Barriers to Health Information
and Building Solutions
Talya Miron-Shatz, Ingrid Mühlhauser, Bruce Bower,
Michael Diefenbach, Ben Goldacre, Richard S. W. Smith,
David Spiegelhalter, and Odette Wegwarth
Most stakeholders in the health care system—doctors, patients, and policy makers—
have not been taught to apply evidence-based information to the many decisions that
must be made daily. Little awareness of this problem exists, yet a better use of evidence
could improve outcomes for patients, increase patient satisfaction, and lower costs.
This chapter considers how the use of information that emerges from evidence-based
medicine could be improved.
“Health literacy” constitutes the rst step. After a discussion of the barriers that ex-
ist to health literacy (e.g., lack of incentive to search for health information, non-stan-
dardized reporting of health results, and poor comprehension), possible remedies are
presented. Raising health literacy by targeting individual stakeholder groups, such as
patients and health care professionals, is debated as is the option of focusing on change
in the overall health system. What is required to achieve a change both at the individual
and system levels? Solutions are unlikely to generate systemic changes in center-based
treatment variations. However, a change at one level may set off change in another. Fi-
nally, increasing awareness beyond the immediate professional community is necessary
if systemic changes are to be made. The promotion of health literacy requires careful
consideration to reach the various stakeholders throughout the health care system.
Health literacy is a broad, social concept. The Institute of Medicine (IOM
2004) denes it as “the degree to which individuals can obtain, process, and
understand basic health information and services they need to make appropri-
ate health decisions.” Statistical literacy can be interpreted as the ability to
grasp the meaning of numbers, proportions, and probabilities. However, al-
though statistical competency is important, it is not the essential component
Chapter from “Better Doctors, Better Patients, Better Decisions: Envisioning Health Care 2020,” edited by G. Gigerenzer
and J.A.M. Gray. Strüngmann Forum Report, vol. 6, J. Lupp, series ed. Cambridge, MA: MIT Press. ISBN 978-0-262-01603-2
192 T. Miron-Shatz et al.
in health information. This was recognized by the World Health Organization,
which included social competencies and skills into its de nition of health lit-
eracy (WHO 2010): “Health literacy has been de ned as the cognitive and
social skills which determine the motivation and ability of individuals to gain
access to, understand, and use information in ways which promote and main-
tain good health.”
To appreciate the complexity of health literacy, it is important to realize that
it is not a static concept. Experiences in personal health, the health care system,
as well as cultural and societal in uences all shape a person’s ability to obtain,
process, and understand basic health information. Zarcadoolas, Pleasant, and
Greer (2006:5–6) advocate a broader view of health literacy, which they de ne
as “the wide range of skills, and competencies that people develop over their
lifetimes to seek out, comprehend, evaluate, and use health information and
concepts to make informed choices, reduce health risks, and increase quality
of life.” They specify four different domains of health literacy:
1. Fundamental literacy includes the traditional view of literacy accord-
ing to the IOM de nition.
2. Science literacy means that a stakeholder appreciates fundamental
scienti c concepts and processes, and understands that technological
relationships are often complex and can rapidly change. They also un-
derstand that there is an inherent element of uncertainty to the scienti c
process. Zarcadoolas et al. (2006) suggest that only between 5–15% of
the general public are scienti cally literate.
3. Civic literacy covers the domains of information acquisition and infor-
mation source. Speci cally, civic literate individuals are able to judge
the information source as credible, possess media literacy skills, have
a basic knowledge of civic and governmental systems, and are also
aware that personal behaviors can affect the community at large.
4. Cultural literacy includes the ability to recognize, understand, and use
the collective beliefs, customs, world view, and social identity of di-
verse individuals. Cultural literate individuals are capable of interpret-
ing and acting on cultural information and context, and are aware that
certain information can only be interpreted through a cultural lens.
This expanded concept has profound implications when constructing appro-
priate messages intended to be understood easily by health literate individu-
als. Clearly, the use of plain language and simple numbers is not enough; the
broader civic and cultural context must also be considered.
Steckelberg et al. (2007, 2009) suggest that teaching competencies that
enhance consumers’ autonomy in health care issues requires combining the
concepts of evidence-based medicine, evidence-based health care, and health
literacy. Competencies originating from this new concept are referred to as
critical health literacy. Some aspects are absent from existing de nitions (e.g.,
the information that patients are not receiving and their ability to request such
Chapter from “Better Doctors, Better Patients, Better Decisions: Envisioning Health Care 2020,” edited by G. Gigerenzer
and J.A.M. Gray. Strüngmann Forum Report, vol. 6, J. Lupp, series ed. Cambridge, MA: MIT Press. ISBN 978-0-262-01603-2
Barriers to Health Information and Building Solutions 193
information). Patients need to know how to pose questions and to be suspi-
cious when, for example, a brochure speci es the bene ts of a drug or treat-
ment but neglects to list possible harms or the merits of alternative routes of
action. Indeed, everyone exposed to health information should be able to pose
questions as to the knowledge source and its bias.
We suggest that de nitions of health literacy apply to all stakeholders. The
levels at which any one particular stakeholder needs to obtain, process, and
understand information will obviously vary, as will the speci c area of focus.
However, all stakeholders can be included in a de nition if the extent of the
knowledge to be obtained, processed, and comprehended depends on whoever
is pursuing it. Notably, differences will exist not only between but also within
stakeholder categories; information should be available so that individuals can
choose how much or how little of it they need.
For the clinician, information must be obtained from the patient: What are
the patient’s preferences, life circumstances, and history? Once elicited, this
information should be used to establish the alternatives which the professional
presents and recommends. It is the clinician’s responsibility to communicate
health information to a patient and to ensure its comprehension. Improving
skills to obtain, process, and understand patient-speci c information remains
For journalists, policy makers, and public communicators, health informa-
tion must be conveyed so that the patient, reader, consumer, or citizen un-
derstands it. However, this must be done without compromising the evidence
base. When communicating to the general public, journalists, policy makers,
or public communicators must consider the potential hurdles that their stories
or accounts will invoke and help the public overcome these. Reports need to be
intelligible, though not at the cost of accuracy.
