RESEARCH DOSSIER
Health Care Access, Institutional Trust, and Service Inequity
Abstract. Health care access is the practical possibility of obtaining appropriate, timely, acceptable, affordable, and continuous care when a health need arises. Institutional trust concerns expectations about the competence, integrity, fairness, and accountability of the organizations through which care is financed, allocated, delivered, and governed. Service inequity emerges when socially patterned differences in these conditions systematically alter access, the experience of care, or the likelihood that clinically appropriate services are received. This fundamental research monograph analyzes access as a multistage process and as a relation between people and institutions. It reconstructs the development of access theory; distinguishes availability, affordability, navigation, encounter quality, and effective receipt; reviews research on distrust, discrimination, insurance, geography, and digital exclusion; and evaluates randomized, observational, qualitative, and quasi-experimental evidence. A critical synthesis shows why increased nominal coverage may improve utilization and financial protection without necessarily eliminating disparities in provider availability or quality. It identifies causal-inference difficulties, psychometric questions, ethical tensions, and research priorities for accountable health systems in the Artificial Era.
Keywords: health care access; institutional trust; medical mistrust; health services utilization; service inequity; health equity; discrimination in health care; insurance coverage; affordability; geographic accessibility; patient experience; health system accountability; causal inference; digital health equity.
The Scientific Object of Health Care Access
Access becomes a scientific problem when there is a difference between the existence of medical services and a person's ability to benefit from them. A clinic may be present while its appointments are unobtainable, its premises inaccessible, its language unfamiliar, its charges unaffordable, or its procedures sufficiently threatening that people postpone contact. A person may possess coverage yet encounter a narrow provider network. Another may attend an appointment but receive a different clinical recommendation under otherwise comparable conditions. Access therefore refers to a sequence of possible transitions from recognizing a need to obtaining an appropriate service and realizing its potential benefits. The problem requires investigation of both institutional arrangements and the psychological processes through which those arrangements are perceived, navigated, and experienced. Aday, L. A., & Andersen, R (1974) placed utilization within a broader framework of health policy, delivery systems, and populations at risk.
This monograph defines its object as the relation among three domains: the opportunity structure of care, the expectations and actions of potential patients, and the distribution of actual service receipt. Its central questions are how barriers are produced, how trust is warranted or damaged by institutions, which observed differences constitute inequities, and which interventions alter outcomes rather than merely service counts. It treats trust as relevant when it changes disclosure, consent, return visits, adherence to agreed care plans, or expectations about fairness. The unit of analysis may be a patient encounter, a provider network, a local service area, or a population moving through a care pathway. Each unit yields different evidence and demands explicit causal boundaries.
Defining Access, Utilization, and Effective Receipt
A useful scientific vocabulary distinguishes potential access, realized access, and effective access. Potential access is a set of enabling conditions: available professionals, eligible financing, physical routes, communication accommodations, and appointment opportunities. Realized access concerns actual service use, such as an attended visit or completed diagnostic test. Effective access adds whether care is appropriate to clinical need, adequately delivered, and able to contribute to a relevant health outcome. These dimensions are related but can diverge. Increased appointment counts could accompany insufficient time per consultation, and reduced emergency department visits could represent improved preventive care or harmful avoidance. Evaluation must therefore connect each indicator to a defined clinical or population question.
Utilization is an imperfect proxy for need satisfaction because people with greater morbidity may appropriately use more care. Equally distributed use across groups with unequal clinical need may itself signal service inequity. Conversely, a high rate of service use may indicate fragmentation, unnecessary testing, or repeated failed attempts to secure help. Levesque, J.-F., Harris, M. F., & Russell, G (2013) frame access as a progression through identifying needs, seeking help, reaching care, obtaining services, and having health needs addressed. Measurement must specify the denominator—eligible population, diagnosed patients, people with symptoms, or those attempting an appointment—and the stage at which an inequality arises.
Historical Origins of the Access Problem
Access research developed across medical sociology, social medicine, health economics, epidemiology, and health services research. In 1971 Julian Tudor Hart articulated the inverse care law, focusing attention on the tendency for the availability of good medical care to vary inversely with population need under certain market conditions. The original article, Hart (1971)92410-X), is historically important as a critique of how medical resources can be distributed. It offers a research question about institutions rather than an invariant law that applies identically to every system. Its empirical relevance depends on measured provider supply, health need, financing, and administrative rules.
Aday, L. A., & Andersen, R (1974) subsequently offered a framework connecting the characteristics of delivery systems and populations to utilization and satisfaction. Penchansky, R., & Thomas, J. W (1981) identified dimensions of fit between service and patient. By the late twentieth century, an increasingly sophisticated literature examined insurance, physician supply, care quality, patient preferences, and organizational processes. The 2003 Institute of Medicine report Institute of Medicine (2003) synthesized evidence on racial and ethnic disparities within United States health care and located the problem in interactions among systems, clinicians, and patients. Modern health-equity research extends this lineage by linking local encounters to laws, financing, spatial arrangements, and histories of unequal treatment. Historical concepts remain valuable when their assumptions and measurement contexts are made explicit.
Andersen's Behavioral Model of Service Use
The behavioral model groups explanations for service use into predisposing characteristics, enabling resources, and need. Predisposing conditions include demographic and social patterns that shape the likelihood of seeking care; enabling conditions include coverage, income, transport, and the existence of providers; need includes perceived symptoms and clinically evaluated illness. Andersen, R. M (1995) revisited the model after decades of applications and revisions, emphasizing that access and use must be understood within a changing social and institutional environment. The model encourages researchers to distinguish a person's observed need from the resources that translate that need into service receipt.
Its explanatory strength lies in specifying where variables belong, but its categories should not be converted into a deterministic checklist. Income may alter transportation, appointment flexibility, internet access, and insurance choice; discrimination may affect both individual expectations and institutional delivery. The same variable can operate at several causal levels. For example, an association between educational attainment and screening might reflect knowledge, provider recommendation, paid leave, or geographical sorting. A satisfactory study uses the framework to generate rival hypotheses, measures pathways separately, and determines whether the effect of an enabling resource varies with illness severity and social position.
The Five-Dimension Fit Between Patients and Services
Penchansky, R., & Thomas, J. W (1981) described access through availability, accessibility, accommodation, affordability, and acceptability. Availability concerns the volume and types of services relative to need. Accessibility concerns travel, geographic location, and physical reachability. Accommodation describes the organization of appointment and service systems. Affordability relates charges to household resources and financial arrangements. Acceptability refers to the fit between users' and providers' attitudes and characteristics. These distinctions give researchers concrete hypotheses about why superficially similar populations may experience different access.
An adequate evaluation should resist collapsing the five dimensions into a single convenience score without first verifying its measurement model. Two neighborhoods might have equal numbers of clinicians per resident but different average travel times or waiting lists; the apparently more accessible area may contain only services excluded by local insurance plans. Similarly, changing clinic hours can improve accommodation without changing affordability or acceptability. Access dimensions are partly substitutable and partly sequential: a mobile clinic may reduce travel burden, but its impact will remain limited if it lacks continuity or interpretation. Such interactions should be studied rather than assumed.
