Informing Patients

Informed patient consent to medical treatment has been fundamental to discussions of patient autonomy. The idea of consent to treatment was formally recognized as early as 1914 in a US Supreme Court judgment that held performing surgery without the patient’s consent to be assault, rendering the surgeon liable for damages.Reference Tauber²⁴ However, this did not address the question of what had to be disclosed to the patient to inform their consent decision. Informed consent did not appear until 1957 in a California appellate decision,Reference Faden and Beauchamp²⁶ and has been debated from both legal and ethical standpoints ever since. Over that period, informed consent has shifted from a basis of beneficence to one of autonomy. Under beneficence, the physician sought to provide medical benefits, and information was provided to support that provision. Today, the physician is expected to provide a great deal more information to support patient decision-making out of respect for the patient’s autonomy. Withholding information and lying to the patient are, of course, inconsistent with this view.

At first glance, this vision would argue for complete transparency – the physician should tell the patient everything. Consistent with this view, patients have increasing access to their medical records. Particularly with the increased use of electronic medical records, patient access has been mandated by law in many countries. In the UK, the NHS App allows a patient to view much of their GP medical record, including laboratory and radiology reports, clinical notes, and doctor–patient communications. Similarly, in the US, patients have a right to access medical records, case notes, billing and payment records, including insurance, laboratory tests, and radiology reports and images.

This stance raises three questions: What does ‘everything’ include? Is it possible to tell the patient everything? Is telling the patient everything the best way to support patient autonomy? Because the patient has far less medical expertise as a background to use the physician’s disclosures, ‘everything’ includes much more than a diagnosis, prognosis, and possible treatments and their benefits and risks. To make sense of this information, relevant background information must be provided as well. If the decision outcome is to serve the patient’s concerns and goals, information may have to be interpreted relative to them. The information would also have to be tailored to the patient’s level of health literacy. However, both tailoring and interpretation could be construed as manipulation by the physician, so the problem remains of how to present this information best.

Even if complete transparency on the part of the physician were possible, could the patient use the information to make an informed decision? Some research suggests that providing too much information can be overwhelming, leading to greater reliance on heuristics or rules of thumb that simplify the choice but can result in worse outcomes.Reference Keller and Staelin⁴⁸ Thus, autonomy viewed as independent patient decision-making presents dangers. At the other extreme, Tauber suggests that asking the physician to make the treatment decision (as in ‘if you were me, what would you do?’) does not automatically imply loss of autonomy. The patient is making an autonomous judgement that the physician can be trusted to use knowledge and expertise to seek the best outcome for the patient. If appropriately informing the patient is this complicated in the structured and well-meaning environment of informed consent, how much more complex is the issue of informing patients in the outside world?

Direct-to-Consumer Advertising of Prescription Medications

Direct-to-consumer advertising – the promotion of prescription medicines to the public in the lay media, is allowed only in the US and New Zealand. There have been industry efforts in many countries to relax restrictions on DTCA,Reference Velo and Moretti^49,Reference Between⁵⁰ notably in the European Union although DTCA there remains forbidden. In the US, it was not until 1981 that Merck ran the first print advertisement for its pneumococcus vaccine.Reference Ventola⁵¹ The first broadcast ad ran in 1983. Following a clarification of Food and Drug Administration regulations in 1985 and relaxation of the information requirements in 1999 and 2004, both print and broadcast DTCA grew rapidly. US DTCA expenditures increased over fivefold from around $800 million in 1996 to $4.8 billion in 2008.Reference Thaul⁵²

One might ask why firms would advertise products that only physicians can prescribe for patients, but DTCA in the US is effective in increasing sales and represents one of the fastest-growing forms of pharmaceutical marketing.Reference Liang and Mackey⁵³ The US General Accounting Office estimates that 8.5 million consumers annually request and receive a prescription for a particular drug from their physician in response to seeing DTCA.Reference Marks⁵⁴ It is cited by many patients as prompting questions for their doctor, including asking about a specific brand.⁵⁵

The core of the debate over allowing DTCA is whether the advertising benefits patients and should be promoted, or places patients at risk and should be curtailed or abolished (and not introduced in other countries).Reference Mintzes^56,Reference Frosch, Grande, Tarn and Kravitz⁵⁷ DTCA is regulated, and in the US, the ad must not be misleading, must provide information that relates to the medications’ uses, and include balanced information about the medication’s benefits and particularly about its risks. According to promoters of DTCA, pharmaceutical companies have more accurate, balanced, and scientifically based information than any other sources.Reference Delbaere and Smith⁵⁸ As such, they are in an exclusive position to provide people with adequate information on the safe use of medication, as well as to create effective knowledge for evaluating the benefits and risks of drug products, and generally assisting people in managing their health autonomously and appropriately.Reference Aikin⁵⁹ Also, DTCA has been credited with decreasing the under-diagnosis and under-treatment of medical conditions.Reference Frosch, Grande, Tarn and Kravitz^57,Reference Donohue, Cevasco and Rosenthal⁶⁰

Opponents emphasize that DTCA may misinform and mislead patients,Reference Almasi, Stafford, Kravitz and Mansfield⁶¹ partly because it overemphasizes the drug benefits and partly because risk information is not properly presented.Reference Davis^62,Reference Frosch, Krueger, Hornik and Barg⁶³ In this view, DTCA’s primary aim is to create name and brand recognition with the goal of increasing the use of the advertised products. Pharmaceutical companies are further criticized for promoting medicalization in DTCAReference Conrad⁶⁴ and disease mongering.Reference Moynihan, Heath and Henry⁶⁵ DTCA is depicted as being devoid of value in increasing patient literacy but rather as encouraging drug-overutilization to increase the financial gains of pharmaceutical firms.

From the standpoint of providing health information, DTCA differs from professional advice. The information provided by DTCA is no longer customized to each recipient. While a doctor uses the diagnosis of the individual patient to guide his or her advice, advertising messages are uniform. The uses, benefits, and risks of the medication are presented, and the consumer is left to decide whether to seek it. Further, because one primary goal of advertising is to increase demand for the advertised drug, the presentation of information is naturally favourable. Information regarding risks and contraindications is typically presented in small type and technical language (print) or in very rapid speech with no visual support (broadcast). As suggested by opponents of DTCA, this places consumers at some risk of misunderstanding the advertising message (e.g., the appropriateness of the medication for them) and for suffering patients’ potential motivation to focus on benefits rather than risks. As in the case of informed consent, increasing the amount of information provided does not automatically benefit consumers and patients. Thus, even good intentions (the health professional) and regulation (DTCA) cannot ensure that the content and presentation of information will best serve the patient or consumer. This issue is magnified in the absence of such constraints, leading us to the fundamental changes to the health information environment brought by the internet.

The Internet and the Changing Landscape of Health Information

For those who have grown up with the internet, life before it is a quaint anachronism characterized by the inefficiencies of standardized information – paper telephone directories issued annually and out of date by the time they appeared, paper maps for planning trips with no customized guidance, and news limited to a few broadcasters, newspapers, and magazines. Much information was simply not available – health information included. Getting information about health problems meant making an appointment with the doctor and meeting face-to-face. Email existed in various proprietary forms from the 1980s but even in 1997, the most popular email services were Microsoft’s Hotmail, with a mere 11 million users, and AOL, with 10 million.⁶⁶

The technology underlying the internet evolved from developments dating back to the 1960s but only came into being as we know it today in the early 1990s. The basic tools that enabled the World Wide Web, developed by Tim Berners-Lee, were released in 1991.⁶⁷ Use of and users of the internet grew phenomenally rapidly. In 1995, there were an estimated 16 million internet users.Reference Coffman and Odlysco⁶⁸ Web browsers and search engines had been introduced. Commerce entered the internet in 1994 with the first banner ad by AT&T and in 1995, Amazon.com was launched as one of the earliest exclusively online retailers. Blogs were launched, as were online communities, precursors of social media. By 2000, there were an estimated 361 million internet users; by 2010, that had grown to 2 billion; by 2021, over 5 billion or some 66% of the world’s population.⁶⁹ Social media is a relative latecomer. MySpace, in 2004, was the first social media site to reach a million users. Facebook launched in 2004 and grew by 2019 to 2.3 billion users.Reference Roser, Ritchie and Ortiz-Ospina⁷⁰ The digital world we take for granted today is barely thirty years old.

