To the Editor: Dr Al-Khatib and colleagues1 attempted to quantitate the subset of ICD implantations in the United States that are inappropriate by virtue of being non–evidence based. However, the magnitude of the inappropriate device implantation problem may actually be greater than that reported if procedures that conform to evidence-based guidelines in appearance only are taken into account—namely, implantations performed as a result of erroneous fulfillment of accepted criteria or as a result of overdiagnosis of conditions for which ICD therapy is legitimately warranted. For example, ICD implantation is indicated for primary prevention of sudden cardiac death in post-MI patients, subject to various additional criteria, as long as the left ventricular ejection fraction is below a specified cutoff value (≤30%, or ≤35% with prior congestive HF). In such patients, therefore, the decision to implant an ICD hangs on the accuracy of a measured imaging parameter. Yet quantitation of ejection fraction may vary by interpreter or by institution, when based on the commonly used echocardiographic visual estimation technique,2 and can also vary as a function of the imaging modality, ie, echocardiography vs radionuclide ventriculography. If ejection fraction measurements are systematically skewed toward lower values via particular interpreters or imaging modalities at certain institutions, a subset of the ejection fractions so generated may be below the requisite cutoff value, resulting in an excess number of unnecessary—but seemingly indicated—ICD implantations.