Heart failure: chest X-rays were not very good at diagnosing or excluding it

Clinical bottom line (level 2a)

  1. Chest X-rays alone could not safely diagnose or exclude left ventricular dysfunction in patients with dyspnoea, a recent myocardial infarction or from the general population.
  2. Cardiomegaly, venous redistribution or congestion and interstitial edema were little different from each other in distinguishing increased preload or decreased ejection fraction.
Badgett et al: Journal of General Internal Medicine 1996; 11: 625-634
Expires October 2003

The study

Systematic review of all studies of
  • Patients: with suspected left ventricular dysfunction
  • Intervention: chest X-ray compared with left ventricular ejection fraction or left ventricular preload
  • Outcome: left ventricular dysfunction


    • Articles found in English using Medline, 1966 to 1995 (search terms: heart failure and X-ray or radiograph ) and reference lists from relevant reviews and primary articles and textbooks, the authors' files and data from experts and original authors.

    Selection criteria: see above
    Appraisal criteria: detailed in text addressing whether studies were blinded and performed in a continuous spectrum of patients
    Articles excluded if:
    • more than 80% of patients had valvular heart disease


    29 studies were found. These were pooled following stratification for radiographic findings, criterion standard and clinical setting.
    Studies were found to be heterogeneous.

    The evidence

    • increased preload
      • cardiomegaly: cardiothoracic ratio > 0.5 (elective and peri-infarction) LR+ 4, LR- 0.7
      • cardiomegaly (pre-transplant) LR+ 1, LR- 1
      • redistribution/congestion: upper lobe veins > lower lobe veins LR+ 2.0, LR- 0.52
      • interstitial edema: hilar haze + other findings LR+ 2, LR- 0.7
    • decreased ejection fraction
      • cardiomegaly LR+ 2.4, LR- 0.62
      • redistribution/ congestion LR+ 3, LR- 0.6
      • interstitial edema (peri-infarction) LR+ 4, LR- 0.9

    Comments

    1. Prevalence of left ventricular preload was assessed in different groups of patients
      • general population: 5%
      • peri-infarction: 24% to 75%
      • dyspnoea: 26% to 31%
      • known severe systolic dysfunction 62% to 86%
    2. Prevalence of decreased ejection fraction was assessed in various groups of patients
      • general population: less than 5%
      • peri-infarction: 12% to 50%
    3. Data was incompletely reported - likelihood ratios have been extracted from a chart
    4. Interobserver agreement for individual signs was moderate
    5. Some clinicians have a fundamental disagreement with the equating of reduced ejection fraction / increased pre-load with the diagnostic category of 'heart failure' and prefer to think of the latter as a clinical syndrome.

    Citation

    1. Badgett RG, Mulrow CD, Otto PM, et al: How well can the chest radiograph diagnose left ventricular dysfunction?. Journal of General Internal Medicine 1996; 11: 625-634
    Contributor: Chris Ball and Clare Wotton, October 1999
    Reviewer: Edward Havranek

    Clinical Question.
    Patient suspected heat failure
    Intervention or Exposure CXR
    Outcome left ventricular dysfunction