Abstract
Background and purpose
Diagnosing heart failure (HF) remains difficult in the acute setting where multiple diagnoses are in play. Objective evidence of pulmonary congestion by chest X-ray (CXR) is one criteria for the recent universal definition of heart failure (UniHF). But, since CXR is known to have a low diagnostic value, we hypothesized that a chest CT (CT) would outdo the CXR to diagnose decompensated HF in acute breathless patients. This study's primary objective was to examine if the CT has higher accuracy than the CXR to diagnose HF in the acute setting; and, secondly, to identify what pre-test characteristics would predict a false negative CXR or CT.
Methods
We performed a single-centre, prospective observational study and included consecutive adult patients with dyspnoea in the emergency department. Patients underwent immediate clinical examination, blood tests, CXR, CT and an echocardiogram. Congestion on CXR and CT was defined as the congruent verdict by two expert thorax radiologists, blinded to each others reading and all other clinical data. The absence of congestion was defined as the congruent verdict of “no congestion”.
Congestion of CXR and CT was held up against UniHF ascertained by an expert panel of cardiologists where the pulmonary congestion component primarily was based on elevated filling pressures from the simultaneous comprehensive echocardiogram. Univariate- and multivariate logistic analyses identified factors associated with a false negative chest x-ray and CT.
Results
Of 228 patients with a mean age of 74,5 years, 129 (56,5%) were male, 98 (43%) had UniHF, and 139 (61.0%) had pulmonary disease. Congestion on the CXR diagnosed UniHF with a 54% sensitivity and 95% specificity, with almost similar figures for the CT with 54% and 99% respectively. A marginally better performance of the CT was shown by a significantly lower Akaike Information Criterion for pulmonary congestion by CT than for CXR. However, the net reclassification improvement by CT was 4% (p:0.5586). The CXR and CT were false negative for UniHF in 46% (45/98) for both modalities (Table 1). The only independent pre-test predictor of a false negative radiology examination in multivariable logistic regression analysis was NT-proBNP (CXR: OR 1.670 per log(BNP), p: <0.001) and CT: OR 1.693 per log(BNP), p: <0.001).
Conclusions
For the first time, CT has been directly compared with CXR to diagnose HF in consecutive breathless patients from the emergency department. The chest CT was marginally more specific than the CXR to diagnose HF, but with a similar sensitivity. Approximately half the patients obeying the universal definition of HF have no definite congestion on CXR nor CT, and these can only be identified by a high proBNP.
FUNDunding Acknowledgement
Type of funding sources: None. Table 1
A Report of the Heart Failure Society of America, Heart Failure Association of the European Society of Cardiology, Japanese Heart Failure Society and Writing Committee of the Universal Definition of Heart Failure Consensus Conference