Review: Chest radiography lacks utility for detecting left ventricular dysfunction
ACP J Club. 1997 May-Jun;126:77. doi:10.7326/ACPJC-1997-126-3-077
Badgett RG, Mulrow CD, Otto PM, Ramírez G. How well can the chest radiograph diagnose left ventricular dysfunction? J Gen Intern Med. 1996 Oct;11:625-34.
To determine the usefulness of chest radiography for diagnosing increased left ventricular (LV) preload or reduced ejection fraction (EF).
Studies were identified using MEDLINE (1966 to February 1995) with the keywords heart failure, x-ray, or radiograph; bibliographies of review articles, primary articles, and textbooks; investigators' files; and by contacting authors of studies.
Studies were selected if they compared chest radiographic findings with an acceptable diagnostic standard for heart failure (reduced EF or increased LV preload). Studies were excluded if >80% of the patients had valvular heart disease.
Data were extracted in duplicate on patient numbers and characteristics, definitions of abnormal findings, sensitivity and specificity of radiographic findings, diagnostic standard (LV preload or EF), and methodologic criteria (interpreter training and blinding, film technique, timing and clinical stability between radiography and diagnostic standard).
29 studies met selection criteria. Chest radiographs were compared with LV EF in 10 studies, LV preload in 18 studies, and both in 1 study. Sensitivity, specificity, and likelihood ratios for a positive (LR+) and negative (LR-) test result are listed in the Table. Bayesian analysis found that chest radiography can only exclude heart failure (post-test probability < 5%) and a decreased EF in patients who are asymptomatic and can never confirm heart failure or a reduced EF (post-test probability > 95%).
The radiographic findings of redistribution and cardiomegaly are most effective for diagnosing increased left ventricular preload and reduced ejection fraction, respectively. Neither finding alone, however, has adequate diagnostic utility for left ventricular dysfunction.
Sources of funding: San Antonio Cochrane Center and the Audie L. Murphy Memorial Veterans Affairs Hospital.
For article reprint: Dr. R.G. Badgett, Division of General Medicine, Department of Medicine, University of Texas Health Science Center, 4200 Floyd Curl Drive, San Antonio, Texas 78284-7879, USA. FAX 210-223-6166. E-mail firstname.lastname@example.org.
Table. Test Features for the Diagnosis of LV Dysfunction
|Clinical State||Radiographic Finding||Sensitivity, % (95% CI)||Specificity, %||LR+*||LR-*|
|Increased LV preload||Cardiomegaly||44 (29 to 60)||88||3.67||0.64|
|Redistribution||65 (55 to 75)||67||1.97||0.52|
|Interstitial edema||45 (32 to 59)||81||2.37||0.68|
|Reduced EF||Cardiomegaly||51 (43 to 60)||79||2.43||0.62|
|Redistribution||37 (25 to 51)||85||2.47||0.74|
|Interstitial edema||18 (2 to 57)||97||6.00||0.85|
*Numbers calculated from data in article.
Many cardiac disorders manifest themselves as LV systolic dysfunction, from subtle and subclinical dysfunction to overt congestive failure. The history and physical examination are the first step in identifying LV dysfunction and often its cause. Cardiac imaging is commonly used to confirm, exclude, or assess the severity of LV dysfunction. In a high-technology world of imaging options (ultrasonography, radionuclides, angiography, computed tomography, and magnetic resonance imaging), the relatively inexpensive, familiar, and widely available chest radiograph continues to be frequently used.
The review by Badgett and colleagues delineates the sensitivity and specificity of chest radiography in the setting of suspected LV dysfunction. Although the authors did a careful study with excellent statistical methods, 3 caveats should be observed. First, as the authors indicated, they evaluated individual radiographic signs in isolation, whereas clinicians usually use clusters of findings from the history, physical examination, electrocardiogram, and chest radiograph to make diagnoses, which is probably a more powerful approach. Second, the criterion standards used in the studies varied in the definition of abnormality (e.g., criteria for abnormal EF ranged from < 35% to < 50%), potentially adding additional "noise" to the data. Third, the authors identified patients in whom the radiographic signs could confirm or exclude increased preload or LV dysfunction (patients with high or low pretest probability, respectively). In patients in whom LV dysfunction is almost certainly present or absent, chest radiography is unlikely to substantively alter the post-test probability. It is for patients with intermediate pretest probability (e.g., patients with dyspnea of uncertain origin after myocardial infarction) in whom chest radiography may yield diagnostically important information by altering the post-test probability of disease. Future research could focus on identifying radiographic and clinical clusters for diagnosis in patients with intermediate risk.
Chest radiography is best used when the clinician wishes to estimate cardiac size and configuration and the state of the pulmonary vasculature with 1 examination. When detailed data on LV size and function are required, chest radiography should be supplanted by echocardiography or other quantitative, noninvasive imaging methods.
David J. Skorton, MD
University of Iowa Hospitals and ClinicsIowa City, Iowa, USA
David J. Skorton, MD
University of Iowa Hospitals and Clinics
Iowa City, Iowa, USA