152 Chapter 7 For the measurement of pain catastrophizing, the Pain Catastrophizing Scale (PCS) was used (30). The PCS score ranges from 0 to 52, with higher scores corresponding to higher levels of pain catastrophizing. The Dutch Central Sensitization Inventory (CSI) was used to assess central sensitization (31). The CSI score ranges from 0 to 100, with higher scores corresponding to higher levels of central sensitization. Self-efficacy was assessed with the General Self-Efficacy Scale (GSE Scale) (32,33). The GSE scale score ranges from 10 to 40, with higher scores corresponding to greater selfefficacy. Self-management ability was measured using the Dutch language version of the shortform Patient Activation Measure (PAM 13-Dutch) (34). The PAM 13-Dutch score ranges from 0 to 100, and a higher score corresponds to higher levels of self-management. Health-related quality of life was assessed with the EuroQol-5D-5L (35). A higher score (range 0-1) corresponds to higher health-related quality of life. Treatment group allocation in the e-Exercise LBP parent trial (19) was the last potential confounder of interest. Data analysis Data preparation was performed using SPSS 27 (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp.). Descriptive statistics were used to report patient characteristics, utilizing means and standard deviations (SDs) for normally distributed data and medians with interquartile ranges (IQRs) for data that were not normally distributed. Subsequent analyses were performed using R (R Foundation, Vienna, Austria). For an EXAS score, a minimum of two treatment sessions are required, which excluded patients with only one treatment session from the analyses. Multivariate imputation by chained equations was used to impute missing data in R using the ‘mice’ package (36,37). In 52% of the cases, at least one measurement was missing, and an imputed dataset was generated for every percentage of cases with missing data, resulting in 52 imputed datasets. The analyses and computations of the pooled results were performed on all imputed datasets using the ‘miceafter’ extension package for ‘mice’. Linear regression and binomial logistic regression were used to test the relationship between adherence and the outcomes. The changes in the ODI and NPRS between baseline and after three months and recovery from LBP were used as outcomes. The mean EXAS score over all treatment sessions and the previously determined trajectory of adherence classes were used as determinants of adherence. All three outcomes were modelled using both determinants of adherence separately and adjusted for confounding
RkJQdWJsaXNoZXIy MTk4NDMw