Bias in completeness of birthplace data for Asian groups in a population-based cancer registry (United States)

Cancer Causes Control. 2004 Apr;15(3):243-53. doi: 10.1023/B:CACO.0000024244.91775.64.

Abstract

Data on place of birth are routinely collected by population-based cancer registries in the United States and are used to study effects of immigration on cancer patterns in Asian migrants, who comprise about a quarter of the US immigrant population. However, the quality of this research, which has the potential for informing cancer etiology and control, is unclear because registry birthplace information is incomplete, and its accuracy has not been examined. We quantified misclassification of birthplace data for Asian cancer patients in the Greater Bay Area Cancer Registry in northern California by comparing registry birthplace information with self-reported birthplace from interview, and then identified sociodemographic and hospital characteristics associated with birthplace completeness and misclassification. Of the 1836 eligible Asian patients, 649 (35%) had unrecorded registry birthplace. For all except Vietnamese, these persons were less likely than those with recorded birthplace to be foreign-born (OR = 0.5, 95% CI = 0.4-0.7), to be diagnosed in public than private hospitals (OR = 0.7, 95% CI = 0.5-0.8) and in teaching than non-teaching hospitals (OR = 0.8, 95% CI = 0.6-1.1), and were more likely to have been diagnosed at a large regional health maintenance organization (OR = 1.7, 95% CI = 1.3-2.2) and after 1995 (OR = 1.6, 95% CI = 1.1-2.1). Among Asians with registry birthplace information (n = 1187), sensitivity and predictive value positive for birthplace exceeded 90% for both US- and foreign-born, except for Japanese (predictive value positive = 85.7%). Among US-born Asians, those misclassified as foreign-born were more likely than those correctly classified to prefer a non-English primary language (OR = 29.4, 95% CI = 1.9-459.9). These results suggest that cancer registry birthplace data for Asians should not be used if they continue to be differentially incomplete for a large proportion of the subjects.

Publication types

  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Adult
  • Aged
  • Asian Americans*
  • Bias*
  • California / epidemiology
  • Cohort Studies
  • Emigration and Immigration
  • Female
  • Humans
  • Language
  • Male
  • Middle Aged
  • Neoplasms / epidemiology*
  • Neoplasms / ethnology*
  • Odds Ratio
  • Registries / standards*
  • Registries / statistics & numerical data*
  • Reproducibility of Results
  • Residence Characteristics / statistics & numerical data*
  • SEER Program / statistics & numerical data*