RESEARCH DOSSIER
Health Literacy, Risk Communication, and Informed Understanding
Abstract. Health literacy, risk communication, and informed understanding form an integrated research problem at the intersection of psychological information processing, medical decision science, public health, institutional accessibility, and digital communication. This fundamental critical narrative monograph reconstructs the development of the field from functional literacy tests and framing experiments to multidimensional assessment, organizational health literacy, patient decision aids, and artificial information intermediaries. It distinguishes access, comprehension, risk perception, calibration, values-congruent choice, behavior, and clinical outcome; evaluates numerical, linguistic, visual, dialogic, and digital communication mechanisms; and critically compares observational, experimental, psychometric, intervention, and systematic-review evidence. Primary studies and major syntheses establish effects of framing and frequency formats, measurement limits of narrow literacy tests, and substantial improvements in patient knowledge and risk-perception accuracy from decision aids. Structural inequality, cross-cultural validity, misinformation, uncertainty communication, and emerging artificial systems introduce unresolved causal and governance questions. The study offers a methodological synthesis of existing scholarship and no new participant experiment, registered systematic review, quantitative pooled estimate, or individualized clinical recommendation.
Keywords: health literacy; health communication; risk communication; informed understanding; health numeracy; natural frequencies; decision framing; absolute risk; probability comprehension; risk perception; shared decision making; patient decision aids; teach-back; organizational health literacy; critical health literacy; eHealth literacy; digital health literacy; misinformation; uncertainty communication; health equity; psychometrics; informed consent; Artificial Era
Scientific Overview and the Research Object
Health literacy is an organized capacity to locate, interpret, appraise, communicate about, and use information and services for health-related judgment and action. Risk communication concerns the production, representation, exchange, and interpretation of information about possible outcomes, their probability, severity, time horizon, and uncertainty. Informed understanding is the demonstrable grasp of the relevant options and consequences sufficient for a person to deliberate in relation to their own purposes. These three objects are linked by a common problem: an empirical probability or a clinical explanation acquires practical meaning only as it enters a human cognitive and institutional setting. The Sørensen, K., Van den Broucke, S., Fullam, J., Doyle, G., Pelikan, J., Slonska, Z., & Brand, H. integrated model and the World Health Organization. glossary locate this setting in the interaction of knowledge, appraisal, and opportunity.
The study of understanding has to differentiate four endpoints. Information exposure is the encounter with a message; comprehension is an accurate representation of its propositions; informed judgment requires comparing meaningful alternatives and uncertainties; and a health outcome follows through a much longer causal chain involving access, action, treatment, social conditions, and biology. The distinctions explain why a readable brochure can improve recall without establishing long-term health gain and why a well-informed person can reasonably decline a clinical option. Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. document associations between limited literacy and health outcomes while leaving the pathways and confounding to be investigated.
This monograph evaluates research on the construction of messages, cognitive processing of probabilities, instruments that measure capabilities, experimental risk formats, institutional responsibility, interpersonal verification, misinformation, and digital mediation. Its method is a critical narrative synthesis of identifiable primary studies, conceptual foundations, systematic reviews, and official sources. No new participant experiment, preregistered literature search, or pooled meta-analysis is claimed. Clinical examples are illustrative research cases rather than individual medical recommendations.
Research Questions, Construct Boundaries, and Levels of Analysis
The primary questions are how people derive meaning from health information; why numerically equivalent claims generate different interpretations; which instruments discriminate reading skill from broader navigation and appraisal; how communicative interventions affect measured comprehension; and which social environments enable informed decisions. An adequate explanation must place message properties, cognitive resources, emotional significance, opportunities for clarification, and institutional accessibility within one research design.
Literacy can refer to functional reading and numeracy, interactive communication, critical evaluation, or broader health-system navigation. Risk perception denotes a subjective representation of possible harm; it can be affected by personal experience independently of accurate probability knowledge. Decision quality refers to whether choices are informed and aligned with the chooser's values, and differs from outcome quality because favorable decisions sometimes precede adverse events. Informed consent is an ethical and legal process, whose particulars depend on context, and cannot be equated with a correct score on a recall quiz.
Units of analysis range from an individual's response to a graph, through a consultation dyad, to a health-care organization and a population's information environment. Berkman, N. D., Davis, T. C., & McCormack, L. show why historical definitions developed from functional literacy into a wider social construct. The present dossier concentrates on validity of information uptake and calibrated risk understanding; negotiation of care responsibilities and the general clinician relationship are separate research problems.
Historical Formation of Health Literacy Research
The clinical literature of the twentieth century connected adult literacy to patients' ability to navigate written instructions, appointments, and consent forms. In 1995, Parker, R. M., Baker, D. W., Williams, M. V., & Nurss, J. R. introduced the Test of Functional Health Literacy in Adults using health-related reading and numerical tasks. During the same year, Williams, M. V., Parker, R. M., Baker, D. W., Parikh, N. S., Pitkin, K., Coates, W. C., & Nurss, J. R. examined limited functional literacy among patients at two public hospitals. Both belong to the early movement that treated comprehension problems as measurable features of care delivery.
The Weiss, B. D., Mays, M. Z., Martz, W., Castro, K. M., DeWalt, D. A., Pignone, M. P., Mockbee, J., & Hale, F. A. Newest Vital Sign subsequently used a nutrition label and a six-item task to provide a brief assessment. These instruments helped make the phenomenon empirically visible, yet they select different portions of the construct: pronunciation or reading comprehension is not a direct measure of a person's capacity to evaluate competing scientific claims or navigate an institution. Historical development therefore includes diversification of measurement rather than a single improvement in scale length.
A related tradition in judgment research established that identical numbers can be interpreted differently when displayed in alternative forms. Tversky, A., & Kahneman, D. analyzed decision framing, and McNeil, B. J., Pauker, S. G., Sox, H. C., Jr., & Tversky, A. investigated treatment preferences when outcome descriptions changed. This research preceded many later health-literacy frameworks and supplies a crucial empirical insight: comprehension depends partly on the representation, even where the recipient's underlying abilities are unchanged.
Health Promotion and the Expansion of the Concept
In Nutbeam, D. public-health formulation, health literacy included functional, interactive, and critical capacities connected to education and health promotion. The shift extended the domain from performance on clinical paperwork to participation in social and political conditions affecting health. The three levels remain analytically useful: someone may read a prescription correctly while finding it difficult to ask questions, assess competing claims, or engage in a public health discussion.
The Sørensen, K., Van den Broucke, S., Fullam, J., Doyle, G., Pelikan, J., Slonska, Z., & Brand, H. synthesis organized access, understanding, appraisal, and application across health care, disease prevention, and health promotion. Its conceptual model includes both personal competencies and contextual determinants. Nutbeam, D. retrospective reinforces the continued relevance of a public-health interpretation after twenty-five years. The evolution is cumulative: decoding remains necessary in many settings, but contemporary evidence demands broader domains.
Expanding a definition changes what a study must measure. A scale of reading accuracy may support claims about reading and cannot on its own establish critical evaluation, collective agency, or empowerment. Cross-national survey responses may show perceived difficulty while also reflecting institutional expectations. A sound monograph therefore treats definitions, operational indicators, and explanatory hypotheses as separate analytical components.
Personal and Organizational Health Literacy
The updated Centers for Disease Control and Prevention. definitions distinguish personal health literacy, the ability to find, understand, and use information and services, from organizational health literacy, the extent to which organizations equitably enable such activity. This approach places the burden of comprehensibility partly on service design: clear navigation, accessible language, interpreters, consistent forms, usable digital interfaces, and opportunities to ask questions are properties of a system rather than attributes of a reader.
