Methods: We utilized 3 years of respiratory specimen data, seasons 2011 – 2014. Inclusion criteria included quasi-complete surveillance questionnaire (based on CD questions) originating from the continental United States and tested using molecular RT-PCR and viral culture. The sensitivity and specificity of each CD were calculated.
Results: Between October 2, 2011, and September 27, 2014, 10,662 respiratory specimens were submitted for testing. Of these, 7,931 met the inclusion criteria and were tested. Overall, 2,608 specimens were positive for influenza (33%) via PCR and/or culture. Almost all symptom onset was reported to be within 10 days or less from collection date (86%) with 75% of those being collected within 1-3 days of illness onset. The most common reported symptom from patients was a cough (81%). Fever (≥100.5°F) was reported in 65% of subjects with all measures of central tendency at 102°F.
The CD for the DoD influenza surveillance program had a sensitivity of 54% and specificity of 63%. When the DoD CD was modified to include fever that was recorded at the clinic only, the sensitivity of the CD dropped to 32% and the specificity increased to 81%. The sensitivity and specificity of the WHO CD using DoD data were 39% and 75%, respectively. The ranges of sensitivity and specificity for the other CDs analyzed were 43-96% and 22-69%.
Conclusion: Sensitive CDs capture a larger proportion of all cases, but this is at the cost of testing more specimens. Definitions with higher specificity result in fewer false negatives, but may miss more cases. CDs should be selected according to the objectives of the surveillance system and resources available.
J. Voss, None