Effects of a risk-based online mammography intervention on accuracy of perceived risk and mammography intentions

Abstract

Objective: This experiment tested the effects of an individualized risk-based online mammography decision intervention. The intervention employs exemplification theory and the Elaboration Likelihood Model of persuasion to improve the match between breast cancer risk and mammography intentions.

Methods: 2918 women ages 35-49 were stratified into two levels of 10-year breast cancer risk (<1.5%; ≥1.5%) then randomly assigned to one of eight conditions: two comparison conditions and six risk-based intervention conditions that varied according to a 2 (amount of content: brief vs. extended) x 3 (format: expository vs. untailored exemplar [example case] vs. tailored exemplar) design. Outcomes included mammography intentions and accuracy of perceived breast cancer risk.

Results: Risk-based intervention conditions improved the match between objective risk estimates and perceived risk, especially for high-numeracy women with a 10-year breast cancer risk ≤1.5%. For women with a risk≤1.5%, exemplars improved accuracy of perceived risk and all risk-based interventions increased intentions to wait until age 50 to screen.

Conclusion: A risk-based mammography intervention improved accuracy of perceived risk and the match between objective risk estimates and mammography intentions.

Practice implications: Interventions could be applied in online or clinical settings to help women understand risk and make mammography decisions.

Keywords: Communication intervention; Decision aid; Mammography; Numeracy; Perceived risk; Risk communication.

Figure 1

Effects of condition on accuracy of perceived risk (overestimation) as a percentage for women with an estimated 10-year breast cancer risk <1.5%, by numeracy level. This figure illustrates the moderating effect of numeracy level on the accuracy of risk depending on condition when perceived risk is measured as a percentage. To create a measure of risk accuracy, the participant's 10-year estimated objective risk for BC (or NCI BCRAT score) was converted to a percentage and subtracted from the participant's perceived risk reported as a percentage. Thus, the numbers presented represent the difference between perceived and estimated objective BC risk (e.g., a woman who thinks her risk is 20% when it is actually 5% would have an accuracy [overestimation] score of 15).

Figure 2

Effects of condition on accuracy of perceived risk (overestimation) as a frequency for women with an estimated 10-year BC risk <1.5%, by numeracy level. This figure illustrates the moderating effect of numeracy level on the accuracy of risk depending on condition when perceived risk is measured as a frequency out of 1000. To create a measure of risk accuracy, the participant's 10-year estimated objective risk for BC as a frequency out of 1000 was subtracted from the participant's perceived risk reported as a frequency out of 1000, and the measure was divided by 10 to put it on the same scale as perceived risk reported as a percentage. Thus, the numbers presented represent the difference between perceived and estimated objective BC risk divided by 10 (e.g., a woman who thinks her risk is 200 out of 1000 when it is actually 50 out of 1000 would have an accuracy [overestimation] score of 15).