Prediction Rule to Identify Low Risk Patients with Heart Failure Final rept. (Sept. 30, 2000-Sept. 29, 2003)

摘要

Over 3 million patients are hospitalized each year in the U.S. for heart failure, with most admitted from hospital emergency departments. Provider prognostic estimates at the time of admission decisions in patients with new or worsening heart failure are often inaccurate. We sought to derive a clinical prediction rule using clinical information available in the emergency department that identifies patients with heart failure at low risk of an adverse medical outcome. We analyzed data for all 33533 patients with a primary hospital discharge diagnosis of heart failure in 1999 admitted from emergency departments in Pennsylvania. We used classification tree analysis to derive a rule that identified at least 10% of the cohort to be at low risk of inpatient death (rate <1%) or survival to hospital discharge with a serious medical complication (rate <3%). Within the entire cohort, 4.5% died and 11.2% died or survived to hospital discharge with a serious medical complication. Our prediction rule used 19 prognostic factors to classify 17.1% of patients as low risk; 19 (0.3%) died and 75 (1.3%) died or survived with a serious medical complication. Our clinical prediction rule identifies patients traditionally hospitalized from the emergency department for the management of heart failure who are at low risk of inpatient death or serious medical complications. If validated, this rule could be used to identify candidates for whom outpatient therapy may be appropriate.