Yoga Program for Improving the Quality of Life in Couples Coping With Cancer

This analysis will be imbedded in a linear mixed-effects (alternatively known as the multilevel) modeling (MLM) analysis, with repeated measures over time (i.e., at end of treatment, 3- and 6-month follow-ups). Potential interaction terms will be included and tested, in particular, between the intervention and time. The repeated measures correlation structure will be selected using the Bayesian information criterion (BIC). The difference between dyadic yoga (DY) and waitlist control (WLC) at 3MFU is considered the primary and confirmatory comparison and will be tested at a two-sided 0.025 significance level. Additionally, on an exploratory basis, will conduct an omnibus test on the overall differences in 6MWT across the three groups at 3-month follow-up, as well as pairwise comparisons between the patient yoga (PY) and WLC and between DY and PY. Exploratory comparisons will be primarily interpreted based on clinically significant difference, which is 70m for the study population.

Analyses will be similar to those for the primary outcome, except that will use a dyadic model assuming additional dependence in outcomes between patient and partner outcomes. Repeated measures over time will include at the end of treatment, 1-, 3- and 6-month follow-ups. Will examine the individual SF-36 domains as well as the physical (PCS) and mental component summaries (MCS). Patient and partner overall quality of life (QOL) will be measured with the Medical Outcomes Study 36-item short-form survey (SF-36) assessing 8 distinct domains: physical functioning, physical impediments to role functioning, pain, general health perceptions, vitality, social functioning, emotional impediments to role functioning, and mental health. The SF-36 will take up to 7 minutes to complete and will be assessed at each follow-up time point (time point 1 [T1] - time point 6 [T6]).

To assess the differences between groups in patient healthcare utilization will test the patient emergency department (ED) visits, hospital admissions (HA) admissions and feeding tube (FT) insertions. Will use a Chi-squared or Fisher's exact test, as appropriate, to compare the probability of healthcare services using a binary outcome (yes/no) between groups at a two-sided 0.025 significance level. Additional analyses will examine the number of times a patient visited the ED during the specified period. As a follow-up analysis, will examine group differences with logistic regression using randomization factors as model covariates (i.e., age, sex, ethnicity, stage, smoking status) as these factors may be associated with the study outcomes. Productivity loss will be measured by the Work Productivity and Activity Impairment General Health (WPAI-GH) assessing presence at the work place. Both patients and partners will complete items pertaining to productivity loss due to their health.

Sexual problems will be assessed with Majerovitz and Revenson's 6-item measure of sexual problems, which have successfully been used in our previous research in couples coping with cancer. Emotional intimacy will be measures with the validated and frequently used 6-item Personal Assessment of Intimacy in Relationships (PAIR) instrument, which has been validated in cancer. Will explore if the DY intervention improves physical and emotional intimacy following the procedures as described for the QOL (secondary) outcome. Will control for values of outcomes at baseline, as well as the randomization factors, in all the above analyses to improve power of the comparisons.

Will examine if treatment efficacy on outcomes measured at 3MFU (T4) are achieved through improved symptoms and symptom management skills at the end of RT (T2) using linear regression analysis in both the intervention-mediator: (a) and intervention plus (+) mediator-outcome (b) paths. Ninety-five percent Bootstrap confidence intervals (CIs) will be constructed to assess the significance of the indirect (or mediation) effects defined using the product-of-coefficient approach. Will also use multilevel modeling (MLM) to test for interaction effects between intervention and the priori moderators, baseline depressive symptoms and dyadic adjustment. Significant interactions will be probed using simple slope analyses following the procedure developed for multilevel modeling by Preacher, Curran, and Bauer. These analyses will use a two-sided 0.05 significance level.