Addressing health disparities – why minority recruitment isn’t enough

Fewer births, declines in mortality related to cardiovascular disease and cancer, and international migration are contributing to unprecedented demographic shifts in the U.S. Increasingly the population is older and more diverse and in the next two decades, racial and ethnic minorities will comprise 42% of older adults. Without intervention, existing health disparities among minority older adults are likely to be exacerbated. The pressing need to address these health disparities is perhaps most evident in Alzheimer disease (AD), which increases in prevalence with age and is disproportionately higher among African Americans and Hispanics/Latinos.

AD prevalence is expected to triple by 2050 and threatens to overwhelm the health system unless significant advances are made. Despite numerous failures of AD drug trials, there remains optimism that some AD treatments may be effective, if started very early, likely in the preclinical (asymptomatic) period. Diagnosis of preclinical AD currently relies on cerebrospinal fluid (CSF) biomarkers and/or molecular or functional neuroimaging, and much of the evidence supporting this approach was generated from studies with fewer than 5% minorities.¹ However, recent studies have found racial differences – lower CSF tau² and higher uptake of radioligands for amyloid³ in African Americans compared to Whites – highlighting the possibility of worsening disparities if scientific evidence is generalized to populations not included in the research.

These reported racial differences in AD studies have heightened calls for more minority participation in research, which remains low despite the decades-old National Institutes of Health (NIH) guidelines on inclusion of minorities in clinical research and more recent NIH requirements that Phase III clinical trials include valid analyses by race. Researchers are keenly aware of the imperative to increase minority enrollment, but lack of trust, prior research abuses, and ineffective communication with minority communities contribute to low participation. Despite the barriers, successful minority recruitment has occurred using strategies such as hiring staff experienced working with diverse communities, partnering with community organizations, and using communication channels known to reach diverse communities.

While inclusion of minorities in research is critical, minority recruitment alone is unlikely to provide the evidence needed to understand and address health disparities. Many studies have found an association between AD and years of education, quality of education, socioeconomic status, and neighborhood characteristics. Thus, AD studies should be expected to consider these contextual and life experiences, especially among racial and ethnic minorities who disproportionately are more likely to be socioeconomically disadvantaged. Many studies comparing groups by race use few variables to adjust for socioeconomic factors (e.g. years of education and income level) and these insufficiently capture important differences in life experiences between races that could impact risk of disease. Unfortunately, analyzing findings by race without appropriate contextual data could lead to inaccurate, misleading, or stigmatizing conclusions that may detract from the overall goals of diversity in research – to enhance the accuracy, utility, and generalizability of scientific evidence.

Race, as defined by the U.S. Census and Office of Management and Budget (OMB), is a sociocultural construct that is not biologically, anthropologically, or genetically based. Because race is socially determined, dependent on self-identification and typically associated with physical characteristics such as facial features, hair texture and skin color, it is a fluid construct that can change over time and vary by location and culture. Studies of race and ethnicity demonstrate greater genetic heterogeneity within than between racial/ethnic groups,^4,5 yet there is a common misperception that racial groups are more genetically homogeneous.

Although substantial evidence exists to contradict claims that race is biologically based, and without evidence that genetic or biological characteristics can be inherently structured into racial categories,⁶ long-held assumptions about innate biological differences between races continue to permeate medicine and biomedical research. This is in part because minority race is tightly interwoven with sociopolitical factors such as systemic racism, discriminatory policies, and access to care, which are linked to biophysiological changes such as neuroendocrine dysregulation, cellular aging, and elevated inflammatory cytokines.⁷ These biological responses to social factors can contribute to premature morbidity and are not unique to minorities, but are more prevalent due to the higher burden of psychosocial stress across the lifespan.

