Multivariable logistic regression analysis
The effects of training on post-operative and survival outcomes are expressed as odds and hazard ratio’s in Table 3. Effect sizes were corrected for age, stage (<2/≥2) and grade (low grade/high grade) using multivariable logistic regression and cox proportional hazard analysis. Variables were included in multivariable logistic and cox proportional hazard analyses if they rendered clinical relevance and showed a significant effect on disease-free and disease-specific survival in univariate regression analysis.
After correction training did not increase the odds ratio’s (OR) for intra-operative complications (0.6, p = 0154), post-operative complications (1, p = 0.994), Clavien-Dindo complications, (0.9, p = 0.583), readmissions < 30 days (0.6, p = 0.202), return to theatre (0.6, p = 0.553) or lymphoedema (0.6, p = 0.110). Training cases had a lower odds for LOS > 1day (0.6, p = 0.004).