Strategies to Increase Under-represented Minorities in Pharmacy Education Take Time to Yield Results

Introduction

From 2007-2012, the Office of Recruitment, Development, and Diversity Initiatives (ORDDI) at the UNC Eshelman School of Pharmacy conducted various integrated recruitment events to diversify the student body of their school. Students who participated in these ORDDI events were considered part of the “ORDDI cohort”.

After implementing these programming changes, ORDDI reported that about 30% of admitted students each year between 2007 to 2012 had participated in their programming. 80% of prospective students who participated in their programming and applied were admitted, but less than 40% of these prospective students came from minority backgrounds. However, among admitted students, there were significantly more Black students among the ORDDI cohort than the general student body. Still, although the average percent of White students at the school decreased almost 10% after implementation of ORDDI and the average percent of minority students increased by 6%, the majority of that increase was found among Asian American students–who are, as a group, well-represented in higher education–and among students identifying as other or unknown.

A diverse health workforce, including among pharmacists, is important to enhance trust between providers and patients. This is especially true because racial demographics are shifting in the US towards a majority minority composition, and minority populations already receive less healthcare and feel less satisfied by the quality of healthcare they receive. Intentional efforts to recruit, retain, and develop underrepresented minorities to enroll and succeed in pharmacy education is crucial to diversifying the workforce. 

The authors were all affiliated with the UNC Eshelman School of Pharmacy, and the first author has gone on to become a dean at the school.

Methods and Findings

The ORDDI began its work in 2007 by defining 8 strategic goals related to leadership and accountability, targeting recruitment efforts and creation of a pipeline, and connecting current students with prospective students. Specifically, ORDDI programs included:

• entrance test prep for prospective students, 
• leadership and mentorship programs to connect current students with prospective ones, 
• opportunities to introduce pharmacy careers to more students and to strengthen the pipeline of high school and undergraduate students interested in pharmacy careers
• educating high school administrators and teachers about pharmacy education and careers so they can educate their students, 
• creation of new external communications, 
• professional development events to retain engaged students.

Primary findings:

• Over 6 years, the ORDDI facilitated 812 recruitment events, and 80% of students who participated in ORDDI events (the ORDDI cohort) and applied to UNC Eshelman were admitted. 
• On average, the ORDDI cohort represented between 22-39% of all admitted students in a year and 20-56% of the ORDDI cohort were from minority backgrounds. 
• The ORDDI cohort contained about 11% more Black students than the student body as a whole and 80% of Black students interfaced with the ORDDI. 
• However, although the average percent of White students dropped from 79% to 68% of the student body since ORDDI began recruitment, the average percent of Black students (7% to 6%), Indigenous students (<1% to <1%), and Latinx students (1% to 2%) stayed largely the same, with only the Asian/Pacific Islander (10% to 16%) and Other/Unknown (2% to 8%) student populations increasing. 
• All together, this represents an increase from 19% minority to 25% minority students, but only when counting Asian American students, who as a whole are not underrepresented in higher education.

Conclusions

The authors recommend centralizing recruitment efforts, in order to maximize the efficiency of personnel and the intensity of efforts. They point to the need for committed resources, including infrastructure, leadership, accountability, and vision, to improve diversity and inclusion of underrepresented students. In the short term, their results were mixed in increasing the recruitment, admission, and retention of students from under-represented backgrounds, particularly for Black students. Thus, the authors acknowledge that this type of recruitment effort takes time and continued effort. However, the high level of engagement with ORDDI activities and staff among Black students point to the additional intangible benefits that students derive from having dedicated resources for diversity, inclusion, and belonging on campus.

Algorithms Can Replicate or Remedy Racial Biases in Healthcare Resource Allocation

Introduction

Obermeyer et al. note both the growing attention to potential racial and gender biases within algorithms and the difficulty of obtaining access to real world algorithms – including the raw data used to design and train them – in order to understand how and why bias could appear in them. This study is important because it has obtained access to the inputs, outputs, and real world outcomes of a health care algorithm that performs a widely used function within the healthcare sector. Further, it is widely representative of the type of logic used by algorithms in other social sectors. In particular, this algorithm identifies which patients to recommend for a care management program where they will receive additional resources. The algorithm simplifies this task into identifying the patients with the greatest care needs, patients at the top 3 percentile of need automatically qualifying for entry into the program, and those at the top 45 percentile of need assessed for entry by their primary care physician.

As prominent researchers at the nexus of machine learning and health, the researchers were able to convince the manufacturer of this algorithm, a leader in the field, to consider changing its algorithm, with the hope that this may change the norms within the entire sector.

