Using multiple surveillance systems to assess influenza activity levels in Minnesota

Abstract

Influenza is responsible for a significant burden of morbidity and mortality each year throughout the world. Thorough surveillance continues to be vital to public health – influenza surveillance contributes to vaccine strain selection, monitoring for novel strains of influenza, as well as important knowledge regarding risk factors and health care disparities. Surveillance for every case of influenza is generally not possible nor desirable due to the self-limiting nature of this illness for many people; in fact, many patients do not seek medical care or testing. Influenza surveillance is focused on specific circumstances and groups with the results being used to estimate overall activity and severity of influenza and influenza-like illness (ILI).There are a multitude of influenza surveillance systems conducted by health departments worldwide which attempt to characterize circulating strains and risk factors and health disparities for severe illness and poor health outcomes, as well as measuring general illness trends and activity to inform public health action. Surveillance on a global scale may include characterization of circulating influenza strains for vaccine strain selection, while surveillance on a local scale may include tracking outbreaks in long-term care (LTC) facilities to provide infection control guidance to limit the spread of influenza in the facility. Attempting to conduct surveillance for influenza has led to many innovative ideas worldwide, including digital crowdsourcing to collect real-time data, such as the Flu Near You (FNY) application, which was renamed Outbreaks Near Me in 2020, to which participants report whether they are experiencing influenza-like symptoms. These data are then made available to local health departments at no cost. These data may uniquely cover non-medically attended influenza infections, which would not otherwise be included in conventional disease surveillance. Minnesota uses six surveillance systems to measure overall influenza activity: sentinel outpatient surveillance of ILI (ILINet), laboratory data on influenza positivity (Minnesota Laboratory Survey [MLS]), Long-Term Care (LTC) facility and K-12 school outbreak surveillance, case-based influenza hospitalization surveillance (FluSurv-NET), and case-base influenza-associated death surveillance. This dissertation analyzed influenza surveillance data in Minnesota, including FNY data, over nine influenza seasons (2011-12 through 2019-20). These analyses sought to determine if FNY data is sufficient for estimating overall influenza activity, to describe the activity and severity of influenza as determined by each surveillance system and whether these systems corroborate with each other in terms of influenza activity over a season, and to evaluate each surveillance system to identify strengths and weaknesses. FNY data is not likely to provide useful data on influenza activity levels at the current rate of participation in Minnesota but may provide excellent supplemental data in areas where participation is broader and more consistent. During the 2011-12 through 2019-20 seasons, participation was largely inconsistent, and significantly over-represented females aged 50 years and older who live in urban areas of the state. The myriad influenza surveillance systems used in Minnesota largely correlate with each other and each gives a fairly accurate representation of general influenza trends of the season. Without an accurate standard of true influenza activity for comparison, systems were compared to each other, and each was able to suggest when influenza activity was increasing, had peaked, or was decreasing. Notable exceptions to this are that influenza-associated deaths generally showed increases, peaks, and decreases 1-2 weeks later than other systems, and FNY data showed early increases and early decreases, potentially due to fading user participation during the season, rather than a decrease in activity. Performing a surveillance program evaluation for each of the six influenza surveillance systems and FNY showed that significant influenza data could be collected with minimal resources by using sentinel outpatient surveillance (ILINet) and laboratory data (MLS), if there is sufficient participation from clinics and laboratories, respectively. The most comprehensive data can be collected via case-based surveillance (FluSurv-NET/hospitalized influenza surveillance), but this is resource intensive, and not possible in all jurisdictions. The conclusions of this dissertation are that while comprehensive surveillance data for influenza is more advantageous, general trends regarding an influenza season can still be determined with more limited surveillance. While this may not contribute to global monitoring of influenza viruses and overall knowledge of influenza disease, this has important implications for resource-limited jurisdictions that wish to monitor influenza illness outbreaks to inform the need for direct public health action, such as vaccination campaigns, education, or infection control guidance.