Helen is sixteen years old, sitting in a Chicago emergency room, coughing badly enough that she needed to come in at night. A resident stops her mid-sentence and asks if she smokes. She says no. He asks again: “Are you sure you’re not a smoker? Your mom’s not here so it’s fine.” The question is not clinical curiosity. It is a verdict delivered twice.

This essay argues that Helen’s ER encounter, read against Marshall’s markedly different relationship with healthcare institutions, reveals that epistemic injustice in healthcare is not primarily a failure of individual sensitivity but a structural condition that clinical documentation systems actively reproduce. The claim is disputable: one could attribute Helen’s encounter to poor bedside manner or a single resident’s bias. But the evidence points to something more systemic—a deflation of patient credibility that was organized, predictable, and left no trace in the record.

Miranda Fricker distinguishes two forms of epistemic injustice. Testimonial injustice occurs when a speaker’s credibility is deflated because of identity prejudice on the part of the hearer. Hermeneutical injustice occurs when a structural gap in shared interpretive resources prevents someone from making their experience intelligible—to others or to themselves. Both are at work here, but in different registers.

The resident’s double questioning is a textbook instance of testimonial injustice. Helen had answered clearly. The repetition—”your mom’s not here so it’s fine”—signals that the resident’s prior credibility assessment of Helen was shaped by an identity composite: young, Black, on Medicaid, presenting with a respiratory complaint in an urban ER. Smoker is the interpretive frame that composite activates. Her denial is filtered through that frame rather than registered as testimony. The resident does not need new clinical information; he needs her to confirm what he has already inferred. Her actual account of her respiratory history—interrupted twice—never fully enters the clinical encounter as evidence. Importantly, it likely never entered the medical record either. What was documented was almost certainly the chief complaint and the eventual diagnosis, not the credibility transaction that preceded them.

Hermeneutical injustice operates differently in Helen’s earlier account of her father’s death in Johannesburg. The hospital discharged him after a stroke with anti-nausea medication. Helen says, carefully: “I like to say he passed away because of medical malpractice—I don’t know if that’s accurate.” That hesitation is not ignorance. It is the mark of someone who lacks access to the conceptual vocabulary—clinical negligence, duty of care, failure to diagnose—that would allow her to name what happened with the authority it deserves. The gap is not personal; it is structural. Refugee patients in South Africa in the 1990s had no institutional pathway through which their experiences of medical abandonment could be formulated as actionable claims. The interpretive resources simply did not exist for them in that context.

Marshall, by contrast, grew up in a stable middle-class household in Boise with attorney parents, studied neuroscience at BYU, and describes his relationship with healthcare as one of general trust: “I don’t think I’ve ever had an experience with a healthcare system that I was really discontent with.” That is not a footnote. It is the analytical pivot of the case. The same healthcare encounter that left Helen braced for dismissal—prompting her to brief Marshall in advance to advocate for her if she was medicated or unconscious—registered to him as competent, informative, and not pushy. Two people in the same room, the same pregnancy, the same medical team, and structurally different relationships to the epistemic authority of that institution.

A properly designed informatics system would need to address this at the point of data capture, not after the fact. FHIR’s current patient-facing resource profiles allow structured documentation of chief complaints and social determinants, but they do not capture what was interrupted, doubted, or dismissed in a clinical encounter. The 21st Century Cures Act’s information-blocking provisions ensure patients can read their notes—but if a note was written from the frame of presumed drug use rather than from the patient’s testimony, transparent access to that note does not correct the original epistemic wrong. A system designed to address testimonial injustice would need structured fields for patient-provided narrative that persists independently of clinician interpretation, with provenance metadata distinguishing what the patient said from what the clinician inferred. A system designed to address hermeneutical injustice would need to surface, at the point of care, interpretive frameworks relevant to a patient’s specific historical and cultural context—not as a demographic flag, but as a prompt for the clinician to ask what conceptual tools this patient may or may not have access to when describing their experience. These are engineering requirements, not aspirations. The question is whether the field is prepared to treat them as such.

