How to validate clinically important change in health‐related functional status. Is the magnitude of the effect size consistently related to magnitude of change as indicated by a global question rating?

Abstract

Some clinical trials perform repeated measurement over time and estimate clinically relevant change in an instrument's score with global ratings of perceived change or so-called transition questions. The conceptual and methodological difficulties in estimating the magnitude of clinically relevant change over time in health-related functional status (HRFS) are discussed. This paper investigates the concordance between the amount of serially assessed change with effect size estimates (the researcher's perspective) and global ratings of perceived change (the patient's perspective). A total of 217 patients who were scheduled for diagnostic examination were included, and the Minnesota Living with Heart Failure Questionnaire, extended with MOS-20 items, was assessed before and after medical intervention (percutaneous transluminal coronary angioplasty, coronary artery bypass grafting or pharmaco-therapy). Global questions were applied to assess perceived change over time for every item from domains of physical and emotional functioning and used as the external criterion of relevant change in the analysis of items. Global questions corresponding with overall change in these domains were used in the comparison of change in physical and emotional functioning scales. Two effect size indices were used: (i) ES (mean change/SDpooled) and (ii) ES (mean change/SDchange). A method is described to calculate a value indicating the extent of discordance between the researcher's interpretation of magnitude of change and the external criterion (the patient's perspective). Findings suggest that effect size (ES) (mean change/SDpooled) was in keeping with the magnitude of change indicated by patients' judgements, or their category of subjective meaning, for all scales. Furthermore, in cases in which the magnitude of change estimated with the SRM (mean change/SDchange) was not confirmed empirically by the external criterion ratings, the discordance could be interpreted as a trivial discordance.