Data quality and adjusted Hispanic mortality in the United States, 1989-1991.

OBJECTIVES Hispanics appear to live longer compared to other ethnic groups in the United States. Our main objective was to determine whether data quality biases mortality statistics. We calculated the impact of misclassification of ethnicity on death certificates in order to create adjusted mortality estimates. METHODS We used the National Mortality Follow-Back Survey of 1993 (NMFS) for our assessment of ethnicity misclassification. We then created misclassification estimates for 10-year age-sex groups, and used these to correct mortality estimates for 1989-1991. RESULTS The overall predictive value positive (PV+) and sensitivity were 0.981 and 0.805, respectively, for men; and 0.994 and 0.902, respectively, for women. Age-specific adjustment for misclassification on death rates caused Hispanic male life expectancy to go from a 1.01 year advantage over White non-Hispanics, to a 1.83 year deficit, with a similar pattern found for females. CONCLUSIONS The apparent Hispanic advantage in life expectancy is influenced by misclassification of ethnicity. Misclassification of ethnicity on death certificates biases Hispanic mortality downward, thereby falsely inflating life expectancy.