Ellen Peters, Louise Meilleur, and Mary Kate Tompkins, Psychology Department, Ohio State University
Numbers are used to instruct, inform, and give meaning to information in order to help us make better judgments and healthier choices in our everyday lives. However, research has demonstrated that not all people can understand and use numbers effectively. In particular, people differ in numeracy. Among uninsured adults, we estimated that 28.8 percent are at a Below Basic level of numeracy, 33.4 percent are at a Basic level, 29.3 percent are at an Intermediate level, and only 8.6 percent are at a Proficient level of numeric literacy. Numeracy skills needed to select a health plan, choose treatments, and understand medication instructions include education-based skills and emergent decision-based abilities. We estimate that the skills needed to make many complex, informed health decisions (e.g., management of chronic diseases) require a Proficient level of numeric literacy, given how numeric information is often provided. However, if health information providers present information to patients and consumers in an evidence-based manner, a greater proportion of the population will be successful in making informed health and health-related decisions. We identify five main communication themes and discuss evidence-based strategies for communication under each theme.
1Paper commissioned by the Roundtable on Health Literacy, Institute of Medicine.
Numbers are used to instruct, inform, and give meaning to information in order to help us make better judgments and healthier choices in our everyday lives. However, research has demonstrated that not all people can understand and use numbers effectively. In particular, people differ in numeracy. Numeracy has been variously defined as the ability to use basic probability and mathematical concepts (Peters et al., 2006b) and as “the degree to which individuals can obtain, process, and understand the basic [quantitative] health information and services they need to make appropriate health decisions” (Ratzan and Parker, 2000, p. vi). Berkman et al. (2011, p. 1) further describe the concept of health numeracy as representing “a constellation of skills necessary to function effectively in the health care environment and act appropriately on health care information.” Even highly educated individuals can be innumerate (Lipkus et al., 2001).
Previous reports have focused on what is known about the relation of numeracy to health outcomes and disparities (Berkman et al., 2011). With so many Americans lacking basic numeracy skills, it is important to know whether and how numeracy influences health outcomes and health disparities. Berkman et al. (2011) conducted a systematic review of numeracy. They concluded that the strength of current evidence was insufficient with respect to the relation of numeracy to health outcomes such as knowledge, risk perception accuracy, and accurate interpretation of health information. Numeracy, however, did appear to mediate some health disparities (e.g., between race and levels of hemoglobin A1c and between gender and HIV medication management capacity), although the strength of evidence was low. Conclusions could not be drawn about the relation of numeracy to use of health care services. Numeracy does appear to be more highly correlated with health outcomes than is health literacy, although possible ceiling effects on health literacy could have clouded the health literacy effects.
In the present commissioned paper, our assignment was to consider the following statement of task: “With the implementation of health care reform, there will be an influx of previously uncovered individuals who have limited knowledge, understanding, and ability to navigate the health care choices available. Of particular importance will be numeracy skills needed to make informed choices about which health plan best meets individual needs, how to make informed treatment decisions (e.g., X treatment has a 5 percent greater risk than Y), and understanding medication instructions. The roundtable will hold a meeting July 18, 2013, in Washington, DC, to explore such issues.”
This commissioned paper addresses three questions:
- What does research show about people’s numeracy skill levels?
- What kinds of numeracy skills are needed to select a health plan, choose treatments, and understand medication instructions?
- How can providers communicate with those with low numeracy skills?
QUESTION 1: WHAT DOES RESEARCH SHOW ABOUT PEOPLE’S NUMERACY SKILL LEVELS?
Numeracy can be assessed with objective measures (e.g., “If person A’s chance of getting a disease is 1 in 100 in 10 years and person B’s risk is double that of A, what is B’s risk?”) (Cokely et al., 2012; Lipkus et al., 2001; Weller et al., 2013) and subjective measures (e.g., “How good are you at working with fractions?”) (Fagerlin et al., 2007b). There are also general health numeracy measures, such as the Numeracy Understanding in Medicine instrument (NUMi) (Shapira et al., 2012) and various numeracy measures specific to health domains such as asthma, diabetes, and anticoagulation control (Apter et al., 2006; Estrada et al., 2004; Huizinga et al., 2008). Other studies have simply tallied how well individuals can do specific health-related numeric tasks. For example, in an online survey representative of the U.S. population, 79 percent of parents claimed to have seen a growth chart before, and most think they understand them well (Ben-Joseph et al., 2009). However, when provided with multiple-choice questions and answers, only 64 percent could identify a child’s weight when shown a plotted point on a growth chart and up to 77 percent misinterpreted charts that included both height and weight measurements. Like other innumeracy-related health examples, this may be important because parents may use their (inaccurate) understanding to guide related health decisions for their children.
As suggested above, Americans have limited numeracy skills. A recent probabilistic sample of Americans answered fewer than two-thirds of simple statistical numeracy questions correctly (Galesic and Garcia-Retamero, 2010). Even for the easiest question (“If the chance of getting a disease is 10 percent, how many people would be expected to get the disease out of 1,000?”), 17 percent answered incorrectly. For the most difficult item (“In the Daily Times Sweepstakes, the chance of winning a car is 1 in 1,000. What percentage of tickets for the Daily Times Sweepstakes win a car?”), participants had to translate 1 in 1,000 to a percentage; only 24 percent did so successfully. Wide disparities in numeracy also existed such that higher scores existed for men versus women, younger adults versus older adults, more educated adults versus less educated adults, and higher versus lower income adults (independent effects existed for only sex, education, and income).
The 2003 National Assessment of Adult Literacy (NAAL) defined numeracy (also called quantitative literacy) as “the ability to understand and use numbers in daily life” (Kutner et al., 2007). They estimated the proportion of Americans who fall into Below Basic, Basic, Intermediate, and Proficient quantitative literacy performance levels. The survey was administered to more than 19,000 adults (ages 16 and older) living in households or prisons. To be classified into each quantitative literacy level, one has to exhibit a set of specific quantitative skills and not exhibit the specific skills of the quantitative literacy level above it.
Key abilities that adults needed to demonstrate to be classified into each level can be found in Table A-1. For example, key abilities at the Below Basic level include finding numbers and using them to perform simple operations (mostly addition) when the information is familiar and concrete. For example, adding two numbers to complete an ATM deposit slip is a task categorized at the Below Basic level of quantitative literacy. In contrast, a sample task from the Intermediate level involved determining what time a person can take a prescription medication, given instructions on taking the medication in relation to eating. A sample task at the highest performance level, Proficient, involved calculating the yearly cost of life insurance using a table that gives the cost per month for each $1,000 of coverage. Individuals with less than Proficient abilities (those at Below Basic, Basic, or Intermediate levels) are expected not to be able to perform this sample life insurance task.
Results from the NAAL indicated that 22 percent of American adults are at the Below Basic level, 33 percent are at the Basic level, 33 percent are at the Intermediate level, and 13 percent are at the Proficient level of quantitative literacy. Results also indicated demographic differences
|Quantitative Literacy Level||% of Adults in Each Level (NAAL findings)||Estimated % (#) of Uninsured Adults in Each Level*||Estimated % (#) of Insured Adults in Each Level*||Key Abilities Associated with Level (NAAL)|
|Below Basic||22%||28.8% or 9,169,063||18.2% or 30,596,144||Locating numbers and using them to perform simple quantitative operations (primarily addition) when the mathematical information is very concrete and familiar|
|Quantitative Literacy Level||% of Adults in Each Level (NAAL findings)||Estimated % (#) of Uninsured Adults in Each Level*||Estimated % (#) of Insured Adults in Each Level*||Key Abilities Associated with Level (NAAL)|
|Basic||33%||33.4% or 10,656,748||31.9% or 53,702,419||Locating easily identifiable quantitative information and using it to solve simple, one-step problems when the arithmetic operation is specified or easily inferred|
|Intermediate||33%||29.3% or 9,339,640||35.3% or 59,508,631||Locating less familiar quantitative information and using it to solve problems when the arithmetic operation is not specified or easily inferred|
|Proficient||13%||8.6% or 2,749,954||14.6% or 24,505,031||Locating more abstract quantitative information and using it to solve multistep problems when the arithmetic operations are not easily inferred and the problems are more complex|
NOTE: Individuals at each level of quantitative literacy are thought to have the skills identified at that level, but are thought to not have the skills at levels above their own (e.g., an individual with Below Basic quantitative literacy should have the skills located in that row, but would not have the skills located in the rows for Basic, Intermediate, or Proficient literacy).
* These estimates are not based on perfectly comparable samples. The sample from the NAAL consists of people ages 16 years and older living in households or prisons whereas the sample from the 2009-2011 Census is a civilian noninstitutionalized population 25 years and over. Both samples also include older adults (65 years and older) who are not as relevant to Patient Protection and Affordable Care Act concerns because most are covered by Medicare. Although older adults tend to be less numerate, their inclusion likely affects the uninsured estimates very little (because most are insured), but means that the insured population of younger individuals 18 to 64 years old likely has higher quantitative skills than reflected in Table A-1.
in quantitative abilities. Males scored higher than females, high-income individuals scored higher than low income, and the more educated scored higher than the less educated. In addition, scores among white and Asian/Pacific Islander adults were higher than scores for black and Hispanic adults. No analyses were available concerning whether any single demographic variable predicted quantitative literacy scores over and above other demographic variables.
The proportions of individuals at each quantitative literacy level are based on the overall U.S. population, however, and may not accurately reflect the numeracy abilities we should expect from previously uninsured adults who will now enter the health care system as the result of the Patient Protection and Affordable Care Act (ACA). By combining the NAAL data with the 2009-2011 Census Bureau data (U.S. Census Bureau, 2009-2011), we estimated the proportion of uninsured and insured American adults who fall into Below Basic, Basic, Intermediate, and Proficient quantitative literacy categories (see Table A-1). For these estimates, we used the 2003 NAAL that provides data on the proportion of adults in each quantitative literacy level based on their educational attainment; the Census Bureau (2009-2011) provides data on the proportion of uninsured and insured adults at each level of educational attainment.
Using these two data sources, we estimated that among uninsured adults 28.8 percent are at the Below Basic level, 33.4 percent are at the Basic level, 29.3 percent are at the Intermediate level, and only 8.6 percent are at the Proficient level. Among insured adults, we estimated that 18.2 percent are at the Below Basic level, 31.9 percent are at the Basic level, 35.3 percent are at the Intermediate level, and 14.6 percent are at the Proficient level. See Table Annex A-1 in this appendix for a more detailed explanation of how these estimates were calculated and the limitations of these estimates. Given these estimates, roughly 29 percent (9,170,000) of uninsured adults and 18 percent (30,600,000) of insured adults lack the Basic quantitative skills necessary to locate quantitative information and use it to solve simple one-step arithmetic problems. Approximately 62 percent (19,800,000) of uninsured adults and 50 percent (84,300,000) of insured American adults lack the Intermediate quantitative skills necessary to locate less familiar quantitative information and use it to solve problems in which the arithmetic operation is not specified.
Dual-Process Theories and the Potential Influence of Time Pressure, Stress, and Illness on Reductions to Health Numeracy Skills
Research in numeracy has been associated with what are known as “dual-process theories” in decision making (Peters et al., 2006b). Information in decision making appears to be processed using an analytic mode
of thinking and an affective/experiential one (Epstein, 1994; Loewenstein et al., 2001; Reyna, 2004; Sloman, 1996; also called Systems 1 and 2, respectively, Kahneman, 2003; Stanovich and West, 2002). In particular, numeracy is considered an analytical skill—one has to think to do number calculations. Both modes of thought are important to forming decisions. The experiential mode is primarily based on affective (emotional) feelings, and processing using this mode is relatively effortless, automatic, and spontaneous. As shown in a number of studies, the affective feelings that are primary to this mode of thought provide both meaning and motivation to choice processes (Damasio, 1994). Processing in the analytic mode, on the other hand, is conscious, deliberative, reason based, verbal, and relatively slow. The analytical mode of thinking is more flexible and provides effortful control over more spontaneous experiential processes. Both modes of thinking are important and good choices are most likely to emerge when affective and analytical modes work in concert and decision makers think as well as feel their way through judgments and decisions (Damasio, 1994). Research, however, has demonstrated that the experiential mode (and affect, in particular) has a relatively greater influence when analytical capacity is lower due to cognitive load or time pressure (Finucane et al., 2000; Shiv and Fedorikhin, 1999).