Brach, C., Keller, D., Hernandez, L. M., Baur, C., Parker, R., Dreyer, B., Schyve, P., Lemerise, A. J., & Schillinger, D. proposed attributes of health-literate health-care organizations, connecting leadership, integration, workforce preparation, communication, and accessibility. The framework is a practical institutional model, not experimental proof that every suggested attribute independently produces a clinical improvement. The World Health Organization. definition similarly emphasizes that competencies are mediated by organizations and available resources.
An interactional account suggests that the relevant predictor is often a mismatch between demands and available supports. Identical people may perform differently in an urgent department, a familiar outpatient service, or a multilingual online portal. Research should therefore assess both persons and the specific system encountered, rather than inferring a stable deficiency from one difficult transaction.
Functional, Interactive, and Critical Competence
Functional competence includes interpreting instructions, schedules, dosage labels, appointment forms, and elementary numeric claims. Interactive competence concerns asking for clarification, explaining circumstances, recognizing a choice, and maintaining a shared informational record. Critical competence includes questioning source quality, evaluating the evidentiary status of claims, detecting inappropriate comparisons, and recognizing conflicts of interest. These functions can be mutually supportive, yet each requires different observational indicators.
The classification introduced by Nutbeam, D. invites empirical questions about transfer: whether better reading skill improves evaluation of a risk ratio, whether dialog training changes questions actually asked, and whether source-appraisal instruction protects against misleading health information. A successful intervention in one domain should not be reported as a generalized improvement without testing transfer in the others.
Conceptual overlap with scientific literacy, media literacy, information literacy, and general numeracy is substantial. Researchers can refine construct boundaries by estimating discriminant validity, demonstrating prediction beyond general education, and testing whether an intervention changes a target capacity more than a neighboring capacity. The practical implication is a multidimensional assessment matched to the decision rather than an unqualified global score.
The Cognitive Architecture of Understanding
Health messages pass through attention, language processing, memory encoding, numerical interpretation, inferential integration, and retrieval during a decision. These stages are affected by distraction, fatigue, anxiety, familiarity, prior beliefs, and presentation. A person can accurately remember a stated percentage but fail to connect it to a specified population and time period. Conversely, a recipient may retain the qualitative meaning of a comparison after forgetting its exact numerical values.
Judgment research by Peters, E., Västfjäll, D., Slovic, P., Mertz, C. K., Mazzocco, K., & Dickert, S. indicates that numeracy shapes not only calculation but also the affective meaning drawn from numbers. Reyna, V. F., Nelson, W. L., Han, P. K., & Dieckmann, N. F. review how numerical ability, gist representations, and decision processes jointly influence medical risk comprehension. These findings support models in which comprehension is structured by representation and prior knowledge, without treating human cognition as a uniform calculation engine.
Working-memory demands are especially relevant when a clinician communicates several probabilities, different treatment horizons, and multiple outcomes in one passage. The testable intervention is to externalize comparisons, keep denominators consistent, identify the baseline, and provide a durable written or visual aid. A study can measure immediate recall, delayed understanding, calibration, and decision concordance separately to identify which processing stage changes.
Language, Readability, and Semantic Precision
Medical terminology carries distinctions between diagnosis, suspicion, prognosis, likelihood, benefit, harm, and evidence strength. Difficulty arises when a term has a technical meaning in a clinician's discipline and another meaning in everyday speech. The expression “positive test,” for example, denotes a laboratory result; its value for a patient depends on sensitivity, specificity, prevalence, and subsequent confirmation. Accurate communication requires defining the referent and making relevant conditions explicit.
Plain language is a design principle for making the essential claim recoverable without unnecessary decoding. It does not license replacing scientifically necessary distinctions with oversimplified certainty. The guidance synthesized by Fagerlin, A., Zikmund-Fisher, B. J., & Ubel, P. A. integrates clear wording with explicit numerical comparators. A message is well designed when readers can identify the outcome, comparison group, time horizon, and magnitude of change.
Readability formulas can estimate structural difficulty in sentences and word lengths; they do not assess factual accuracy, inference demands, visual hierarchy, emotional impact, cultural relevance, or whether a decision is correctly understood. Valid usability work combines readability checks with interviews, comprehension tasks, and observation of actual navigation. Testing with users of different educational and linguistic backgrounds is required for meaningful generalization.
Health Numeracy and Quantitative Representation
Health numeracy involves interpreting proportions, percentages, ratios, rates, and probability statements in context. At its simplest it includes comparing two frequencies; at higher levels it entails reasoning about conditional probabilities, statistical uncertainty, temporal risk, and uncertainty in effect estimates. Reyna, V. F., Nelson, W. L., Han, P. K., & Dieckmann, N. F. describe how numeric comprehension interacts with cognition and medical decision processes.
A population rate and an individual's probability are related by a model with assumptions about the population and the target person. A 2% event rate over five years does not communicate the chance of that event next week, nor does it necessarily apply to someone with substantially different characteristics. Units, denominators, baseline hazards, and time horizons are constituents of meaning. Removing them may create apparent simplicity while increasing ambiguity.
Numeric confidence and numeric accuracy also diverge. Surveys may assess self-reported ease while objective tasks assess actual operations, and both can be affected by mathematical anxiety and context. Peters, E., Västfjäll, D., Slovic, P., Mertz, C. K., Mazzocco, K., & Dickert, S. provide experimental evidence that numeracy changes how numbers influence preference, but do not establish a universally beneficial cutoff applicable to every clinical decision.
Natural Frequencies and Bayesian Inference
A conditional probability such as the chance of a disease given a positive test requires joint consideration of prevalence and the test's error rates. Gigerenzer, G., & Hoffrage, U. demonstrated that natural-frequency representations can simplify Bayesian reasoning for many inferential tasks. A frequency tree makes the reference classes explicit and allows the target group to be counted rather than inferred from nested percentages.
For illustration only, consider a hypothetical population of 1,000 individuals, of whom 10 have a condition. Suppose a test detects 9 of those 10 and returns positive for 99 of the remaining 990. Then 108 results are positive and 9 are true positives: the fraction among positive tests is 9/108, approximately 8.3%. The example is a mathematical illustration with invented inputs and is not a diagnostic estimate for a real disease or test. Its purpose is to show the distinct roles of prevalence, sensitivity, and false-positive results.
The psychological advantage of frequency displays depends on the task and exact representation. Frequencies can be misleading when denominators shift, base populations are unspecified, or a simple rate is dressed up as conditional information. Risk communication experiments should therefore identify the inferential operation, test comprehension using equivalent formats, and preserve semantic consistency across compared groups.
Absolute Risk, Relative Risk, and Denominators
An intervention changing an illustrative event probability from 4 in 100 to 3 in 100 has an absolute reduction of 1 percentage point and a relative reduction of 25%. Both statements can be arithmetically true, but they convey different impressions if presented separately. Fagerlin, A., Zikmund-Fisher, B. J., & Ubel, P. A. recommend clear communication of absolute changes and baselines because the recipient needs to know the starting level before evaluating the practical size of a reduction.
The number needed to treat is the inverse of an absolute risk reduction under an appropriate fixed horizon, with assumptions that must be made explicit. It can aid comparisons but is sensitive to baseline risk and follow-up and may be misunderstood when its endpoint is unspecified. Relative effects may be useful in scientific analysis; a public-facing message can pair the relative statistic with expected event counts to preserve interpretability.
Denominator neglect also arises when a message uses “one in a thousand” for a small harm but “one in a hundred” for a larger benefit. Consistent units reduce avoidable cognitive work; the appropriate denominator depends on the magnitude and purpose, especially for rare outcomes. Trevena, L. J., Zikmund-Fisher, B. J., Edwards, A., Gaissmaier, W., Galesic, M., Han, P. K. J., King, J., Lawson, M. L., Linder, S. K., Lipkus, I., Ozanne, E., Peters, E., Timmermans, D., & Woloshin, S. provide a research-based primer on these design choices.
