Press 1 to Speak to a Representative

See Article by Lee et al

Clinicians who care for patients with heart failure in the hospital are familiar with these and other questions that occur at discharge:

• “When will I be seen next?”

• “Who will see me in clinic?”

• “What happens if I don’t feel well or gain extra fluid?”

The age-old dilemma of timing discharge, follow-up, and interim problem management is familiar ground for the teams of clinicians seeing patients with heart failure. There is often a myriad of options and answers to the above questions, but for patients, it may feel like being on hold for customer service with unanswered questions and have other questions that they should have asked.

It has been a particular focus of hospital systems and payors in certain regions of the world to understand and pivot care towards the first 30 days after discharge, to potentially reduce the risk of readmission. This, although at the same time shortening the hospital stay and often skeletonizing the outpatient network required for complex health issues, is a recipe for disaster. Indeed, there is much controversy surrounding the Hospital Readmissions Reduction Program¹ in the United States as to whether or not it has led to improvements, stagnation, or potential harm for patients by focusing on a less well-validated measure as a target. Perhaps Goodhart’s law is at play, often paraphrased as “when a measure becomes a target, it ceases to be a good measure.”² With these targets and outcomes in mind, observational data have demonstrated the potential and laid the groundwork but remains just that: observational.^3,4 Data from the highest level of evidence (the randomized trial) is limited and thus the area has often focused on common-sense approaches, which is fraught with potential unintended consequences.

In this issue of Circulation: Quality and Outcomes, the THRIVE Study (The Heart Failure Readmission Intervention by Variable Early Follow-up) investigators attempted to fill a knowledge gap with a randomized controlled trial: does it matter who or how patients are seen in the first 7 days, as long as they are seen by a clinician with experience and expertise in the disease of note?⁵ In this case, the trialists designed an intervention focusing on the timing and type of follow-up rather than the preparedness for discharge or quality of care. Much like other strategy trials done in a pragmatic clinical system, the fidelity of the intervention is difficult to control, and the trialists are to be commended for conducting a much-needed trial in the manner it was done.

The THRIVE trial identified no obvious advantage to early structured nonphysician phone support, despite achieving a 13% difference in the number of visits within 7 days for the interventional arm (an early nonphysician phone-based strategy). At first glance, this would appear to have disproved the hypothesis that the early nonphysician strategy has advantages that should lead to a system-wide practice change. However, careful review of the interventional arm (the phone strategy) identifies that 44% had a phone visit, 39% a phone plus clinic visit, and 9% clinic visit alone, whereas the clinic strategy had a remarkable 77% of patients seen in person within 7 days by a physician. This is a very high percentage when compared to other health systems and commendable—few interventions could beat that follow-up—and, therefore, it would be hard to impact subsequent outcomes. In terms of outcomes, there was no statistical difference despite a 2% absolute reduction in heart failure readmissions in 30 days (and a 1.8% reduction in all-cause readmissions). The THRIVE sample size was based on an ambitious 3.75% absolute risk difference: very few interventions lead to that degree of difference and in an area (heart failure hospitalization) where many other failures have been seen. If by moving up the visits to be within 7 days and with a broader team, this seems like a laudable goal.

In evaluating the THRIVE trial, there are a few positive signals to consider. First, the trial aimed to and achieved inclusion of the elderly (mean age 78 years), of women (44% women) and 41% were persons of color. This is very similar to the recently completed Patient-Centered Care Transitions in HF trial⁶ in which 2494 patients in 10 hospital clusters randomly allocated to a transition bundle, focused on discharge preparedness with similar follow-up mechanisms as the THRIVE trial. Both of these trials have demonstrated that it is highly feasible to design trials that are inclusive and generalizable and likely relevant to the hospital system and patients they serve. Second, where possible, the trial captured data directly from the electronic health records rather than manual abstraction of data. This has improved substantially in the past few years but can be subject to accuracy issues similar to paper chart data on subjective information. But it has a sizeable advantage on numerical values (eg, hemoglobin, ejection fraction) and can thus lead to more accurate, efficient, and potentially lower-cost data acquisition. However, caution should be exercised as the database coding, timing of measurement, and numerous other data-decisions made along the way can impact interpretation.

THRIVE leaves many questions unanswered. First, was the quality of care for patients with heart failure with a reduced ejection fraction (HFrEF) upon discharge low to begin with or after either strategy, altered? Because many of the medications have an early reduction in morbidity end points for HFrEF, were the quality of care metrics improved upon with early intervention (within 7 days), setting the stage for long-term benefit? Second, a surprisingly high proportion of patients in the trial are on medications with less adequate evidence, such as calcium channel blockers (33%) and few on those with a stronger evidence-base across the ejection fraction spectrum such as mineralocorticoid antagonists (8%). This recognizes the range of patients with different ejection fractions and renal function but likely is a signal that much more appropriate choices could be made and potentials for deprescribing abound. As such, there is much to be improved upon both in-hospital and soon after discharge and applying opportunistic or all-or-none composite measure to assess these changes.⁷ Third, would the fidelity of the intervention have been further amplified if clinicians known to the patient had seen the patient—in either arm? Data support that the usual care physician sees a patient before and after heart failure hospitalization or emergency department visit is associated with fewer subsequent events.⁸ Reporting on the above data would help shape future projects.

So where does that leave us, knowing the THRIVE and other results? First, more randomized trials are sorely needed to prove (or disprove) common-sense ideas as the unintended consequences, cost, and patient outcomes are important. Second, pragmatic designs such as THRIVE demonstrate the ability of a health system to conduct randomized controlled trials, yet these need to occur more often, faster, and integrated into clinical care. THRIVE had to utilize research coordinators to organize and deliver follow-up—should this not be the role of the health system and patients and clinicians to improve the environment for all to benefit? The return on investment of research is a multiple for those that undertake the effort. Third, combinations of care elements need to be brought together from the local environment—what works in one jurisdiction may not work elsewhere, so a bespoke approach would be wise for those undertaking system redesign trials. Fourth, and possibly the most important, is that digital health and technology facilitate care but does not obviate the need for human to human communication and building teams that communicate well around a patient. Mistaking technology for care has led to many failed experiments with telehome-monitoring and virtual hospitals, not recognizing the expertise of the entire team (such as dieticians, nurses, pharmacists, physiotherapists, occupational therapists, and others), and undervaluing the touchpoint that an admission to hospital provides for altering the longer trajectory for a patient.

In summary, THRIVE demonstrates some of the possibilities, and although nonstatistically significant, shows a window of opportunity. To answer the questions posed above, the answers should be (1) this week, in person or by phone or video; (2) by a team member who you have already met; and (3) here are our phone, email, text, and video link numbers. We need to press 1 for our patients, provide better more efficient service, and improve upon system redesign by robust randomized clinical trials.

Footnotes