Graphic Format and the Patient’s Perception of Risk and Reality

One question that practitioners might consider prior to developing any campaign aimed at describing risk is how does the use of graphic format affect risk perceptions and the perceived truth of the data presented? Health communication practitioners often find themselves in a situation where they need to communicate probabilistic information (e.g., "9 out of 10 individuals are at a minimal risk"). Displaying this probabilistic information graphically such as bar charts, pictures, or graphs makes this communication easier. However, when ever practitioners choose to display probabilistic information in a graphic format they run the risk of causing unintended effects on those who they wish to communicate to due to the potential of misinterpreting the graphic information.

Research Questions

Schapira et al.’s study explores three vital questions regarding the use of graphic format to convey probabilistic information:

1. Are there differences in perceptions of risk when using different formats to convey identical information?
2. Are there differences in perceptions of truth when using different formats to convey identical information?
3. Are there differences in preference for graphic format amongst the target audience?

Method

In order to explore these questions Schapira et al. surveyed a sample of women and asked them to respond to questions after showing them identical information presented in various graphic formats. Schapira et al. gathered a sample of 254 women, measured their sociodemographic information, ascertained their ability to handle numbers and mathematical concepts (numeracy), and calculated their breast cancer risk factor. Next the participants viewed a series of risk communications regarding the breast cancer life time risk of a hypothetical patient. The information presented in the risk communication was identical except that the type of format it was presented in differed. Participants were asked to answer questions regarding the perceived risk magnitude, the perceived truth of the information, and were asked to rate the preference for the format.

Results

Risk Magnitude

• A hypothetical person’s risk is viewed as greater with a pictoral display than with a graph.
• Cancer risk is seen as greater when presented using random vs. consecutive highlighting of disease incidence.

Perceived Truth

• Risk is seen as more true when using: (1) smaller vs. larger forms of numbers, (2) a bar graph with comparable vs. single risk estimates, and (3) a pictoral display with random vs. consecutive highlighting.

Preference for Graphic Format

• A pictoral display is preferred to a bar graph when presenting a single risk, however.
• A bar graph is preferred to a pictoral display when presenting multiple risks.

Discussion

It should be stressed that Schapira et al. are not trying to discover the optimal format to be used for communicating risk. Rather, they state that health communicators need to be consistent with the format they use to minimize unintended effects due to misinterpretation of the information. As is always the case health communicators should strive to enhance patient understanding and ability to participate in the decision-making process regarding their health care.

The Authors

	Marilyn M. Schapira, MD, MPH A native of Montreal, Canada, Dr. Schapira attended Oberlin College (B.S. in 1981) and the University of Michigan (MD, MPH in 1986). She trained in Internal Medicine at Emory University and Moses Cone Hospital followed by a General Internal Medicine Research Fellowship. Dr. Schapira is currently an Associate Professor of Internal Medicine and the Director of Research and the Fellowship program in the Division of General Internal Medicine at the Medical College of Wisconsin in Milwaukee, WI. Dr. Schapira has a clinical practice in Women’s Health at the Zablocki VA Medical Center where she provides primary care to female veterans. Dr. Schapira is an investigator in the field of doctor-patient communication and medical decision-making. One of Dr. Schapira’s research interests is to understand how we can best communicate evidence from the scientific literature to patients and support a patient centered approach to medical decision-making. She is currently funded by the NIH on a study evaluating numeracy as a potential link between education and health. [Curriculum vitae]
	Ann Butler Nattinger, MD, MPH	Professor of Medicine and Health Services Research Chief, General Internal Medicine Lady Riders Professor of Breast Cancer Research Director, Center for Patient Care and Outcomes Research
	Dr. Nattinger has broad experience in health services research including use of various survey methods and analysis of administrative data, Medicare claims and tumor registry data. A nationally recognized cancer health services researcher, Dr. Nattinger's extensive research work includes studying treatment of breast cancer patients, therapy treatment options, physician communication with patients, the utilization of screening tests -- where she developed strategies practitioners can use to improve the utilization of screening mammography by their patients -- and the development and use of videotapes for aiding patients in medical decision making. She has is experienced in studying disparities in care, geographic variations in treatment, quality of life issues, the effect of health policy at the legislative level on the treatment patients will seek, and research methodology. Dr. Nattinger has been the recipient of a number of grants funded by the National Institues of Health and the Department of Defense. [Curriculum vitae]
Timothy L. McAuliffe, PhD		Professor and Director, CAIR Quantitative Core
Dr. McAuliffe's research interests include methodological issues in the design and analysis of clinical and field trials in AIDS prevention and cancer. His research also includes issues of reliability and validity of self-reported sexual behavior and the improvement of HIV risk assessment based on self-reported sexual behavior, the evaluation of group-randomized (community) intervention programs, and the analysis of frequency data (e.g., sexual activity) that often cannot adequately be modeled using standard methodology.

Know of an article that should have our spotlight? Send nominations to riskcenter@umd.edu.