Table 6.
Accuracy measures among CERC survey respondents from the adult sites (N = 9, 185). Frequencies (Nc) and prevalence estimates (P) were computed using self-reported data, and accuracy measures (Se, Sp, PPV, NPV)a were computed by comparing self-reported data (ie, gold-standard) to the EHR and/or USC-derived data. These summaries are also presented by the data source used to define the sampling strata (EHR, USC, or combined)b
| Data Source Used to Define Sampling Strata (Adult Respondents, N = 9185) |
||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
EHR data Only |
USC data Only |
EHR and USC data (combined) |
||||||||||
| Nc | p | Se/Sp | PPV/NPV | Nc | p | Se/Sp | PPV/NPV | Nc | p | Se/Sp | PPV/NPV | |
| Age | ||||||||||||
| < 35 years | 2211 | 24.8 | 99/97 | 91/100 | – | – | – | – | 2211 | 24.8 | 99/97 | 91/100 |
| ≥ 35 years | 6690 | 75.2 | 97/99 | 100/91 | – | – | – | – | 6690 | 75.2 | 97/99 | 100/91 |
| Gender | ||||||||||||
| Female | 5042 | 56.0 | 97/98 | 99/97 | – | – | – | – | 5042 | 56.0 | 97/98 | 99/97 |
| Male | 3969 | 44.0 | 98/97 | 97/99 | – | – | – | – | 3969 | 44.0 | 98/97 | 97/99 |
| Race | ||||||||||||
| White | 4210 | 48.5 | 78/92 | 90/82 | 142 | 53.2 | 62/51 | 59/54 | 4352 | 48.7 | 77/91 | 89/81 |
| African American | 989 | 11.4 | 94/97 | 79/99 | 15 | 5.6 | 60/82 | 17/97 | 1004 | 11.2 | 93/96 | 76/99 |
| Asian | 1440 | 16.6 | 85/96 | 83/97 | 10 | 3.7 | 70/93 | 28/99 | 1450 | 16.2 | 85/96 | 82/97 |
| AI/AN | 240 | 2.8 | 82/94 | 30/100 | 2 | 0.7 | 50/100 | 100/100 | 242 | 2.7 | 82/95 | 30/100 |
| NH/PI | 103 | 1.2 | 71/97 | 21/100 | 0 | 0 | −/100 | −/− | 103 | 1.2 | 71/97 | 21/100 |
| Other | 1692 | 19.5 | 35/88 | 42/85 | 98 | 36.7 | 10/83 | 26/62 | 1790 | 20.0 | 33/88 | 42/84 |
| Ethnicity | ||||||||||||
| Not Hispanic /Latino | 6697 | 80.8 | 90/89 | 97/69 | 495 | 85.5 | 64/73 | 93/26 | 7192 | 81.1 | 89/88 | 97/65 |
| Hispanic /Latino | 1594 | 19.2 | 89/90 | 69/97 | 84 | 14.5 | 73/64 | 26/93 | 1678 | 18.9 | 88/89 | 65/97 |
| Education | ||||||||||||
| < HS | – | – | – | – | 670 | 7.6 | 21/91 | 17/93 | 670 | 7.6 | 21/91 | 17/93 |
| HS + some college | – | – | – | – | 3415 | 38.9 | 66/57 | 49/72 | 3415 | 38.9 | 66/57 | 49/72 |
| ≥ Bachelor’s | – | – | – | – | 4684 | 53.4 | 52/78 | 73/59 | 4684 | 53.4 | 52/78 | 73/59 |
| Ruralityc | ||||||||||||
| Rural | – | – | – | – | 3919 | 42.7 | – | – | 3919 | 42.7 | – | – |
| Suburban/Urban | – | – | – | – | 5266 | 57.3 | – | – | 5266 | 57.3 | – | – |
Nc = number who completed the survey item, Se = sensitivity, Sp = specificity, PPV = positive predicted value, NPV = negative predictive value, AI/AN = American Indian/Alaska Native, NH/PI = Native Hawaiian/Pacific Islander, HS = high school.
Accuracy measures are based on the subset of respondents who provided self-reported data (ie, those who complete the demographic questionnaire item of interest). For example, 284 = 9185—2211—6690 respondents did not complete the survey item related to age.
To define sampling strata, age and gender were based on EHR data for all, as availability of both in the EHR was an inclusion criterion, and education and rurality were based on USC data for all. For race and ethnicity, data presented in the USC Only column correspond to the subset of survey respondents who did not have race and/or ethnicity data recorded in the EHR. For this subset, USC data were used to define their sampling strata.
Accuracy estimates were not calculated for rural living, as the true value is based on the address and not on a participant response.