The third panel presentations were from Marguerite Holloway, from the Columbia University Graduate School of Journalism, and Jessica Ancker, an assistant professor at the Center for Healthcare Informatics and Policy at Weill Cornell Medical College. Their presentations examined the role of the media in communicating health information to the public and the challenges faced by communicators in effectively communicating risk and uncertainty.
Marguerite Holloway, M.S.
Columbia University Graduate School of Journalism
Holloway said the issue of numeracy and health journalism is a vital one. A number of things can lead to confusion and mistakes in the coverage of science news, including misinterpretation of results, institutional spin, discarded caveats and context, time constraints, and the fixed beliefs of the audience. These problems are as familiar to audiences as they are to journalists. The issues of numeracy and medical reporting have been written about a great deal and there are excellent books, reports, and studies to consult, Holloway said.
Numeracy in science and medical reporting remains an issue of ongoing professional discussion and concern. In preparation for her talk, Holloway noted that she spoke with a manager at the Association of Healthcare
Journalists who confirmed that numeracy and improving health reporting remain areas of high priority for the organization and areas in which they frequently do training and outreach. Journalism, generally, not just in health and medical reporting, is relying increasingly on data analysis, numeracy, and statistical savvy. Holloway said that her presentation addresses the challenges posed to journalists by numbers in health reporting, strategies that journalists can or should use, and the ethical issues that can arise.
Holloway said she could find no definitive numbers on how many Americans get what proportion of their information about health from journalists. It is clear from responses to news stories that many people get and use health information presented by the media and that media outlets respond to consumer demand by presenting health stories, particularly on personal health. In this context getting the numbers wrong or creating hype has the potential to have significant consequences, Holloway said. People can make bad choices about care, treatments, or lifestyle, which can lead to poor decisions such as refusing vaccines for children. Media stories can give people false hope or great disappointment and no hope. For example, people can become desperate for a new cancer drug that they later learn is not available or has only been tested on animals or they can waste money on medications that are no different from cheaper ones already on the market. Holloway pointed out that media coverage can drive research and funding into areas that might not be as significant for public health as others and can cause people to lose trust in science and medicine.
Most journalists are aware of the impact their stories can have and take that responsibility seriously, Holloway said. Reporters face a number of challenges, however, some shared by society at large and others that are specific to the field of journalism. First is the baseline challenge that many, if not most, people have some trouble with numbers. If something can be numerically expressed, Holloway said, it carries with it a sense of authority and fact. As a result numbers are influential, and they can have long lives even if they are errors, being cited again and again and shaping public understanding of a topic. Yet although people respect numbers and attribute power to them, they want to engage with numbers as little as possible.
According to Holloway, this peculiar combination of skittishness and reverence becomes evident in the fact that numbers in media stories often exist side by side with basic mathematical mistakes. A 2012 examination of one daily newspaper found, for instance, that nearly half of the stories, a total of 536, published over the course of 1 month included or required some kind of mathematical information or calculation (Maier, 2012). The study also found that errors were prevalent in these and other stories. The author of the article identified 11 types of mistakes, including incorrect addition, misinterpretation of numbers, sensationalization using dramatic numbers, and unquestioning use of figures. Many errors were of elementary
math, errors that common sense could easily catch, but few reporters or editors had turned their attention to the numbers. Basic prevalent attitudes toward math are one fundamental challenge, Holloway added.
A second major challenge is a lack of understanding of scale and scalability, Holloway said. Researchers such as Gail Jones at North Carolina State University have shown that appreciating scale is a key to scientific thinking. Problems of scale as it relates to dimension can extend to people’s problems interpreting and contextualizing numerical health information. Numbers of cases or rates of disease are difficult to scale up or down in an accurate or meaningful way. A number can have one meaning or implication when considering an individual’s personal circle of friends and acquaintances or community, and another with regard to the U.S. or global population. Few people are able to move fluidly up and down those scales to see the personal and the big picture accurately, Holloway said.