This distinction is important because being involved in health decisions often involves factors that reduce how well patients and consumers think (e.g., time pressure for a patient to make an informed choice in a physician’s office, being sick, being stressed, or being overwhelmed with too much information). As a result, the numeric abilities of representative U.S. populations may overestimate the numeracy levels of patients making health decisions. This is because reported numeracy abilities for the U.S. population are usually measured in healthy individuals who are not under time pressure, whereas patients seen by health care providers may be subject to one of the factors above (e.g., being sick). These reduced numeracy abilities may lead to numeric sources of information being less well understood and used in health decisions, while less relevant sources of affect and emotion play a larger role. Little health research, however, exists concerning this possibility.
QUESTION 2: WHAT KINDS OF NUMERACY SKILLS ARE NEEDED TO SELECT A HEALTH PLAN, CHOOSE TREATMENTS, AND UNDERSTAND MEDICATION INSTRUCTIONS?
Education-Based Numeracy Skills
Apter and colleagues (2008) presented a hierarchy of mathematical skills required to successfully complete numeric tasks while making health
decisions. Higher level tasks include estimation, understanding probabilities, problem solving (the ability to decipher when and how to apply a numerical skill), understanding variability and error in measurement, and risk assessment. See the education-based numeracy skills of Table A-2 adapted from Apter et al. (2008). Having the skills to successfully complete these tasks is expected to allow patients and consumers to locate numeric information and transform it in ways that allow them to make more effective decisions about their health. Education-based skills are divided into four main skill categories: basic, computational, analytical, and statistical numeric skills. The basic skill to understand numeric information is necessary for many health-related tasks. When choosing a health plan from a health insurance exchange, for example, consumers must be able to read and understand basic fees and use simple arithmetic operations, such as adding costs together. Such understanding is a fundamental building block to deciding which health plans they prefer and can afford. Similarly, taking medications correctly requires the ability to read and understand dosage and timing instructions. Computational skills to do tasks such as estimating sizes and understanding how to work with frequencies and percentages are particularly important when making treatment decisions because options can be described based on the likelihood of risks and benefits in frequentistic form (e.g., 10 out of 100 patients) or percentage form (e.g., 10 percent of patients).
For tasks requiring analytical skills, patients and consumers must be able to apply numeric information to solve problems, make inferences, and interact with complex displays of information such as tables, graphs, and maps. For example, understanding numeric information provided in formats such as frequencies and percentages may not, by itself, be sufficient for accurate risk perception. Peters et al. (2006b) demonstrated that less numerate individuals were susceptible to format effects, presumably because the less numerate, although they likely understood the numbers in the sense that they could repeat them back accurately, did not transform numbers from one format to another. Specifically, in Peters et al. (2011), experimenters presented participants with the likelihood of an adverse event from a prescribed medication either in a frequentistic format (10 of 100 patients get a bad blistering rash) or a probabilistic format (10 percent of 100 patients get a bad blistering rash). Both formats are normatively equivalent. The experimenters found that less numerate individuals perceived a greater risk of an adverse event when the likelihood estimate was described in a frequentistic format (10 of 100) than when it was described in a probabilistic format (10 percent of 100). By contrast, highly numerate individuals rated the level of risk as the same in each information format. Normatively, the frame of information should not change the risk perception judgment.
|Skill Categories||Numeracy-Related Tasks|
Reading numbers, counting, telling time
Estimation of size, trend
|Education-based numeracy skills||
Problem solving and inferring the mathematical concepts to be applied
|Estimating error, uncertainty, variability|
|Statistical||Relative versus absolute|
|Risk (cumulative, relative, conditional)|
Seeking numeric information rather than avoiding it
Willingness to perform computation
More likely to attend to numeric information in a complex display
Able to disregard irrelevant information presented with numeric information
|Emergent decision-based numeracy skills||Memory||Recall numeric information from memory|
Sensitivity to numeric information sources
Sensitivity to non-numeric information sources when numeric sources are available
|Affective meaning||Derive affective meaning (i.e., a sense of goodness or badness) from numeric information. Note: Affect comes into play when developing preferences and making decisions. National Assessment of Adult Literacy comparison examples do not include choice|
In taking medications, other kinds of format issues appear. For example, with liquid medication, patients often use inaccurate measurement devices such as household spoons, and they often confuse teaspoons and tablespoons (Madlon-Kay and Mosch, 2000). In selecting health plans, consumers sometimes want to estimate annual costs. To do so correctly, they must transform some numbers (e.g., monthly premiums and biannual physician visits to annual) in order to add them to other numbers (e.g., annual deductibles). Such calculations require analytical skills and knowing how to apply numeric information to solve problems.
Finally, Apter includes concepts related to probabilistic reasoning in the Statistical skill category. This includes the understanding of variability and randomness, being able to evaluate relative versus absolute comparisons, and being able to compare different risk assessments (cumulative, relative, and conditional). Such skills are important because inclusion of preventive care services in plans offered in health exchanges means that the newly insured will need to choose between treatment options, and also choose whether or not to obtain preventive health screenings and treatments. To do so, consumers first must realize they are susceptible to a given disease (e.g., understand concepts of randomness and variability), and then understand the risks from the disease as well as the risk reduction from taking preventive steps (both relative and absolute risks). For example, imagine a patient who accurately understands that his risk of developing type 2 diabetes is greater based on the percent chance (probability) of developing disease at his current weight. He can then estimate how much his risk will be reduced with effortful changes to diet and exercise, and he can choose to develop healthier behaviors. He may also be better able to follow through on effective behaviors due to superior understanding of how to count calories or do other number-related tasks. In another example, imagine a 50-year-old woman with no family history of breast cancer. Although her known risk factors are low, if she is highly numerate, she may understand that the inherent variability and randomness of health risks still means she is at risk. Understanding that risk, she may be more likely to pursue recommended screening procedures.
Apter et al.’s hierarchy focuses on math education and the computational skills necessary to function in a complex environment. Table A-2 lists the education-based numeracy skills as discussed by Apter et al. (2008). With respect to the NAAL quantitative literacy levels (Below Basic, Basic, Intermediate, and Proficient), The skills in Table A-2 do not align directly with particular levels, but Apter et al. (2008) listed the education-based skills in order of difficulty from least to most difficult. Table A-2 also includes emergent decision-based numeracy skills adapted from Peters (2012). The two types of skills are separated by a dashed line in Tables A-2 through A-5 for clarity.
Emergent Decision-Based Numeracy Skills
Berkman et al. (2011) concluded that having a theoretical basis to interventions was an important component of effective interventions to reduce health disparities. As a result, we briefly review what is known about the psychological theory underlying numeracy’s relation to health decision making. It is thought that numeracy exerts its influence on health outcomes in part through its effects on health decision making (Peters, 2012; Reyna et al., 2009). Understanding these underlying mechanisms should help in the design of more effective interventions in the future.
Psychological research on numeracy and decision making indicates that numeracy is also associated with emergent decision-based abilities not formally taught in school (see emergent decision-based abilities in Table A-2). Previous research has shown that higher numeracy is related (not surprisingly) to more comprehension of provided numeric information in a variety of domains, but it is also associated with a greater likelihood to seek out, attend to, and remember numeric information. Higher numeracy has also been associated with more precise number-related affect, a greater sensitivity to numbers in judgments and decisions, and less influence of non-numerical information (Peters et al., 2012). Some emergent decision-based numeracy abilities (e.g., more numerate individuals are less susceptible to various framing effects) have previously been identified by Apter et al. (2006) as being part of education-based skills, so we leave them categorized as education-based skills.
To begin, the highly numerate appear to be more motivated with respect to numeric information; they are more likely to seek it out, whereas less numerate individuals may avoid numeric information (Ancker and Kaufman, 2007; Keller, 2011; Lipkus and Peters, 2009). Such information seeking (and lack of information avoidance) is important when choosing whether to find out about the likelihood of a disease such as breast cancer (and possibly be screened for it) or deciding whether to take a new medication that has less than certain benefits and may cause adverse events. Highly numerate individuals, for example, might be more likely to examine detailed consumer medication information to find out about possible side effects and their associated likelihoods. They may also be more likely to pursue information about how to minimize the likelihood of potential medication side effects (e.g., eating and exercise behaviors when taking Coumadin®). Second, when faced with a complex display of information, higher numeracy is associated with a greater likelihood of attending to provided numeric information (Keller, 2011), as well as a greater ability to ignore irrelevant information (e.g., hospital information not related to the quality of care it offers) (Peters et al., 2007b). In the case of choosing a health plan, less numerate individuals might seek out, attend to, and be
|Quantitative Literacy Level||Comparative NAAL Item||Example Task: Health Plan Selection||Skill Categories (from Table A-2)|
(28.8% of uninsured population)
|Calculate the price difference between two appliances, using information in a table that includes price and other information about the appliances.||Compare and calculate the difference between monthly premiums of two plans.||Basic; Analytical
Information Seeking; Attention
|Basic||Calculate the cost of a sandwich and||Select the health plan with the||Basic; Computational; Analytical|
|(33.4% of||salad, using prices||lowest cost based|
|uninsured||from a menu.||on the annual||Information Seeking;|
|population)||premium and annual deductible for a family.||Attention; Memory|
|Intermediate||Calculate the cost of raising||Calculate the coinsurance cost||Basic; Computational; Analytical|
|(29.3% of||a child for 1||of an emergency|
|uninsured||year in a family||room visit bill||Information Seeking;|
|population)||with a specified income, based on a newspaper article that provides the percentage of a typical family’s budget that goes toward raising children.||for $500 from a table of different coinsurance rates.||Attention|
|Proficient||Calculate an employee’s share||Estimate total annual cost||Basic; Computational; Analytical; Statistical|
|(8.6% of||of health insurance||of the health|
|uninsured||costs for 1 year,||plan, including||Information Seeking;|
|population)||using a table that shows how the employee’s monthly cost varies with income and family size.||premiums, copays, and deductibles, based on expected health care needs (e.g., estimating costs due to chronic illnesses such as diabetes or asthma).||Attention; Memory; Information Sensitivity|
|Performance Level||Comparative NAAL Item||Example Task: Treatment Selection||Skill Categories (from Table A-2)|
|Below Basic (28.8% of uninsured population)||Compare two prices by identifying the appropriate number and subtracting.||Compare and calculate the difference in copay amounts between generic and name-brand prescription drugs.||Basic; Analytical Information Seeking; Attention|
|Basic (33.4% of uninsured population)||Perform a two-step calculation to find the cost of three baseball tickets, using an order form that gives the price of one ticket and the postage and handling charge.||Calculate the difference in percentages of patients who survive one treatment compared to another.||Basic; Computational; Analytical; Statistical Information Seeking; Attention|
|Intermediate (29.3% of uninsured population)||Calculate the cost of raising a child for 1 year in a family with a specified income, based on a newspaper article that provides the percentage of a typical family’s budget that goes toward raising children.||Calculate the proportion of patients who will suffer at least one adverse event based on patient age and three possible adverse events (assume independence of adverse events).||Basic; Computational; Analytical Information Seeking; Attention|
|Proficient (8.6% of uninsured population)||Calculate the yearly cost of a specified amount of life insurance, using a table that gives cost by month for each $1,000 of coverage.||Calculate the 5-year risk of fracture from an osteoporosis medication for a female patient of a given age, using a table that gives annual risk for each gender by age group.||Basic; Computational; Analytical; Statistical Information Seeking; Attention|
|Performance Level||Comparative NAAL Item||Example Task: Understanding Medication and Treatment Instructions||Skill Categories (from Table A-2)|
|Below Basic (28.8% of uninsured population)||Calculate the change from a $20 bill after paying the amount on a receipt.||Locate the risks of different side effects for the medication. Identify which side effect is most likely to occur.||Basic; Computational; Analytical Information Seeking; Attention|
|Basic (33.4% of uninsured population)||Perform a two-step calculation to find the cost of three baseball tickets, using an order form that gives the price of one ticket and the postage and handling charge.||24 pills remain in a bottle of prescription medication. If you take 2 pills per day and refilling a prescription can take up to 7 days, in how many days should you order a refill to make sure that you do not run out of your prescription?||Basic; Computational; Analytical Information Seeking; Attention|
|Intermediate (29.3% of uninsured population)||Determine what time a person can take a prescription medication, based on information on the prescription drug label that relates timing of medication to eating.||“The patient forgot to take this medicine before lunch at noon. What is the earliest time he can take it in the afternoon? GARFIELD, Robert M. Dr. LUBIN, Michael DOXYCYCLINE 100MG Take one tablet on an empty stomach 1 hour before a meal or 2 to 3 hours after a meal unless otherwise directed by your doctor.”||Basic; Analytical Information Seeking; Attention; Memory (if time of last meal was not provided)|
|Performance Level||Comparative NAAL Item||Example Task: Understanding Medication and Treatment Instructions||Skill Categories (from Table A-2)|
|Proficient (8.6% of uninsured population)||Determine the number of units of flooring required to cover the floor in a room, when the area of the room is not evenly divisible by the units in which the flooring is sold.||Diabetes management—understanding glucose meter readings, interpreting sliding-scale regimes, titrating oral medications or insulin, adjusting insulin for carbohydrate content. (Note: This example is much more complex than any of the NAAL examples used, but it is a realistic example of what patients are required to do.)||Basic; Computational; Analytical Information Seeking; Attention; Memory; Information Sensitivity; Affective Meaning|
more easily influenced by anecdotes that describe the friendliness of an insurance provider’s staff (e.g., from a neighbor or in marketing materials). At the same time, they may fail to adequately attend to the large annual deductible or copays required by the plan. In one study, for example, less numerate participants could usually understand which consumer-directed health plan had the lowest monthly premiums (we estimate this task to require Below Basic ability), but only about a third of them were able to identify which health plan was better if the patient needed a lot of care (a more difficult task that likely requires at least Intermediate ability) (Greene et al., 2008).