Existing de nitions of health literacy have focused on the individual per-
son—primarily the patient. They take into account the individual’s ability and
skills, yet ignore elements that could in uence and help the individual under-
stand and interpret medical information. Some of these factors are internal,
such as the patient’s beliefs and their impact on comprehension. Others are
broader in context: cultural and social aspects, or even the extent to which
health literacy is pursued. The way systems are structured affects a patient’s
ability to obtain, process, and understand health information. If such infor-
mation exists only in specialized journals, if it is presented in an incomplete
manner, if media reports sensationalize it or omit “boring” details (e.g., sample
size or numbers needed to treat), then obtaining, processing, and understanding
health information is greatly impaired. Focusing on any stakeholder at the in-
dividual level may not be the most bene cial way of promoting health literacy.
We suggest a dual approach: top down and bottom up. We discuss remedies
at the individual level, but also at the level of the information as conveyed
throughout the health system and beyond, including industry and government.
Chapter from “Better Doctors, Better Patients, Better Decisions: Envisioning Health Care 2020,” edited by G. Gigerenzer
and J.A.M. Gray. Strüngmann Forum Report, vol. 6, J. Lupp, series ed. Cambridge, MA: MIT Press. ISBN 978-0-262-01603-2
194 T. Miron-Shatz et al.
Obtaining, Processing, and Understanding Health Information
In an ideal world, society would be made up of health literate citizens, yet
this is not the case. Here we discuss potential barriers that exist to hinder the
process of accessing health information. Not every obstacle will be easy to
overcome. Still, there is value in drawing attention to them so that their impact
can be addressed in the future.
Social Construction of Roles
The chain of events whereby a person sets out to obtain information requires
motivation. A person who is content with the information they have is unlikely
to contemplate obtaining further knowledge. People need to feel unsettled and
be urged into actively seeking information. While acquiring new knowledge
requires action on the part of an individual, the motivation to do so does not
solely stem from the individual. Motivation can involve a shift in the way
society constructs the “sick” role; that is, whether a patient is a mere recipient
of information or someone who assumes an active role in health-related deci-
sion making, including the decision of whether to engage with the medical
establishment. Such construction of prevailing narratives has been shown to
be in uential in selecting treatment options (Wong and King 2008). Likewise,
for health care professionals, it would be conducive to health literacy if greater
emphasis were put on re ective practice. The active pursuit of information
by a patient does not necessarily equate to an inherent mistrust of health care
professionals, but is often perceived as such. This needs to be addressed in the
role construction of both the patient and health care professional.
Perceiving Information Seeking as Mistrust in Health Care Professionals
If patients and other stakeholders perceive information seeking as a sign of
mistrust, they might be reluctant to engage in this behavior or to encourage
others to do so. The misinterpretation of trust as blind acceptance is most often
imposed on patients, though it may apply to other stakeholders as well. Those
who provide information often struggle with numbers or may have con ict of
interest. Thus, blind trust in the health care professional who presents informa-
tion can be a barrier to health literacy.
Onora O’Neill argues that trust has fallen dramatically in many institutions:
“Mistrust and suspicion have spread across all areas of life” (O’Neill 2002:8).
The usual answer to the growth in mistrust is to insist on transparency and
greater accountability in everything. Further, O’Neill argues that “if we want
to restore trust we need to reduce deceptions and lies….well placed trust grows
out of active inquiry, not blind acceptance” (O’Neill 2002:70, 76).
One would hope that health care professionals and other stakeholders, who
are sometimes mistrusted, would view the opportunity to refer patients to in-
formation they can check and assess themselves as advantageous rather than
Barriers to Health Information and Building Solutions 195
threatening. For example, Nannenga et al. (2009) demonstrated that a decision
aid designed to help patients decide about statin use was associated with both
increased knowledge and enhanced trust on the part of patients. Trust increased
with patient participation.
David Mechanic (2004) reminds us that patients cannot always assess their
health care professionals’ skills, which then leads to trust or mistrust accord-
ingly. He mentions other bases for trust, however, which coincides with the no-
tion of professionals both conveying information and eliciting patient-speci c
input: “Patients’ trust is how doctors communicate and whether they listen and
are caring. Patients do not expect intimacy but they do seek respect and respon-
siveness…patients want to know that their doctors are committed to protecting
their interests” (Mechanic 2004:1419).
When health care professionals, policy makers, insurers, or other stakeholders
avoid presenting evidence to patients, decline to share uncertainty, and/or make
decisions on their behalf, they erect barriers to health literacy. Paternalism sup-
ports the notion of a patient as a passive, non-inquisitive recipient of whatever
information the professional wishes to impart. Health professionals may do
this with the best of intentions, believing that it actually protects patients, as
has been the traditional style of doctors for centuries. Yet increasingly, doctors
do share evidence and uncertainty with patients. Unfortunately, many health
care professionals continue to insist that patients do not want to share in the
decision-making process. It is undoubtedly true that some will choose, under
some circumstances, to allow health care professionals to make decisions on
their behalf, but strong evidence indicates that doctors assume all too often
a paternalistic attitude (Stevenson et al. 2004; Coulter and Jenkinson 2005).
Professionals should adopt a default position that patients want to be informed
and, at the very least, should ask patients how much information they would
like and whether they would like to share in the decision-making process.
Paternalism is not con ned, however, to the doctor–patient encounter. It is
evident between the media and the public, the government and the public, and
health care professionals and policy makers.
Perceived Inadequacy in Handling Medical Information
A de-motivating factor for seeking health information is the feeling that even
if accessible, it will not be understood. In a study on prostate cancer patients
(van Tol-Geerdink et al. 2006), about half of the patients (69 of 148) indicated
an initial low preference to participate in medical decisions. However, after re-
ceiving information about treatment options from a decision aid, 75% of these
patients wanted to be involved in choosing their radiation dosage. Contrast this
196 T. Miron-Shatz et al.
with the group of patients who expressed an initial high participation prefer-
ence, where 85% wanted to be involved in choosing their radiation dose.
