Statistical analysis

The mean numbers of each cell type were calculated for each arm at 24 h post-treatment and compared to those in the pre-treatment samples using a simple t test, with p =  < 0.05 as the level of significance. Since none of the measurements passed this test, no further analysis was performed. Cytokine levels were obtained from the Luminex data. The mean level plus or minus standard error of each analyte were calculated for each arm at each time point (0, 4, 24 h post-treatment). For the control and ivermectin arms, levels at 4 h and 24 h post-treatment were compared to the levels pre-treatment using simple t tests to determine if any individual results required further investigation, taking p =  < 0.05 as the level of significance. The change in levels 4 h and 24 h post-treatment were also compared between the placebo- and ivermectin-treated groups. Since none of the individual analyte results met this criterion, no further analysis was carried out. The NanoString data were analysed on the proprietary software, nSolver™ (NanoString Technologies, Inc., Seattle, WA, USA). The data passed quality control criteria. The software calculated the log₂ geometric mean levels of each mRNA measured and used t tests to determine statistically significant changes in expression between 4 and 24 h post-treatment for the control and ivermectin arms. We also compared the changes in expression at 4 h and 24 h post-treatment between the placebo- and ivermectin-treated groups. The Benjamini-Yekutieli false discovery rate [35] method was used to account for the expectation that significant changes in genes may be correlated with or dependent on each other, and the subsequent resulting false discovery rate adjusted p-value of < 0.05 used to determine those changes that were deemed to be statistically significant. In all the analyses, each individual subject was treated as a biological replicate, giving N = 4 for the placebo arm and N = 8 for the ivermectin arm.