A third challenge relates to statistical thinking or understanding probabilities, ranges, risks, and ratios. This represents a mathematical skill or habit of mind that can be particularly challenging for the press and public alike. A 2002 study surveyed 165 journalists and found that 84 percent of the reporters, 96 of the respondents, had never been trained in understanding health statistics (Voss, 2002). The importance of correctly interpreting statistics and the difference it can make is captured in an essay written in 1985 by Stephen Jay Gould titled “The Median Isn’t the Message,” Holloway said (Gould, 1985). The essay is widely known and is very helpful in thinking about patients and numeracy. Gould writes of learning of his cancer diagnosis and reading in the medical literature that patients with this cancer have a median mortality of 8 months. He notes that most people would take this to mean that they had only 8 months to live. Gould writes that this conclusion must be avoided because it is untrue, and attitude matters a great deal in approaching serious illness. He goes on to explain that one must understand that variation is the reality and mean and median are abstractions. He wonders whether he might be in the groups of patients who will live longer than 8 months, which he learns he is. He then learns that the distribution is right skewed with a long tail and he may live years beyond the median, which he does. Holloway said that Gould explains the statistics so clearly that his essay is a model of how to explain commonly used statistical concepts in terms that are easily understood. The essay also provides a reconciliation of what Gould calls the unfortunate and invalid separation between heart and mind or feeling and intellect. As an example of this, Holloway said studies about public perception of climate change information illustrate the difference between the experiential processing system and the analytical processing system. The tension between these two ways of processing information is another challenge for reporters covering health.
Some challenges faced by reporters, however, are specific to journalism. Many media outlets are losing money and decreasing numbers of staff, and many reporters have to do more in less time, Holloway said. The profession has always been driven by deadlines and intense competition, but now reporters and writers in many places must be constantly producing content. There is often less time to be reflective and little incentive to wait to publish a story to analyze the numbers, the implications, and the context.
The academic journal culture is an additional challenge for the press. Holloway said that science, health, and medical information largely come from studies that are embargoed to build a “news peg” to create buzz and often revenue for the journals themselves. Daily journalists are embedded in a cycle that is nearly impossible to escape from while keeping pace with the competition. A reporter on a deadline often does not have the time to examine a number too closely and there is little opportunity to think historically under those conditions or take time to understand the information in context. This culture does not favor an appreciation of medicine or science as incremental and uncertain, and stories with too many caveats are not “newsy,” Holloway said. In these conditions, most reporters do the best they can and set aside questions for more in-depth examination for longer stories or perhaps a trend story. These are more analytical reflective pieces where issues are examined more deeply and with nuance and where numeracy is handled in much better or clearer ways.
Holloway presented several strategies that journalists use or should use and that journalism students are taught. Some of the strategies are not specific to numeracy issues, but are generalizable ways of thinking about approaches to reporting on science, health, and medicine. Journalists should be familiar with various types of studies and their limitations, Holloway said. Reporters should not misrepresent or overinterpret the significance or implications of findings and should have a roster of basic questions to ask about any study. The reporting about how research works must be transparent and present the strengths and weaknesses of different types of studies. It should also look beyond the one study to reviews or meta-analyses that may have been done in the field. Journalists should find a statistician they can rely on and turn to for advice and guidance, Holloway said. As news organizations adapt to the new environment, there is more collaboration with statisticians and data experts in newsrooms. Holloway recommended that numbers be presented with transparency and in a variety of ways, and that journalists give both relative and absolute risk. According to Holloway this practice is not as routine as it should be; a review of 500 news stories in 2008 found that only 18 percent gave both relative and absolute risk (Schwitzer, 2008).
An example can be found in the review of a story about suicide rates by Paul Raeburn, who assesses the press coverage of science for the Knight
Science Journalism Tracker (Raeburn, 2013). The original story reported that the suicide rate had increased by 30 percent in those ages 35 to 64 between 1999 and 2010, from 13.7 deaths per 100,000 to 17.6 deaths per 100,000. As Raeburn notes in his review, but the original story failed to do, a 30 percent increase sounds large, but the absolute numbers are much smaller. The increase amounts to approximately 4 more people per 100,000 and suicide remains a rare event for this age group, occurring in much less than 1 percent.
Holloway noted that it is important that journalists maintain skepticism about the study and the numbers being reported. This can mean thinking about the history of a figure and not taking anything for granted. A good example of investigating a widely used statistic can be found in “How Long Can You Wait to Have a Baby?” in The Atlantic, which looks at the assumptions underlying fertility (Twenge, 2013). The author of the story noted that most sources reported that one-third of women between 35 and 39 would not be able to get pregnant within 1 year of trying and that women in their late 30s had a 30 percent chance of never having a child. After tracking down the source of these numbers, the author discovered that they are from an analysis of French birth records from between 1670 and 1830. The author notes that there are not many well-designed studies of female age and natural fertility that include women born in the 20th century, but those that do have different, and more optimistic, results.