This suggests that although patients may have initially refrained from seek-
ing information or making health decisions, preferring to defer to professional
opinion instead, once information is available in a clear and understandable
way, these same patients now feel equipped to participate in decision making.
Barriers to Obtaining Health Information
Lack of Evidence-based Information
Although there is an abundance of communication on health issues targeting
patients and the general public, it rarely adheres to criteria deemed by research-
ers as necessary for making decisions on health issues (Bunge et al. 2010;
Steckelberg et al. 2005; Trevena et al. 2006). Information must be relevant to
the individual patient, its content must be based on the best available evidence,
and its presentation must follow speci c criteria to allow understanding of the
information. Elements that are often missing include quantitative estimates of
the bene ts and harms of treatment options, diagnostic or screening procedures
using numbers, and verbal and graphical presentations. All options should be
presented, including the option not to intervene or to postpone the intervention.
Outcomes relevant for patients should be reported rather than surrogates. At
present, such information is provided only for a few areas of interest; structures
are missing to ensure that updated, comprehensive, and understandable infor-
mation can be used by all stakeholders.
It is dif cult to develop the skills to seek out health information if such infor-
mation is not reported in a consistent form. Clear, transparent templates need to
be developed and consistently utilized: in textbooks, drug information, patient
brochures, newspaper reports, and white papers. Standardized reporting is also
a key component in the development of a scheme to address existing and miss-
Existing and Missing Knowledge
To pursue information, one needs a scheme involving the kind of information
that is needed within a particular decision situation. If such a scheme is ab-
sent, one does not know what is missing, or even that the information they are
receiving is incomplete. An overview of questions to ask, to know what infor-
mation must be sought, has been provided by Gigerenzer et al. (2007). These
questions can serve as informal guidelines for designing risk information.
Barriers to Health Information and Building Solutions 197
Listed below are some of the most important questions and concepts that need
to be clari ed when examining health information.
• Risk of what? Does the outcome refer to dying from the disease, not
getting the disease, or developing a symptom?
• What is the time frame? Time frames such as the “next ten years” are
easier to imagine than the widely used “lifetime” risks. They are also
more informative because risks can change over time but ten years
may be long enough to enable action to be taken. For example, for ev-
ery thousand female nonsmokers (age 50), 37 will die within the next
ten years; however, for every thousand smokers (age 50), 69 will die
(Woloshin et al. 2008).
• How big? Since there are few zero risks, size matters. Size should be
expressed as absolute rather than relative risks.
• Bene ts and harms: Screening tests, as well as treatments, have ben-
e ts and harms. Thus information on both is needed. Always ask for
absolute risks of outcomes with and without treatment.
• Screening tests can make two errors: false positives as well as false
negatives. Understand how to translate speci cities, sensitivities, and
other conditional probabilities into natural frequencies. Ask how com-
mon the condition is in the population: Out of 1,000 sick people, how
many will the test correctly identify (sensitivity) and how many will it
miss (false negative rate)? Out of 1,000 healthy people, how many will
the test correctly identify (speci city), and how many will it mistake as
ill (false positive rate)?
Con icts of Interest
A con ict of interest occurs when an individual or organization is involved in
multiple interests, one of which could possibly in uence the other. A con ict
of interest can only exist if a person or testimony is entrusted with some impar-
tiality. Thus, a modicum of trust is necessary to create it.
An Institute of Medicine report on the topic states that “con icts of inter-
est threaten the integrity of scienti c investigations” (Lo and Field 2009:2),
and “evidence suggests that...relationships [between industry and researchers]
have risks, including decreased openness in the sharing of data and ndings
and the withholding of negative results” (Lo and Field 2009:9). However, the
authors are also explicit in saying that empirical evidence is limited and that
“data are suggestive rather than de nitive” (Lo and Field 2009:4).
In 2001, American-based pharmaceutical companies sent out some 88,000
sales representatives to doctors’ of ces to hand out nearly US$11 billion worth
of “free samples” as well as personal gifts (Angell 2004; Chin 2002). The
expectation is that when the free samples run out, doctors and patients will
continue to prescribe or take them. An editorial in USA Today (2002) painted
198 T. Miron-Shatz et al.
a vivid picture: “Christmas trees; free tickets to a Washington Redskin game,
with a champagne reception thrown in; a family vacation in Hawaii; and wads
of cash. Such gifts would trigger a big red “bribery” alert in the mind of just
about any public of cial or government contractor.”
Even though the Of ce of the Inspector General of the U.S. Department of
Health and Human Services issued a warning in 2003 that excessive gift-giv-
ing to doctors could be prosecuted under the Anti-Kickback Law, critics stress
that the laws are still full of loopholes (Angell 2004). In recognition of these
pervasive in uences, some major hospitals in the United States have denied
pharmaceutical representatives entry onto their premises and have forbidden
the sponsoring of meals where products are advocated. In addition, existing
con ict of interest regulations have been tightened to enforce strict reporting
of any incomes or stock holdings of health care-related companies, of hospital
employees, as well as their family members. It is too soon to tell whether these
efforts have been effective in curbing the in uence of the pharmaceutical in-
dustry on health care delivery.
Alongside health care professionals, patient advocacy groups also meet the
criteria for assumed impartiality. Such groups have become a target of the in-
dustry’s marketing efforts (Grill 2007) and have been found to be strongly en-
tangled with pharmaceutical companies, who often design their web sites and
pay their public relations agencies (Angell 2004). Company representatives
sometimes even take leading positions on their boards (Schubert and Glaeske
2006). Such entanglement may cloud any objective view of a pharmaceutical
product portfolio and prevent doctors’ and patients’ advocacy groups from ei-
ther requesting or reporting transparent numbers on the real bene ts and harms
of a drug or medical intervention. In general, little is known about the exact
extent of in uence that pharmaceutical promotions have on the behavior of
patient advocacy groups.
Barriers to Processing Health Information
During elementary, middle, and high school education, children are seldom
taught risk, probability, and the concept of uncertainty. This lack of education
is at the root of the prevailing levels of low statistical literacy. Teaching school-
children how to approach frequencies and probabilities helps prepare them for
the complexities and uncertainties of the modern world; it also equips them
to make sound decisions throughout their lives (Bond 2009; Gigerenzer et al.