Baseline Risk, Time Horizons, and Competing Outcomes
A probability statement is incomplete without specifying the event, eligible population, observation interval, and relevant counterfactual. The same relative treatment effect can imply different absolute benefits in populations with different starting risks. Long-term cumulative risk cannot be compared directly to short-term adverse-event probabilities unless a time-adjusted framework is explained. Apparent disagreement between reports may arise from distinct denominators or periods rather than from conflicting evidence.
Health decisions often involve competing endpoints: symptom relief, serious complications, inconvenience, loss of function, and other outcomes valued differently by different people. These endpoints cannot always be reduced to a single objective scale. Informed understanding requires that the outcomes considered relevant are named and the evidence behind their estimates explained. The Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. review demonstrates the importance of providing structured information in patient decision aids.
Researchers evaluating communication can use scenario-based tasks in which the same effect is described with multiple follow-up intervals. Correct responses should involve matching time horizons, distinguishing cumulative and instantaneous rates, and identifying populations rather than simply choosing the numerically largest figure. Such tasks separate genuine understanding from acquiescence to persuasive wording.
Framing Effects and Preference Formation
Framing occurs when logically related descriptions emphasize different aspects of the same outcomes. Tversky, A., & Kahneman, D. demonstrated shifts in preference across alternative formulations of decision problems. McNeil, B. J., Pauker, S. G., Sox, H. C., Jr., & Tversky, A. investigated how information formats affected choices between treatment alternatives among patients, graduate students, and physicians. The historical evidence made framing a central problem in medical judgment research.
Equivalent survival and mortality statements can elicit different emotional responses because attention is directed toward different features of an outcome. A valid message should consider both gains and losses where appropriate, while avoiding a presentation that strategically favors one choice. Researchers can randomize presentation order and test whether respondents understand numerical equivalence after exposure.
The existence of a framing effect does not prove that people cannot make reasoned choices. It demonstrates that formats participate in the construction of decision meaning. The appropriate response is evidence-based design, transparent comparison, and assessment of whether framing changes comprehension, informed preference, or only immediate affective reaction.
Visual Aids, Icon Arrays, and Information Design
Icon arrays represent groups of people with shared denominators and make the numerator visually countable. Bar graphs and other displays can emphasize magnitude, comparison, or trajectory. In a cross-national investigation in the United States and Germany, Garcia-Retamero, R., & Galesic, M. studied how visual aids affected understanding of treatment-risk changes. Their findings support thoughtful visual formats, particularly for people with difficulty interpreting numbers, while the published correction and design-specific context caution against treating every graphic as equivalent.
A pictograph showing events among 100 hypothetical people can foreground absolute difference and baseline. But graphic choices such as color saturation, icon arrangement, nonuniform axes, omitted denominators, and animated movement may distort perceived seriousness. Usability evaluation should inspect not only correct calculations but also false impressions of frequency and ease of comparing alternatives.
Visual formats are especially important when accessibility requirements make ordinary graphics insufficient. A visually accessible presentation should also provide a text description, readable numbers, and a semantic explanation for screen readers. Trevena, L. J., Zikmund-Fisher, B. J., Edwards, A., Gaissmaier, W., Galesic, M., Han, P. K. J., King, J., Lawson, M. L., Linder, S. K., Lipkus, I., Ozanne, E., Peters, E., Timmermans, D., & Woloshin, S. evaluate quantitative presentation options and the tradeoffs relevant to decision-aid development.
Verbal Probability Terms and Numerical Anchoring
Words such as “rare,” “likely,” and “possible” often have different numerical interpretations across speakers and contexts. Hashim, M. J. reviewed studies mapping verbal expressions to numerical interpretations, but its heterogeneous evidence and proposed ranges should not be taken as universally applicable dictionary equivalents. In a high-stakes communication, a numerical estimate, reference population, and time period can clarify what the speaker means.
Numbers also imply precision. A point estimate derived from a small or heterogeneous evidence base may encourage unwarranted certainty if confidence intervals or plausible ranges are concealed. The scientific purpose of numerical anchoring is to make meaning explicit while exposing uncertainty honestly. Vocabulary such as “very likely” should be tested in relevant languages and populations.
A strong experiment would compare a verbal-only message with numerical-only and combined presentations, recording both the perceived risk and the respondent's stated interpretation. If the number improves accuracy but reduces recall or increases anxiety, the evidence calls for a multidimensional judgment rather than a binary declaration that one format is superior in all circumstances.
Diagnostic Tests, Screening, and Conditional Risk
Screening combines a decision to test, a distribution of underlying condition status, and the performance of an assay within a population. The probability of an abnormal result differs from the probability that an abnormal result indicates a condition. Comprehension depends on baseline prevalence and both false-positive and false-negative rates. The frequency-format findings of Gigerenzer, G., & Hoffrage, U. provide a general reasoning framework applicable to these inferential structures.
Additional complexity arises when follow-up testing changes the relevant probabilities, when the test population was selected because of symptoms, or when verification is incomplete. Research should distinguish statistical calculation from the clinical implications of subsequent assessment. A communication aid cannot guarantee informed screening if it suppresses overdiagnosis, downstream procedures, uncertainty, or alternatives.
The illustrative task for a participant is to identify which denominator answers a specific question: all screened people, everyone receiving a positive test, or everyone who actually has a condition. Correct denominator selection is a more discriminating indicator of understanding than a general claim of familiarity with test accuracy.
Uncertainty, Evidence Quality, and Epistemic Calibration
Uncertainty in health care has multiple origins. Han, P. K. J., Klein, W. M. P., & Arora, N. K. distinguish sources and manifestations of uncertainty across scientific knowledge, probabilistic outcomes, and interpersonal or organizational ambiguity. Random variation remains even when a population model is well estimated. Scientific uncertainty concerns incomplete or conflicting evidence; ambiguity can concern the meaning of a risk category or the implications of available options.
Politi, M. C., Han, P. K. J., & Col, N. F. examine communicating uncertainty about harms and benefits. A well-calibrated message differentiates estimated probability from certainty about that estimate, describes evidence limitations proportionately, and explains which future observations might revise the conclusion. Presenting uncertainty is compatible with clear advice when that advice is grounded in an appropriate professional context.
Calibration can be measured by comparing stated confidence with response accuracy and asking respondents to distinguish what the evidence supports from what remains unknown. The communication objective is a justified correspondence between belief strength and available support. Neither maximally confident speech nor endless qualification substitutes for accurate calibration.
Emotion, Perceived Threat, and Trust
Risk information arrives in a motivational environment. Fear, anticipated regret, prior illness experience, perceived vulnerability, and interpersonal trust can determine which parts of a message receive attention. A person may accurately state that a complication is uncommon while assigning it high personal importance; this can be coherent when severity, past experience, and values are considered. Risk perception is therefore not reducible to arithmetic error.
Peters, E., Västfjäll, D., Slovic, P., Mertz, C. K., Mazzocco, K., & Dickert, S. show that numeric interpretation and affect are intertwined. A communication study must distinguish the factual judgment “how often does this happen?” from the evaluative judgment “how troubling would that be to me?” and from the behavioral intention “what will I do?” Conflating these produces weak measures of decision quality.
Trust is especially consequential when the audience cannot independently reconstruct technical evidence. Institutional transparency, acknowledgment of uncertainty, and a recognizable pathway for questions can support warranted trust. However, popularity or confident tone is insufficient evidence of reliability. The relevant psychological construct is trust calibration: the degree to which reliance follows demonstrated competence and accountable processes.