Health organizations were early adopters of the internet as a medium for delivering information. Medscape, a site providing peer-reviewed information for medical professionals, was launched in 1995. WebMD, the most popular health information site in the US, was founded in 1996. In addition to informational sources, the internet has also seen the development of many web-based educational sites focused on the needs of specific patient groups. For example, an online Arthritis Self-Management Program developed by Lorig and colleagues at Stanford University is built around a six-week series of online sessions focused on tools participants could use to manage pain and improve function.Reference Lorig, Ritter, Laurent and Plant⁷¹

These developments have been accompanied by profound changes in consumer behaviour that have made the internet a powerful force in the provision of health information. By 2007, over 70% of US adults used the internet and between 75% and 80% of them had searched for health information online at least once.Reference Fox⁷² Online information often led patients to ask doctors new questions or to get a second opinion. Among patients with chronic conditions, 75% indicated that the information affects a decision about treating illness. In short, in a little more than a decade, the internet became a major force in informing patients about health and medical concerns. By 2012, 81% of US adults reported using the internet and 72% of them had looked for health information online in the past year.Reference Fox and Duggan⁷³ An Australian survey of 2,944 patients found that 63.4% accessed the internet in the previous month and 28.1% sought health information.Reference Wong, Harrison, Britt and Henderson⁷⁴ Older patients were less likely to use the internet, as were those who were socioeconomically disadvantaged. By 2019, according to the National Cancer Institute Health Information National Trends Survey (HINTS), an estimated 72.7% of people who looked for information about health or medicine went to the internet first; 38.7% had watched a health-related YouTube video in the last year; 9.5% had participated in an online support group for people with a particular health issue. By 2020, 84.6% had used a health or wellness app.

The popularity of the internet for seeking health information is understandable. Vast amounts of information about medicine, including diagnosis and treatment of illness, medications, medical research and innovations, medical care providers, and related health topics are available on the internet – information provided by healthcare professionals and organizations, the traditional media, pharmaceutical companies, university researchers, and government units. On one hand, the potential of the internet as a tool for enhancing patients’ ability to cope with health conditions is argued to be extraordinary.Reference Kirsch and Lewis^75,Reference Kravitz and Bell⁷⁶ It supports the accumulation of information by individuals and the dissemination of information. Social media such as Facebook, Twitter, YouTube, and online support groups are becoming more and more important as sources of health information.Reference Chou, Hunt and Beckjord⁷⁷

The internet has also become a commercial force in healthcare. For example, while traditional broadcast media may not be allowed to carry DTCA in countries outside the US and New Zealand, information about prescription medications, including promotional communications, is today available to consumers everywhere via the internet. Through social networking tools such as Facebook pages, Twitter accounts, YouTube channels, blogs, and corporate webpages, companies have multiple new online marketing opportunities beyond any geographic borders.Reference Egilman and Druar⁷⁸ Beyond commercial interests, the absence of any regulation makes it straightforward for any group or individual to post whatever they want related to health. As a result, while the public has greater access than ever to health information, the quality of information available to consumers is more variable; hence consumers are increasingly vulnerable to exposure to misleading information. Thus, the internet has complicated the problem of providing information to health consumers and patients.

Misinformation on the Internet

The potential for medical misinformation on the internet was recognized early in its development. By 1998, dietitians were called to evaluate websites to find and expose quackery and fraud.Reference Vozenilek⁷⁹ Another 1998 study examined the first 300 references identified in an internet search on management of acute gastroenteritis in young children; 60 of the documents retrieved were from traditional medical sources, and only 12 conformed to the current American Academy of Pediatrics recommendations. Thus, not only was fraudulent information appearing on the internet, even legitimate sources could be presenting inaccurate information. The problem has only grown. While information of questionable quality can and has appeared in many traditional media and spread through social groups, the problem is magnified by the internet because the cost of production is low and reach can be wide.Reference Levi^80,Reference Eysenbach and Diepgen⁸¹ ‘Context deficit’ can exacerbate problems for the reader – lack of indicators that the information is meant for professionals or particular types of patients rather than general readers increasing potential misinterpretation or misapplication, easily skipped ‘fine print’ pages containing disclaimers, anonymity of authorship, and unclear source materials, for example, reliance on personal experience rather than clinical data.

Even predating social media, information distributed through the internet was problematic. In a 1994 study of an online forum (bulletin board) devoted to arm and hand pain due to repetitive use, 1,658 messages were posted by 313 participants over a five-month period.Reference Culver, Gerr and Frumkin⁸² Only 5% of the messages were written by healthcare providers; the great majority came from affected persons. Over half concerned medical topics, and of those, 70% were from participants without professional training, with much of the advice based on personal experience. Today, social media such as Twitter and Facebook are heavily contaminated by misinformation. A recent review found that in 69 studies published between 2013 and 2019, misinformation is particularly prevalent in the domains of vaccination, smoking (e-cigarettes and hookahs), and drug use (particularly opioids and cannabis).Reference Suarez-Lledo and Alvarez-Galvez⁸³

Outside of highly constrained domains, people are exposed to information that is incomplete (e.g., advertising for prescription medications), of questionable relevance (e.g., testimonials from friends and social media), or simply wrong (e.g., anti-vaxxer websites). The internet has enabled direct consumer access to vast amounts of health information, and while online patient decision support aids abound,Reference Dewalt, Berkman, Sheridan, Lohr and Pignone^41,Reference Stacey, Légaré and Lewis⁸⁴ 2/3 of searches for health information on the internet start from a general-purpose search engine such as Google.Reference Fox^85,Reference Eysenbach and Köhler⁸⁶ Unlike professionally managed programs, search engines provide little guidance about the integrity of the links they provide or the quality of information that can be found under these links. Different links often provide conflicting viewpoints on what information is important and how judgements and decisions should be made.Reference Anderson, Steen and Stavropoulos^87,Reference Fast, Deibert, Hruby and Glassberg⁸⁸ In political debate, such variety might be seen as a virtue, but when it comes to health information, this lack of structure can be a curse. For a person searching the internet for health information using a general search engine, assumptions of veracity, accuracy, and relevance of the accessed information are almost certainly heroic.

To understand the impact and implications of this misinformation, we begin by considering in the next section the nature of knowledge and the character of misinformation. Then we explore in Section 3 psychological processes that help to explain individual acceptance and persistence of misinformation.

Beyond access, consumer acceptance of internet information can exacerbate the problem. Often, health consumers accept internet information uncritically, particularly when it aligns with their prior views and preferences (confirmation bias), failing to consider the quality of websites in terms of the accuracy and relevance of the presented information.Reference Ahmad, Hudak, Bercovitz, Hollenberg and Levinson^89,Reference Berland, Elliott and Morales⁹⁰ As a result, information of dubious quality is accessed, accepted, and used by consumers.Reference Gauld and Williams^91,Reference Sommerhalder, Abraham, Zufferey, Barth and Abel⁹² This misinformation damages trust between patient and physician as well as between the public and the public health system. As a result, these incorrect and dangerous claims, as well as false conspiracy theories, pose serious public health threats.Reference Ball and Maxmen^6,Reference Oliver and Wood⁹³ Before these threats can be addressed, it is necessary to understand how and why people adopt dubious claims and incorrect information, and why such beliefs are difficult to dislodge and correct.Reference Schulz and Nakamoto⁹⁴

Knowledge: Fallible But Justified

Justification, too, is a long-contested epistemological issue. Of interest here, however, is the point that knowledge requires a reason, which suggests more generally that a belief doesn’t stand alone. Knowledge is contextual – meaning and evaluation of new knowledge are built on a background of pre-existing knowledge.Reference Williams⁹⁸ As a basis for justification, there must be a link between the background and new knowledge. Aristotle proposed association as that link and proposed four types – contiguity in space and time, similarity, contrast, and frequency. When events are noted as close together spatially or temporally, when they appear similar with or notably different from each other, or when they co-occur frequently, the association can form the basis for learning. In a sense, the association helps us make sense of or understand the new phenomenon. More recently, Thagard considered the problem of making sense of things we encounter like texts, people, or events and argued that we construct an interpretation that fits the available information and choose that interpretation that fits best.Reference Thagard¹⁰⁰ He suggests that coherence arises from relations among elements (representations) of things. Some relations increase coherence, such as association, deduction, and facilitation; others decrease coherence, like inconsistency, incompatibility, and negative association. The best interpretation is one that is most coherent. Haack proposes that a belief is embedded in (coherent with) a network of other beliefs as well as supported by (founded on) direct experiences. She likens her ‘foundherentist’ view to the support for an entry in a crossword puzzle.Reference Haack¹⁰¹ The entry (piece of knowledge) is consistent with the intersecting entries in the puzzle (other pieces of knowledge) forming a coherent structure, and consistent with (supported by) its clue, which Haack equates to experience. These characteristics, she argues, provide justification for the belief.

