Building Better Medicine: Translational Justice and the Quest for Equity in US Healthcare

Abstract

Despite considerable scientific progress and the evolution of regulatory pathways to ensure safety and efficacy, US healthcare continues to see increasing health disparities. This suggests that clinical translation in of itself cannot be the only measure of its own success, especially when the most marginalized patients, are neglected in the development and implementation of medical innovations. This raises the question of whether a system that is narrowly focused on technical achievement can meet the moral obligations of medicine and public health. We argue that traditional technocratic standards are failing to integrate normative considerations into biomedical translation. What is needed is a translational domain that moves beyond safety and efficacy toward anticipating how proposed technologies will be effective in society as it exists. We propose an additional metric of success: translational justice.

“Translation is the process of turning observations in the laboratory, clinic and community into interventions that improve the health of individuals and the public—from diagnostics and therapeutics to medical procedures and behavioral changes.”

—National Center for Advancing Translational Sciences¹

INTRODUCTION

The 20th century in the United States (U.S.) saw a concerted effort from national actors, most notably the National Institutes of Health (NIH), to accelerate the pace of clinical translation of scientific advancements. The Bayh-Dole Act of 1980 was designed to incentivize clinical translation of basic biomedical science toward actionable medical interventions by providing U.S. academic institutions and researchers with financial interests toward commercialization (Sampat 2006). Clinical and Translational Science Awards were likewise designed to accelerate the translation of innovative research into viable clinical and commercial products (Kim, Lee, and Woo 2023). Translational science includes multiple stages of evidence generation and review, beginning with initial trial investigations focused on technical issues of safety/efficacy or validity/utility. If the technology progresses it will be subject to a comprehensive review of the outcomes by the Food and Drug Administration (FDA), which assesses safety and efficacy across the entire evidence base. If safety—probable benefits outweigh probable risks when used according to its intended use—and efficacy—high quality evidence of reproducible impact under controlled clinical circumstances—are demonstrated, a technology will undergo technology assessment by payors, including the Centers for Medicaid and Medicare Services (CMS), state-level bodies, and private insurers (Darrow, Avorn, and Kesselheim 2020; Kim and Basu 2021; Pochopień et al. 2021). Professional societies may also conduct their own evidence reviews when making clinical practice advisories, employing many of the same sources.

This system of regulatory pathways has shown undeniable success in standardizing processes of technology translation and minimizing the use of unsafe or ineffective interventions. Repeated incidents throughout the early 20th century of fraudulent or irresponsibly formulated “cures” causing disability and death motivated this mission. The passage of legislation has led to the designation of safety and efficacy as the gold standards for successful regulatory approval (Hilts 2004). As with many other forms of governmental risk assessment, the FDA’s current approach is informed by the 1983 National Research Council’s influential report that set the stage for the next 35 years of biopharmaceutical research and development oversight (U.S. Congress, Office of Technology Assessment 1984). Subsequent policy changes often responded to both corporate and patient advocacy for faster approval processes (averaging 8 years throughout this period); the current regulatory paradigm can be understood as attempting to balance these considerations of speed of new innovation with rigorous evidence of generation to establish safety and efficacy (Darrow, Avorn, and Kesselheim 2020). The ensuing field of risk assessment as a larger technocratic field of inquiry tended to distinguish between measurable, quantitative aspects of proposed interventions, which were the responsibility of the FDA, and more values-based considerations, which were considered outside regulatory purview (Hilts 2004; Jasanoff 2011).

As a result, it is less clear that this system has contributed to other desirable outcomes of technology translation, such as values-concordant care, equity in access, and sustainable incentives to meet the needs of U.S. population health. Historically, clinical translation in and of itself has been perceived as the appropriate end goal of biomedical research, while issues of inequity and unequal application were instead the responsibility of the healthcare system. This perception has led, in part, to disproportionate investment in therapies aimed at the greatest financial yield or lowest technical barriers to clinical translation, rather than those addressing the highest unmet clinical needs or burden. Prioritizing clinical translation over other considerations has inevitably led to new therapies that have been developed for those populations who can afford and access them: predominantly patients at academic medical centers, who tend to be comparatively affluent individuals from dominant ethnic groups. An underlying assumption appears to be that once these populations have access to therapy, access will naturally expand into other clinical care contexts, including federally qualified health care centers and community clinics where most patients with low socio-economic status receive their care. This argument is similar to the dominant U.S. economic policies of much of the 20th century, which held that improving the economic health of the wealthy would naturally result in economic benefits for the rest of society (Greenwood and Holt 2010). Much as “trickle-down economics” failed to reduce economic inequality, “trickle-down equity” in health care has likewise failed to address health and health care disparities (Center for American Progress 2013a; 2013b; Goldenberg 2024; Hope and Limberg 2020; Roosevelt Institute 2024; Urquiza 2021). Translation in of itself cannot be the only measure of its own success, especially when the most marginalized patients (e.g., minoritized populations, publicly insured, and rare disease patients), are neglected in the development and implementation of medical innovations. These disparities raise the question of whether a system that is narrowly focused on technical achievement can meet the moral obligations of medicine and public health.