2007). Yet the attainment of these necessary skills is not part of the curricula in
every school and in every country.
Health statistics and randomized trials are an indispensable part of clinical
practice. In 1937, an editorial in The Lancet stressed the importance of statis-
tics for both laboratory and clinical medicine, and criticized the “educational
blind spot” of physicians. In 1948, the British Medical Association Curriculum
Barriers to Health Information and Building Solutions 199
Committee recommended that statistics be included in medical education. Ten
lectures were proposed with additional time for exercises, ranging from teach-
ing core concepts such as chance and probability to interpreting correlations
(Altman and Bland 1991). Despite this, it took until 1975 before statistics be-
came a mandatory subject in medical schools within the University of London,
and it took an additional ten years before adoption in Austria, Hungary, and
Italy (Altman and Bland 1991). Although statistics has received a mandatory
status, statistics and risk communication are far from being an essential part of
medical education: Only 3% of the questions asked in the exam for certi ca-
tion by the American Board of Internal Medicine cover the understanding of
medical statistics (clinical epidemiology), and risk communication is not ad-
dressed at all. Similarly, biostatistics and clinical epidemiology do not seem to
exceed the benchmark of 3% in curricula of medical schools, and transparent
risk communication is completely lacking (cf. Wegwarth and Gigerenzer, this
volume). Another important aspect in improving literacy is an understanding
of what information is missing. For example, in the case of screening, does the
report include numbers needed to treat, does it include health outcomes of con-
trol groups that have not received screening, and does it include information
on potential harms? If society wishes to have literate doctors, medical schools
need to devote time in their curricula to teach the concepts of transparent and
nontransparent statistics, as well as which information is necessary and which
is not. Teaching medical students transparent representations fosters under-
standing (Hoffrage et al. 2000) and so does teaching doctors (Gigerenzer et al.
2007). Every medical school should require graduates to exhibit at least mini-
mal statistical literacy, with a longer-term goal of requiring more advanced
Barriers to Understanding Health Information
There are transparent and nontransparent formats for statistical information.
Much of the mental confusion that denes nontransparency seems to be caused
by the reference class to which a health statistic applies (Gigerenzer and
Edwards 2003). Single-event probabilities, by denition, specify no classes
of events, and relative risks often refer to a reference class that is different
from the one of which a patient is a member. Sensitivities and speci cities
are conditional on two different reference classes (patients with disease and
patients without disease), whereas natural frequencies all refer to the same
reference class (all patients). Survival and mortality rates differ crucially in
their denominator; that is, the class of events to which they refer. Clarity about
the reference class to which a health statistic refers is one of the central tools
in attaining health literacy.
Understanding could also be improved by providing numbers as well as
words. Patients have the right to know the extent of the benets and harms
200 T. Miron-Shatz et al.
of a treatment, and qualitative risk terms are notoriously unclear. Contrary to
popular belief, studies report that a majority of patients do prefer numerical in-
formation to care alone (Hallowell et al. 1997; Wallsten et al. 1993). Providing
patients with accurate, balanced, and accessible data on disease risk and treat-
ment bene ts could, however, in uence their choices in ways that doctors may
consider undesirable. Patients may be very surprised at how negligible many
of the risks or bene ts are. Consequently, they may dismiss interventions that
physicians might deem extremely valuable.
In one example, participants in a study were very optimistic about the ef-
fectiveness of three different drugs; in each case, these perceptions dropped
substantially after seeing the actual data (Woloshin et al. 2004). This was a
cause for concern, since one of the drugs—a statin used to treat men with high
cholesterol but no prior myocardial infarction—showed a reduction of overall
mortality over ve years from 4 in 100 patients to 3 in 100 patients. It seemed
that many respondents did not appreciate the real magnitude of this effect. Few
drugs now being manufactured can match this reduction in all-cause mortality
among relatively healthy outpatients.
To judge how well a drug (or other intervention) works, people need a con-
text; that is, some sense of the magnitude of the bene t of other interventions.
Reactions to bene t data will change as people have more exposure to them.
Consumers will be better able to discriminate among drugs and interventions
as they become better calibrated to effect sizes. This context is necessary so
that people do not discount small but important effects.
How Can Stakeholders Be Motivated to
Obtain Good Quality Evidence?
We live in an era where knowledge abounds and time is pressing. Assuming
that a general practitioner needs only ve minutes to skim an academic article,
it has been estimated that 600 hours per month are required to read all of the
published academic items pertinent to primary care alone (Alper et al. 2004).
Work on cognitive limitations and information overload leads us to question
whether simply providing more information will result in a higher comprehen-
sion levels (Shaughnessy et al. 1994). Clearly, the existence of information
is not enough. For information to be utilized, it needs to be deemed useful, a
concept Shaughnessy et al. de ned as:
usefulness relevance validity
time required to obtain
While clinicians often claim that they consult medical journals or the library
when they are unsure how to proceed, research suggests they may actually
consult with their colleagues instead (Shaughnessy et al. 1994). This clearly
fullls the requirements of efciency and relevance (a sense of local practice
Barriers to Health Information and Building Solutions 201
norms may also be a source of reassurance to the anxious practitioner), but the
information may not be of the highest quality.
It has proved challenging to develop systems capable of ensuring that in-
formation is relevant at the point of delivery. Even when there are attempts
to make clinical advice context dependent (e.g., noti cations on evidence in
practitioners’ computer systems, like EMIS), these may be deactivated by
practitioners, who regard them as intrusive or unhelpful. A system that sets red
ags, when several medications prescribed to the same patient are determined
algorithmically to be incompatible, may be disenabled if it raises alerts so as
to be overwhelming. Thus too much information, given in an automated and
unsolicited way, may not be the ideal structure.
Humans are good at judging relevance. “Information prescriptions,” giv-
en explicitly by health care practitioners to patients, provide an example and
could be improved by considering when the information is most relevant. For
example, it may be most useful for people to have access to information con-
textualizing a discussion with a clinician before an appointment, requiring that
information prescriptions are routinely sent out in anticipation of an appoint-
ment, and ensuring that such information exists. The concept of alerting pa-
tients to a range of treatment options prior to meeting with clinicians is the sub-
ject of research in shared decision making. At the Decision Laboratory (2010)
in Cardiff University, parents of children referred for possible tonsillectomy
are sent a checklist, based on a national guideline, which stipulates the criteria
for surgery. Parents are asked to assess whether their child meets these criteria.