Misinformation, Correction, and Persistent Beliefs
Health misinformation can enter comprehension through repetition, apparent social consensus, misleading narratives, and partial truths. The synthesis by Lewandowsky, S., Ecker, U. K. H., Seifert, C. M., Schwarz, N., & Cook, J. examined how corrected false claims may continue to influence subsequent inference. A person can repeat a correction while retaining a causal story previously built around the false information. Evaluation must therefore measure inferred beliefs and decisions, not simply whether a corrective fact was noticed.
Vraga, E. K., & Bode, L. discuss correction strategies within social media environments. Corrections work within platforms where sources, recommendation systems, and peer reactions shape credibility. The causal effects of a specific correction vary by audience and context; content that appears successful in a brief experiment may have less durable impact under repeated exposure.
Health literacy research gains specificity when it distinguishes epistemic skills, source recognition, factual recall, susceptibility to manipulative framing, and willingness to update beliefs. Source-evaluation interventions should test genuine and fabricated material across multiple topics, as memorizing a single approved answer does not establish a transferable capacity to evaluate evidence.
Artificial Systems as Health Information Intermediaries
Generative artificial systems can translate terminology, summarize documents, adapt reading levels, and respond interactively to questions. These capacities create a research object at the interface of health literacy, linguistic interaction, human trust, and automated knowledge production. The central issue is whether mediated communication preserves the evidence, population, uncertainty, and options of the originating source while remaining understandable.
Abeo, A. N. A., Armstrong, S., Scriney, M., & Goss, H. reviewed AI techniques in health-literacy applications and emphasized limitations in direct health-literacy outcome measurement. The 2026 synthesis by Alshanqiti, A., Alghabban, H., Surrati, A. M. Q., Ibrahim, S., & Owaydhah, W. found limited and heterogeneous evidence across digital communication modalities, with AI-enabled interventions sparsely represented and knowledge effects not automatically transferable to clinical outcomes. Its published correction is part of the provenance and must accompany any close reading of numerical syntheses.
The World Health Organization. governance guidance identifies risks from inaccurate or biased large multimodal model output and automation bias. For an Artificial Era research agenda, the appropriate tests include factual fidelity, risk-format equivalence, source traceability, calibration, differential performance across populations, and human capacity to challenge the system. These are empirical evaluation requirements rather than claims of established clinical benefit.
Informed Understanding, Consent, and Autonomy
Informed understanding means that a person can identify the relevant decision, available options, expected benefits and harms, material uncertainties, and the implications for their values. A document can contain all legally required words while leaving a participant unable to explain the actual choice. Conversely, knowledge demonstrated in an interactive conversation may be richer than performance on a written checklist. The ethical construct of autonomy requires meaningful possibility to deliberate and decide.
The Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. Cochrane review provides strong evidence that patient decision aids improve several measures of knowledge and accuracy of risk perception. These endpoints provide a firmer basis for evaluating informational quality than assuming that a particular choice is objectively correct. In many preference-sensitive situations, people can understand the same evidence and legitimately choose differently.
Consent comprehension should be assessed by asking what will happen, why, which risks matter, what alternatives exist, and whether participation or treatment is optional in the given context. Assessment itself must protect privacy and avoid coercion. The present psychological analysis does not claim uniform legal criteria for every jurisdiction or clinical setting.
Teach-Back and Verification of Comprehension
Teach-back invites a recipient to explain information or a planned action in their own words, enabling a communicator to detect a mismatch between intended and received meaning. Schillinger, D., Piette, J., Grumbach, K., Wang, F., Wilson, C., Daher, C., Leong-Grotz, K., Castro, C., & Bindman, A. B. investigated the closing-the-loop practice in communication with patients with diabetes and limited health literacy. The approach operationalizes understanding as demonstrated reconstruction rather than a nod of apparent agreement.
Dinh, T. T. H., Bonner, A., Clark, R., Ramsbotham, J., & Hines, S. systematically reviewed teach-back in chronic disease education, examining adherence and self-management outcomes. Heterogeneity in intervention delivery, personnel, follow-up, and measured endpoints limits any universal causal claim. The method's direct purpose is to make misunderstanding observable, while longer-term health effects require additional evidence.
A rigorous teach-back protocol records which concept was tested, how the prompt was phrased, whether the respondent could answer without leading cues, what clarification followed, and whether understanding endured. The feedback relationship should avoid making a patient responsible for an unclear message. It is the communicator who uses the response to improve the explanation.
Measurement Instruments and Their Construct Validity
Parker, R. M., Baker, D. W., Williams, M. V., & Nurss, J. R. developed TOFHLA to assess functional reading and numerical skills using healthcare material. The shorter functional tests reduce administration burden but also limit the observed domain. Weiss, B. D., Mays, M. Z., Martz, W., Castro, K. M., DeWalt, D. A., Pignone, M. P., Mockbee, J., & Hale, F. A. introduced the Newest Vital Sign, a nutrition-label task, which assesses brief functional and numeric performance. The instruments serve different practical purposes and are not interchangeable measures of all health-literacy competencies.
Osborne, R. H., Batterham, R. W., Elsworth, G. R., Hawkins, M., & Buchbinder, R. developed and initially validated the Health Literacy Questionnaire, drawing on experiences of patients, practitioners, and communities to represent multiple dimensions. Its structure responds to the limitation of narrow decoding tests; nonetheless, self-reported experience and objective skill performance capture different phenomena. Scale selection must begin with the construct and decision context.
Evidence of validation involves more than internal consistency. It requires content relevance, structural fit, responsiveness where intended, relations to relevant external variables, appropriate translations, and evidence that score interpretation is suitable for the population. A measurement instrument cannot receive unlimited validity from one developmental study.
Psychometrics, Cross-Cultural Invariance, and Response Processes
Psychometric assessment must examine whether the same item describes the same underlying competence for people with different languages, educational histories, ages, impairments, and clinical experiences. Differential item functioning can arise when a health system uses unfamiliar conventions or a question assumes access to services not available locally. Testing factor structure alone cannot establish conceptual equivalence.
The European survey analyzed by Sørensen, K., Pelikan, J. M., Röthlin, F., Ganahl, K., Slonska, Z., Doyle, G., Fullam, J., Kondilis, B., Agrafiotis, D., Uiters, E., Falcon, M., Mensing, M., Tchamov, K., van den Broucke, S., & Brand, H. demonstrates the possibilities of comparing reported health-literacy challenges across countries. Such comparisons require scrutiny of sample recruitment, translation, response categories, local health-system demands, and the extent to which a self-report scale can support population rankings. Cross-national differences in ease of navigating care may partly measure differences in institutional complexity.
Cognitive interviews, think-aloud protocols, item-response analysis, test-retest reliability, and multi-group measurement models provide complementary evidence. Longitudinal programs must also test whether instrument meaning remains stable as digital services and clinical workflows change. A score becomes scientifically useful through explicit interpretive limits.
Experimental Paradigms of Risk Communication
Randomized communication experiments can manipulate the numerator-denominator format, absolute versus relative risk, framing, display type, visual density, linguistic precision, or source cue while holding the underlying quantities constant. Well-controlled manipulation supports conclusions about the causal effect of the message format on a measured response under the study conditions. It does not automatically establish durable clinical behavior change.
The historical experiments of Tversky, A., & Kahneman, D., McNeil, B. J., Pauker, S. G., Sox, H. C., Jr., & Tversky, A., and Gigerenzer, G., & Hoffrage, U. demonstrate how different representations affect decision tasks. More clinically specific studies such as Garcia-Retamero, R., & Galesic, M. test visualization across heterogeneous audiences. A methodological synthesis must examine the exact question format, incentives, literacy distribution, baseline probabilities, and ecological validity of each paradigm.