We argue, based on considerable evidence (National Academies Press, 2017; O’Brien et al. 2020; Petersen et al. 2019; Zavala et al. 2021), that traditional technocratic standards are failing to integrate normative considerations into biomedical translation. What is needed is a translational domain that moves beyond safety and efficacy toward anticipating how proposed technologies will be both effective and equitable in society as it exists. We propose an additional metric of success: translational justice.

TRANSLATIONAL JUSTICE

We have defined translational justice as “procedural and outcomes-based attention to how clinical technologies move from bench to bedside in a manner that equitably addresses the values and practical needs of affected community members, with attention to the needs of the most morally impacted (Allyse et al. 2023).” Here, we distinguish between equity and justice in ways historically formulated in the sociology literature (Cook and Hegtvedt 1983; Deutsch 1975; Walster and Walster 1975). Equity is the project of ensuring that everyone has access to health in ways that actively compensate for the fact that not all persons experience the same stressors on their overall health (Braveman et al. 2018). Access programs in which drugmakers offer free access to a drug for patients without health insurance or to those whose health insurance declines coverage are attempts at health equity. Justice also encompasses addressing the systems that create multiple tiers of healthcare access and quality, driven by income inequality and the historic subordination of entire groups of people. Justice as a framework is necessitated by the centuries of harm caused by colonial systems and programs. Their legacy ensures that certain populations worldwide continue to be systematically excluded from equal access to health. As formulated by Professor Rachel Fabi: “Health rights are not sufficient to ensure access– true health justice requires action to address structures that put health out of reach for some on the basis of morally arbitrary characteristics.”²

Translational justice is a conceptual, empirical, and normative tool. As a conceptual tool, it encourages reforming our translational pathways to include end-users as co-designers from the outset. By the time a technology is implemented, its design has already incorporated many elements that impact its accessibility and responsiveness to the needs and interests of a variety of end-users. As an empirical framework, translational justice facilitates the integration of end-user perspectives into every phase of translation in ways that can be evaluated for rigor, inclusivity, and procedural quality. This approach is consistent with calls from both patients and professional bodies to ensure that equity considerations must be present throughout the technological innovation process rather than postponed until the implementation stage (National Academies Press 2023). Finally, translational justice as a normative framework focuses attention on the societal values that can and should guide policy regulating clinical translation. Shared community understandings of moral values, societal priorities, and ethical principles such as equity and solidarity form the pillars of regulatory and governance systems where authority is administered, oversight is exercised, and accountability is held.

Below we offer a fuller explanation of how considerations of justice may be integrated into translational processes. Using a framework approach that breaks the translational process down into five elements—Problem, Assumptions, Approach, Implementation, and Evaluation—we explore how translational justice might be integrated into the development of new therapies at all stages in ways that would decrease the potential for uneven and inequitable outcomes of medical technologies. As an example, we will discuss how this framework may be applied to emerging gene editing technologies, while emphasizing that it may inform the development of any biomedical innovation.

Problem – What Challenges Face Us?

An underlying implication of translational justice is that considerations of equity must begin at the basic science level, in the identification of the research topics and questions for study. Historically, this idea has been controversial (Ruzycki and Ahmed 2022; Venkateswaran et al. 2023); many scientists have argued that their obligation is to “follow the science” with no consideration of broader social implications that might be seen to bias their work (Akkermans 2024; Bird 2014; Resnik and Elliott 2016; Science 2024). However, numerous scholars of science and technology studies have pointed out that this argument relies on a fundamentally flawed view of objectivity that is unreachable by socio-technical systems (Akkermans 2024).

The field of science and technology studies has spent much of the latter half of the 20th century elucidating how science, conceptualized during the Enlightenment as an objective reflection of the natural world, is a human endeavor, conducted inside a complex web of cultural, social, economic, and legal constraints (Hess 1997; Latour 1987; MacKenzie and Wajcman 1999). This scientific ecosystem is shaped by both formal and informal incentives that influence which conditions and therapies are prioritized for investigation. The most obvious to many scientists is the need for funding. Most scientists and scientific institutions are subject to the capacities and priorities of large-scale institutions, most notably the NIH, which remains the largest funder of scientific research globally. The NIH’s agenda is not set in isolation; it is influenced by various political sources, including Congress, regarding its priorities for scientific innovation (Pielke 2007). Another compelling source of input comes from patient advocacy groups, who can bring lobbying pressure on the scientific funding system to address their areas of interest and/or are able to mobilize enough resources to offer private funding (Epstein 1996). Moreover, researchers are often lauded for being “the first,” which may influence the prioritization of clinical targets that have extensive preclinical data but less robust ethical or medical justification or application to legitimate medical needs.

