Validation of a Mortality Prediction Model for Acutely Decompensated Heart Failure Patients

Overview

The purpose of this protocol is to determine if the 3 variable mortality prediction model established using data from a retrospective, multi-center patient registry (The ADHERE Acute Decompensated HEart FailuRE National Registry) will hold for a prospective, observational outcome study of OSU patients diagnosed in the Emergency Department (ED)with Acute Decompensated Heart Failure (ADHF). In the retrospective registry, three parameters were found to be highly predictive of inpatient mortality for patients admitted with ADHF - on admission, BUN \> 43 mg/dL, systolic BP \< 115 mmHg, and Cr \> 2.75 mg/dL. However, the very nature of the ADHERE database limited the analytic potential of this model, as the data captured by ADHERE is retrospective and limited to the inpatient stay, and the only outcome evaluated was inpatient mortality. Due to its inherent limitations, the model did not and could not address longer term outcomes, such as repeat visits to the emergency department after discharge, or need for readmission to an acute care setting, which frequently occurs soon after discharge in patients who survive to discharge after being admitted with ADHF. This observational study will create a registry of patient information obtained from an interview with the patient and a review of the patient's medical record. Follow up information at 30 days post discharge will be obtained by phone interview with the patient and a review of the patient's OSUMC visit history. Most ED patients diagnosed with ADHF are admitted, as emergency physicians are aware that heart failure in general carries a very high mortality rate. However, as risk stratification for ADHF is a severely under researched area, it is not at all clear which patients with acutely decompensated heart failure will have a poor outcome in the short and intermediate term. With an improved understanding of the risk profile of our ADHF patients, more appropriate decision making and disposition assignment can be made.

Congestive heart failure (CHF) affects nearly 5 million people in the United States and is steadily increasing in prevalence. It is the single most expensive diagnosis for CMS in the United States, responsible for one million hospitalizations per year. There is a wealth of evidence based management strategies for the long term care of the patient with chronic, stable heart failure. Unfortunately, there is a dearth of evidence for the management of the patient with acutely decompensated heart failure. By way of illustration, the American College of Cardiology and the American Heart Association (ACC/AHA) issued a voluminous guideline for the management of chronic CHF; to date no recommendations for the management of ADHF have been issued by the ACC/AHA. As this is an observational study, the experimental portion is solely data collection. The research team will not attempt to guide clinical care. Demographic data, vital signs at presentation, medical history, advance directive status, laboratory values on admission (BNP, BUN, Cr, troponin), use of intravenous vasoactive medications, laboratory values if discharged alive (discharge BNP, BUN, Cr), critical care utilization, final discharge diagnosis, and hospital length of stay will be collected. Patients will be contacted by phone at 30 days from initial presentation to assess survival, ED usage, and rehospitalization. Specific Aim #1: We will evaluate the performance of the three factor model (BUN \> 43 mg/dL, systolic blood pressure \< 115 mmHg, Cr \< 2.75 mg/dL) using a prospective cohort of all patients admitted with a primary diagnosis of ADHF to predict ICU utilization and hospital length of stay. Specific Aim #2: We will test the ability of the model to predict an expanded combined endpoint of mortality, readmission to an Emergency Department, and readmission to an inpatient or observation setting within 30 days of presentation with ADHF. Specific Aim #3: We will incorporate and test additional variables into the Fonarow model to derive a rule to predict the 30 day combined endpoint of death, readmission to an Emergency Department, and readmission to an inpatient or observation setting within 30 days of presentation with ADHF.