Methods and Findings

Obermeyer et al. collected input data of all primary care patients enrolled in risk-based contracts at a large academic hospital from 2013-2015. They defined the population of Black and white, non-Hispanic patients based on patient self-identification. The researchers also used outcomes from electronic health records to assess patients’ health needs and insurance claims to assess patients’ costs including: all diagnoses, key quantitative laboratory studies, vital signs, utilization, outpatient visits, emergency visits, hospitalizations, and health care costs.

Primary findings:

• Based on the number of comorbid conditions and severity of markers of chronic disease, Black patients with the same level of predicted risk as white patients, according to the algorithm, were substantially sicker than their white counterparts. Thus, if the algorithm identified high risk patients solely on health needs, significantly more Black patients should have been included in the care management program. 
• Previous health care costs were the driving force behind Black patients’ lower entry to the care management program. For unknown reasons, the gap between care needed and care received among Black patients was significantly larger. Thus, a seemingly neutral factor of previous health care spending could turn into a racially biased one, due to external social conditions. This demonstrates the difficulty of “problem formulation” in data science – how to turn complex, interactive, and vague social conditions into a concrete, measurable variable in a dataset.
• The study explores other ways to define the input variables that the algorithm considers when distinguishing patients’ health risks, including considering only total costs, avoidable costs (based on emergency visits and hospitalizations), and health needs (based on number of chronic conditions). These three options all do fairly well in predicting patient outcomes, but considering health needs results in almost twice the number of Black patients that make it into the highest risk group compared to considering only total costs. 
• Doctor judgment can marginally increase the number of Black patients that make it into the care management program compared to the cost-based algorithm. But, an algorithm that is adjusted to take health needs into consideration is even better than doctors at identifying high-risk Black patients. Thus, an improved algorithm has greater potential for improving the ratio of Black patients who are rated as “high-risk” than relying on individual doctors’ judgement.  
• The manufacturer of this algorithm independently confirmed Obermeyer et al.’s findings. Working together, the researchers and manufacturer demonstrated that adjusting the algorithm to take health needs into account reduces racial bias in outcomes by 84%.

Conclusions

Obermeyer et al. recommend caution in identifying and defining what measures algorithms are trained to look for to ensure that the algorithm collects truly relevant input and output data for health outcomes. Although this seems difficult and costly, the private sector has clearly demonstrated that it is possible, and technically the additional labor is merely in validating the conceptual relationships within the algorithm rather than the statistical technique it relies on. With the high stakes involved in the health and social sectors, this kind of investment is necessary, and can yield great results while minimizing harm. 

With the start of a partnership with the manufacturer of this algorithm, Obermeyer et al. hope to find solutions to this type of error together, and with the combined leadership in academia and industry, they hope that their findings could change the norms used to create algorithms within the healthcare sector and wider social sectors.

Unconscious Bias Training Doesn’t Work

Introduction

Unconscious bias training is based on the methodology in social psychology that an individual’s response time when presented with 2 images reveals how closely the viewer unconsciously connects the two. By using two sets of images — one of people with different racial/ethic backgrounds and one of negative or positive attributes, prior researchers have demonstrated that most people respond faster to the combination of positive associations with white people and negative associations with Black people. Since the introduction of “unconscious bias” in the 2000s, a new form of diversity training has emerged in response. These trainings aim to support people in acknowledging their own unconscious racial biases, with the understanding that awareness of how racism impacts your behavior and decisions at an unconscious level is the first step in changing behavior. However, this type of “training” is not evidence-based, and there is not yet enough evidence to show that it has any outcomes related to diversity, racism, or bias. 

Methods and Findings

This author suggests that the lack of outcomes from unconscious bias trainings may be due to the challenges of transferring a psychological concept into a sociological training setting without addressing the implications and underlying theory and assumptions. Many of unconscious trainings are predicated on the assumption that knowing you have a bias means that you will change your behavior – for example, the author points to one study that showed that unconscious bias training resulted in organizational backlash rather than prompting antiracist change. In addition, the authors note, much of the discourse on unconscious bias ignores the breadth and impact of institutional racism and focuses instead on solely interpersonal racism.

Conclusions

Unconscious, individual bias is a big problem, but unconscious bias training is not solving it,despite the catchy name; these trainings fail to acknowledge group dynamics and structural racism. There is no evidence that knowledge about bias results in behavior change. and the design of unconscious bias training does not inspire the type of self-reflection that would be needed to affect behavioral change. Lastly, the approach itself does not have clearly defined tools and techniques nor robust evaluation mechanisms to gauge outcomes.

Additional critical examination is needed into the methods and outcomes of unconscious bias training, including accumulation of sufficient data to support its underpinnings. This article brings much needed critique of “mainstream” fads in anti-racism training, particularly the crucial point that even if unconscious bias training works in some contexts, it is explicitly limited to individual racism and entirely ignores (and thus absolves) institutional, structural, and organizational racism. The author notes that even if the stated goal of unconscious bias training — to nudge individuals to recognize that they are biased — was successful, the approach also does nothing to motivate people to change or show them what they could do differently to enact more equitable behavior and a more equitable workplace.

