Quality control practices in centralized tumor registries in North America.
 A survey of quality control practices was mailed to 73 central registries in the U.S.
 and Canada.
 The response rate was 88%, with respondents representing a wide range of registry characteristics and reporting strategies.
 While registries expressed different priorities in data use, 80% of respondents felt quality control data were important in the identification of problems.
 The most common method of quality control was acceptance sampling (used by 97% of respondents), and took the form of visual review, recoding and edit checking.
 Computer-based edit checks were almost universally used (95%).
 Process control methods of any sort were used by only 22% of respondents with less than 4% of registries reporting formal quantitative criteria.
 Sixty-one percent of respondents reported conducting one or more designed studies (e.g.
 reabstracting or casefinding studies) but only 20% of those made the results public.
 Greater emphasis should be placed on development of quantitative process controls, experimental design of quality control studies, and formal analyses and reporting of study results.