One highly topical example is the comparative trajectories of technologies to address cystic fibrosis (CF) and sickle cell disease (SCD). Both are single-gene disorders with significant sequelae on quality and duration of life. The etiologies of both were determined at roughly the same time and have thus been a legitimate target of scientific attention for the same duration (Wailoo and Pemberton 2006). However, in that same amount of time, CF, which is over-represented in populations from northern Europe, and SCD, which is over-represented in individuals of West African origin, have received drastically different amounts of attention (Wailoo and Pemberton 2006).³ The life expectancy of individuals with CF has expanded to a mean life expectancy of close to 50 years (Scotet, L’Hostis, and Férec 2020). A variety of new and increasingly effective therapies and a medical focus on proactive care has meant that there are now more adults living with CF than children (Wailoo and Pemberton 2006). By contrast, the first therapy directly targeting SCD was approved by the FDA only in 2023. And only six years prior was the approval of hydroxyurea, which increases fetal hemoglobin production to reduce, but not eliminate, pain crises and the need for blood transfusion (Carden and Little 2019). Treatment of individuals with SCD remains complicated by continuing practices and assumptions in medicine, including perceptions of drug seeking for legitimate pain crises, that have long been shown to be racist (Hoffman et al. 2016; Lin et al., 2024; Trawalter, Hoffman, and Waytz 2012). The large, concerted efforts made to lobby for CF research were lacking in SCD research (Farooq et al. 2020), resulting in decades of slow to no progress, while CF has long been framed as an urgent medical problem requiring well-funded research (Burke 2023; Gerido 2023; Halley et al. 2023).

CF and SCD are also instructive in that they demonstrate a different type of disparity, what we term “mutational discrimination”, which threatens to greatly exacerbate the inequity of access to participation in clinical research. The recent history of CF drug development serves as a preview of the mutational discrimination that is likely to shape the development of genetic therapies. CF is relatively common among rare diseases, preferentially affecting people of Northern European ancestry (1 in 3,000 newborns in that population) (O’Sullivan and Freedman 2009). Pivotal to the development of effective CF medications (small molecules rather than genetic therapies) by Vertex Pharmaceuticals was substantial funding from the Cystic Fibrosis Foundation that ultimately added up to hundreds of millions of U.S. dollars; indeed, it was the availability of that funding that persuaded Vertex to work on CF (Herper and Matthew 2019). The first drug out of that pipeline was Kalydeco (ivacaftor), which was primarily targeted at one of the most common CF mutations (CFTR G551D) out of the more than 2,000 mutations that have been identified, covering ≈4% of patients, largely of European descent. Subsequent efforts focused on treating the most common CF mutation (CFTR del508), which affects about ≈70% of patients in the U.S., resulting in the development of Orkambi (lumacaftor/ivacaftor). Vertex then sought to develop a drug that could address more mutations, culminating in Trikafta (elexacaftor/tezacaftor/ivacaftor), which can treat ≈90% of patients based on the spectrum of specific mutations for which use of the drug is approved (McGarry and McColley 2021).

Despite this tremendous progress in the treatment of CF, the question of how to address the remaining ≈10% of CF patients, for whom the Vertex drugs are ineffective, is fraught. It has become clear that the “10% left behind” are disproportionately skewed toward non-White populations. A U.S. study of CF patients found that 7.6% of non-Hispanic white patients, 24.4% of Hispanic patients, 30.3% of patients of African descent, and 19.5% of patients of other races have mutations that make them ineligible to receive any of the Vertex drugs (Desai et al. 2022). A study of CF patients in the United Kingdom found similar results, with 6.5% of patients of European descent, 59.4% of patients of Asian descent, and 50% of patients of African descent having mutations that make them ineligible (McGarry et al. 2022).

On the one hand, drug development was skewed toward a higher-frequency rare disease that disproportionately affects individuals of European descent, prioritizing variations prevalent in individuals of European descent (driven by the large total addressable market and the availability of ample funding from a foundation) (Marshall et al. 2009). On the other hand, the longstanding challenge of some populations being underrepresented in clinical research in this case has a pernicious negative feedback effect: CF mutations that are exclusive to those populations were not assessed in the clinical trials sponsored by Vertex, which meant there was no possibility of generating positive data with which the company might seek regulatory approval for the drug(s) to have expanded use in patients with those mutations. These same factors will be in play when it comes to the development of genetic therapies for rare, and especially ultra-rare, diseases.

