|| Checking for direct PDF access through Ovid
We compare the effects of voluntary and mandatory submission of data to a colorectal clinical registry and the implication for national statistics and unit benchmarking. We analyse data from the National Bowel Cancer Audit Programme (NBOCAP) in the UK and compare outcomes between units that do, and do not, submit data to this registry from a compulsory administrative national database-the Hospital Episode Statistics (HES).Colorectal cancer patients undergoing major colorectal resections over a 12 month period were compared. Thirty-day mortality, length of stay and readmission rates were compared for submitter and non-submitter units. Multifactorial analyses were performed to identify independentpredictors of outcome differences.17722 patients underwent resections. No demographic differences existed between the groups. Unadjusted 30-day in-hospital mortality rates were higher in non-submitting trusts compared with submitters (5.2% vs 4.0%, P = 0.005). On multifactorial regression ‘submitter’ status was independently associated with lower 30-day mortality risk (OR 0.76, 95% CI for OR 0.61-0.96, P = 0.021).Voluntarily submitted data for use in benchmarking can only be considered as estimates. In England units that do not voluntarily report data to a national colorectal cancer registry are associated with higher postoperative mortality risk.