Highly numerate individuals also remember numeric information better than the less numerate (Sagara, 2009). Such numeracy effects, however, may be greatest soon after learning health information, then lessen over time. Galesic and Garcia-Retamero (2011), for example, studied how well participants recalled the consequences of health-related behaviors, such as being overweight or exercising, and cardiovascular health. Such recall
may be important to following through on recommended behaviors. They found that highly numerate individuals recalled the consequences of health-related behaviors better than the less numerate after 10 minutes. Memory for both groups had declined after 3 weeks, and no statistically significant memory differences existed between the groups at this later time point. Of course, even the short-term memory advantage could be helpful in following the complex treatment plans required in management of chronic diseases such as diabetes. In these cases, patients either have to remember pertinent numeric information (carbohydrate consumption, blood glucose levels, insulin doses, times administered, etc.) in order to take the next step in managing their disease effectively, or they have to be diligent about recording it in the moment.
Previous research also has shown that highly numerate individuals draw more precise affective meaning from numbers than less numerate individuals. Using a paradigm modified from Denes-Raj and Epstein (1994), Peters et al. (2006b) presented participants with drawings of two bowls of jellybeans with different numbers of red and white jellybeans. Participants were told to imagine they could pick one bean and they would win $5 if the bean they selected was red. The larger bowl of 100 jellybeans had a higher number, but a smaller proportion (9 in 100 or 9 percent) of red jellybeans than the smaller bowl. The smaller bowl of 10 jellybeans had one red jellybean and a larger proportion (1 in 10 or 10 percent) of red jellybeans. Both bowls had the objective percentage of colored jellybeans labeled under each bowl. Participants were asked which bowl they would prefer to choose from and how clear a feeling they had about the goodness or badness of the larger bowl’s 9 percent chance of winning. Peters et al. (2006b) found that the less numerate were more likely to choose bowl A, the suboptimal choice, than were more numerate individuals. The reason for this difference appeared to be that highly numerate individuals developed more precise feelings about the 9 percent chance of winning than the less numerate.
Being able to derive affective meaning from numbers and number comparisons is important in a health environment to compare treatment effectiveness or health care costs. Individuals can have strong affective reactions to risk and other numeric information, and this affect appears to guide risk perceptions and decisions (Slovic et al., 2005; Zikmund-Fisher et al., 2010). Studies have shown that without affect, numbers are not used in judgment and choice (Bateman et al., 2007; Peters et al., 2006b). In one study, for example, Fagerlin and colleagues (2005b) found that women asked to estimate their risk of breast cancer tended to overestimate that risk. Then, when told their actual risk, these women appeared to draw affective meaning from the number comparison. Compared to women who had not estimated their own risk first, they were quite relieved and perceived their cancer risk as lower than when they were simply told their
cancer risk without having made their own estimate first. This is important because it may help to explain why counseling women about breast cancer risks decreases screening compliance. Although numeracy was not explicitly studied, highly numerate women may be more likely to show this and similar effects. As a result, although having greater numeracy generally leads to superior judgments and decisions because they are more likely to attend to numbers and number comparisons and derive affective meaning to guide their choices (Peters et al., 2006b), the highly numerate may sometimes demonstrate worse judgments than the less numerate.
Perhaps because of their greater abilities to attend to numeric information and draw affective meaning from it, highly numerate individuals tend to show a greater sensitivity to numeric information in health compared to the less numerate. For example, Lipkus et al. (2010) presented women with early-stage breast cancer with their chances of being cancer free during the next 10 years under four preventive cancer treatment decisions. They found that more numerate patients were sensitive to differences in cancer-free survival estimates for the treatments (they perceived themselves, on average, as more likely to survive when provided higher survival chances such as 92 percent than lower chances such as 63 percent); perceptions of the less numerate patients were almost completely insensitive to these same differences in survival odds. Among the women with the highest provided survival odds (average survival odds were about 92 percent), the less numerate were very pessimistic and perceived their 10-year survival odds as quite low on average (less than 45 percent); the highly numerate were also pessimistic, but perceived their odds as considerably higher (more than 75 percent).
The differences in sensitivity to numbers may also cause (or be caused by; the research is not clear on this point) an opposing difference in sensitivity to non-numeric information. In contrast to the highly numerate, less numerate individuals have shown a greater sensitivity to non-numeric and often emotional sources of information such as provided information frames (survival versus mortality rates are potential sources of emotion) and current mood states. In a study by Västfjäll and colleagues (in preparation), researchers manipulated participants’ moods to be either positive or negative using a presumably unrelated recall task, and then asked participants to price a lottery ticket. Results indicated that less numerate participants were more influenced by the mood induction than highly numerate participants. In particular, the less numerate participants set higher prices for the lottery ticket in the positive-mood condition than the negative-mood condition. This is important because patients and consumers make many health judgments and decisions while in emotional states (e.g., the joy of a positive result; the anxiety of a new diagnosis). In Peters et al. (2009), less numerate participants also relied on their moods to judge the quality of care of a hospital rather than using provided numeric quality-of-care indicators;
the highly numerate used some of the provided numeric information and did not rely on their mood states in the moment.
Thus, previous research has shown that greater numeracy generally leads to better decision making. More numerate individuals tend to understand numbers better than the less numerate (and comprehension is a fundamental building block of good decisions). In addition, however, greater numeracy has been associated with a greater likelihood to seek out, attend to, and remember numeric information; to derive more precise number-related affect; to be more sensitive to numbers in judgments and decisions; and to be less influenced by non-numerical information (Peters, 2012). In general, the highly numerate do more work with numbers than do the less numerate, and these habits of the mind appear to coalesce and allow them to make superior number-based decisions.
The emergent decision-based abilities have not been linked explicitly with the four NAAL quantitative literacy levels. The extant research, however, supports these emergent abilities being present more among highly numerate individuals than the less numerate. As a result, individuals with higher levels of quantitative literacy will tend to exhibit more of these abilities. In particular, the emergent numeracy abilities are likely to be associated with either Intermediate or Proficient quantitative literacy levels. These abilities, however, can also emerge due to experience and/or motivation; individuals with lower numeracy will sometimes use these abilities nonetheless (e.g., seek out numeric information) if they have had experience in the health domain and understand its importance or if they are motivated in some other way (Hibbard et al., 2007b). Women, for example, although less numerate on average than men, often show what is likely a health care experience-based gender advantage (i.e., women tend to be more involved in family health decisions) (Ben-Joseph et al., 2009; Hibbard et al., 2007b).
Below we provide additional health examples in the three requested areas (following medication instructions, making health plan choices, and choosing treatments). We also attempt to match the examples, where possible, to the four levels of quantitative literacy identified in the NAAL to provide the reader with an idea of the approximate proportion of the previously uninsured population who will likely be able to do each task.
Example Skills Needed to Select a Health Plan at Each Level of Quantitative Literacy
The ACA (section 1302) broadly defines the levels of coverage and the essential health benefits that must be included in new health insurance plans. It also leaves considerable room for variation between plans. Bronze, Silver, Gold, and Platinum plans must, respectively, cover 60, 70, 80, and 90 percent of the value of the benefits included in the plan, but a great deal
of flexibility exists in how plans are implemented. Consumers are faced with financial decisions based on premiums, copayments, coinsurances, and annual deductibles. The broad definition of essential health benefits also allows for variation in the services covered, adding another layer of complexity to the decision process.
Table A-3 provides examples of tasks related to health plan selection that patients should be able to complete at each level of quantitative literacy. For each example, a comparative NAAL task is included for reference along with relevant skill categories from Table A-2 that we believe are needed to perform the task listed. Note that, as in Table A-2, the education-based skills and emergent decision-based skills are separated by a dashed line (education-based skills are above the dashed line in each row; emergent decision-based skills are below).
To make a health plan choice, consumers first must be able to locate and understand relevant pricing information. This initial task can be complicated by unfamiliarity with terms such as copay and coinsurance (Quincy, 2012). However, most consumers (even those with Below Basic ability) should be able to locate information, although one should keep in mind that the Below Basic group includes individuals with very low-level skills. In fact, previous attempts to assess how well younger and middle-aged adults locate cost and quality-of-care information in tables and charts indicate that about 9 percent errors might be expected even at this basic building block of the health plan selection process (Hibbard et al., 2001). It is not entirely clear how to adjust this finding for the group of previously uninsured individuals who will soon be making these choices, although the proportion of comprehension errors will be largest in the Below Basic group.
To compare different plans, consumers must be able to calculate differences in monthly premiums; this is expected to be a Below Basic skill that most consumers can perform successfully (see Table A-3 for the tasks and relevant skill categories). Selecting the health plan with the lowest cost based on the annual premium and deductible for a family is expected to be a Basic skill doable by about 71 percent of the uninsured population (everyone except those at the Below Basic level). With at least an Intermediate level of NAAL performance, more comprehensive evaluations of health plan costs are more likely, including calculations such as coinsurance costs based on a percentage of the cost of treatments. More complex calculations (e.g., calculating annual costs based on monthly premiums, estimated out-of-pocket expenses from flat-rate copayments, and estimated out-of-pocket expenses from percentage-based coinsurance amounts that meet annual deductibles), however, require much greater proficiency. Only an estimated 8.6 percent of the currently uninsured population is expected to have reached this Proficient quantitative literacy level. Moreover, consumers must be able to estimate their own future health care needs. For example, a
patient with a chronic illness, such as asthma, needs multiple prescription drugs and may be best served by a plan with higher monthly premiums that covers a greater percentage of prescription drug costs. To determine which health plan best meets her needs, the patient must recall how much each prescription costs her and how many prescriptions she fills per year, add together the cost of these prescriptions, calculate the annual premium amount, and then calculate total costs for each plan and compare the total cost across multiple plan offerings. This is also expected to require Proficient quantitative literacy. Given the small proportion of individuals at the Proficient level in both the insured and uninsured groups (see Table A-1), it is not surprising that researchers have found consumers to be anxious, confused, and overwhelmed when making health plan choices (Day and Nadash, 2012; Quincy, 2012).