Holloway said it is also important to blend the statistics and stories of people in a compelling way in order to capture both the data and the human experience. Many journalists do this beautifully when they have time and some space and support, she said. She added that journalists should repeat as often as possible that correlation is not causation.
Holloway concluded by noting that everyone tries to force information, including numbers, into the frame of the mental model they already have. It is difficult to absorb information that runs counter to expectations. When journalists tell stories of individual experience that are well supported by and help illustrate numerical data, the complexity and nuance of the issue can come to life. In this way numbers can have a transformative effect. Engaging with numbers, getting them right, understanding their implications, and then presenting them in the public realm is for the greater good and benefit for society.
Jessica S. Ancker, Ph.D., M.P.H.
Center for Healthcare Informatics and Policy,
Weill Cornell Medical College
Ancker’s presentation explored the ways in which numeracy issues relate to shared decision making and the concept of explaining risks and uncertainty. She began with an example of a patient decision that generated a number of headlines early in 2013, when Angelina Jolie publicly disclosed that she had tested positive for mutation at BRCA1 and chose a prophylactic bilateral mastectomy as a preventive measure. Jolie’s decision received a great deal of attention because of her celebrity and the radical nature of her choice, Ancker said. It is instructive to examine the editorial Jolie wrote explaining her decision. In the editorial, Jolie discusses her family history of breast and ovarian cancer and her desire for more information, which led to seeking out the BRCA tests. When Jolie tested positive, she was given a lifetime risk of developing breast cancer of more than 80 percent, which is obviously very high. Jolie writes that she explored her options and weighed the advantages and disadvantages. Ancker said the reason she chose the Jolie example and what makes it interesting is that it conforms to societal expectations for shared and informed decision making. In this situation the patient gathered relevant information, understood the risks and the options, and engaged in the decision-making process to the extent that she was comfortable.
Ancker noted that there is variability in medical decision making, with some patients choosing to have the doctor take the lead role while other patients would rather be more engaged. Generally, the Jolie example is what is meant by informed decision making in health; informed decision making conducted collaboratively between the patient and a physician is known as shared decision making. Such shared decision making is difficult to achieve in a population with low numeracy skills, especially if the expectation is that patients must really understand risk as part of this shared decision making.
Risk in the medical realm is usually thought of as “epidemiologic risk,” which is the probability of developing a certain disease or the number of people who contract a disease over a specific time period over the total number at risk. Many people are also familiar with an economic form of risk, which is the “probability multiplied by utility” concept, Ancker said. As an example, most people in the room would probably agree that a 1 percent risk of developing breast cancer is worse than a 1 percent risk of
developing a cold because of the higher “disutility” placed on breast cancer rather than a cold, even when the epidemiologic risks are equivalent. She noted that within the discussion of risk, there is potential for miscommunication with patients if they are thinking about the economic type of risk as opposed to the epidemiologic type of risk.
Uncertainty is also a problem in situations in which it is reasonable to assume there is some risk, but its magnitude is unknown, Ancker said. She noted that a classic example of this is a newly approved drug for which the long-term safety profile is unknown. Another example is confidence intervals where there is a range of plausible estimates for what the risk might be.
Ancker focused her presentation on epidemiologic risks because, she said, in the context of shared decision making the health care professional usually communicates the epidemiologic risks to the patient. At that point the patient is expected to talk about his or her personal utilities, or the personal values the patient holds. This iterative discussion is shared decision making.
According to Ancker, much of the risk communication literature comes from the public health realm in which persuasive communication is the norm and is considered ethically appropriate. In the public health context, informing an individual of his or her personal risk of developing lung cancer is not the goal. The goal is to persuade that individual to quit smoking. This differs from the context of shared decision making, which is a narrower realm. In shared decision making, the goal is to help people have a better understanding of their own risk for a certain disease or condition. As a result, a great deal of the public health risk communication literature focuses on nonquantitative ways of expressing risk, on framing, or on fear appeals and not quantitative risk communication. In shared decision making, however, there is a strong emphasis on ensuring that patients understand their quantitative risk. There is fairly good evidence that people with lower numeracy skills are more reluctant to engage in shared decision making, Ancker said.
