Reply to "Varicella-like exanthem as a specific COVID-19-associated skin manifestation: multicenter case series of 22 patients": Discussing specificity.

To the Editor: We read with interest the article by Marzano et al addressing the specificity of varicella-like exanthem to diagnose coronavirus disease (COVID-19). Although this type of COVID-19–associated rash is rare, the authors claim that it is more specific than others without having performed a diagnostic accuracy study. This type of study would calculate the association of COVID-19 status (yes/no) with the type of exanthem (varicella-like/non–varicella-like) in a predetermined number of patients (to ensure sufficient statistical power) and in a defined population with a known prevalence of COVID-19 disease.

Starting from the assumption that the data are not sufficient to draw such a robust conclusion, we would like to contribute our vision of such an important issue.

Firstly, for clinicians, specificity is of little value. The positive predictive value, that is, the probability that a person with a varicella-like rash has COVID-19, is much more relevant, because a high value would justify severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) testing. The positive predictive value is affected by the prevalence of the disease. Consequently, as we are going through the peak of the COVID-19 pandemic, it is evident by applying the Bayes theorem that the conditional probability of COVID-19 given a varicella-like rash is enormous. However, this probability is as high given a dengue-like, erythematous, or urticarial rash. Therefore, the positive predictive value for COVID-19 of skin rashes, regardless of which, is high.

For us, the main contribution of the study by Marzano et al is that with all exanthems currently having a high positive predictive value for COVID-19, that of varicella-like rash is probably the highest because vesicular rash is more specific for viral disease than others. In addition, at the current stage of the pandemic, COVID-19 is more frequent in adults, where varicella is not. This gives varicella-like rash additional COVID-19 positive predictive value compared with the rest of the rashes. The main factor that at this moment reduces the positive predictive value of rashes for COVID-19 is adverse drug reactions. Unfortunately, this topic is being less discussed in the literature.

To conclude, because things will change fast as the pandemic progresses, as the adult population becomes infected, the age at diagnosis of COVID-19 is expected to decrease. At that time, this rash will be an important differential diagnosis of varicella itself. In this context, the Tzanck test and varicella-zoster polymerase chain reaction should be subjected to new diagnostic accuracy studies.