TAKE-THE-BEST APPROACH TO DECISION AIDS

PURPOSE   Besides informing patients about the benefits and harms of available options, decision aids often include guidelines for the decision making process itself. These guidelines typically involve complex procedures such as weighing and adding of a number of considerations. It has been shown, in other decision contexts, that simple decision strategies based solely on the most important reasons are faster, and can be equally or even more accurate than complex strategies (Gigerenzer, Todd, & the ABC Research Group, 1999). In this study, we compare the effectiveness of one simple strategy (take-the-best heuristic, TTB) with a standard complex procedure (Analytic Hierarchy Process, AHP; Dolan, Isselhardt, & Cappuccio, 1989).

METHOD   Based on these different decision processes, we developed decision aids for choosing among three screening options, with six relevant considerations, for colorectal cancer. The AHP version described options and considerations, then required pair-wise comparisons of all considerations and of all options relative to each of the considerations (Dolan & Frisina, 2002). The TTB version described options and considerations, and then asked about only the most important consideration and the option that best satisfied it. A control version described options and considerations, with no decision support.   Participants (n=53, 32 women, 51-75 years) read one of three decision aids. They were then asked which screening option they would pick if they were choosing now, completed the decisional conflict scale (O’Connor, 1995), and evaluated how well they understood the criteria and the screening options on 5-point scales.

RESULTS   Take-the-best predicted the final decision better than the Analytic Hierarchy Process, yielding correct predictions for 94% vs. 41% of participants, respectively. Final decisions were similar across the three versions of aids (74% colonoscopy, 24% stool test, 2% Wait&See). Decisional conflict was comparable for all versions, with one reliable difference found in the values clarity subscale (AHP version: 35.3, SE=2.9, TTB version: 46.9, SE=4.5; Control: 50.1, SE=3.8). Self-rated evaluation of understanding was similar across the versions (AHP version: 4.1, SE=.24, TTB version: 4.4, SE=.20; Control: 4.2, SE=.19; higher is better). Compared to the AHP version (18.6 minutes), the TTB decision aid took much less time to complete (11.6; Control: 10.5).

CONCLUSIONS   Simple heuristics represent a promising alternative to more complex guidelines for making decisions about treatment and screening options.