Ironically, despite being relatively ignored for decades, SCD has arguably become the exemplar for mutational discrimination in the development of gene editing therapies. Although SCD preferentially affects populations of non-European ancestry, there are an estimated 100,000 individuals with SCD in the U.S. SCD also has an unusual distinction among genetic disorders in that every affected individual harbors the same mutation (HBB E6V)—making it feasible to develop “one-size-fits-all” editing solutions, in contrast to CF, and charge the premium price of millions of U.S. dollars per treatment that has now become standard for genetic therapies (Basu et al. 2024; Nuijten 2022). As such, SCD has attracted outsize interest as a commercial opportunity—given the large total addressable market in the U.S.—and has become highly competitive, with more than 10 companies and academic teams developing gene editing therapies for SCD; the first such therapy, Casgevy (exa-cel), recently received FDA approval and earned the distinction of being the first CRISPR drug (FDA 2023). Yet hundreds of hemoglobinopathies that could be treated with gene editing therapies like exa-cel have seen relatively little investment into their development, because these disorders are all ultra-rare in frequency and caused by a diversity of gene mutations, meaning that these therapies will have to be highly personalized to the patients. Thus, it is highly likely that many patients will be “left behind” with unjust translational outcomes because there will be no efforts to develop gene editing therapies for them.

It is important to consider whether the conditions being addressed in biomedical research are targets of opportunity or are selected with an eye to addressing active disparities. One way of integrating translational justice into basic science is to encourage early-stage assessment of disease targets, including the impact on existing health disparities and the potential creation of new disparities. If emphasized by funders as a research priority, such assessment will encourage basic researchers to reconfigure their approaches to team science upstream along the translational pipeline, incorporating health disparities expertise earlier.

Assumptions – What is Being Taken as Truth?

In 2019, Obermeyer et al. published an analysis of an algorithm used by a large academic health system to identify post-surgical patients most likely to decompensate and require readmission. The goal of the algorithm was to proactively offer additional support to such patients to avoid readmission. In practice, the algorithm repeatedly identified white patients as being sicker and requiring more services than Black patients, even though they were more seriously ill. The reason was not an explicitly racist bias but algorithmic bias in the formative assumption in its creation, which equated medical billing with the severity of illness (Obermeyer et al. 2019). This assumption is not unreasonable; as the authors pointed out, “all other things being equal, people who are sicker tend to use more healthcare.” However, all things are not equal in the US; Black patients receive significantly less intense levels of healthcare based on a variety of factors that include both structural and economic forces (Hoffman et al. 2016; McCarthy et al. 2016; O’Brien et al. 2020; Petersen et al. 2019; Peterson et al. 2020; Valdes 2021).

This example points to the fact that assumptions made at multiple levels of translation can fundamentally alter the downstream impact of a given technology. In this case, the assumption was that healthcare is essentially a frictionless system in which the amount of healthcare received is equal to the amount of healthcare needed. Given that this is not the case in any known healthcare system in the world, this assumption doomed the technology to an unjust translational outcome.

Astute critiques of biomedical science have noted that a lack of diversity among researchers can limit their ability to understand the needs of potential end users and the social ramifications of novel technologies (BrckaLorenz, Haeger, and Priddie 2021; Miriti 2020; Obermeyer et al. 2019; Palid et al. 2023). These gaps in understanding often lead to poor assumptions and can mean that translation heads down unhelpful paths and blind alleys. As an example, some hospital systems have responded to the increased need for remote healthcare by pushing for a technological solution: full-scale broadband access in even the most sparsely populated areas, on the assumption that these populations need equal access to live video-based telehealth care (Anderson and Carr 2022; Healthcare Value Hub 2017; Pitsor 2022). Such assumptions may stem, at least in part, from the perspectives of individuals working in well-resourced areas with constant access to high-speed internet and video communication, who envision this kind of access as a solution for everyone. While telehealth access may be part of an overall solution, solutions for rural health inequity must focus on other areas of rural inequity, including innovating products and services that meet the needs of remote patients without requiring a costly, time-consuming video visits and directing resources toward larger problems identified by rural communities themselves, such as physical and financial access to medication (Goins et al. 2005). Similarly, technological fixes will be ineffective if they are disconnected from the contextual realities of the impacted populations. Even beyond the general issues of access and availability of care, they must consider population-specific barriers including lack of culturally competent care, low technological uptake, and mistrust in the health care system, among other lived realities (DeGuzman et al. 2022; Kappel et al. 2022; O’Shea et al. 2024). The need for contextualized solutions necessitates empirical work throughout the translational process to ensure that the correct problems and approaches are being made to support a just outcome—informed by the voices of end users—rather than basing decisions on potentially biased assumptions.