Race, particularly in the U.S., is associated with markedly different life experiences, including gross social injustices, that affect health. Painful legacies of overt racism, abuse, and trauma continue to negatively impact minority communities. Native Americans experienced historical traumas including exploitation and forced removal from their land, resulting in a ripple of economic, cultural, and social losses that continue to disadvantage their communities today. Such injustices are not limited to the distant past. In the 1950s and 1960s, most schools in the Southern U.S. were racially segregated and many public places, including hospitals, had separate entrances for African Americans, if they were allowed to enter at all. There were substantial differences in the quality of the education at schools for African Americans in the South, many of which operated on split terms to allow two months off in the Fall to pick cotton. In 2002, Unequal Treatment, the landmark Institute of Medicine consensus report, found that racial and ethnic minorities in the U.S. receive lower quality health care even when adjusting for insurance status, income, age, and severity of conditions. Racial bias, prejudice, and stereotyping by healthcare providers were found to contribute to these differences in care.

To mitigate health disparities, more comprehensive research assessments are needed to disentangle the myriad of factors contributing to them. Researchers must begin with sufficient numbers of racial and ethnic minorities, and ideally, the study sample should reflect the population of people with the disease. For example, if Hispanics/Latinos have higher incidence of AD, that should be expected in the study. Race and ethnicity should be collected by self-report using questions that allow individuals to identify as multiracial or multiethnic. Because only 22% of older adults born outside of the U.S. speak English, language spoken at home and English proficiency should also be collected. As important, researchers should collect contextual and environmental information that may explain or be associated with racial differences. See Box 1. for variables that would be useful in assessing individuals in an AD study.

Adapting and expanding data collection to more accurately assess environmental and sociocultural factors affecting health is critical not only for addressing health disparities, but for science more broadly. Less than 25% of the global population is of European ancestry, thus studies primarily comprised of Whites make lack generalizability, particularly if groups with the highest disease burden are excluded. Basic scientists and clinical investigators must learn to consider and identify social determinants of health and integrate these factors into hypothesis generation and study design. Social scientists and individuals experienced with minority health should become valued members of clinical research teams, not because they will help with minority recruitment, but because they will improve the quality of the science. By increasing the diversity of research participants, and integrating the contextual information needed to understand diverse life experiences, we will finally begin to address the racial differences in health and disease.

Box 1.

Key sociodemographic information to understand social and environmental factors relevant Alzheimer’s disparities.

Demographic and Social Information to Collect in an Alzheimer’s Study.
1. Race/ethnicity, self-reported+  Which of these describe you? Select all that apply.  ● American Indian or Alaska Native (For example: Aztec, Blackfeet Tribe, Mayan, Navajo Nation, Nome Eskimo Community, etc.)  ● Asian (For example: Asian Indian, Chinese, Filipino, Japanese, Korean, Pakistani, Vietnamese, etc.)  ● Black, African American, or African (For example: African American, Ethiopian, Haitian, Jamaican, Nigerian, Somali, etc.)  ● Hispanic, Latino, or Spanish (For example: Colombian, Cuban, Dominican, Mexican or Mexican American, Puerto Rican, Salvadoran, etc.)  ● Middle Eastern or North African (For example: Algerian, Egyptian, Iranian, Lebanese, Moroccan, Syrian, etc.)  ● Native Hawaiian or other Pacific Islander (For example: Chamorro, Fijian, Marshallese, Native Hawaiian, Tongan, etc.)  ● White (For example: English, European, French, German, Irish, Italian, Polish, etc.)  ● None of these fully describe me (display optional free text)  ● Prefer not to answer
2. Primary language spoken at home (or preferred language)
3. Education: total years of education; school characteristics (public vs private, rural vs urban vs suburban); parents’ total years of education
4. Annual household income (current and at age 40)
5. Perceived social class: occupational prestige, housing type, sources of income
6. Neighborhood characteristics: walkability, availability of healthy foods, social cohesion, and neighborhood violence*
7. Perceived discrimination: 9-item Everyday Discrimination Scale**

⁺A single question for race and ethnicity minimizes missing data from individuals who do not identify a race and acknowledges ethnicities other than Hispanic/Latino.

^*Christine PJ et al. JAMA Internal Medicine. 2015; 175(8): 1311–1320

^**Williams DR et al. Journal of Health Psychology. 1997;2:335–351.