Access as a Patient and Health-System Interface
Levesque, J.-F., Harris, M. F., & Russell, G (2013) further specified service accessibility in terms of approachability, acceptability, availability and accommodation, affordability, and appropriateness, corresponding to patients' abilities to perceive, seek, reach, pay, and engage. The model expands access beyond clinic entry to a dynamic interface. A service must be visible as an option before it can be approached, culturally and ethically intelligible before it can be sought, reachable before it can be obtained, and sufficiently appropriate for continued engagement. These capacities are influenced by literacy, disability, social resources, information, and past experience with institutions.
This interface model has an important methodological consequence: barriers accumulate across stages. If 80 percent of an eligible population knows that care is available and 80 percent of those aware can obtain appointments, only a smaller proportion reaches the next stage even before financing and appropriateness are considered. This is an illustrative arithmetic relationship rather than an empirical estimate. The model directs research toward transition probabilities, time to service, attrition points, and interactions among system-side and person-side capabilities. It also cautions against interpreting every instance of nonuse as an individual decision independent of institutional design.
Institutional Trust as a Psychological Construct
Institutional trust is a prospective expectation that an organization will act competently, reliably, honestly, and fairly in circumstances where individuals are vulnerable and possess incomplete information. In health care, patients typically cannot verify technical competence in real time and frequently disclose sensitive information or accept interventions whose consequences extend beyond the visit. Hall, M. A., Dugan, E., Zheng, B., & Mishra, A. K (2001) synthesized the emerging literature on trust in physicians and medical institutions, showing why it must be distinguished from satisfaction, confidence, and other adjacent attitudes. Trust is relational: it depends on a particular object, domain, and exposure to potential harm.
A person's trust in a familiar clinician may coexist with distrust of insurers or hospitals. Trust in medical research differs from expectations about the confidentiality of electronic records, and confidence in professional knowledge differs from approval of billing practices. Researchers should therefore define the object of trust before interpreting group differences. Psychological mechanisms may include learning from prior encounters, interpreting professional explanations, assessing procedural fairness, anticipating discrimination, and observing whether institutions acknowledge mistakes. Trust matters to access insofar as it affects stages of help-seeking and continued participation; its role should be tested as a potential mediator or moderator rather than asserted as the universal reason for differences in use.
Measuring Trust and Distrust
Trust measurement faces an immediate construct problem: a survey item about honesty may capture perceived integrity, whereas an item about providing equal treatment may assess fairness and a question about protecting records may assess privacy. Shea, J. A., Micco, E., Dean, L. T., McMurphy, S., Schwartz (2008) developed a revised Health Care System Distrust scale, offering a way to study distrust of institutions instead of inferring it from discontent with one physician. Armstrong, K., Rose, A., Peters, N., Long, J. A., McMurphy (2006) examined associations between health-system distrust and self-reported health in the United States. These contributions establish measurable research programs without proving that every observed association represents an effect of trust.
Scientific evaluation requires factor structure, test–retest reliability when appropriate, measurement invariance across languages and populations, and discriminant validity relative to generalized cynicism or political distrust. Researchers must decide whether trust and distrust occupy opposite ends of one scale or contain distinguishable components. Social-desirability effects, prior illness, service familiarity, and survey timing may alter responses. Cross-sectional relations cannot settle causal direction: people with poor health may need more contact, experience more institutional failures, and consequently express more distrust. Better designs follow trust and encounters over time, test specific policy changes, and examine whether measured shifts predict observable steps in service receipt.
Historical Experience, Discrimination, and Institutional Credibility
Medical institutions possess histories, and their histories become information that people can use when estimating risk. Trust can be damaged by perceived or documented unequal treatment, opaque charges, disrespectful interactions, or failures of accountability. Boulware, L. E., Cooper, L. A., Ratner, L. E., LaVeist, T. A (2003) investigated racial differences in trust concerning physicians, hospitals, and health insurance plans through a cross-sectional survey. The distinct institutional objects in that study matter: a respondent may evaluate hospital administration differently from the physician they see. Group-level variation in distrust should be investigated in relation to experienced treatment, historical structures, and contemporary institutional incentives.
A central interpretive safeguard is to avoid presenting mistrust itself as a population defect. The research question concerns how expectations are formed and whether institutions supply credible grounds for confidence. Institute of Medicine (2003) organized evidence that differences in treatment can persist in clinical settings beyond differences in insurance status and health need. Modern research must measure perceived discrimination alongside audited service decisions and organizational conditions. An intervention that asks patients to become more trusting while leaving discriminatory allocation unchanged risks confusing a survey score with improvement in the care system.
Structural Exclusion and Health System Segregation
Structural exclusion refers to patterned institutional arrangements that distribute the chance of obtaining appropriate health services unevenly across populations. Its mechanisms include residential segregation, unequal local investment, insurance network segmentation, restrictive eligibility rules, hospital closure, and employment conditions that make attending care costly. Bailey and colleagues (2017)30569-X) conceptualized structural racism as the mutually reinforcing effects of social institutions, including health care, housing, employment, education, and public policy. In the present dossier, the empirical focus is narrower: which parts of that system create measurable differences in entry to clinical care, the quality of encounters, and continuity after treatment begins? A study of structural inequality can specify these pathways without attributing every health outcome to one cause.
One mechanism is the separation of service markets. If insurance plans, specialist referral relationships, or transit systems channel groups toward different facilities, the average technical quality of an area's hospitals is a poor representation of the care that each group can actually obtain. Bach, P. B., Pham, H. H., Schrag, D., Tate, R. C., & Hargrav (2004) demonstrated an important historical example of uneven patient distribution across primary care physicians in United States Medicare data. Such a pattern may operate through residential location, accepted insurance products, physician practice organization, and patients' accumulated experiences. These channels must be investigated individually when causal effects are asserted. The ecological fact that two groups use different physicians does not reveal whether each difference was caused by patient choice, referral restrictions, service supply, discrimination, or some combination.
The distinction between structural and encounter-level mechanisms is essential for policy evaluation. A hospital may adopt communication training while its referral network continues to exclude many residents; an insurance expansion may improve legal eligibility while specialist wait lists remain different across neighborhoods. A rigorous study should use at least two denominators: people with comparable clinical need in the catchment population, and those who successfully enter a specific service. Comparing only treated patients can hide selection into care. Spatial supply, referral records, appointment attempts, administrative denials, clinician workload, and patient-reported experience can together reveal the institutional sequence. Levesque, J.-F., Harris, M. F., & Russell, G (2013) provide a model for linking system accessibility with individual capacities rather than treating structural and psychological processes as separate explanations.
Historical analysis also requires temporal precision. Discriminatory laws, financing systems, migration, service closures, and population health have changed across decades; effects identified in one location should not be assigned to all systems or historical periods. The value of a structural account lies in generating testable predictions: where service supply is reorganized, where networks become more inclusive, or where transport routes change, a particular stage of access should improve for particular groups. Strong designs identify the relevant exposure and measure what occurs between a changed institutional rule and the completed clinical outcome. Policy accountability then concerns observable opportunity and service quality in addition to formal nondiscrimination commitments.
Implicit Bias, Explicit Conduct, and Professional Decision-Making
The psychology of unequal clinical encounters includes the possibility that professionals respond differently to information about patients' social identities. Implicit associations, explicit beliefs, expectations about adherence, differential attention, time pressure, and institutional decision rules are potential mechanisms. Hall, W. J., Chapman, M. V., Lee, K. M., Merino, Y. M., Thom (2015) systematically reviewed research on implicit racial and ethnic bias among health professionals and its relation to health-care outcomes. Their work helps locate implicit bias within a broader explanatory program; measures of implicit association and observed clinical behavior answer different questions and should be evaluated separately. An implicit association score is neither a clinical outcome nor proof of discriminatory intent in a particular encounter.
