196 Chapter 9 ECG and blood pressure values were collected from the electronic patient records at each timepoint, with a minimum follow-up of 12 months. An automated ECG interpretation was generated and assessed by the treating physician. Heart rhythm, heart rate, electrical axis, conduction intervals, P-wave morphology, QRS complex morphology, and ST-T segment changes were documented. If any abnormal findings were detected, consultation with a cardiologist was sought for further interpretation and evaluation. The evaluations and interpretations of the ECGs were documented in the electronic patient file and later extracted for the database (BvdA and JdR). In the case of abnormal ECG findings, they underwent a meticulous review for a second time, with the assistance of a cardiologist for in-depth analysis and assessment. Statistics Our primary outcome was the occurrence of CV adverse events during treatment with anti-CGRP(R)-mAbs. Secondary outcomes were the change in ECG values (HR, PR-interval, QRS-complex and QTc-interval) and clinical blood laboratory values and over time compared to baseline. Sample size was based on the available data. Baseline characteristics included sex, age, headache diagnosis, baseline monthly headache days (MHD), monthly migraine days (MMD) and systolic and diastolic BP and were summarized using means, standard deviations, frequencies and proportions. Descriptive statistics were employed to analyze the data on CV adverse events. The descriptive analysis included determining the frequency and prevalence of adverse events. To identify potential outliers, a principal component analysis (PCA) was conducted on all laboratory values. For all laboratory and ECG values, a linear mixed model was fitted with time, sex, age, body mass index and treatment (and in case of erenumab including the dosages 70 and 140 mg) as fixed effects and the patient as a random effect, to adjust for potential confounders. For serum creatinine and eGFR, the systolic BP and diastolic BP values measured at baseline were added to the model as additional covariates. For all linear mixed models the assumptions were checked. All missing data is assumed to be missing at random, except for patients who discontinued treatment. To address this issue, complete case analyses were additionally performed. This approach was used for laboratory and ECG values in addition to the main analyses. We used the Bonferroni correction to counteract the multiple comparisons (n=20) and assessed each hypothesis at α = 0.05/20 = 0.003. The corrected p-values are displayed. The analyses were performed in R version 4.2.1 and the lme4 package was used to fit linear mixed-effects models.
RkJQdWJsaXNoZXIy MjY0ODMw