The Interviews

Introductions and backgrounds

Healthcare experiences before pregnancy

Marshall’s healthcare relationship and information seeking

The pregnancy — information, data, and the c-section

Race, trust, and the delivery

Key Medical Concepts

Informatics and Epistemic Themes

Health Equity and Structural Racism

Health equity names the condition in which every person has a fair opportunity to attain their full health potential, and structural racism describes the way that laws, policies, and institutional practices—not only individual prejudice—systematically produce differential health outcomes across racial and ethnic groups. Helen’s account of her father’s death in Johannesburg is a precise illustration: as a Rwandan refugee, he occupied a legal category that excluded him from routine hospital care, was sent home with antiemetics despite a hemorrhagic brain injury, and died when a private-practice scan three days later confirmed what timely intervention might have addressed. The structural mechanism here is not a single clinician’s bias but a layered architecture—refugee legal status, a single-hospital safety net, resource triage that treated non-emergency presentations as non-urgent—that concentrated fatal risk in a particular population. Marshall’s trajectory offers a counterpoint that sharpens the analysis: raised in a professional, upper-middle-class household with attorney parents, he describes a qualitatively different relationship to institutions, suggesting that class position can partially buffer the effects of racial marginalization even when it does not eliminate them. The tension worth naming is that health equity frameworks, however analytically powerful, risk remaining descriptive unless they are operationalized: the FHIR Social History and Observation resource categories that might capture refugee status, housing instability, or documentation status are largely optional and inconsistently populated in U.S. EHRs, meaning that the structural conditions Helen describes remain invisible to the data systems that are supposed to support equitable care.

Testimonial Injustice

Miranda Fricker defines testimonial injustice as occurring when a speaker receives a deflated credibility assessment from a hearer, and that deflation is caused by identity prejudice—the hearer’s negative stereotypes about a social group to which the speaker belongs distort their judgment of how much epistemic trust the speaker deserves. The ER encounter Helen describes as a teenager in Chicago is a textbook instance: a resident interrupts her clinical history twice to ask whether she smokes, and on the second occasion explicitly tells her that her mother is absent and therefore it is “fine” to admit it—a move that encodes the presumption that a young Black immigrant on Medicaid is both likely to smoke and likely to lie about it in a parent’s presence. The credibility deflation is not incidental; it is the mechanism by which her testimony about her own body is structurally discounted before she can finish delivering it. What makes this analytically important beyond the individual slight is that the interrupted history is also a corrupted clinical dataset: if the differential diagnosis is organized around a smoking assumption the patient has denied, downstream reasoning is compromised by the same prejudice that produced the injustice. From an informatics standpoint, this moment has a direct analog in structured data entry: if a clinician encodes “tobacco use” in a FHIR Observation resource based on presumption rather than verified patient report, that datum propagates through the record, potentially influencing CDS Hooks alerts, insurance risk stratification, and future clinical encounters in ways Helen has no mechanism to contest. The 21st Century Cures Act’s information blocking provisions and the resulting patient access rights under 42 U.S.C. §300jj-52 give Helen the legal right to read that note—but the right to read a record that misrepresents you is not the same as the epistemic repair that Fricker’s concept demands, and patient portals do not currently offer structured pathways to formally dispute a clinician’s documented inference.