Beyond the difficulties posed in making the calculations to compare different health plans, individuals with Below Basic performance may be more prone than other individuals to focus on the most salient cost involved in health insurance, which, based on an analysis of Medicare Part D choices, is likely to be monthly premiums rather than out-of-pocket expenses (Abaluck and Gruber, 2011). Such patients may simply choose the plan with the lowest premium, not understanding that their total annual cost of services may be much higher than another plan with only slightly higher premiums (see also Greene et al., 2008).
Example Skills Needed to Select Treatments at Each Level of Quantitative Literacy
Although decisions among health plans may rely largely on price calculations and comparisons, the decision of which treatment to choose is much less likely to include price as a component. This difference is primarily due to the ambiguity and variability of treatment costs and difficulty in obtaining them. Patients frequently do not receive cost information before treatments are administered (and may find out about or pay attention to only their portion of the costs afterward). Moreover, recent data highlighted the extreme price variability that exists among hospitals for similar treatments (CMS, 2013). As a result, even when patients want to evaluate treatment cost differences, accurate cost information can be complicated and difficult to obtain (Rosenthal et al., 2013). Treatment decisions tend to be based instead on the health care provider’s recommendation and (when patients share in the decision) on the convenience of administration, medication copayments, and perceived risks and benefits of treatment options.
Table A-4 provides example NAAL tasks paired with treatment decision tasks estimated to fall into each performance level of quantitative literacy; relevant skill categories are also included. As in previous tables, the
education-based skills and emergent decision-based skills are separated by a dashed line (education-based skills above, emergent decision-based skills below). Because the NAAL examples focus heavily on calculations of costs, the examples are not directly matched with our treatment option example tasks. In each example, the patient must be able and willing to seek out numeric information and attend to it; such ability and willingness could be derived from numeracy skills or from a motivation to care for the self or others (e.g., patient activation) (Hibbard et al., 2007a). Most patients, including those with Below Basic performance, likely will be able to compare the copay amounts between a generic and name-brand prescription drug. With at least a Basic level of quantitative literacy (an estimated 71.2 percent of the uninsured population), patients should be able to calculate the difference in survival rates between two treatment options when provided with the percentage of patients who survive. Having at least Intermediate quantitative literacy (an estimated 37.9 percent of the uninsured) would be necessary to complete a medication cost comparison based on the recommended dosage and unit cost of a medication (e.g., comparing the number of pills per dose and cost per pill in generic acetaminophen versus Tylenol in order to choose the less expensive option). Only those with Proficient quantitative skills (an estimated 8.6 percent of the uninsured) are expected to be able to calculate cumulative risks and benefits of treatments accurately and compare them to make treatment decisions based on tradeoffs that are acceptable to them. For example, a woman with osteopenia might be advised to take a bisphosphonate for 3 to 5 years, but must choose whether or not to take it based on information about annual rates of risks and benefits.
Example Skills Needed to Understand Medication Instructions at Each Level of Quantitative Literacy
Properly following medication instructions can be a difficult task for some patients (e.g., taking a prescription drug in their own homes). Although prescription drugs are labeled with dosage instructions, patients must be able to read and understand them, remember what time to take any medication, determine how to handle inadvertently missed doses, and, when appropriate, determine when to have prescriptions refilled to avoid running out of daily medications.
Table A-5 provides examples of the skills needed to follow medication and treatment instructions at each level of quantitative literacy performance, a comparative NAAL example, and relevant skill categories. Note that, as in Table A-2, the education-based skills and emergent decision-based skills are separated by a dashed line (education-based skills above, emergent decision-based skills below). Individuals with Below Basic abilities
can be expected to locate the risks of side effects in a table in a decision aid or in a relatively simple insert located on a prescription drug bottle and to determine which side effect is most likely. With at least a Basic level of performance, patients can be expected to anticipate and plan for medication needs, such as determining how soon a prescription must be ordered based on the number of pills left and the number required each day. One NAAL task identified as at the Intermediate level of performance requires patients to understand medication information and infer, based on instructions, how to handle a missed dose, taking into consideration the time since their last meal.
The management of chronic diseases such as diabetes and asthma pose particular challenges, even for those with Proficient quantitative literacy. Diabetics must know how to accurately use and understand the readings from glucose meters, and modify their insulin dosage based on glucose levels, level of activity, and carbohydrate content. The information needed to make these calculations is found in a variety of formats, such as sliding scales and tables that include nutritional information. Diabetic patients need to be able to perform relatively complex calculations correctly, understand numeric information presented in different formats, and recall numeric information and/or keep an accurate record of it. This combination of tasks is likely more difficult than any of the NAAL examples at the Proficient level. As a result, even the most numerate likely find chronic disease management challenging, although they would perform better than those at lower performance levels.
QUESTION 3: WHAT DO WE KNOW ABOUT HOW PROVIDERS SHOULD COMMUNICATE WITH THOSE WITH LOW NUMERACY SKILLS?
A series of recent papers has reviewed how to present numeric information to maximize informed decision making (e.g., Ancker et al., 2006; Apter et al., 2008; Berkman et al., 2011; Fagerlin and Peters, 2011; Fagerlin et al., 2007a; Hibbard and Peters, 2003; Lipkus, 2007; Lipkus and Hollands, 1999; Peters et al., 2007a). They have come to many of the same conclusions. In this section, we summarize the literature by focusing on five main communication themes that are consistent with the process goals identified in Figure A-1 (especially lowering cognitive effort and highlighting meaning) and that are updated based on more recent results with less numerate patients and consumers.
FIGURE A-1 Data presentation approaches that facilitate informed decision making and the use of information in choice.
SOURCE: Hibbard and Peters, 2003.
- Provide numeric information (as opposed to not providing it);
- Reduce the cognitive effort required from the patient or consumer and require fewer inferences (i.e., do the math for them);
- Provide evaluative meaning, particularly when numeric information is unfamiliar;
- Draw attention to important information; and
- Set up appropriate systems to assist consumers and patients.
Most strategies targeted toward the education-based numeracy skills from Table A-2 can be found in the section focused on reducing cognitive effort. Strategies targeted at the emergent decision-based skills are found in the sections on reducing cognitive effort, providing evaluative meaning, and drawing attention to important information. Table A-6 summarizes recommended strategies for communicating with patients and consumers with low numeracy skills. In the text that follows, we describe the evidence underlying each of these recommendations.
|What Communicators Should Do||Specific Strategies|
|Provide numeric information (as opposed to not providing it)||Self-explanatory|
Reduce the cognitive effort required from the patient or consumer and require fewer inferences (i.e., do the math for them)
Provide fewer options
Provide less information
Present absolute risks, not just relative risks
Keep denominators and time spans constant
Use numbers consistent with how people use the number line
Do the math for them
Use appropriate visuals
Provide evaluative meaning, particularly when numeric information is unfamiliar
Carefully use evaluative labels and symbols
Carefully use frequency versus percentage formats
Use other, more imaginable data formats
Use emotion to persuade
Draw attention to important information
Order information with the most important information first or last
Highlight the meaning of only the most important information
Use a framework to provide an overview
Use fonts that draw attention to important information
Set up appropriate systems to assist consumers and patients
Identify communication goals
Choose information presentation formats strategically
Consider the use of default options and other choice architecture
Use computer-aided decision tools
Use information intermediaries
Provide Numeric Information (as Opposed to Not Providing It)
In consumer domains such as purchases of homes and lottery tickets, numeric information (e.g., mortgage rates and likelihoods of winning) is provided to better inform choices. In health domains, numbers are sometimes provided consistently (e.g., copay amounts in insurance choices), but other times are rarely provided (e.g., likelihoods of benefits and side effects when choosing a medical treatment). Providing numbers (compared to not providing them) even in these latter circumstances has been found to influence patient understanding and willingness to take medications in two ways (Berry, 2006; Lipkus, 2007). First, qualitative labels such as “low chance” or “common” are interpreted differently by different people. To
one person, common might mean 50 percent whereas to others it means 25 percent (Berry, 2006). Second, the average person tends to overestimate risk likelihood when provided with only non-numeric information (e.g., risk labels such as “common,” “rare”) compared to when they are provided numeric information (Berry et al., 2002, 2003a,b). Of course, providing numeric information can be problematic, particularly in less numerate populations. As a result, policy makers and others have questioned whether less numerate populations can “handle” numeric information (Schwartz, 2011). Results of a recent study, however, did not support this view (Peters et al., in review). Both more and less numerate respondents were less likely to overestimate risks and were more willing to take the prescribed medication when provided numeric information about medication side effects as opposed to providing only non-numeric information. Although less numerate individuals have more difficulty with numeric information than do the more numerate, they nonetheless benefitted from its provision at least in the context of medication side effects.
The fact that less numerate individuals do have more problems with numeric information, however, emphasizes the need to understand how to provide comprehensible and usable numeric information to them.
Reduce the Cognitive Effort Required from the Patient or Consumer and Require Fewer Inferences from Them (i.e., Do the Math for Them)
Provide Fewer Options
A breast cancer communication tool called “Adjuvant Online!” (http://www.adjuvantonline.com) was designed to help oncologists communicate the benefits for patients receiving hormonal therapy and chemotherapy (Ravdin et al., 2001). Typically, patients are presented with the risks of no additional treatment, each treatment alone, or both hormonal therapy and chemotherapy. However, for most women, only two choices are appropriate. Zikmund-Fisher and colleagues (2008) tested the impact of providing only those two choices and found that, when fewer options were presented, knowledge increased significantly. Medical and other health experts should identify more and less critical elements of a decision (e.g., dominated options that are worse than other available options on every important dimension) so that information providers can delete them from the consideration set or strategically choose how to present them.
Although having more choice options can have advantages, recent research has pointed toward the notion of a “paradox” or “tyranny” of choice. For example, psychological research has demonstrated that having more options can lead to worse choices and lower satisfaction (Hanoch et al., 2009; Schwartz, 2005). In particular, researchers have suggested that an
overabundance of choice can lead to information overload (Huffman and Kahn, 1998; Reutskaja and Hogarth, 2009; Scammon, 1977), decreased motivation, an inability to choose (Dhar, 1997; Iyengar and Lepper, 2000; Iyengar et al., 2004), decision-related anxiety (Garbarino and Edell, 1997), and outcome dissatisfaction and regret (Botti and McGill, 2006; Schwartz, 2000, 2004). Schwartz et al. (2002) further found that the combination of large choice sets and a desire to choose the best were related to more regret, reduced happiness, and less overall choice satisfaction. The notion of providing fewer options may be particularly relevant to health plan selection. Numeracy effects have not been studied to the best of our knowledge, but it seems likely that providing fewer options would be especially helpful to the less numerate.
Provide Less Information
Information is provided to respect consumer and patient autonomy and to help them make better informed decisions. Cognitive drawbacks exist, however, to providing more information. Peters and colleagues (2007a) tested whether providing lay decision makers with less information, rather than more, could result in the best outcomes. The results indicated that providing less information in hospital quality reports (non-quality-of-care information, e.g., the number of general care beds was removed) resulted in better decision making through improved comprehension and higher quality choices, particularly among participants with lower numeracy skills. Health information providers are faced with a challenge to communicate important content to patients and consumers (e.g., through patient portals and mobile apps) and, simultaneously, not communicate too much content as the presence of extraneous information appears to confuse those who are less numerate (see also Kaminski and Sloutsky, 2013).
