BACKGROUND: In 2009, the Centers for Disease Control and Prevention obtained funding to expand data collection among central cancer registries (CCRs) in 10 states within the National Program of Cancer Registries in support of comparative effectiveness research (CER). This expanded data collection included patient occupation and industry (O/I) information. CCRs were also encouraged to recode textual O/I information into census numeric codes, using the NIOSH Industry and Occupation Computerized Coding System. We examined the feasibility and completeness of O/I data collection submitted through the Enhancing Cancer Registry Data for CER project.
METHODS: The quality of O/I data for 304,637 cancer cases diagnosed in 10 states during 2011 was investigated. Both textual and recoded census numeric values of O/I data were examined for completeness by state.
RESULTS: Among the eight states that submitted O/I code, three had fewer than 10% missing data. Within these states, 63% of occupational codes were classified as “Unemployed” and 61% of industrial codes were “Unknown.” For O/I textual data, seven states submitted data with fewer than 10% missing for occupational data and five submitted data with fewer than 10% missing for industrial data. O/I completeness varied by state, but 36% of occupational and 33% of industrial text were “Retired,” “Unemployed,” or “Unknown.”
CONCLUSIONS: Although states within the CER project were specifically funded and encouraged to collect O/I data, the quality of information collected fell below expectations. States reported similar challenges to the collection of these variables: O/I information from source data such as hospitals were more likely incomplete or not easily coded; the coding process was time consuming and required manual editing; and privacy issues inhibited follow-up with facilities. These findings suggest that more training may be needed to improve the quality of O/I data within medical records as well as central cancer registries.