The need for hospitalization is a sentinel event in the life of a patient with heart failure ((1),2). Within 30 days of hospital admission with heart failure, nearly 1 in 10 patients is dead and 1 in 4 has been readmitted, half of these because of recurrent symptoms of heart failure (3). Readmission rates approaching 50% at 6 months contribute to an annual Medicare expenditure of nearly $17 billion ((4),5). Because of casual retrospective estimates that nearly three-fourths of early readmissions may be preventable (6), public and private payers have increasingly targeted reduction in readmission rates as a primary focus of pay-for-performance initiatives. Financial penalties for 30-day readmissions as part of the demand for “accountable” care have shifted hospital incentives toward support for improving education, post-discharge care transitions, and palliative care integration for patients with heart failure (7).

Even with widespread implementation of a package of post-discharge strategies that successfully addresses the triggers of readmission, episodes of heart failure decompensation will continue to occur. Earlier recognition of clinical deterioration in well-managed populations should increasingly permit timely intervention in the ambulatory setting to restore compensation. However, patients concerned about clinical changes and physicians discomfited by the heart failure diagnosis frequently use the emergency department (ED) as the first point of call. Faced with a broad array of urgent conditions, the ED has not been an efficient point of triage for patients with heart failure. Fewer than 20% of patients with heart failure presenting to the ED are discharged directly to home (8), and even fewer are likely to remain home, given that recurrent event rates for patients with heart failure after discharge from the ED have in some cases exceeded those for hospitalized patients ((5),9).

Physiologic investigation has shown that more than 90% of heart failure hospitalizations follow gradual increases in intracardiac filling pressures that are restored to baseline during therapy in hospital (10). The only real therapeutic change during most heart failure readmissions is the administration of intravenous diuretic agents, with an average fluid loss of about 4 kg and monitoring only by bedside clinical assessment and routine laboratory tests (11). The average length of stay for patients with heart failure in the United States has fallen considerably in recent years, with nearly 25% of patients now discharged within 4 days of admission (12). Because resting symptoms are frequently relieved within 24 h (13), it is reasonable to ask whether hospital admission is truly necessary for patients who 1) present with a low-risk profile for adverse events during treatment, 2) respond rapidly to initial treatment, and 3) can be followed closely in the ambulatory clinic.

In this issue of the Journal, Collins et al. (14) articulate a strong theoretical case for inserting the heart failure observation unit (OU) as an intermediate step between home discharge from the ED and inpatient admission. The choice of the term “observation” is partly strategic, as OU stays (<24 h) are currently exempt from penalties imposed on 30-day readmissions and might therefore provide a lower-cost alternative to hospitalization for selected patients. In this framework, patients with heart failure would undergo rapid stratification of risk on arrival to the ED on the basis of a limited initial evaluation and early response to doses of intravenous diuretic agents. High-risk patients would be triaged to inpatient admission, while low-risk and intermediate-risk patients unsuitable for immediate home discharge would be sent to the OU for additional evaluation and management. The investigators speculate that up to 50% of those triaged to the OU in this fashion might be sufficiently improved within 24 h to permit home discharge without the need for admission, while the rest would require extension to a conventional inpatient stay. Encouraged by the success of OUs for managing low-risk patients presenting to the ED with chest pain and a small pilot experience in patients with heart failure (15), they propose that a randomized trial powered to examine the impact of the OU approach on mortality and readmission rates in heart failure is now warranted.

The need for alternate routes to steer around heart failure hospitalization is indisputable, as is the need to embark on them without delay. However, there are daunting challenges to the immediate implementation of a randomized clinical trial to test the incremental value of this approach over routine care. The term “observation” itself is appropriate for chest pain of unknown etiology but seriously misleading when applied to heart failure decompensation, which may be mild but is never entirely benign. Regardless of whether triage takes place under the supervision of ED staff members or heart failure providers in a dedicated ambulatory unit, many active steps are necessary to ensure that the decompensation event is successfully reversed and the long-term course is stabilized (Table 1). Practically, local variation in both geography and personnel providing heart failure care (physicians, specialty nurses, pharmacists, social workers) may create substantial heterogeneity in how this transition hub should be structured to address these multiple goals.

