Background Patients presenting with advanced ovarian cancer can be managed in a variety of ways. No clear selection algorithms exist to guide decision-making and there is significant geographical variation in practice. Decision-making takes place in specialist multidisciplinary team meetings (MDTs). We wanted to examine whether format and behaviour within these meetings could explain the geographical variation in treatment patterns seen in England Methods Observational study of five cancer centre MDTs over a six-week period. Data were recorded for overall MDT performance. The GO-MDT-MODe tool was used to provide a measure of participation and quality of case discussion for all cases of advanced ovarian cancer. MDT scores were correlated with surgical and survival data extracted from national audit data. Results A total of 870 case discussions, including 145 cases of advanced ovarian cancer, were observed. MDTs varied in structure, format and time allocation between centres. Cluster analysis showed significant variation in quality and participation of discussion between centres (p<0.0025) and this correlated with the proportion of patients in the wider cancer alliance undergoing surgery, but not with overall survival Conclusions We have shown that at least part of the variation in practice seen in the UK correlates with different behaviours within MDTs. Increasing time for discussion and encouraging participation from all staff groups may increase proportions of patients undergoing optimal treatment regimens.

Objective The role of surgery for patients with advanced ovarian cancer is well established but wide variations in the percentage of patients undergoing surgery exist. Design Observational Study Setting The 19 cancer alliances in England, UK Methods We undertook a comparative analysis to examine the effects of rates of surgery on one and five year disease specific survival across the 19 English Cancer Alliances. We also undertook a propensity score analysis of a single cohort of patients with advanced ovarian cancer to compare patients who did, and did not, undergo surgery as part of their primary management. Main outcome measures Correlations between rates of surgery, demographics, and cancer alliance performance against one and five year survival Results In multivariable analysis, rates of surgery undertaken by a Cancer Alliance remained an independent predictor of both one and five year survival (correlation coefficient 0.81, and 0.70, respectively) although this relationship is likely non linear with optimal rates of surgery still to be defined. Using propensity score analysis a subgroup of patients who gain no benefit from surgery can be identified. Conclusions Overall survival of patients with advanced ovarian cancer is likely to be improved with an increase in overall rates of surgery however there are likely to be optimal maximal rates, above which patients will be disadvantaged. Mechanisms for identifying individual outcome predictions are required.