166 | Chapter 8 Data collection For each patient, baseline demographics, treatment with antiarrhythmic medication, and presence of an ICD were recorded. ECG, Holter and CMR data from the study closest to baseline echocardiography and prior to occurrence of primary outcome were used. The 7 variables included in the ARVC risk calculator (www.ARVCrisk.com) were collected independently in each centre according to uniform definitions, and included: age at diagnosis, sex, recent syncope (<1 year ago), number of inferior/anterior leads with T-wave inversion, non-sustained ventricular tachycardia (NSVT), 24-hour premature ventricular complex (PVC) count and RVEF measured by CMR. All genetic variants reported were adjudicated according to the American College of Medical Genetics and Genomics guidelines.17 The echocardiographic deformation imaging protocol was previously described in more detail.12,15 We used a GE Vivid 7, E9 or E95 scanner and loops were stored for post-processing with EchoPac version 203 (GE Healthcare, Horten, Norway). All measurements were performed by a single observer according to the current guidelines18,19 and blinded to clinical outcome. Data on inter- and intra-observer variability are provided in Supplemental Table 1. Myocardial deformation was assessed in three apical views of the left ventricle (LV) and in the RV focused 4-chamber view. Predictive value of the following five deformation parameters was tested: LV global longitudinal strain (GLS), LV and RV mechanical dispersion (LVMD and RVMD, respectively), RV free-wall longitudinal strain (RVFWLS) and regional RV deformation patterns. LV GLS was calculated as the peak negative strain from the averaged regional 16-segment LV model.18 Mechanical dispersion was calculated as the standard deviation (SD) of the segmental time intervals until maximum shortening of 16 segments in the LV and 6 segments in the RV.12 RV FWLS was defined as the peak negative strain from the averaged regional RV free wall deformation curve. Deformation patterns were analysed in the basal, mid and apical segment of the RV free wall. The RV deformation pattern was classified as normal if the pattern was normal in all three segments and abnormal if at least one segment showed abnormal deformation (Figure 1).10,15 Missing data Missing data were assumed to be missing at random and imputed using multiple imputation with chained equations as described previously.4 The multiple imputation model included all pre-specified predictors, proband status and genotype together with the outcome, and cumulative baseline hazard estimation.20,21 A total of 25 imputed datasets were generated, and the final inference estimations were combined using Rubin’s rules.22 All covariates of the ARVC risk calculator and the deformation imaging variables had <5% missingness, except for 24-hour PVC count (11%) and RVEF by CMR (19%). Complete case analyses without imputation and 24-hour PVC count and RVEF were performed as sensitivity analyses (Supplemental Table 2).
RkJQdWJsaXNoZXIy MTk4NDMw