The experimental reasoning behind this research can be stated directly. If investigators present clinicians with otherwise comparable clinical information while varying a social cue, differences in recommendations may reveal a response to that cue within the designed task. The Schulman, K. A., Berlin, J. A., Harless, W., Kerner, J. F (1999) used scripted presentations for precisely this kind of question. Because the clinical scenarios were standardized, the contrast helps isolate differences in decision making in that setting. Its limitations also matter: a simulated encounter lacks some features of sustained clinical relationships, and actual treatment depends on multiple later steps. Experimental findings need replication under different conditions and corroboration with clinical quality data.
In natural encounters, differences in diagnostic tests and specialist referrals are more difficult to interpret. Patients differ in clinical risk, preferences, contraindications, service history, and the information available to clinicians. Yet adjusting for every recorded factor may itself be problematic when records reflect earlier unequal access. For example, a prior diagnosis can be a mediator of previous service availability. Analysts should present results under alternative adjustment sets and distinguish clinical differences justified by need from disparities lacking a documented basis. Recordings, direct observation, case audits, matched standardized patients, and prospective patient follow-up each offer different checks against biased interpretation.
Interventions should therefore target observable behavior and institutional incentives. Structured clinical protocols may reduce discretionary variation where guidelines are appropriate, while communication support and protected consultation time can increase the opportunity to elicit patient concerns. Performance monitoring should identify differences in recommended, accepted, and completed care rather than using a single attitude score as a surrogate for equity. The literature permits investigation of implicit and explicit psychological mechanisms, but improvements in access must be demonstrated through decisions and patient outcomes.
Clinical Encounters and Differential Recommendations
Access continues inside the examination room. Schulman, K. A., Berlin, J. A., Harless, W., Kerner, J. F (1999) used scripted patient presentations to study whether physicians' recommendations for cardiac catheterization varied by patient race and sex. The experiment demonstrated that clinical decision making could be studied under standardized presentation conditions. Its physician sample, hypothetical scenarios, and historical setting require attention when generalizing to contemporary practice. The study does not by itself identify the mechanism behind every disparity or imply that all physicians make the same decisions. It does show why measuring visits alone is insufficient: recommendations and subsequent care can differ even after contact occurs.
van Ryn and Burke (2000)00338-X) examined how patient race and socioeconomic status related to clinicians' perceptions during a post-angiogram encounter. Such findings point to social categorization and differential expectations as plausible components of encounter inequity. Observational analyses of clinician perceptions remain sensitive to contextual confounding, whereas vignettes trade some real-world validity for standardization. A mature evidence program combines recordings of actual encounters, blinded chart review, standardized patients, service-quality outcomes, and prospective follow-up. The aim is to identify specific moments when clinically relevant information is discounted, follow-up is not arranged, or appropriate alternatives are not explained.
Patient–Clinician Concordance and Communication
One proposed mechanism of differential experience concerns concordance between patients and clinicians in race, ethnicity, language, or other social characteristics. These categories are proxies for multiple processes and should not be treated as universal biological explanations. In an experiment in Oakland, Alsan, M., Garrick, O., & Graziani, G (2019) randomly assigned Black male participants to Black or non-Black male physicians and examined preventive-service choices before and after the clinical interaction. Greater uptake after interaction with racially concordant clinicians was particularly notable for some invasive preventive services. The design strengthens causal inference for the assignment in that setting; it does not establish that clinician identity is the only mechanism or that the resulting uptake automatically produces the projected mortality effect.
A distinct Takeshita, J., Wang, S., Loren, A. W., Mitra, N., Shults, J (2020) assessed associations between patient–physician racial/ethnic and gender concordance and patient experience ratings. Such ratings reveal perceptions of care but are vulnerable to survey nonresponse and case-mix differences. The two designs answer different questions: a randomized assignment can identify the effect of a particular encounter configuration, while experience ratings describe patterns that warrant further explanation. Practical implications concern language services, respectful communication, workforce diversity, consistent informed consent, and trustworthy quality measurement, each requiring direct evaluation on its own terms.
Insurance, Entitlement, and Financial Access
Insurance coverage changes the expected price of care, the risk of medical debt, the providers people may use, and sometimes the administrative tasks required to obtain services. It does not fully determine practical access because deductibles, copayments, prior authorization, network adequacy, transportation, and appointment supply remain relevant. In United States research the association between coverage and care use is substantial, but causal interpretation requires attention to differences between insured and uninsured populations. Poorer health may induce some people to seek insurance, while employment and household income affect both insurance and access.
The review by Sommers, B. D., Gawande, A. A., & Baicker, K (2017) integrates evidence concerning coverage and health. It must be read alongside the individual studies rather than substituted for them. Coverage reforms may influence utilization, financial protection, self-reported health, psychological distress, and some clinical indicators on different schedules. Health outcomes are more distal than insurance enrollment, and studies with short follow-up may capture intermediate benefits before changes in morbidity or mortality are detectable. The exact policy, population, and comparator should accompany every generalization about expanding access.
The Oregon Medicaid Experiment
The Oregon Health Insurance Experiment is central because a lottery offered a comparatively strong design for examining coverage among low-income uninsured adults. Finkelstein, A., Taubman, S., Wright, B., Bernstein, M., Gru (2012) studied the first year after lottery assignment and reported increased insurance coverage, greater use of several forms of care, reduced financial exposure, and improvements in self-reported health. Lottery selection altered the opportunity to apply for Medicaid rather than guaranteeing coverage, so the causal estimand must be described carefully. The paper reports an approximately 25-percentage-point insurance difference between selected and nonselected groups in the first year.
A later evaluation by Baicker, K., Taubman, S. L., Allen, H. L., Bernstein, M., Gr (2013) assessed outcomes roughly two years after the lottery, including physical measurements, depression screening, service use, and spending. Their report found increases in access and financial security and a reduction in depression, while detecting no statistically significant effects on several measured physical-health indicators over that interval. Failure to detect a change in particular biomarker outcomes is not evidence that coverage has no effect on health, just as greater use is not evidence that all care is beneficial. The experiment shows why access research should predefine intermediate and distal endpoints and disclose statistical uncertainty. It also establishes the limits of transporting an experiment in one state and historical period to every contemporary insurance system.
Quasi-Experimental Evidence from Coverage Expansions
Natural experiments exploit policy differences when random assignment is unavailable. Sommers, B. D., Baicker, K., & Epstein, A. M (2012) compared states that expanded Medicaid eligibility with neighboring states before and after the expansion and reported an association with lower adult mortality and improved access. Their adjusted analysis estimated a relative mortality reduction of 6.1 percent in the specified comparison. Because states selected policies rather than being randomized, the inference depends on comparability, parallel underlying trends, and control of concurrent changes. A credible synthesis therefore distinguishes the stronger internal identification of a lottery from the broader geographical reach and assumptions of a policy comparison.
