Presentation (CTI)

OBJECTIVES: A lesson learned from the COVID-19 pandemic is that, to prevent or mitigate future, similar events, we need not only respiratory illness surveillance, but also the ability to forecast trends on a national scale. This research explored opportunities to use pharmacy and public health data to predict future surges.
METHODS: The datasets utilized in this study included location-specific testing results for respiratory viruses (influenza viruses and SARS-COV-2), antiviral prescription (nirmatrelvir/ritonavir for treating COVID-19 and oseltamivir for treating influenza) claims, over-the-counter (OTC) treatment sales, and SARS-COV-2 viral levels detected in nationwide wastewater monitoring data collected by CDC. One year of each dataset was collected except CDC wastewater data to build an individual time series predictive model in SAS Enterprise Guide. The data for the post-period was forecasted and compared to the pre-period using each predictive algorithm to determine if there was a projected surge for each dataset. Finally, results were combined across data sources to determine whether there was an imminent surge in the coming month at each location after the viral level changes in nearby wastewater sites were added to the algorithm.
RESULTS: The combined data sources revealed a specific seasonal pattern: nirmatrelvir/ritonavir claims and COVID-19 positivity displayed two major peaks in summer and winter, which aligned with the viral levels in wastewater. Our results showed that the forecasting accuracy (true positive and true negative cases out of total cases) for COVID-19 positivity reached 87% (81% for oseltamivir fills, 77% for OTC sales, and 61% for nirmatrelvir/ritonavir claims). After combining each data source together with CDC wastewater information, the accuracy improved to 97%.
CONCLUSIONS: Our model successfully leveraged multiple data sources in forecasting the respiratory illness trend nationally. It can proactively inform of geographic variation of impending surges for public awareness and seasonal planning.