Learning curve
Based on significant predictors in univariate logistic regression,
multivariate risk models were fitted. The final multivariate risk model
used for RA-CUSUM analysis included age (odds ratio = 1.03, 95% CI
1.00-1.06, p=0.03) and adjuvant or adjusted treatment (odds ratio =
3.15, 95%CI 1.41-7.04, p=0.01), which summarises multiple prognostic
factors significant at univariate analysis (e.g. lymph node status, LVSI
and tumour size). Adjusted treatment was defined as replacing radical
hysterectomy with chemoradiation because of an intraoperative finding of
positive lymph node at frozen section.
Figure 1 shows the expanded RA-CUSUM charts. From the
introduction of the novice surgeon (group 3) no new peak in RA-CUSUM+
chart is observed and the chart moves around zero, indicating
satisfactory results with regard to the observed recurrence rate. The
RA-CUSUM- chart, indicating further improvement of surgical performance
by moving more negatively, has stabilized in group 3 and does not show
further improvement.