Implicit Racial Bias is Malleable but Stable in Young Adults

Introduction

Recent research on implicit social cognition suggests that implicit associations may be malleable to change. However, the majority of studies on modifying implicit associations only evaluate short term results, with only 3.7% of these 585 studies attempting to look at longer-term change. Of these 22 studies, roughly the same number of publications showed lasting impact, no change, or mixed results. The study reviewed here attempts to shed light on these conflicting data points by systematically evaluating 9 interventions that had previously successfully changed short-term implicit bias and also translated to long term change hours and days afterward.

The researchers found that even when interventions modify implicit racial bias immediately they do not maintain this change days after the intervention. Nor do they impact explicit racial preferences, motivation to reduce prejudice, or support for affirmative action. However, the authors point to other research that suggests that implicit bias reduction interventions may be more effective at producing lasting change when they are introduced earlier in childhood.

This is an important finding that should guide the overall strategy of antiracist training and education. Instead of attempting to change implicit bias as a means for changing explicit racial preferences in adults, strategists should aim to bring implicit bias reduction interventions to settings where children learn their values, such as schools and the elementary school level curriculum.

Methods and Findings

The authors tested 17 interventions aimed to change implicit racial bias and measured whether they were successful. They compared this to a control group. Success was defined as significantly different reaction times in the Implicit Association Test before and after the intervention.

The researchers then conducted a second round of experiments among mostly white, female college undergraduates in the US. Using 9 of the previously successful interventions they tested for short-term effectiveness immediately after and again anywhere from 1 to 4 days. In addition to measuring implicit racial bias through the race implicit association test, researchers also evaluated self-reported explicit racial preferences. Finally, they also measured support for affirmative action and internal and external motivations to respond without prejudice. In the first round of experiments, researchers recruited ~ 1,000 participants, then repeated this experiment with a larger set of ~5,000 participants.

In the first study, although half of the 9 previously successful interventions significantly changed participants’ implicit racial bias immediately afterward, none continued to show this effect 2-4 days later in the follow-up assessment. Similarly, in the second study, 8 of the previously successful interventions significantly changed participants’ implicit racial bias immediately afterward, but none continued to show this effect 1-2 days later in the follow-up. None of the 9 interventions impacted explicit racial preference immediately or in the follow-up in either study. Only 25% of participants supported affirmative action in the workplace (6% supported it in higher education), and this support was not related to their implicit or explicit racial preference. 

Overall, participants were motivated to respond without prejudice both based on internal values and external pressure, but their motivation did not impact their implicit racial bias. Of note, participants with high internal motivation had lower implicit and explicit pro-white/anti-Black preferences, while participants with high external motivation had higher implicit and explicit pro-white/anti-Black preference. Researchers also found that the proportion of white and Black students on campus weakly correlated with higher implicit and explicit pro-white/anti-Black preference, which echoes recent findings that people in states with a higher proportion of Black residents tend to show more bias on the implicit association test.

Conclusions

This research that even when interventions change implicit bias, they are ineffective at maintaining this change even hours or days afterward. “Malleability” of implicit preference in the short term does not necessarily lead to lasting change. This is supported by other recent studies in developmental psychology that show that children learn implicit preferences for their own social groups (whether race, gender, religion, etc) within the first year of life, and these preferences are stable throughout development. The researchers also conclude that changing implicit bias does not impact explicit racial preferences, explicit support for affirmative action, or internal/external motivation.

This work considers whether it is truly possible to effect long term change in implicit preference, and poses the possibility that implicit bias reduction interventions may need to be repeated and/or may need to occur earlier in childhood development in order to impact long term implicit preferences. This insight suggests that future studies should prioritize the evaluation of longer exposures to implicant bias modification and examination of the impact of implicit bias reduction interventions in children. This recommendation is significant because it hones in on the importance of early intervention in childhood – and thus the importance of bringing challenges to implicit bias into the classroom and other settings tailored tow

Beyond Equal Access

Introduction

There is a large, consistent, and persistent gap in many different indicators of health status, across the lifespan, between Black and white populations. The author attempts to break this problem down to its root cause: first, the author notes that while it may be possible that there could be biological or genetic differences between Black and white populations, it is highly unlikely that this is the only reason for the gap in health outcomes. Next, the author notes that while Black patients receive lower-quality healthcare than white patients, this also does not explain the gap in health outcomes, since it ignores the fact that health status is also impacted by individuals’ socioeconomic and psychological factors. Finally, the author argues there are two ways that structural racism impacts the effect of socioeconomic and psychological factors on Black people’s health: “differential exposures” (lower access to protective socioeconomic and psychological advantages and higher amounts of health risks) and “differential effects”, or “diminished gain” (less “bang for the buck” out of the resources that they do have: the same economic, social, and/or psychological resources are less protective for Black people than for white people). The author points to evidence from large national surveys with multiple years worth of data from thousands of individuals over a wide range of participant age to show that typically protective social factors (employment, education, social networks) and psychological factors (self-efficacy, sense of control over life) have less impact on extending life expectancy or preventing premature mortality among Black patients than white ones. 

