Imaging for the Management of Community-Acquired Pneumonia: What to Do if the Chest Radiograph Is Clear

Michael Niederman

CHEST 2018; 153: 583-585

The diagnosis of community-acquired pneumonia (CAP) is made when a patient has symptoms of a respiratory infection, a compatible physical examination, and a new lung infiltrate on chest radiograph. We have known for some time that some patients with respiratory infection symptoms and a negative chest radiograph can have infiltrates on chest CT scan.¹ However, the clinical significance of these CT findings is not known, nor is it clear if these patients need antibiotic therapy. Virtually all clinical trials of antibiotic therapy for pneumonia require an abnormal chest radiograph, and in the absence of a lung infiltrate, a respiratory infection is usually diagnosed as bronchitis.

In 1998, a Scandinavian study evaluated 47 patients (including 28 outpatients) with symptoms of respiratory infection who had both a chest radiograph and a CT scan.¹ Eighteen patients had pneumonia by both imaging modalities, but an additional eight patients had a lung infiltrate on CT scan only. Bilateral pneumonia was present in only six patients using chest radiograph and in 16 by chest CT scan. In general, the CT scan detected bronchopneumonia more often, whereas lobar pneumonia was more common on chest radiograph. A much larger study in 12 US EDs enrolled 3,423 patients who had both a chest radiograph and CT scan (often done to evaluate for possible pulmonary embolus).² Pulmonary infiltrates were present in 309 patients on chest radiograph, but in only 191 patients by CT scan. The positive predictive value for a chest radiographic opacity to be seen on CT scan was only 26.9%, whereas the sensitivity that the chest radiograph would show an infiltrate seen on CT scan was 43.5%. Of the 309 patients with abnormal chest radiograph, 226 had a chest CT scan without opacities, and of the 3,114 patients without an infiltrate on chest radiograph, 108 had an infiltrate on CT scan (3.5%). Similar findings were reported in a French study of 319 ED patients.³ Chest radiograph showed a parenchymal infiltrate in 188 patients, but in 56 of these patients there was no infiltrate on CT scan, whereas in 40 patients, CT scan found an infiltrate not present on chest radiograph. The CT scan findings were used to start antibiotics in 51 patients, and to discontinue them in 29 patients.

In the current issue of CHEST, Upchurch et al⁴ examined the clinical and microbiologic findings of patients with CAP, including those with pneumonia by CT scan criteria only, and with a negative chest radiograph. All the patients were adults who required admission for CAP, including 2,185 patients with radiographic pneumonia and 66 additional patients with CT scan-only pneumonia. Antibiotic therapy was given to all 2,251 patients who had an infiltrate on either chest radiograph or CT scan. Those with a variety of forms of immune suppression or recent hospitalization were excluded, and all imaging was done within 48 h of admission. The decision to do a CT scan was at the discretion of the treating physician, and a total of 748 patients had a CT scan. In 652 patients radiographic pneumonia was confirmed, in 30 patients with radiographic pneumonia the CT scan showed no pneumonia, and in 66 patients an infiltrate was present on CT scan but absent on chest radiograph. Among those 66 patients, 65% had both studies on the same day, and 35% had the CT scan on the next day.

Patients with CT scan-only pneumonia were younger than those with radiographic pneumonia, and were more likely to have chest pain, a lower pneumonia severity index risk class, a greater likelihood of obesity, and a lower serum procalcitonin. Although both groups received the same types of antibiotics, the CT scan-only pneumonia group received them less often in the first 6 h. The distribution of pathogens was similar in all patients, with viruses being the most common pathogens in both groups, as in the Etiology of Pneumonia in the Community study overall.⁵ Clinical outcomes including length of stay, ICU admission, mechanical ventilation, and mortality were similar in both groups. Based on these findings, the authors concluded that patients with CT scan-only pneumonia should be treated with antibiotics, using the same approach as for those with radiographic pneumonia.

However, a number of key questions remain. First, should we get a CT scan for all patients with respiratory infection symptoms and a negative chest radiograph, or should we simply treat these patients for pneumonia based on clinical suspicion alone? This may be especially important when patients are severely ill, or are elderly and frail, a clinical condition that is known to lead to false-negative chest radiographs. In a study of 169 multimorbid elderly patients, 96 were judged to have pneumonia, but the chest radiograph was negative in 51 patients.⁶Another implication of these findings is that certain patients at high risk for pneumonia, such as those with COPD who are taking inhaled corticosteroids, may indeed have pneumonia if properly evaluated, and not just a bronchitic exacerbation, even with a negative chest radiograph. The finding of lower procalcitonin in those with CT scan-only pneumonia vs radiographic pneumonia raises the question about whether CT scan is detecting pneumonia at an earlier stage than chest radiograph, or if it is finding viral pneumonia. However, antibiotics are probably still needed because 12% of patients with CT scan-only pneumonia did have a documented bacterial infection. On the other hand, patients with a negative CT scan and a positive chest radiograph probably do not need antibiotics because only one of 30 patients had a bacterial pathogen identified.

I do not think that all patients with suspected pneumonia should have a chest CT scan if the radiograph is negative; however, as the Etiology of Pneumonia in the Community study showed, nearly one-third of all patients with pneumonia will have a CT scan anyway, for a number of reasons. A more judicious approach might be wider use of lung ultrasound. In a study of 96 elderly patients with pneumonia, chest radiograph was positive in only 45 patients, whereas the lung ultrasound was positive in 88 patients.⁶ In a meta-analysis comparing ultrasound with chest radiograph, the ultrasound had a higher sensitivity and specificity, using either the discharge diagnosis or CT findings as the gold standard for diagnosis.⁷ In another meta-analysis, ultrasound had a high sensitivity and specificity for the diagnosis of pneumonia, probably better than chest radiograph.⁸