Present Absolute Risks, Not Just Relative Risks
When treatment information is presented in a relative risk format (e.g., using hormone replacement therapy doubles the risk of breast cancer), their risks seem larger and treatments are viewed less favorably than when the same information is presented using an absolute risk format (Baron, 1997; Forrow et al., 1992; Malenka et al., 1993). This is as true for the lay public as it is for medical students (Chao et al., 2003). Although not studied with respect to numeracy, it is quite likely that effects would be as big or bigger among the less numerate. Other relative risk examples are ambiguous (“Treatment X has a 5 percent greater risk than Y”). If treatment Y has an absolute risk of 20 percent, 5 percent more risk means that X has a risk of either 21 percent or 25 percent. Providing absolute risk numbers disam-
biguates the situation and reduces cognitive effort and potential confusion by doing the math for the patient.
Keep Denominators and Time Spans Constant
Patients experience greater difficulty comparing across treatments when different denominators are used (Fagerlin and Peters, 2011). A single denominator should be chosen for comparisons (e.g., 1 in 10,000 and 400 in 10,000 rather than 1 in 10,000 and 4 in 100). In addition, whole numbers (e.g., 1 in 10,000) are better understood than fractions and decimals (0.01 in 100). Similar advice exists for time spans. To facilitate comparisons, use the same time frame when presenting risks and benefits (e.g., provide annual costs for all health plans rather than monthly costs for some and annual costs for others).
Use Numbers in a Direction Consistent with People’s Expectations
Peters and colleagues (2007a) found that less numerate consumers, in particular, understood more when provided information requiring less cognitive effort. They presented hospital quality-of-care information either in a format in which a higher number meant better (the number of registered nurses per 100 patients) or in the more usual format where a lower number meant better (the number of patients per registered nurse). Putting the numbers in a direction consistent with people’s expectations (i.e., usually higher numbers mean something “better” than lower numbers) facilitated comprehension and helped respondents make better choices. Results were even stronger among the less numerate than among the highly numerate. This concept applies equally to other information formats common in medicine. For example, when explaining risks associated with treatment, some information providers use the Number Needed to Treat (NNT). If considering the benefits of chemotherapy, for example, NNT is the number of women needed to take chemoprevention to prevent cancer in one of them; here, larger numbers mean a less effective treatment. NNT is a difficult format for people to understand and it should not be used with laypeople (and arguably not with physicians either who can also be innumerate) (Anderson et al., 2011; Sheridan and Pignone, 2002).
Do the Math for Them
When evaluating healthy behaviors such as taking medication, eating better, or exercising more, consumers and patients are often told about risks over one time period and they are expected to extrapolate to other time periods. For example, Nina might be informed of the annual risk of
taking birth control pills, but she intends to take them for many years, say 10. Understanding this 10-year risk requires a level of numeracy that most people do not have. In one study, for example, well-educated participants were asked a problem that required a similar mathematical solution: “Imagine that, when the Columbus Clippers and the Eugene Emeralds minor league baseball teams have played each other, the Columbus Clippers won only 10 percent of the time. If the teams have a four-game series, by your calculations, what are the chances that the Clippers will win at least once?” (Correct answer: 34 percent) (Peters et al., 2012). Only 1 percent of their college student sample answered this question correctly. Similar cumulative-risk comprehension issues exist in the long-term, false-positive rates from annual cancer screenings in some groups (Gigerenzer, 2002; Sakr et al., 1996; USPSTF, 2011; Welch et al., 2011). Providing estimates for risks over longer time periods by doing the math for consumers would go a long way toward helping them understand the cumulative implications of their choices.
Use Appropriate Visuals
Presenting event rates with visual aids such as pictographs (also called icon arrays), bar charts, or flow diagrams may aid accurate understanding of numeric information such as probabilities. This appears particularly true in less numerate populations. Visual displays have been shown to reduce several biases, including denominator neglect (Garcia-Retamero et al., 2010), framing effects (Garcia-Retamero and Cokely, 2011), and the use of anecdotes over more reliable statistical information (Fagerlin et al., 2005a). Icon arrays, in particular, have been tested extensively in recent health communication research, and some nuances to their use have arisen. For example, the icons should be arranged in blocks (e.g., of those with versus without the disease) rather than being scattered randomly (although scattering them randomly can facilitate the perception of randomness, e.g., who gets a disease). Numerator size may also be an important factor when presenting the changes in numeric outcomes for events out of 1,000 among adults with lower education and literacy (McCaffery et al., 2012). Where the outcome is less than 100/1,000, icon arrays were better understood and processed more quickly than bar charts, particularly if the difference between event rates was small. However, for more common outcomes (greater than 100/1,000), bar charts were better, possibly because the icon arrangement was more complicated. In addition, the role of shading in processing the part-to-whole relationship of icon arrays is still not well understood. Most importantly, usually single icon arrays have been tested, and little is known about the effects of icon arrays in those health situations that would likely require integration across multiple arrays (e.g., displaying
the 10 possible adverse effects of a prescribed medication). It seems probable that the complexity of multiple icon arrays would disadvantage the less numerate in particular.
Finally, some graphs appeared better suited for particular tasks (e.g., line graphs for trends over time, bar graphs for comparison across groups) (Lipkus, 2007; Lipkus and Hollands, 1999). One final note: Just because consumers or patients prefer some graphs does not necessarily mean they will understand them better than nonpreferred graphs. For an excellent systematic review of the use of graphs in health communication (that did not focus on numeracy, however), see Ancker et al. (2006).
Provide Evaluative Meaning or Highlight Meaning
Some of the approaches recommended above lower cognitive effort by providing cues to transform the information to an evaluative good/bad scale (Hsee, 1996, 1998). Instead of having to think hard about how to evaluate the goodness or badness of information about an option, an evaluable display reduces the analytical effort required by providing these evaluations in a simpler form. It also may motivate further information processing and behaviors (Peters et al., 2009). The concept of evaluability is simple but profound. Information varies in the degree to which it conveys evaluative meaning. Particularly in unfamiliar domains, we may not know what a measure means (e.g., a measure of quality of care, expressed by the percentage of people satisfied with their care). Research on evaluability demonstrates that even if we understand the numbers used (e.g., a medication that has a 2 percent elevated risk of stroke) at some fundamental level, we may not have an emotional or affective understanding of it (e.g., we do not know how bad this elevated risk is). When information lacks emotional meaning, it lacks evaluability and is not weighted properly in decision making (Slovic et al., 2002). We can determine meaning through considerable effort in comparing and contrasting available information; this is especially true for the highly numerate (Peters et al., 2006a). However, it appears that consumers do not always go to this extra effort and may rely instead on information that is a priori more evaluable. In health contexts, for example, money may be one of the variables that is most evaluable and easily understood; other important variables such as quality-of-care measures are less evaluable and, thus, are less weighted in choice despite their importance to the long-term quality of our health care system. As we will review, however, information evaluability can be improved in a variety of ways. By improving evaluability, we can lower the effort required of the analytical system and highlight the meaning of the information at the same time.
Altering the evaluability of information means that we can help consumers transform data into meaningful information and, by so doing, affect
the degree to which the information is actually used in choice (Hibbard et al., 2002). These evaluability changes make all of the information about a choice available in a simple good/bad form (so that consumers can compare apples to apples). This simpler information then influences the interpretation and comprehension of information about the choice attributes. By providing information in an explicitly evaluative form, it can be used more easily to evaluate the overall goodness or badness of any one option. Experimental findings indicate that evaluable displays of comparative data influence the degree to which information such as quality of care is actually weighted and used in choice.
Carefully Use Evaluative Labels and Symbols
People making decisions can be quite poor at using numeric information in making decisions. Interpreting the meaning of numeric information (e.g., telling patients how good or bad a 9 percent risk is) can have a robust influence in health judgments and choices across diverse adult populations (Peters et al., 2009). In one series of studies, providing evaluative labels (poor, fair, good, and excellent) with numeric quality-of-care information resulted in its greater use in judgments and less reliance on an irrelevant affective state among the less numerate. Follow-up studies in this paper demonstrated that consumers given evaluative labels processed the numeric information (and did not ignore it due to the presence of labels). Instead, the evaluative labels appeared to increase the relative accessibility of valenced feelings about the choice options over valenced thoughts about the same options. In another study, evaluative labels for test results (that a test came back “positive” or “abnormal”) induced larger changes to risk perceptions and behavioral intentions than did numeric results alone (Zikmund-Fisher et al., 2007). The normative appropriateness of changes in this latter study were unclear, however, thus highlighting that evaluative labels should be applied with great care.
Carefully Use Frequency Versus Percentage Formats
The choice between frequencies and percentages can affect people’s perceptions of provided information, especially risk information (Slovic et al., 2000). For instance, Peters and colleagues asked participants to imagine they had severe headaches and that a medicine existed that could decrease headache frequency (Peters et al., 2011). Participants read about a possible side effect of the drug in a percentage format (10 percent of patients get a blistering rash) or in a frequency format (10 patients out of 100 get a blistering rash). Less numerate participants (but not the highly numerate) perceived the medicine as less risky when side effect information was pre-
sented using percentages rather than frequencies. Peters et al. interpreted their results as being due to the frequency formats eliciting greater emotional imagery compared to percentage formats (thought to be perceived as relatively abstract and meaningless). Because information providers have to choose some format to provide likelihood information (and no format is neutral), they should think carefully about whether they would recommend taking a medication that has a possible side effect (in which case, they should use a percentage format in conveying possible risks) or they think the patient should seriously consider the side effect (they might use a frequency format instead). The choice of format will make little difference to the highly numerate, but will matter to the less numerate (see also Dieckmann et al., 2009; Peters et al., 2006b).
Use Other More Imaginable Data Formats
Just as data presented in a frequentistic format may be easier (and more emotional) to imagine than presented in a probabilistic format, changes in life expectancy appear easier to imagine than changes in disease risk. Galesic and Garcia-Retamero (2011) found that, when information about consequences of risky behaviors was presented as months of life lost or gained, recall was better than when it was presented in terms of risks of a disease. The effect held for both short-term and longer-term memory for the information and for individuals higher and lower in numeracy. The improved recall seemed to be due to better imaginability of changes in life expectancy. These results are consistent with recent research demonstrating an effect of displaying the minutes of brisk walking needed to burn calories for menu items (as opposed to having only calorie counts) on how many calories were ordered and consumed (James et al., 2013).
Use Emotion to Persuade
Diverse studies have demonstrated that affective reactions are powerful sources of information when deriving perceptions of risk (Loewenstein et al., 2001; Slovic et al., 2004). Emotional manipulations can influence risk evaluations (Loewenstein et al., 2001; Slovic et al., 2004) and increase thoughts about behavioral change (Diefenbach et al., 1999; Romer and Jamieson, 2001). Tobacco, for example, is the leading cause of preventable death worldwide, killing one person every 6 seconds (CDC, 2012; WHO, 2012). To combat this epidemic, some countries have implemented health warnings on the front and back of cigarette packages that include basic statements of health risks (e.g., “smoking kills”) and large graphic images illustrating the risks. In contrast to basic text-only warnings, which are forgettable and ineffective (Bansal-Travers et al., 2011; Borland et al., 2009;
Hammond et al., 2007; Moodie et al., 2009), graphic pictorial warnings create negative affect toward smoking (Peters et al., 2007c) and encourage smokers with those reactions to think about quitting (Hammond, 2011; White et al., 2008). It is thought that the graphic labels may have greater effects among less educated (including less numerate) populations. Health care providers should consider the use of emotion such as with graphic verbal or visual representations in situations where persuasion is an acceptable tool.
Draw Attention to Important Information
Order Information So That the Most Important Information Is First or Last
Ordering information can help consumers by reducing the cognitive effort required to locate and understand the goodness or badness of information and by drawing attention to important information. Hibbard et al. (2002), for example, found that ordering health plans by performance within premium cost strata resulted in more choices of higher performing plans compared with presenting the information unordered. It is not clear from the literature whether ordering might have a differential effect based on consumer numeracy level, but it is likely that the effect is larger among the less numerate, who generally have more difficulty understanding the meaning of numeric information.
