Assessing an electronic self-report method for improving quality of ethnicity and race data in the Veterans Health Administration.

Abstract

Evaluate self-reported electronic screening (eScreening) in a VA Transition Care Management Program (TCM) to improve the accuracy and completeness of administrative ethnicity and race data. We compared missing, declined, and complete (neither missing nor declined) rates between (1) TCM-eScreening (ethnicity and race entered into electronic tablet directly by patient using eScreening), (2) TCM-EHR (Veteran-completed paper form plus interview, data entered by staff), and (3) Standard-EHR (multiple processes, data entered by staff). The TCM-eScreening (n = 7113) and TCM-EHR groups (n = 7113) included post-9/11 Veterans. Standard-EHR Veterans included all non-TCM Gulf War and post-9/11 Veterans at VA San Diego (n = 92 921). Ethnicity: TCM-eScreening had lower rates of missingness than TCM-EHR and Standard-EHR (3.0% vs 5.3% and 8.6%, respectively, P < .05), but higher rates of "decline to answer" (7% vs 0.5% and 1.2%, P < .05). TCM-EHR had higher data completeness than TCM-eScreening and Standard-EHR (94.2% vs 90% and 90.2%, respectively, P < .05). Race: No differences between TCM-eScreening and TCM-EHR for missingness (3.5% vs 3.4%, P > .05) or data completeness (89.9% vs 91%, P > .05). Both had better data completeness than Standard-EHR (P < .05), which despite the lowest rate of "decline to answer" (3%) had the highest missingness (10.3%) and lowest overall completeness (86.6%). There was strong agreement between TCM-eScreening and TCM-EHR for ethnicity (Kappa = .92) and for Asian, Black, and White Veteran race (Kappas = .87 to .97), but lower agreement for American Indian/Alaska Native (Kappa = .59) and Native Hawaiian/Other Pacific Islander (Kappa = .50) Veterans. eScreening is a promising method for improving ethnicity and race data accuracy and completeness in VA.