Chapter 5 88 Health-related quality of life (HRQL) was assessed [19]. Patients reported HRQL by answering a single question; ‘can you indicate on a scale of 0 to 10 how you currently perceive your health-related quality of life, where 0 = very poor and 10 = very good’ [20]. With this single question, longitudinal changes in HRQL within patients during 12 months could be measured. Covariates The aim of this study was to determine the course of exercise adherence over a 12month period in patients undergoing prolonged rehabilitation so, covariate of interest was; time points (baseline, 3, 6 and 12 months). Statistical analysis Data were analyzed using R version 4.0.3. For missing data first, the amount of missingness for each variable was calculated; the difference between the sample size and the number of useable observations. Second, Fisher exact tests were used to analyze differences in baseline characteristics between patients with missing and complete data. Finally, multiple imputation was used to create and analyze five multiple imputed datasets. Incomplete variables were imputed under fully conditional specification, using the default settings of the mice 3.0 package [21]. The parameters of substantive interest were estimated in each imputed dataset separately, and combined using Rubin’s rules. Considering the hierarchy in the data, with adherence (and 6MWD and HRQL) nested in patients, multilevel regression analysis [22] was used with adherence (and 6MWD and HRQL) at the first level and patients at the second level. All patients had different degrees of adherence at baseline and adherence may have changed differently over time for each patient [10] so, we let both the intercepts and slopes of the association between time and adherence vary at the patient level. This random intercept and random slopes analysis gives information about whether the association between adherence and time is different in each patient. Next, a model was developed to study the influence of time on adherence during twelve months. Fixed effects The results are shown as beta’s (b) with t-values and degrees of freedom. Results were considered statistically significant when p < 0.05. Random effects Second level variance (variation between patients) was calculated regarding adherence (i.e., the intercepts in the multilevel regression), and the second level variance regarding the association between time and adherence (i.e., the slope variance in the multilevel regression). Covariance between intercept and slope residuals were also calculated. The covariance gives information about whether the
RkJQdWJsaXNoZXIy MjY0ODMw