Hermeneutical Injustice

Fricker’s hermeneutical injustice is structurally distinct from testimonial injustice: it arises not from a prejudiced hearer but from a gap in the collective hermeneutical resources—the shared concepts, frameworks, and vocabulary—available to a community, such that a person whose experience falls outside those resources cannot fully articulate, and therefore cannot fully communicate or even fully understand, what is happening to them. This transcript offers suggestive but incomplete evidence for the concept. Helen’s description of her father’s death—”I like to say he passed away because of medical malpractice, I don’t know if that’s accurate”—gestures toward a real interpretive gap: she has a felt sense that something wrongful occurred, but the legal and medical conceptual apparatus for naming that wrong (malpractice, negligence, standard of care) was not available to her family in that moment in Johannesburg, and may still feel like borrowed rather than fully owned vocabulary. The hedging (“I like to say,” “I don’t know if that’s accurate”) is characteristic of what Fricker calls the “strained, gappy, and inarticulate” quality of experience that lacks adequate interpretive resources. However, the transcript does not give us enough evidence to claim full hermeneutical injustice in Fricker’s strict sense, which requires that the gap be structural and collective, not merely individual unfamiliarity with a technical term—Helen’s MPH training and healthcare employment mean she has acquired significant interpretive resources since, and the uncertainty she expresses may reflect genuine epistemic ambiguity about a complex clinical event rather than a structural lacuna. What the case does illustrate is the temporal dimension of hermeneutical injustice: even if the concept is eventually acquired, the damage done during the period of interpretive absence—grief unprocessed, accountability unassigned, a family’s understanding of their loss forever incomplete—is not undone by later literacy.

Cultural Humility in Clinical Communication

Cultural humility, as distinguished from the more static concept of cultural competence, names an ongoing, self-critical orientation in which the clinician treats their own cultural frameworks as contingent and positions the patient as the primary authority on the meaning of their experience—rather than applying a fixed inventory of group-level cultural “facts.” The ER resident’s conduct toward Helen fails this standard in a specific and instructive way: rather than treating her as an individual whose history he does not yet know, he imports a population-level stereotype (young Black patients on Medicaid are more likely to smoke and to conceal it) and uses it to override her self-report, which is precisely the epistemically closed posture that cultural humility is designed to interrupt. Marshall’s background—neuroscience-trained, from a professionally successful family, generally trusting of institutions—represents a different kind of test for cultural humility: not the challenge of engaging across distance and distrust, but the subtler challenge of not assuming that class markers and scientific literacy translate into equal comfort or equal safety in every clinical encounter. Helen explicitly prepared him to advocate for her precisely because she knew that her race meant their experiences in the same room might not be identical. The limit of cultural humility as a framework is that it is primarily an account of individual clinical disposition, and therefore cannot by itself address the structural conditions that produced the Johannesburg hospital triage or the Chicago Medicaid intake process; a clinician of perfect humility operating within those systems would still be constrained by them. In informatics terms, cultural humility has a design corollary: patient-facing intake tools and portal interfaces that offer rigid, checkbox-based demographic fields—race, ethnicity, language—enact a kind of structural cultural presumption, and FHIR’s relatively flexible extension mechanisms for capturing patient-preferred identifiers and communication needs are a technical affordance that humility-oriented design could use but rarely does.

Social Determinants of Health in Data Models and FHIR

Social determinants of health (SDOH) are the non-clinical conditions—housing, food security, legal status, income, transportation, social support—that research consistently shows account for a substantial portion of health outcomes, and their representation in data models is the project of encoding these conditions in structured, computable form so they can be acted upon systematically rather than noted anecdotally. Helen’s narrative is almost entirely a story of SDOH: refugee legal status determining hospital access in South Africa, Medicaid status triggering a specific intake workflow in Chicago, years of geographic displacement shaping which dental care was available and under what conditions. The Gravity Project, which produces the SDOH-relevant value sets now incorporated into FHIR R4 and US Core, includes Observation and Condition profiles for food insecurity, housing instability, and transportation problems—but notably does not have mature, standardized representations for refugee or asylum-seeker status, documentation status, or the compounding effect of multi-country displacement on health history continuity. This gap is not incidental: it reflects the same structural blind spots in the healthcare system that Helen’s story exposes, because data models are not neutral containers but codifications of what a system has decided is worth measuring and acting on. The practical limit is bidirectional: even well-designed SDOH data models require clinical workflows and institutional will to populate them, and Helen’s Chicago ER encounter suggests that the encounter likely generated a Medicaid billing record but almost certainly no structured SDOH data—meaning the conditions most relevant to her care were the least visible in the system that was supposed to serve her.