The role of this networked pre-existing knowledge in treating new information can be seen in examples like the physician’s diagnosis of disease in a patient. A patient presents with difficulty breathing and crackling sounds when listening to the lungs, and a chest X-ray reveals opacity. The differential diagnosis (plausible explanations) includes viral pneumonia, bacterial pneumonia, and lung cancer. The patient reports further that the onset of breathing difficulty was abrupt and accompanied by a high fever. The sudden onset of symptoms makes cancer less likely, and the presence of opacity on chest X-ray favours a bacterial over a viral source, so the physician diagnoses bacterial pneumonia as the best explanation of the patient’s symptoms. In this vignette, the physician’s expertise (her background) in terms of both knowledge of diseases and experience with a range of cases (thus variation in presentation, treatment, and outcome) enables her to claim justified belief in her diagnosis – what we could call knowledge. Clearly, this inferential process leans heavily on the physician’s background knowledge and diagnostic skill. The process by which a physician analyses a case (new information) to reach a diagnosis has been characterized as abduction or, more specifically, inference to the best explanation.Reference Douven¹⁰²

Abduction is not limited to expert analysis. Douven suggests it is ubiquitous and a regular feature of everyday judgement.Reference Douven¹⁰² He gives as an example a vignette where you hear of two friends who have fought and ended their friendship. Yet some time later, you hear they were seen jogging together. You conclude that they must have made up. The basis of your conclusion is that this is the most likely explanation (among any others that might come to mind). However, it is likely that you draw on other things you know about the friends, for example, neither tends to hold a grudge. Like the physician, you are using available data, coupled with your understanding of the context, to make a justified inference regarding the best (most likely) explanation. Notice that abductive inference is not proof. A physician’s diagnostic inference might lead her to call for more tests to provide more conclusive evidence for a final diagnosis. Of course, even experts like physicians can make mistakes – applying their best judgement and medical expertise to the signs and symptoms of a particular patient’s case but reaching the wrong conclusion (diagnosis or treatment). Certainly, you could be mistaken in concluding the friends made up.

This possibility of error doesn’t destroy our search for confidence in knowledge, but accepting the fallibility of knowledge changes the way it is envisioned. The focus turns from a normative goal of truth to a pragmatic concern for knowledge as a strong working hypothesis. It is well supported by pre-existing knowledge (allied beliefs and empirical evidence), and major inconsistent claims have been found to be implausible. While it could still be wrong, it is accepted in the absence of a compelling challenge. Popper argues that this is how science advances – through falsification.Reference Popper¹⁰³ A theory is accepted provisionally as one that explains phenomena of interest and has not yet been falsified. If the theory is later found wanting, a new theory may be required. From the history of science, a vivid example often discussed is the case of Newtonian physics, whose predictions regarding the motions of objects from apples to planets were supported by innumerable observations and experiments. Yet, Newton’s analysis was ultimately found to be wrong (albeit an extraordinarily successful approximation in everyday contexts) and Einstein’s theory of relativity took its place as the best explanation based significantly on a famous demonstration by Arthur Eddington that light was bent by the sun. Yet even today, questions remain as to whether the theory of relativity is somehow flawed, and experiments are conducted seeking evidence for flaws or phenomena the theory can’t explain. Similarly, in medicine, tests are often performed to ‘rule out’ possible diagnoses. Again, given the fallibility of knowledge, it is as important to rule out wrong answers as it is to link evidence favouring the answer we think is right.

Understanding: Building on the Background

Understanding and making use of new information relies on networked pre-existing or background knowledge. This background is central. When people interact with the world, they don’t just sense it (see, feel, smell, etc.); they interpret it using their background knowledge. This is how people come to understand things they encounter.Reference Knight and Tsoukas¹⁰⁴ In fact, some argue that people do more than understand; they construct the things they encounter by adopting a frame of reference.Reference Caputo¹⁰⁵ A diagnosis of hypertension may mean very different things to a doctor versus a patient. For the physician, training and experience frame hypertension as a significant medical problem that increases the risk of future catastrophic events like a heart attack or stroke, so it needs to be addressed. For the patient, hypertension is invisible. It has no symptoms, and the patient, given his experiences, sees himself as healthy. As such, he might discount the seriousness of the condition and the urgency for treatment. Moreover, the side effects of some medications are uncomfortable – why make yourself feel sick?

The frame of reference arises from a background far beyond the particulars of the diagnosis. It is a function of experience, culture, values, attitudes, and other social influences. For someone raised in a Western European tradition, ginger is a slightly exotic spice. For someone raised in an Asian, particularly Chinese tradition, ginger is also a medicine useful for the treatment of a range of health problems, including nausea and vomiting, migraine, and arthritis.Reference Shahrajabian, Sun and Cheng¹⁰⁶ The background, then, is implicated in activities as basic as making sense and understanding new information and applied to further uses like judgement (medical diagnosis), decision (whether or not to call the doctor), and forecasting (my sprained ankle will take two weeks to recover). It isn’t possible to enumerate the components of one’s background, but there are some hints as to how it is constituted.

While background knowledge certainly includes pre-existing information about the world, the example of abduction suggests that it includes something more. Gilbert RyleReference Ryle¹⁰⁷ drew an important distinction between ‘know-that’ (propositional knowledge) and ‘know-how’ (non-propositional knowledge). Propositional knowledge is what one might typically think of as knowledge. It is presented in statements that can be examined and analysed to see whether they should be accepted or believed. The propositional structure is central: knowledge means to know something about something. Whoever claims to know something is thereby stating their belief that something specific is or is not the case.

Non-propositional knowledge includes abilities, skills, and competences, as well as practical knowledge and experience. This type of knowledge is important in allowing us to use (understand, interpret, evaluate, and apply) propositional knowledge. In concert, the two types of knowledge enable access to a certain area of reality. In contrast to propositional knowledge, however, non-propositional knowledge can neither be described by its relatedness to a specific object nor can it be adequately objectified. Even if one can make statements about a particular ability or skill, the content of that competence can never be fully presented in statements and communicated to another. How an experienced doctor reaches his diagnosis, for example, cannot be described or specified in every detail, and certainly not in enough detail to allow a novice to diagnose new cases. The skills involved are developed through training and experience and are not simply the sum of propositional knowledge. In addition, such non-propositional knowledge is inherent to a person in a different way than is the case with propositional knowledge. It cannot simply be adopted; it can only be acquired by oneself, which explains the limited ability to communicate it. Try, for example, to explain how you are able to read this text.

Non-propositional knowledge is essential for the acquisition of new information and knowledge. How else could one assimilate new information? Searle identifies key functions of the background: enabling linguistic and perceptual interpretation, structuring consciousness (in a sense similar to that suggested by Thagard and Haack), assigning categories to objects and narrative meaning to sequences of experiences, and thereby setting up conditional expectations regarding interaction with the objects and other experiences, and mobilizing motivational and behavioural dispositions.Reference Searle^108,Reference Searle¹⁰⁹ In this way, the background enables meaning to be attached to new information.

Subjectivity

People learn about new things in many ways. Sometimes they observe something new directly. For example, walking down the street, you see a sign in a shop window saying a health clinic will be opening soon. Or they might experience it – their doctor is on an extended trip and they are seen by a locum tenens physician taking her place for a period of time. However, much of what we learn is not direct, but things we are exposed to via communication. We learn about health from TV news and advertising, friends and neighbours, and social media. When we experience a health problem, we consult the internet or a health professional like a physician, therapist, or pharmacist. This means that much of our knowledge is socially constructed. Other people are providing us with information and telling us what it means – how to interpret and understand it, that is, develop knowledge about it.

This social aspect of knowledge adds a new complexity to what we can claim to know. Searle proposes some clarifying distinctions.Reference Searle¹⁰⁸ He first notes our everyday or commonsense distinction between objective and subjective knowledge. He terms this epistemic, relating to statements like ‘The carrot in my refrigerator weighs fifty grams’ versus statements like ‘I prefer the flavour of carrots to that of broccoli’. The first statement is epistemically objective in that we can test its truth by referencing a feature of the external world, for example, the reading of a scale. The second statement is not testable in this way. It is epistemically subjective; the ‘truth’ of the statement lies in the judgement of the person making the claim, and like any preference statement de gustibus non est disputandum. One may feel strongly about preferences, believe in them, and make decisions based on them. One can also present an argument (justification) for them. However, the truth of the statement is not testable by reference to a feature of the external world.

Searle proposes an additional distinction. He argues that there are ontologically subjective and objective facts. Ontological subjectivity relates to the existence of the entity itself – the existence of an ontologically subjective entity is ‘observer relative’. The fact that a chemist has synthesized 2-(4-isobutylphenyl) propionic acid in a laboratory is ontologically objective – the compound exists whether or not I am aware of it. On the other hand, the fact that I take ibuprofen (the pharmaceutical name for that compound) to lessen my knee pain is ontologically subjective. Ibuprofen would not be an analgesic if I were not there to experience it. In other words, the fact that ibuprofen is an effective analgesic was created by people and shared with others who accepted it. The analgesic function is certainly tied to its chemical composition, but if there were no people to experience it, ibuprofen could not be an analgesic. Thus, ibuprofen as an analgesic is ontologically subjective.

Much of a person’s interaction with health products and services is subjective, either ontologically or epistemically (I like ibuprofen because it helps to get rid of my headache) or both. This is because we interact with them with a purpose in mind, and as such, their existence as products must be ontologically subjective. Ibuprofen is an analgesic because people use it as one. The shared recognition of use is significant. Among ontologically subjective facts, Searle distinguishes a class that is: (1) intentional, that is, they reflect purposes and goals, and more specifically (2) functional in an agentive sense, that is, they reflect the use of an object relative to a purpose or goal, and (3) collective, that is, they are shared by members of a relevant group. These are social facts – shared mental facts that relate to the function assigned to an entity to fulfil some purpose. When the pharmaceutical company Boots introduced ibuprofen in the UK as a prescription medication under the brand name Brufen, it was introduced to physicians as an effective analgesic for rheumatoid arthritis. While advertising and promotion increase awareness, they are often viewed with scepticism and suspended disbelief. As physicians gained experience with Brufen, they found it efficacious and grew to accept the advertised claim. Thus, through its marketing, Boots created the social fact that Brufen is a product effective for managing arthritic pain.