Competent use of quantitative information is not solely dependent on the patient’s skills, Ancker said. Numeracy is often perceived as a quality, skill, or ability that the patient brings to the situation along with previous knowledge and perception. However, the person providing the information brings a set of skills to the situation as well. This person may be a good communicator or a poor communicator. The competent use of the quantitative information comes from the interaction, not solely from the patient skills. Frequently the discussion is supported or informed by a document, website, or other artifact that contains information. These supports may be designed well or poorly; they may help compensate for low numeracy or increase the numeracy burden on the patient. It is important to remember that the patient is making sense of the information in the context of a social
network, a physical environment that may carry risks or no risks, and an information environment that contains news stories and television, radio, and online sources.
Keeping the complexity of the interaction in mind, what are the options for explaining risks? Broadly speaking, information providers can use words, numbers, and pictures. Ancker noted that she would not make a distinction between written and oral or spoken presentation because there has not been sufficient high-quality research about differences between spoken and written communication of numbers.
Evidence indicates that people at the lower levels of numeracy say that they prefer verbal descriptions of risks and that they trust information more when it comes packaged in that format, Ancker said. Words and phrases like “big risk,” “small risk,” and “common” or “uncommon” are familiar to patients and convey the affective impact that patients say they want. But the disadvantage of this is that such words are not very specific. People also tend to overestimate the number associated with a particular risk word, she said. As a result, it is difficult to use words alone to make good comparisons. For example, “Is a very small risk better or worse than a rare risk?” There is no way to know without more information.
In shared decision making, there is a strong preference for providing information numerically, Ancker said. For example, the International Patient Decision Aid consensus1 states that patients should be provided with numbers. One option is to present information as percentages. Percentages are generally familiar and are independent of sample size. Two percent of a small group is the same thing as 2 percent of a large group. There can be problems with presenting information in this way, however. Often people do not know how to manipulate percentages and, particularly at lower numeracy levels, do not know how to calculate them. There is good evidence that people perceive them as abstract and may, therefore, feel the information does not apply to them. The difficulty in manipulating percentages is a longstanding problem, Ancker said. She gave the example of providing the risk level and the information that an intervention will reduce that risk by 30 percent. Many times a patient cannot perform the calculation needed to understand what this means for his or her decision making.
Some advocate providing information in terms of frequencies rather than percentages, changing 23 percent to 23 in 100. This is the standard of communication for such factors as genetic risk. Good evidence shows that people perceive this as vivid and personal, said Ancker. Yet this approach has disadvantages. Among the less numerate, this approach may inflate the perceived risk compared to percentages. In addition, there are two other
1 For more information see Elwyn et al. (2006), http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1553508 (accessed October 30, 2013).
problems: denominator neglect and denominator confusion. “Denominator neglect” is the standard term for a longstanding, well-known phenomenon where people, particularly the less numerate, may not recognize that two risks with different denominators are equivalent (e.g., 23 in 100 versus 230 in 1,000) because they focus on the numerators and compare only those two numbers.
“Denominator confusion” is a term coined by Ancker to describe a common misconception that arises when risks are presented to people with different denominators. According to Ancker, there is good evidence that people with lower educational attainment are likely to focus on the numerator and not recognize that 1 in 20,000 is in fact quite a bit smaller than 1 in 5,000. As a result, there has been some promotion of the idea that information should always be presented with the same set of denominators (e.g., a “natural frequencies” format). Presenting the comparison as 4 in 20,000 versus 1 in 20,000 would allow people to make that comparison more easily.
The third option that information providers have is graphics. People like graphics because they are visually interesting and attractive, and they exploit not only learned skills, but also automated visual perception, or things that we do not have to learn. An example is determining which of two bars on a page is larger is an automated task, Ancker said. There is not much, if any, learning involved. As long as the two bars are on the same X-axis or the same horizon, people can automatically tell which one is larger. Judging how much larger one is than the other is also fairly automated. Numeracy skills have something to do with how well an individual can verbalize the size difference, but understanding the differential happens at an automatic level. The learned part is having an understanding of what the two bars mean, knowing the significance of the X axis and Y axis, and how the information applies to the individual. This knowledge is not independent of learned skills at all, Ancker said.
