HF imposes a staggering economic burden on the healthcare system, with most of the costs for the care of patients with HF attributable to hospitalizations. More than 1 million hospitalizations for HF occur in the United States annually. Also, HF has the greatest 30-day rehospitalization rate of any diagnosis among Medicare beneficiaries. Accordingly, the initial and subsequent hospitalizations among patients with HF constitute a major public health problem and have been a recent focus in the healthcare reform debate.

Despite considerable discussion regarding the appropriateness of the metric, much of the energy of policymakers and private payers has centered on the reduction of 30-day rehospitalizations, particularly for resource-intensive diagnoses such as HF. Since 2009, the Centers for Medicare and Medicaid Services have publicly reported 30-day rehospitalization rates after HF hospitalization. Within the Affordable Care Act, Congress has directed the Centers for Medicare and Medicaid Services to financially penalize hospitals with greater than expected 30-day rehospitalization rates. With a goal of reducing preventable rehospitalizations, researchers have been fervently exploring methods to predict rehospitalizations after HF admission to identify potential targets for intervention.

Despite intensive effort, predicting rehospitalizations has proved to be challenging, and available models have suffered from poor discrimination and have often pinpointed factors, such as patient demographics, that are not amenable to intervention or modification. The number of actionable factors identified has been limited, with a major focus placed on modification and improvements during the patient’s transition from hospital to home. In particular, early physician outpatient follow-up within 7 to 10 days of hospital discharge has been adopted as a critical component of the home transition, because greater rates of early follow-up were associated with lower readmission rates in a single study, although in a separate analysis, this metric had no association with general medicine readmissions. Although at least one half of hospitalizations of patients with HF are from noncardiovascular causes, rehospitalizations for recurrent HF might be particularly amenable to preventive efforts. It is well-established that pulmonary capillary wedge pressure might increase well in advance of the onset of clinical HF symptoms. Thus, if it were possible to identify these patients at an early stage and intervene, rehospitalizations could potentially be avoided. Implantation of CRT devices has been shown to decrease mortality and, as such, are an American College of Cardiology/American Heart Association class I indication in patients with established HF who continue to experience New York Heart Association functional class III or IV symptoms with an ejection fraction (EF) of ≤35% and QRS duration of ≥120 ms, despite optimal medical therapy. The number of patients with advanced HF with CRT devices continues to increase, and these devices contain potentially important information regarding the patient’s volume status and cardiac function. The role that implantable device technology might play in identifying patients with HF before overt decompensation in which providers can intervene and avoid rehospitalization is particularly intriguing and relatively unexplored.

In this issue of the American Journal of Cardiology, Whellan et al report provocative findings from a pooled analysis of patients with CRT with defibrillator (CRT-D) devices enrolled in 4 studies. The 166 participants with HF experienced 34 rehospitalizations within 30 days of HF hospital discharge. Information obtained from the CRT-D device diagnostics 7 days after hospital discharge aided in the identification of patients at the greatest risk of 30-day rehospitalization for HF. Although limited by the small number of events, their results are thought provoking and worthy of additional investigation in larger patient populations.

Whellan et al examined the data available from patients enrolled in the PARTNERS-HF, OFISNER, FAST, and CONNECT studies who had had an HF hospitalization and 30 days of follow-up data after hospital discharge. The authors then investigated the association between 4 variables (ie, intrathoracic impedance, atrial fibrillation burden plus ventricular response rate, nocturnal heart rate [NHR], and heart rate variability [HRV]), obtained from the Medtronic CRT-D devices at 7 days after discharge, and the risk of rehospitalization within 30 days after hospital discharge. From the association with rehospitalization, the thresholds for each device parameter and a combined diagnostic score were developed to aid in the stratification of patients into low-, intermediate-, and high-risk categories for rehospitalization. The total 30-day rehospitalization rate was 35%, 10%, and 2% in the high-, intermediate-, and low-risk groups, respectively. Although the number of events was small, the authors did an excellent job of reducing the complicated device diagnostic data into a risk score that could be easily applied in the clinical setting. These findings build on their previous work in the PARTNERS study of 694 patients with Medtronic CRT-D devices, in which they established that an abnormal finding using a similar device diagnostic algorithm predicted hospitalization for HF within the next month. The pooled data presented now demonstrate that a similar approach can be useful at hospital discharge to identify those patients at risk of HF readmissions.