Ableist assumptions also pervade healthcare research and translation. Biomedical research agendas often assume that disability—framed in many different senses—is a fundamentally undesirable state of being that requires alleviation through medical interventions. In contrast, individuals with disabilities regularly report a high quality of life and often frame these perceptions of disability as products of social and built environments that do not support a wider variety of ways of being (Shakespeare 2013). A classic example of this conflict is the introduction of cochlear implants in the 1980s. While proven safe and effective at restoring some level of hearing, if introduced in early childhood, many in the Deaf/deaf community framed this intervention as pathologizing deafness and seeking to erase a culture of sign language and the non-hearing (Christiansen and Leigh 2002). In the early 2000s, the Human Fertilization and Embryology Authority (HFEA) added congenital deafness to the list of covered conditions for preimplantation genetic diagnosis, allowing future parents to screen embryos for deafness and select those that would most likely result in a hearing child. Segments of the Deaf/deaf community were likewise angered by the assumption that deafness was a legitimate target of elimination along with fatal and life-limiting conditions included on the list (Porter and Smith 2013).

More recent therapies designed for children with achondroplasia have focused on increasing height in individuals with achondroplasia, with the stated goal of achieving population-average height and “independent” living in a world designed for average-height individuals (Savarirayan et al. 2024; Willyard 2023). Within achondroplasia communities, however, the Little People of America have issued a statement that “focusing on growth velocity is a pharmaceutical solution for a societal problem. We want to reframe these priorities in research to the most meaningful ones to our members” including reducing spinal compression and increasing range of limb motion (Little People of America Biotech Industry Liason Committee 2022). Researchers in this area have responded that the interventions also confer secondary benefits, and that height represents an objective proxy measure for drug effectiveness. However, many in the community felt that the assumption of height as a primary focus misrepresented their lived experience and their needs.

There is added complexity when considering conditions with a wide range of phenotypic expression and those which vary by age of onset and/or evade early clinical detection. Individuals with the same condition can differ in medical and support needs and in the ways in which they perceive their condition and its relationship to their identity. For these reasons, condition communities may have substantial internal disagreement regarding medical interventions targeting central aspects of the condition. For example, parents of children with autism spectrum disorder (ASD) level 3 may welcome a medicalized approach to their child’s health and investment in therapies that promote “typical” neurocognition for independence, improved communication, and reduction of physically harmful self-stimulating behaviors. Many individuals with ASD level 1, however, are not diagnosed until adulthood due to their more internalized presentation and camouflaging of autistic traits, and may be more critical of medical interventions to “normalize” what they perceive to be a positive difference in neurocognition (Atherton et al. 2022; Grosvenor et al. 2024; Jadav and Bal 2022). Some communities have shifted their language from pathologizing terms (“disorder”, “symptoms”) to “neurodivergence” or “neurotype” to reflect these attitudes.⁴ Moreover, individuals with ASD level 1 often have vastly different needs from children and adults with more compromised functioning that may not be sufficiently distinguished by the medical and scientific community.

Without inclusion of the voices of lived experience, treatments may be developed for communities that do not desire therapies, or for endpoints that are not supported by patients or families. From a market perspective, these mismatched solutions also have negative effects on science and industry via limited uptake of therapies and misallocation of resources that could have been directed toward therapies sought by communities. Additionally, varied perceptions and values regarding traits or symptoms call for heightened sensitivity around targeting these, as well as engagement with all sectors of the community (e.g., both parents of children and affected individuals) to avoid unintentional stigmatization and selective engagement with sub-groups perceived to be more supportive of the intervention (e.g., unaffected parents of a child). It will be crucial to consider whether communities perceive a therapy for their condition to be appropriate, and what metric or end goal is the most appropriate proxy for success (e.g., length of survival, reduction of specific or global symptoms, a complete “cure”).

These examples highlight again the need for engagement with end users early in the scientific process, most importantly to clarify assumptions and co-design therapeutic options that are sensitive to their values and needs and responsive to their priorities. We recommend that end users with lived experience be included in the early stages of technology development, well before human trials, and that engagement continue during the entire translational cycle, integrating the legitimate needs and priorities of members of these communities and ensuring congruence and just translation.

Approach – How Do we Address the Problem?