Searle proposes one further refinement of social facts. Some social facts arise from the nature of the object, for example, using ibuprofen as an analgesic. Other social facts are unrelated to an object per se but are created and continue to exist only because people treat them as facts. For example, a patient about to undergo a surgical procedure signs a consent form, creating a social fact called informed consent. This status is recognized by the patient, her healthcare providers, and others as a fact – that the patient understands what is to be done and states that this is acceptable. More generally, it is accepted that informed consent is expected and officially sanctioned by the government for all such procedures. Searle calls these institutional facts. Institutional facts exist because (and continue to exist only so long as) people collectively accept them.

Given social facts, then, knowledge involves more than the attainment of objective facts about the world. Moreover, even knowledge about epistemically objective facts is fallible. We can believe and provide justification for belief and make a strong case for it using our cognitive skills and pre-existing background knowledge, but we could still be wrong. With social facts, which are created and continue to exist through collective acceptance, even a direct link to objective facts may not be invoked. This doesn’t mean that social facts are completely arbitrary. They can be supported by (coherent with) justifying data – for example, a consensus of expert opinion, the best available evidence, and institutional sanction. However, justification as a process is not enough to address fallibility. Lay consensus for incorrect information could result in a social fact that is not true.Reference Vraga and Bode¹¹⁰

Using Knowledge: Applying Facts

The point of requiring that knowledge consist of justified beliefs is to increase the likelihood that knowledge is veridical. In professional discussions of health and medicine, the focus is ‘the facts’, which refers to information that is epistemically objective. The importance of veridical objective knowledge in the domain of health is very practical; without it, health professionals cannot provide advice that improves patients’ health, and patients risk making choices that will not improve and may worsen their health. The question, of course, is how to know what information qualifies as veridical. This is one of the most important functions of non-propositional knowledge – assessing the veracity of information.

Vaccination against childhood diseases, including measles, is safe and effective in preventing serious possible complications, including lifelong disabilities. Of course, this requires a definition of ‘safe and effective’. Vaccines against childhood illnesses today are 85–90% effective in preventing disease.¹¹¹ Clinical studies and epidemiological data show that serious adverse events are rare and in many cases are shown to be caused by factors unrelated to vaccination. Cases of actual adverse events due to vaccination are extremely rare. The rate of anaphylactic reaction to vaccines in general is estimated at about 1.3 per million vaccine doses.Reference McNeil¹¹² To give some context, the odds of being struck by lightning – the prototypical benchmark for rare events – is 1 in 15,300 in an 80-year lifetime,¹¹³ 85 times more likely than vaccine anaphylaxis. Based on thousands of studies of vaccine safety and efficacy, the safety of vaccines is an epistemically strongly justified fact. The vast majority of healthcare professionals form a community of belief, and government agencies like the US Food and Drug Administration lend official approval and supportive infrastructure for vaccination. All these facts provide a strong evidentiary basis for a professional to justify medical advice. Thus, given the purpose of protecting people’s health, that vaccination is safe, effective, and therefore to be encouraged is an institutional fact. This network of background knowledge leads to the professional’s recommendation that unless there is a specific contraindication, a person should be vaccinated – an epistemically subjective statement but again, coherent with and justified by a network of background knowledge.

The internet has given patients and consumers access to these statistics and professional advice, but it also gives them access to contrary arguments – that vaccination is neither safe nor effective. However, the evidentiary base for these arguments relies not on systematic research but anecdote, speculation, and emotional plea. The stories may be compelling in their tragedy, but they provide no systematic evidence for the purported danger of vaccination. Knowledge may be fallible, but that doesn’t mean that any belief is as justified as any other. It is not enough that a child’s diagnosis of autism followed a vaccination. Temporal contiguity is only one aspect of a causal link. Other possible causes must be ruled out and using far more evidence, inference made to the best explanation. In that process, a vast body of research has shown no systematic link between vaccination and autism. To provide a ground for sound judgement and decision, the network of beliefs must conform to the full range of external evidence.

Evaluating Information: The Importance of Objective Facts in Health

In some domains, the links between cause and effect, and thus the criteria appropriate for evaluating claims, can be difficult to discern. In political economy, the impact of policies is often difficult to predict with any precision, and their adoption often has unintended consequences. Globalization was promoted as increasing economic efficiency (thus wealth) but grossly underestimated the dislocations for those whose jobs moved to other countries. Those defending globalization might argue that this reflects a clash of goals and that wealth creation is the greater good. Beyond such reasoned disagreements, social facts with weak justification thrive in a range of contexts. Some of the most notorious in recent years can be found in politics, where ‘fake news’ – a term used to cast doubt on any media coverage that disagrees with a preferred position – invites the hearer to ignore the disagreeable information regardless of the evidence, including even direct observation. When Donald Trump was inaugurated in 2017, his press secretary claimed that the audience was the largest ever, and when challenged by photographic evidence, Trump claimed the criticism to be fake news. A senior advisor described the claim as not a falsehood but an ‘alternative fact’.Reference Knight and Tsoukas¹⁰⁴ The torrent of such unsupported claims (disinformation), particularly from highly visible political commentators, lends greater legitimacy to the claims – true or false. The prevalence of alternative facts and widespread belief in them, for example, among millions of avid Trump supporters in the US, has been more broadly described as ‘post-truth society’.Reference Strong¹¹⁵ This phenomenon is ascribed at least in part to extreme political polarization in which core values and goals are in conflict among political groups; that is, the selective acceptance and interpretation of presented information is driven by a political frame of reference which, in turn, gives rise to particular goals and objectives such as justifying denial of the legitimacy of the 2020 presidential election.

One would like to think that health is different from politics. While people argue about what constitutes a good state or government, the idea of good health seems clearer. The WHO summarizes this idea as ‘Health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity’.¹¹⁶ It is hard to imagine that less well-being and more disease would be better health objectives. In the domain of health, then, even social facts relate to actual health goals and outcomes. Accurate information about health, health promotion, and medical treatment helps to advance the goal of improving health or at least not making it worse. This feature of health information is critically important in that it makes it possible to move beyond differences of opinion. In measuring health knowledge, for example, we can appeal to expert consensus to distinguish between fact and fiction so that it is meaningful to say that the better a person’s ability to make this distinction – identify correct statements as correct and incorrect statements as incorrect – the greater the person’s health knowledge.

However, thinking about health doesn’t take place in a vacuum; it is embedded in a larger background that can distort it. Recent controversies over the management of Covid-19 highlight the importance of this feature of health knowledge. With Covid-19, the ground (background) and perspective of discussion shifted for many from health to politics. Those arguing against mandated wearing of masks, social distancing, and vaccination positioned the policies as government intrusion attacking personal choice. Opposing or refusing to comply with such mandates was seen as ‘standing up to’ government encroachment on personal freedom. This shifted the focus away from objective health knowledge to subjective political opinion and social facts about government, where charges and countercharges of fake news and alternative facts held sway and choices could be seen as serving more important goals and information inconsistent with those goals could be rejected.

The foregoing discussion of knowledge might suggest a very deliberate and probative approach to accepting new information. In everyday life, we do not test our explanations rigorously. We make an observation (I have a migraine) or an inference (My headache is a migraine because it is too severe to be something else) and act on it. Rarely do we consider whether we might be mistaken about the headache and generate and test alternative explanations for it. Much of the time, we are likely to treat our knowledge as verified and sufficient and use it that way. Only if challenged or when we feel we don’t know enough are we likely to consider whether we could be wrong. The next section elaborates on the ways this can lead people to be misinformed.

