Chi-square tests were used to assess the association between categorical variables and the 2-level VAS pain measures at baseline and at time of healing, while independent
t-tests were used for continuous variables. Variables with a non-normal distribution were analyzed with appropriate non-parametric procedures, Mann–Whitney
U test for unpaired data and Wilcoxon signed ranks test for paired data. Multivariable logistic regression was then used to identify the subset of significant predictors
of pain at healing. The primary outcome for the predictive model was defined as pain present at time of healing (VAS>0/10). All regression procedures used simultaneous
entry. Variables were eliminated one at a time in successive regressions if p≥0.10, and retained if p