We return to genetic therapies, which, after a relative hiatus following the death of Jesse Gelsinger (Smith and Byers 2002), made a resurgence in the 2010s, accelerated by the introduction of more precise gene editing technologies. While genetic therapies have shown promising results in several previously intractable conditions, mechanisms must be developed to specifically address the challenges inherent in ensuring their equitable development (Herzog, Cao, and Srivastava 2010). One such mechanism might be a top-down, prescriptive approach. For example, regulatory agencies such as the FDA might mandate that sponsors of clinical research on genetic therapies show that they have considered equity issues in their basic science or translational research, perhaps by including patients with lived experience as research advisors and/or addressing ultra-rare or even n-of-1 mutations to ensure inclusion of the entire community. In practice, such a mandate would likely be untenable and unenforceable; even though the FDA has a substantial interest in fostering diversity in clinical research and recently published a guidance document outlining best practices, the document explicitly states that the recommendations are non-binding (FDA 2020).

A more realistic approach would be for regulatory agencies to heavily incentivize diverse clinical research. For example, the FDA might offer a supported approval pathway for research that exclusively involves patients in underrepresented groups and/or with undertreated conditions/variants, although communication efforts should stress its scientific rigor and historically marginalized populations should be reassured that this is an effort to address underinvestment, rather than perpetuate past racist selection of certain populations for human subjects research In such cases, more limited preclinical datasets could suffice for clinical trial applications, greatly reducing the time and work needed to reach clinical trials for research involving underrepresented conditions/pathological variants. Such an incentive would be particularly attractive for genetic therapies in which most components of the drug products have been previously approved (i.e., the delivery vehicles), and only one component is being changed to personalize each therapy for a specific mutation. Slashing the time and resource expenditures needed to develop each new therapy, incentivizing for-profit companies to see better value for the therapy and enabling nonprofit groups and academic groups to develop therapies, will be a necessary step to mitigate mutational discrimination. Revisiting the example of CF, the FDA has facilitated the expanded use of Vertex drugs for ultra-rare CF mutations by allowing for in vitro cellular data for each mutation to be used for approval, rather than insisting that each mutation be studied in the context of a clinical trial (Durmowicz et al. 2018).

Another angle from which to tackle the problem of mutational discrimination is to address funding for the development of genetic therapies. Even if the costs of developing therapies and launching clinical trials are substantially reduced by regulatory streamlining, there will be large gaps not be served either by the financial interests of investors in for-profit companies or by philanthropy (notwithstanding highly exceptional circumstances such as the Cystic Fibrosis Foundation committing $500 million USD to the development of therapies that serve the 10% of CF patients who are not eligible for Vertex drugs) (Joseph 2019). The readiest solution—though by no means a complete one—lies in funding agencies like the NIH, which can preferentially direct funds to preclinical and clinical research that involves patients in underrepresented populations and/or with underrepresented conditions/mutations. Nascent NIH programs such as the Bespoke Gene Therapy Consortium (Kingwell 2021), the Platform Vector Gene Therapies Project (Brooks et al. 2020), and the Somatic Cell Genome Editing Consortium (Saha et al. 2021) represent commitments to the development of genetic therapies on the order of hundreds of millions of U.S. dollars; none of these programs are explicitly focused on underrepresented populations and/or under-addressed mutations, although the first two programs have prioritized ultra-rare diseases for which there are no therapies in development in industry. Future funding programs need to be much more attentive to diversity in all these respects.

Implementation – How Does the Technology Operate in the Real World?

Phase 1 trials, sometimes called safety trials, are designed to demonstrate that interventions do not cause toxicity outside of the intended therapeutic target. Satisfactory completion of safety trials at multiple doses is a requirement to progress with efficacy trials. In practice, however, participants in clinical trials are significantly constrained to a narrow range of phenotypes based on availability and to ensure data consistency. The past decade has seen an increase in attention to the lack of diversity across the research spectrum. The National Institute on Minority Health and Health Disparities has acknowledged that “historically, clinical trials did not always recruit participants who represented the individuals most affected by a particular disease, condition, or behavior.”⁵ Within the US, clinical trials have overwhelmingly relied on individuals of European descent. This reality is reflective of both the location of most biomedical research inside academic medical centers, which over-represent higher income individuals, and the lack of community engagement in diverse communities. As a result, many therapies, including those targeting conditions over-represented in ethnic minority populations, reach the market having been studied only or primarily in individuals from a European background. The result is not only a lack of information on how therapies may perform in diverse populations, but also an understandable concern among those populations about accepting novel therapies (Alsan et al. 2022; Riggan et al. 2023).

