A COMPARATIVE STUDY OF RISK COMMUNICATION FORMATS

Purpose: To compare patient preferences and accuracy of interpretation for 5 different risk communication formats across a wide range of risk likelihoods.

Method: Questionnaires with 5 different combinations of formats (table, flowchart, scale, bar, and icons) and risks (lifetime risk of disease (range: 10  to 48 per 1,000), risk of dying from disease (range: 4 to  22 per 1,000), risk of serious screening test side effect (range: 2 to 17 per 5,000), risk of dying from serious screening test side effect (range: 2 to 17 per 5,000), and risk of a false positive screening test result (range: 100 to 500 per 1,000)) were distributed to study subjects over the internet. Patient preferences were measured by 10-point Likert scale ranging from 1 (confusing, very unhelpful) to 10 (clear, very helpful). Understanding was measured by the percentage of correct ordinal judgments between pairs of risk likelihoods. The results were analyzed using descriptive summary, analysis of variance (ANOVA) single factor analysis, and correlation analysis.

Result: 279 valid responses were received. The table format was most preferred with a mean score of 7.40, followed by flowchart (6.72), scale (5.94), bar (5.10), and icons (4.59). Results from ANOVA analyses suggested that significant variation in preferences was due to risk presentation format (p<0.001) but not due to the nature or the range of the risk involved (p=0.80). The accuracy of risk comparisons varied by presentation format: table (83%), bar (81%), flowchart (79%), scale (79%), and icons (79%). Results from ANOVA analyses suggested that, once again, variation in accuracy was due to the type of format (p=0.004) but not to the nature or the range of the risk (p=0.44). No significant correlation was detected between patient preferences and accuracy of interpretation (r=0.28). Using Chi Square tests, no statistically significant differences in these results were detected with respect to respondents’ age (p=0.35), gender (p=0.70), ethnic (p=0.87), race (p=0.40), education (p=0.22), literacy (p=0.29), and numeracy (p=0.40).

Conclusion: Different presentation formats, but not the type of risk or risk magnitude, play an important role in comparative risk communication across a variety of risks and a wide range of risk likelihoods. Consistent with prior studies, formats most preferred by patients are not necessarily the best for conveying comparative risk information.