LAUSR

Abstract Background Outbreaks of influenza can result in significant morbidity, including secondary bacterial infections. Invasive group A streptococcal (iGAS) infections are associated with a 12% case fatality rate. We used surveillance data to examine if there was a correlation between hospitalized influenza and GAS cases. Methods Minnesota Department of Health conducts population-based surveillance for hospitalized lab-confirmed influenza and iGAS (sterile site isolation) cases in the Minneapolis–St. Paul area as part of the CDC Emerging Infections Program. Cases were categorized by week during October–April of each year for 2010–2016, based on specimen collection date. Using STATA (v15), the correlation between the number of influenza (N = 11,768), and overall iGAS (N = 687), iGAS septic shock (n = 104), and iGAS pneumonia cases (n = 59) was assessed in weekly time periods using the Granger causality test. Results The number of hospitalized influenza cases was associated with an increase in the overall number of iGAS cases (Wald χ2 = 10.22, P = 0.04). Hospitalized influenza cases were associated with an increase in iGAS septic shock cases; every 1,000 increase in case counts were associated with one case of iGAS septic shock 1 week later (P = 0.02). Similarly, every 1,000 increase in hospitalized influenza cases were associated with one case of iGAS pneumonia 1 week later (P < 0.01). While the effect of Granger causality is cumulative when describing the causal relationship between hospitalized influenza and total iGAS, the correlation between influenza and the iGAS subgroups is best described with a 1-week lag. Conclusion Granger causality tests are commonly used in economic modeling but have not been routinely applied to infectious diseases. Using this test, we found a strong correlation between weekly cases of hospitalized influenza and iGAS cases, with a 1-week lag between influenza and iGAS septic shock or pneumonia. This approach can provide insight into the potential impact of developing prevention interventions for infections with strong correlation. Further exploration of Granger tests in infectious disease modeling should be considered. Disclosures All authors: No reported disclosures.