Outcomes should be distinguished: exact quantitative comprehension, gist accuracy, preference stability, decision confidence, decisional conflict, retention, and observable choice. A display could increase correct answers on a math task while altering risk perception in an unintended direction. Preregistered contrasts, adequately powered samples, and reporting of null effects strengthen interpretation.
Survey, Cohort, and Population Research
Cross-sectional surveys estimate the distribution of health-literacy responses and their association with education, socioeconomic resources, chronic illness, institutional access, or self-reported health. Sørensen, K., Pelikan, J. M., Röthlin, F., Ganahl, K., Slonska, Z., Doyle, G., Fullam, J., Kondilis, B., Agrafiotis, D., Uiters, E., Falcon, M., Mensing, M., Tchamov, K., van den Broucke, S., & Brand, H. supplies one comparative European example; Okan, O., Bollweg, T. M., Berens, E.-M., Hurrelmann, K., Bauer, U., & Schaeffer, D. provides a context-specific pandemic-period example. These designs identify patterns for investigation but do not settle whether literacy influences health or whether health-system experience influences literacy scores.
Longitudinal cohorts improve temporal ordering by observing baseline competencies and later behaviors or outcomes. However, both literacy and health are influenced by prior education, disability, socioeconomic position, access to care, and disease burden. Appropriate analysis requires temporal confounder control and transparent handling of attrition and differential measurement. Mediation through treatment adherence should be tested rather than assumed.
Population prevalence estimates are especially vulnerable to definitional change. A figure derived from a particular screening threshold, nation, instrument, or year cannot be represented as a universal rate of low literacy. For this reason the present synthesis focuses on verified study designs and conceptual comparisons rather than importing apparently precise, nontransferable global percentages.
Health Literacy and Health Outcomes
The updated systematic review by Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. linked lower literacy with several adverse health-related outcomes in the reviewed literature. The evidentiary pathway remains multidimensional: reading difficulties can affect appointment navigation, comprehension of instructions, adherence, and service access, while health status, resource deprivation, and organizational complexity can also influence performance on literacy measures.
An observational association between literacy score and hospital use may reflect confounding by illness severity or access to preventive care. A randomized intervention improving comprehension supplies stronger evidence about one mediator, but still may not demonstrate changes in disease progression. Hoffmann, T. C., & Del Mar, C. showed that expectations about benefits and harms can diverge from evidence, supporting knowledge calibration as a relevant intermediate outcome.
A defensible causal model distinguishes upstream educational and structural factors, information access, message processing, judgment, chosen behavior, treatment availability, and clinical endpoints. Each arrow should be evaluated with an appropriate design. The outcome hierarchy prevents the scientifically unjustified practice of using a favorable knowledge score as a surrogate for proven therapeutic effectiveness.
Education, Socioeconomic Position, and Structural Inequality
Health literacy arises within unequal distributions of time, education, language access, stable connectivity, disability accommodations, and the ability to obtain clinical follow-up. A person who comprehends a prescription may still lack the resources to fill it. Conversely, an institution that communicates in inaccessible terminology can create barriers across educational levels. Brach, C., Keller, D., Hernandez, L. M., Baur, C., Parker, R., Dreyer, B., Schyve, P., Lemerise, A. J., & Schillinger, D. and World Health Organization. make the organizational environment central to improvement.
Education, age, and income can predict survey scores, yet these associations do not justify treating demographic groups as intrinsically incapable. The scientifically relevant question is which resources, experiences, message demands, and institutional policies explain observed differences. Research should report the sample's context and examine effect modification rather than ranking populations by presumed ability.
Equity-focused evaluation compares outcomes across groups for the same message, checks whether benefits persist in routine settings, and asks whether the intervention changes access for underserved participants. A system-wide average improvement may conceal worsening exclusion for those who rely on translation, screen readers, or low-bandwidth access.
Culture, Translation, and Multilingual Communication
A literal translation of a medical message may preserve words while altering pragmatic meaning. Languages differ in expressions of likelihood, authority, obligation, uncertainty, and politeness. Cultural expectations about clinician roles can affect the readiness to ask questions and the interpretation of an invitation to choose. Reliable research therefore tests whether recipients reconstruct the intended probabilities, alternatives, and conditional statements in the target language.
The internationally oriented model in Sørensen, K., Van den Broucke, S., Fullam, J., Doyle, G., Pelikan, J., Slonska, Z., & Brand, H. and European comparison in Sørensen, K., Pelikan, J. M., Röthlin, F., Ganahl, K., Slonska, Z., Doyle, G., Fullam, J., Kondilis, B., Agrafiotis, D., Uiters, E., Falcon, M., Mensing, M., Tchamov, K., van den Broucke, S., & Brand, H. offer foundations for examining cross-cultural variability. Equivalent scales require documented adaptation, cognitive interviewing, and evidence of measurement comparability; a translated instrument cannot assume its psychometric properties remain invariant.
Multilingual information systems also encounter literacy variation within a language community. Professional interpretation and accessible written follow-up address different needs. Evaluation should specify whether the intervention changes comprehension, confidence to ask questions, access to services, or clinical behavior, because these are separate possible mechanisms of benefit.
Development, Aging, and Family Mediation
Health-information tasks change across the life course. Adolescents encounter preventive information through schools, families, and digital networks; adults often manage competing obligations and chronic-condition decisions; older adults may navigate increasingly complex care and medications. Age alone cannot explain comprehension without considering education, cognitive function, disease experience, vision, hearing, and interface accessibility.
Family members and informal caregivers can support the interpretation and practical use of health information, but the involvement of another person changes confidentiality, authority, and whose preferences are represented. A triadic interaction may improve retention while also creating a risk that the patient's own understanding is unobserved. Research instruments should distinguish the recipient's demonstrated comprehension from the household's collective coordination.
Longitudinal designs can reveal skill development and contextual adaptation when participants face new diagnoses or technologies. The important outcome is whether people are enabled to understand and act according to their circumstances at each life stage, not conformity to a single standardized educational profile.
Disability, Accessibility, and Universal Information Design
Comprehension barriers can arise through visual, auditory, motor, language, or cognitive access demands, independently of the substantive evidence. A risk graphic without alt text, a video without captions, or an inaccessible appointment interface can impede informed understanding. Universal design asks whether essential information can be perceived, navigated, and verified through multiple usable channels.
Accessible alternatives should preserve semantic and quantitative equivalence. A textual description of a chart needs its denominator, relative positions, comparison groups, and uncertainty, rather than merely announcing that a chart is present. Audio and conversational interfaces may enable access but must permit correction, repetition, and a permanent record. The institutional frameworks of Brach, C., Keller, D., Hernandez, L. M., Baur, C., Parker, R., Dreyer, B., Schyve, P., Lemerise, A. J., & Schillinger, D. and Centers for Disease Control and Prevention. support treating accessibility as part of organizational competence.
An accessibility audit must include participatory testing with intended users, not only automated checks. Outcome measures should examine whether the alternative representation supports the same decision-relevant inferences as the original. This connects disability inclusion directly to the epistemic validity of health communication.
Screening, Medication, and Chronic Disease as Communication Contexts
Screening communication concentrates on prevalence, test accuracy, downstream procedures, overdiagnosis, and preferences under uncertainty. Medication communication concentrates on anticipated benefit, adverse effects, practical burdens, and monitoring. Chronic disease education often includes repeated actions, changing goals, and information retained across many encounters. Each context tests distinct aspects of health literacy while preserving a common requirement for comprehensible evidence.
