The long learning curve of gynaecological cancer surgery: an argument for centralisation

Objective: To study the development of surgical performance of an unchangin g surgical team over 13 years. Design: Prospective, observational study. Setting: A university hospital, The Netherlands. Participants: Three hundred and eight women who underwent surgical treatmen t for early cervical cancer. Interventions: Radical hysterectomy and pelvic lymphadenectomy between 1 Ja nuary 1984 and 31 December 1996. Results: The surgical procedure and indication for treatment remained uncha nged during the study period. This applied also to the surgical team. The w omen's age increased significantly during the study years, as was the case with the number of nodes removed. The depth of infiltration by the tumour i ncreased steadily throughout the study, but this failed to reach statistica l significance. The distribution of FIGO stages, percentage of positive lym ph nodes, radicality of the surgical margins and post-operative morbidity r emained the same. Overall, the five year survival rate was 83%; for women w ith negative nodes 91%, and for women with positive nodes 53%. Survival ten ded to improve during the course of the study, but this was not statistical ly significant. Blood loss during surgery decreased consistently during the whole study period, from a mean of 1515 mL at the beginning of the study t o a mean of 1071 mL at the end (P < 0.0001). The operating time also dimini shed significantly by 8 minutes per year (P < 0.0001). In 1985 the average operating time was 270 minutes, compared with 187 minutes in 1996. Conclusions: These findings indicate that it takes a long time to acquire s kill in the surgical treatment of early cervical cancer. Centralisation of relatively infrequent operations for cancer should be encouraged.