Sommers, B. D., Blendon, R. J., & Orav, E. J (2016) compared access changes after different forms of Medicaid expansion in Arkansas and Kentucky with nonexpansion in Texas. Difference-in-differences methods used survey waves to investigate policy-associated changes. This evidence adds institutional variation: financing methods and plan design may differ while expansions improve opportunities for care. Nevertheless, telephone-survey composition, self-reported utilization, regional trends, and unmeasured contemporaneous policies remain limitations. Researchers should examine pretrends, robustness to state-specific trajectories, heterogeneity among subgroups, and the durability of effects over time.
Affordability, Medical Debt, and Foregone Care
Affordability is a relationship between costs and resources, not a property of a posted price alone. The full cost of attending a visit can include travel, childcare, unpaid leave, medications, follow-up tests, and the uncertainty of receiving a bill. A household may technically qualify for covered care while experiencing substantial liquidity constraints. Researchers must distinguish premiums, cost sharing, out-of-network charges, medical debt, catastrophic expenditure, and delayed or foregone care. These measures illuminate different mechanisms. Financial protection is an intrinsically important outcome because the consequences of illness include losses in consumption and household security beyond clinical events.
National evidence demonstrates the continuing relevance of cost barriers. The Federal Reserve Board (2026) found that 26 percent of surveyed adults had forgone some form of medical care in that year because of cost, down from 28 percent in 2024. This is a self-reported survey indicator across types of care rather than an estimate of clinically avoidable deaths. The proportion and survey frame should not be mechanically combined with administrative insurance statistics. A policy evaluation should ask whether reduced cost sharing changes timely appropriate use, out-of-pocket financial strain, and differences between those with higher and lower health need.
Geographic Access and Transportation
Distance to a facility, transit availability, travel time, traffic, physical terrain, and the need to accompany dependents can alter care seeking. Geography becomes a psychological burden when reaching care requires uncertainty, repeated planning, or loss of income. Syed, S. T., Gerber, B. S., & Sharp, L. K (2013) synthesized 61 United States studies addressing transport barriers in continuing primary or chronic-disease care. They found evidence that transportation impediments affect care access, particularly for lower-income and uninsured populations, while emphasizing variability in definitions, measurements, and outcomes. Their review supports investigating transport as a substantive barrier without establishing a universal effect size for every region.
Access mapping should account for actual routes, clinic schedules, service eligibility, accessible vehicles, and whether a provider is accepting new patients. A straight-line distance may understate travel burden when routes are indirect or public transportation is infrequent. Population averages conceal uneven exposure among people with mobility limitations, rural residents, and households without a vehicle. Studies of transport interventions should evaluate attended necessary appointments, delayed diagnoses, continuity, patient time, and total system cost, while guarding against selection effects in voluntary transport programs. Geographic proximity is a useful input; the capacity to complete an appropriate care pathway is the outcome of interest.
Provider Distribution, Primary Care, and Continuity
The number of clinicians in a population is an incomplete measure of access. Profession, specialty, appointment availability, referral relationships, insurance participation, working hours, and cultural and language competencies alter effective supply. Bach, P. B., Pham, H. H., Schrag, D., Tate, R. C., & Hargrav (2004) analyzed visits involving older Medicare beneficiaries and found that Black and White patients were often treated by different segments of the primary care physician workforce. The result illustrates how structural segregation in provider markets can expose populations to different practice resources. Historical and system-specific findings should be assessed against contemporary workforce distribution rather than presumed constant.
Starfield, B., Shi, L., & Macinko, J (2005) reviewed evidence connecting primary care to health systems and equitable health outcomes. Primary care's potential contribution depends on first contact, continuity, comprehensiveness, and coordination. A patient who repeatedly searches for a new clinician may face repeated history-taking, duplication, and weak follow-up even where many appointments occur. Continuity can also be problematic if poor-quality or discriminatory relationships persist, so its benefits should be evaluated together with technical and interpersonal quality. Effective supply is best understood as a network of reliable services over time rather than a static count of facilities.
Appointment Systems and Organizational Friction
Administrative requirements make health care access a process of labor. Patients may need to verify eligibility, navigate referral rules, complete forms, wait through telephone queues, resubmit documents, obtain prior authorizations, and reconcile conflicting instructions. These tasks consume time and attention and can be especially burdensome during illness. The accommodation dimension of Penchansky, R., & Thomas, J. W (1981) anticipates this organizational problem, while the framework of Levesque, J.-F., Harris, M. F., & Russell, G (2013) clarifies how institutional processes interact with people's capacity to reach and engage with services.
To study friction rigorously, researchers need workflow data: the number of contacts required to secure a visit, elapsed days from clinical indication to completed care, cancellation and no-show patterns, authorization denials, appeal reversals, and differences between groups. A shorter average waiting time may still mask an inaccessible tail of the distribution. Measurements should separate a patient's preference to defer care from deferral caused by system rules. A successful appointment intervention may shift pressure downstream to diagnostics or specialists, so analysis should follow the entire sequence instead of celebrating a faster first click.
Nonfinancial Barriers and Unequal Time Budgets
Cost-based measures can overlook substantial barriers among people who can technically afford an appointment. Time away from paid work, childcare, transportation changes, inaccessible telephone systems, inflexible clinic hours, and the difficulty of finding an available clinician may all prevent appropriate care. Kullgren, J. T., McLaughlin, C. G., Mitra, N., & Armstrong (2012) examined nonfinancial barriers and unmet or delayed need among United States adults. Their study provides empirical grounds for treating practical constraints as separate from insurance and household income; its survey-based associations still require careful interpretation of self-reported barriers and population selection.
The concept of an unequal time budget connects health services to work organization and caregiving without assuming that every patient has the same schedule or resources. Someone whose employment permits paid leave can complete a brief visit with little economic disruption, whereas an hourly worker may need to forgo wages or risk job security. A parent may require two transport legs and alternative childcare to attend one appointment. These costs are partially monetary but can also take the form of scheduling uncertainty, fatigue, and lost control over daily activity. From an access perspective, the relevant question is whether an individual can complete the necessary service sequence under actual conditions, not whether a facility's nominal appointment slot exists.
Measuring these barriers demands more than asking patients why they missed a scheduled visit. Individuals who were never able to schedule are absent from that denominator. Researchers can record time from first contact to booked appointment, unsuccessful telephone attempts, total travel and waiting time, cancellation reasons, work hours lost, number of required visits, and changes in referrals. Survey questions should distinguish preferences from constraints and permit multiple coexisting barriers. Cross-sectional associations may reflect sickness severity, household composition, labor-market conditions, or accumulated experiences with providers; longitudinal designs can measure whether service redesign changes particular obstacles before outcomes shift.
The practical consequences extend to intervention assessment. Adding evening appointments could reduce missed work, but benefits depend on transport availability and workforce support. Digital scheduling could remove telephone queues while excluding people who lack devices or who need human assistance. A transportation voucher might affect completed visits while leaving referral delays unchanged. The institution's obligation is to measure which needs become easier to meet, for whom, and under what operating conditions. Penchansky, R., & Thomas, J. W (1981) remains useful because it focuses on the fit between service organization and everyday circumstances.
Medical Interpretation and the Quality of Participation
Health service access depends on the patient's ability to communicate symptoms, questions, values, and consent in a form usable by qualified professionals. Interpretation is therefore part of clinical quality rather than an optional convenience. A review by Flores, G (2005) surveyed the literature on the effects of medical interpreter services and found that limited language support was associated with poorer care and that trained interpreters or bilingual clinicians improved several communication-related outcomes compared with untrained ad hoc arrangements. Because the review incorporated heterogeneous studies, it should be used as a synthesis of a historically important evidence base rather than a single causal effect estimate.