In her review of the graphical literature, Ancker said she identified some core principles that apply to all types of graphics (Ancker and Kaufman, 2007). First, whether the part-to-whole relationship is visible and easily identified is critical to how well people can judge the graphic. Figure 4-1 is an example of a graphic in which the part-to-whole relationship is not visible. The risk represented on the left is 10 percent and on the right, 7 percent. But the graph lacks the context of the entire 100 percent. Extending the Y axis clarifies how the 10 and 7 percent relate to the whole. The graphic on the left inflates the apparent difference and the graphic on the right somewhat minimizes it or at least places it in context. Another example from icon graphics illustrates the point further. Individual risk is a certain number of icons while average risk is another number of icons, but once placed within the context of a larger number, the part-to-whole
FIGURE 4-1 Judgment is affected by whether the part-to-whole relationship is visible.
SOURCE: Ancker, 2013.
relationship can be shown. The part-to-whole relationship has to be easily visible before people can do this on an automated level, Ancker said.
Figure 4-2 shows two ways of expressing risk where the part-to-whole relationship is visible. Ancker said some good qualitative research suggests that the arrangement on the left does a better job of expressing the idea that anybody can be at risk, that risk is haphazard and random, and that anybody might be affected (Ancker et al., 2011a,b). By contrast, the figure on the right seems more controllable and tidy, as though risk is concen-
FIGURE 4-2 Graphic representation of risk in a population.
SOURCE: Ancker, 2013.
trated in the population and not distributed throughout. This example shows there are different dimensions of communication effectiveness that must be considered.
The second principle, Ancker said, is that judgments are most accurate when only one dimension varies at a time. In Figure 4-3 there are two bars that change in height and two icon graphics that change in area, height, and width. In general, people are not able to easily judge the difference between two graphics that differ in more than one way. Their perception skills decrease further when circles are used, Ancker added. Pie charts can be a good way of demonstrating part-to-whole relationships, but comparing pie charts of different sizes can lead to confusion.
Ancker noted that three types of graphics have been well studied (see Figure 4-4). Icon graphics, in the form of either smiley faces or small stick figures, are widely used and well understood. There is a very clear visual analogy with the risk information. Icon graphics exploit people’s ability to visually process sizes at an automatic level and rely somewhat less on learned information. These graphics are well accepted by people with low numeracy and fulfill the part-to-whole principle requirement.
Bar charts are also fairly well accepted and are familiar to many people,
FIGURE 4-3 Judgments are most accurate when only one dimension varies.
SOURCE: Ancker, 2013.
FIGURE 4-4 Three common types of graphics.
SOURCE: Ancker, 2013. Horizontal risk scale with magnifier adapted from Woloshin et al., 2000.
Ancker said. She noted that there is some evidence that patients think of them as more abstract and are less able to relate to the information personally when it is in bar chart form. She added that there is also evidence that patients prefer graphics with fewer visual elements in them because the graphics look simpler. Information in a bar graph may seem easier to understand just because there are fewer elements to the graphic.
Finally, risk ladders or risk scales are a type of graphic that is not seen much in health risk communication, but is used more widely in environmental risk communication. The advantage of risk ladders or risk scales is that only one dimension varies at a time, Ancker said. People can easily compare positions along that line. This type of graphic permits the addition of comparative information or some sort of action threshold that indicates that once a risk reaches a certain level, a certain action becomes important or necessary or the risk has exceeded the average for a particular population.
Given that different types of graphics have different advantages and
disadvantages, it is important to determine what patients want and find acceptable, Ancker said. Unfortunately, the research has not yielded many clear answers for what patients want. Patients sometimes prefer the more visually simple graphics even if they contain less information. But patients do not necessarily answer comprehension questions better when given information with the graphic type they prefer, and neither do doctors. There is also the question of whether it is better for patients to layer information or have multiple representations of the same information. It is important to guard against the possibility of cognitive overload in the patient instead of better understanding.
Ancker conducted research in which she used an innovative game-like interaction to educate patients about their risk. Instead of merely telling people what their risk was, the game encouraged individuals to click on an icon graphic to uncover which figures in the graphic had the disease. Results showed that the more often people clicked on the icons looking for one with the disease, the more anxious they felt, which was exactly the opposite of what she had hypothesized. The game was attractive and people liked it, but it didn’t impart the information in the way that the designers thought it would. She cautioned that this shows that innovative ways of communicating with patients should be studied carefully before they are employed.
