External validation of a colorectal cancer model against screening trial long term follow-up data

Objectives: The ScHARR Bowel Cancer Screening Model has been used to make decisions about screening strategies in England. The objective of this study was to perform external validation of the ScHARR model against long-term follow-up data about colorectal cancer (CRC) incidence and mortality reductions due to screening, from the Nottingham Trial of guaiac Faecal Occult Blood Testing for CRC, and the UK Flexible Sigmoidoscopy Screening Trial.

Methods: The ScHARR model was adapted prior to validation to reflect the setting of each trial in terms of population characteristics, details of screening and surveillance programmes, uptake of screening and further investigations and study follow-up. The impact of using current versus historical CRC incidence and mortality data in the validation was also examined by carrying out a series of analyses in which historical data from different years was included in the model.

Results: The ScHARR model was able to predict CRC incidence and mortality rate/hazard ratios from both trials to well within the 95% confidence intervals in the observed data. Whilst it was less accurate in predicting absolute incidence and mortality rates, modelling historical incidence and mortality data enabled these predictions to be improved considerably.

Conclusions: The ScHARR model is able to replicate the long-term relative benefit from screening observed in two large-scale UK based screening trials and can therefore be considered to be an appropriate tool to facilitate decision making around the English bowel cancer screening programme.