The study of Schillinger, D., Piette, J., Grumbach, K., Wang, F., Wilson, C., Daher, C., Leong-Grotz, K., Castro, C., & Bindman, A. B. concerns closing communication loops in diabetes care; Dinh, T. T. H., Bonner, A., Clark, R., Ramsbotham, J., & Hines, S. synthesized teach-back interventions involving chronic conditions. Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. evaluated decision aids across many treatment and screening decisions. The variation in setting matters: results from one condition or an elective choice cannot be applied mechanically to acute care or another disease.
Context-specific outcomes should include correct identification of a treatment task, knowledge of alternatives, understanding of warning signs in the relevant clinical instructions, and appropriate escalation channels. The monograph examines how these constructs are studied and does not issue instructions about a particular drug or diagnosis.
Decision Aids and the 2024 Cochrane Evidence
Decision aids organize evidence about options and consequences in a form intended to support deliberation. The Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. Cochrane review included 209 studies with 107,698 participants, covering 71 different decisions. It reported high-certainty improvements in knowledge and accuracy of risk perceptions compared with usual care, while also evaluating values congruence and decisional conflict. These outcomes closely match the present dossier's concept of informed understanding.
The review reported a mean knowledge difference of 11.90 points on a 100-point scale (95% CI 10.60 to 13.19) across 107 studies involving 25,492 participants. These are review-level results rather than estimates for every decision aid. The same synthesis assessed risk of bias, noting limitations in reporting of selective outcomes and blinding in portions of the evidence. A high-quality summary retains both the reported benefit and the scope of the comparison.
Decision aids supplement informed discussions; they do not substitute for clinical assessment, individualized interpretation, or genuine opportunity to express values. Their effectiveness can be evaluated through content accuracy, accessibility, patient knowledge, choice awareness, values concordance, and whether a resulting choice is informed.
Risk Communication Intervention Evidence and Its Limits
Intervention studies range from short framing experiments to organization-wide educational programs and decision-aid trials. A clear causal claim requires alignment between what was randomized and what was measured. Randomly assigning a graphic permits attribution of differences in immediate task accuracy to the graphic under study conditions; evaluating a system rollout requires attention to site effects, clinician adaptation, implementation fidelity, and differential access.
The recommendations summarized by Fagerlin, A., Zikmund-Fisher, B. J., & Ubel, P. A. are based on heterogeneous experimental and conceptual evidence. The Légaré, F., Adekpedjou, R., Stacey, D., Turcotte, S., Kryworuchko, J., Graham, I. D., Lyddiatt, A., Politi, M. C., Thomson, R., Elwyn, G., & Donner-Banzhoff, N. review concerns professional shared-decision-making behavior; Dinh, T. T. H., Bonner, A., Clark, R., Ramsbotham, J., & Hines, S. concerns teach-back in chronic disease education. Pooling these different outcomes under a single label of “effective communication” would obscure which mechanism has been demonstrated.
The strongest actionable synthesis is mechanism-specific: consistent denominators can improve quantitative comparison, natural frequencies can support certain Bayesian tasks, clarification can expose misunderstanding, and decision aids can improve measured knowledge. Effects on adherence, health, costs, or mortality need their own designs and cannot be inferred merely from comprehension.
Health Communication During Emergencies and Infodemics
Public health emergencies create rapidly changing evidence and intense pressure for simple recommendations. The information environment can mix provisional scientific knowledge with rumors, institutional messages, and social transmission. Under such conditions the task of literacy expands to interpreting revisions, distinguishing uncertainty from error, and understanding why guidance may change as new evidence accumulates.
Okan, O., Bollweg, T. M., Berens, E.-M., Hurrelmann, K., Bauer, U., & Schaeffer, D. examined coronavirus-related literacy during an infodemic and Lewandowsky, S., Ecker, U. K. H., Seifert, C. M., Schwarz, N., & Cook, J. supplied a more general cognitive framework for misinformation persistence. These literatures imply that effective messages should name their evidence status, date, intended population, and planned updating process. A single alarming or reassuring sentence may be interpreted as timeless even when the underlying recommendation is provisional.
Evaluation should study durable comprehension across message revisions, including whether people can identify what changed and why. Transparent update mechanisms, accessible question channels, and provenance of official information are testable institutional interventions rather than merely stylistic preferences.
Study Design, Confounding, and Measurement Error
A critical synthesis must distinguish foundational concepts, randomized tests of display format, survey correlations, clinical process interventions, and reviews. Confounding is especially important in literacy-outcome research because education, resources, cognitive functioning, illness experience, and organization quality affect both exposure and outcome. Measurement error can attenuate a true association or manufacture differences when tests are culturally unfamiliar.
Selection bias occurs when surveys exclude people unable to access a web form or written consent material; attrition threatens longer trials when the least supported participants are least likely to return. Publication bias can favor visually attractive interventions with positive immediate effects. Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., & Crotty, K. offers observational synthesis, whereas Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. provides comparative intervention evidence for selected decision-aid outcomes.
Methodological adequacy requires documented recruitment, prespecified endpoints, appropriate allocation and concealment for trials, reliable measurement, uncertainty intervals, missing-data analysis, and replication in independent populations. This manuscript is a critical narrative review and does not claim exhaustive retrieval, formal risk-of-bias scoring for every study, or a new quantitative effect estimate.
Reproducibility, Source Transparency, and Evidence Hierarchies
A durable research record must identify original studies by authors, date, journal, and persistent link; separate data from interpretation; and flag where a conclusion rests on an abstract rather than the full report. DOI identification confirms a bibliographic object but does not guarantee that all data, materials, or analytical decisions are available. Reproducibility also depends on task stimuli, questionnaire scoring, code, and transparent outcome definitions.
The 2026 digital communication synthesis Alshanqiti, A., Alghabban, H., Surrati, A. M. Q., Ibrahim, S., & Owaydhah, W. has an accompanying published correction, which illustrates the importance of tracking article-level updates. A synthesis should verify the corrected record before reusing exact pooled statistics and treat uncertain numerical details conservatively. The same principle applies to corrected visual-aid literature and to evolving international guidance.
Evidence hierarchies are question-specific. A randomized test of icon arrays directly assesses comprehension effects; a qualitative interview can reveal why a numerical sentence was misunderstood; a psychometric study evaluates whether a score represents the intended construct; a cohort can examine durable associations; and an organizational evaluation tests implementation. None of these designs alone answers all questions about informed understanding.
The Organization of Communication Systems
A system produces health understanding through a sequence of decisions: which evidence is selected; which outcomes are emphasized; which language and format are used; who delivers the message; how questions are received; and how the message is revised. Organizational responsibility follows from this chain because any step can introduce ambiguity or obstruction. The Brach, C., Keller, D., Hernandez, L. M., Baur, C., Parker, R., Dreyer, B., Schyve, P., Lemerise, A. J., & Schillinger, D. framework and World Health Organization. glossary describe the institutional dimension of these processes.
A practical research model can trace message provenance from scientific finding to guideline, patient-facing artifact, consultation, and observed decision. At each step an audit can compare the communicated proposition with the underlying evidence, including its population, numbers, uncertainty, and date. This approach integrates cognitive science with information governance without erasing their distinct mechanisms.
The resulting unit of study is the information pathway. A patient-facing text cannot be evaluated fully without examining where it appears and what assumptions its user brings; similarly, system-level policy needs measurement at the point of encounter. The purpose is to identify sources of distortion and verify that corrective interventions actually change understanding.
Contemporary Research in 2025 and 2026
The recent literature increasingly addresses digital mediation and the difference between better access to content and demonstrated comprehension. Abeo, A. N. A., Armstrong, S., Scriney, M., & Goss, H. concluded that health-literacy outcomes were often insufficiently measured in reviewed AI applications. Alshanqiti, A., Alghabban, H., Surrati, A. M. Q., Ibrahim, S., & Owaydhah, W. differentiated knowledge, behavior, clinical, and empowerment outcomes in digital communication research, finding substantial heterogeneity and comparatively sparse evidence for some emerging modalities.
