Did Finerenone Improve Health Status in the FINEARTS Trial?: A Critical Reevaluation of the Analysis of Patient-Reported Outcomes in Heart Failure.

Patient-reported health status is an important assessment of patients with heart failure, but current approaches have substantial methodological and analytical limitations. Changes in the Kansas City Cardiomyopathy Questionnaire (KCCQ) are commonly presented as a measure of the effect of drugs and devices, most often as the between-group difference in population means or as the odds of showing threshold changes of 5, 10, 15, and 20 points. However, the presentation of mean differences is based on statistical assumptions that are routinely violated in most trials. The presentation of threshold changes is based on the belief that a within-patient change in KCCQ of 5 points represents a significant treatment difference across diverse populations and trial settings, but the minimal clinically meaningful difference varies substantially depending on patient characteristics, comorbidities, and trial duration and design, with most values for minimally clinically important difference for KCCQ ranging from 10 to 20 points. Furthermore, the assessment of between-group differences is highly distorted by the assignment of a large proportion of randomized patients with very good health status as having substantially improved even if they showed no change after treatment. Any responder analysis is highly sensitive to differences in variance between the 2 treatment groups and cannot account for the stability of changes in the KCCQ score. The imposition of number-to-treat presentations onto KCCQ scores further compounds the lack of interpretability of reported changes. It is therefore not surprising that trials have reported substantial discrepancies between the effect of treatment on KCCQ and on the risk of hospitalizations for heart failure. In the FINEARTS (Finerenone Trial to Investigate Efficacy and Safety Superior to Placebo in Patients with Heart Failure) trial, finerenone produced an 18% reduction in the risk of hospitalizations for heart failure but yielded uninterpretable and clinically questionable changes in KCCQ, with the small possibility of a modest benefit in <2% of randomized patients. Most physicians are unaware of the critically important methodological concerns summarized in the current paper, and, therefore, may make clinical decisions that hinge on unwarranted impressions of the effects of an intervention on health status.