Coding of Central Cancer Registry Industry and Occupation Information: The Texas and Louisiana Experiences.

Background: Usual industry and occupation text information have been collected by central cancer registries but few have had the resources to code these data, limiting their usefulness for assessing occupational cancer risks.

Study Aims: This project was undertaken to use software available from the National Institute for Occupational Safety and Health (NIOSH) to code industry and occupation information in cancer records reported to the Texas Cancer Registry (TCR) and the Louisiana Tumor Registry (LTR) and to assess the feasibility of its use in ongoing registry operations; to assess the quality of the reported information; and to determine its usefulness in occupational cancer research.

Methods: De-identified data files of TCR (n = 103,276) and LTR (n = 26,090) cancer records were obtained for diagnosis years 2010 and 2011, respectively, for cases aged 14 years and older, with industry and occupation text. These data fields were coded to the 2000 US Census Bureau using the NIOSH Industry and Occupation Computerized Coding System (NIOCCS) software at the high level confidence (90% or greater accuracy) and through manual code assignments for records not coded by NIOCCS.

Results: NIOCCS assigned a code for 37.2% of TCR records and 59.9% of LTR records. Examination of the quality of the coded data found 44.2% of TCR records and 31.1% of LTR records to have missing, unknown, or otherwise insufficient text for assigning a specific industry and occupation code. Additionally, the vague noninformative category of "retired" was reported for 14.9% and 11.2% of TCR and LTR records, respectively. Records with "homemaker/housewife" or those with terms indicating that they never worked represented 7.2% of TCR cases and 9.7% of LTR cases. Excluding the unknown, never worked, and retired categories, no one specific industry or occupation major grouping represented more than 5% of cases in either of the registries.

Conclusion: NIOCCS is a helpful tool for coding industry and occupation text and continues to improve, but other registry resources are required for implementation into ongoing operations. Improvement in data quality of reported text information in cancer records is paramount to maximize the efficiency of NIOCCS and improve the availability of coded, specific industry and occupation information for occupational cancer research.