These reviews temper the inference that a conversational interface is intrinsically more helpful than a static explanation. Its interaction might support clarification, but generated text can introduce ungrounded claims or alter a risk ratio. The World Health Organization. guidance provides a governance rationale for auditing accuracy, equity, and automation bias.
The same period saw renewed emphasis on health literacy as a public-health determinant in World Health Organization. and on its conceptual development in Nutbeam, D.. Together these sources support a research agenda that connects individual comprehension, organizational design, and trustworthy digital information rather than reducing the problem to a reading grade level.
Critical Scientific Discussion
Three persistent tensions characterize the field. First, broad definitions capture realistic health participation but risk construct proliferation and weak measurement; narrow instruments offer reliability for specific tasks but cannot claim to measure full agency. Second, the communication format can be experimentally optimized for immediate understanding while remaining weakly connected to clinical outcomes. Third, institutional health-literacy commitments require evaluation that separates changes in organizational inputs from changes in users' actual capabilities.
The theoretical models differ in emphasis. Nutbeam's functional, interactive, and critical framework is useful for educational objectives; Sørensen's access-understand-appraise-apply model organizes information activities; cognitive and numerical decision research explains representation effects; organizational models explain how systems create or reduce demands. These perspectives form complementary but empirically distinct levels of explanation. A credible synthesis identifies the observable variables and hypotheses belonging to each.
Evidence is strongest for specific propositions, especially certain risk-format manipulations and the knowledge and perception outcomes of patient decision aids. It is less settled for long-term generalized literacy improvements from one-off digital interventions, the transportability of risk displays across languages and disabilities, and consistent clinical gains from platform-mediated explanation. The challenge is to integrate these findings without granting any single measure the status of the entire phenomenon.
Scientific Synthesis and a Multilevel Research Framework
The reviewed evidence supports a multilevel account of informed understanding. Source fidelity supplies valid propositions; representation determines which inferences are cognitively accessible; personal experience and numeracy shape interpretation; dialogue detects and repairs divergence; institutional accessibility determines whether the exchange is possible; and a person's values connect knowledge to an actual choice. A complete study specifies variables for each layer and the causal pathways hypothesized to connect them.
At the message level, tests assess numbers, wording, comparators, and visual structure. At the individual level, they assess numeracy, knowledge, confidence, and prior experience. At the interaction level, teach-back and observed dialogue reveal whether a shared interpretation was achieved. At the organizational level, access and accommodations are measured. At the decision level, knowledge accuracy, calibrated belief, and values congruence are studied before clinical consequences are considered.
This organization is a synthesis of the cited literature, not a newly validated predictive theory. Its value is methodological: it prevents the collapse of source accuracy, literacy skill, comprehension performance, decision autonomy, and clinical outcome into an undifferentiated success measure. Future testing can compare its specified pathways against simpler explanatory models.
Open Problems and Testable Research Priorities
A priority is the development of task-specific comprehension batteries with documented construct validity across cultures, ages, and accessibility modes. Researchers should compare whether a participant can reconstruct numerical meaning, identify missing evidence, and apply a comparison to a stated goal. Measurement invariance and predictive utility across settings are more informative than adding loosely related items to a single index.
A second priority is longitudinal randomized evaluation of communication systems rather than isolated content. Studies can compare plain language, visual aids, teach-back, navigational support, and conversational assistance factorially, testing both interactions and sustained effects. Preregistration, shared stimuli, consistent outcome measures, reporting of null results, and transparent intervention fidelity are especially valuable.
A third priority concerns artificial intermediaries. Source-constrained generation, numeric consistency tests, uncertainty preservation, citation accuracy, and human override mechanisms should be examined against standard materials under real use. Error monitoring should include differential effects on people with low numeracy and limited language access. Patient well-being and clinical safety require dedicated research in addition to knowledge tests.
Conclusion
Health literacy, risk communication, and informed understanding constitute a coherent psychological research object centered on the transformation of health information into justified, usable meaning. Its history joins functional literacy measurement, public-health education, experimental judgment science, interpersonal communication, and institutional design. The central evidence demonstrates that both recipient competencies and representations of information influence comprehension.
The most defensible results are specific. Natural-frequency presentations can facilitate certain conditional inferences; framing alters preference in identifiable experiments; visual and plain-language formats can improve selected aspects of comprehension; teach-back makes misunderstandings observable; and patient decision aids improve measured knowledge and risk-perception accuracy across a substantial reviewed trial corpus. The links from these intermediate endpoints to durable health improvement vary by design, population, implementation, and access.
The scientific future lies in transparent, inclusive, longitudinal investigation of the complete informational pathway. Within the Artificial Era, new systems become intermediaries in this pathway and must be evaluated for fidelity, calibration, accessibility, and accountability. A scientifically informed health decision is an event in which reliable evidence is made intelligible, relevant uncertainty is understood, and the person has a meaningful opportunity to act according to their values.
References
- Abeo, A. N. A., Armstrong, S., Scriney, M., & Goss, H. (2025). Artificial intelligence techniques and health literacy: A systematic review. Mayo Clinic Proceedings: Digital Health, 3(4), 100269. https://doi.org/10.1016/j.mcpdig.2025.100269
- Alshanqiti, A., Alghabban, H., Surrati, A. M. Q., Ibrahim, S., & Owaydhah, W. (2026). Enhancing patient knowledge and behaviour through digital health communication: A systematic review and random-effects meta-analysis of mobile, web-based, social media, telehealth, and AI-enabled interventions. Frontiers in Public Health, 14, 1741936. https://doi.org/10.3389/fpubh.2026.1741936
- Berkman, N. D., Davis, T. C., & McCormack, L. (2010). Health literacy: What is it? Journal of Health Communication, 15(Suppl. 2), 9–19. https://doi.org/10.1080/10810730.2010.499985
- 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
- Brach, C., Keller, D., Hernandez, L. M., Baur, C., Parker, R., Dreyer, B., Schyve, P., Lemerise, A. J., & Schillinger, D. (2012). Ten attributes of health literate health care organizations. Institute of Medicine discussion paper. https://nam.edu/perspectives/ten-attributes-of-health-literate-health-care-organizations/
- Centers for Disease Control and Prevention. (2024). What is health literacy? https://www.cdc.gov/health-literacy/php/about/index.html