Three different failures can occur in an inadequately supported encounter. First, the patient may not be able to communicate clinically important information, altering diagnostic reasoning. Second, the clinician may provide accurate advice that is misunderstood or cannot be acted upon. Third, shared decision making may fail because the patient cannot ask clarifying questions or express preferences. These mechanisms concern the accuracy and reciprocity of information, patient autonomy, and the continuity of care. Measuring language access as the availability of translated brochures alone misses the live conversational work required during diagnostic uncertainty, consent, and follow-up.
A strong evaluation compares clinically meaningful communication outcomes. Appropriate measures include the proportion of encounters with qualified interpretation when needed, clinically important omissions or mistranslations, patient understanding assessed in a preferred language, completion of indicated follow-up, and documented opportunities to ask questions. Patient satisfaction is informative but cannot substitute for these measures. Studies must distinguish language proficiency, literacy, hearing or speech disabilities, familiarity with the medical system, and the availability of support persons. Each dimension can require different accommodations. It is also important to protect privacy and the patient's choice about which information may be shared with accompanying relatives.
Digital translation and speech-recognition interfaces add further challenges. Performance can vary with accent, domain-specific vocabulary, background noise, network reliability, and rare clinical terminology. An apparent improvement in conversational speed is insufficient if diagnostic meaning or informed consent is degraded. Auditing digital language services requires clinical error rates, subgroup performance, secure data handling, and clear routes to human interpretation. The broader scientific principle is that communication access is a measurable stage of appropriate service receipt. Its success is established by patient participation and accurate clinical exchange rather than the mere presence of an interface.
Disability and Accessible Service Design
For people with disabilities, access depends on more than the number and location of facilities. Examination tables, diagnostic equipment, communication alternatives, sensory environments, transport, digital interfaces, and staff training may determine whether a person can safely participate in clinical assessment. Eligibility and reimbursement policies also affect the provision of accommodations. These mechanisms fit both the geographic-access and organizational-accommodation dimensions of the established frameworks, yet disability research requires direct consultation with affected populations and condition-sensitive measures of effective care.
A meaningful evaluation distinguishes architectural entry, appointment completion, the technical adequacy of an examination, and the person's control over disclosure and consent. Administrative encounter rates may overstate accessibility when clinically indicated tests are never completed. Researchers should examine disability categories only insofar as the categories correspond to actual barriers and accommodations, avoiding the assumption that all disabled patients share an experience. Participatory design, accessibility audits, and paired observational and administrative measures can reveal which changes improve service receipt. Equity is assessed by the fit between the person's needs and the service's obligations, resources, and procedures.
Digital Health, Telemedicine, and New Access Barriers
Telemedicine changes the location of some care encounters and can reduce transport time, work disruption, and exposure to physical barriers. It simultaneously introduces requirements for devices, broadband, private space, digital literacy, accessible interfaces, and permission to use telephone rather than video. Eberly, L. A., Kallan, M. J., Julien, H. M., Haynes, N., Kha (2020) examined telemedicine access among 148,402 scheduled patients in a large United States health system early in the COVID-19 pandemic. Their analysis found group differences in visit completion and video use associated with age, language, insurance, income, and racial or ethnic categories. The findings are observational and specific to a rapidly changing emergency period; they nevertheless illustrate why counting telehealth offerings does not establish equitable access.
Digital access should be measured as a chain from receiving an invitation to establishing a connection, completing a clinically appropriate consultation, accessing any subsequent tests, and understanding the resulting plan. A video-only design may disadvantage patients who could have received meaningful telephone services; a telephone-only design may be inadequate for some assessments. Researchers should compare clinical appropriateness and accessibility of modalities under defined circumstances rather than assuming technological sophistication is equivalent to quality. Privacy and data governance also affect willingness to seek help, especially when sensitive information is handled by platforms whose policies are difficult to interpret.
Cumulative Inequity Across the Care Pathway
Service inequity is often cumulative. Consider two groups with comparable clinical indications that differ modestly in the probabilities of recognizing a service, securing an appointment, reaching the facility, receiving a recommended test, and returning for follow-up. Multiplying the transition probabilities can produce a sizeable final difference even when no single stage appears overwhelming. This formulation is a conceptual model of sequential attrition, not a claim that the transitions are independent in actual populations. In practice, earlier failure may increase later burden, and repeated encounters may alter both health and trust.
The pathway approach makes it possible to compare bottlenecks across insurance, geography, scheduling, clinical recommendations, and ongoing care. It also clarifies why an intervention aimed at one stage may produce small final effects when other stages remain blocked. Institute of Medicine (2003) and the Levesque, J.-F., Harris, M. F., & Russell, G (2013) provide complementary system and process perspectives. Longitudinal administrative records linked to patient experiences can identify loss points, but privacy-protective data linkage and careful handling of missing observations are required. A population that never contacts the system will be absent from encounter-based datasets by construction.
Measuring Access Across Different Data Sources
No single dataset captures the whole access process. Household surveys detect perceived unmet need and foregone care among people who never appear in clinical records; insurance files reveal enrollment and charges; health-system scheduling logs show attempts and wait times; electronic health records document encounters and some quality measures; geospatial data estimate physical accessibility; qualitative interviews reveal why a nominally available option was unusable. Each source has its own selection process, missingness, and construct limitations. Combined measurement can triangulate observed patterns, provided that links across data are legally, ethically, and statistically defensible.
The denominator deserves special attention. A percentage of completed visits among scheduled patients excludes individuals unable to schedule. A measure of specialist referrals completed among those referred excludes patients whose clinicians never offered referral. Readmission and emergency-use measures require clinical risk adjustment and should not automatically be classified as indicators of access failure. Analysts should report absolute differences, relative differences, uncertainty, and the health need against which differences are judged. Surveys of trust should be evaluated for language and group measurement invariance; administrative categories often encode institutional histories that may distort comparisons across periods.
Causal Designs and the Interpretation of Inequity
Randomized assignment of a coverage opportunity, a navigation program, or appointment assistance may identify causal effects under clear assumptions. The Oregon experiment illustrates the value of randomization while also making visible imperfect treatment uptake and generalizability limits. Quasi-experimental designs, including difference-in-differences, interrupted time series, and regression discontinuity, can evaluate policies when randomized allocation is infeasible. Such designs require documented exposure timing, stable measurement, appropriate comparison units, and explicit examination of underlying trends. Sommers, B. D., Baicker, K., & Epstein, A. M (2012) and Sommers, B. D., Blendon, R. J., & Orav, E. J (2016) offer examples of policy-focused approaches with different data and identification constraints.
For trust, mediation analysis is especially demanding. If an institution changes scheduling, professional staffing, and communication simultaneously, any observed increase in trust may accompany improvements without constituting their causal mechanism. Conversely, policy effects may operate partly through trust while surveys miss the relevant domain. Longitudinal designs, prespecified outcomes, negative controls, replication across settings, and qualitative process evidence strengthen inference. There is also a risk of adjusting away a pathway that constitutes inequity: controlling for insurance in an analysis of a social disparity can answer a conditional question while concealing part of the institutional disadvantage being investigated. Researchers must state whether they estimate total disparities or direct effects conditional on mediators.