The impact of this checklist is being evaluated.
Guessing what consumers of information will nd relevant, however, is not
enough: the information needs to be evidence based. Most decision support
systems assume that patients would like to know, for example, the probability
of suffering a heart attack with different treatments over a one- and ve-year
period. Thus, decision support systems are designed to answer these questions,
even though they may not be the ones that a patient asks. Some support systems
do present patients with alternatives in vivid terms. The Prostate Interactive
Education System (PIES 2010), for example, displays the harms and bene ts
of prostate cancer treatments in lm format, using real doctors.
An individual’s own search activities provide a valuable source of informa-
tion and may prove the most valuable resource. Similarly, just as we must iden-
tify areas of clinical uncertainty to drive the generation of primary and second-
ary research (trials and systematic reviews), we may need to consider systems
to identify where there is the most need for the evidence-based information.
Good quality evidence-based information should be available online.
Ideally, such information needs to be organized rationally, preferably in a
standardized format, and accessible with consistent search tools. The Finnish
Medical Society, Duodecim, has been producing national evidence-based
current care guidelines since 1995. This web-based tool is administered by
expert groups, who select core clinical questions to be answered; systematic
202 T. Miron-Shatz et al.
literature searches are conducted, articles critically appraised and summarized,
and detailed evidence reviews linked to the text as background documents.
Guidelines also include a patient version and an English summary. Ninety-
eight guidelines are available from a single web site (Current Care 2010) free
of charge. In 2004, physicians accessed 3.2 million articles from this database;
this corresponds to each Finnish physician reading one guideline per work-
ing day (Kunnamo and Jousimaa 2004). In 2009, the number of annual read-
ings had reached 12 million, which means that health professionals consult the
guidelines four times per day on average (Jousimaa, pers. comm.). This statis-
tic is encouraging; it indicates that when up to date, standardized information
is accessed by people who need it, when they need it.
Standardized Reports and Common Health Literacy Vocabulary
Non- standardized reporting impairs comprehension as well as health literacy
by increasing the risk that awed or misleading reporting can become accepted
as the truth. If we could develop a prototype of a transparent and comprehen-
sive report, information transfer might be optimized and health literacy might
be more easily attained. Due to the nature of reporting, meeting this goal will
not be easy. However, unless we impose a structure in this complex and rapidly
changing eld, the framing and transfer of information will continue to be un-
necessarily dif cult.
We deliberately interpret the term reports loosely. We envisage standard-
ized reporting in (a) professional medical journals and subsequent press re-
leases; (b) newspaper stories, feature articles, commercials, or advertisements;
(c) patient communications (e.g., brochures, support-group web sites, or mam-
mogram screening invitations); and (d) of cial documents intended for policy
makers, commissioners, insurers, and other major players. In terms of medi-
cal content, our suggestions relate to individual decisions, whether concerning
treatment, prevention, or diagnostics, as well as to population decisions regard-
ing public health interventions.
Are There Precedents?
Whether discussing an intervention, screening or diagnostic test, or even a
recommended behavior change, information needs to be presented and clearly
delineated so that any stakeholder can weigh the possible bene ts and harms.
At a minimum, these can be listed and informally weighted. Often there is suf-
cient evidence to quantify the likelihood of speci c bene ts and harms. In our
discussion of how these “probabilities” might be displayed, we reviewed sev-
eral suggestions for conveying information in a strati ed, standardized fashion
(for a review, see Politi et al. 2007).
Barriers to Health Information and Building Solutions 203
The Cochrane Collaboration utilizes a layered structure in its reports and
has developed a plain-language summary, which includes the use of num-
bers. The inclusion of a “summary of ndings” table for the main comparison
in the review is gradually becoming the standard (Schünemann et al. 2008).
Cochrane appears to believe that this does not compromise the quality of their
reports and has positively evaluated the format. Cochrane has standardized
as many elements of the reports as possible (Higgins and Green 2009); for
example, although the effect of the intervention may be reported as an odds
ratio, the absolute risks with and without the intervention are reported as fre-
quencies (with the same denominator used throughout the report). Both “high”
and “low” baseline risks are considered, and “ GRADE” assessments (GRADE
Working Group 2004) of the quality of evidence are provided for each bene t
Review users often become overwhelmed when more than seven outcomes
are presented; more than this is confusing and cannot be integrated. Thus, in
line with memory constraints, Cochrane Tables list no more than seven bene ts
and harms (or, more commonly, outcomes) (Higgins and Green 2009). These
are ordered by degree of bene t, followed by potential harms. Cochrane also
includes a “plain language summary” intended for patients, but accessible to
all (Higgins and Green 2009, Chapter 11).
In Germany, guidelines are presented at three levels: a short version (no
numbers) for doctors, a longer version (with numbers) for doctors, and a ver-
sion for patients that includes practical suggestions (e.g., what to do if you are
at particular coronary risk and feeling unwell: should you rush to the hospi-
tal or wait until the morning). The German Disease Management Guidelines
(NVL 2010) also includes an evaluation of the quality of the evidence; this
transparency allows clients to be critical users of the information.
The “ drug facts box” is another clear example of a successful precedent.
Developed by Schwartz and Woloshin (this volume), its goal is to provide ac-
curate information on prescription drugs to the public. By utilizing a standard-
ized 1-page table format, it summarizes the bene ts and side-effect data on
prescription drugs in a concise manner. Descriptive information about the drug
is provided: its indication, who might consider taking it, who should not take
it, recommended testing or monitoring, and other things to consider doing to
achieve the same goals as the drug.
How Can Statistical Information Best Be Presented?
