Statistical methods
For between timepoint statistical comparisons, parametric variables (SPT
wheal size, HRQL) were analyzed using paired t-test with data summarized
as mean and standard deviation (SD). Non-parametric variables (sIgE,
sIgG4) were analyzed using Wilcoxon rank-sum test with data summarized
as median and inter-quartile range (IQR) and standard t-test on
logarithmic scaled values with data summarized as geometric mean and
95% confidence interval (95% CI). The number and percentage of
participants who achieved desensitization and SU were presented for all
enrolled participants (intention-to-treat) and those who had an
end-of-treatment DBPCFC outcome (complete case population). Persistence
of SU was expressed as the proportion of participants with SU at T3 who
maintained SU at 3-years post-treatment (T5). For HRQL outcomes, primary
analysis used all available data at all time points. Robustness of
findings and impact of missingness was investigated with a sensitivity
analysis of complete series data to T4. P < 0.05 was
statistically significant. Statistical analyses were performed using
Stata release 15 software (StataCorp, College Station, Tex) and R (R
version 3.6.1).