Validation of LexisNexis® Accurint® in the Georgia Cancer Registry's Cancer Recurrence and Information Surveillance Program.

BACKGROUND Lexis® Accurint® is a database of ~84 billion public records that includes an individual's location of residence. Its ability to track residences longitudinally has not been validated. This study used the Georgia Cancer Registry's Cancer Recurrence and Information Surveillance Program (CRISP) to validate U.S. state of residence and to examine characteristics of patients not included or who had an inaccurate entry in LexisNexis. METHODS The Georgia Cancer Registry is routinely linked to the National Death Index (NDI), providing information regarding state of residence in which the patient died. We compared the state of residence reported in LexisNexis with the NDI gold standard state of residence at death. Multivariate logistic regression analyses estimated associations between demographic information and: (1) having a mismatch between LexisNexis and NDI, and (2) being missed in LexisNexis. RESULTS Of the 69,494 patients in the CRISP cohort, 65,890 (95%) were found in LexisNexis and 9,597 (14%) had died. Among a subset of patients who were deceased, the sensitivity of LexisNexis for identifying persons who left Georgia was 42% and the specificity was 89%. Minority groups were more likely to be missed in the LexisNexis database as well as to have a discordance between LexisNexis and NDI state of residence at death. CONCLUSIONS LexisNexis Accurint failed to identify the emigration of more than half of deceased CRISP patients who had left Georgia, but correctly identified most who had remained. The validity of state of residence is important for studies using LexisNexis as a tool for follow-up.