Color-blind and classless: The bias problems with health risk appraisals

U.S. racial/ethnic health disparities have been documented over the past decades, yet disparities persist despite national/regional efforts to improve care access and quality. Health risk appraisals (HRA) can be important tools in assessing the impact of risk factors 1) potentially under the control of the individual, and 2) that may be amenable to social, public health, medical or pharmacological interventions. HRA's can provide patients with specific information about their health risks, what they can do about them, and give healthcare providers a tool to improve communication with patients regarding health, risks and prevention issues. Conspicuously missing from all or nearly all HRA's available today is any use of race/ethnicity or socioeconomic status (SES) in the actual risk calculation engines. One design group commented that attempts to create a better fit by creating race-specific values for use in the risk engines generated unreasonably different estimates for males of different races with otherwise identical characteristics. This suggests that current risk algorithms in most HRA's 1) lack statistical precision and validity across racial/ethnic groups and SES, 2) under-estimate the risk levels for African Americans, and 3) fail to transparently identify health disparities. As an example, we have documented these statements by empirically verifying the exclusion of race/ethnicity in risk calculations in HRA's used by the CDC, Marine Corp, Healthier People Network, Dr. Oz's Real Age and several others. We suggest possible solutions to this bias that include a rigorous review and incorporation of evidence-based research on risks related to race/ethnicity/SES and inclusion of personalized genetic information in HRA's.