Comparative and Global Perspectives
Health systems differ in public financing, workforce organization, formal entitlements, geographical coverage, out-of-pocket payment, quality regulation, and methods of patient participation. The World Health Organization (2025) situates unequal opportunity within broader distributions of power and resources. This contextual analysis is adjacent to but broader than the present dossier, which focuses on mechanisms connecting people to clinical services. Comparing countries requires operational consistency: having legal coverage in one jurisdiction may entail very different practical service availability, waiting times, or financial exposure in another.
The joint WHO–World Bank World Health Organization & World Bank (2025) tracks service coverage and financial hardship with indicators referring to different calendar years. Its global coverage index increased from 54 in 2000 to 71 in 2023, while the share experiencing financial hardship from out-of-pocket health spending was estimated at 26 percent in 2022 under the report's definition. These figures characterize global monitoring constructs, not identical outcomes observed for every country or person. Equity research should compare within-country gradients and disparities in effective care rather than ranking systems solely by national averages.
Evaluating Interventions to Improve Access
Interventions can be organized by the barrier they are intended to remove. Coverage and cost-sharing changes address financial exposure. Expanded services and workforce incentives target availability. Transportation assistance and mobile care address reachability. Extended hours, simplified registration, and transparent referral processes address accommodation. Interpretation, inclusive equipment, and respectful communication address appropriateness and acceptability. Some programs, such as patient navigation, connect several dimensions. An evaluation should articulate the intended mechanism before measuring outcomes and specify the services whose receipt is clinically appropriate for the population.
The strongest useful result is a change in need-adjusted effective receipt or a defensible intermediate measure linked to it. An apparent increase in screening may reflect useful prevention, but studies should also examine follow-up after positive results and potential overuse. Satisfaction may improve without clinical care improving; conversely, an ethically justified difficult conversation may temporarily lower satisfaction. Experimental trials, pragmatic evaluations, and natural experiments can complement interviews that document how programs function in practice. The evidence from Alsan, M., Garrick, O., & Graziani, G (2019) demonstrates that interpersonal conditions can change preventive choices, while Finkelstein, A., Taubman, S., Wright, B., Bernstein, M., Gru (2012) show that financing opportunities influence use and financial exposure. Their mechanisms are different and should be interpreted at their respective levels.
Institutional Accountability and the Ethics of Trust
Trust is warranted when institutions demonstrably perform the duties on which people depend. Ethical responsibilities therefore include informed communication, fair access rules, professional competence, usable complaint procedures, protection of sensitive information, and public explanation of consequential decisions. These responsibilities have a psychological consequence: experiences of consistency and fair treatment provide evidence from which people can update expectations. A health system that improves its messaging while maintaining opaque denials or discriminatory practices may raise short-term approval scores without earning stable trust. The literature on trust in institutions, beginning with Hall, M. A., Dugan, E., Zheng, B., & Mishra, A. K (2001), provides a basis for distinguishing the interpersonal reliance required in medicine from endorsement of an organization's reputation.
Institutional justice also entails hearing from populations who are absent from routine service-feedback systems. Surveys limited to current patients may omit those excluded by geography, payment, scheduling, language, or prior harms. Transparent reporting should present both quality among patients reached and unmet need among those not reached. Participation in service design, independent auditing, and accessible channels for contesting decisions are researchable features of accountability. Their value must be evaluated through actual changes in experience, service allocation, and clinical opportunity. Ethical analysis informs what ought to be monitored; empirical research establishes what a given procedure actually changes.
Artificial Systems and the Future of Health Service Access
In the Artificial Era, digital interfaces and artificial systems increasingly mediate information, scheduling, eligibility, communication, and clinical decision support. Their presence shifts the access question from physical and administrative institutions alone to a combined human–artificial service architecture. A digital triage assistant may reduce waiting and help explain a referral, but its effects depend on language coverage, accessibility, data quality, escalation routes, and the ability to challenge a mistaken classification. These are testable system properties. An optimization model that shortens average processing time can still worsen inequity if errors or delays concentrate among populations poorly represented in the data.
Institutional trust in an artificial service interface has multiple objects: the digital system, the health organization operating it, the professional supervising its use, and the governance procedures that determine accountability. A user's willingness to disclose symptoms to a chatbot should not be equated with a validated measure of trust in a hospital or the appropriateness of clinical decisions. Evaluation should distinguish automation performance from health outcomes, human oversight from mere nominal availability, and accessible alternatives from forced digital participation. A scientifically responsible research agenda includes subgroup calibration, prospective safety studies, independent audits, human escalation, privacy governance, and measurement of completed care pathways. These propositions define research priorities; they are not claims that contemporary artificial systems have established equivalent clinical competence or psychological status.
Algorithmic Allocation, Proxy Outcomes, and Unequal Need
Algorithmic systems can influence which patients receive additional nursing support, specialist outreach, disease management, or urgent follow-up. Such decisions depend on the outcome used to train the model. In a well-known investigation, Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S (2019) analyzed a population-health risk algorithm and identified racial bias related to its use of predicted health-care costs as a proxy for health need. At the same risk score, Black patients had more signs of illness than White patients. The study reported that changing the target from future expenditures toward measures more aligned with clinical need would substantially alter who qualified for additional assistance. This finding concerns a studied algorithm and its allocation context rather than every algorithm in medicine.
The underlying mechanism is important beyond the historical example. Expenditure is shaped by access to paid services, negotiated prices, insurance products, provider availability, and prior referral decisions. A model trained to predict costs can therefore learn patterns of institutional spending alongside patterns of illness. If groups with comparable health need have unequal historical expenditures, the target embeds an inequity already present in the service system. Even excellent prediction of cost may fail to estimate the clinical quantity that allocation policy actually aims to prioritize. The statistical problem begins with the choice of target and training data, before threshold selection or dashboard appearance is considered.
This raises a series of methodological questions for any artificial service allocator. What is the intended construct: clinical severity, avoidable hospitalization risk, probable benefit from support, or cost? Which observable variable represents it? Are training labels affected by historical differences in contact with care? Are false-negative referrals concentrated in groups with fewer previous encounters? Can clinicians override a model, document their reasons, and secure review for patients excluded by a threshold? Does a service intervention improve clinical need-adjusted access after deployment, and are those changes persistent across institutions? An audit should report subgroup outcome calibration and downstream resource allocation; conventional predictive accuracy can obscure inequity when the wrong target is optimized.
The trust question becomes specific once an artificial system is placed within an institution. Patients rely on the organization to define a valid objective, collect appropriate data, monitor errors, and provide meaningful appeal and human escalation. Explanations should identify what a decision means for the availability of care, who is accountable, and how it can be contested. Trust should be studied as an expectation concerning these real governance arrangements rather than as a generalized willingness to accept automated recommendations. The Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S (2019) result supplies a concrete empirical rationale for this research program: allocation fairness depends on the relationship between the statistical target, historical inequality, and actual clinical need.