Ancker noted that there is another law in addition to the Patient Protection and Affordable Care Act (ACA) of 2010 that is impacting the health care system. The Health Information Technology for Economic and Clinical Health (HITECH) Act of 2009 may not be as obvious to people who are not doctors, but it is changing the practice of medicine. The HITECH Act creates an electronic health record (EHR) incentive program in which doctors and hospitals can receive financial incentives for computerizing their medical records. If they do not make the change, they receive penalties. Over the past 5 years, there has been a massive increase in the number of health care institutions that use EHRs. Many are unaware that there is a mandate that in order to collect the financial incentives, doctors and hospitals must make electronic medical data directly available to patients. Ancker said she believes this is a nationwide experiment in health communication. Patients will be able to access the system and see their medical records, including diagnoses, medication lists, and details of lab tests and radiology reports. There is some concern, however, that this could lead to patient anxiety because they will not understand the meaning of the information they access.
Ancker concluded by listing the areas where she would like to see additional research. The first is the area of novel communication modalities that are available online, such as games and interactive tools. It is not known whether using these types of tools in health and risk communication is beneficial or not. The second area is the experiment in releasing medical
records directly to patients, which raises questions such as whether patients will understand the information, how the information will affect shared decision making and risk perception, and what can be done to make this information more useful to patients.
Moderator: Paul Schyve
Robert Logan from the National Library of Medicine commented that two important points might get lost in the discussion. First, there is serious research about public health information, the public understanding of science, and news coverage of science and health. A number of journals publish this research, such as the Journal of Health Communication, Science Communication, and Journalism and Mass Communication Quarterly. Not all of the journals that publish science and health communication research are in PubMed because they also publish a great deal of other social science content that is unrelated to health. Some of content can be accessed easily, but some cannot. Logan said he hopes that PubMed will be able to extend its content to include more of these journals in the future so that health researchers will have better access to this field of study.
The second point, Logan said, is that there are organizations that undertake serious criticism and analysis of health news reporting and how journalists communicate research data to readers. For example, the website Holloway mentioned in her talk, HealthNewsReview.org, offers critiques of the reporting from major news organizations on a daily basis. Logan also commented that the Association of Health Care Journalists engages in a great deal of professional development in the field and encouraged those present to visit the Association’s website, healthjournalism.org. He added that the Columbia Journalism Review also has excellent criticism of how the news media covers health insurance–related issues.
Rima Rudd, roundtable member, noted that the presentations confirmed the notion that journalism is public health’s best partner and that training journalists in how to communicate science and health concepts is vital. Rudd said that mathematical terms are often taken for granted and that more attention should be paid to explaining terms such as “normal,” “range,” and “risk.” She noted that individuals and cultures react differently to the word “risk” and related an anecdote from a colleague that in certain Native American cultures, risk was assumed to convey inevitability. Efforts should not just be about communicating numbers, Rudd said, but also concepts. Rudd also recommended a tool developed by Mosenthal and Kirsch (1998) called the PMOSE-IKirsch assessment tool that rates the com-
plexity of documents and document elements, including charts and graphs.2 Ancker replied that the PMOSE-IKirsch tool is a great resource. She added that every component of health communication, including the clinician, the patient, and the document or materials used, should be included in determining ways to make communication better because miscommunication arises from a problem in the interaction between parties, not just from a patient’s inability to understand. Rudd noted that the idea of health literacy arising from interactions rather than being embedded in an individual is a central point of the Institute of Medicine’s 2004 report on health literacy.3
Patrick McGarry, roundtable member, asked whether any studies had been done on people’s perceptions of case studies versus population-based statistics. He noted that although suicide is low prevalence, it is a very real problem. Journalists must be careful how they present the data and what conclusions they draw for their readers. Holloway answered that she was unaware of any research that looks specifically at case studies versus population studies and its impact on media coverage. There is always a tension in journalism between focusing on personal stories versus population-level data. Holloway said that when a journalist can take a personal story and relate it to the larger population, it has more effect, but that journalists must be careful because sometimes using people’s personal stories can feel exploitive to readers.
Susan Pisano, another roundtable member, commented that medical journals assume that their audience consists only of practicing physicians, yet members of the public are reading these journals more and more often. No effort is made to translate these studies into something that would be easier for journalists to write about or for the average person to understand. Holloway responded that some medical journals think about the members of their audience that are not physicians. They provide video clips, press releases, and other information. Sometimes these help journalists to report more accurately, but they can also be misleading and manipulative. Journalists should be careful about relying too much on a single source that may be advancing a particular interest. The time demands on journalists can lead to them relying too much on press releases rather than doing their own reporting. They need to be careful about that, she said.
