STATISTICS
The association between severity of COVID-19 and both the atopy status and the clinical co-factors recorded was studied in both univariate and multivariate analyses. Each variable of interest was dichotomized as negative or positive to study the proportion of subjects with a given clinical status. Categorical variables were analyzed using the Pearsons’ χ2 or Fisher’s exact test when indicated. Multiple logistic regression was performed to estimate the degree of association of the main exposure variable (i.e., atopy) with COVID-19 severity after simultaneously adjusting for all the other variables of interest. P values <0.05 were considered significant.
Relatedness of COVID-19 disease grading was tested by applying unsupervised Eisen’s hierarchical cluster methods to the data set, encompassing patient’ comorbidities and using as agglomeration rule the average linkage clustering as implemented in the program Genesis 1.7.2 (10,11). Unsupervised clustering involved the sorting of both COVID-19 clinical status and patients comorbidities. The COVID-19 disease grading tree was computed based on a full data set and the distance between samples was computed by using the Pearson correlation as similarity measures. As a result, coexistent clinical comorbidities were grouped as hierarchical clusters and presented as heat-maps. Each square in the heat-map represents the presence (red) or the absence (black) of any given tested comorbidity.
The SPSS/PC+ statistical package for statistical evaluation (SPSS, version 15, Chicago, IL) was used to analyze the data.