What do we need to know before launching into a trial of such a program? The first roadblock is how best to stratify risk at the initial point of triage. Divergent secular trends in lengths of hospital stay and readmission rates for patients with heart failure (12) underscore that the selection of appropriate patients for early discharge remains a major hurdle. There are few data to formalize a decision about which patients with heart failure can be safely and effectively managed out of the hospital. The investigators have proposed a limited set of parameters (blood pressure, blood urea nitrogen, serum creatinine, and cardiac biomarkers) that discriminate the risk for mortality in the hospital with acute decompensated heart failure (16), but these have not been validated as a guide for sending patients home before full stabilization. Unmentioned factors such as cognitive impairment and inadequate social support may occasionally be of greater importance than laboratory and hemodynamic criteria in this regard.

The risk for early mortality is not the only relevant criterion for admission. The relative benefits of hospitalization over home discharge vary according to the reason for heart failure exacerbation and the location along the overall trajectory of illness (Figure 15_gr1). As systems are redesigned, care must be taken to contain excessive aversion to hospitalization that could become detrimental in complex situations for which an inpatient stay will still offer the best setting to integrate care for the rest of the journey.

Recurrent decompensation in the high-risk period early after hospital discharge (point 1 in Figure 15_gr1) may reflect incomplete treatment or accelerating renal dysfunction, for which readmission may be necessary, or care coordination failure that could be addressed during a social work consultation during an intravenous diuretic infusion. A superficially similar event disrupting the stable plateau phase (point 2) may reflect dietary indiscretion or medication nonadherence that can be rapidly addressed in the ED, or the appearance of a new condition (point 3), such as atrial fibrillation or thyroid disease, that will require complex decisions. Patients with an accelerating pattern of ED presentations in the pre-terminal phase of illness (point 4) may merit hospital admission to consider advanced heart failure therapies or redefine overall goals of care, but those in the end stage of their disease (point 5) might reasonably be discharged home if the appropriate ambulatory supports for palliative care are in place. Thus, even the first step of initial triage is probably not ready for a uniform approach to risk stratification.

For those admitted to the OU, the second triage point is uncharted territory. What is the optimal method for determining readiness to leave the hospital after <24 h? Available discharge risk scores apply only to traditional inpatient stays, with goals of complete decongestion and stabilization of fluid balance on oral diuretic agents, treatment of exacerbating factors, patient education regarding self-management, and titration of neurohormonal antagonists for long-term benefit. These will not all be achieved in a stay of <24 h, so the criteria for defining “adequate,” if not “optimal,” treatment need to be clarified. Are urine output and symptomatic response the most relevant measures? Patients often report symptom relief well before adequate decongestion has occurred, and residual elevation in filling pressures is a major determinant of rates of readmission and subsequent mortality (17). Natriuretic peptide levels correlate strongly with prognosis, but it may not be clear what level would be too high for discharge in a patient with a level low enough to have passed initial triage for observation status. Moreover, recurrent heart failure and related cardiovascular conditions account for only about half of readmissions in patients with heart failure; for the remainder who present with exacerbations of noncardiovascular illness, it remains unclear how the OU strategy should be deployed with regard to the management of medical comorbidities (18). This aspect may be particularly relevant for the nearly half of patients with heart failure with preserved ejection fraction who constitute an increasing proportion of the heart failure burden but have few options for evidenced-based medical treatment.

Early discharge is likely to be feasible only when directed to a well-developed ambulatory framework for continuing care that can properly address lingering or unresolved issues identified during the shortened hospital stay. This framework must function as efficiently on Friday night as on Monday morning, so that there is no longer a discrepancy between weekday and weekend presentations (19). Once patients leave the ED, reinforcement of the ambulatory infrastructure to permit better upstream detection and management of heart failure and prevent a return to the ED will be crucial (7).

Now that heart failure admission has moved from the profit to the loss column, every hospital is struggling to deploy and enhance available resources effectively to decrease heart failure admissions. Innovation is necessarily rapid and ongoing, even in the absence of compelling data to guide the optimal approach. The emergence of a single, uniformly effective strategy is not likely. An environment with such limited standardization of care and widespread practice variation may not be hospitable to implementation of a randomized trial, because the outcome of a fixed protocol may be outdated before it is analyzed. Until “best” practice is more clearly defined, hospitals may need to look to collaborative learning forums such as the American College of Cardiology's Hospital to Home (H2H) Initiative (20) to help select reasonable or “sound” clinical strategies that are good matches for decreasing readmissions in their own practice environments. Iterative analysis and modification of these strategies according to quality improvement processes such as the Standardized Clinical Assessment and Management Program (21) method may help practices evolve effectively in real time. It is fervently to be hoped that the universality of the challenges presented by heart failure readmissions will inspire not only new approaches but new ways to assess and share them.