Evaluating the breast screening programme: the need for surgical audit.

It is to be anticipated that a reduction in population mortality attributable to mammographic screening would be heralded by an increasing proportion of breast cancer cases diagnosed at earlier stages and by an improvement in case fatality. Few cancer registers routinely produce incidence or survival data by stage at diagnosis and thus improvement in these will be the harder to assess. By thorough casenote review and follow up, this study has determined the usual presentation and survival of breast cancer in Northern Ireland in 1986 before the introduction of screening. Overall, 85% of cases were Manchester stage I or II, figures which accord with other British studies. Five-year survival ranged from 77.8% for stage I (95% confidence limits 71.7%, 82.7%) to 35.7% for stage IV (95% confidence limits 13.0%, 59.4%). Forty-three per cent of cases treated in non-teaching hospitals could not be pathologically staged, more than twice the figure for teaching hospitals (Chi-squared = 15.7, df = 1, P < 0.001). Since many tumours not detected by screening will be treated outside teaching centres, this difference will reduce the statistical power to detect the true shift in stage distribution and improvement in survival from screening. Comprehensive surgical audit would help to resolve the inadequacies in existing data collection and improve the ability to evaluate the outcome of the screening programme.