RESUMO
STUDY OBJECTIVE: While development is under way of accurate, point-of-care molecular tests for influenza infection, the optimal specimen type for molecular tests remains unclear. Compared with standard nasopharyngeal swab specimens, less invasive nasal swab and midturbinate swab specimens may cause less patient discomfort and be more suitable for routine emergency department (ED) testing, although possibly at the expense of diagnostic accuracy. We compare both the accuracy of a polymerase chain reaction molecular influenza test and discomfort between these 3 intranasal specimen types. METHODS: A convenience sample of adult and pediatric patients with influenza-like illness and presenting to 2 Northern California EDs and 2 EDs in Santiago, Chile, was prospectively enrolled during the 2015 to 2016 influenza season. Research nurses collected nasopharyngeal swab, midturbinate swab, and nasal swab specimens from each subject and assessed discomfort on a validated 6-point scale. Specimens were tested for influenza A and B by real-time polymerase chain reaction at reference laboratories. Outcome measures were comparison of test performance between nasal swab and midturbinate swab, when compared with a reference standard nasopharyngeal swab; and comparison of discomfort between all 3 specimen types. RESULTS: Four hundred eighty-four subjects were enrolled, and all 3 swabs were obtained for each subject; 14% were children. The prevalence of influenza (A or B) was 30.0% (95% confidence interval [CI] 26.0% to 34.8%). The sensitivity for detecting influenza was 98% (95% CI 94.25% to 99.65%) with the midturbinate swab versus 84.4% (95% CI 77.5% to 89.8%) with the nasal swab, difference 13.6% (95% CI 8.2% to 19.3%). Specificity was 98.5% (95% CI 96.6% to 99.5%) with the midturbinate swab versus 99.1% (95% CI 97.4% to 99.8%) with the nasal swab, difference -0.6% (95% CI -1.8% to 0.6%). Swab discomfort levels correlated with the depth of the swab type. Median discomfort scores for the nasal swab, midturbinate swab, and nasopharyngeal swab were 0, 1, and 3, respectively; the median differences were nasopharyngeal swab-midturbinate swab 2 (95% CI 1 to 2), nasopharyngeal swab-nasal swab 3 (95% CI 2 to 3), and midturbinate swab-nasal swab 1 (95% CI 1 to 2). CONCLUSION: Compared with the reference standard nasopharyngeal swab specimen, midturbinate swab specimens provided a significantly more comfortable sampling experience, with only a small sacrifice in sensitivity for influenza detection. Nasal swab specimens were significantly less sensitive than midturbinate swab. Our results suggest the midturbinate swab is the sampling method of choice for molecular influenza testing in ED patients.