Highlight the Meaning of Only the Most Important Information
In Peters et al. (2007a), making only a more important quality measure easier to evaluate through the use of evaluative symbols such as those used by Consumer Reports (rather than making all indicators easier to evaluate) led to more choices of higher quality hospitals. These results were particularly strong among the less numerate. When the meaning of nonessential information is highlighted (along with more important information being highlighted), it may actually worsen health choices among those with lower numeracy.
Use a Framework to Provide an Overview
Greene et al. (2008) examined consumer understanding and use of information when making a choice between a more familiar type of health plan and a less familiar one. They found that less numerate consumers understood less of the information provided about the new type of health plan at the same time as they were substantially more likely to choose it.
Providing an overarching framework to explain and highlight the differences between the two types of health plans boosted comprehension of items related to the framework message. However, it reduced comprehension of items not related to the framework, particularly among the less numerate. The study highlighted the difficulty that many consumers, and especially the less numerate, have in understanding comparative plan information and in making informed health care choices similar to what will be provided as a result of the ACA. Providing a framework can help, but information providers will need to take care that all important information is mentioned in the framework (with the more detailed information following the framework) to ensure comprehension among the less numerate.
Use Fonts That Draw Attention to Important Information
One reason that health information may not be used is because consumers never attended to it in the first place. With numeric information, this may be particularly true for less numerate consumers (see review of attention effects in the section on emergent decision-based numeracy skills). Methods can be used, however, to explicitly draw attention to numeric information in these cases. Stimuli that are perceptually salient draw attention (Parkhurst et al., 2002) and tend to have greater influence on choice (Bettman et al., 1998). For example, in a men’s clothing store, a red tie placed in a display of neutrally colored ties may capture attention and be chosen more often than the same red tie in a display of vibrant colors.
The visual salience of health information can be manipulated in a variety of ways, including through larger or bold fonts. In an unpublished dissertation, for example, Sagara (2009) found that participants were more sensitive to different levels of numeric information when the numbers were printed in a font that contrasted more with other information provided. In particular, numeric product information that was italicized and printed in grey (in contrast with the regular black font of the surrounding information) appeared to increase the salience of the numeric information, and to result in a greater impact of the numbers on participants’ product judgments. In two studies in an unpublished master’s thesis, Meilleur (2012) varied the risks associated with a vaccination and the font size in which the risks were printed to increase salience and draw attention to the risks. Meilleur found that increasing the font size of the numeric risk information drew participants’ attention toward it, increased their sensitivity to risk, and altered vaccination decisions.
Set Up Appropriate Systems
Identify the Goals of the Communication
To communicate effectively, communicators (whether health care providers or insurance providers) need to identify the goal or goals of a communication and what information the decision maker needs to receive. Without this identification of what matters and to whom, communication efforts will be inadequate. For the previously uninsured population, low numeracy is likely to be an issue. Communication efforts (how to present information) should address this issue in an evidence-based manner. For each type of decision the previously uninsured population will need to make, effective communication will depend in part on identifying information that is more and less important and identifying options that are dominated and dominant. Doing so will allow communicators to take some of the recommended steps to reduce cognitive effort, highlight evaluative meaning, and draw attention to important information in ways that facilitate appropriate comprehension and use of numeric information.
Information Presentation Formats
Communication should be viewed as a strategic process that begins with identifying which information the patient or consumer should know and use or wants to know and use. Then information presentation should proceed in an evidence-based manner to best reach the identified communication goals. One of the most important points is that communications should be tested prior to their use and in appropriate populations (e.g., in a less numerate population if that is the ultimate target for the communication).
Default Options and Other Choice Architecture
If a health provider wants to promote behavior change (as opposed to simply informing a patient or consumer), the notion of choice architecture offers alternative approaches to promoting better health decisions. Choice architecture is a term coined by Thaler and Sunstein (2008) that reflects the fact that many ways exist to present a choice to decision makers, and that what is chosen often depends on how the choice is presented. Although few of these tools have been examined with respect to individual differences such as numeracy, they hold some promise. Johnson et al. (2012) provide a brief review that identifies, describes, and categorizes some of the many tools that could be tested within health environments.
One of the primary tools tested thus far is the use of default options.
Defaults are choice options that are chosen a priori by policy makers and that are applied to individuals who do not take active steps to change away from them (Brown and Krishna, 2004). The default is “chosen” if the consumer does nothing. These are already in wide use; consider, for example, a physician who has a recommended treatment. She usually just writes out the appropriate prescription at that point, although the patient could continue to discuss alternative treatments. Defaults have been shown to have strong effects on choices concerning investments (Cronqvist and Thaler, 2004; Madrian and Shea, 2001), insurance (Johnson et al., 2003), and organ donation (Johnson and Goldstein, 2003). They appeal to a wide audience in their ability to guide choice while preserving freedom of choice. In another example, providing calorie information has not consistently improved individuals’ food choices. However, providing healthful default options on a menu has significantly increased choices of lower calorie foods (Wisdom et al., 2010). Greater use of defaults may be particularly useful in health insurance selection to encourage enrollment and, if defaults are carefully selected, result in consumers who are more likely to be satisfied with their choice.
Computer-Aided Decision Tools
Health care and related providers do not need to be the sole communicators with the ACA population. Many of the same strategies (e.g., reducing cognitive burden and highlighting meaning) can be accomplished through the use of carefully designed, computer-aided decision tools. Use of such tools can structure and simplify the decision process at the same time as important factors and trade-offs are highlighted for consideration. Calculators (e.g., for health plan costs for those needing a lot of health care because of chronic disease or those expecting few health care costs) can be built into such tools or can be provided as stand-alone tools. Such strategies may be quite important given the small proportion of the ACA population expected to have Proficient levels of quantitative literacy and to be able to perform such calculations (see Table A-3).
Individuals sometimes require greater assistance, particularly individuals with less computer experience, lower numeracy, and other limitations with respect to health literacy. An information intermediary can perform a similar, but more personalized, function to computer-aided decision tools.
The expected influx of previously uninsured individuals into our nation’s health care system will present a variety of challenges, including the challenges of communicating with less numerate individuals who have limited knowledge and abilities to navigate this unfamiliar and often numeric world. This population will vary considerably in education-based numeracy skills (from basic arithmetic to understanding cumulative risk) and emergent decision-based numeracy skills (from seeking out numeric information to deriving affective meaning from it). Providers have an opportunity in the coming months and years to better understand who these people are (in terms of their abilities) and to apply the science of communication to help these patients and consumers make informed decisions and maximize their health and well-being given new ACA benefits.
Abaluck, J., and J. Gruber. 2011. Choice inconsistencies among the elderly: Evidence from Plan Choice in the Medicare Part D program. American Economic Review 101(4):1180-1210.
Ancker, J. S., and D. R. Kaufman. 2007. Rethinking health numeracy: A multidisciplinary literature review. Journal of the American Medical Informatics Association 14(6):713-721.
Ancker, J. S., Y. Senathirajah, R. Kukafka, and J. B. Starren. 2006. Design features of graphs in health risk communication: A systematic review. Journal of the American Medical Informatics Association 13:608-618.
Anderson, B. L., N. A. Obrecht, G. B. Chapman, D. A. Driscoll, and J. Schulkin. 2011. Physicians’ communication of Down syndrome screening test results: The influence of physician numeracy. Genetics in Medicine 13:744-749.
Apter, A. J., J. Cheng, D. Small, I. M. Bennett, C. Albert, D. G. Fein, M. George, and S. Van Horne. 2006. Asthma numeracy skill and health literacy. Journal of Asthma 43:705-710.
Apter, A. J., M. K. Paasche-Orlow, J. T. Remillard, I. M. Bennett, E. P. Ben-Joseph, R. M. Batista, J. Hyde, and R. E. Rudd. 2008. Numeracy and communication with patients: They are counting on us. Journal of General Internal Medicine 23(12):2117-2124.
Bansal-Travers, M., D. Hammond, P. Smith, and M. Cummings. 2011. The impact of cigarette pack design, descriptors, and warning labels on risk perception in the U.S. American Journal of Preventative Medicine 40(6):674-682.
Baron, J. 1997. Confusion of relative and absolute risk in valuation. Journal of Risk and Uncertainty 14:301-309.
Bateman, I., S. Dent, E. Peters, P. Slovic, and C. Starmer. 2007. The affect heuristic and the attractiveness of simple gambles. Journal of Behavioral Decision Making 20:365-380.
Ben-Joseph, E. P., S. A. Dowshen, and N. Izenberg. 2009. Do parents understand growth charts? A national, Internet-based survey. Pediatrics 124:1100-1109.
Berkman, N. D., S. L. Sheridan, K. E. Donahue, D. J. Halpern, A. Viera, K. Crotty, A. Holland, M. Brasure, K. N. Lohr, E. Harden, E. Tant, I. Wallace, and M. Viswanathan. 2011. Health literacy interventions and outcomes: An updated systematic review. Evidence Report/Technology Assessment No. 199. (Prepared by RTI International–University of North Carolina Evidence-based Practice Center under contract No. 290-2007-10056-I.) AHRQ Pub. No. 11-E006. Rockville, MD: Agency for Healthcare Research and Quality.
Berry, D. C. 2006. Informing people about the risks and benefits of medicines: Implications for the safe and effective use of medicinal products. Current Drug Safety 1:121-126.
Berry, D. C., P. R. Knapp, and D. K. Raynor. 2002. Provision of information about drug side effects to patients. Lancet 359:853-854.
Berry, D. C., P. Knapp, and D. K. Raynor. 2003a. Communicating risk of medication side effects: An empirical evaluation of EU recommended terminology. Psychology, Health & Medicine 8(3):251-263.
Berry, D. C., D. K. Raynor, P. Knapp, and E. Bersellini. 2003b. Patients’ understanding of risk associated with medication use: Impact of European Commission Guidelines and other risk scales. Drug Society 26(1):1-11.
Bettman, J. R., M. F. Luce, and J. W. Payne. 1998. Constructive consumer choice processes. Journal of Consumer Research 25(3):187-217.
Borland, R., N. Wilson, G. T. Fong, D. Hammond, K. M. Cummings, H. H. Yong, W. Hosking, G. Hastings, J. Thrasher, and A. McNeill. 2009. Impact of graphic and text warnings on cigarette packs: Findings from four countries over five years. Tobacco Control 18(5):358-364.
Botti, S., and A. L. McGill. 2006. When choosing is not deciding: The effect of perceived responsibility on satisfaction. Journal of Consumer Research 33:211-219.
Brown, C., and A. Krishna. 2004. The skeptical shopper: A metacognitive account for the effects of default options on choice. Journal of Consumer Research 31(3):529-539.
CDC (Centers for Disease Control and Prevention). 2012. Smoking & tobacco use: Fast facts. http://www.cdc.gov/tobacco/data_statistics/fact_sheets/fast_facts (accessed June 25, 2014).
Chao, C., J. L. Studts, T. Abell, T. Hadley, L. Roetzer, S. Dineen, D. Lorenz, A.Y. Agha, and K. M. McMasters. 2003. Adjuvant chemotherapy for breast cancer: How presentation of recurrence risk influences decision-making. Journal of Clinical Oncology 21:4299-4305.
CMS (Centers for Medicare & Medicaid Services). 2013. Medicare provider charge data. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-andReports/Medicare-Provider-Charge-Data/index.html (accessed June 25, 2014).
Cokely, E. T., M. Galesic, E. Schulz, S. Ghazal, and R. Garcia-Retamero. 2012. Measuring risk literacy: The Berlin Numeracy Test. Judgment and Decision Making 7(1):25-47.
Cronqvist, H., and R. H. Thaler. 2004. Design choices in privatized social-security systems: Learning from the Swedish experience. American Economic Review 94(2):424-428.
