Pneumonia or Atelectasis? Here's a trick to tell them apart
“New York is cold, but I like where I'm living … There's music on Clinton Street all through the evening.”
It’s the end of December; we collectively reflect on the year that was and try to find our footing for the next. In the short winter days of 2016, it is unexceptional to find oneself atelectatic – collapsed, blundered, but easily recruited perhaps with some sun, a trip to California. Yet, distinguishing simple existential atelectasis from a socked-in, ruddy and coarse consolidation can be fraught with subtlety. For this knotty differential, a drive through the soaring expanses of Big Sur can cool one’s innards providing both diagnosis and therapy. It is opportune, therefore, that the radiographic differentiation of atelectasis from consolidation became more objective in 2016.
As we are often left with a CT angiogram for pulmonary embolus [PE] to explain a patient’s non-thromboembolic respiratory presentation, diagnostic assistance when contemplating the commonly noted ‘atelectasis versus consolidation’ provides long-desired cognitive relief. It is oft taught that, on a contrast-enhanced chest CT scan, atelectasis glows brighter than consolidation. In a terrific little study, Edwards and colleagues have put numbers to this teaching.
The authors analyzed about one year’s worth of CT angiograms and included patients with area of lung which appeared consolidated or atelectatic. These lung regions had to be a minimum of 1 cm by 1 cm for at least 4 slices on the CT scan [1.25 mm cuts]. Very important exclusion criteria included: patients with malignancy, PE, pulmonary contusion/chest trauma, and CT not performed using 120 kilovolt (peak) imaging protocol.
The gold standard for the diagnosis of pneumonia included a clinical scoring system which assigned one point for each of the following: fever and/or cough, leukocytosis, antibiotic treatment for pneumonia, and documentation of pneumonia as discharge diagnosis. If 2 or more were met, the patient was considered to have pneumonia, if only 1 or less, then the patient was considered to have atelectasis. Patients clinically, and retrospectively parsed into having either atelectasis or consolidation then had their radiographic opacities compared in terms of Hounsfield Units [HU].
152 CT angiograms for PE [CTPA] were evaluated and 97 excluded for the following reasons: lack of non-aerated regions or non-aerated regions too small, PE, pulmonary contusion, respiratory motion artifact. 55 studies were included, 30 of which were deemed to have atelectasis and 25 to have pneumonia. In their protocol, the average HU attenuation of the ascending aorta was 221 in those with pneumonia and 258 in those with atelectasis. The HU attenuation of the main pulmonary artery was 283 and 298, respectively.
In those with atelectasis by the aforementioned, retrospective, clinical criteria, the average attenuation was 119 HU in the non-aerated lung regions of interest. In comparison, the average attenuation of non-aerated lung regions in those diagnosed with pneumonia was 62. The difference [119 versus 62] was statistically significant and confirmed the common teaching that atelectasis is brighter than consolidation.
The authors then went on to find helpful diagnostic cutoff values and found that < 92 HU predicted a 97% sensitivity and 85% specificity for pneumonia; that is, a 3% false negative rate and a 15% false positive rate for the diagnosis of pneumonia. The ‘optimal’ HU value for the diagnosis of pneumonia was < 85 HU which predicted a 90% sensitivity [i.e. 10% false negative rate] and 92% specificity [i.e. 8% false positive rate].
Studies like this make me smile, because they are simple, intuitive and clinically meaningful. I file them away with other studies that I wish that I had conceived. There are certainly reasons to pause here, however. Notably, there is some problem with their clinical prediction model for pneumonia in that the criteria are mechanistically linked. For example, ‘pneumonia as a discharge diagnosis’ is worth one point but it is clinically-uncommon having pneumonia as a discharge diagnosis without receiving at least a dose of antibiotics [it is important to highlight here that the authors make no distinction between bacterial and viral pneumonia or, conversely, causes of high attenuation that are not atelectasis]. In totality, however, their scoring system for pneumonia follows what most clinicians use inherently, so there is external validity here. It also falls in line with other clinical prediction scores utilized for pneumonia.
The authors astutely bring to our attention that the timing of contrast and attenuation of the ascending aorta and main pulmonary artery should be similar when comparing this study to others or to one’s patient. Further, changing the peak kilovolt setting will alter the HU threshold between patients.
I also appreciate how the authors predict clinical settings in which using different HU diagnostic cutoffs may be more pertinent. For example, they mention that when the diagnosis of pneumonia cannot be missed, the higher HU  should be utilized – such patients may be those in the ED who are to be discharged without follow up, or those who are immunocompromised. In patients where clinical follow up is close, or cross-sectional in time [e.g. those being admitted], the lower HU threshold could be utilized.
While these cutoffs are useful, and putatively codify and objectify experience and/or anecdotal approaches to differentiating consolidation from atelectasis on the chest CT, like every other diagnostic test, these cutoffs must be nested within clinical context. I am imaging a patient with an imperceptible airway narrowing causing lower lobe atelectasis and post-obstructive pneumonia. Here, there may be very high HU from a predominant atelectatic event, yet with abrupt onset fevers, chills, productive sputum and leukocytosis, the patient’s radiographic findings mustn’t be held as sacrosanct.