Ruth Parker, roundtable member, pointed out that the field of health literacy is somewhat nascent, and that she believes health literacy is entering its second generation. This has offered a chance for reflection in the field and an opportunity to chart the course for the next 20 years. Parker
2 The tool can be found online at http://www.hsph.harvard.edu/healthliteracy/files/2012/09/pmose.pdf (accessed November 3, 2013).
said that health communicators and people in the media are natural allies and asked if the speakers had any thoughts or suggestions on how to forge stronger ties between the two fields. Ancker answered that in some ways, health communicators and journalists are not natural allies. For example, journalists do not focus on literacy differences within their readership. Journalists tend to assume that their readers are all at the same level, she said. Journalists are interested in getting information to the public, but do not see any additional level of explanation or translation as being within their role. In addition, Ancker said, journalists are often skeptical about their sources even when those sources are doctors or public health authorities. To do a good job, journalists have to stand a little apart and avoid the risk of being seen as a spokesperson for a particular view. Ancker said she would encourage more communication between the two fields, but that their perspectives are very different. Holloway agreed and added there is some opportunity to work with professional associations and strengthen training in health communication for journalists.
As a follow-up, Parker gave the example of the upcoming enrollment period for the health care insurance marketplaces opening under the ACA. She noted that there are people who are interested in communicating to the public about how to use the exchanges in ways that are understandable and actionable. Journalists are also interested in providing accurate information to the public. Parker asked how the two groups could work together to provide accurate and usable information. Holloway answered that in that example, the best way forward would be for the health communicators to contact individual journalists and publications. But she cautioned that the journalists would not see it as a collaboration, but rather information provided to them so they could report on it independently.
Paul Schyve, roundtable member, commented that in Chicago public officials have turned to investigative reporters to advise them on oversight issues. The reasoning behind this is that reporters are trained to be skeptical and determine the validity of information for themselves rather than relying on the assurances of others. Schyve added that reporters and health communicators could learn from each other without necessarily collaborating on work.
Cindy Brach, another roundtable member, said she was interested in Ancker’s point that people do not necessarily better understand the graphics they prefer. The research on this topic is mixed. Generally people prefer simpler materials, but there is not strong evidence that this improves comprehension. She asked if any research shows that presenting information a certain way will improve comprehension. Ancker answered that there is a concept within informatics called “task technology fit,” which is that in human factors in informatics, there is no single technology that solves all problems. A technology’s usefulness is dependent on the context of the task.
She added that the research literature on graphics is confounded by the fact that different people are asking subjects to do different tasks, and thus getting different results. For example, if a researcher asks people to compare icon graphs to bar charts, but only one of the graphs for comparison makes the part-to-whole relationship visible, then the research results are unreliable. In addition, there is another layer of variability, which is that familiarity has an effect on comprehension. This means there is a sociological element to understanding graphics because some types of graphics will be familiar to different audiences at different times. Researchers are making inferences based on studies about how people are going to relate to electronic information, but that relationship will change over time.
Roundtable member Steven Rush asked about the effectiveness of “infographics” for presenting information visually in the news or for providing health information and reminders. Ancker answered that there is a growing emphasis on visual communication and that, in many ways, this is an improvement over previous methods. The effectiveness of a specific type of communication or piece of information would depend on the context both in terms of the communication itself and the audience. Rush then asked if there is a way to calculate the numeracy burden of a specific piece of information or material. Ancker said she is not aware of a way to measure the demands of a document or education material, but she would be interested in developing one and in shifting the focus away from measuring the skills of the patient. Rudd added that doctoral students in her program had adapted Apter’s hierarchy for numerical information by adding numbers to produce a scale that helps determine the numeric burden of information. This method is in its early stages and has not been tested, but it is a first step toward developing a tool that measures numeracy burden. Holloway added that within journalism, there is a growing emphasis on data visualization, and a great deal of work is being done on how to best present various types of information so they are easily understood by the reader.
