The Potential of Molecular Diagnostics and Serum Procalcitonin Levels to Change the Antibiotic Management of Community-Acquired Pneumonia

Background: Traditional methods identify an etiology in fewer than 50% of patients with community acquired pneumonia (CAP).  We assessed the influence of two diagnostic bundles on empiric antibiotic therapy in CAP patients admitted to Providence Portland Medical Center (PPMC).   

Methods: Diagnostic “bundles” were used in all CAP patients admitted to the hospital January – March 2015.  Core tests were performed in all patients: blood and sputum cultures; urine antigens for S. pnemoniae and L. pneumophila, serogroup 1; nasopharyngeal (NP) polymerase chain reaction (PCR) swabs for S. aureus and S. pneumoniae; and serum procalcitonin (PCT) assay. Respiratory NP specimens were assessed by either a PPMC “standard” PCR platform, which probes for 5 viruses or the FDA-approved Biofire FilmArray multiplex platform that probes for 7 viruses and 3 bacteria.

 

Results: 210 patients were randomized; 127 were evaluable (68 standard, 59 FilmArray). Use of sputum, blood cultures, and urine antigens yielded a diagnosis in only 22% of patients.  Adding NP PCR for S. pneumoniae and S. aureus detected a potential pathogen in 64% of patients.  Viral PCR was positive in 49 patients; one half in combination with a bacterial pathogen.  A potential pathogen was detected in 90 of 127 patients (71%). The number of viruses detected was similar for the FilmArray and standard PCR tests; the mean turn-around-time of 2.1 hours vs 26.5 hours, respectively. Serum PCT levels were lower in those with pure viral infections vs those with bacterial or mixed viral/bacterial infections, p = 0.003, figure 1.  Of 25 patients identified with a viral pneumonia (viral pathogen, low PCT), empiric antibiotics were discontinued in only 8.  The median cost of therapy per 1000 patient days was lower in the FilmArray vs standard group ($3037 vs $7952, p = 0.02). The length of therapy (LOT) per 1000 patient days was significantly lower in viral compared to bacterial infections (p = 0.04).

Conclusion: The microbial etiology of CAP was identified in 71% of patients using a bundle of diagnostic tools. The combination of pathogen identification and serum PCT levels facilitates the transition from empiric to specific therapy of CAP.   

Fig 1. The PCT values in patients with CAP due to bacteria or a virus + bacteria compared to virus alone, p=0.003.