Intersectional Heterogeneity and the Unit of Equity Analysis
Health service inequity cannot always be described adequately by comparing one category at a time. Insurance status, disability, language, age, household resources, racialized experiences, region, and employment constraints may combine to create specific access pathways. A person who needs interpretation and an accessible examination room may face a barrier that is not visible in the average outcome of either language or disability group considered independently. Bailey and colleagues (2017)30569-X) emphasize interacting institutional arrangements in the production of inequalities, while the Levesque, J.-F., Harris, M. F., & Russell, G (2013) identifies multiple capacities required at successive access stages. Together they motivate empirical examination of interactions rather than treating single-axis categories as complete explanations.
Intersectional analysis requires disciplined interpretation. In statistical models, an interaction term represents departure from a specified additive or multiplicative scale; its size and sign depend on the chosen outcome and link function. Absence of a significant interaction in a small sample cannot establish that joint disadvantage is absent. Conversely, documenting a larger disparity in a subgroup does not identify the causal institutions responsible. Researchers need prespecified comparisons, appropriate sample sizes, uncertainty intervals, meaningful denominators, and careful attention to missing data among groups least likely to appear in administrative records. Qualitative studies can identify the combinations of rules and experiences that should be tested quantitatively.
The sequence of care matters for this analysis. One subgroup may have comparable chances of scheduling but markedly lower rates of referral completion; another may rarely appear in scheduling records because of transport or documentation barriers. Reporting only an overall rate of appointments can erase these differences. Researchers should consider transition-specific conditional probabilities alongside population-level effective receipt. Because decisions at earlier stages determine who is observed later, conventional regression among treated patients can introduce selection bias. Evaluating a policy solely in the final patient sample may thus miss the very exclusion it was designed to reduce.
Equity evaluation also entails a normative question about clinical need. Different levels of use are sometimes justified when clinical conditions differ, whereas equal use can conceal unmet need in a population with higher disease burden. Analysts should disclose how need is measured and whether the measure itself depends on prior care. A diagnosis-based risk score may be lower among patients who were not evaluated often enough to receive diagnoses. The scientific standard is therefore an explicitly stated outcome, a defensible need measure, and multiple forms of evidence about the pathway linking institutional opportunities to care. These requirements become more important as service systems use artificial models to forecast demand or allocate scarce interventions.
Critical Scientific Discussion
The reviewed literature supports several robust conceptual distinctions. First, the presence of services, legal coverage, use, quality, and benefit are separate empirical outcomes. Second, trust is multidimensional and object-specific; its measurement and causal role require independent examination. Third, unequal service receipt can arise through financing, geography, organizational rules, differential recommendations, and communication, with barriers accumulating across stages. Evidence ranges from established access frameworks to strong randomized and quasi-experimental studies and observational accounts of institutional practice. These sources can be integrated without treating their designs as equally probative for each claim.
Three interpretive tensions organize the field. The first concerns selection: people who are unwell may seek more care and also encounter more obstacles, so simple associations of use with health can be misleading. The second concerns scale: a national insurance reform may raise access on average while leaving pronounced differences in provider networks, transportation, or communication; institutional effects must be disaggregated. The third concerns trust: distrust may inhibit engagement, yet it may also be an understandable response to unequal treatment. Research that targets only the patient's beliefs misses organizational mechanisms. In the opposite direction, structural reform can improve access while communication and credibility remain poor. A full explanation follows both routes and tests where they converge.
The evidence is heterogeneous in historical era, population, jurisdiction, clinical need, and outcome definition. The Baicker, K., Taubman, S. L., Allen, H. L., Bernstein, M., Gr (2013) provide strong identification of a particular coverage opportunity; Alsan, M., Garrick, O., & Graziani, G (2019) identifies a specific interpersonal configuration; national policy comparisons depend on stronger assumptions; and surveys of trust offer observational associations. It would be methodologically incorrect to aggregate these studies into a common numerical estimate of “the effect of access.” The present monograph is a critical narrative synthesis, not a preregistered systematic review or meta-analysis, and does not claim exhaustive database coverage. Its conclusions are weighted by design, construct match, plausibility, replication needs, and the narrowness or breadth of each source's demonstrated result.
Research Priorities
A priority is to link exposure to access barriers with longitudinal indicators of need, attempt, receipt, quality, and outcome. This requires collecting data on people who did not enter clinical systems, measuring referral and authorization failures, and preserving time order. Multi-stage process models can estimate where attrition occurs and whether a reform creates new bottlenecks downstream. Research should also develop comparable, accessible instruments for institutional trust that distinguish physicians, insurers, hospitals, and digital platforms, testing measurement invariance and sensitivity to actual changes in practice.
A second priority is stronger evidence about mechanisms of equitable intervention. Trials of scheduling redesign, transportation assistance, language services, and navigation should measure health-relevant outcomes and report who was unable to participate. Policy studies should publish pretrends and subgroup estimates and use multiple control strategies. Qualitative research can specify mechanisms that administrative metrics conceal, while participatory methods can improve the selection of outcomes meaningful to underserved communities. The global perspective requires harmonization that respects national institutional differences rather than erasing them.
A third priority concerns algorithmically mediated care. Researchers should audit allocation and triage systems across protected and clinically relevant groups, report false-negative access decisions, test escalation to qualified professionals, and examine whether digital substitutes reduce or increase the work patients perform to receive care. Ethical and legal oversight must accompany technical evaluation. An appropriate standard for progress is measurable improvement in access to timely, safe, clinically suitable services, with reductions in unjustified differences between populations rather than gains in average automation throughput alone.
Conclusion
Health care access is a structured social and psychological process by which medical need becomes—or fails to become—appropriate service receipt. Institutional trust influences participation when patients must rely on professionals and organizations under uncertainty. Service inequity is produced through differential opportunities, costs, organizational requirements, clinical judgments, communication, and continuity. The historical development from the inverse care law and behavioral models to patient-centered access theory established the language required to examine these mechanisms; experimental and observational research provides different forms of evidence about their operation.
The strongest synthesis is a multistage account of access linking opportunity, credible institutions, individual capacity, clinical encounter quality, and effective benefit. Insurance expansion can reduce financial barriers and increase use; equitable service requires additional attention to distribution, navigation, clinical recommendations, and accountability. Scientific progress depends on explicit constructs, need-sensitive denominators, causal identification, longitudinal measurement, and scrutiny of who remains invisible to conventional service records. In the Artificial Era, these standards extend to algorithmically mediated pathways whose effectiveness and fairness must be demonstrated through observable outcomes.
