LAUSR

The creation and use of disease-specific risk models to assign prognosis and predict outcomes for individual patients has pervaded health services research in the last decade. Development of such models has been driven by the increased availability of data and the need for tailoring treatment to risk, as underscored by the Institute of Medicine’s vision for patient-centered care and shared decision-making. However, despite a surplus of validated, published risk models for various disease states, and a culture of medicine that claims to emphasize transparency with patients, there is a paucity of data evaluating the implementation and clinical utility of these models from patients’ perspectives. In this issue of the Journal of Cardiac Failure, Narayan et al present a qualitative study of patients’ desires for, and reactions to, estimates of their predicted survival based upon the Seattle Heart Failure Survival Model (SHFM). Though the SHFM was created over a decade ago, to our knowledge this study represents the first attempt to assess how valuable a personalized estimate of prognosis is to patients. The authors should be lauded for their efforts, since a primary goal for all predictive models should be to inform patients of their prognosis as a foundation for treatment planning and shared decision-making. Of 26 patients approached for this qualitative research study, 24 agreed to participate and 17 wished to see their individualized survival estimates. Nine of the 17 did not feel it increased their anxiety to see their projected life expectancy and 15 of the 17 felt it was useful to review their results with a provider. It is both reassuring and in keeping with prior literature to know the majority of patients didn’t feel increased anxiety after seeing their results. However, nearly 40% of patients either felt that viewing their results was not useful or did not want to view the results in the first place. This begs the question—what makes a risk model valuable, from the patient’s perspective? Narayan and colleagues offer several important insights into patients’ perceptions of receiving risk estimates to inform healthcare decisions. First, the majority of patients expressed not only a desire to receive such estimates, but also a desire to receive them repeatedly over time as a “useful benchmark for how things are going”. Meeting this need requires a tectonic shift in the current practice of medicine, in which the infrastructure for generating and discussing prognosis needs to become more standardized. While electronic medical records could conceivably be programmed to regularly execute and update risk estimates, most currently do not do so. Moreover, it is not clear what models are most valid and useful. We would propose that guidelines committees move beyond the generic articulation of the importance of sharing patients’ prognoses with them to emphasizing the need both to identify the best and most valid models to use and to create the capacity of EMRs to integrate these into clinical documentation and patient portals. Second, patients in the study did not feel that the model results were personalized to them, despite the fact that the SHFM explicitly integrates a number of patient-specific factors to personalize survival estimates. Though the SHFM performs well at the population level, its ability to predict individual one-year survival has been previously questioned. This shortcoming notwithstanding, the failure of patients to understand that these estimates were, in fact, specific to them, highlights the need to better explain to patients what a risk model represents. Furthermore, the SHFM does not integrate important factors such as patients’ psychosocial and socio-economic status; this highlights the need to continually test, update and improve clinically-implemented risk models as new advancements in knowledge are achieved through research. Our group has created a number of risk models that do include more patient-specific health, psychosocial, socio-economic and stress variables, but many of these predictors are not routinely collected in registries and trials. They would need to be, in future studies, for patients to believe From the Saint Luke’s Mid America Heart Institute/UMKC, Kansas City, Missouri. Manuscript received February 10, 2017; revised manuscript accepted February 10, 2017. Reprint requests: John A. Spertus, MD, MPH, Saint Luke’s Mid America Heart Institute, 4401 Wornall Rd., Kansas City, MO 64111. Tel: +816 932 5613; Fax: +816 932 5179. E-mail: [email protected] See page ■■ for disclosure information. 1071-9164/$ see front matter © 2017 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cardfail.2017.02.002 ARTICLE IN PRESS