Table 1.
Agreement Between E-Tech Estimate of Race and Self-reported Race for PRISM Study Particpants (n = 3375), North Carolina, 2005a
| Category | Black | White | American Indian,Asian,or Native Hawaiian | Total |
|---|---|---|---|---|
| Self-reported race | 365 | 3010 | 0 | 3375 |
| E-Tech–estimated race | 245 | 3106 | 24 | 3375 |
| Correctly identified | 174 | 2916 | 0 | 3090 |
| Incorrectly identifiedb | 71 | 190 | 24 | 285 |
PRISM indicates Personally Relevant Information About Screening Mammography.
E-Tech did not identify race codes for 115 of the 3490 PRISM study participants who self-reported black or white, resulting in 3375 total participants. Sensitivity of E-Tech estimates of black participants = 47.7% (174/365); positive predictive value of E-Tech estimates of black participants = 71.0% (174/245); specificity of E-Tech estimates of white participants = 96.9% (2916/3010). κ = 0.53; 95% confidence interval, 0.48–0.58.
Of the 71 participants who were incorrectly identified as black, all self-reported as white. Of the 190 participants who were incorrectly identified as white, all self-reported as black. Of the 24 participants incorrectly identified as American Indian, Asian, or Native Hawaiian, 1 self-reported as black and 23 self-reported as white.