We suggest that this can best be achieved by combining the strengths of the
“drugs facts box” (Schwartz and Woloshin, this volume) and the “summary of
ndings table” (Schünemann et al. 2008) using a layered format and a vocabu-
lary accessible to all stakeholders. Principles for “basic” representation are as
204 T. Miron-Shatz et al.
1. Represent probabilities as whole numbers using a constant denomina-
tor (e.g., 1000). Probabilities could also be shown as miniature icons
2. Provide a list of probabilities with and without intervention (as two col-
umns) to allow comparison between options. The difference between
interventions could then be presented in a third column.
3. Include harms and bene ts.
4. Reproduce tables for different baseline risks (but do not include rela-
5. Include an indicator of quality of evidence (but do not include con -
dence intervals in the basic presentation).
6. Provide a common format for patients, health care professionals, policy
makers, and the media.
7. Encourage an electronic version, with links to supplemental informa-
tion for those suf ciently motivated. This would allow multiple graphi-
cal formats, animations, resources, references, etc., depending on the
desires and sophistication of the user.
These principles will not apply in all circumstances, but are intended to spur
further development. The challenge and goal for the future will be to develop
standardized reporting formats for qualitative as well as for diagnostic research.
Who Can Help Make Standardized Reporting Happen?
Multiple strategies—both bottom up and top down—are necessary to imple-
ment standardized reporting. It will never be enough to embed it within guide-
lines alone, since guidelines are often not adhered to or enforced.
The U.K. Academy of Medical Sciences is proposing uniform requirements
for reporting bene ts and harms in medical journals. There is already an ex-
panded CONSORT statement to include harms (Ioannidis et al. 2004), but no
evidence on compliance. Editors and referees of peer-reviewed and other jour-
nals are best positioned to ensure adherence to guidelines. We note, however,
that editors in the mass media may lose in uence if publishing continues along
its present trajectory. Continuing medical education (CME), or an equivalent
professional organization for journalists, could help disseminate the prescribed
reporting formats, should editors become obsolete.
In line with the mandate to provide patients with helpful, agenda-free in-
formation, patient advocacy groups could help enforce implementation. Some
major charitable organizations (e.g., the U.S. National Breast Cancer Coalition)
already present information in this way.
The pharmaceutical industry provides another avenue, since it is in their
interest to make clear the balance of harms and bene ts. Hormone replace-
ment therapy information lea ets provide current examples of risk report-
ing. Paradoxically, this is one place where legislators can mandate the ways
Barriers to Health Information and Building Solutions 205
information needs to be presented. Even if a company cannot list all alterna-
tives to their proposed product, they could be required to note the ef cacy of
their treatment compared with a placebo or no treatment, as is the case in the
drug facts box.
A model by which good practice can be enforced may be found in Finland.
There, a consumer organization reviews every type of patient-oriented com-
munication according to pre-determined and well-publicized criteria, and ei-
ther grants or withholds its stamp of approval.
Standardized reporting could also be enforced through a bottom-up ap-
proach. If consumers are trained in reading statistics (when these are provided
in natural frequency format) and become familiar with a standard format, we
can expect that they will, in turn, demand that information is presented in this
way—be it from the media, industry or other source. Trained citizens would
not be susceptible, for example, newspaper stories consisting solely of relative
risk information and, likewise, would be wary of advertisements devoid of
information on numbers-needed-to-treat. Equipped with the necessary tools
and skills, informed consumers will be able to demand better, clearer, more
transparent information from their physicians, media, and others.
What Could Stop Standardized Reporting from Happening?
We should expect the call for standardizing report formats to be met by objec-
tions. One possible source could come from health care professionals, who
might perceive that standardized information (and patient aids) would interfere
with their professional judgment—making them obsolete or turning them into
automatons. Yet there is a big difference between presenting numbers clearly
and implementation. Accessible information to the patient makes the doc-
tor’s role all the more crucial: doctors will be able to discuss implications and
patient-speci c recommendations, rather than regurgitate information already
available. Rather than debilitating, doctors should view this change as emanci-
pating and enabling. Standardized reporting does not eliminate health care pro-
fessionals; it simpli es and streamlines the processing and understanding of
information, thus leaving room for more substantial discussions to take place.
Another objection concerns the modi cation to original data, which in-
volves judgment. Take the Cochrane list of harms and bene ts for example:
Who determines the importance of each of these? A reasonable way to address
such objections is to make the judgment process transparent. In addition, the
multi-layered presentation used allows anyone to go back to the original paper
or raw data to generate their own impressions and ratings.
Inevitably, various groups and stakeholders have agendas and would prefer
to select information to coincide with their interests. Thus, it is imperative for
the public to become trained in reading standardized reports to detect informa-
tion gaps and distortions.
206 T. Miron-Shatz et al.
Improving Health Literacy Skills
In this section, we discuss different ways to improve public comprehension
of health information. Our ideas are not comprehensive, but serve to highlight
areas which should be further explored.
Targeting Young People
The need to change existing educational school curricula to promote statis-
tical literacy throughout the population has been discussed elsewhere (e.g.,
Gaissmaier and Gigerenzer; Smith, this volume). Early statistical teaching
must overcome the assumptions that the mathematics of certainty (e.g., al-
gebra and geometry) is more important than the mathematics of uncertainty.
Martignon and Kurz-Milcke (2006) provide an encouraging study that shows
the effectiveness of early training: fourth graders were able to master com-
bined probabilities after participating in structured games using colorful cubes.
While structured education offers one method, another avenue might be
through computer and board games. Games appeal to young people and could
be an appropriate platform to turn something that appears dif cult to under-
stand into something that is assimilated naturally. For example, in the United
Kingdom, computer games designed to teach probabilities have utilized as-
pects of gambling and have proven to be very appealing to young people
(Spiegelhalter 2010). At a later age, visualization software such as Fathom
(Finzer and Erickson 2006) and TinkerPlots (Konold and Miller 2005) help
young people explore and manipulate data sets (Gar eld and Ben-Zvi 2007).
By starting with concrete representations of risks, children can accumulate
con dence in basic concepts and will less likely develop a math phobia when
more complex concepts are introduced at a later stage.
