Uncertainties dose evaluation induced by non-exhaustivity data

Introduction The individual dose estimation to organs at risk has known important evolution last years, with new tools whose permits to obtain histogram dose-volume. However, dose reconstruction based on retrospective epidemiology studies are often realized dozens of years after initial radiotherapy, based sometimes on non-exhaustive data, both on patient anatomy as the beams. The aim is to establish the link between the missing data rate and the uncertainties induced on out-of-field dose by the non-exhaustivity of the data. Methods and materials Statistics are used for evaluating the missing data rate in patient radiotherapy files from french, british, Italian and dutch cohorts. The around twenty variables contributing to the patient anatomy modelisation and dose calculation are evaluated. Statistics methods (multiples imputation) are used to complete the missing data. For uncertainties dose calculation, the formalism defined in TRS 384 is applied, variances are deducted from the statistical study. Results The primary results indicate that, according to the considered parameter, missing data rate is very inconstant and strongly linked to the year of treatment and the institution. The dose uncertainties induced is very linked to the considered parameter. Conclusion The results of the study should allow a better knowledge of the uncertainties in retrospective dose estimations used for long term studies of effects of radiotherapy.