We propose to use subjective well-being (SWB) measures to determine patient-relevant treatment benefit. Benefit can be measured either prospectively (pre-post) or retrospectively, but both approaches can be biased: Prospective evaluation may be subject to response shift; retrospective evaluation may be subject to recall bias. As prospective and retrospective evaluations often differ in effect size and since there is no gold standard to compare against, the extent of the two biases needs to be determined. Response shift includes reprioritization, reconceptualization, and recalibration. We argue that in SWB measures only recalibration, but not reprioritization and reconceptualization are validity threats. We review approaches to quantify recall bias, response shift, or both in the measurement of health-related quality of life. We discuss which of these approaches are most suitable for application to SWB measurement, where only recall bias and recalibration are to be quantified, ignoring the other two response shift types. Some approaches of bias detection will not be applicable to SWB measurement, because they do not distinguish between recalibration and other types of response shift, or quantify reprioritization and/or reconceptualization alone. For other approaches, it is unclear whether underlying assumptions apply to SWB measurement. Anchor recalibration, structural equation modelling, and ROSALI are most suitable, the latter two with some limitations. Anchor recalibration was considered by its developers to be too difficult for participants to understand in its current form. Refining the anchor recalibration method may provide the most promising way to quantify both scale recalibration and recall bias.
health-related quality of life thentest response shift recall bias scale recalibration subjective well-being