[Skip to Content]
[Skip to Content Landing]
January 1, 2003

Risk Communication: Problems of Presentation and Understanding

Author Affiliations
  • 1University of California, Berkeley
JAMA. 2003;289(1):95. doi:10.1001/jama.289.1.95

Although patients frequently make decisions about the risks of medical treatments, their understanding of such risks may not be completely objective. Risk perception is affected not only by individual factors such as the patient's sex, prior beliefs, and past experience,1,2 but also by how the risk information itself is presented.

Identical risk information may be presented in different ways, resulting in "framing bias." Perceptions of risk are particularly susceptible to framing effects.3 For example, patients are much more likely to favor radiation treatment over surgery when radiation is presented as having a 90% survival rate than when it is presented as having a 10% mortality rate. Although both numbers describe identical risks, the latter is perceived as more dangerous.4 Another common framing effect involves absolute and relative risks. For example, if a medication reduces an adverse outcome from 20% to 15%, then the absolute risk reduction is 5% and the relative risk reduction is 25%. Although the absolute and relative risk estimates are derived from the same data, patients are more strongly persuaded by the larger changes in relative risk.4

The impact of framing on risk perception is often overshadowed by the effects of low numeracy. Numeracy refers to the ability to use numerical concepts and to perform basic probability operations. In one study of randomly selected women at a Veterans Affairs hospital, 46% could not correctly answer how many coin flips out of 1000 would turn up heads. Among these women, numeracy scores were related more strongly than framing effects to their ability to interpret risk estimates correctly. After controlling for demographic factors and whether the risk was framed as relative or absolute, women with high numeracy scores were 13 times more likely to interpret risk estimates correctly than women with low numeracy scores.5

Despite this propensity to misinterpret statistical risk information, patients often prefer quantitative over qualitative explanations of risk.2 Preferences for numerical risk estimates may stem from an inaccurate perception that these numbers represent some objective certainty. In one study of risk communication in genetic counseling, only 3 of 46 counselees understood that their risk estimates involved a degree of uncertainty.2

For patients with low numeracy skills, qualitative explanations may improve understanding. Despite variation in people's understanding of words like "rarely," "sometimes," and "often," these descriptors can be contextualized by comparing them to everyday risks like being involved in an automobile crash.3 This allows patients to compare the medical risk with risks whose severity and frequency they already understand.6 Narratives about risks facing people who are similar to the patient are also useful in helping patients contextualize risk. For example, personalized accounts by HIV positive people have been shown to lead to increased perceptions of risk among at-risk patients who identified with the individuals depicted in the stories.7

There are instances when it is difficult to avoid a quantitative discussion of risk. In these situations, clinicians can employ various techniques to improve patient understanding. Individualized risk estimates use the individual's personal risk factors (eg, age, sex, race) to compute the probability of developing a specific health problem in a given period of time.8 Individualized risk estimates have been shown to influence patients' treatment choices more strongly than presenting general risk information, and may also result in increased screening behavior.9 Visual displays of risk information may also increase patient understanding more than qualitative or quantitative explanations alone. These include risk ladders, which place risks in decreasing order of magnitude alongside equivalent comparisons of everyday risks, and the Wall of Balls, in which a risk of 1 in 1000 would be presented as one colored dot on a page with 999 different dots. Such displays improve understanding of the difference in magnitude between risks.10

The goal of risk communication is to help patients make informed decisions about treatment options, medication regimens, and lifestyle changes. To make such communication a useful decision-making aid for the patient is thus an arduous task, but one that can be aided by employing a mix of techniques that accommodate the varying preferences and abilities of different patients.

Shiloh  S, Saxe  L.  Perception of risk in genetic counseling.  Psychol Health. 1989;3:45-61.Google ScholarCrossref
Hallowell  N, Statham  H, Murton  F, Green  J, Richards  M.  "Talking about chance": the presentation of risk information during genetic counseling for breast and ovarian cancer.  J Gen Counseling. 1997;6:269-285.Google ScholarCrossref
Edwards  A, Elwyn  G.  Understanding risk and lessons for clinical risk communication about treatment preferences.  Quality in Health Care. 2001;10:i9-i13.Google ScholarCrossref
Malenka  DJ, Baron  JA, Johansen  S, Wahrenberger  J, Ross  JM.  The framing effect of relative and absolute risk.  J Gen Intern Med. 1993;8:543-548.Google ScholarCrossref
Schwartz  LM, Woloshin  S, Black  WC, Welch  HG.  The role of numeracy in understanding the benefit of screening mammography.  Ann Intern Med. 1997;127:966-972.Google ScholarCrossref
Bogardus  ST, Holmboe  E, Jekel  JF.  Perils, pitfalls, and possibilities in talking about medical risk.  JAMA. 1999;281:1037-1041.Google ScholarCrossref
Rothman  AJ, Kiviniemi  MT.  Treating people with information: an analysis and review of approaches to communicating health risk information.  J Natl Cancer Inst Monogr. 1999;25:44-51.Google ScholarCrossref
Benichou  J, Gail  MH, Mulvihill  JJ.  Graphs to estimate an individualized risk of breast cancer.  J Clin Oncol. 1996;14:103-110.Google ScholarCrossref
Edwards  A, Hood  K, Matthews  E,  et al.  The effectiveness of one-to-one risk-communication interventions in health care: a systematic review.  Med Decis Making. 2000;20:290-297.Google ScholarCrossref
Lipkus  IM, Hollands  JG.  The visual communication of risk.  J Natl Cancer Inst Monogr. 1999;25:149-163.Google ScholarCrossref