Prevalence of Lab-confirmed Influenza and Outcomes by Admission Diagnosis in Hospitalized Adults

In the United States, from 2010-2013 influenza accounted for an estimated 114,192–624,435 hospitalizations, 18,491–95,390 ICU admissions, and 4,915–27,174 deaths per year. Clinical presentation of influenza is widely variable in hospitalized adults. For instance, a broad spectrum of diseases was described during the 2009 (H1N1) virus pandemic ranging from mild upper respiratory tract infections and febrile influenza-like illnesses, to progressive pneumonia and gastrointestinal symptoms such as nausea and vomiting2. Variability in presentation may be attributed to factors such as age, underlying comorbid conditions, respiratory bacterial superinfections, and immune status. The most common symptoms observed in patients with laboratory confirmed influenza include: cough, fever, sore throat, myalgias, and headache.

Overlap in these influenza symptoms and other clinical features, including syndromes, with different respiratory pathogens and progressive diseases has been well documented. In effect, with exacerbation of comorbidities, suspicion of influenza infection may be low and have a substantial impact on treatment, accuracy, and timing of diagnoses by clinicians. Public health rationale: Much research has been done describing the clinical symptoms of hospitalized patients with influenza infections. However, there is little work detailing epidemiology of influenza presentation by admission diagnosis and the HAIVEN acute respiratory screening criteria provides a unique opportunity to do so. An understanding of influenza in this manner is important in increasing awareness of the association between clinical diagnoses at presentation and their predictive value of influenza infections and associated outcomes. Moreover, a heightened awareness will aid infection prevention measures, earlier testing and detection of influenza, and may help guide treatment patterns.

Research question: What are the clinical diagnoses that influenza presents as in adult hospitalized patients and what is the difference in treatment patterns and outcomes? Primary objective(s):

• Determine the frequency of each diagnosis category and its proportion for influenza positivity and the frequency of fever by syndrome and age group.
• Determine the outcomes of each diagnosis category.

Data variables: All admission ICD codes within first 24 hours of admission, all enrollment variables, all clinical test results, antiviral variables, and outcome variables.

Methods and Analysis Plans

Methods: We will perform retrospective reviews on all HAIVEN enrollees during two seasons (2016-2018). Subjects will be classified into one of four categories defined as: Respiratory Infection Dx, Exacerbation of Chronic High-Risk Dx, Sepsis or Stroke Dx, Other Qualifying Symptoms (Appendix 1). These will be defined using most relevant ICD codes within the first 24 hours of admission. The purpose of the fourth category is to include all enrolled subjects who didn’t receive a diagnosis necessary for classification in one of the three categories. In order to determine frequency of syndrome by fever it will not be included in subject classification. A descriptive analysis will also be done to determine outcome of each diagnosis category and the delay between illness onset and admission. Outcome will be defined as, antiviral treatment, length of stay, ICU admission, length of ICU stay, chest radiography (and infiltrate detected), mechanical ventilation, discharged alive.

Analysis: The statistical analyses will be reported using summary tables. Continuous variables will be summarized with means, standard deviations (SD), standard errors (SE), medians, Q1, Q3, minimums, and maximums. Categorical variables, such as clinical diagnosessymptoms (ICD 10 code), will be summarized by counts and by percentages of subjects in corresponding categories.

No Imputation on Missing Data Will Be Made

The proportions of flu positive, fever, the treatment of antiviral and antibiotic will be tabulated by different diagnosissymptom groups. The length of stay, ICU admission will be explored. The proportion of positive flu, fever, treatment of antiviral drug, treatment of antibiotic drug varies across different symptom groups will be detected using Mann-Whitney test. The odds ratio of influenza between different symptom groups will be compared using multivariate logistic regression, adjusted for age groups, race, gender, health status, vaccination status, and time from disease onset to enrollment. Statistical tests will be 2-sided at the alpha=0.05 significance level and 2-sided 95% confidence interval will be used. All p-values will be presented as nominal p-values. No adjustment on multiplicity will be made. All statistical analysis will be performed in SAS 9.4 (SAS Institute Inc., Cary, NC).

Deliverable(s): Site-specific analysis and manuscript IX. Timeline (Please indicate number of seasons’ data needed for analysis/completion): Two seasons (2016-2018) of site-specific data will be analyzed.