Quality of cancer registry birthplace data for Hispanics living in the United States

Abstract

Patient birthplace from the SEER population-based cancer registries is potentially useful for identifying disparities in cancer occurrence and for studying cancer etiology. However, for Hispanics, completeness and accuracy of registry birthplace is unknown. By comparing registry birthplace to self-reported birthplace from 13 interview studies, we determined the completeness and accuracy of this variable and the associations of these measures with patient and hospital characteristics in the Greater Bay Area. Registry birthplace was unrecorded for 46% of 1277 Hispanic cancer cases, and unrecorded birthplace (i.e., incompleteness) was associated with younger age, higher education, English language preference, US birthplace, and admission at certain hospitals. For 691 Hispanics with available registry birthplace, sensitivity and positive predictive value compared to self-report (i.e., accuracy) were 96.3 and 97.3 among foreign-born, and 96.8 and 95.6 among US-born. US-born Hispanics misclassified in the registry as foreign-born were more likely to have unavailable education information, be deceased, prefer a language besides English, and be diagnosed at a smaller hospital or before 1996. Among self-reported foreign-born Hispanics, those misclassified as US-born were less likely to have been diagnosed at an HMO. Although the completeness and accuracy of birthplace information may vary across registries, this variable appears to be limited for analyses involving Hispanics.