Methods: Retrospective analysis of clinical samples obtained from the NCH microbiology laboratory from January 2011 - December 2014 and sent for BR-PCR to the University of Washington Molecular Diagnostics Lab. Medical record review collected data on patient characteristics, clinical scenario and laboratory results, and a determination of clinical value was assigned.
Results: 256 samples from 157 patients were identified of which 45% (N=71) were ICP and 77% (N=196) of samples were from patients who had received antibiotics prior to sample acquisition. BR-PCR detected a causative organism in 39 samples and was the sole positive diagnostic result for 20 samples (8%). BR-PCR failed to detect the causative agent in 27 (10%) of samples that were confirmed by other diagnostic methods (N = 24 by culture, N= 5 by serology, N=1 by pathology, N=2 by culture/pathology). Yield was lowest among acid-fast (AFB) organism infections diagnosed in our cohort; BR-PCR did not detect AFB DNA in all 5 cases of microbiologically proven infection. Among samples, bronchoalveolar lavage (BAL) was the type most commonly sent (N=77); however, no organisms were detected by BR-PCR that were not already confirmed via a conventional method. BR-PCR detected an organism from tissue samples more than from BAL or other fluids (p= 0.05, Chi square).
Conclusion: In our cohort, BR-PCR was helpful as an adjunctive diagnostic in identifying infectious organisms in patients previously exposed to antibiotics, but lacked sensitivity for detection in BAL samples and AFB infections. Additional data is needed to better define the optimal clinical scenarios and specimen type in which it can provide the highest diagnostic yield.
M. I. Ardura, None