Precision and accuracy of the bedside examination in detecting an ejection fraction of less than 40% following acute myocardial infarction.

摘要

BACKGROUND: Left ventricular function is the most important determinant of prognosis following myocardial infarction. METHODS: A prospective analytical cohort study of 33 cardiac care unit survivors of acute myocardial infarction was performed to assess the accuracy, reproducibility and observer variation of the bedside Valsalva response in predicting an ejection fraction (EF) less than 40%. RESULTS: Agreement between physicians for the clinical Valsalva response was excellent (kappa coefficient [k] 0.75), as was the comparison between physicians' clinical response to the Finapres (Ohmeda, USA) hemodynamic response (weighted k=0.85). The EF was significantly higher among patients with a normal Valsalva response (56.9%) than in patients with either an absent overshoot (48.4%) or a square wave (28.3%) response (P<0.001). Physicians were very accurate at estimating whether the EF was greater or less than 40%. In 40 of 66 situations, the clinicians were confident, based on the clinical examination and the Valsalva response, that the EF was either greater or less than 40%. In these situations, agreement (95.0%) and k (0.89) were both excellent. When all patients were considered, the degree of agreement (90.8%) and k (0.80) diminished slightly. A square wave response had poor sensitivity (37.5%) but excellent specificity (92.7%), whereas any abnormal response had excellent sensitivity (91.7%) but poor specificity (54.8%). CONCLUSION: Compared with many other aspects of the clinical examination, the bedside Valsalva manoeuvre has acceptable degrees of interobserver variability. A normal response tends to rule in an EF of greater than 40%, whereas a square wave response rules out an EF of greater than 40%. This simple bedside manoeuvre may be useful in predicting low EF following acute myocardial infarction.