Objective To define and use a minimum clinical dataset for prospective data collection in order to audit the surgical management of cervical cancer in the South West of England. To compare this data set with a retrospective audit allowing assessment of the quality of care offered to patients.

Design Prospective collection of a defined dataset on paper forms which were put into a computerised database for analysis. Registrations validated against histopathology databases and hospital coding.

Setting All 13 hospitals in the South West of England which participated in the retrospective audit.

Participants One hundred and sixty-five women with cervical cancer diagnosed in 1997.

Main outcome measures Distribution of cases by hospital and surgeon; workload of individual surgeons; adequacy and accuracy of FIGO staging; adequacy of histological information; and adequacy of surgery.

Results There is a trend to centralisation of cancer care and radical surgery in the region. Prospective collection of data has dramatically improved FIGO staging with 92% of all cases staged. For cases greater than Stage la, 98% were staged suggesting that a target of 100% staging is feasible. The histological dimensions of tumours were not measured in a high proportion of cases (20% of tumour diameters and 28% of tumour thicknesses). Apparent inadequacies in surgical management are explored. In 10/165 cases (6%) inappropriate conservative surgery may have been unavoidable, suggesting that a quality standard of 95% for appropriate radical surgical management of cervical cancer can be achieved. An anatomically complete removal of pelvic node-bearing tissue, yielding greater than 10 nodes in more than 95% cases, should be achievable with each surgeon/pathologist achieving a mean of more than 20 nodes.

Conclusion Regional audit of cervical cancer management is feasible. It can be used to improve the quality of information on management and guide improved service provision.