- Dinh, T. T. H., Bonner, A., Clark, R., Ramsbotham, J., & Hines, S. (2016). The effectiveness of the teach-back method on adherence and self-management in health education for people with chronic disease: A systematic review. JBI Database of Systematic Reviews and Implementation Reports, 14(1), 210–247. https://doi.org/10.11124/jbisrir-2016-2296
- Elwyn, G., Frosch, D., Thomson, R., Joseph-Williams, N., Lloyd, A., Kinnersley, P., Cording, E., Tomson, D., Dodd, C., Rollnick, S., Edwards, A., & Barry, M. (2012). Shared decision making: A model for clinical practice. Journal of General Internal Medicine, 27(10), 1361–1367. https://doi.org/10.1007/s11606-012-2077-6
- Fagerlin, A., Zikmund-Fisher, B. J., & Ubel, P. A. (2011). Helping patients decide: Ten steps to better risk communication. Journal of the National Cancer Institute, 103(19), 1436–1443. https://doi.org/10.1093/jnci/djr318
- Garcia-Retamero, R., & Galesic, M. (2010). Who profits from visual aids: Overcoming challenges in people's understanding of risks. Social Science & Medicine, 70(7), 1019–1025. https://doi.org/10.1016/j.socscimed.2009.11.031
- Gigerenzer, G., & Hoffrage, U. (1995). How to improve Bayesian reasoning without instruction: Frequency formats. Psychological Review, 102(4), 684–704. https://doi.org/10.1037/0033-295X.102.4.684
- Han, P. K. J., Klein, W. M. P., & Arora, N. K. (2011). Varieties of uncertainty in health care: A conceptual taxonomy. Medical Decision Making, 31(6), 828–838. https://doi.org/10.1177/0272989X11393976
- Hashim, M. J. (2024). Verbal probability terms for communicating clinical risk: A systematic review. Ulster Medical Journal, 93(1), 18–23. https://pubmed.ncbi.nlm.nih.gov/38707974/
- Hoffmann, T. C., & Del Mar, C. (2015). Patients' expectations of the benefits and harms of treatments, screening, and tests: A systematic review. JAMA Internal Medicine, 175(2), 274–286. https://doi.org/10.1001/jamainternmed.2014.6016
- Légaré, F., Adekpedjou, R., Stacey, D., Turcotte, S., Kryworuchko, J., Graham, I. D., Lyddiatt, A., Politi, M. C., Thomson, R., Elwyn, G., & Donner-Banzhoff, N. (2018). Interventions for increasing the use of shared decision making by healthcare professionals. Cochrane Database of Systematic Reviews, 2018(7), CD006732. https://doi.org/10.1002/14651858.CD006732.pub4
- Lewandowsky, S., Ecker, U. K. H., Seifert, C. M., Schwarz, N., & Cook, J. (2012). Misinformation and its correction: Continued influence and successful debiasing. Psychological Science in the Public Interest, 13(3), 106–131. https://doi.org/10.1177/1529100612451018
- McNeil, B. J., Pauker, S. G., Sox, H. C., Jr., & Tversky, A. (1982). On the elicitation of preferences for alternative therapies. New England Journal of Medicine, 306(21), 1259–1262. https://doi.org/10.1056/NEJM198205273062103
- Norman, C. D., & Skinner, H. A. (2006). eHealth literacy: Essential skills for consumer health in a networked world. Journal of Medical Internet Research, 8(2), e9. https://doi.org/10.2196/jmir.8.2.e9
- Nutbeam, D. (2000). Health literacy as a public health goal: A challenge for contemporary health education and communication strategies into the 21st century. Health Promotion International, 15(3), 259–267. https://doi.org/10.1093/heapro/15.3.259
- Nutbeam, D. (2025). Health literacy as a public health goal: 25 years on. Health Promotion International, 40(4), daaf119. https://doi.org/10.1093/heapro/daaf119
- Okan, O., Bollweg, T. M., Berens, E.-M., Hurrelmann, K., Bauer, U., & Schaeffer, D. (2020). Coronavirus-related health literacy: A cross-sectional study in adults during the COVID-19 infodemic in Germany. International Journal of Environmental Research and Public Health, 17(15), 5503. https://doi.org/10.3390/ijerph17155503
- Osborne, R. H., Batterham, R. W., Elsworth, G. R., Hawkins, M., & Buchbinder, R. (2013). The grounded psychometric development and initial validation of the Health Literacy Questionnaire (HLQ). BMC Public Health, 13, 658. https://doi.org/10.1186/1471-2458-13-658
- Parker, R. M., Baker, D. W., Williams, M. V., & Nurss, J. R. (1995). The test of functional health literacy in adults: A new instrument for measuring patients' literacy skills. Journal of General Internal Medicine, 10(10), 537–541. https://doi.org/10.1007/BF02640361
- Peters, E., Västfjäll, D., Slovic, P., Mertz, C. K., Mazzocco, K., & Dickert, S. (2006). Numeracy and decision making. Psychological Science, 17(5), 407–413. https://doi.org/10.1111/j.1467-9280.2006.01720.x
- Politi, M. C., Han, P. K. J., & Col, N. F. (2007). Communicating the uncertainty of harms and benefits of medical interventions. Medical Decision Making, 27(5), 681–695. https://doi.org/10.1177/0272989X07307270
- Reyna, V. F., Nelson, W. L., Han, P. K., & Dieckmann, N. F. (2009). How numeracy influences risk comprehension and medical decision making. Psychological Bulletin, 135(6), 943–973. https://doi.org/10.1037/a0017327
- Schillinger, D., Piette, J., Grumbach, K., Wang, F., Wilson, C., Daher, C., Leong-Grotz, K., Castro, C., & Bindman, A. B. (2003). Closing the loop: Physician communication with diabetic patients who have low health literacy. Archives of Internal Medicine, 163(1), 83–90. https://doi.org/10.1001/archinte.163.1.83
- Sørensen, K., Van den Broucke, S., Fullam, J., Doyle, G., Pelikan, J., Slonska, Z., & Brand, H. (2012). Health literacy and public health: A systematic review and integration of definitions and models. BMC Public Health, 12, 80. https://doi.org/10.1186/1471-2458-12-80
- Sørensen, K., Pelikan, J. M., Röthlin, F., Ganahl, K., Slonska, Z., Doyle, G., Fullam, J., Kondilis, B., Agrafiotis, D., Uiters, E., Falcon, M., Mensing, M., Tchamov, K., van den Broucke, S., & Brand, H. (2015). Health literacy in Europe: Comparative results of the European health literacy survey (HLS-EU). European Journal of Public Health, 25(6), 1053–1058. https://doi.org/10.1093/eurpub/ckv043
- Stacey, D., Lewis, K. B., Smith, M., Carley, M., Volk, R., Douglas, E. E., Pacheco-Brousseau, L., Finderup, J., Gunderson, J., Barry, M. J., Bennett, C. L., Bravo, P., Steffensen, K., Gogovor, A., Graham, I. D., Kelly, S. E., Légaré, F., Sondergaard, H., Thomson, R., Trenaman, L., & Trevena, L. (2024). Decision aids for people facing health treatment or screening decisions. Cochrane Database of Systematic Reviews, 2024(1), CD001431. https://doi.org/10.1002/14651858.CD001431.pub6
- Trevena, L. J., Zikmund-Fisher, B. J., Edwards, A., Gaissmaier, W., Galesic, M., Han, P. K. J., King, J., Lawson, M. L., Linder, S. K., Lipkus, I., Ozanne, E., Peters, E., Timmermans, D., & Woloshin, S. (2013). Presenting quantitative information about decision outcomes: A risk communication primer for patient decision aid developers. BMC Medical Informatics and Decision Making, 13(Suppl. 2), S7. https://doi.org/10.1186/1472-6947-13-S2-S7
- Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453–458. https://doi.org/10.1126/science.7455683
- Vraga, E. K., & Bode, L. (2020). Correction as a solution for health misinformation on social media. American Journal of Public Health, 110(S3), S278–S280. https://doi.org/10.2105/AJPH.2020.305916
- Weiss, B. D., Mays, M. Z., Martz, W., Castro, K. M., DeWalt, D. A., Pignone, M. P., Mockbee, J., & Hale, F. A. (2005). Quick assessment of literacy in primary care: The newest vital sign. Annals of Family Medicine, 3(6), 514–522. https://doi.org/10.1370/afm.405
- Williams, M. V., Parker, R. M., Baker, D. W., Parikh, N. S., Pitkin, K., Coates, W. C., & Nurss, J. R. (1995). Inadequate functional health literacy among patients at two public hospitals. JAMA, 274(21), 1677–1682. https://doi.org/10.1001/jama.1995.03530210031026
- World Health Organization. (2021). Health promotion glossary of terms 2021. World Health Organization. https://www.who.int/publications-detail-redirect/9789240038349
- World Health Organization. (2024). Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models. World Health Organization. https://www.who.int/publications/i/item/9789240084759
- World Health Organization. (2025). Low health literacy is costing health. World Health Organization. https://www.who.int/publications/i/item/B09619