DECISION AIDS: THE EFFECT OF LABELING OPTIONS ON FINAL DECISIONS AND DECISION-MAKING PROCESS

Sunday, October 19, 2014
Poster Board # PS1-21

Candidate for the Lee B. Lusted Student Prize Competition

Olena A. Cherkasky, MD, Peter J. Veazie, PhD, MS, Nancy Chin, PhD, MPH and James G. Dolan, MD, University of Rochester, Rochester, NY
Purpose: Decision aids are designed to help patients participate in informed decision-making process regarding their health and health-care options. If the options are identified before users compare their attributes, however, it is possible they might base their decision on preconceived beliefs and emotions instead of the objective and clinically-relevant information provided by a decision aid. We sought to evaluate the effect of labeling the options in a decision aid on final decisions and to discern if labeling brings additional relevant information into the decision or error into the decision due to use of heuristics and biases.

Method: 523 participants recruited from an on-line survey panel completed a decision aid regarding options to prevent future cardiovascular disease. One group of panelists had a decision aid with labeled options (Aspirin and Statins), another group had blinded options (Medication A and B).  Following the decision aid, participants were asked to indicate their preferred choice, confidence in their choice, and an ease-of use question. To discern the mechanism of the labeling effect, participants in blinded and labeled groups were asked to evaluate relative risks and benefits of each option. 

Result: Among the group with labeld options, the odds of choosing Aspirin over Statins increased 3.8 times (coeff=1.33, P<0.000); the odds of being 1 point more confident on a 10-point scale were 30% greater (coeff=0.26, P<0.087);  the odds of rating the task as 1 point easier on a 7-point scale were 79% higher (coeff=0.58, P<0.000)

   When rating relative risks and benefits, those in labeled group had 45% less likelihood of rating Statin’s benefits 1 point higher (on a 9-point scale) than benefits of Aspirin compared to blinded group. The effect of labeling on evaluating risks was even larger: 71% less likelihood of favoring Statins’ risks among labeled groups compared to blinded group. Controlling for literacy, numeracy, age, gender, and education, did not change the results.

Conclusion: Even though it is natural to think that knowing the options gives people more meaningful information and enhances their decision, our study showed that people interpreted the same information on risks and benefits differently when  knew the options. When decision aids include options familiar to patients it may be necessary to reveal the names of the options only after an initial blinded comparison phase.