Comparing treatment trends for colorectal cancer in clinical database and cancer registry data: implications for monitoring cancer care

Abstract

Recent UK policy has promoted routine monitoring of the performance of cancer services. Clinical databases are one method of collecting the detailed data required for clinically relevant analyses and of supplementing that collected by cancer registries. To describe characteristics of patients, their treatment and surgical care recorded in the North Thames Prospective Audit of Colorectal Cancer database compared with the Thames Cancer Registry database between 1999 and 2004. We defined the geographical area served by the prospective audit and extracted a registry dataset for the same area. We compared the numbers, age, sex, cancer site and staging of patients in each database. The proportions recorded as receiving surgery, radiotherapy and chemotherapy were compared over time, and clinical standards recorded for preventing post-operative complications and for specialist care were determined from the clinical database. Clinical and registry databases differed in case mix. The clinical database began to capture data on a larger number and wider range of patients over time, and generally included more complete staging data. Treatment trends differed between the databases, with the registry data showing lower proportions undergoing surgery and a slight increase in rectal cancer surgery over time. There was a suggestion that a higher proportion of men received radiotherapy and chemotherapy than women. Recording of surgical standards in the clinical database was generally poor, although decreased leak and infection rates and post-operative mortality were suggested in the initial years of the audit. The clinical database contributed some additional information on patient characteristics and the performance of services, but variable data quality and completeness prevented many planned analyses. Collecting additional data requires funded structures and quality assurance systems and such effort should focus on the data required to answer the most pressing clinical questions.