Quality of life in EMPEROR-Reduced: emphasizing what is important to patients while identifying strategies to support more patient-centred care

This editorial refers to ‘Empagliflozin and health-related quality of life outcomes in patients with heart failure with reduced ejection fraction: the EMPEROR-Reduced trial’†, by J. Butler et al., on page 1203. The presentation of study-specific data is challenging for patients to understand and for physicians to apply, whereas consistent benchmarks of clinically-significant scores and changes in scores are much more readily interpreted. The presentation of study-specific data is challenging for patients to understand and for physicians to apply, whereas consistent benchmarks of clinically-significant scores and changes in scores are much more readily interpreted. Patients with heart failure seek care to not only improve their survival, but also to improve their health status; their symptoms, function, and quality of life. In fact, this latter goal is often most important as it has been previously demonstrated that many patients would trade some of the duration of their survival to improve its quality.1 Yet, despite the importance of health status, clinical trials have been relatively slow in adopting and rigorously testing the impact of new treatments on these outcomes. The recent emphasis by the US Food and Drug Administration (FDA) on patient-reported outcomes (PROs), however, is rectifying this situation,2 and more trials, such as the Empagliflozin and Health Related Quality of Life Outcomes in Patients with Heart Failure with Reduced Ejection Fraction (EMPEROR-Reduced) trial, are systematically collecting and reporting these data. In the report by Butler et al. in this issue of the Euopean Heart Journal,3 the authors sought to accomplish two key goals. The first was to examine whether or not the illness of patients, as categorized by tertiles of their Kansas City Cardiomyopathy Questionnaire (KCCQ) Clinical Summary scores, was associated with more or less benefits from empagliflozin, a sodium–glucose co-transporter 2 (SGLT2) inhibitor, in terms of subsequent clinical events of death and hospitalizations. The second was to provide a more complete understanding of the health status benefits of empagliflozin on patients’ health status. Understanding the heterogeneity of treatment benefits as a function of patients’ baseline health status is important so that treatment can be tailored to risk. For example, with intracardiac defibrillators, the benefits of treatment were far more apparent in those at lower risk of death, presumably due to arrhythmic death, than in those at the highest quintile of risk who most likely died of pump failure - a cause of death that is not impacted by defibrillation.4 Since KCCQ scores have been shown to be strongly and independently associated with subsequent clinical events,5,6 illuminating whether the benefits in avoiding death and hospitalization are similar across the range of patients’ health status is important so that the benefits of treatment can be clearly communicated to patients. In EMPEROR-Reduced, there was no interaction or trend for better or worse clinical event rates by tertiles of KCCQ Clinical Summary scores, suggesting that, even in those with the best health status, where health status improvements are not to be expected, there is a benefit from treatment on reducing the risks of clinical events. The second goal of better understanding the health status benefits of treatment is exceedingly important so that patients can better understand how treatment is likely to make them feel. The investigators found an early benefit of treatment across multiple KCCQ domains within 3 months of therapy that persisted throughout the year of observation. Most importantly, the authors sought to render the mean differences clinically interpretable by demonstrating important and statistically significantly greater odds of clinically important improvements across tertiles of baseline scores. Rendering PROs clinically interpretable is a challenge because currently the range of scores is not intuitively familiar to providers or patients. Fundamentally, instruments such as the KCCQ simply quantify patients’ experiences with their disease—the symptoms they experience and how it impacts their function and quality of life. Yet, when transformed to a 0–100-point scale, the meaning of individual scores is difficult to comprehend. To address this challenge, efforts have been made to convert changes in PRO scores into clinically important magnitudes of change, with patients generally reporting smaller mean changes as being important than their providers.7,8 Thus, to help interpret changes in KCCQ scores between groups, the proportions of patients in a clinical trial who change by different magnitudes of clinical importance can be described and the differences in proportions can be converted into the number-needed-to-treat for one patient to benefit (or avoid deterioration), which is a much easier concept for patients and providers to understand.9 While the odds ratios for these clinical thresholds were provided, this is not as clinically tangible as the number-needed-to-treat, and an even simpler presentation of the data could further enhance its clinical interpretability for patients and providers alike. Building on the concept of simplifying data analysis to facilitate easier interpretation of the findings, we believe that future reports can present health status data in simpler ways. First, the authors chose to use tertiles to categorize their participants’ baseline health status, as opposed to fixed ranges of scores (e.g. 0–24, 25–49, 50–74, and 75–100). While this approach maximizes their statistical power by creating similar sized samples in each group, translating these findings to clinical care (scores <62.5 vs. 62.6–85.4 vs. 85.4) is much more difficult to apply to future patients than simple ranges of