Low Literacy

A second reason for knowledge failure is inability – the problem of low literacy, particularly limited ability to read and comprehend text, which is termed functional literacy. A low-literate person could lack knowledge even if they seek to become more informed. In the domain of health, measures of functional health literacy assess a person’s ability to pronounce medical words (REALM),Reference Davis, Crouch and Long¹¹⁸ measure reading and numeracy skills by asking respondents to fill in gaps in texts (TOFHLA),Reference Baker, Williams, Parker, Gazmararian and Nurss¹¹⁹ and answer questions about a nutrition label (Newest Vital Sign).Reference Weiss, Mays and Martz¹²⁰ These measures tap the ability to attain the correct answer to the questions posed. Some 26% of US adults and 34% of the elderly exhibit inadequate ability to read and understand medical information, generally consistent with prevalence reported for countries in the European Union.Reference Baccolini, Rosso and Di Paolo¹²¹ The limited ability to access and make use of health information has predictably deleterious consequences – less knowledge and comprehension of healthcare resources, lower rates of compliance with medical advice, and worse outcomes, including increased hospitalization and health costs.Reference Andrus and Roth¹²

Researchers in health literacy realized that functional health literacy is incomplete as a basis for patient decision and action so broader conceptualizations have been developed, such as that of the US National Institutes of Health:

These abilities point to skills more complex than functional literacy. Sørensen et al. enrich the construct to include access, understanding, appraisal, and application of knowledge.Reference Sørensen, van den Broucke and Fullam¹²³ Such abilities depend not only on reading skills but also on background knowledge that makes health literacy contextually specific – mastery of medical information and decisions about particular medical issues so health literacy becomes context-specific.Reference Schulz and Nakamoto¹⁵ Such domain-specific propositional and non-propositional knowledge is foundational to a person’s ability to make sound judgements and decisions that will lead to improved health outcomes. As such, health literacy includes factual knowledge in a particular domain.Reference Sørensen, van den Broucke and Fullam^123,Reference Ishikawa and Kiuchi¹²⁴ In addition, skills such as interpreting medical information by comparing data to criterial values, assessing risk, and calculating probabilities tap more demanding skills related to access, understanding, assessment, and application of medical information.Reference Schwartz, Woloshin and Welch¹²⁵ Lacking these skills, the person becomes more dependent on others to select and interpret information and recommend actions. Selecting the credible other then becomes an essential task, calling on, for example, ‘knowledge and skills necessary to become a more mindful and sceptical news consumer’.Reference Vraga and Tully¹²⁶

Beyond topical specificity, the expansion of health literacy makes it personal. It invokes not only information that is immediately needed, but also the background knowledge necessary to comprehend the information in the context of the person’s health, to apply the information as it pertains to the person’s needs and goals, and to use the information to make personal health decisions. Low health literacy, then, refers not only to the inability to read and comprehend information, but also the inability to personalize and apply the information.

Problems due to limited health literacy have been examined in many domains. In the area of medical informed consent, lack of comprehension as well as cursory reading and limited recall are recognized as problems associated with low education and literacy.Reference Lavelle-Jones, Byrne, Rice and Cuschieri^46,Reference Glaser, Nouri and Schenker¹²⁷ Studies of consumer understanding of nutrition labels suggest problems using presented information such as sodium content as relating to salt, particularly among those with less education.Reference Cowburn and Stockley¹²⁸ In these cases, lack of relevant information or skill using it is the foremost concern. For example, in an Australian study, pregnant women were highly motivated to consume a healthy diet and believed they were capable of doing so but actually lacked the requisite knowledge as reflected in poor adherence to recommended nutritional guidelines.Reference Bookari, Yeatman and Williamson¹²⁹

Health literacy also has a subjective component – self-impressions of the ability to execute or perceived difficulty of health-related tasks. For example, the European Health Literacy SurveyReference Sørensen, van den Broucke and Pelikan¹³⁰ uses scales of perceived difficulty: ‘On a scale of from very easy to very difficult, how easy would you say it is to ….’ Tasks attached to the stem include ‘find information about symptoms of illnesses that concern you’, ‘understand what your doctor says to you’, and ‘judge the advantages and disadvantages of different treatment options’. Perceptual approaches to health literacy again focus on core dimensions like accessing, understanding, appraising, and applying information and provide an important bridge to empowerment – the patient’s motivation to take an active role in decision. However, these perceptual measures relate imperfectly to performance-based ones. Carlson et al. report in a meta-analysis of fifty-four studies an average correlation of .37, suggesting that objective knowledge predicts only about 14% of the variation in subjective knowledge.Reference Carlson, Vincent, Hardesty and Bearden¹³¹

Health literacy rests on a central assumption – that it is veridical, that is, the underlying knowledge is correct and the interpretation and application of that knowledge is sound.Reference Schulz and Nakamoto¹⁵ This is necessary for people to use their literacy skills to successfully navigate contexts such as healthcare, disease prevention, and health promotion.Reference Sørensen, van den Broucke and Fullam¹²³ The assumption is reflected in the measurement of health literacy as the presence of the requisite skills and knowledge. The questions posed have right answers, and literacy is measured as the ability to answer correctly. The low-literate patient lacks the skills, which increases the likelihood that the patient will make mistakes and fail to make health-promoting choices.

The typical remedy proposed for limited health literacy is education – to provide the patient with relevant declarative (propositional) knowledge and procedural (non-propositional) knowledge and skills. For example, the UK Expert Patient program envisions a patient who is well informed and has access to important information about his or her chronic health condition. While a newly diagnosed diabetic might know little about the condition and its management because it was never relevant, the program makes available information filtered for accuracy and applicability, as well as professional guidance in the interpretation of that information. As the patient learns more in this program, literacy – both knowledge and skill – should increase. This expertise enables patients to take a more active role in managing their condition, improving health outcomes.Reference Tyreman¹³²

Outside professionally managed programs, as discussed in Section 1, most searches for information begin with a search engine such as Google, which provides little guidance about the integrity of the websites it lists or the quality of information offered. As such, it is almost certain that some of the information will constitute misinformation. Beyond access, consumer acceptance of internet information can exacerbate the problem. Health consumers often accept internet information uncritically, particularly when it aligns with their prior views and preferences (confirmation bias), failing to consider the quality of websites in terms of the accuracy and relevance of the presented information.Reference Ahmad, Hudak, Bercovitz, Hollenberg and Levinson^89,Reference Berland, Elliott and Morales^90,Reference Ahluwalia, Murray, Stevenson, Kerr and Burns¹³³ As a result, information of dubious quality is accessed, accepted, and used.Reference Gauld and Williams^91,Reference Sommerhalder, Abraham, Zufferey, Barth and Abel^92,Reference van de Belt, Engelen and Berben¹³⁴ As in the case of ignorance, low literacy can lead to misinformation. The broader implication is, as HuntReference Hunt⁹⁹ argues, if a respondent gets it wrong (e.g., says a correct statement is incorrect), it is misleading to infer that the respondent is uncertain of the answer. It could be that the respondent believes strongly in the wrong answer. In other words, the respondent, rather than being uninformed, is misinformed.

This third type of knowledge failure – being misinformed – arises despite the person’s interest in and ability to assess a health issue. The question in this case is how a person comes to ‘know’ things that are incorrect. Research in social psychology and the psychology of judgement and decision-making has revealed a great deal about how people attend to, perceive, and evaluate information and then use it to make judgements and decisions. In contrast to classical economic models, for example, people do not attend objectively and comprehensively to presented information and do not use the information to make consistent, informed choices. They selectively access and attend to information and process it in heuristic and biased ways. The question remains, however, why at least some of the information and conclusions drawn from it will lead to people becoming misinformed.

Confirmation Bias

Confirmation bias relates to the evidence used to reach a conclusion, such as the truth of a proposition. The bias refers to the preferential use of evidence that supports the position one already favours and discounting evidence contrary to it. The bias is often unconscious. In the classic selection task developed by Wason, participants are shown four cards (Figure 1) and told that each has a number on one side and a letter on the other.Reference Wason¹³⁵

Figure 1 Wason Selection Task.

They are then presented with a proposition: ‘If a card has an A on one side then it has a 4 on the other side’ and asked which cards need to be turned over to determine whether the proposition is true or false. Logically, the proposition is an implication so a disconfirming card would be one with an A and a number other than 4, which could potentially be the case for cards A and 7, that is, if card A had a 5 or card 7 had an A, the proposition would be disconfirmed. Cards B and 4 provide no information. For B, any number would be consistent with the proposition and for 4 any letter would be. Most study participants, however, choose card A or cards A and 4 suggesting they are looking at the cases that could be consistent with the proposition (confirm it) rather than seeking information that could falsify it. This tendency to seek evidence consistent with a belief or hypothesis one holds rather than disconfirming evidence is called confirmation bias.

In an extensive review, Nickerson notes ‘A great deal of empirical evidence supports the idea that the confirmation bias is extensive and strong and that it appears in many guises’.Reference Nickerson¹³⁶ In information search, people tend to seek information they interpret as supporting their beliefs. People also tend to ignore or reject information they see as contrary to their beliefs or distort its interpretation to render it supportive. When asked to retrieve reasons from memory, people are more likely to recall reasons supporting their position. This approach, of course, makes it unlikely that a person will revise their beliefs or discover any shortcomings. Further, cognitive heuristics such as likelihood estimation based on availability – the ease of remembering – will reinforce these biases in judgement. These and analogous tendencies have been examined in a wide range of contexts – unwarranted optimism in gambling, social stereotyping, exacerbating hypochondria, even the credibility of fortune tellers and astrologers. The (historical) real-world dangers of confirmation bias discussed by Nickerson include jury members focusing on and preferentially recalling information consistent with a tentative verdict reached early in the trial, bias toward guilt in witch hunt trials, rationalization and justification of policies by government, theory persistence in science in the face of inconsistent data, and medicine, where either through haphazard ‘testing’ or chicanery, numbers of procedures like bleeding and various ineffective concoctions were promoted on the basis of ‘cures’ – simply biased reporting of sick people receiving the treatment recovering and suppression of cases where either the patient did not recover or recovered without treatment.

