Date of Award

January 2023

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Daniel Weinberger


Background: SARS-CoV-2, an emerging epidemic respiratory virus, has circulated in the United States since early 2020. The dynamics of Influenza (flu) and Respiratory Syncytial Virus (RSV) have shifted, causing atypical timing and levels of illness across the country and in particular age groups. Fever, recorded on at-home bluetooth-enabled thermometers, is a promising new tool with which to monitor respiratory disease. Quantifying the proportion of fever attributable to these three viruses, at a particular time or within a particular age group, would strengthen the collective understanding of viral circulation in the U.S., and enable the use of fever data in the prediction and prevention of respiratory disease.

Methods: The study period, July 2016-March 2023, encompassed the first years of the COVID-19 pandemic. A negative binomial Bayesian hierarchical model was fit to fever with different measures of viral covariates, accounting for temporal changes in diagnostic testing behavior. MCMC simulations informed estimation of the proportion of fever attributable to respiratory viral causes. Attributable proportions were compared across groups and across a Pre- and Post-COVID temporal stratification.

Results: Prior to the COVID-19 pandemic, both flu and RSV were strongly correlated with fever. Post-COVID, the timing, magnitude, and fever correlation of both viruses shifted. The model attributed more Pre-COVID fever to flu than RSV. In the Post-COVID period, fever was less attributable to viral causes than before. SARS-CoV-2 contributed to fever incidence more than flu or RSV with the models that used viral covariates adjusted for testing behavior. Pre-COVID, viruses were most responsible for fever in children 5-17 and least responsible for fever among elders 65-100. SARS-CoV-2 only minimally impacted fever incidence of pediatric populations.

Conclusions: This study emphasizes the utility of fever for capturing and understanding respiratory disease in the U.S. and across age groups. As new variants emerge, or SARS-CoV-2 stabilizes and develops endemic seasonality, it will be necessary to continue monitoring the age distribution of viral diseases, and nonclinical fever surveillance is a strong supplement to traditional methods.


This is an Open Access Thesis.

Open Access

This Article is Open Access