Predictors of Clinical Respiratory Virus Testing Among Adults Hospitalized with Acute Respiratory Illness (ARI) (2015–2016)

Abstract

BackgroundVaccine effectiveness (VE) studies require an accurate indicator of influenza infection, often obtained through research testing independent of clinical (physician-ordered) testing. Clinical testing could be used to detect influenza in these studies if factors associated with clinical testing for influenza were better understood.MethodsAdults hospitalized with ARI at three study sites during the study period were enrolled in CDC’s 2015–16 Hospitalized Adult Influenza Vaccine Effectiveness Network (HAIVEN) study and tested for influenza by RT-PCR. Clinical testing information, presenting symptoms, and patient characteristics were collected from medical records and patient interview. Logistic regression was used to estimate odds of receiving a clinical test based on age, vaccination status, comorbidities, presentation with influenza-like illness (ILI; defined as fever and cough or sore throat) and other factors.ResultsOf 895 enrollees, 571 (64%) patients meeting study inclusion criteria received physician-ordered testing. Of these, 53% had a multipathogen panel, 13% had a rapid antigen test, 7% had singleplex PCR, <1% had viral culture, and 27% had multiple tests; influenza infection was detected in 55 (6%) patients. Of 150 influenza cases identified by study testing, 25 (17%) were not tested clinically. Enrollees who did not receive clinical testing were older, had longer time to admission, and were less likely to present with ILI. Immunosuppressive disorders (aOR=2.05), non-COPD lung conditions (aOR=1.68), presentation with ILI (aOR=4.03), and admission ≤2 days from symptom onset (aOR=1.89) were positively associated with receiving a clinical test (P < 0.01 for all; Figure 1). After adjusting for these factors, enrollees with influenza vaccination were 37% less likely (aOR=0.63) to receive a clinical test (P < 0.01).ConclusionPatients with ARI who were clinically tested for influenza differed from those not tested. A lower likelihood of testing among influenza positive vaccinees could potentially bias VE estimates upward and requires further evaluation. Clinical testing alone may fail to detect a substantial proportion of influenza cases.Disclosures All authors: No reported disclosures.