Background: Decision aids often provide statistical information and patient testimonials to guide treatment choices. This raises the possibility that the testimonials will overwhelm the statistical information.
Methods: Prospective jurors in Philadelphia County were presented with hypothetical statistical information about the percentage of angina patients who benefit from angioplasty and bypass surgery (50% and 75%, respectively). They were also given written testimonials from hypothetical patients who had benefited or not benefited from each of the two treatments. The numbers of patients benefiting and not benefiting were varied to be either proportionate to the statistical information or disproportionate. In study 1, all participants received 1 testimonial from a patient who had benefited from angioplasty and 1 from a patient who had not. Participants receiving the proportionate questionnaire version were also given 3 testimonials from patients who benefited from bypass surgery and 1 from a patient who did not, coinciding with the hypothetical statistical information. In contrast, participants receiving the disproportionate questionnaire version received only 1 testimonial from a patient who benefited from surgery and 1 from a patient who did not. In study 2, all participants received 2 examples of patients who benefited from angioplasty and 2 who did not. Participants with the proportionate questionnaire version received the same testimonials regarding surgery as in study 1. Those receiving the disproportionate questionnaire version received 2 testimonials from patients who benefited from bypass and 2 from patients who did not. Finally, a separate set of participants in study 2 received a questionnaire with no testimonials.
Results: In study 1, 30% of participants receiving the disproportionate questionnaire version chose bypass surgery versus 44% of those receiving the proportionate questionnaire (P = 0.002 by chi2). In study 2, 34% of participants receiving the disproportionate questionnaire version chose bypass surgery versus 37% of those receiving the proportionate questionnaire (P = 0.59 by chi2). Of those receiving no patient testimonials, 58% chose bypass surgery.
Conclusions: The inclusion of written patient testimonials significantly influenced hypothetical treatment choices. Efforts to make the mix of positive versus negative testimonials proportionate to statistical information may, under some circumstances, affect choices in ways that cannot automatically be assumed to be optimal.