Although computer probability games for children exist already on sev-
eral Internet sites, the extent to which young people take advantage of these
games is unknown. Use of those designed explicitly for children, such as Club
Penguin (2010) or Webkinz (2010) provides a starting point to incorporate
modules to teach probability. These popular resources often have modules
geared toward spelling, have attractive designs, and award children points for
successful performance. Providing incentives to commercial companies to in-
tegrate relevant modules into resources widely used may be a feasible means
of reaching and capturing the attention of children.
Another example can be drawn from a curriculum of critical health lit-
eracy for secondary school students, age 17 years, successfully piloted by
Steckelberg et al. (2009). The curriculum’s objectives were to develop and
enhance statistical knowledge and competencies, to appraise medical infor-
mation critically, and to gain understanding of how medical information is
Barriers to Health Information and Building Solutions 207
From Patients to Doctors
Simple decision aids for interpreting the meaning of medical test results pro-
vide a possible means of educating patients. Human mediators (e.g., health
coaches who inform patients on how to pose questions about the bene ts and
risks of tests administered by their physicians) might also facilitate access to
Increasingly, patients’ interests are being represented on health care boards,
agencies, and institutions. Training courses for patient and consumer representa-
tives to cope with these roles have been successfully piloted (Berger et al. 2010).
Computer games that involve probability might help health care consum-
ers, medical students, and clinicians explain the bene ts and risks of various
tests and screening instruments to their patients. In addition, handheld devices
(e.g., touch phones) could be equipped with probability guides so that medical
students and clinicians can calculate key measures (e.g., base rates of diseases,
frequencies of medical conditions, and the absolute risk of having a disease
after testing positive) or access information to key questions.
A more direct route to medical students would be through the Medical
College Admission Test (MCAT), which is administered by the American
Association of Medical Colleges (AAMC). Since every prospective medical
student is required to take the MCATs, there is high motivation to perform
well. Thus, if probability, uncertainty, and other core concepts of statistical
literacy were integrated into the exam, students would be motivated to acquire
this competency. As with children’s computer games, the trick is getting people
to want to know.
Journalists exert a powerful in uence on how the public perceives health and
health care. Much of what people––including many physicians––know and
believe about medicine comes from the print and broadcast media. Yet jour-
nalism schools tend to teach everything except the understanding of numbers.
Journalists generally receive no training in how to interpret or present medical
research (Kees 2002). Medical journals communicate with the media through
press releases; ideally, they provide journalists with an opportunity to get their
facts right. Unfortunately, press releases suffer from many of the same prob-
lems noted above (Woloshin and Schwartz 2002). Some newspapers have be-
gun to promote correct and transparent reporting, and efforts are underway to
teach journalists how to understand what the numbers mean. As in Finland,
statistical training courses for journalists exist at several universities and are
conducted by some government agencies. Opportunities for statistical learning
remain rare in the media, however.
Editors play a central and powerful role in the presentation of new stories, at
least in traditional media outlets. Even a minimal amount of training for editors
208 T. Miron-Shatz et al.
(e.g., knowing which questions to ask about statistical procedures in health and
medical studies written by reporters) could improve general news coverage.
Another push to make journalists report more carefully about medical facts
could come from statistically literate observers. Internet sites in the United
States and Australia, for instance, rate the quality of medical reporters and
medical news in major news publications. One such site (Health News Review
2010) evaluates the quality of health reporting in major U.S. newspapers and
magazines. Although this form of watchdog effort has not yet been evaluated,
it seems plausible to assume that by intensifying the monitoring of press cover-
age of medical statistics, improved reporting will result.
Change from Within
It is dif cult to de ne the level at which change should be pursued. It is people
who understand information, but the problem neither starts nor ends at the in-
dividual level. Should the focus be on the patient and his/her maladies, history,
and choices? Or should the emphasis be on the policy maker who determines
funding, resource allocations, treatments, screenings, and bed numbers for
The best way to change a system is from within. Those who set health poli-
cy and fund health care are in the strongest position to ensure that more health
promotion and care is based rmly on evidence, that formats are standardized,
and that people are helped and encouraged to obtain, process, and understand
There is strong evidence to indicate that the amount of health care that peo-
ple receive is largely determined by the supply of health care institutions (i.e.,
“supplier-induced demand”) (Fisher et al. 2004). For example, people in Los
Angeles receive twice as much health care as people in Minneapolis simply
because there are twice as many health care providers in Los Angeles. Far from
producing bene t this leads to increased adverse outcomes Los Angeles: peo-
ple are more likely to suffer from errors that are common in the health system
because they come into contact with it more often. Wennberg (pers. comm.)
argues that improving health literacy among the people of Los Angeles is un-
likely to reduce the amount of care they receive. Yet providing his compelling
data to those who organize and fund health care could potentially reduce the
health care supply in Los Angeles and encourage care based on evidence.
Do We Need to Demonstrate the Benets of Health Literacy?
In a room full of psychologists and physicians who have been practicing shared
decision making for decades, the benets of health literacy appear obvious.
Yet, this skill is neither prevalent nor actively fostered by most major players
in the health and education arena. To promote health literacy, especially if this
Barriers to Health Information and Building Solutions 209
endeavor entails a major change in the way risks, bene ts, and probabilities
are conveyed in every report, we need to transmit the merits of health literacy
beyond the scope of its current advocates. Presently, two main justi cations
exist for promoting health literacy, representing two different viewpoints that
are not necessarily contradictory. We discuss these bene ts (ethical and eco-
nomic) in turn. The widespread exploitation of this lack of literacy through
biased report of evidence can also be seen as a moral issue (see Gigerenzer and
Gray, this volume).
Improving Health Literacy: An Ethical Imperative
Since the Age of Enlightenment, efforts have focused on educating citizens
for their personal and the greater societal good. It has become an ethical ob-
ligation to educate young members of society. Over time, general education-
al efforts have been applied to health issues: individuals’ knowledge about
diseases, risks to health, and ways to maintain a healthy lifestyle. Yet, health
literacy levels, as de ned previously, have generally been found to be insuf -
cient among large segments of the population. A recent report by the American
Medical Association (AMA) estimates that over 89 million American adults
have limited health literacy skills (Weiss 2007). This estimate is based on the
AMA’s interpretation of the results of a National Assessment of Adult Literacy
survey conducted in 2003, wherein AMA reclassi ed anyone “below basic” or
“basic” (Weiss 2007:10) health literacy as being “limited” and multiplied this
percentage by the number of adults in the population.
