Cardiothoracic Ratio on CXR Does Not Predict Ejection Fraction in CHF


Synopsis: Philbin and colleagues found that the CXR determination of cardiothoracic ratio is not a valid way of estimating EF or systolic vs. diastolic dysfunction.

Source: Philbin, et al for the Digitalis Investigation Group. Arch Intern Med 1998;158:501-506.

A differentiation between systolic and diastolic dysfunction is an important component of the diagnosis, prognosis, and treatment of congestive heart failure (CHF). The ability to use a common diagnostic test such as a chest x-ray (CXR) would be attractive to differentiate systolic and diastolic dysfunction.

Philbin and colleagues had the luxury of the Digitalis Investigation Group data. These New England Journal of Medicine data dealt with CHF and the use of digoxin. Also required in the study were CXRs and some other measurement of ejection fraction (EF) such as radionuclide study, echocardiogram, and angiogram.

Philbin et al took a look at all 7476 of the patients who had entered the trial. A total of 254 patients were excluded on the basis of valvular disease, which can confuse the dilatation in CHF. Another 60 or so patients were excluded because of inadequate data.

The EF had been calculated in the remaining patients using radionuclide (66%), echocardiograph (29%), or angiogram (5.5%). The cardiothoracic ratio was measured in the conventional fashion. By comparing these two and statistically analyzing with correlative and categorical analysis, Philbin et al were able to show a weak negative correlation. (What one would have wanted was a strong negative correlation.)

Philbin et al have shown that the CXR determination of cardiothoracic ratio is not a valid way of estimating EF or systolic vs. diastolic dysfunction. A more expensive test is needed.


Every primary care physician is faced with a patient with new onset of CHF. One of the more common tests ordered in this scenario is CXR. The residents are taught not to use this to make the diagnosis, but it certainly corroborates the clinical suspicion. Our eyes naturally drift to the size of the heart. It would be attractive if the size of the heart would help us differentiate systolic and diastolic dysfunction. The reason we need this differentiation is because of the different therapeutic choices in these two syndromes. I must admit, it was attractive to estimate whether new onsets of CHF were from systolic or diastolic causes based on that CXR.

What I had always been taught in medical school is that there were significant respiratory variations and systolic and diastolic differences in size of the heart on CXR. There was no way to "gate" the timing of the CXR in systole or diastole sequential chest x-rays; even if only 30 seconds apart, they could vary widely in the size of the myocardia.

In an acute CHF episode with fluid overload, a diastolic dysfunction can look like a systolic dysfunction (even on echocardiogram). For that reason, I have cautioned my residents to wait several days before ordering an echocardiogram on a CHF patient. Order it acutely if absolutely necessary for other reasons (like ruling out a valvular abnormality that might make an acute therapy necessary). I firmly believe that waiting just prior to discharge for that echocardiogram or maybe even a week later will give you a more accurate representation of the difference between diastolic and systolic dysfunction.

Philbin et al have proved that the CXR cannot be used to tell the difference between systolic and diastolic dysfunction. The percentages of how they measured EFs are slightly different than my practice. I noted that they had almost twice as many radionuclide EF studies as echocardiogram EF studies. Regardless, their study indicates that we do need to bump up to the more expensive test of either radionuclide, angiogram, or echocardiography to make that differentiation between systolic and diastolic dysfunction. Most clinicians had already known this and were already doing this in order to estimate EF. The "rough" look of the heart that we get from the CXR might give us a clue as how to initially treat a CHF patient, as a systolic or diastolic dysfunction can be erroneous.

Primary care clinicians should not use the cardiothoracic ratio on CXR to help determine whether a patient's etiology of CHF is from systolic or diastolic dysfunction.