References
- Aday, L. A., & Andersen, R. (1974). A framework for the study of access to medical care. Health Services Research, 9(3), 208–220. https://pmc.ncbi.nlm.nih.gov/articles/PMC1071804/
- Alsan, M., Garrick, O., & Graziani, G. (2019). Does diversity matter for health? Experimental evidence from Oakland. American Economic Review, 109(12), 4071–4111. https://doi.org/10.1257/aer.20181446
- Andersen, R. M. (1995). Revisiting the behavioral model and access to medical care: Does it matter? Journal of Health and Social Behavior, 36(1), 1–10. https://pubmed.ncbi.nlm.nih.gov/7738325/
- Armstrong, K., Rose, A., Peters, N., Long, J. A., McMurphy, S., & Shea, J. A. (2006). Distrust of the health care system and self-reported health in the United States. Journal of General Internal Medicine, 21(4), 292–297. https://doi.org/10.1111/j.1525-1497.2006.00396.x
- Bach, P. B., Pham, H. H., Schrag, D., Tate, R. C., & Hargraves, J. L. (2004). Primary care physicians who treat Blacks and Whites. New England Journal of Medicine, 351(6), 575–584. https://doi.org/10.1056/NEJMsa040609
- Baicker, K., Taubman, S. L., Allen, H. L., Bernstein, M., Gruber, J. H., Newhouse, J. P., Schneider, E. C., Wright, B. J., Zaslavsky, A. M., & Finkelstein, A. N. (2013). The Oregon experiment—Effects of Medicaid on clinical outcomes. New England Journal of Medicine, 368(18), 1713–1722. https://doi.org/10.1056/NEJMsa1212321
- Bailey, Z. D., Krieger, N., Agénor, M., Graves, J., Linos, N., & Bassett, M. T. (2017). Structural racism and health inequities in the USA: Evidence and interventions. The Lancet, 389(10077), 1453–1463. https://doi.org/10.1016/S0140-6736(17)30569-X
- Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. (2011). Low health literacy and health outcomes: An updated systematic review. Annals of Internal Medicine, 155(2), 97–107. https://doi.org/10.7326/0003-4819-155-2-201107190-00005
- Boulware, L. E., Cooper, L. A., Ratner, L. E., LaVeist, T. A., & Powe, N. R. (2003). Race and trust in the health care system. Public Health Reports, 118(4), 358–365. https://doi.org/10.1093/phr/118.4.358
- Eberly, L. A., Kallan, M. J., Julien, H. M., Haynes, N., Khatana, S. A. M., Nathan, A. S., Snider, C., Chokshi, N. P., Eneanya, N. D., Takvorian, S. U., Anastos-Wallen, R., Chaiyachati, K., Ambrose, M., O’Quinn, R., Seigerman, M., Goldberg, L. R., Leri, D., Choi, K., Gitelman, Y.,. Adusumalli, S. (2020). Patient characteristics associated with telemedicine access for primary and specialty ambulatory care during the COVID-19 pandemic. JAMA Network Open, 3(12), e2031640. https://doi.org/10.1001/jamanetworkopen.2020.31640
- Federal Reserve Board. (2026). Report on the economic well-being of U.S. households in 2025: Economic hardships. https://www.federalreserve.gov/publications/2026-economic-well-being-of-us-households-in-2025-economic-hardships.htm
- Finkelstein, A., Taubman, S., Wright, B., Bernstein, M., Gruber, J., Newhouse, J. P., Allen, H., & Baicker, K. (2012). The Oregon Health Insurance Experiment: Evidence from the first year. Quarterly Journal of Economics, 127(3), 1057–1106. https://doi.org/10.1093/qje/qjs020
- Flores, G. (2005). The impact of medical interpreter services on the quality of health care: A systematic review. Medical Care Research and Review, 62(3), 255–299. https://doi.org/10.1177/1077558705275416
- Hall, M. A., Dugan, E., Zheng, B., & Mishra, A. K. (2001). Trust in physicians and medical institutions: What is it, can it be measured, and does it matter? Milbank Quarterly, 79(4), 613–639. https://doi.org/10.1111/1468-0009.00223
- Hall, W. J., Chapman, M. V., Lee, K. M., Merino, Y. M., Thomas, T. W., Payne, B. K., Eng, E., Day, S. H., & Coyne-Beasley, T. (2015). Implicit racial/ethnic bias among health care professionals and its influence on health care outcomes: A systematic review. American Journal of Public Health, 105(12), e60–e76. https://doi.org/10.2105/AJPH.2015.302903
- Hart, J. T. (1971). The inverse care law. The Lancet, 1(7696), 405–412. https://doi.org/10.1016/S0140-6736(71)92410-X
- Institute of Medicine. (2003). Unequal treatment: Confronting racial and ethnic disparities in health care. National Academies Press. https://doi.org/10.17226/10260
- Kullgren, J. T., McLaughlin, C. G., Mitra, N., & Armstrong, K. (2012). Nonfinancial barriers and access to care for U.S. adults. Health Services Research, 47(1 Pt 2), 462–485. https://doi.org/10.1111/j.1475-6773.2011.01308.x
- Levesque, J.-F., Harris, M. F., & Russell, G. (2013). Patient-centred access to health care: Conceptualising access at the interface of health systems and populations. International Journal for Equity in Health, 12, Article 18. https://doi.org/10.1186/1475-9276-12-18
- 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. https://doi.org/10.1126/science.aax2342
- Penchansky, R., & Thomas, J. W. (1981). The concept of access: Definition and relationship to consumer satisfaction. Medical Care, 19(2), 127–140. https://doi.org/10.1097/00005650-198102000-00001
- Schulman, K. A., Berlin, J. A., Harless, W., Kerner, J. F., Sistrunk, S., Gersh, B. J., Dubé, R., Taleghani, C. K., Burke, J. E., Williams, S., Eisenberg, J. M., & Escarce, J. J. (1999). The effect of race and sex on physicians' recommendations for cardiac catheterization. New England Journal of Medicine, 340(8), 618–626. https://doi.org/10.1056/NEJM199902253400806
- Shea, J. A., Micco, E., Dean, L. T., McMurphy, S., Schwartz, J. S., & Armstrong, K. (2008). Development of a revised Health Care System Distrust scale. Journal of General Internal Medicine, 23(6), 727–732. https://doi.org/10.1007/s11606-008-0575-3
- Sommers, B. D., Baicker, K., & Epstein, A. M. (2012). Mortality and access to care among adults after state Medicaid expansions. New England Journal of Medicine, 367(11), 1025–1034. https://doi.org/10.1056/NEJMsa1202099
- Sommers, B. D., Blendon, R. J., & Orav, E. J. (2016). Both the 'private option' and traditional Medicaid expansions improved access to care for low-income adults. Health Affairs, 35(1), 96–105. https://doi.org/10.1377/hlthaff.2015.0917
- Sommers, B. D., Gawande, A. A., & Baicker, K. (2017). Health insurance coverage and health—What the recent evidence tells us. New England Journal of Medicine, 377(6), 586–593. https://doi.org/10.1056/NEJMsb1706645
- Starfield, B., Shi, L., & Macinko, J. (2005). Contribution of primary care to health systems and health. Milbank Quarterly, 83(3), 457–502. https://doi.org/10.1111/j.1468-0009.2005.00409.x
- Syed, S. T., Gerber, B. S., & Sharp, L. K. (2013). Traveling towards disease: Transportation barriers to health care access. Journal of Community Health, 38(5), 976–993. https://doi.org/10.1007/s10900-013-9681-1
- Takeshita, J., Wang, S., Loren, A. W., Mitra, N., Shults, J., Shin, D. B., & Sawinski, D. L. (2020). Association of racial/ethnic and gender concordance between patients and physicians with patient experience ratings. JAMA Network Open, 3(11), e2024583. https://doi.org/10.1001/jamanetworkopen.2020.24583
- van Ryn, M., & Burke, J. (2000). The effect of patient race and socio-economic status on physicians' perceptions of patients. Social Science & Medicine, 50(6), 813–828. https://doi.org/10.1016/S0277-9536(99)00338-X
- World Health Organization & World Bank. (2025). Tracking universal health coverage: 2025 global monitoring report. https://www.who.int/teams/health-financing-and-economics/global-monitoring-report/2025
- World Health Organization. (2025). World report on social determinants of health equity. https://www.who.int/publications/i/item/9789240107588