Information search today is heavily concentrated online, utilizing the internet and social media. In the domain of health, confirmation bias has been found to affect internet information search.Reference Suzuki and Yamamoto^137,Reference White¹³⁸ People tend to construct web searches that will generate results that include information consistent with current beliefs and are more likely to click on search results that appear consistent, which strengthens those beliefs – an effect described as reinforcing spiralsReference Slater¹³⁹ and in the case of social media, a homogeneity of views termed an ‘echo chamber’.Reference Cinelli, de Francisci Morales, Galeazzi, Quattrociocchi and Starnini¹⁴⁰

One suggestion is that this search behaviour is another consequence of low health literacy. In some cases, increased health literacy appears to help. In one study in Japan, those with lower health literacy appear to be affected by confirmation bias more than those with higher health literacy.Reference Suzuki and Yamamoto¹⁴¹ In that study, participants were first queried about the safety of genetically modified foods (negative, neutral, or positive) and their level of health literacy measured. Among those with low health literacy, those in the negative group clicked on fewer pages conflicting with their views than those in the neutral group. They also spent less time reviewing the list of results and clicked on higher-ranked webpages (reflecting shallower search). Those with high health literacy reversed this pattern, with those in the negative group clicking on more pages inconsistent with their views than those in the neutral group. However, other studies suggest that confirmation bias can persist even with high health literacy. For example, in a study of vaccination information in the Netherlands, among those with negative beliefs toward vaccination, health literacy had no influence on perceived credibility and usefulness of vaccination information.Reference Meppelink, Smit, Fransen and Diviani¹⁴²

The Problem of Overconfidence

One characteristic of the homogeneous information generated by confirmation bias is the lack of dissonant feedback. This gives rise to the problem of unknown unknowns. Dunning discusses the twofold nature of this problem in task performance.Reference Dunning¹¹ First, of course, lack of expertise degrades performance. Second, the respondent often will be unaware of this lack of expertise because ‘evaluating the correctness of one’s (or anyone else’s) response draws upon the exact same expertise that is necessary in choosing the correct response in the first place’ (p. 261). As a result, they are unaware of their lack of expertise – an example of an unknown unknown. This lack of awareness has been demonstrated in a variety of settings, including text comprehension,Reference Epstein, Glenberg and Bradley¹⁴³ political knowledge,Reference Alter, Oppenheimer and Zemla¹⁴⁴ and knowledge of clinical procedures.Reference Watts, Rush and Wright¹⁴⁵ These unknown unknowns impact self-perceptions of competence or subjective knowledge. Kruger and Dunning show that subjects (in their case students) who perform poorly on a test (lack expertise) have little sense of their lack and overestimate their performance both in absolute terms and relative to others.Reference Kruger and Dunning¹⁴⁶ Dunning notes two heuristic reasons for this overestimation. First, the subjects may believe they know more than they do – drawing on intuitive ‘knowledge’ or general impressions to derive an answer. These intuitions are often simply wrong and lead subjects to a wrong answer but a perception that their answer is rational and sound. Second, subjects may draw on a perceived false consensus – that other people have similar levels of knowledge and will perform similarly. Even the objectively worst performing quartile of subjects estimate their performance to be average or above. By contrast, high performers typically understate their performance relative to others. Dunning argues that this is due to a perception that, since the problems seem simple and straightforward, high performers assume that others experience the problems similarly.

As a result of this overestimation, one who is misinformed may feel well informed and capable of making health judgements and decisions – overconfident in their expertise. For example, anti-vaccination views can be explained as a problem of unknown unknowns – a person’s overconfidence in knowledge of vaccination even though it is wrong.Reference Motta, Callaghan and Sylvester¹⁴⁷ The task of correcting misinformation can be more difficult than providing correct information. The misinformation is often consistent with the person’s preferences, other beliefs (e.g., distrust of government or medical institutions), and subject to confirmation bias when exposed to new information, ignoring that which is inconsistent with the misinformed beliefs.

Distinguishing Health Literacy and Empowerment

Much literature on patient empowerment has examined initiatives in which health professionals design programs that show increased self-management by patients improves their access to services, their ability to manage their disease, and their health outcomes. More generally, as stated for the British Expert Patient program, the vision is that ‘many more patients with chronic diseases are well informed about their condition and medication, feel empowered in their relationship with health care professionals, and have higher self-esteem’.Reference Tyreman¹³² Thus, the programs take as a goal both situational and psychological empowerment of the patient. However, the mechanics of the program rely heavily on education seeking to improve patient literacy regarding their health condition. Work on health literacy regularly includes empowerment as a goal but seems to assume that high levels of expertise will naturally lead to effective involvement in medical decisions or beneficial self-management. For too many patients, this assumption is heroic. For example, parents who refuse vaccination for their children are exercising empowerment but it is based on information that is wrong or unrelated to vaccination. The problem of a misinformed and empowered patient is that he or she could well make dangerous choices.Reference Schulz and Nakamoto¹⁵ At the other extreme is the highly literate patient lacking in psychological empowerment who remains highly dependent on health professionals despite their apparent ability to make well-informed decisions for themselves. Thus, health literacy and empowerment are distinct and health outcomes will depend on both (Figure 2).

Figure 2 Health Literacy, Empowerment, and Patient Behaviour.

The idealized vision of advocates of both improved health literacy and patient empowerment is to move people from the lower and left cells of the figure to the upper right one – helping patients to become more effective self-managers of their health, using healthcare resources appropriately to optimize their health outcomes. Self-management doesn’t mean they have to do it on their own. Effective self-management also includes the use of available resources like educational programs and expert advice. Making such resources available is an important part of seeking to advance health literacy and effective self-management. Nevertheless, as suggested by the examples above, a central problem in a health management context may lie in the mismatch of literacy and empowerment, so that any intervention must be sensitive to the impact of the other construct.

Bounding Literacy and Empowerment

Accepting that the patient needs to know more raises the question of what information makes a person health literate. The expanded visions of health literacy discussed earlier involve the patient assessing the meaning of information, analysing it relative to a health issue, and using that analysis to participate in health decisions. Literacy programs are directed at patients lacking in knowledge and skills, so the programs proceed from a vision that more is better. Similarly, analyses of patient empowerment often are responding to the vestiges of paternalism and, as a result, advocate for greater empowerment. It is important to remember, however, that self-management need not mean completely autonomous decision-making. Nor does health literacy mean that patients gain the expertise of a physician or even some sort of pale shadow of one. The patient has unique knowledge of the symptoms, character, and consequences of his or her health problems, that is, the experience of the health problems. This knowledge needs to be communicated to the health professionals so that they can bring their expertise to bear. As noted in Section 1, empowered care becomes a constrained collaboration.

In other words, patient literacy and empowerment must be bounded. Even an empowered patient should not be diagnosing and treating himself. The literate patient not only has knowledge but also has a sense of the limits of that knowledge and when to seek help. In the expanded literacy framework, literacy could enable the patient to make judgements about when it is advisable to seek expert advice from a health professional, find the appropriate one, explain the health problem and personal concerns that made the consultation necessary, and interpret the professional’s guidance as it relates to their own health. Effective self-managing patients are active in their healthcare decisions, but also making appropriate use of available professional advice and support.

When this goes wrong because of misinformation, the intersection of health literacy, patient empowerment, and misinformation creates a complex problem for health professionals and public policy managers in seeking to convince patients to make healthy choices. This is the challenge considered in the next section.

Cognition: The Impact of Heuristic Processing

People are not aware of being misinformed; they believe their information is true. The problem is that their bases for judging truth are unreliable.Reference Brashier and Marsh¹⁵⁵ In the face of an overwhelming amount of information, people rely on mental shortcuts to make judgements about what to believe and what information to use in making judgements. Tversky and Kahneman proposed that these shortcuts take the form of heuristics – easier and faster algorithms for making judgements.Reference Tversky and Kahneman¹⁵⁸ They showed that ease of generating examples increased judgements of frequency – termed the availability heuristic, and that featural similarity predicted judgements of category membership – the representativeness heuristic. With respect to judgements of truth, people likewise unthinkingly employ fast and easy heuristics. One important heuristic for assessing truth is reliance on a feeling of comfort with the content. Coherence with information in memory is an important basis for such feeling. However, comfort can be enhanced by mechanisms seemingly unrelated to the content. For example, repeated exposure to claims makes them familiar and more likely to be accepted as true – an effect called illusory truth.Reference Hasher, Goldstein and Toppino¹⁵⁹ This is why advertising campaigns employ repeated presentations of messages. People are more likely to be exposed to them and repeated exposure increases acceptance of the messages. It is also why the spread of information on social media, creating repetitive exposure, is influential even in the case of retweets of messages not opened. Just repeated exposure to a headline can make an impact on perception of truth.

