Comparing Graphical Formats for Feedback of Clinical Practice Data

Abstract

Summary Objectives: Although graphical formats used to feedback clinical practice data may have an important impact, the most effective formats remain unknown. Using prevention of postoperative nausea and vomiting by anesthesiologists as an application, the objective of this study was to assess which graphical formats for feedback of clinical practice data are the most incentive to change practice. Methods: We conducted a multicenter cross-sectional study among anesthesiologists randomized in two groups between March andJune 2014. Each anesthesiologist assessed 15 graphical formats displaying an indicator of either prescription conformity or prescription effectiveness. Graphical formats varied by: type of graph (bar charts, linear sliders, or pictographs), presence or not of a target to reach, presence or not of a contrast between a hypothetical physician and his/her team, direction of the difference between the physician and his/her team, and restitution or not of the quality indicator evolution over the previous six months. The primary outcome was a numerical scale score expressing the anesthesiologists’ motivation to change his/her practice (ranging from 1 to 10 points). A linear mixed model was fitted to explain variation in motivation. Results: Sixty-six anesthesiologists assessed the conformity indicator and 67 assessed the effectiveness indicator. Factors associated with an increased motivation to change practice were: (i) presence of a clearly defined target to reach (conformity: β = 0.24 points, p = 0.0046; effectiveness: β = 1.11 points, p < 0.0001); (ii) contrast between the physician and his/her team (conformity: β = 0.38 points, p < 0.0001; effectiveness: β = 0.33 points, p = 0.0021); (iii) better results for the team than for the physician (conformity: β = 0.65 points, p < 0.0001; effectiveness β = 1.16 points, p < 0.0001). For the effectiveness indicator, anesthesiologists were more motivated to change practice with bar charts (β = 0.24 points, p = 0.0447) and pictographs (β = 0.45 points, p = 0.0001) than with linear sliders. Conclusions: Graphs associated with a defined target to reach should be preferred to deliver feedback, especially bar graphs or pictographs for indicators which are more complex to represent such as effectiveness indicators. Anesthesiologists are also more motivated to change practice when graphs report contrasted data between the physician and his/her team and a lower conformity or effectiveness for the physician than for his/her team.