Disclosure Risk Assessment for Population-based Cancer Microdata

• Activity Number: 338
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Mandi Yu*+ and David Stinchcomb and Kathleen Cronin
• Companies: National Cancer Institute and Westat Inc. and National Cancer Institute
• Keywords: Cancer Registry Data ; Disclosure Risk ; Population Unique ; Small Area ; ACS ; PUMS

Abstract

The authors developed and tested a non-parametric method of estimating the risk of disclosing information about whether known population individuals have cancers. This method matches cancer patients diagnosed in 2000 in a research data file from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program with individuals in Census 2000 Public Use Microdata Sample (PUMS) by county of residence and several common demographic key variables and estimates the proportion of patients who are unique in both files. Older racial and ethnic minorities residing in less populous areas are at higher risks of being identified. Both imputation methods produce conservative risk estimate and the magnitudes of upwards bias are 3-4 times. The bias tends to be greater for areas with larger risks. This research is the first attempt to systematically evaluate the risk of disclosing the attribute of whether an individual has cancer, and it builds substantial foundation for establishing routine procedures for assessing such risks using yearly updated microdata from the American Community Survey (ACS) in the future.