Fig. 3.

COVID-19 associated symptoms in healthcare workers and risk stratification in an unbiased decision tree. a Percentage of SARS-CoV-2 Ab⁺ and Ab⁻ HCWs who reported having experienced at least one of nine symptoms shown in b. P-value was calculated using Fisher’s exact test. b Frequency of individual symptoms in SARS-CoV-2 Ab⁺ and Ab⁻ staff members with at least one self-reported symptom as a percentage of the respective group. c Percentage of SARS-CoV-2 Ab⁺- and Ab⁻ staff reporting the indicated number of symptoms. Numbers beside data points indicate number of staff members per group. d Frequency of co-occurrence of pairs of symptoms in Ab⁺ (red) and Ab⁻ (blue) staff members. Squares on the diagonal represent the frequency of single symptoms. e A conditional inference tree (decision tree) was trained in R using the ctree function implemented in the party package, using default parameters. All significant parameters from the logistic regression were included in the training dataset. Depicted is the resulting decision tree with the stop-criterium for tree splits set at a significance level of α = 0.05. Numbers underneath bars represent the total number of HCWs in the respective group, numbers in braces those of Ab⁺ staff members