Reducing the influence of anecdotal reasoning on people's health care decisions: Is a picture worth a thousand statistics?

Background. People's treatment decisions are often influenced by anecdotal rather than statistical information. This can lead to patients making decisions based on others' experiences rather than on evidence-based medicine. Objective. To test whether the use of a quiz orpictograph decreases people's reliance on anecdotal information. Design. Two cross-sectional survey studies using hypothetical scenarios. Participants read a scenario describing angina and indicated a preference for either bypass surgery or balloon angioplasty. The cure rate of both treatments was presented using prose, a pictograph, a quiz, or a pictograph and quiz combination. Participants read anecdotes from hypothetical patients who described the outcome of their treatment; the number of successful anecdotes was either representative or unrepresentative of the cure rates. Setting and Participants. Prospective jurors at the Philadelphia County Courthouse and travelers at the Detroit-Wayne County Metropolitan Airport. Measurements. Proportion of respondents preferring bypass over balloon angioplasty. Results. In study 1, when statistical informotion was presented in prose, treatment choices were influenced by anecdotes, with 41% of participants choosing bypass when the anecdotes were representative and only 20% choosing it when the anecdotes were unrepresentative (chi(2) = 14.40, P < 0.001). When statistics were reinforced with the pictograph and quiz, anecdotes had no significant influence on treatment decisions (38% choosing bypass when anecdotes were representative v. 44% when unrepresentative, chi(2) = 1.08, P > 0.20). In study 2, the tradeoff quiz did not reduce the impact of the anecdotes (27% v. 28% choosing bypass after receiving or not receiving the quiz, chi(2) < 1, P > 0.20). However, the pictograph significantly reduced the impact of anecdotes, with 27% choosing bypass after receiving no pictograph and 40% choosing bypass after receiving a pictograph (chi(2) = 6.44, P < 0.001). Conclusions. Presenting statistical information using a pictograph can reduce the undue influence of anecdotal reasoning on treatment choices.