ASE: Predicting Heart Filling Pressure -- Sensitivity or Specificity, Not Both

SEATTLE, June 20 -- When estimating left ventricular filling pressure to assess cardiac risk, physicians can choose between clinical or echocardiographic criteria, depending on whether they want sensitivity or specificity.
That mixed message came from a comparison of two prognostic algorithms to estimate 114 patients' risk for hospitalization or death presented by Kofo O. Ogunyankin, M.D., of Queen's University in Kingston, Ontario, at the American Society of Echocardiography meeting here.
Dr. Ogunyankin and colleagues evaluated all the patients clinically and by echocardiography and left ventricular catheterization. Each patient was assigned to a risk category on the basis of clinical variables and to one of three diastolic function classes determined by echocardiography.
After three years of follow up, the choice of prognostic scheme remains a toss up, they said.
"Both clinical and echocardiographic methods for estimating high filling pressure help identify subsets of patients with an increased likelihood of death or cardiac hospitalization," Dr. Ogunyankin said. "The clinical parameters work if one just follows well-recognized criteria in an organized way. However, the echo parameter is more specific."
Dr. Ogunyankin and colleagues developed a clinical algorithm for estimating the probability of a high left ventricular filling pressure and a three-stage echocardiographic classification of diastolic function. In an earlier study, the group showed that the two approaches to risk assessment have comparable accuracy, as determined by catheterization and measurement of filing pressure (Echocardiography. 2006;23:817-828).
In this report, Dr. Ogunyankin presented the long-term outcomes from the study.
On the basis of clinical criteria for predicting elevated filling pressure, 44 of the patients had been deemed high risk, 42 intermediate risk, and 28 low risk. The risk stratification results were compared with the echo criteria for the classes of diastolic function: normal, impaired relaxation, and severe diastolic dysfunction.
By three years after initial assessment, 29 patients had died or had been hospitalized. Of those 16 (55%) had been classified clinically as high risk and 10 (34%) had met echo criteria for severe diastolic dysfunction (P=0.014 in favor of clinical assessment).
Of the 85 patients who had no clinical events during follow up, 57 (67%) had been classified clinically as "non-high risk" and 73 (86%) had "non-severe" diastolic dysfunction by echocardiography (P=0.002 in favor of echo).
The odds ratio for death/hospitalization in the high-risk clinical group was 2.5 compared to the rest of the patients (P=0.04). The severe category of echo-determined diastolic function was associated with an odds ratio of 2.7 (P=0.04).
Noting that the two stratification schemes were validated in a group of patients with a high prevalence of coronary artery disease, Dr. Ogunyankin said the algorithms require validation in patients with other types of cardiac pathology.
He also said the results also did not answer the key question about the two risk stratification methods: Can they be used to guide patient management?