Benard Dreyer, roundtable member, asked the presenters about the best way to communicate risk and risk reduction to patients. Ancker answered that showing the part-to-whole relationship is an important part of communicating risk and risk reduction because it gives people context. She also noted that providers might have different goals for the communication depending on the situation, and that this would affect how information was presented. For example, a clinician who is trying to get a patient to quit smoking would present information differently than one who was trying to help a patient choose between two therapeutic options. Dreyer commented that a good example of the difficulty of risk communication is vaccines. Because most of the diseases that are vaccinated against have been all but eradicated, the absolute risk reduction from immunizations is quite small. The relative risk reduction, however, is very high. Dreyer said
that if he focuses on absolute risk reduction, then vaccination might not seem worth it. Yet if people do not take the vaccines, these diseases could reappear. Ancker replied that there are a number of issues contained in that example. This might be an area in which non-quantitative risk communication is more effective. Quantitative risk communication is not the only tool available and is not always the most appropriate tool.
Schyve pointed out that this discussion is another example of the ethical ramifications of many of these issues. The ethics of a specific situation and the goals must be considered. George Isham, roundtable chair, asked whether it would be ethical for a professional in a situation where the vast number of patients cannot evaluate the information to consciously present information in a way calculated to persuade rather than inform. He reminded the group of the controversy surrounding the change in the recommendations regarding mammograms. In that case the change in absolute risk was very small, but many women feel that risk on a personal level. Isham concluded that there are serious ethical issues around the topic of numeracy and health literacy and the responsibility of professionals to communicate ethically, and that those issues should be explored. Ancker agreed that there are ethical issues involved in whether a health communicator’s job is to persuade or merely give information. She stressed that the issue is highly context specific. For example, society and the public health community are very comfortable with using scare appeals to persuade people to quit smoking. Both would feel differently about attempting to persuade people to get more X-rays or take more prescription drugs. She noted that professional communities will nearly always have more information than the general public and that they will always have some control over how that information is communicated.
Schyve ended the discussion by noting that these ethical issues will require a great deal more discussion and analysis in the future.
Ancker, J. 2013. Issues and challenges in the era of shared decision making: Explaining risk and uncertainty. PowerPoint presentation, Institute of Medicine Workshop on Health Literacy and Numeracy, Washington, DC, July 18.
Ancker, J. S., and D. Kaufman. 2007. Rethinking health numeracy: A multidisciplinary literature review. Journal of the American Medical Informatics Association 14(6):713-721.
Ancker, J. S., E. U. Weber, and R. Kukafka. 2011a. Effects of game-like interactive graphics on risk perceptions and decisions. Medical Decision Making 31:130.
Ancker, J. S., E. U. Weber, and R. Kukafka. 2011b. Effects of arrangement of stick figures on estimate of proportion in risk graphics. Medical Decision Making 31:143.
Elwyn, G., A. O’Connor, D. Stacey, R. Volk, A. Edwards, A. Coulter, R. Thomson, A. Barratt, M. Barry, S. Bernstein, P. Butow, A. Clarke, V. Entwistle, D. Feldman-Stewart, M. Holmes-Rovner, H. Llewellyn-Thomas, N. Moumjid, A. Mulley, C. Ruland, K. Sepucha, A. Sykes, T. Whelan, and The International Patient Decision Aids Standards (IPDAS) Collaboration. 2006. Developing a quality criteria framework for patient decision aids: Online international Delphi Consensus process. British Medical Journal 333(7565):417.
Gould, S. J. 1985. The median isn’t the message. Discover, June.
Maier, S. R. 2012. Numbers in the news: A mathematics audit of a daily newspaper. Journalism Studies 3(4):507-519.
Mosenthal, P., and I. Kirsch. 1998. A new measure for assessing document complexity: The PMOSE/IKirsch document readability formula. Journal of Adolescent and Adult Literacy 41:638-657.
Raeburn, P. 2013. Shaky ground: Speculating on the causes of suicide. Knight Science Journalism Tracker, May 6.
Schwitzer, G. 2008. How do U.S. journalists cover treatments, tests, products, and procedures?: An evaluation of 500 stories. PLoS Medicine 5(5):e95.
Twenge, J. 2013. How long can you wait to have a baby? The Atlantic, July/August.
Voss, M. 2002. Checking the pulse: Midwestern reporters’ opinions on their ability to report health care news. American Journal of Public Health 92(7):1158-1160.
Woloshin, S., L. M. Schwartz, S. Byram, B. Fischhoff, and H. G. Welch. 2000. A new scale for assessing perceptions of chance: A validation study. Medical Decision Making 20(3):298-307.