Project Overview +

Guide to Decide uses a multi-phased experimental process to explore methods of communicating risk regarding tamoxifen or raloxifene prophylaxis to women at high risk for breast cancer.

Aims +


Aim 1. Develop and test a tamoxifen prophylaxis decision aid by screening three risk messages and three graphical formats using fractional factorial designs.

Aim 2. Screen for individual characteristics (e.g., ability to think numerically) that will point the way towards methods of improving decision aids by tailoring risk messages.

Aim 3. Test two novel methods of evaluating whether decision aids improve decisions.


Aim 1. Refine our understanding of the factors that increase the effectiveness of the decision aid in Phase I, by isolating specific components of those factors.

Aim 2. Refine and further explore promising interactions between risk messages/graphical formats and specific individual characteristics.

Aim 3. Determine whether improved knowledge of the risks and benefits of tamoxifen translates into behavioral outcomes such as consulting with a physician.

Participants +

1,184 women in two Cancer Research Network HMOs.

Intervention +


In Phase I, we design a web-based decision aid for women at an increased risk of developing breast cancer that discusses the risks and benefits of taking tamoxifen to prevent a first breast cancer. A screening experiment uses a fractional factorial design, to test the effect of various risk messages and graphical formats on women's risk perceptions and decisions regarding tamoxifen prophylaxis. All risk information is tailored to the individual women based on their personal risk of breast cancer according to the Gail model.

Phase I tests the following factors:

  • Pictographs to display risk information (present versus absent)
  • Statistics: Simple (out of 100 women, no relative risk) versus elaborate (out of 1000 women, presence of relative risk)
  • Risk presentation (additional risk [incremental increase/decrease in risk] versus total risk)
  • Order (risk information first versus benefit information first)
  • Risk context (information about other health risks-present versus absent)

A total of 644 subjects from two HMOs - Henry Ford Health System in Detroit and Group Health in Seattle.


In Phase II, approximately 640 women (roughly 2/3 of total enrolled) are randomly assigned to the Intervention group, and receive the Decision Aid, an individually-tailored program based on combinations of five tailoring factors (described below) at baseline. Results from Phase I and new clinical research results from the STAR trial helped us to identify new factors to guide our research in Phase II:

  • Order - whether participants receive information on risks of chemoprevention first or information on benefits of chemoprevention first.
  • Describing decision as having 2 vs. 3 treatment options - whether participants receive an assessment of benefits and risk in terms of no medications vs. medications, or no medications vs. tamoxifen vs. raloxifene.
  • All About You (AAY) - Personalization and Testimonials - whether participants receive testimonials written from the viewpoint of individuals who have or have not used chemoprevention, as well as testing using 2nd person vs. 3rd person voice. The testimonials describe how the woman in the testimonial (age and race matched to the subject) used the decision aid to help her make her decision.
  • Comparative risk information - whether participants receive information on the average woman's risk of breast cancer in addition to their own personal risk.
  • Summary table - whether participants receive all the information on risks and benefits in table form after receiving the same information as text and pictographs.

Findings +

PHASE I Findings

  • Women have very little interest in looking for more information about tamoxifen, talking with their doctor about tamoxifen, or taking tamoxifen to prevent a first diagnosis of breast cancer.
  • Women's lack of interest in tamoxifen is largely due to their perception of the risks of tamoxifen, particularly in that they do not view the benefits of tamoxifen are worth the risks associated with taking it.
  • Women's comparative risk perceptions were more important than their objective risk in predicting responses to a decision aid about tamoxifen (e.g., their anxiety, knowledge, and behavior).
  • Higher comparative risk perceptions were associated with more anxiety about breast cancer, more knowledge about the risks and benefits of tamoxifen, greater intentions to take action, and three months later, engaging in behaviors consistent with an interest in taking tamoxifen.
  • The order in which the risks and benefits of a medication are presented has a significant impact on people's risk perceptions, knowledge, and behavior. When risks are presented last, people are more worried about the side effects of the medication and perceive them as more likely. The information presented last is remembered better. When risks are presented last, people are less interested in the medication.
  • Pictographs make risk statistics easier to interpret, eliminating biases caused by other design factors.
  • When using pictographs to present risk and benefit information, it is best to present information using an incremental risk format.
  • A measure of subjective numeracy is a good predictor of health communication predictor and has similar predictive ability as an objective numeracy measure.

PHASE II Findings

  • Development of Subjective Numeracy Scale. This scale has been translated into Spanish, Portuguese, Dutch, and Japanese. The scale has been used in numerous studies and cited in over 60 articles.
  • After completing a number of preliminary, hypothetical studies to determine the best methods for illustrating the incremental risks associated with a drug (i.e., how does taking a drug increase risk of side effects above and beyond baseline risks of those same side effects), we tested it with a sample of women who were deciding about chemoprevention. We found that providing information using an incremental risk format, decreased women’s perceived side effects to be significantly less common, less likely, and less worrisome when the presentation reported the incremental risk (F = 12.56, p < 0.001).
  • Confirmed that use of pictographs (vs. plain text) increases knowledge of key risk/benefit information (F = 3.44, p = 0.064).
  • Showed that women at a high risk of breast cancer are not interested in raloxifene after reading a decision aid about tamoxifen and raloxifene. Only 0.5% of women who read a decision aid started chemoprevention within 3 months of reading the decision aid (these 2 women chose raloxifene). Only 44.7% of those reading a decision aid believed that their risk of breast cancer would decrease if they took tamoxifen, whereas 51.5% perceived no change in their risk. Similarly, 42.9% perceived that raloxifene would decrease their risk of breast cancer whereas and 55.4% perceived that it would not. This study was one of the first studies that has been published showing women’s attitudes regarding raloxifene, particularly after receiving a decision aid.
  • Illustrated that the fractional factorial design can be successfully used in risk communication studies.

Conclusion +

PHASE I Conclusion
How you present information to patients has a significant impact on their decision making. When designing decision aids, it is extremely important to consider how to present the risks and benefits of treatment options. Our findings have led us to recommend using pictographs and incremental risk format to communicate the additional risk a patient is exposed to due to a medical intervention. Furthermore, the order in which individuals discuss the risks and benefits of treatment has a substantial impact on knowledge and treatment preferences.

Women have very low interest in taking tamoxifen to prevent a first diagnosis of breast cancer. This is primarily due to their concerns regarding the side effects of tamoxifen. Furthermore, women do not view the benefits of tamoxifen worth the risks of taking tamoxifen.

PHASE II Conclusion
The work of this project has encouraged others to rethink how decision aids are evaluated. We are moving beyond just examining knowledge and anxiety, but testing decision aids to determine if they introduce cognitive biases and if certain risk communication techniques eliminate the presence of cognitive biases.

Guide to Decide

07/01/2003 - 08/31/2008


National Cancer Institute

Principal Investigators:

Victor J. Strecher, PhD, MPH
Angela Fagerlin, PhD
Peter A. Ubel, MD


Sharon M. Hensley Alford, PhD
Daniel F. Hayes, MD
Paula M. Lantz, PhD, MS
Jennifer B. McClure, PhD
Priti R. Shah, PhD
Dylan M. Smith, PhD
Stephen H. Taplin, MD, MPH
Brian Zikmund-Fisher, PhD