Predictors and Outcomes of Hospitalization for Influenza: Real-World Evidence from the United States Medicare Population

Abstract

Introduction: The purpose of this study was to identify predictors of initial hospitalization and describe the outcomes of high-risk patients hospitalized with influenza. Methods: Data were taken from the 5% national US Medicare database from 2012 to 2015. Patients (aged at least 13 years) were required to have at least one diagnosis for influenza and have continuous health plan enrollment for 6 months before (baseline) and 3 months (follow-up) after the date of influenza diagnosis. Patients who died during follow-up were included. Patients were categorized as initially hospitalized if hospitalized within 0-1 day of diagnosis. High-risk initially hospitalized patients were defined as patients aged at least 65 years or those that had a diagnostic code for chronic lung disease, cardiovascular or cerebrovascular disease, or weakened immune system during baseline period. Logistic regression models were developed to determine predictors of initial hospitalization. Results: The study population included 8127 high-risk patients who were initially hospitalized and 16,784 who were not hospitalized. Among high-risk patients, 89.3% were diagnosed in the emergency room, whereas 7.5% and 3.2% were diagnosed in a physician's office or other Medicare settings, respectively. Chronic obstructive pulmonary disorder, congestive heart failure, chronic kidney disease, older age, being male, other comorbidities, number of comorbidities, and baseline healthcare resource use were the predictors of hospitalization. Median length of stay for the hospitalization was 5.0 days, and the 30-day readmission rate was 14%. All-cause mortality rate was 5.1% during the inpatient stay and 9.2% within 30 days of diagnosis. Hospitalized patients with influenza incurred an increase of $16,568 per patient in total all-cause healthcare costs from pre-influenza to post-influenza diagnosis. Conclusion: The study characterized the burden of hospitalization for influenza and found that hospitalized high-risk patients experience greater comorbidity burden, higher likelihood of multiple inpatient admissions, and costly Electronic supplementary material The online version of this article (

Using national electronic health records for pandemic preparedness: validation of a parsimonious model for predicting excess deaths among those with COVID-19–a data-driven retrospective cohort study

“…4,5 All-cause and disease-specific mortality prediction in research and clinical practice has included underlying conditions or "baseline mortality risk", often derived and validated using EHR. [6][7][8] Underlying non-communicable diseases (NCDs) are important mortality predictors in infectious diseases [9][10] , but baseline mortality risk based on NCDs is largely neglected in pandemic preparedness, which emphasises infection transmissibility and severity, using metrics such as case fatality ratio, infection fatality ratio and reproduction number. [11][12][13][14] Although COVID-19 is increasingly viewed as a "syndemic" 15 (with interaction between infectious diseases and NCDs, requiring cross-speciality expertise), efforts to predict excess mortality have focused on dynamic transmission modelling without consideration of baseline risk or use of anonymised, individual-level, population-scale EHR 16,17 .…”