SARS-CoV-2 Seroprevalence: Demographic and Behavioral Factors Associated With Seropositivity Among College Students in a University Setting

Abstract

Purpose: Examine SARS-CoV-2 seroprevalence and the association of seropositivity with demographic, geographic, and behavioral variables among University of North Carolina Chapel Hill (UNC-CH) undergraduate students enrolled in the fall 2020 semester.

Methods: All UNC-CH undergraduate students were invited to participate in the Heelcheck study; participants were weighted to the UNC-CH undergraduate population using raking methods. We estimate SARS-CoV-2 seroprevalence at study entrance (11/12/2020-12/10/2020) and bivariable associations using log-binomial regression.

Results: SARS-CoV-2 seroprevalence was 7.3% (95% confidence interval (CI): 5.4%-9.2%) at baseline. Compared to students who were living off-campus in the Chapel Hill/Carrboro area (CH) for the Fall 2020 semester (8.6% seroprevalence), students who never returned to CH had lower seroprevalence (1.9%, prevalence ratio (PR), 95% CI: 0.22, 0.06-0.81), whereas, students who started the semester on-campus and moved to off-campus CH housing had 18.9% seroprevalence (PR, 95% CI: 2.21, 1.04-4.72) and students who spent the semester living in a Sorority/Fraternity house had 46.8% seroprevalence (PR, 95% CI: 5.47, 2.62-11.46). Those who predicted they would join an indoor party unmasked had 3.8 times the seroprevalence of those who indicated they would not attend (PR, 95% CI: 3.80, 1.58-9.16). Compared to students who disagreed with the statement "…I am not going to let COVID-19 stop me from having fun…", those who agreed had higher seroprevalence (14.0% vs. 5.7%; (PR, 95% CI: 2.45, 1.13-5.32)).

Discussion: Increased seroprevalence was associated with congregate living and participation (actual or endorsed) in social activities. During pandemics, universities must create safe socializing opportunities while minimizing transmission.

Keywords: COVID-19; Campus life; College life; SARS-CoV-2; Serology; Seroprevalence.

Figure 1

Estimated associations between SARS-CoV-2 seropositivity and demographic/geographic characteristics in the weighted sample. CH: Chapel Hill; Weighted prevalence ratio (PR) estimates and 95% CIs were calculated using log-binomial regression with a robust error variance; PR presented on the log scale; Reference group remains constant until noted; PR not estimated if group had five or fewer individuals and/or zero outcomes (subgroups: Black/African-American, students who were in CH off campus on Aug 10 and then left CH).