Repetition works because it makes things more likely to be recalled (available). It also makes it easier to recognize and process that information when presented again – an effect called fluency. Repetition is not the only route to fluency.Reference Brashier and Marsh^155,Reference Schwarz and Jalbert¹⁵⁶ Simply making statements easier to read using a more readable font, higher colour contrast, or more readable handwriting increases fluency compared to a font that is difficult to read. Neither repetition nor ease of reading relates to the truth of the statement, so fluency is not a reliable indicator of a statement’s veracity. Yet, fluency acts as a strong cue for judgements of truth – the ease of accessing fluent information makes individuals feel it is (should be, must be) true. Repeated presentation of false information increases ratings of truth. Even when respondents know the information is not true, they may give increased truth ratings for repeated false statements apparently relying on fluency rather than checking their knowledge.Reference Fazio, Brashier, Keith Payne and Marsh¹⁶⁰ Fluency can even stand in for coherence of information because inconsistency with prior knowledge disrupts the processing of the new information, decreasing fluency. Recalling the confirmation bias discussed in Section 3, people rely on confirmation to assess truth. Rather than ‘sifting the evidence’ in search of contradiction to evaluate information, they can rely on fluency. Because fluency makes information ‘feel’ right, it can also feel coherent, thus more likely to be true.

Heuristic processing of information also has a social component. For example, particularly when people don’t process information extensively, credibility of information is signalled by the credibility of the source. Studies of cognitive response to persuasion have long recognized the importance of a credible source for the acceptance of persuasive messages. Credibility can be inferred from actually informative characteristics, such as evidence of expertise like educational attainment and important achievements, but it can also be inferred from characteristics that are not really related to expertise, such as familiarity and attractiveness.Reference Pornpitakpan¹⁶¹ The use of celebrities to promote products in advertising draws on their notoriety as a cue to credibility. The actor or athlete wearing a brand of watch actually offers little information beyond their own attractiveness but can still convey to fans a perception of quality, style, and exclusivity.

A related cue is perceived acceptance of the information by others.Reference Ecker, Lewandowsky and Cook¹⁶² Widespread acceptance, particularly among in-group members or people viewed as similar, promotes information acceptance. Homophily is a powerful cue. ‘People like me’ are naturally credible, and when a lot of them endorse information, it is more likely to be accepted. The problem again is that an inference of widespread acceptance is affected by factors other than actual consensus.Reference Schwarz and Jalbert¹⁵⁶ Repeated exposure to information, even from the same person, increases belief in its accuracy. An account that appears in multiple social media posts makes readers more confident of the account’s accuracy, even if it is based on the same source and even the same interview with that source. Thus, many of the cues people use to assess the veracity of information – fluency, availability, and source credibility – are influenced by factors other than information content. As a result, they can lead to erroneous judgements, and they also create opportunities for persuasive manipulation.

Affect: The Impact of Emotions

The heuristics discussed focus on cognitive processes. Another important influence on judgement and decision-making is affect, particularly emotions.Reference Lerner, Li, Valdesolo and Kassam^163,Reference Ferrer, Klein, Lerner, Reyna and Keltner¹⁶⁴ One prominent view of this process is the appraisal tendency framework, which holds that ‘an emotion, once activated, can trigger a cognitive predisposition to assess future events in line with the central appraisal dimensions that triggered the emotion’ (p. 805).Reference Lerner, Li, Valdesolo and Kassam¹⁶³ The framework incorporates emotional appraisals beyond valence. Fear and anger, for example, both have a negative valence but elicit different responses because they involve other appraisals such as certainty (predictability), individual control, and others’ responsibility. Anger is high in all three of these dimensions while fear is low in certainty and individual control and medium in others’ responsibility. As a result, anger promotes a vision of control, leading to a perception of lower risk and riskier choices. Fear, on the other hand, promotes a perceived lack of predictability and control, leading to a perception of higher risk and a choice of less risky options.

Some emotions are elicited by the judgement or decision at hand. Such integral emotions are influential. For example, faced with a decision about treatment of cancer, the fear induced by the diagnosis itself can lead to selecting options perceived to be safer. This is also consistent with cuing of a prevention mindset as suggested in research on regulatory focus,Reference Higgins¹⁶⁵ where focus on preventing negative outcomes leads to choosing lower risk options. On the other hand, a patient diagnosed with mesothelioma (a type of lung cancer) associated with exposure to asbestos might well be angry at his employer because he was likely exposed on the job and, as a result, seek more aggressive treatment even if it involved greater risk. Whether this effect of emotion contributes optimally to improving the patient’s health is not certain. It is unclear whether choosing a treatment that has a greater potential for longer remission but greater risk of disability is the best choice for quality of life.

Returning to the theme of misinformation, incidental emotions unrelated to the decision at hand can also influence decision-making. Emotions cued by an unrelated newspaper article or even the weather (a sunny versus rainy day) can influence decision-making, particularly when the decision is complex, novel, or unanticipated.Reference Johnson and Tversky^229, Reference Schwarz and Clore²³⁰ This, once again, means that judgement and decision may be influenced by emotional responses that have little bearing on the issue at hand. It also once again opens the door for manipulation; by inducing feelings, a decision-maker can be nudged toward different choices. If one is angered by an internet blog and that anger carries over to a healthcare deliberation, it could foster greater risk-taking. A patient choosing such a treatment might suffer later regret about the decision, viewing it as an error.

The problem posed by the effects of incidental and unrelated sources of fluency, false consensus (the erroneous perception that others share my views), and affect is that they can give rise to a mistaken self-perception of literacy. In terms of the literacy skills noted earlier, these effects degrade the person’s ability to evaluate the accuracy and relevance of information. In addition, biases toward confirmation and overconfidence combined with these effects lead people to overestimate the sufficiency of their abilities and knowledge as bases for making a decision. Correcting these problems presents challenges quite different from the tasks of educating the uninformed.

Continued Exposure: The Information Environment

As the information environment has evolved, so too has the opportunity to spread misinformation. Because the internet is relatively unfiltered for accuracy, when people search for information about a health problem, they are likely to encounter inaccurate as well as accurate information and the presentation of misinformation often mimics that of accurate information from authoritative sources. Unproven or disproven cures are promoted using testimonials from people with advanced degrees (MD or Ph.D.) or simply claiming the title ‘Doctor’, seeking credibility. Testimonials from success stories suggest the cure works for ‘regular’ people suggesting widespread utility (social consensus). A story of the inventor’s heroic struggle to provide a miraculous cure despite harassment from moneyed corporate interests and corrupt government agencies might offer a distorted or even fictitious narrative but one that is coherent and emotionally compelling.

With the advent of widespread adoption of social media, the opportunity for repeated exposure to misinformation has greatly increased. A study of 126,000 true and false news stories on Twitter, for example, found they were shared (retweeted) by 3 million people 4.5 million times.Reference Vosoughi, Roy and Aral¹⁶⁷ False news stories spread more widely than true stories and did so more quickly. The tools that people use to access information have become a major contributor to peoples’ acceptance of misinformation.

Cementing Misinformation: The Coherence of Knowledge

In discussing misinformation, the focus is often on a particular item of information – for example, a person’s belief that ivermectin (an anti-parasitic agent) can cure Covid-19 (which multiple studies show it cannot do) – and on correcting the erroneous belief. As discussed in Section 2, that belief is embedded in a network of beliefs and buttressed by them. For example, hesitancy regarding Covid vaccination (a new item) is associated with related Covid beliefs such as the virus being manmade, that the vaccine was introduced too quickly to be safe, and pre-existing beliefs such as general vaccination hesitancy, belief in alternative medicine, and distrust of science, the pharmaceutical industry, and government.Reference Kricorian, Civen and Equils^168,Reference Jennings, Stoker and Bunting¹⁶⁹ This embedding suggests that addressing the one belief by itself rarely addresses the broader problem of the person’s misinformation and that seeking to correct multiple linked misinformed beliefs will be more difficult.

Making the task even more difficult is the attraction of misinformation. Conspiracy theories provide a prime example. These are ‘explanatory beliefs, involving multiple actors who join together in secret agreement and try to achieve a hidden goal that is perceived as unlawful or malevolent’.Reference van Prooijen and Acker¹⁷⁰ Their attraction lies in peoples’ need to make sense of their social environment, particularly when faced with uncontrollable (therefore threatening), complex, and distressing events, for example, a pandemic. Conspiracy theories provide a way to understand (make sense of) these events. Because it lends coherence, the fact that the theory could be not just wrong but even ludicrous or fanciful does not, for some people, make it less compelling, and their belief in the theory could be very strong. For those harbouring a strong distrust of government, the idea that Covid-19 was invented to allow for increased government control of the population and that Covid vaccines contain microchips that enable the government to track recipients is believable. Likewise, belief in other false claims such as ‘natural’ herbal concoctions billed as boosting the immune system to prevent Covid and ‘alternative’ remedies to cure it as substitutes for vaccination and medical intervention have been embraced. Of course, not everyone subscribes to conspiracy theories. Certain individual differences, for example, appear to be associated with belief in conspiracy theories, notably proneness to magical thinking, Machivellianism (seeking to manipulate others), selfishness, lack of caring, and a callous nature.Reference March and Springer¹⁷¹

Thus, the coherence of networked beliefs, their ability to address feelings of powerlessness, and perceptions of regaining control make misinformation attractive, reflecting a motivational component as well as a cognitive one. These are features that make misinformation believable and even attractive, thus more challenging to correct.Reference Ecker, Lewandowsky and Cook^162,Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook¹⁶⁶

Knowing You’re Right: Commitment to Misinformation

A further difficulty is the confidence that people have about their knowledge. As discussed in Section 3, information search is often biased in favour of confirming current beliefs. In the context of the internet, the algorithms that respond to search queries and those determining content delivered on social media favour information that is consistent with prior viewing. This creates a cycle in which the searcher finds more and more information consistent with their current beliefs and grows more confident in those beliefs because of the constant presentation of consistent information. This has been called an echo chamber effect or reinforcing cycle in terms of media use.Reference Cinelli, de Francisci Morales, Galeazzi, Quattrociocchi and Starnini^140,Reference Slater¹⁷² This effect is exacerbated by overconfidence; people are often overly confident that they are informed and capable of making decisions and also feel they are typical so that their choice would be endorsed by relevant others.

