Detection of respiratory viruses in sputum from adults by use of automated multiplex PCR.

Respiratory tract infections (RTI) frequently cause hospital admissions among adults. Diagnostic viral reverse transcriptase PCR (RT-PCR) of nose and throat swabs (NTS) is useful for patient care by informing antiviral use and appropriate isolation. However, automated RT-PCR systems are not amenable to utilizing sputum due to its viscosity. We evaluated a simple method of processing sputum samples in a fully automated respiratory viral panel RT-PCR assay (FilmArray). Archived sputum and NTS samples collected in 2008-2012 from hospitalized adults with RTI were evaluated. A subset of sputum samples positive for 10 common viruses by a uniplex RT-PCR was selected. A sterile cotton-tip swab was dunked in sputum, swirled in 700 μL of sterile water (dunk and swirl method) and tested by the FilmArray assay. Quantitative RT-PCR was performed on "dunked" sputum and NTS samples for influenza A (Flu A), respiratory syncytial virus (RSV), coronavirus OC43 (OC43), and human metapneumovirus (HMPV). Viruses were identified in 31% of 965 illnesses using a uniplex RT-PCR. The sputum sample was the only sample positive for 105 subjects, including 35% (22/64) of influenza cases and significantly increased the diagnostic yield of NTS alone (302/965 [31%] versus 197/965 [20%]; P = 0.0001). Of 108 sputum samples evaluated by the FilmArray assay using the dunk and swirl method, 99 (92%) were positive. Quantitative RT-PCR revealed higher mean viral loads in dunked sputum samples compared to NTS samples for Flu A, RSV, and HMPV (P = 0.0001, P = 0.006, and P = 0.011, respectively). The dunk and swirl method is a simple and practical method for reliably processing sputum samples in a fully automated PCR system. The higher viral loads in sputa may increase detection over NTS testing alone.