Algorithmic Bias and Disparity Amplification

Algorithmic bias in healthcare refers to the way that predictive models, clinical decision support tools, and risk stratification algorithms can reproduce or amplify existing health disparities when they are trained on historically biased data or embed assumptions that systematically disadvantage particular populations. This transcript does not contain direct evidence of algorithmic bias—neither Helen nor Marshall describes an encounter with a risk score, a predictive model, or an automated recommendation—and it would be intellectually dishonest to claim otherwise. What the transcript does provide is the upstream conditions that make algorithmic bias consequential: Helen’s fragmented, multi-country health history, her family’s use of emergency care rather than primary care as teenagers, and her father’s death in a system that generated no transferable record are precisely the kinds of data absences and distortions that, when a population’s records are used to train a model, produce systematically underfit representations of that population’s health needs. The well-documented example of Obermeyer et al.’s 2019 finding—that a widely used commercial algorithm used healthcare cost as a proxy for health need, thereby underestimating illness severity in Black patients whose prior utilization had been suppressed by access barriers—is the structural analog to what Helen describes: the algorithm did not introduce bias, it faithfully learned from a system already shaped by it. The technical limit is that bias auditing tools, fairness constraints, and disaggregated performance reporting can reduce but not eliminate this problem so long as the training data reflects access-unequal utilization, which means that algorithmic fairness interventions are necessary but not sufficient without the upstream structural changes that health equity frameworks demand.

Discussion Questions

  1. Helen describes being born in the “poor people’s hospital” versus her sister’s birth in the “rich people’s hospital,” and later receiving emergency dental extractions in informal clinics across multiple countries — yet she arrives in the U.S. with an MPH and works inside a healthcare system. Using Fricker’s distinction between testimonial and hermeneutical injustice, identify which form of epistemic injustice is more likely to operate when a clinician encounters Helen: is the problem that her testimony about her own health history will be discounted, or that neither she nor her clinician has adequate shared conceptual resources to make sense of a medical biography that spans refugee camps, informal care, and missing institutional records? What are the clinical and informatics consequences of your answer?

  2. Marshall describes his relationship with healthcare as one of general trust and satisfaction, while Helen — in the same delivery room, with the same medical team — prepared him in advance to advocate for her in case she was incapacitated, because she anticipated being treated differently based on her race. Neither of them reports overt mistreatment during the birth itself. What does this convergence and divergence reveal about how epistemic injustice operates: does its absence from conscious experience mean it was absent from the encounter, and what tools — clinical, informatics, or otherwise — could detect a credibility differential that neither party may have named in the moment?

  3. Helen’s medical history is distributed across Rwanda, the Democratic Republic of Congo, Kenya, Swaziland, South Africa, and Chicago, with significant portions existing only in her mother’s oral account. FHIR’s patient matching and record-linking specifications assume a persistent, government-issued identity and a documented care history. Analyze the specific technical and epistemic failure modes that emerge when FHIR-based record retrieval is applied to a patient whose identity and care trajectory were shaped by statelessness — and explain what clinical knowledge is structurally unrepresentable in that architecture, not merely missing from it.

  4. The 21st Century Cures Act’s information blocking rules are designed to guarantee patients like Helen and Marshall access to their own health data through patient portals and open APIs. Identify a concrete scenario from either interview in which full, immediate data transparency — the stated goal of those rules — could produce harm rather than remedy it, and then articulate the genuine tension this creates: not between transparency and privacy in the abstract, but between an informatics policy designed to correct institutional power asymmetries and the specific ways that power asymmetry is experienced by patients whose distrust of institutions is historically grounded.

  5. Helen’s medical history is held partly in her own memory, partly in her mother’s oral account, and partly in records distributed across five countries — most of which are inaccessible. Marshall’s history is comparatively continuous, documented, and retrievable. FHIR’s patient-facing access APIs and the information blocking rules of the 21st Century Cures Act were designed to give patients like both of them access to their data — but the design implicitly assumes a documented past. What clinical knowledge about Helen is structurally unrepresentable in a FHIR-based record system, not merely missing from it — and what obligation does that gap place on a clinician encountering her for the first time?

