HF constitutes a heavy burden to healthcare systems worldwide. Specifically, HF patients suffer high mortality rates and are often readmitted [10]. Consequently, there is an urgent need for developing reliable risk scores which will help with identifying high-risk HF patients, especially those with an acute decompensated HF event. We have demonstrated here that using a simple echocardiographic score, HF patients could be reliably classified into distinct risk groups. The fact that these results were observed within a 30-day period might assist with focusing medical efforts in preventing HF readmissions to high-risk groups of HF patients. Notably, the highest-risk group (eKillip Class IV), which persistently showed statistically significant increased risk for both HF hospitalization and/or mortality, comprised about one-third of the patient population. These important data imply that once appropriately classifying this group as a high-risk one, a focused and dedicated follow-up (and potential interventions) may assist with significantly decreasing overall rehospitalizations rates.

The need for developing clinical, laboratory or imaging tools for identifying high-risk HF patients prone for rehospitalizations and mortality continues to challenge the medical world and different models have been suggested [16,17,18]. The CHAMPION trial showed that implanting a pressure-monitoring device into one of the pulmonary artery branches provided remote monitoring of pulmonary pressures and induced a significant reduction in rehospitalizations rates [19]. Other devices and models are currently being investigated [20, 21], but all of them require an invasive procedure and dedicated monitoring. Other, simpler methods, such as natriuretic peptides measurement, showed conflicting results as to their ability in predicting post-discharge recurrent hospitalizations and mortality [22,23,24].

Echocardiography is an accessible, noninvasive, reproducible and reliable tool that is often used to evaluate patients with HF. Studies by the working group on HF of the Italian society of cardiology [8, 9] have shown that different hemodynamic profiles, as assessed by echocardiography, can predict prognosis in HF patient evaluated in the ambulatory setting. Patients with low flow state and elevated filling pressure have the worst outcome.

Consequently, different echocardiographic scores were developed and showed good predictive abilities regarding post-discharge clinical outcomes. Nevertheless, most were either cumbersome or used sophisticated echocardiographic methods [25,26,27]. For example, Thavendiranathan et al. examined the additive effect of echocardiographic findings to an elaborate risk-prediction tool (the Yale-CORE HF readmission score) and showed that elevated right atrial pressure and left-sided filling pressures added to the predictive ability of the model [27]. In another study, Saito et al. showed that reduced left ventricular global longitudinal strain was associated with worse post-discharge clinical outcomes [26]. Although important, these studies emphasize the need for a straightforward tool which will assist the everyday clinician with identifying HF patients at risk for rehospitalizations and mortality. Our suggested eKillip fits to this description well. The suggested parameters in our model (SVI and E/E') are regularly examined during echocardiography in most facilities. We intentionally chose filling pressure indices which do not require sinus rhythm (i.e. "a wave") and can be applied to the entire HF population including those with AF. Also, the cutoffs which we have used do not significantly differ from the ones used in the routine evaluation of HF [11].

Examining the performance of our model shows that although the categorization of risk did not always reach a statistical significance, it was repeatedly able to categorize the highest-risk group (i.e. eKillip Class IV) appropriately, including following corrections for both clinical and echocardiographic parameters. Notably, this group did not differ from the overall population in other important features such as age, kidney function, the presence of CAD or discharge medication use, emphasizing the added predictive ability of the eKillip.

The pathophysiological basis of our findings emerges from the one which dictated the original Killip score since it captures the fundamental function of the left ventricle. That is, to be able to produce normal perfusion while maintaining normal intracavitary pressures and thereby preventing lung congestion. Numerous trials have demonstrated the importance of SVI and diastolic function on patients' outcomes [28,29,30,31,32,33]. Furthermore, the predictive ability of both SVI and diastolic function on survival was shown to be superior to LVEF in a recent study conducted in cardiac intensive care patients, emphasizing the importance of perfusion and congestion over systolic function in the acute setting [34].

Our study has a few limitations. First, although large and comprehensive, this is a single-center, retrospective study which did not include the initiation time or the adherence to medical therapy. Second, though echocardiography was done during the index admission, its exact timing might have influenced the results. Third, neither cardiac output (as a surrogate for peripheral perfusion) nor diastolic function (as a surrogate for pulmonary congestion) were fully evaluated. Nevertheless, our aim was to produce a simple tool which will assist with everyday clinical practice and decisions.

In conclusion, we have demonstrated that a simple and reproducible echocardiographic score was able to identify HF patients at risk for 30-day readmissions and mortality. Further studies are needed to test the consistency of our findings in other cohorts. While echocardiographic scores might be found as promising tools for identifying patients at risk, they should not be considered as a substitute for a full echocardiographic assessment.