Of the individual diagnostic parameters examined, thoracic impedance appeared quite useful in discriminating low-, intermediate-, and high-risk patients. This is not surprising, because thoracic impedance has been shown to inversely correlate with the pulmonary capillary wedge pressure and could represent a surrogate marker for volume overload in patients with HF. Similar to the slow increase in pulmonary capillary wedge pressure observed before the clinical signs and symptoms of HF occur, thoracic impedance might decrease before any signs of volume overload occur. In the Medtronic Impedance Diagnostics in Heart Failure Trial (MidHeFT), a decrease in thoracic impedance preceded the onset of dyspnea by 15.3 ± 10.3 days, considerably earlier than the actual onset of symptoms that led to hospitalization (3 ± 2.5 days). In the OFISNER study, a multicenter, retrospective, cohort study of patients with CRT-D, the patients who frequently crossed a nominal threshold for thoracic impedance were at increased risk of HF admission during the study period. Similarly, 1 pilot study implemented an Internet monitoring system that uploaded OptiVol assessment, triggering a nurse to telephone the patient when the thoracic impedance crossed a specified threshold (>60 Ω for ≥10 days). The investigators found a high number of clinically important events, including hospital admission, dietary indiscretion, or medication changes, in those who crossed this threshold. Changes in thoracic impedance might hold particular promise in identifying those at the greatest risk of decompensation from HF who might be amenable to intervention.

Although these findings are intriguing, 2 important questions are raised by these data. First, what will be the effect on the prediction of HF rehospitalization if a device diagnostic approach is added to the early outpatient follow-up visit after a HF hospitalization? Despite their increasing use, only a fraction of the HF population are eligible for CRT devices, because most patients with HF have an EF >35%. However, patients with preserved EF are at a similar rehospitalization risk, and these data would not apply to that important population. Furthermore, HF is the primary indication for admission in only a fraction of the rehospitalizations in patients with a reduced EF. Although device diagnostics might aid in identifying patients with CRT devices who are at increased risk of HF readmissions, noncardiovascular hospitalizations are increasingly common and would not be predicted by these data. Finally, although the national 30-day rehospitalization rate for HF hovers around 24%, many of these readmissions occur within 7 days after hospital discharge and would occur before the 7-day outpatient follow-up that was strongly advocated. In the report by Whellan et al, 12 of 34 rehospitalizations (35%) occurred in the 7 days after discharge. Also, although the device diagnostic algorithm still predicted rehospitalizations when these events were removed, the predictive ability was attenuated. Therefore, a device diagnostic approach is not likely to be effective as a standalone strategy in the identification of patients at greatest risk of readmission, although it could be considered for incorporation into a comprehensive management strategy for patients with HF after hospitalization.

Second, even if we can accurately pinpoint patients at the greatest risk of readmission, how can we translate this into the prevention of rehospitalization? Although the study by Whellan et al, as well as numerous other published studies, have focused on identifying the patient- and hospital-level factors most associated with rehospitalization, actually preventing readmission is likely to prove much more difficult. Despite promising observational data, several telemonitoring trials of HF have already shown no benefit in preventing readmissions. Among the hospitals enrolled in the Hospital-to-Home Quality Improvement Initiative, 90% reported that reducing readmissions was a hospital-wide objective. However, many of the instituted efforts, although important in providing coordinated and integrated patient care, have had little proven benefit in reducing rehospitalization. Although pay-for-performance has provided strong motivation to hospitals to take action to reduce rehospitalizations, little underlying evidence is available to guide their efforts.

Although we must commend Whellan et al for their efforts in taking advantage of existing device technology to improve patient care, their study is primarily hypothesis generating. Larger studies are needed to confirm their findings, and a trial of an intervention based on abnormal device diagnostic data would be needed before the routine implementation of this approach in clinical practice.

There is no easy answer to the rehospitalization problem. However, we must not forget that patients with HF are experiencing significant morbidity at the center of this metric. Taking full advantage of implantable device diagnostic data and incorporating this information into the care of our patients with HF could help us to provide higher quality, more patient-centered care, laudable goals that are sometimes overlooked in the economically focused healthcare debate.