An elderly patient presents with cough and shortness of breath. The resident says the lung examination is normal and the chest x-ray is negative and wants to send the patient home. But you hear crackles posteriorly at the right base. And the oxygen saturation is 94%. And the patient looks ill. What do you do?


If your practice is to use chest radiograph as the gold standard for pneumonia (following IDSA guidelines – Mandell 2007) then you might be missing pneumonia. In medicine we tend (alas) to assume our reference standards are disease-defining until we compare them against something else.


Against CT scan, chest radiograph is probably about 75% sensitive and specific for pneumonia. Often the literature does not give exact numbers but false negatives and false positives are extensively documented. Where sensitivity is reported, it can be as low as 44%.(Self Am J Emergency Medicine 2013 – from a datebase of PE workups) A more representative finding for sensitivity is 77% (Ye PLoS One 2015). Good, but not perfect. Other studies show significant false positive rates (Clasessens Am J Resp Crit Care Med 2015). Specificity is probably also about 75% but we do not yet have a precise estimate of that. And of course there are significant differences between a supine 1 view chest radiograph in an altered patient and a 2 view standing radiograph with tidal volume inspiration.


Your overall accuracy is not doomed by limited tests. Your accuracy depends on how well you formulate Bayesian “prior probability” BEFORE applying results of tests. Of course, if you don’t want to do that there is always pulmonary ultrasound, with 95% sensitivity (Ye PLoS One 2015). But let’s talk bedside medicine.


The classical clinical signs of pneumonia are:




-asymmetric breath sounds


-progression of symptoms after 3-5 days


Steven McGee’s Evidence-Based Physical Diagnosis has an excellent chapter on this topic. It essentially supports the classical findings but reminds us of their limitations too.


Back to the case. You have a moderate clinical suspicion and you are aware that the gold standard is flawed. Do you order an ultrasound? CT? Do you start empiric antibiotics? Do you recommend next day follow-up? In this case you desire to admit the patient because she appears ill. But the utilization case manager and hospitalist argue that the admission does not meet Interqual criteria. They say it is impossible for a patient to have pneumonia without a positive chest x-ray and besides, the patient does not even have a fever (oral temperature).


You palpate the temperature at the neck with the back of your hand and leave it there for 5 seconds, and feel a tingle – this is a fever. The RN confirms the rectal temperature is 103. You perform a bedside ultrasound and find mild hepatization and a few air bronchograms in the area where you heard crackles. They do not accept the legitimacy of that test, despite the literature.


You order a CT scan and the radiologist confirms a moderate size consolidation, interpreted as pneumonia, at the exact area where you heard crackles hours before.



-Chest x-ray has limitations in sensitivity and specificity for pneumonia (probably about 75% for each)

-If you do a thorough physical examination with attention to heart rate, oxygenation, and auscultation then the CXR is probably adequate for most patients.

-When you need more information, consider pulmonary ultrasound, or CT scan of the chest.