Ethnicity Misclassification in Cancer Registries

Background and Objectives: NAACCR's Latino Research Group developed the Hispanic Identification Algorithm (NHIA) to improve ascertainment of Hispanic ethnicity in US cancer registries (CRs). This study applies NHIA to assess possible misclassification of Hispanic/non-Hispanic cancer incidence rates (CIRs).

Methods: NHIA was applied to all primary cancer cases (PCC) in Rhode Island residents (1997-2007). NHIA assigns Hispanic ethnicity if probability of being Hispanic at birth country is high; gender is male and surname is predominantly Latino (per US Census Bureau); gender is female and maiden name is predominantly Latino; or, maiden name is absent/indeterminate and surname is predominantly Latino.

Results: Age-adjusted (2000 US population) CIRs for Hispanics were slightly higher than for non-Hispanics. CIRs for Hispanic and non-Hispanic females were not statistically different (463 vs. 454/100,000) but significant (p<0.05) for Hispanic/non-Hispanic male CIR differences (665 vs. 621/100,000). Most common sites for Hispanics included breast, prostate, lung-bronchus and colon-rectum. Hispanic females had higher CIRs for colorectal cancer (57 vs. 50 in non-Hispanics) and lower for breast (119 vs. 130) and lung-bronchus (50 vs. 62). Among Hispanic males, CIRs were higher for prostate (222 vs. 164) and lower for lung-bronchus (86 vs. 98) and colon-rectum (64 vs. 71).

Conclusions: CIRs (1999-2003) calculated with the CR ethnicity field were lower for Hispanic males/females compared to NHIA. The former method also yielded smaller differences between Hispanic and non-Hispanic female (436 vs. 455) and between male CIRs (635 vs. 631). Findings suggest that Hispanics are under-identified and non-Hispanics over identified in the Rhode Island CR.