When proposing a new therapy, rheumatologists must inform patients of a range of therapeutic options and support them towards making an informed decision. This article introduces definitions of equipoise and a good decision, contrasts persuasion from informed patient choice, and discussed the effects of patient characteristics including cognition on decision making. It also describes and offers examples of techniques and visual formats utilized in patient decision aids to present risk estimates to reduce cognitive bias and maximize patient comprehension.
Every day in the clinic rheumatologists guide patients with decisions about initiating complex therapies.
A good decision is informed, consistent with patient values and acted on.
Cognitive biases, in part arising from reduced health literacy, can impair patient decision making.
When sharing risk information, bias and framing effects can be in part accommodated by presenting absolute risks with supportive visual aids and context.
Using decisions aids to structure and support deliberation after the office visit may improve patient decision making.
This monograph discusses the broad landscape of antirheumatic risks that rheumatologists must regularly address with patients. These include minor and serious adverse events, which may occur frequently or only rarely. Individual classes of medications (ie, disease-modifying antirheumatic drugs [DMARDs] or bisphosphonates) have unique risks and the target audience can vary greatly in demographics (ie, pregnancy, children–parental dyad, the elderly) or by circumstance (ie, at the time of surgery) in the setting of concurrent illnesses (ie, hepatitis C or latent tuberculosis). In some cases patient risk may vary based on individual characteristics and sometimes this tailored risk is known. In other cases only a global estimate of risk can be honestly offered. Despite incomplete data, for legal and ethical reasons, physicians must strive to inform patients about potential benefits and harms of medication options.
A good decision is informed, consistent with personal values, and acted on. In an informed choice the decision maker understands relevant information and the choice reflects their values. The capability of an individual patient to deliberate and their approaches to narrowing options and choosing may vary but an informed choice should at least involve the patient in decision making to the extent they desire. This can be assessed with the following questions :
Did the doctor make you aware of the different treatment options?
Do you know the advantages and disadvantages of treatment or not having treatment?
Are you adequately informed about the issues important to the decision?
Has the doctor given you a chance to be involved in the decision?
Do you believed an informed choice has been made?
There can be a gap between believing one is informed and actually being informed. Being informed means knowing the options, being aware of the nature and frequency of the common and most serious medication risks and benefits. In addition, patients must know and be prepared to do the procedures for obtaining, administering, and keeping up with needed safety monitoring.
It is important to differentiate an informed choice from persuasion. Persuasion is a form of communication that is intentional; interpersonal; and involves creating, reinforcing, or modifying beliefs, intentions, motivations, or behaviors. The overall goal is to influence the thoughts or actions of the decision maker. In the case of many pharmaceutical advertisements, this may use symbolic expression, which emphasizes pictures instead of words. In addition, persuasive messages can appeal to fear (ie, of progression of rheumatoid arthritis [RA] joint damage) or protective motivation, which leads the patient to evaluate a health threat and their ability to cope and protect against the undesirable outcome. To illustrate these techniques, Fig. 1 simulates a promotional message for a hypothetical DMARD. Typically, within industry promotional fliers (which are sometimes labeled “decision guides”) there is a summary of risks displayed, but no estimates of the probability of outcomes or reference to other options. Specific information is printed on the “medication guide,” which is tucked in a pocket in the back cover of the brochure.
In contrast, patient decision aids seek to support patients in making an informed choice by disclosing options and relevant information about the consequences of treatment in an accurate, balanced, and understandable manner. Patient decision aids then guide the consumer through steps to clarify the value they place on these consequences. This recognizes that in some decisions there is a need to portray equipoise. This exists where there is a choice among two or more therapeutically equivalent options. Based on medical evidence, the professional may have no preferred recommendation; however, individual treatment attributes or outcomes may be more important or valued by an individual patient. Consider the choice of a bisphosphonate for osteoporosis. Patients must make a trade-off between cost, convenience, hassle of administration, and gastrotoxicity when choosing among a generic oral, once weekly, a branded oral, once monthly, and a branded, intravenous, yearly agent. The expected increase in bone density and fracture risk may be the same; however, inconveniences, costs, and risk of toxicity do vary among the choices. Thus, when equipoise exists, the deciding factors should be the patient’s personal values and preferences related to the potential outcomes of treatment. However, to be aware of one’s preferences, a patient must be informed of treatment options and outcomes, have the time and ability to reflect, and in some cases need structured support to clarify what they believe are the factors that are to them most important.
