Published in Quality of Life Research, 2021
Meaningful Change Thresholds (MCTs) are used to represent meaningful changes in patient-reported outcome scores. Typically, multiple methods to estimate MCTs are used, resulting in a range of estimates, so triangulation is applied to produce a single estimate or small range, making it possible to consistently interpret outcome scores.
Although current approaches to triangulation are fine in principle, there is a lack of clarity around the vital step of moving from the range of estimates to a single value or small range, and this process is subjective in nature. Our aim was to develop a more robust form of triangulation.
We focused on anchor-based estimates, which are recognised as the most valid and developed approach. A correlation-weighted average of MCT estimates is recommended to triangulate multiple MCT estimates, with increased weighting given to stronger anchor measures. The choice of method used to triangulate estimates from a range of published studies is dependent on the availability of information within those publications. MCTs designed for between-group differences, within-individual changes and within-group changes have to be considered seperately.
Our recommendations provide a reliable and transparent triangulation approach, which is preferable to a simple mean of estimates or the visual estimation of plotted data. We encourage the adoption of these methods, but also urge researches to be aware of the limitations in each method, and of further nuances in study design that result in heterogeneity. Our recommendations provide a suitable starting value for inferences, but we encourage sensitivity analyses with a range of plausible values for inferences. We also highlight unresolved issues in triangulation, which require further exploration.