Similarly, a vast majority of interventions are validated for safety only in animal models or biological males; pregnancy and lactation have historically been considered standard exclusion criteria for almost all clinical trials, meaning that a wide swath of drugs has never been tested on women of reproductive age (Foulkes et al. 2011; Martin 2016). The exclusion of pregnant individuals is often linked to the discovery that the drug thalidomide (Madhusoodanan 2020), widely prescribed to pregnant individuals in Europe and Canada as an analgesic for morning sickness in the 1970s, causes severe and potentially fatal birth defects (Lo and Friedman 2002). It is worth noting, however, that the drug was presumed safe for use during pregnancy because no adverse effects had been observed in limited animal studies; no studies for teratogenic effects in pregnant animal models were conducted prior to its introduction in humans. The lack of any safety studies was the reason the drug was blocked by an FDA reviewer in the US, suggesting that the appropriate lesson to be learned was about the necessity of more safety trials in pregnancy, rather than the permanent exclusion of pregnant individuals from biomedical research (Costantine, Landon, and Saade 2020; FDA 2019; Lyerly, Little, and Faden 2008; Riley 2024).

A more recent example can be seen in the COVID-19 mRNA vaccines, which were developed through an accelerated clinical trial pathway to address the pandemic. The exclusion of pregnant and lactating individuals from those trials led many to question whether the vaccines were safe for use during pregnancy (Farrell, Michie, and Pope 2020). Many pregnant individuals or those contemplating pregnancy hesitated or refused vaccination out of concern for fetal impact (Huang et al. 2022). Data later showed that vaccination was not associated with any risk to maternal-fetal health, and in fact conferred benefits, such as vertical transmission of protective antibodies to the fetus. By the time this data was published, however, there was ample data that contracting the COVID-19 virus during pregnancy was significantly correlated with adverse pregnancy outcomes, including preterm birth and maternal morbidity and mortality (Jorgensen et al. 2023; Norman et al. 2024; Torche and Nobles 2023). The systematic exclusion of pregnant and lactating individuals from clinical studies has created a lacuna in the clinical safety data that leaves a host of relatively common medications without safety data for pregnant individuals (National Academies of Sciences, 2024; Riley 2024). Even the Centers for Disease Control has acknowledged that “Fewer than 10% of medicines approved since 1980 have enough information to determine their safety during pregnancy. This is because pregnant people are often not included in studies that determine the safety of new medicines.”⁶ In 2016, a draft document was circulated proposing a revised framework for integrating pregnant or lactating individuals into clinical studies at an earlier stage. At the time of writing, however, no action has been taken to implement changes to inclusion criteria. From a translational justice perspective, this is insufficient: clinical trials must be broadly inclusive and specifically designed to be representative of the population the drug/therapy is intended for, both to engender trust and ensure therapeutic effectiveness. This historic and present-day context has important ramifications for recruitment into clinical trials during the prenatal period and the pregnant populations who may be more willing and able to participate.

Evaluation – What Counts as Success?

Success metrics in the medical system often include indicators such as patient outcomes, quality of care, patient satisfaction, and cost-effectiveness. While these metrics of success were intended to assess the effectiveness and efficiency of healthcare delivery, they can inadvertently contribute to health inequities in a multitude of ways. Translational justice calls on us to address multiple drivers of health inequity, including (1) how implicit bias shapes patient outcome measures, (2) intersecting sources of marginalization, (3) the need for health economic metrics to integrate patient perspectives on equity, and (4) underfunding of safety-net hospitals in the context of values-based purchasing.

Biases in outcome measures may arise in favoring specific patient populations over others due to socioeconomic status, academic background, and cultural background, leading to disparities in perceived quality of care. A recent study comparing patients of different socioeconomic backgrounds participating in cancer trials found that disparities in care, follow-up, and support services contributed to poorer survival rates among individuals from lower socioeconomic backgrounds in these trials, compared to individuals from high socioeconomic backgrounds (Unger et al. 2021). Another study found significant differences in medical student and physician resident beliefs regarding biological differences between Black and white patients. These beliefs led to racial biases in treatment recommendations based on perceived clinical outcomes and disparities in pain assessments, with Black patients being less likely to receive appropriate pain management (Hoffman et al. 2016). Additionally, a growing body of research highlights biases in clinical outcome measures about gender. Several studies of gender differences in the presentation and management of acute coronary syndrome have found that women often present with different symptoms compared to men, which have not historically been included in medical training curricula. This lacuna leads to delays in diagnosis and treatment for women presenting with these symptoms (Kawamoto, Davis, and Duvernoy 2016; Mousavi et al. 2023). Women were found to be less likely to receive guideline-recommended treatments and had worse outcomes compared to men.