As patients consider options and choose they may use one or more cognitive strategies. Prospect theory suggests that decision makers carefully weigh multiple attributes of available treatment options before arriving at a rational choice that maximizes benefit. Kahneman, a Nobel laureate in economics, describes that as relying on “System 2,” which is the mode of thinking that directs conscious, deliberate, effortful activities, such as choice. Alternatively, patients may use a heuristic approach, making simpler more cognitively efficient decisions using rules of thumb. Kahneman labels this as directed by “System 1,” which is the mode of thinking that is unconscious, fast, intuitive, and the source of impressions and feelings at the source of many explicit beliefs. Many decisions made by patients and expert physicians are heuristic based. However, the way in which people process information and use personal heuristics can introduce cognitive bias in decision making and lead to suboptimal decisions.
Patient characteristics can also influence decision making. Age and general health can effect concentration and cognition, which are necessary in making good decisions. Health literacy is a measure of multiple domains, including reading; ability to locate and use information; and doing simple mathematical tasks (numeracy). Health literacy is influenced not only by formal education but also by cognitive function including recall and critical thinking. In a survey of a large cohort of community patients with RA it was found that health literacy, independent of low educational achievement or other demographic, was a common predictor of risk perception and willingness to take a proposed DMARD. In addition, it was observed that risk perception was increased by negative RA disease and treatment experience, whereas willingness to take a proposed DMARD was reduced as perception of current RA control improved. This demonstrates the influences of patient disease experiences on decision-making processes. These findings are consistent with the observations of other investigators that individuals with low numeracy may have a higher susceptibility to information-framing effects. In addition, they may be less effective or exert less effort trying to decode medication information and more likely to rely on established heuristics; rules of thumb, such as “if I had a side effect before, it will happen again”; or antidotal reasoning.
Mood is another potentially important patient characteristic. There is evidence that a depressed person’s diminished ability to think and concentrate or indecisiveness may impair ability to participate in decision making. However, in a study of community patients with RA, no significant effect of depression or happiness on risk perception or willingness to take a proposed DMARD was found. This confirms previous observations in a separate RA cohort that history of major depression was not significantly related to patient satisfaction with decision.
Finally, an understudied area is the effect of the doctor-patient interaction on treatment decisions. In a previous study it was found that patients’ trust in their rheumatologist had nearly seven times the effect on their confidence in a DMARD decision than any other predictor including numeric literacy and DMARD-related knowledge. This supports the position that despite physicians’ efforts to educate, many patients make decisions based on factors that are peripheral to the substance of health information and focus on clues, such as physician characteristics of goodwill, expertise, or trustworthiness, when choosing a medication.
If a good decision is dependent on an individual’s ability to understand and evaluate options and to make judgments that are relatively free of bias, then physicians should consider if individual patient characteristics may complicate comprehension. Physicians often overestimate patient literacy and because of unawareness or shame if not questioned, patients may not disclose their limitation. This may justify routine health literacy screening to identify patients who need low literacy materials or consider using enhanced strategies to support patient decision making. Fortunately, there are simple methods, such as screening questions and word recognition tests, that make routine screening for low or marginal health literacy feasible in the clinic.
