Laennec’s stethoscope has survived more than 200 years, much longer than some of his contemporaries predicted. But will it survive the challenge of bedside ultrasonography and other technologic advances?
The physical examination, with its roots extending at least as far back as Hippocrates, may be at a crossroads as the mainstay of diagnosis. Physical signs can be subjective and lack sensitivity and specificity. Modern imaging and laboratory studies may already be more trusted.
If the physical examination is to survive, it must be accurate, reproducible, and efficient. Needed is a simple, evidence-based approach to the physical examination that enhances its diagnostic accuracy while maintaining bedside efficiency.
Here, we analyze the accuracy of the physical signs that are most effective in the clinical diagnosis of 4 common cardiopulmonary conditions that often present with dyspnea: pneumonia, pleural effusion, chronic obstructive pulmonary disease (COPD), and congestive heart failure.
To grasp the significance of physical findings, it is necessary to understand the concept of likelihood ratios, which are widely accepted measures of the accuracy of a test or clinical finding.1,2 The positive likelihood ratio is the probability of a disease being present when the test is positive or the clinical finding is present, while the negative likelihood ratio is the probability that the disease is present when the test is negative or the clinical finding is absent. They are calculated as follows1:
Positive likelihood ratio = sensitivity / (1 – specificity)
Negative likelihood ratio = (1 – sensitivity) / specificity
Table 1 shows how the likelihood ratio of a test changes the posttest probability that a condition is present or absent, according to an analysis by McGee.2
Pneumonia is a common disease, with more than 2 million cases annually in the United States. It is most often diagnosed by standard chest radiography, although computed tomography can identify it earlier and with higher sensitivity and specificity.5 The amount of published data on physical examination findings in pneumonia is surprisingly small.
Asymmetry in chest expansion: Specific, reproducible, but not sensitive
The physical finding with the highest positive likelihood ratio for diagnosing pneumonia is asymmetry in chest expansion.6,7
In a 1984 study of 1,819 patients presenting to an emergency department with acute cough, Diehr et al6 evaluated several physical signs of pneumonia. Asymmetric chest expansion had a specificity and positive predictive value of 100%, but its sensitivity was only 4.3%. Thus, it is not a good screening test, but it is a good diagnostic or confirmatory test. From these numbers, Metlay et al8 calculated that the positive likelihood ratio was infinity and the negative likelihood ratio was 0.96.
McGee,7 on the other hand, calculated the positive likelihood ratio of asymmetric chest expansion at 44.1. McGee also found chest expansion to be a highly reproducible finding, with an interobserver agreement kappa score of 0.85.7 (A kappa score of 1.0 would indicate perfect interobserver agreement.) Interestingly, chest radiographs interpreted for pulmonary infiltrates have an interobserver kappa score of only 0.38.7 Further studies of this physical sign could shed more light upon this area of uncertainty.
Other signs of pneumonia
None of the other physical signs studied for the diagnosis of pneumonia has as high a positive likelihood ratio as asymmetric chest expansion.6–12
Egophony is a high-pitched or nasal quality of the patient’s voice heard on auscultation over lung tissue that is consolidated or fibrosed, due to enhanced transmission of high-frequency sound across fluid. It is often described as the “E-to-A change.” Although listening for egophony is widely done and easy to do, we calculate that this sign has a positive likelihood ratio of only 6.8 based on pooled data from 3 trials with a total of 3,245 patients.6,10,11
Faring less favorably, in descending order of diagnostic accuracy, are:
Percussion dullness (positive likelihood ratio 5.7 based on 4 studies with 3,653 patients)6,10–12
Bronchophony or bronchial breath sounds (positive likelihood ratio 3.3 based on 1,118 patients)10
Crackles have long been taught as a common physical finding in pneumonia. Bohadana et al pointed out that “crackle” can be defined acoustically but does not suggest any means or site of generation.4 Pooled data from 4 studies in 3,647 patients6,10–12 result in a positive likelihood ratio for crackles in the diagnosis of pneumonia of only 3.2.
Diminished breath sounds (positive likelihood ratio 2.5 based on 3 studies with 1,828 patients).10–12