Clinical outcome measures may also prioritize health outcomes that are more easily achievable for certain patient groups, further exacerbating disparities in health outcomes (Elliott et al. 2010; Fiscella and Sanders 2016; Haviland et al. 2005). In these cases, the healthcare system is structured to meet the needs of more advantaged groups, further marginalizing those who are already disadvantaged. As an example, a study investigating cancer patient outcomes found that socioeconomic disparities in outcomes are partially due to differences in the quality of care received, highlighting a need to implement clinical outcome measures that are equitable and consider the diverse needs of all community groups (Esnaola and Ford 2012). In addition, medical care success metrics tend to emphasize cost-effectiveness or efficiency, which could contribute to healthcare providers prioritizing patients who are perceived to have lower healthcare needs or more positive outcomes. The use of success metrics in this results in underinvestment in resources and services for underserved communities. For example, pay-for-performance programs make up about 20% of medical care institutions in the US (CMS 2024; Sorbero et al. 2014). These programs can unintentionally incentivize healthcare systems to engage in a practice known as “lemon-dropping” or focusing on patients who are more likely to achieve positive outcomes, neglecting high-risk patients who are more likely to have negative outcomes (Kerbel 2021). As a result, these programs can inadvertently exacerbate disparities by prioritizing resources and attention toward patients with fewer healthcare needs, underinvesting in the care of more vulnerable high-risk populations (Mendelson et al. 2017; Rosenthal et al. 2016).

Value-based purchasing models have also been widely adopted across the US healthcare system, with value-based purchasing programs affecting payment for inpatient stays in more than 3000 hospitals nationwide (American Hospital Association 2024). This program ties Medicare payments to the quality of care provided rather than the quantity of services performed. This, in turn, incentivizes hospitals to improve patient outcomes and safety. However, numerous studies have found that this model may disproportionately penalize safety net hospitals, which serve a higher proportion of low-income vulnerable patients and thus may struggle to achieve the same performance metrics as more affluent institutions, leading to financial penalties and further underinvestment in resources for underserved communities (Chiu et al. 2022; Pandey et al. 2023). Effective implementation of these value-based programs also requires robust infrastructure and data analytic capabilities. Thus, institutions with fewer resources will inevitably struggle to implement these programs effectively, leading to inequitable outcomes (CMS 2024; Sorbero et al. 2014). A reinvigorated health economics discourse has echoed our call for translational justice: i.e., innovating and building health data infrastructure that measures cost and access in ways that reflect the perspectives of patients and families, rather than treating patient satisfaction and cost as distinct, unrelated phenomena (Smith et al. 2023).

Our call for translational justice suggests opportunities for defining success as early integration of health disparities expertise, both for outcomes metrics and the challenges of implementing a therapy into an inequitable health care system, which are unrelated to the efficacy of the therapy itself (Pelfrey, Goldman, and DiazGranados 2021). Even building equity considerations into the selection of the condition itself—as is the case with Vertex and its decision to pursue a genetic therapy to meet the unmet medical needs of SCD patients (Pagliarulo 2023)—does not guarantee that therapy will reach the patients it is designed to serve. More than half of SCD patients are covered under Medicaid insurance and underutilize existing therapies in part due to pre-authorization requirements and other restrictions (Bazell et al. 2019; Scott et al. 2023). Cost analyses project considerable burdens on state Medicaid programs that choose to cover genetic therapies and private insurance coverage is variable (DeMartino et al. 2021; North Carolina Department of Health and Human Services 2024). Access to genetic therapies for SCD and other conditions is further limited by availability only at authorized treatment centers, and loss of work, transportation/travel distance, and childcare costs associated with a lengthy treatment protocol. These challenges are magnified when considering prenatal applications of gene editing and other therapies designed to be introduced during pregnancy. Relatively few tertiary care centers in the U.S. have established maternal-fetal therapy expertise, wraparound support, and follow-on care, even though there are many families with affected pregnancies that could theoretically benefit. Strategies at both the clinical trial and practice will need to be proactively implemented to promote equitable access for all end-users across geographic and socioeconomically diverse populations.

CONCLUSION

Translational justice encourages us to take an approach to clinical translation that is both deep and wide, by prioritizing a detailed understanding of end users’ priorities, values, and therapeutic goals along with individual and systems-level barriers to implementation at each level of clinical translation. Translational justice also encourages broad inclusiveness in therapeutic design and clinical evaluation, to begin the process of undoing historic inequities and injustices, and unequal outcomes. Research into genetic therapies, as described above, is one salient example, having historically focused on a relatively narrow patient population. In gene editing research, as in all biomedical research and translation, we should not be content with the former neglect of community voices and exclusion of health disparities concerns. Our communities and patients must be the primary focus across the entirety of the clinical translational pathway.