As physicians consider how to present information about risks of treatment, several general and specific considerations arise. In general, the context and framing of a decision can be important. For example, in the presentation of economic risk, the endowment effect has been observed where individuals are motivated by the possibility of losing more than gaining something of value. This “I would have been” framing is the persuasive approach in the industry medication message cited previously. It is also known as a negativity bias or loss aversion. Interestingly, meta-analysis evaluating such framing studies in health messages has unexpectedly failed to show a persuasive advantage for this loss-frame approach in health-related messages. Thus, cognitive bias may in some cases be domain specific. Another example of potentially motivating and biasing communication is using personal stories or testimonials that illustrate other patients’ interpretation of decision experiences. Most rheumatologists recall persuasive advertisements with testimonials recommending the use of etanercept by public figures, such as golfer Phil Nicholson or the grandson of the famous aviator Charles Lindberg. In general, personal stories that emphasize specific reasons for a choice or introduce emotional terms may introduce bias, whereas stories that illustrate how others made sense about the facts may aid informed decision making. A good use of patient narratives is illustrated in the design of the UK National Health Service direct decision aid for patients considering treatments of osteoarthritis of the knee ( www.nhsdirect.nhs.uk/DecisionAids/PDAs/PDA_KneeArthritis.aspx ).
Physicians should provide patients with information about treatment options and the possible benefits and side effects. For several years the rheumatology faculty at Michigan State University has used decision aids developed to support the discussion of antirheumatics and structure patient deliberation after the office visit. Although these types of decision supports may not be necessary in all medication discussions, readers may wish to review them ( www.mi-arthritis.com ) as specific exemplars that meet the International Patient Decision Aids Standards for high-quality decision aids. They begin by directly stating the decision to be made. Background about RA, available treatments, and the expected course of RA with or without treatment are outlined. Next presented is specific information regarding the medication recommended by the rheumatologist. This includes the most common or serious side effects. Described are individual medication side effects and what it would be like to experience them. For example, a serious infection is described in this way : “Because MTX reduces the hyperactive immune system in RA, it can also reduce your ability to fight off serious infections like pneumonia, kidney infections, etc. When we refer to serious infections, we mean infections severe enough that you would need to be admitted to the hospital for one or more days to receive antibiotics through the vein and/or other care like IV fluids and oxygen.” It is essential for a patient to comprehend the personal impact of a side effect to be able to construct a meaningful value to that possible outcome.
Next presented is the probability of important outcomes. Several basic principles have been recognized as best practices in communicating risk. The first is to present frequency of a risk as an absolute risk rather than relative risk. Presenting relative risk tends to inflate the perceived benefits of therapy. For example, a hypothetical drug may reduce the relative risk of fracture by 40%. However, if this represents an absolute risk reduction from 1% per year to 0.6% per year a patient may interpret this as a lesser benefit. Another approach is to offer patients information as natural frequencies rather than probabilities, which in essence “does the math” by deconstructing the probability. Visual aids, such as natural frequency trees ( Fig. 2 ) or pictographs, may reduce some bias and framing effects and aid the understanding of incremental risk ( Fig. 3 ). One should also attempt to communicate the uncertainty around numeric risks presented. The estimates presented in decision aids from the University of Ottawa describe the source and quality of risk estimates as very low (+) to high (++++). Other designers qualify risk statements as “roughly 2 in 100”. Han and coworkers take the sophisticated approach of presenting a random pictogram, which can be animated ( Fig. 4 ). In some cases it is possible to present more patient-specific or tailored outcome information. Another approach demonstrated in the University of Ottawa “stepped decision aid” addresses a need to give information tailored to the stage of disease ( decisionaid.ohri.ca/decaids.html ). The patient answers questions about their impairment from osteoarthritis, and from this they are classified into level 0 to 5 disease. Patients then can review treatment options relevant for their level of disease, and the specific benefits and harms for each. Box 1 illustrates another tailored approach used by Dr David Hickman of Oregon Health Sciences University in an Agency for Healthcare Research and Quality funded decision aid that presented age-specific risk of gastrointestinal bleeding. In some cases one must communicate rate of outcome rather than frequency of outcome. For example, if RA or systemic lupus erythematosis is not treated, continuing disease activity may lead to progression of end organ damage. When communicating rate of disease progression it may be preferred to use speedometers rather than a pictograph. Fig. 5 is an excerpt from a methotrexate decision aid that describes how RA joint damage is slowed with treatment. To ensure patients understand the link between clinical arthritis and the progression of joint damage both are contrasted over time. In addition, the effect of methotrexate in percent absolute reduction of rate of progression followed by speedometers that contrast the progression of joint damage without and with methotrexate therapy are described.