their follow-up time as the denominator. In our primary analyses comparing the 63 number of injurious falls, we used a priori specified negative binomial regression models to 64 estimate absolute differences in incidence rate ratios (IRRs) from baseline to 12 months with 65 their corresponding 95% confidence intervals (CIs) comparing TJQMBB and multimodal 66 exercise with stretching exercise and TJQMBB and multimodal exercise. As part of planned 67 group comparisons, we also conducted additional analyses to test whether there was a difference 68 between the injurious fall rates at the end of the 6-month active intervention and at the 6-month 69 post-intervention follow-up between the groups.

70
IRRs were adjusted for important covariates (age, sex, number of falls 6 months prior to 71 the start of intervention, health status, and levels of weekly physical activity at home ascertained 72 at 12 months) and without these covariates. All analyses were based on intent-to-treat (i.e., all 73 participants who were enrolled and randomly allocated to interventions were analyzed in the 74 groups to which they were randomized) and adjusted for the variable follow-up times of 75 participants. Significance testing was 2-sided. Results were considered statistically significant if 76 P < 0.05. No adjustment was made on multiple comparisons. Analyses were conducted using 77 SPSS version 23 (IBM Corp, Armonk, NY).