Prevailing Influenza Activity and Acute Aortic Dissection Seasonality

Abstract

Introduction: Seasonal patterns of acute Stanford type A dissection have been described, with events occurring most commonly in winter months. Attempts to correlate this with such external influences as temperature and barometric pressure have not been productive. Our group has previously described correlation between seasonal influenza activity and cardiac outcomes such as death, myocardial infarction and automated implantable defibrillator triggering. We sought to determine whether seasonal influenza activity could help explain the seasonal periodicity of acute aortic dissection onset. Methods: We reviewed hospital admissions for both acute type A and type B dissection in our center over a period of 15 years. We further retrieved weekly influenza activity for the contemporaneous time period from the US Centers for Disease Control (CDC) outpatient influenza-like illness surveillance program using estimates based on population weights. Data were analyzed by autoregressive internal moving average models (ARIMA) to assess seasonal variation and crosscorrelation, and by VARMAX regression to separate the effects of winter months from reported influenza activity Results: Between January 2001 and November 2016, 605 acute type A and 543 acute type B cases were admitted to our hospital. For type A cases, strong seasonal periodicity was identified over the study period. Highly significant crosscorrelation effects at lags of 6 (p< 0.008), 12, 18 and 24 months (all p< 0.001) were present between influenza and acute type A admissions across time averaged over all study years (Figure 1A). Additionally, individual years showed similar seasonal periodicity for influenza and type A admissions (standardized units, Figure 1B). In univariate VARMAX regression, strong non-lagged effects of winter season were observed, but this disappeared when one-month lagged influenza activity was added to the model, indicating the type A admissions were strongly influenced by influenza activity over the preceding four weeks. Type B dissection admissions did not demonstrate seasonal periodicity (p for moving average and autoregressive terms both p>0.45). Conclusion: Both influenza and acute type A dissection show marked seasonal variation. Type A admission rate is significantly associated with influenza activity beyond the effect of season alone. Direct measurement of flu antigen and vaccination status should be assessed in future studies to help establish mechanism.Figure 1bSmoothed-Spline Plots.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Disclosure: Nothing to disclose