Caution is required in the implementation of 90-day mortality indicators for radiotherapy in a curative setting: A retrospective population-based analysis of over 16,000 episodes

Background: 90-day mortality (90DM) has been proposed as a clinical indicator in radiotherapy delivered in a curative setting. No large scale assessment has been made. Its value in allowing robust comparisons between centres and facilitating service improvement is unknown. Methods: All radiotherapy treatments delivered in a curative setting over seven years were extracted from the local electronic health record and linked to cancer registry data. 90DM rates were assessed and factors associated with this outcome were investigated using logistic regression. Cause of death was identified retrospectively further characterising the cause of 90DM. Results: Overall 90DM was 1.25%. Levels varied widely with diagnosis (0.20%-5.45%). Age (OR 1.066, 1.043-1.073), year of treatment (OR 0.900, 0.841-0.969) and diagnosis were significantly associated with 90DM on multi-variable logistic regression. Cause of death varied with diagnosis; 50.0% post-operative in rectal cancer, 40.4% treatment-related in head and neck cancer, 59.4% disease progression in lung cancer. Conclusion: Despite the drive to report centre level comparative outcomes, this study demonstrates that 90DM cannot be adopted routinely as a clinical indicator due to significant population heterogeneity and low event rates. Further national investigation is needed to develop a meaningful robust indicator that delivers appropriate comparisons and drive improvements in care.