To address this remarkably low level of literacy, efforts to increase health
literacy have been advocated as a necessary condition for a better educated
population—one capable of making appropriate and informed health decisions
and engaging in recommended health behaviors. Increasing health literacy has
even been called an ethical imperative for health care professionals, who are
tasked with ensuring that individuals process and comprehend relevant public
health messages, treatment options, and recommended regimens (Gazmararian
et al. 2005; Woloshin and Schwartz 2002).
Improving Health Literacy: A Means of
Reducing Costs and Improving Care
In the present era of cost containment, it is questionable whether the ethical im-
perative is suf cient to remedy the dismal situation of low literacy levels and
to develop and sustain the resources for an effective health literacy campaign.
One argument put forth by the AMA is that individuals with limited health
literacy incur up to four times greater cost in unnecessary doctor visits and
hospital care, compared to individuals with average health literacy.
Increasing health literacy has been associated with a number of positive
outcomes: improved decision making, better understanding of disease and
210 T. Miron-Shatz et al.
treatment regimens, and adherence to prescribed treatment options. It is thus
possible that increased health literacy could translate to lower costs associated
with increased adherence and fewer unnecessary doctor visits and hospitaliza-
tions. Indeed, studies on the health outcomes of low-income individuals (con-
trolling for education, insurance, race/ethnicity, sex, language, insurance, de-
pressive symptoms, social support, diabetes education, treatment regimen, and
diabetes duration) have found that those with higher health literacy had better
health outcomes than those in the low literacy group (Schillinger et al. 2002).
However, the assessments of detailed costs are hampered by the fact that
multiple outcomes could be considered and standards for assessments do not
exist. The question therefore arises as to how outcomes can best be measured.
What Are the Most Appropriate Outcome Measures?
A number of outcomes have been associated with improved health literacy. For
example, adherence to diabetes treatment regimens (Mühlhauser and Berger
2002) increased after patients received an intervention designed to improve
understanding of the disease and self-management of its treatment. Other out-
comes frequently mentioned are the reduction of unnecessary care and its asso-
ciated monetary savings. Another measure, which resides between the ethical
imperative and economic bene t, is that of quality of life. If health literacy
is associated with reducing the severity of chronic diseases (e.g., diabetes or
hypertension), patients’ lives may become more bearable and acute outcomes
can be avoided.
Marketing Implications: Promoting Systemic Change
There is a strong sense that psychologists and health care professionals have
valuable health literacy tools and information which are underused. If we as-
sume that a need exists among stakeholders in the health domain for obtaining,
processing and understanding information, we can state that our products, as it
were, address a pressing need. That this need is seldom met, or even identi ed,
by stakeholders is a problem that could be solved through improved coordina-
tion and marketing.
Marketing is a word rarely used in the world of health literacy. However,
drug companies, diagnostic tool manufacturers, and other commercial players
use appealing strategies to promote their information. Often, their products do
not represent the totality of the evidence in the most readily comprehensible
fashion, and yet they are widely and efciently disseminated. Evidence-based
patient information and evidence-based information sources targeted at health
care professionals compete in the ideas market alongside these materials.
In marketing our insights, knowledge, and tools, we have several resources
at our deposal. The public and clinicians have considerable trust in academics
Barriers to Health Information and Building Solutions 211
and other health care professionals. Several strategies are available to us to
exploit this receptiveness. Just as individual case studies of patients may be
appealing, case studies of clinicians who had a positive experience around
improved health literacy may be helpful in marketing (e.g., positive experi-
ences in shared decision making, improved statistical literacy). As the media
is often drawn to stories of personal interest, we could attempt to create such
stories demonstrating, for example, how a city, a school, or a retirement home
achieved better health outcomes due to a literacy intervention. There are pub-
lication venues for these positive stories in “personal view” sections of profes-
sional journals, and these could be written or solicited by anyone in the eld.
For resources to be obtained, they need to be easily accessed. To date, how-
ever, there is no single information resource linking all the many available de-
cision aids and risk calculators. At the most basic level, a Wikipedia page could
be created, producing background references and a table of available tools. It
would be trivial––yet surprisingly novel––to install a piece of wiki software
on a server such as “decisionscommons.com” where trusted and registered
academics in the eld could update a central structured list of resources. The
same could readily be done for structured and updatable lists of educational
resources, probability games for children, and other resources.
A top-down approach also needs to be applied, as change cannot only come
from below. Lobbying politicians to improve the quality and transparency of
patient information provided by industry may bear fruit. Sites funded by the
state (e.g., NHS Choices 2010) also represent good examples of state-funded,
patient-facing, evidence-based information. Another strategy would be to offer
assistance in developing statistical features to add into existing online com-
munities and games. This will require funding and infrastructure but would be
extremely helpful in reaching audiences far beyond what conventional inter-
ventions can accomplish.
“Translational medicine” or the concept of knowledge translation (Davis et
al. 2003) is a relatively new principle which is also relevant here. Generally
characterized as taking research work “from the laboratory to the bedside,”
it improves the application of basic science research to clinical research and
practice (Lean et al. 2008). At its simplest level, it may be bene cial for all
who work in the eld to keep in mind that their work could be applied when
communicated in professional and lay media.
In all of these actions, there is a strong argument to have a single profes-
sional organization for those engaged in improving health literacy: one that
will re ect the interests of the community; act as a central contact point for cli-
nicians, academics, journalists, politicians, civil servants, and patients seeking
further information and expertise; coordinate lobbying; fund outreach activi-
ties and share information. Such an organization would also serve as a valu-
able resource for generating and sharing ideas on promoting the eld of health
literacy, in addition to this early and speculative list.
212 T. Miron-Shatz et al.
Our discussions bene ted from the input of Angela Coulter, Glyn Elwyn, Ralph Hertwig,
Marjukka Mäkelä, France Légaré, Albert G. Mulley, Jr., and Holger Schünemann.