scores. Moreover, the tertiles will differ in every study and, if all trials adopt this approach, then comparing different trials will be impossible. A second, albeit minor, critique is the use of different definitions of clinically important changes in the responder definitions. To their credit, they used the well-established thresholds of 5 points to represent a small but clinically important change and 10 points to represent a moderate to large change. However, the 15-point threshold is below the standard recommendation of using a 20-point change to represent a large to very large clinical improvement.8,9 Again, while a relatively minor point, the use of different response definitions challenges the reader to interpret the results from different trials. While many trials are forced by the FDA to re-establish the minimally important clinical differences in their trials,10 there is no obligation to report these thresholds, which seldom differ substantially from the current recommendations of 5, 10, and 20 points, in the medical literature. What is much more important is to create a standard method of reporting these data so that they can be interpreted and the results applied in clinical practice (Graphical abstract). This report of EMPEROR-Reduced also illuminates an increasingly recognized challenge in clinical trial design. For decades, clinical trialists have used the New York Heart Association (NYHA) class as the foundation for defining trial eligibility. Yet, NYHA class is assigned by physicians, rather than patients, and it is increasingly apparent that there is little consistency in patients’ perceptions of their health status as compared with those of clinicians or other objective measures of disease severity.11,12 In EMPEROR-Reduced, 30% of patients had baseline KCCQ scores ≥90, which is more consistent with NYHA class I, despite the intention to enrol patients who were NYHA class II–III. While the authors introduced a creative means of addressing ceiling effects by counting benefit as no significant worsening of their KCCQ scores, as has been done in transaortic valve interventions,13 this diminished the mean population-level benefits of empagliflozin, as it is very difficult to improve the health status of patients who are minimally symptomatic. Using PROs, such as the KCCQ, as entry criteria for a trial is not only simpler to translate into clinical practice (where providers could simply give the 12-item KCCQ14 to determine if their patient would have met the entry criteria for the trial), but is also comparable with efforts the EMPEROR-Reduced investigators made to enrich the study population for more clinical events. Future clinical trials should consider inclusion criteria based on patients’ perspectives of their disease burden, rather than providers’ assessments. These technical issues aside, EMPEROR-Reduced is another important example of the benefits of SGLT2 inhibitors to improve the health status of patients with heart failure. As seen in DAPA-HF and DEFINE-HF,15,16 these drugs exert an impressive effect on patients’ health status, which seems comparable across different agents within the class, but has now been shown in EMPEROR-Reduced to be sustained out to at least a year. These agents thus represent an important new tool in the armamentarium of evidence-based heart failure treatments that markedly enrich the decades-old strategies of beta-blockers and renin–angiotensin–aldosterone system inhibitors. Given that so few patients are treated with guideline-directed medical therapy,17 there is an enormous opportunity to help improve treatment and patient outcomes. As clinicians seek to improve the care of their patients with heart failure, there is now an ever-expanding range of treatment options available. Given that many of these have relatively comparable effects on reducing the risks of clinical events, the improvements in patients’ health status may become a driving factor in optimizing care for each individual patient. While it is difficult to know how each patient may react to treatment, a new opportunity now exists to build protocols that enable new heart failure treatments to be tried and to systematically monitor patients’ health status response to treatment using simple instruments, such as the KCCQ-12.14 For those who benefit, the therapy can be continued and the patient can have concrete insights into why they are taking the medicine, a potentially powerful means of improving adherence. If patients’ health status deteriorates, or does not improve over a period of 3 months, then an alternative treatment can be tried. Such an empirical approach to tailoring treatment to individual patient benefit represents the implementation of personalized medicine, even if not guided by genetics or other biomarkers. The report by the EMPEROR-Reduced investigators provides much-needed information to patients and providers about an important benefit of therapy. While there are additional methods to render the findings even more clinically interpretable and to eventually build risk models to help identify which patients are most likely to benefit, or deteriorate, from empagliflozin, it is increasingly clear that SGLT2 inhibitors can improve the symptoms, function, and quality of life of patients with heart failure and reduced ejection fractions. The key challenge now is for the provider community to develop new strategies to treat and engage their patients in using these treatments so that their health status and clinical outcomes can be improved. Conflict of interest: J.S. discloses that he owns the copyright to the KCCQ. Unrelated to this work, he serves as a consultant to Amgen, Bayer, Myokardia, Novartis, Janssen, Merck, and United Healthcare, and serves on the board of directors for Blue Cross Blue Shield of Kansas City. He has an equity interest in Health Outcomes Sciences. The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal or of the European Society of Cardiology.