Thus, the key finding of this work is that differential exposure to risk and protection doesn’t fully explain the health gap between Black and white people: you can’t “solve” the gap just by removing socioeconomic differences, segregation, or discrimination because the evidence shows that improving socioeconomic status has less impact on health for Black patients than white ones. Social class and economic status modify the impact of race on health; but their intersection also amplifies the Black-white health gap by creating additional barriers to protective psychosocial resources. Secondly, you can’t “solve” the Black-white health gap by only focusing on the healthcare system because health is impacted by social and economic structure. Thus a broader, sociological approach to health disparities is needed. Using these insights, the author poses several social and economic policies that follow from their research. First, they note that income is not as vulnerable to diminished gain, so income redistribution is a crucially important policy strategy. Similarly, diminished gain shows that policies that aim for equality in access will not go far enough to achieve equality in outcomes, so there is a need for policies that directly target barriers at multiple levels and leverage protective factors, like religion and social support. At the same time, policies that expand the Black-white health gap must be avoided and policies that police discrimination must be enforced. 

This article brings together evidence to prove that the Black-white health gap results from social processes that prevent Black populations from fully realizing the protective potential of their resources, rather than individual behavior choices. This means that the health gap can be changed if there is a concerted push to address structural racism in US institutions. Dr. Assari is well-published in the effects of race, ethnicity, gender, and place on health consequences and is a leader in health equity research. He continues to push the research agenda on the impact of social and economic policies on attaining health equity, and also illuminates the complex interplay of race, gender, and class on health.

Methods and Findings

The research team observed data from large national surveys and cohorts: study duration ranged from 2 years to 25 years; study size ranged from 1,500 adults to more than 37,000 adults; and study populations ranged from adolescents to older adults.

The author’s hypothesis is: if Black people are less protected by socioeconomic class, then white people who lose socioeconomic resources should go through a bigger health loss than comparable Black people. The author used historical analyses to show that the Black-white health gap increases as a direct consequence of widening economic (income and wealth) disparities. The author considered five interconnected mechanisms to explain different determinants of “diminished gain” impacts on health, including structural racism in the labor market, purchasing power, chronic exposure to discrimination, cumulative disparities starting with an initial advantage, and cost of upward social mobility.

Conclusions

The author recommends using the evidence of diminished gain to drive development of social and economic policies and to guide future research. The research evidence shows that income demonstrates less effect of diminished gain compared to other socioeconomic and psychological factors, so income redistribution should be a central policy strategy to address diminished gain. Similarly, policies should leverage religion and social support because the evidence shows that they are particularly protective for Black populations. Diminished gain also separates access from outcomes: equal access does not go far enough to close the Black-white gap in health outcomes, so policies must go further to also remove barriers. This includes eliminating policies that widen the Black-white health gap and enforcing policies that aim to police discrimination. Finally, it is crucial that policies target individual, organizational, and institutional racism.

Strategies for income redistribution include raising the minimum wage for jobs typically occupied by BIPOC; closing the racial wage gap; and instituting tax policies that allow low income families to build wealth. Cash assistance and temporary financial incentives may help with deep poverty. Strategies for going beyond equal access include tailoring interventions and programs to address the specific needs of BIPOC communities. This means specifically removing structural and societal barriers to improving health outcomes, which supports the need for affirmative action. The author also points to the impact of the education, employment, housing, criminal justice, and economic (banking and lending practices) sectors on health outcomes.

The author concludes that because Black populations face more social and economic adversity, they have developed increased resilience to additional economic and psychological risk factors, unlike white populations. Thus, Black populations’ poor health outcomes result from their consistent exposure to many contemporaneous risk factors and systematic exclusion from protective factors. The author emphasizes the need to identify, measure, and mitigate health inequities across population groups and to conduct further research to separate the effect of culture and societal structure on creating and reinforcing health inequities. In addition, the author argues for the need for an intersectional framework to guide future research in examining and explaining differences in health status and health risks within Black populations. Finally, the author notes that although the totality of the body of research he draws on is sufficient to substantiate his argument, he points out that the individual studies are correlational not causal (due to ethical concerns). The author suggests that further studies that use more sophisticated analytical methods may further strengthen the argument.