Identifying CHF Patients as Low-Risk for Adverse Outcomes

The Case:
A 65-year-old male presents to your ED complaining of increasing exertional dyspnea and orthopnea limiting daily activity…

While a relatively simple tool like the Acute Heart Failure Index (AHFI) could be a useful aid to the challenging disposition decisions in the management of acute heart failure, the current tool requires broader testing before general use.

The Case:
A 65-year-old male presents to your ED complaining of increasing exertional dyspnea and orthopnea limiting daily activity. He has previously been diagnosed with congestive heart failure secondary to hypertensive cardiomyopathy, but has no history of diabetes or ischemic events. He reports some edema of the lower extremities but no chest discomfort or recent changes in medication, and the remainder of the review of systems is unremarkable.

Physical examination revealed rales and bilateral pedal edema. The electrocardiogram was unremarkable and the blood work including troponin was normal. Following 1.5L diuresis, your patient’s symptoms improve significantly so he wants to go home. Is it safe to discharge him?

The Study:
Validation of the Acute Heart Failure Index. Hsieh M, Auble TE, Yealy DM. Ann Emerg Med. 2008; 51(1):37-44.

Background:
The burden of illness associated with heart failure is enormous, resulting in over a million US ED visits and costs measured in the billions annually. The reported mortality and adverse rates vary significantly, leaving the clinician with little consistent direction in determining those patients requiring admission. Current EM guidelines offer little disposition guidance for CHF patients. Clinical decision rules (CDR) have been shown to enhance physician diagnostic and prognostic decision-making in scenarios of high frequency (Ottawa Ankle Rules) or high risk (Well’s DVT Score). Acute heart failure is both common and high risk so a reliable decision aid could be of great benefit.

The authors attempt to validate their previously derived AHFI by applying it to a computer generated random sample of 5000 patients from each of 2003/2004 in the Pennsylvania Hospitals (~10000 out of over 100000 ED visits for acute heart failure). With such a large and detailed database the authors have been able to derive a rule and test it on a large cohort of patients, something that would be challenging to do prospectively. The rule uses a rather complicated algorithm containing a number of predictors (Figure 1). All branches in the algorithm identify low-risk patients, while all others are high-risk.

The Results:
Application of the AHFI separated patients into two groups (high or low risk) and the primary and secondary outcomes for each are presented in Table 1. From the derivation cohort, the AHFI classified just 17% of subjects as low-risk. The AHFI had a poor positive-Likelihood Ratio for inpatient deaths or complications (1.21), but had a reasonable negative-Likelihood Ratio (0.11) for these outcomes. Even in the low risk group the event rate for death or serious outcomes ranges from 1-3%. Although this equates to a negative-LR of 0.31 (95% CI 0.24-0.41) for 30-day mortality this may not be sufficiently low for many clinicians.

Click on image to view high-res PDF
Figure 1. Heart failure clinical prediction rule
All branches identify low-risk patients; all other patients are at higher risk.
*Test result also applies to patients for whom the test was not ordered. ECG, Electrocardiography; MI, myocardial infarction; BUN, blood urea nitrogen; WBC, white blood cell count; PTCA, percutaneous transluminal coronary angiography; SBP, systolic blood pressure; AMI, myocardial infarction.

Case Review:
After reviewing the article and collecting the patient’s data you look at the rule; any path that leads to the far right is considered low risk. Your patient, with no ECG evidence of myocardial ischemia or infarction, fairly normal vital signs (HR 90, BP 160/80, RR 20, SpO2 96%), unremarkable labs (BUN 34 WBC 9 Sodium 140) and no plural effusion, matches the path second to the bottom and is therefore classified as low risk by the AHFI. You explain to your patient this still probably means a 1-3% risk of death, MI, or V-fib. Nonetheless, with informed consent he chooses to go home.

In Summary:
Heart failure is a major cause of morbidity, mortality and increasing health care expenses. Any tool with the potential to improve a clinician’s decision making can have a major impact.  The AHFI is the best currently available evidence, but prospective validation in different patient populations would be needed before it is considered sufficiently reliable to be used routinely.

John Crossley, MD, FRCPC, is the Emergency Medicine Residency Program Director at McMaster University