Damasio, A. R. 1994. Descartes’ error: Emotion, reason, and the human brain. New York: Avon.
Day, R., and P. Nadash. 2012. New state insurance exchanges should follow the example of Massachusetts by simplifying choices among health plans. Health Affairs 31(5):982-989.
Denes-Raj, V., and S. Epstein. 1994. Conflict between intuitive and rational processing: When people behave against their better judgment. Journal of Personality and Social Psychology 66:819-829.
Dhar, R. 1997. Consumer preference for a no-choice option. Journal of Consumer Research 24:215-231.
Dieckmann, N. F., P. Slovic, and E. M. Peters. 2009. The use of narrative evidence and explicit likelihood by decisionmakers varying in numeracy. Risk Analysis 29(10):1473-1488.
Diefenbach, M. A., S. M. Miller, and M. Daly. 1999. Specific worry about breast cancer predicts mammography use in women at risk for breast and ovarian cancer. Health Psychology 18:532-536.
Epstein, S. 1994. Integration of the cognitive and the psychodynamic unconscious. American Psychologist 49:709-724.
Estrada, C. A., M. Martin-Hryniewicz, B. T. Peek, C. Collins, and J. C. Byrd. 2004. Literacy and numeracy skills and anticoagulation control. American Journal of the Medical Sciences 328(2):88-93.
Fagerlin, A., and E. Peters. 2011. Quantitative information. In Evidence-based communication of risk and benefits: A user’s guide, edited by B. Fischhoff, N. Brewer, and J. Downs. Silver Spring, MD: Food and Drug Administration. Pp. 53-64.
Fagerlin, A., C. Wang, and P. A. Ubel. 2005a. Reducing the influence of anecdotal reasoning on people’s health care decisions: Is a picture worth a thousand statistics? Medical Decision Making 25(4):398-405.
Fagerlin, A., B. J. Zikmund-Fisher, and P. A. Ubel. 2005b. How making a risk estimate can change the feel of that risk: Shifting attitudes toward breast cancer risk in a general public survey. Patient Education and Counseling 57:294-299.
Fagerlin, A., P. A. Ubel, D. M. Smith, and B. J. Zikmund-Fisher. 2007a. Making numbers matter: Present and future research in risk communication. American Journal of Health Behavior 31:47-56.
Fagerlin, A., B. J. Zikmund-Fisher, P. A. Ubel, A. Jankovic, H. A. Derry, and D. M. Smith. 2007b. Measuring numeracy without a math test: Development of the Subjective Numeracy Scale. Medical Decision Making 27(5):672-680.
Finucane, M. L., A. Alhakami, P. Slovic, and S. M. Johnson. 2000. The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making 13:1-17.
Forrow, L., W. C. Taylor, and R. M. Arnold. 1992. Absolutely relative: How research results are summarized can affect treatment decisions. American Journal of Medicine 92:121-124.
Galesic, M., and R. Garcia-Retamero. 2010. Statistical numeracy for health: A cross-cultural comparison with probabilistic national samples. Archives of Internal Medicine 170(5):462-468.
Galesic, M., and R. Garcia-Retamero. 2011. Communicating consequences of risky behaviors: Life expectancy versus risk of disease. Patient Education and Counseling 82:30-35.
Garbarino, E. C., and J. A. Edell. 1997. Cognitive effort, affect, and choice. Journal of Consumer Research 24:147-158.
Garcia-Retamero, R., and E. T. Cokely. 2011. Effective communication of risks to young adults: Using message framing and visual aids to increase condom use and STD screening. Journal of Experimental Psychology: Applied 17:270-287.
Garcia-Retamero, R., M. Galesic, and G. Gigerenzer. 2010. Do icon arrays help reduce denominator neglect? Medical Decision Making 30:672-684.
Gigerenzer, G. 2002. Calculated risks: How to know when numbers deceive you. New York: Simon & Schuster.
Greene, J., E. Peters, C. K. Mertz, and J. H. Hibbard. 2008. Comprehension and choice of a consumer-directed health plan: An experimental study. American Journal of Managed Care 14(6):369-376.
Hammond, D. 2011. Health warnings on tobacco packages: A review. Tobacco Control 20:327-337.
Hammond, D., F. Wiebel, L. T. Kozlowski, R. Borland, K. M. Cummings, R. J. O’Connor, A. McNeill, G. N. Connolly, D. Arnott, and G. T. Fong. 2007. Revising the machine smoking regime for cigarette emissions: Implications for tobacco control policy. Tobacco Control 16(1):8-14.
Hanoch, Y., T. Rice, J. Cummings, and S. Wood. 2009. How much choice is too much? The case of the Medicare prescription drug benefit. Health Services Research 44(4):1157-1168.
Hibbard, J. H., and E. Peters. 2003. Supporting informed consumer health care choices: Data presentation approaches that facilitate the use of information in choice. Annual Review of Public Health 24:413-433.
Hibbard, J. H., P. Slovic, E. M. Peters, M. Finucane, and M. Tusler. 2001. Is the informed-choice approach appropriate for Medicare beneficiaries? Health Affairs 20(3):199-203.
Hibbard, J. H., P. Slovic, E. Peters, and M. L. Finucane. 2002. Strategies for reporting health plan performance information to consumers: Evidence from controlled studies. Health Services Research 37(2):291-313.
Hibbard, J. H., E. Mahoney, R. Stock, and M. Tusler. 2007a. Do increases in patient activation result in improved self-management behaviors? Health Services Research 42(4): 1443-1463.
Hibbard, J. H., E. Peters, A. Dixon, and M. Tusler. 2007b. Consumer competencies and the use of comparative quality information: It isn’t just about literacy. Medical Care Research & Review 64(4):379-394.
Hsee, C. K. 1996. The evaluability hypothesis: An explanation for preference reversals between joint and separate evaluations of alternatives. Organizational Behavior and Human Decision Processes 67:247-257.
Hsee, C. K. 1998. Less is better: When low-value options are valued more highly than high-value options. Journal of Behavioral Decision Making 11:107-121.
Huffman, C., and B. E. Kahn. 1998. Variety for sale: Mass customization or mass confusion? Journal of Retailing 74:491-513.
Huizinga, M. M., T. A. Elasy, K. A. Wallston, K. Cavanaugh, D. Davis, R. P. Gregory, L. S. Fuchs, R. Malone, A. Cherrington, D. A. DeWalt, J. Buse, M. Pignone, and R. L. Rothman. 2008. Development and validation of the Diabetes Numeracy Test (DNT). BMC Health Services Research 8:96.
Iyengar, S. S., and M. R. Lepper. 2000. When choice is demotivating: Can one desire too much of a good thing? Journal of Personality and Social Psychology 79:995-1006.
Iyengar, S. S., G. Huberman, and W. Jiang. 2004. How much choice is too much: Determinants of individual contributions in 401K retirement plans. In Pension design and structure: New lessons from behavioral finance, edited by O. S. Mitchell and S. Utkus. Oxford, England: Oxford University Press. Pp. 83-95.
James, A., B. Adams-Huet, K. Crisp, J. Mitchell, L. Dart, M. Turner, M. Kasper, J. Bowman, S. Joeckel, N. Toomey, H. Heefner, E. Blasco, and M. Shah. 2013. The effect of menu labels, displaying minutes of brisk walking needed to burn food calories, on calories ordered and consumed in young adults. Journal of the Federation of American Societies for Experimental Biology 27:367.2.
Johnson, E. J., and D. G. Goldstein. 2003. Do defaults save lives? Science 302:1338-1339.
Johnson, E. J., S. B. Shu, B. G. C. Dellaert, C. Fox, D. G. Goldstein, G. Häubl, R. P. Larrick, J. W. Payne, E. Peters, D. Schkade, B. Wansink, and E. U. Weber. 2012. Beyond nudges: Tools of a choice architecture. Marketing Letters 23:487-504.
Johnson, R. W., A. J. Davidoff, and K. Perese. 2003. Health insurance costs and early retirement decisions. Industrial and Labor Relations Review 56(4):716-729.
Kahneman, D. 2003. A perspective on judgment and choice: Mapping bounded rationality. American Psychologist 58(9):697-720.
Kaminski, J. A., and V. M. Sloutsky. 2013. Extraneous perceptual information interferes with children’s acquisition of mathematical knowledge. Journal of Educational Psychology 105(2):351-363.
Keller, C. 2011. Using a familiar risk comparison within a risk ladder to improve risk understanding by low numerates: A study of visual attention. Risk Analysis 31:1043-1054.
Kutner, M., E. Greenberg, Y. Jin, B. Boyle, Y. Hsu, and E. Dunleavy. 2007. Literacy in everyday life: Results from the 2003 National Assessment of Adult Literacy (NAAL). National Center for Education Statistics. Institute of Education Sciences. http://nces.ed.gov/naal (accessed June 25, 2014).
Lipkus, I. M. 2007. Numeric, verbal, and visual formats of conveying health risks: Suggested best practices and future recommendations. Medical Decision Making 27:696-713.
Lipkus, I. M., and J. G. Hollands. 1999. The visual communication of risk. Journal of the National Cancer Institute Monographs 9(25):149-163.
Lipkus, I. M., and E. Peters. 2009. Understanding the role of numeracy in health: Proposed theoretical framework and practical insights. Health Education and Behavior 36(6):1065-1081.
Lipkus, I. M., G. Samsa, and B. K. Rimer. 2001. General performance on a numeracy scale among highly educated samples. Medical Decision Making 21:37-44.
Lipkus, I. M., E. Peters, G. Kimmick, V. Liotcheva, and P. Marcom. 2010. Breast cancer patients’ treatment expectations after exposure to the decision aid program Adjuvant Online: The influence of numeracy. Medical Decision Making 30:464-473.
Loewenstein, G. F., E. U. Weber, C. K. Hsee, and E. S. Welch. 2001. Risk as feelings. Psychological Bulletin 127:267-286.
Madlon-Kay, D. J., and F. S. Mosch. 2000. Liquid medication dosing errors. Journal of Family Practice 49(8):741-744.
Madrian, B. C., and D. F. Shea. 2001. The power of suggestion: Inertia in 401(k) participation and savings behavior. Quarterly Journal of Economics 116(4):1149-1187.
Malenka, D. J., J. A. Baron, S. Johansen, J. W. Wahrenberger, and J. M. Ross. 1993. The framing effect of relative and absolute risk. Journal of General Internal Medicine 8(10): 543-548.
McCaffery, K. J., A. Dixon, A. Hayen, J. Jansen, S. Smith, and J. M. Simpson. 2012. The influence of graphic display format on the interpretations of quantitative risk information among adults with lower education and literacy: A randomized experimental study. Medical Decision Making 32(4):532-544.
Meilleur, L. R. 2012. Manipulating attention to improve health behaviors. Master’s thesis. Ohio State University. Retrieved from OhioLink ETD. Pub. No. OSU1354291552.
Moodie, C., A. M. MacKintosh, and D. Hammond. 2009. Adolescents’ response to text-only tobacco health warnings: Results from the 2008 UK Youth Tobacco Policy Survey. European Journal of Public Health 20(4):463-469.
Parkhurst, D., K. Law, and E. Niebur. 2002. Modeling the role of salience in the allocation of overt visual attention. Vision Research 42:107-123.
Peters, E. 2012. Beyond comprehension: The role of numeracy in judgments and decisions. Current Directions in Psychological Science 21(1):31-35.
Peters, E., D. Västfjäll, T. Gärling, and P. Slovic. 2006a. Affect and decision making: A “hot” topic. Journal of Behavioral Decision Making 19(2):79-85.
Peters, E., D. Västfjäll, P. Slovic, C. K. Mertz, K. Mazzocco, and S. Dickert. 2006b. Numeracy and decision making. Psychological Science 17(5):407-413.
