| Background. Diagnosis of the etiologic agent of respiratory viral infection traditionally relies on culture or antigen detection. A novel platform, the mass spectrometry followed RT-PCR (RT-PCR/ESI-MS), which has capability to rapidly identify and quantify multiple pathogens simultaneously, were evaluated in this study. Disposition decision making for patients with influenza can be problematic in emergency departments (ED). Influenza Viral load (VL), which has been shown to be important for many pathogens, has been proposed as a predictor of symptom severity, inpatient length of stay, and clinical course, and could be utilized in ED. A clinical prediction rule (CPR), in order to aid clinicians to prediction VL in overcrowded ED, was derived in this study. Another CPR to help clinicians to stratify the need for hospitalization for patients with influenza, incorporating influenza VL and other clinical information, was derived as well.;Method. The RT-PCR/ESI-MS respiratory virus surveillance kit was designed to detect seven including influenza. Excess frozen archived nasopharyngeal aspirates (NPA) from patients visited ED during 2007-8 respiratory season (N=192) were analyzed by RT-PCR/ESI-MS then compared to clinical virology results to assess "qualitative" performance. Dilution series of Influenza A (H3N2) were spiked into different media (viral transport media and negative NPA) to assess the "quantitative" performance (accuracy and reliability) of RT-PCR/ESI-MS platform. Another retrospective cohort of patients with laboratory confirmed influenza from EDs (N=117) in 2007-2009 was consented to derive CPRs by recursive partitioning algorithm with influenza VL obtained by PCR/ESI-MS.;Results. RT-PCR/ESI-MS has moderately well performance of qualification (sensitivity: 89.1%, specificity: 80.3%) and quantification (range of reliable quantification: 3.2x105--6.4x10 5 genome copies/mL, R2: 0.88-0.99, rho: 0.87-0.95). CPR for VL prediction contains age with six different cutoff points, vaccination history, and absolute neutrophil count (c-statistics: 0.85, sensitivity: 90%, specificity: 83%). CPR for hospitalization prediction contains underlying illness, age, influenza VL level, and vaccination history (c-statistics: 0.84, sensitivity: 76%, specificity: 83%).;Conclusion. RT-PCR/ESI-MS demonstrated capacity to rapidly and multiply detects respiratory viruses with good performance of qualification and quantification. CPR for VL and hospitalization were derived and merits further validation for use in ED disposition decision-making for patients with influenza. |
