Supplemental Material: Comparative Histopathologic Analysis of COVID-19-Associated Pseudo-Chilblains (“COVID-toes”), with Idiopathic Pernio, and Chilblains Lupus Erythematosus
Description
This study is a systematic review which evaluates histopathologic findings of COVID-toes and compares them to that of CHLE and IP, with the aim toward understanding the pathogenesis of COVID-toes. Expanded Methods: A total of 50 articles met inclusion criteria and were included. Due to the small number of subjects per study (range 1-22), the study- and subtype-specific proportions were estimated to be either zero or one for many studies. To compensate, the analysis was conducted using a common effect meta-analysis without considering between-study heterogeneity6,7. Thus, the proportions of patients with each histopathologic finding subtype were estimated by aggregating the cases in all studies and divided by the total number of patients reported in all studies combined. The confidence intervals for the proportions were calculated by the Wald (number of events >5) or Clopper-Pearson method (number of events <5)8. The proportions of patients with COVID-toes were compared pairwise to ChLE and IP, respectively, using Pearson’s chi-squared test. Demographics were compared using Welch’s one-way t-test for mean age and Kruskal-Wallis rank sum test for the proportion of female patients. The unadjusted odds ratio (OR) between the diagnoses and sex were estimated by Firth logistic regression to reduce the bias that may be introduced by the small number of males with ChLE9. Since the patient sex information was not available in all studies, OR was estimated from two datasets: 1) only studies with patient sex information available; 2) all studies, and in the studies where patient sex information was unavailable, number of females and males were estimated in accordance with the overall sex proportion of the study. P-values were adjusted using Bonferroni correction to control the type-I error for multiple tests. The significance level was set at α = 0.05/number_of_tests = 0.05/25 = 0.002. All analysis were performed in R version 4.0.4