REPRESENTING THE THRESHOLD IN THE COST-EFFECTIVENESS PLANE- PROPOSING A NEW GRAPHICAL CONVENTION
Author(s)
O'Mahony JF
Trinity College Dublin, Dublin, Ireland
BACKGROUND: The aim of this presentation is to explain why the standard cost-effectiveness (CE) plane gives an incomplete representation of the cost-effectiveness of intervention strategies and to show how this problem is easily solved. The CE plane is the standard graphical framework for presenting costs, effects and incremental cost-effectiveness ratios (ICERs) of alternative intervention strategies. Comparing multiple strategies typically yields an increasingly steep efficient frontier, indicating that strategies become progressively less cost-effective as the ICER rises. However, standard CE planes generally do not include the cost-effectiveness threshold, therefore the figure itself does not identify which strategies are cost-effective. Consequently, the CE plane does not make it clear what strategy is optimal and the relative scale of costs and effects in the figure is uninformative. ANALYSIS: A new graphical convention is proposed whereby a line with slope equal to the cost-effectiveness threshold is added to the CE plane. While previous examples of CE planes have included such threshold lines, these have drawn the threshold from the origin, often at 45 degrees. Using published cost-effectiveness estimates from the literature, we show why this approach can be problematic: it does not readily identify the optimal strategy; can confuse average and incremental cost-effectiveness ratios; and, can impose a scale on the diagram that obscures relevant detail. Using the same published examples we then demonstrate the proposed alternative of including a threshold line which is set tangential to the frontier rather than passing through the origin and is not necessarily drawn at 45 degrees. This unambiguously identifies the optimal strategy and provides a useful scale that graphically illustrates which interventions are cost-effective. CONCLUSION: The simple inclusion of the threshold avoids several problems with the CE plane. It enhances the usefulness of the figure and better communicates which strategies are relevant and which are not.
Conference/Value in Health Info
2015-11, ISPOR Europe 2015, Milan, Italy
Value in Health, Vol. 18, No. 7 (November 2015)
Code
PRM243
Topic
Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference
Disease
Multiple Diseases