SummaryData from the Thames Cancer Registry were compared with data independently abstracted from medical records for 466 patients with confirmed cancer of the bladder diagnosed in 1982. High levels of agreement were observed for five continuous variables and for tumour morphology. Data England is one of a small number of countries with a national system of population based cancer registries (Waterhouse et al., 1982). The data collected are published at national level (Office of Population Censuses and Surveys, 1985) and provide information concerning the incidence and duration of survival with cancer (Cancer Research Campaign, 1982). Cancer registry data also have uses in epidemiological research and health service planning (Office of Population Censuses and Surveys, 1990), their value as a starting point for auditing the effectiveness of cancer treatment has also been emphasised (Gillis et al., 1991). As cancer registry data may be put to a number of practical uses it is important to evaluate their quality. The completeness with which cases are ascertained by registries has been investigated in several studies (Nwene & Smith, 1982
ResultsThe selection of cases has been reported previously (Gulliford et al., 1991a;Gulliford et al., 1991b