Moreover, research on persuasion suggests that health consumers and patients can employ a range of strategies to maintain current beliefs, attitudes, and behaviours. These include information avoidance, counterargument, source derogation, biased processing strategies that reduce importance or isolate inconsistent information, and focus on consistent knowledge or social validation.Reference Fransen, Smit and Verlegh¹⁷³ These efforts are triggered (motivated) by the person’s perceptions of the persuasive attempts – notably when the attempts are seen as threats to personal autonomy (reactance), when they are not credible because they are inconsistent with prior beliefs and attitudes, or when they raise concerns of deception.

Campaign Strategies

Guidelines for corrective campaigns have still been advanced. Messages can offer fact-based correction, attack logical fallacies in misinformation, and attack the credibility of the misinformation source.Reference Ecker, Lewandowsky and Cook¹⁶² The timing of the campaign can be prior to exposure to misinformation (prebunking) following the strategy of inoculationReference McGuire¹⁷⁷ or after exposure (debunking). Inoculation strategies can address likely misleading persuasive approaches people are likely to encounter. Debunking campaigns must address misinformation already in circulation. Repetition can be used against the misinformation by repeating the retraction.Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook¹⁶⁶ Importantly, the effectiveness of retraction can be enhanced by providing an alternative account or narrative. Some success has been reported when the narrative addresses the motives of the source of misinformation. It has also been suggested that misinformation often offers a simple explanation and people prefer simple explanations. To the extent that the true explanation is more complex, nuanced, or bounded, it is important to emphasize the core of the argument and how it coheres with the larger information context. This may not be possible when the corrective information is inconsistent with the recipient’s other beliefs and their worldview.

A vast literature has shown how difficult it can be to change strongly held attitudes. Corrective campaigns often enjoy limited success. The general limitations of mass media campaigns can be seen in studies of advertising. For example, a field experiment of advertising by the Anheuser-Busch company found that for a well-established brand, even eliminating advertising for a year in a group of markets had no effect on sales despite the continuation of competitor advertising.Reference Ackoff and Emshoff¹⁷⁸ Only after 18 months did sales begin to decline. Thus, ‘a major objective of advertising … may well be simply to reinforce and reiterate the fact that the brand exists, is eminently acceptable for use, and is widely available to consumers’ (p. 233).Reference Weilbacher¹⁷⁹

Rather than attacking specific misinformation by providing corrections, broader goals have been proposed such as improved media, e-media, and news literacy, focusing particularly on the importance of scepticism and careful review of not only credentials but interests of information sources. Media literacy has been considered as a component of educational curricula and as a more focused intervention aimed at educating people on how to challenge media messages and protect themselves from their harmful effects.Reference Potter¹⁸⁰ However, the success of such interventions is mixed. For example, a study of social media self-disclosure found that media literacy had a constraining effect on intentions regarding self-disclosure but not on actual disclosure behaviours.Reference Masur, DiFranzo and Bazarova¹⁸¹

Overall, then, misinformation is difficult to dislodge and the effectiveness of media and other mass audience campaigns is often limited. While there is evidence that deliberation decreases acceptance of misinformation, the problem remains that consumers and patients have limited expertise to discern the quality of medical information they encounter. This means that they must at some point decide who to trust to help them understand and evaluate the information and its implications for their health. This brings us back to the medical expert and the importance of the doctor–patient relationship.

Challenges of Modern Healthcare to Patient Trust

The profound changes to the healthcare delivery system since the 1950s present challenges to patient trust. One important challenge arises from the depersonalization of the doctor–patient relationship. The doctor–patient dyad has long been a guiding principle of medicine.Reference Entralgo^186,Reference Entralgo¹⁸⁷ Within this personal relationship, sometimes compared with friendship, the doctor and patient meet each other without external influences. The duty of confidentiality imposed on the doctor contributes to the preservation of this relationship, as does the fact that the medical professional can expect the patient to place his or her full trust in him or her. Because of this relationship of trust, communication between doctor and patient can flourish despite the knowledge asymmetry between doctor and patient. In recent years, this pattern of personal relationship has been increasingly challenged by the institutional arrangements of healthcare. First, the growth of medical knowledge has led to greater reliance on specialists. Specialized medical knowledge is a hallmark of the progress of medicine since the 1950s, providing incalculable benefits to patients. There is, however, a price to pay. With the increasing level of specialization in medicine, the patient must interact not only with his or her GP but with several specialized physicians. An urban US cancer patient will be referred by the GP to a cancer centre, where the patient will be seen by an oncologist as well as other specialists such as a radiation oncologist, surgeon, nurses who administer many kinds of treatment, a palliative care physician for quality-of-life concerns including pain management, and allied health professionals including dieticians and physical therapists. This panoply of specialized care may shift the bases for trust. As in the past, if the patient sees a physician on a repeated basis (e.g., the GP or oncologist), the continuity provides a basis for the development of a personal relationship in which trust is based on each knowing the other’s likely behaviour in their interactions as well as mutual respect. With the other specialists, trust in technical expertise is the paramount concern as evidenced by credentials and reputation.Reference Wright and Holcombe¹⁸⁸ Patients still want relationships with specialists, but they focus on being treated as an individual and being respected – a member of the care team with legitimate opinions.

Coincident with the growth of specialization, the doctor and patient have been placed increasingly in a corporate structure of healthcare provision with the development of managed care, in which a corporate entity like a Health Maintenance Organization (HMO) or insurance company takes a role in structuring the interface between doctor and patient.Reference Starr²² In the case of insurance companies, the interface highlights economic aspects of access to and quality of care. The HMO places limits on consumer choice of providers. These changes led to ‘anonymization’Reference Wieland¹⁸⁹ or ‘depersonalization’Reference Shorter¹⁹⁰ of the patient’s relationship with the healthcare system, embedding that with the physician. This shift has raised concerns about trust by highlighting a transactional view of medical care constrained by limited choice either within the service network of the HMO or the payment limits imposed by insurance. This is not to say that trust has been lost entirely. In a 2014 study, only 23% of Americans reported high confidence in the healthcare system and 58% reported that physicians could be trusted, much lower than in many other countries,Reference Blendon, Benson and Hero¹⁹¹ for example, Switzerland (83%), UK (76%), and France (75%). However, 56% of Americans reported high levels of satisfaction with the treatment they received when they last visited a doctor, higher than that reported in the UK (51%) and France (38%). Americans hate the system but seem satisfied with their doctors’ care. In most managed care systems, the patient has a primary physician – a GP in the British system or a primary care physician in the US. The maintenance of a trusting doctor–patient relationship is conceptually important to these corporate entities, who recognize its importance to patient participation, compliance, and health outcomes. Nevertheless, building and maintaining that trusting relationship is more challenging than in earlier times. For example, continuity by itself does not guarantee trust. In these new healthcare arrangements, it is possible for a patient to experience continuity with a physician for whom they feel low trust, in which case, unsurprisingly, continuity doesn’t contribute to patient satisfaction.Reference Baker, Mainous, Gray and Love¹⁹² Trust is conditioned on the patient’s perception that the physician is working for their benefit – providing needed information and evidence-based personalized recommendations.Reference Rowe and Calnan¹⁹³

Challenges of the Internet to Patient Trust

The internet has become a central place for patients seeking health-related information, which has also impacted the doctor–patient relationship. For most patients, this impact is not direct. After searching health information on the internet, patients report greater confidence in discussing their health with physicians, more confidence in the physician’s advice, and more empowerment to participate in managing their health.Reference Tan and Goonawardene¹⁸⁵ Most patients feel that bringing information from the internet to their consultation with a physician either doesn’t affect their relationship with their physician or improves it. Patients report their trust in their doctor; they attest to the quality and strength of that trust, and they emphasize the necessity of having their doctor to consult.Reference Guo, Hildon and Loh¹⁹⁴ Patients desire their physicians’ expertise to help them understand and interpret internet information, particularly its accuracy as well as its relevance to their condition. Even with patients’ concern for autonomy and independence,Reference Roter and Hall^195,Reference Caiata-Zufferey, Abraham, Sommerhalder and Schulz¹⁹⁶ information-seeking on the internet is often aimed at the background of a disease rather than replacing the consultation about it with the physician. In other words, patients recognize the critical importance of the physician, especially his or her ability to tailor the responses and advice provided to the patient’s specific medical condition, their expressed concerns, and their ability and readiness to cope with the information. Delivering on these expectations can be challenging in the current healthcare setting.