A randomized comparison between league tables and funnel plots to inform health care decision-making
Anders Anell1, Oskar Hagberg2, Fredrik Liedberg3,4
1Department of Business Administration, Lund University School of Economics and Management, P.O. Box 7080, SE-220 07 Lund, Sweden.
Objective:
Comparison of provider performance is commonly used to inform health care decision-making. Little attention has been paid to how data presentations influence decisions. This study analyzes differences in suggested actions by decision-makers informed by league tables or funnel plots.
Design:
Decision-makers were invited to a survey and randomized to compare hospital performance using either league tables or funnel plots for four different measures within the area of cancer care. For each measure, decision-makers were asked to suggest actions towards 12-16 hospitals (no action, ask for more information, intervene) and provide feedback related to whether the information provided had been useful.
Setting:
Swedish health care.
Participants:
Two hundred and twenty-one decision-makers at administrative and clinical levels.
Intervention:
Data presentations in the form of league tables or funnel plots.
Main Outcome Measures:
Number of actions suggested by participants. Proportion of appropriate actions.
Results:
For all four measures, decision-makers tended to suggest more actions based on the information provided in league tables compared to funnel plots (44% vs. 21%, P < 0.001). Actions were on average more appropriate for funnel plots. However, when using funnel plots, decision-makers more often missed to react even when appropriate.
Conclusions:
The form of data presentation had an influence on decision-making. With league tables, decision-makers tended to suggest more actions compared to funnel plots. A difference in sensitivity and specificity conditioned by the form of presentation could also be identified, with different implications depending on the purpose of comparisons. Explanations and visualization aids are needed to support appropriate actions.
Related Concept Videos
Comparing the Survival Analysis of Two or More Groups
Hazard Ratio
For example, in a clinical trial...
The Mantel-Cox Log-Rank Test
Friedman Two-way Analysis of Variance by Ranks
Types of Biopharmaceutical Studies: Controlled and Non-Controlled Approaches
Non-controlled studies, commonly employed for initial exploration, lack a control group, rendering them susceptible to biases and external influences. In contrast,...
Interpreting Run Charts


