33 SEX- AND AGE SPECIFIC ASSOCIATION OF AF WITH MORTALITY IN COVID-19 To explore the association between all predefined patient characteristics and the development of the most prevalent new-onset arrhythmic disorder (i.e. AF and/or AFL), univariable logistic regression analyses were performed to estimate crude odds ratios (OR) and corresponding 95% confidence intervals (95% CI). Next, the association between development of new-onset AF and/or AFL during hospitalisation and in-hospital mortality in COVID-19 patients was first examined using univariable logistic regression analysis. Second, multivariable logistic regression analysis was performed with sex, a cubic spline function for age, the development of new-onset AF and/or AFL during hospitalisation, and the interaction between the latter two variables. The results of this analysis were depicted in plots for males and females separately. To explore whether other concomitant comorbidities and/or other known risk factors may have contributed to the observed results, we performed a sensitivity analysis where we additionally adjusted for CHA2DS2-VASc score. For all analyses, the different AF subtypes (paroxysmal, persistent, and permanent) were merged. All statistical analyses were performed using R version 4.0.2 with the bias reduction in binomial-response generalised linear models (brglm) function in the package ‘brglm’ version 0.7.1, which implements Firth correction reducing finite sample bias in the regression coefficients compared to default maximum likelihood regression.15 Non-linear relations are graphically displayed using the package ‘rms’ version 6.6.1 and the package ‘ggplot2’ version 3.3.2. In all univariable analyses with age and in all multivariable analyses, a cubic spline function for age (and in the univariable analyses for the association between BMI and new-onset AF and/or AFL also a cubic spline function for BMI) with four knots on recommended locations (on the percentiles 0.05, 0.35, 0.65, and 0.95) was used.16 Missing data for each variable were reported as percentages in the text or as counts in the corresponding tables. Since missing data was overall limited (e.g. maximum n=24 in mortality analyses), we proceeded with analyses of complete cases. Associations with two-sided p-values <0.05 were considered statistically significant. RESULTS A total of 5,782 patients were included in this study. The majority of them were hospitalised in European countries (89.9%). The median duration of hospital admission was 8 (IQR 4–17) days, and 28.8% (n=1664) of all subjects were admitted to the intensive care unit (ICU). Of the total study population, 63.8% was male and the median age was 67 (IQR 56–76) years. 12.5% (n=725) had been diagnosed with an arrhythmic event in the past, of which 93.2% (n=676) consisted of at least one episode of supraventricular arrhythmia and 7.7% (n=56) at least one episode of ventricular arrhythmia. Of all patients, 1.7% (n=96) had been diagnosed with at least one conduction disorder 3
RkJQdWJsaXNoZXIy MTk4NDMw