Peters, E., N. Dieckmann, A. Dixon, J. H. Hibbard, and C. K. Mertz. 2007a. Less is more in presenting quality information to consumers. Medical Care Research & Review 64(2):169-190.
Peters, E., J. Hibbard, P. Slovic, and N. Dieckmann. 2007b. Numeracy skill and the communication, comprehension, and use of risk-benefit information. Health Affairs 26(3):741-748.
Peters, E., D. Romer, P. Slovic, K. H. Jamieson, L. Wharfield, C. K. Mertz, and S. M. Carpenter. 2007c. The impact and acceptability of Canadian-style cigarette warning labels among U.S. smokers and nonsmokers. Nicotine & Tobacco Research 9(4):473-481.
Peters, E., N. F. Dieckmann, D. Västfjäll, C. K. Mertz, P. Slovic, and J. H. Hibbard. 2009. Bringing meaning to numbers: The impact of evaluative categories on decisions. Journal of Experimental Psychology: Applied 15(3):213-227.
Peters, E., S. Hart, and L. Fraenkel. 2011. Informing patients: The influence of numeracy, framing, and format of side effect information on risk perceptions. Medical Decision Making 31(3):432-436.
Peters, E., H. Kunreuther, N. Sagara, P. Slovic, and D. R. Schley. 2012. Protective measures, personal experience, and the affective psychology of time. Risk Analysis 32(12):2084-2097.
Peters, E., S. Hart, M. Tusler, and L. Fraenkel. In review. Numbers matter to informed patient choices: The effects of age and numeracy.
Quincy, L. 2012. State insurance exchanges’ impact on consumers. In Facilitating state health exchange communication through the use of health literate practices: Workshop summary. Washington, DC: The National Academies Press. Pp. 27-47.
Ratzan, S. C., and R. M. Parker. 2000. Introduction. In National Library of Medicine current bibliographies in medicine: Health literacy, edited by C. R. Selden, M. Zorn, S. C. Ratzan, and R. M. Parker. Bethesda, MD: National Institutes of Health.
Ravdin, P. M., L. A. Siminoff, G. J. Davis, M. B. Mercer, J. Hewlett, N. Gerson, and H. L. Parker. 2001. Computer program to assist in making decisions about adjuvant therapy for women with early breast cancer. Journal of Clinical Oncology 19(4):980-991.
Reutskaja, E., and R. M. Hogarth. 2009. Satisfaction in choice as a function of the number of alternatives: When “goods satiate.” Psychology & Marketing 26(3):197-203.
Reyna, V. F. 2004. How people make decisions that involve risk: A dual-processes approach. Current Directions in Psychological Science 13:60-66.
Reyna, V. F., W. L. Nelson, P. K. Han, and N. F. Dieckmann. 2009. How numeracy influences risk comprehension and medical decision making. Psychological Bulletin 135:943-973.
Romer, D., and P. Jamieson. 2001. Do adolescents appreciate the risks of smoking?: Evidence from a national survey. Journal of Adolescent Health 29:12-21.
Rosenthal, J. A., X. Lu, and P. Cram. 2013. Availability of consumer prices from US hospitals for a common surgical procedure. JAMA Internal Medicine 173(6):427-432.
Sagara, N. 2009. Consumer understanding and use of numeric information in product claims. Doctoral dissertation. University of Oregon. Retrieved from ProQuest. Pub. No. AAT 3395194.
Sakr, W. A., D. J. Gringon, G. P. Hass, L. K. Heilbrun, J. E. Pontes, and J. D. Crissman. 1996. Age and racial distribution of prostatic intraepithelial neoplasia. European Urology 30(2):138-144.
Scammon, D. L. 1977. Information load and consumers. Journal of Consumer Research 4:148-155.
Schwartz, B. 2000. Self-determination: The tyranny of freedom. American Psychologist 55: 79-88.
Schwartz, B. 2004. The paradox of choice. New York: Harper Collins Publishers.
Schwartz, B. 2005. Choose and lose. New York Times, January 5.
Schwartz, B., A. Ward, J. Monterosso, S. Lyubomirsky, K. White, and D. R. Lehman. 2002. Maximizing versus satisficing: Happiness is a matter of choice. Journal of Personality and Social Psychology 83(5):1178-1197.
Schwartz, P. H. 2011. Questioning the quantitative imperative: Decision aids, prevention, and the ethics of disclosure. Hastings Center Report 41(2):30-39.
Shapira, M. M., C. M. Walker, K. J. Cappaert, P. S. Ganschow, K. E. Fletcher, E. L. McGinley, S. Del Pozo, C. Schauer, S. Tarima, and E. A. Jacobs. 2012. The Numeracy Understanding in Medicine instrument (NUMi): A measure of health numeracy developed using Item Response Theory. Medical Decision Making 32:851-865.
Sheridan, S. L., and M. Pignone. 2002. Numeracy and the medical student’s ability to interpret data. Effective Clinical Practice 5:35-40.
Shiv, B., and A. Fedorikhin. 1999. Heart and mind in conflict: The interplay of affect and cognition in consumer decision making. Journal of Consumer Research 26(3):278-292.
Sloman, S. A. 1996. The empirical case for two systems of reasoning. Psychological Bulletin 119:3-22.
Slovic, P., J. Monahan, and D. G. MacGregor. 2000. Violence risk assessment and risk communication: The effects of using actual cases, providing instructions, and employing probability versus frequency formats. Law and Human Behavior 24:271-296.
Slovic, P., M. Finucane, E. Peters, and D. G. MacGregor. 2002. Rational actors or rational fools: Implications of the affect heuristic for behavioral economics. Journal of Socio-Economics 31(4):329-342.
Slovic, P., M. L. Finucane, E. Peters, and D. G. MacGregor. 2004. Risk as analysis and risk as feelings: Some thoughts about affect, reason, risk, and rationality. Risk Analysis 24(2): 311-322.
Slovic, P., E. Peters, M. L. Finucane, and D. G. MacGregor. 2005. Affect, risk, and decision making. Health Psychology 24:S35-S40.
Stanovich, K. E., and R. F. West. 2002. Individual differences in reasoning: Implications for the rationality debate? In Heuristics and biases: The psychology of intuitive judgment, edited by T. Gilovich, D. W. Griffin, and D. Kahneman. New York: Cambridge University Press. Pp. 421-440.
Thaler, R. H., and C. R. Sunstein. 2008. Nudge: Improving decisions about health, wealth and happiness. New Haven, CT: Yale University Press.
U.S. Census Bureau. 2009-2011. Selected characteristics of the uninsured in the United States 2009-2011. http://factfinder2.census.gov (accessed June 25, 2014).
U.S. Census Bureau. 2012. Health insurance: Highlights. http://www.census.gov/hhes/www/hlthins/data/incpovhlth/2011/highlights.html (accessed June 25, 2014).
USPSTF (U.S. Preventive Services Task Force). 2011, October. Screening for prostate cancer: A review of the evidence for the U.S. Preventive Services Task Force. http://www.uspreventiveservicestaskforce.org/uspstf/uspsprca.htm (accessed June 25, 2014).
Västfjäll, D., E. Peters, and C. Starmer. In preparation. Numeracy, incidental affect and the construction of prices.
Welch, G., L. Schwartz, and S. Woloshin. 2011. Overdiagnosed: Making people sick in the pursuit of health. Boston, MA: Beacon Press.
Weller, J., N. F. Dieckmann, M. Tusler, C. K. Mertz, W. Burns, and E. Peters. 2013. Development and testing of an abbreviated numeracy scale: A Rasch Analysis approach. Journal of Behavioral Decision Making 26(2):198-212.
White, V., B. Webster, and M. Wakefield. 2008. Do graphic health warning labels have an impact on adolescents’ smoking-related beliefs and behaviours? Addiction 103(9):1562-1571.
WHO (World Health Organization). 2012. Tobacco. http://www.who.int/mediacentre/factsheets/fs339/en/index.html (accessed June 25, 2014).
Wisdom, J., J. S. Downs, and G. Loewenstein. 2010. Promoting healthy choices: Information versus convenience. American Economic Journal: Applied Economics 2:164-178.
Zikmund-Fisher, B. J., A. Fagerlin, K. Keeton, and P. A. Ubel. 2007. Does labeling prenatal screening test results as negative or positive affect a woman’s responses? American Journal of Obstetrics and Gynecology 197:528.e1-528.e6.
Zikmund-Fisher, B. J., A. Fagerlin, and P. A. Ubel. 2008. Improving understanding of adjuvant therapy options via simpler risk graphics. Cancer 113(12):3382-3390.
Zikmund-Fisher, B. J., A. Fagerlin, and P. A. Ubel. 2010. Risky feelings: Why a 6% risk of cancer doesn’t always feel like 6%. Patient Education and Counseling 81(Suppl):S87-S93.
ANNEX TO COMMISSIONED PAPER
Estimating Quantitative Literacy Levels in U.S. Uninsured Adults
The percentage of Americans without health insurance in 2011 was 15.7 percent (U.S. Census Bureau, 2012). Using the 2003 National Assessment of Adult Literacy (NAAL) and 2009-2011 Census Bureau data, we calculated an estimate of the proportion of uninsured and insured American adults who fall into Below Basic, Basic, Intermediate, and Proficient quantitative literacy categories. The 2009-2011 Census Bureau provides data on the proportion of uninsured adults at each level of educational attainment (see Table Annex A-1), whereas the 2003 NAAL provides data on the proportion of adults in each quantitative literacy level by highest educational attainment (see Table Annex A-2). The NAAL sample consists of people ages 16 years and older living in households or prisons, whereas the sample of uninsured from the 2009-2011 Census consists of non-institutionalized civilian adults ages 25 and older. Thus, our comparison is imperfect, although it nonetheless gives an idea of the relative difference in quantitative literacy skills in patients and consumers that the health care system sees now (insured adults) and will likely see soon (previously uninsured adults). Additionally, according to the 2009-2011 Census, less than 1 percent (0.8 percent) of the uninsured population is age 65 and older; as a
|Civilian Non-Institutionalized Population 25 Years and Over||U.S. Population||Margin||Uninsured||Margin|
|Less than high school graduate||14.1%||±0.1||27.2%||±0.1|
|High school graduate, GED, or alternative||28.3%||±0.1||34.1%||±0.1|
|Some college or associate’s degree||29.0%||±0.1||26.9%||±0.1|
|Bachelor’s degree or higher||28.6%||±0.1||11.9%||±0.1|
|Educational Attainment||Below Basic||Basic||Intermediate||Proficient||Total|
|Less than/some high school||64.0%||25.0%||10.0%||1.0%||100.0%|
|High school graduate||24.0%||42.0%||29.0%||5.0%||100.0%|
result, our estimated proportions in the uninsured group would not change drastically if we had been able exclude older adults.
Among uninsured adults, we estimated that 28.8 percent are at the Below Basic level, 33.4 percent are at the Basic level, 29.3 percent are at the Intermediate level, and 8.6 percent are at the Proficient level (see Table Annex A-3). We calculated this estimate first by multiplying the proportion of uninsured adults at each level of education attainment (from Table Annex A-1) by the proportion of adults in each quantitative level at every level of education attainment (from Table Annex A-2). Next, we summed the proportions within each quantitative literacy level (across education levels) to get a total estimate of the proportion of uninsured at each level (see Table Annex A-3).
To estimate the proportion of insured adults who fall into each quantitative literacy category, we used the same procedure. We first calculated the proportion of insured adults at each level of education by using Table
|Educational Attainment||Below Basic||Basic||Intermediate||Proficient||Total|
|Less than/some high school||17.4%||6.8%||2.7%||0.3%|
|High school graduate||8.2%||14.3%||9.9%||1.7%|
|% uninsured adults at each quantitative literacy level||28.8%||33.4%||29.3%||8.6%||100.1%|
Annex A-1. We subtracted the number of uninsured adults from the U.S. population for each education level, and then divided that number by the total number of insured adults. Next, we followed the same multiplication and summation computations previously described with the uninsured population, but used the insured proportions.