The analysis of primary and secondary outcome measurementswill include all randomized patients. To beable to use all patient data, the mixed models approachwill be adopted with individuals at level one and centerat level two. The analysis will demonstrate any improvementsat follow-up, comparing the intervention andcontrol group. In linear mixed models the variable groupis considered main, controlling for baseline value, centerand gender. To judge the quality of the model we willanalyze the residuals. A difference at the P<0.05 levelwill be considered significant in all outcome measuresand analyses. For each outcome the results will bereported as a summary of the outcome in each grouppresented as mean and SD, together with effect size. Fortreatment effect, 95% confidence intervals (CI) and Pvalues will be presented. For each group, informationabout the number of participants included in each separateanalysis and whether theses analyses were by originalassigned groups will be given. Correlation betweenchanges in physical variables and changes in health variableswill be measured by Pearson correlation. A prespecifiedsubgroup analysis of participants randomizedto the intervention group will be performed to investigatedifferences in BORG CR10 rating between elasticresistance exercises to evaluate the strength relationshipbetween involved muscles/movements. Our researchgroup has previously found relatively lower thigh musclestrength in female patients with COPD compared tomale patients with COPD and healthy controls [32]. Inaddition, Skumlien and colleagues [79] found gender differencesregarding effects following a pulmonary rehabilitationprogram and further research regarding thepotential differences in training effects between femaleand male patients with COPD is required. Therefore, weaim to compare primary and secondary outcome measuresfor males versus females in pre-specified subgroupanalyses to gather information for future studies.Intention-to-treat analysis will be used for all outcomemeasurements and involve all patients who are randomlyassigned. Mixed models will be used to handle missingoutcome data. In addition, within the main analysis, anon-treatment analysis will be executed. On-treatment isdefined as at least 80% compliance - in other words, atleast 20 training sessions or three education sessions dependingon group allocation. For data management andstatistical analysis the IBM Statistical Package for SocialSciences (SPSS) version 20.0 will be used.
การแปล กรุณารอสักครู่..