Further Reading

1. Clinical

Geltman, P., Grant-Knight, W., Mehta, S., Lloyd-Travaglini, C., Lustig, S., Landgraf, J., & Wise, P. (2005). “The ‘lost boys of Sudan’: Functional and behavioral health of unaccompanied refugee minors resettled in the United States.” Archives of Pediatrics & Adolescent Medicine, 159(6), 585–591. doi:10.1001/archpedi.159.6.585

Divi, C., Koss, R., Schmaltz, S., & Loeb, J. (2007). “Language proficiency and adverse events in US hospitals: A pilot study.” International Journal for Quality in Health Care, 19(2), 60–67. doi:10.1093/intqhc/mzl069

Crear-Perry, J., Correa-de-Araujo, R., Lewis Johnson, T., McLemore, M., Neilson, E., & Wallace, M. (2021). “Social and structural determinants of health inequities in maternal health.” Journal of Women’s Health, 30(2), 230–235. doi:10.1089/jwh.2020.8882

Jacobs, E., Chen, A., Karliner, L., Agger-Gupta, N., & Mutha, S. (2006). “The need for more research on language barriers in health care: A proposed research agenda.” The Milbank Quarterly, 84(1), 111–133. doi:10.1111/j.1468-0009.2006.00440.x

2. Informatics

Mandel, J., Kreda, D., Mandl, K., Kohane, I., & Ramoni, R. (2016). “SMART on FHIR: A standards-based, interoperable apps platform for electronic health records.” Journal of the American Medical Informatics Association, 23(5), 899–908. doi:10.1093/jamia/ocv189

Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019). “Dissecting racial bias in an algorithm used to manage the health of populations.” Science, 366(6464), 447–453. doi:10.1126/science.aax2342

Hatef, E., Rouhizadeh, M., Tia, I., Lasser, E., Hill-Briggs, F., Marsteller, J., & Weiner, J. (2019). “Assessing the availability of data on social and behavioral determinants in structured and unstructured electronic health records: A retrospective analysis of a multilevel health care system.” JMIR Medical Informatics, 7(3), e13802. doi:10.2196/13802

Bonomi, L., Huang, Y., & Ohno-Machado, L. (2020). “Privacy challenges and research opportunities for genomic data sharing.” Nature Genetics, 52(7), 646–654. doi:10.1038/s41588-020-0651-0

3. Philosophical

Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford University Press. doi:10.1093/acprof:oso/9780198237907.001.0001

Dotson, K. (2011). “Tracking epistemic violence, tracking practices of silencing.” Hypatia, 26(2), 236–257. doi:10.1111/j.1527-2001.2011.01177.x

Medina, J. (2013). The Epistemology of Resistance: Gender and Racial Oppression, Epistemic Injustice, and Resistant Imaginations. Oxford University Press. doi:10.1093/acprof:oso/9780199929023.001.0001

Kidd, I., Medina, J., & Pohlhaus, G. (Eds.). (2017). The Routledge Handbook of Epistemic Injustice. Routledge. doi:10.4324/9781315212043

4. Anthropological / Social Science

Kleinman, A., Eisenberg, L., & Good, B. (1978). “Culture, illness, and care: Clinical lessons from anthropologic and cross-cultural research.” Annals of Internal Medicine, 88(2), 251–258. doi:10.7326/0003-4819-88-2-251

Farmer, P. (2004). “An anthropology of structural violence.” Current Anthropology, 45(3), 305–325. doi:10.1086/382250

Becker, G., Beyene, Y., Newsom, E., & Rodgers, D. (1998). “Knowledge and care of chronic illness in three ethnic minority groups.” Family Medicine, 30(3), 173–178. (DOI unverified)

Merry, L., Pelaez, S., & Edwards, N. (2017). “Refugees, asylum-seekers and undocumented migrants and the experience of parenthood: A synthesis of the qualitative literature.” Globalization and Health, 13(1), 75. doi:10.1186/s12992-017-0299-4