Early diagnosis of Community-Acquired Pneumonia in the Emergency Department: Imaging and Biomarkers

Background
This thesis addresses the early diagnosis of community-acquired pneumonia in emergency departments. The background is an increasing prevalence of antibiotic-resistant bacteria, which pose a significant global health threat. There is a need to rationalise antibiotic consumption in hospitals in order to support efforts to combat antibiotic resistance.

Community-acquired pneumonia is the most common infection requiring hospitalisation. However, correct identification of patients with pneumonia can be challenging because symptoms and findings are non-specific and can be atypical. Chest X-rays are the most commonly used diagnostic imaging tool to confirm the disease; however, chest X-rays have low sensitivity for pneumonia, and the imaging quality is poor in supine patients. More rational antibiotic consumption requires more accurate diagnostics.

Ultralow-dose CT of the thorax is a safe and more precise alternative to chest X-rays and has the potential to improve the radiological diagnosis of community-acquired pneumonia. However, the interpretation of CT scans requires expertise in radiology, which is not always available in the emergency department.

Blood markers for systemic inflammation are not considered central in pneumonia diagnostics but are often used as guidance. C-reactive protein (CRP) is most commonly used, but its utility is limited by low specificity and a delayed response time. No tissue-specific blood biomarkers to reflect acute lung injury are in use.

This thesis describes three clinical studies with the following aims:
1) To evaluate the diagnostic accuracy of emergency department physicians’ interpretations of ultralow-dose CTs of the chest for pneumonic changes.
2) To evaluate the diagnostic accuracy of the lung injury markers SP-D, KL-6, and CC16 in distinguishing between pneumonia and non-pneumonia in patients suspected of having community-acquired pneumonia.
3) To evaluate the diagnostic accuracy of the inflammatory biomarkers CRP, PCT, IL-6, suPAR, and YKL-40, and to develop a diagnostic model based on these biomarkers and symptom duration for patients suspected of having community-acquired pneumonia. 

Methods
All studies were designed as diagnostic cross-sectional studies. Patients with an initial clinical suspicion of community-acquired pneumonia were included in the emergency departments of three Danish hospitals. The participants underwent ultra-low-dose CT of the thorax and had additional blood samples taken to investigate the study biomarkers. 

Ten doctors from the emergency departments assessed 128 ultra-low-dose CT scans after completing a focused, interactive course on identifying changes compatible with pneumonia on ultralow-dose CT. The reference standard was a radiologist's assessment of the same scans.

The biomarker values were compared with the patients' final diagnoses, which were assigned by two medical experts through a medical chart review.

The primary outcome measures were diagnostic accuracy estimates in the form of sensitivity, specificity, positive/negative predictive values (PPV/NPV), and area under the curve (AUC). For the inflammatory markers, we investigated their combined diagnostic value by developing a classification tree. 

Results
We included 411 patients suspected of community-acquired pneumonia, which was the final diagnosis in 58% of the cases.

The physicians’ overall accuracy in assessing ultralow-dose CT for pneumonia was: sensitivity=83%, specificity=70%, PPV=80%, NPV=78%. 

The lung injury markers revealed no ability to distinguish between patients with and without pneumonia, with AUC’s ranging from 0.50 to 0.55.

None of the investigated inflammatory biomarkers were effective in diagnosing CAP. The diagnostic performance of the classification tree did not differ from the performance of IL-6 and CRP alone, and they were effective at differentiating patients without and without infection (AUC=0.86 and 0.88, respectively). 

Conclusions
The results indicate that emergency department physicians with limited training can assess ultralow-dose CT for pneumonic changes with good accuracy. None of the investigated biomarkers can be used to diagnose pneumonia with satisfactory accuracy. IL-6 and CRP have the potential to support the diagnosis of patients suspected of pneumonia by differentiating between infected and non-infected patients. Further studies are needed to assess the impact of ultralow-dose CT, CRP, and IL-6 on antibiotic prescriptions in emergency departments.