Underdiagnosis of Influenza Virus Infection in Hospitalized Older Adults.

Abstract

To describe factors associated with provider-ordered influenza testing in hospitalized older adults. Information on participant demographics, symptoms, and provider-ordered influenza testing were collected by questionnaire and chart review. We conducted prospective laboratory-based surveillance using reverse-transcriptase polymerase chain reaction (RT-PCR), the criterion standard for diagnosis of influenza, to determine how participant characteristics and provider-ordered testing affected accurate influenza diagnosis. One academic and three community hospitals in Davidson County, Tennessee. Adults aged 18 and older with acute respiratory illness or nonlocalizing fever (N=1,422). We compared characteristics of participants with and without provider-ordered testing for influenza using the Wilcoxon test and Pearson chi-square test. Multivariable logistic regression models were used to identify factors predictive of provider-ordered influenza testing. Twenty-eight percent (399/1,422) of participants had provider-ordered influenza testing. Participants who were tested were younger than those not tested (58 ± 18 vs 66 ± 15, p<.001) and more likely to have influenza-like illness (ILI) (71% vs 49%, p<.001). ILI decreased with increasing age (aged 18-49, 63%; aged 50-64, 60%; aged ≥65, 48%). ILI and younger age were independent predictors of provider-ordered testing. Of the 136 participants with influenza confirmed using RT-PCR, ILI was the only significant predictor of provider-ordered testing (adjusted odds ratio=3.43, 95% confidence interval=1.22-9.70). Adults aged 65 and older hospitalized with fever or respiratory symptoms during influenza season are less likely to undergo a provider-ordered influenza test than younger adults. Some, but not all, of this disparity is due to a lower likelihood of ILI. Further strategies are needed to increase clinician awareness and testing in this vulnerable group.