Calculating the best ED decisions for acute heart failure patients

Heart failure has a high fatality rate and, commensurate with this risk, is one of the costliest diseases in all of health care.

Often, the care of heart failure patients has a “one-size-fits all” approach, which means it is not necessarily tailored to their prognosis. There is a lack of “precision” in how clinicians manage heart failure patients because that prognosis is not routinely considered in day-to-day care.

Important decisions such as whether to admit a patient with acute heart failure to hospital are not immune to our tendency to treat all patients uniformly. Thus, the majority of heart failure patients who present to the emergency department (ED) are admitted – and typically spend over a week on the hospital ward. As a result, somewhat “healthier” patients are often admitted to hospital, while “sicker” patients (who may not appear unwell to the treating physician), might be inappropriately discharged home.

What if we can identify the patients who are lower risk and could actually go home instead of stay in-hospital?  Can we identify high-risk patients who should not be discharged home despite a good response to initial treatment? Can these methods accurately improve physician estimates of their patient’s future risk?

These notions are at the heart of an algorithm my team has developed: the Emergency Heart Failure Mortality Risk Grade (EHMRG).

Seven-Day Mortality Risk

EHMRG helps identify a heart failure patient’s risk level of dying by entering a patient’s information into a calculator. The resulting score and graph illustrate the risk probability and risk decile of that patient.

By knowing that patients have a low risk, with a more favourable prognosis, physicians may feel more comfortable discharging them. Alternatively, lower-risk heart failure patients can be discharged earlier than the average eight-day hospital stay. Benefits to patients are clear, and the hospital itself has more control over the flow and usage of beds – which in turn elevates the level of care for all patients.

Openly available online, EHMRG is used now in many countries. A multicentre prospective validation of EHMRG is currently ongoing in Ontario, where one of the objectives will be to compare the risk score with physician-estimated risk. It will definitely determine if EHMRG predictions provide better discrimination or reclassification of patient risk than the physicians’ estimates.

30-Day Mortality Risk

We modified the 7-day model to develop a new application that can simultaneously predict a patient’s risk of death over a full month. This new algorithm charts a patient’s probability at 7 and 30 – critically important since a person could be low-risk at one week and high risk at one month. I believe this type of multi-dimensional prognostication is where risk modelling is headed, as people are interested in multiple outcomes – in this specific case, outcomes at two separate times.

This combined 7- and 30-day model is being investigated now, with results expected in late 2018, under a study called “Comparison of Outcomes and Access to Care for Heart Failure (COACH).” We are trying to determine if care decisions fed by the 7/30-day EHMRG model will improve decision-making in the ED, reduce hospitalizations and improve survival.

This is an area of vast importance in health care. One in 25 heart failure patients die within 30 days of their presentation to the ED. Consequently, there is much fear about sending patients with heart failure home from the hospital.

The prospects of EHMRG are potentially fewer low-risk admissions to hospital and fewer high-risk patients inadvertently discharged from the ED. The net effect could be a reduction in hospitalizations and potentially enormous savings for the health-care system.

Dr. Douglas Lee is the Ted Rogers Chair in Heart Function Outcomes, cardiovascular program lead at the Toronto General Research Institute and cardiologist at the Peter Munk Cardiac Centre (both at UHN), senior scientist at the Institute for Clinical Evaluative Sciences, and associate professor of medicine at U of T.

Comments are closed.