The Centers for Disease Control and Prevention (CDC) partnered with the Wisconsin Division of Health Services (WDHS; Madison, WI, USA) and University of Wisconsin—Oshkosh to investigate COVID-19 among on-campus residents during the fall 2020 semester (September 2–December 19). On-campus residents were housed in 8 dormitories (dorms A–H). In 7 dormitories, students resided in double-occupancy rooms (dorm C included 4 triple-occupancy rooms) and shared bathrooms along with a common area per floor. Dorm D was the only dormitory made up of suites in which <4 students lived in either 4 single-occupancy or 2 double-occupancy bedrooms with the suite’s own bathroom, common area, and kitchen. Not all bedrooms were occupied at their full capacity. After a positive COVID-19 diagnosis, on-campus residents were housed in an isolation dormitory. Students who might have been exposed were housed in a separate quarantine dormitory (9,10). (Appendix).

Data provided by the university included the number of available rooms, dormitory room types, student housing contracts, serial testing records, and a list of all student COVID-19 cases. For our study, we defined students sharing a bedroom with another student at the start of the semester as having a shared bedroom. Students sharing a suite or defined as having a shared bedroom were classified as having a shared living space. In addition, we defined dormitory floor-level occupancy as the number of occupied rooms divided by the number of rooms per floor. We defined a laboratory-confirmed case as a positive SARS-CoV-2 antigen or reverse transcription PCR test result for any on-campus student during the fall semester (11).

All data were analyzed using R version 4.0.2 (R Foundation for Statistical Computing, https://www.r-project.org). We used χ² tests, Fisher exact tests, and t-tests to determine differences between COVID-19 cases and noncases. We modeled the association between student characteristics and a laboratory-confirmed COVID-19 case using univariable and multivariable logistic regression; covariates were age, sex, race, ethnicity, all dormitories, and dormitory floor level occupancy. The dormitory with the lowest COVID-19 positivity for the semester was the reference group. Sharing a bedroom or living space were analyzed in separate models. We conducted our investigation consistent with applicable federal laws and CDC policy (e.g., 45 C.F.R. part 46.102(l)(2), 21 C.F.R. part 56; 42 U.S.C. §241(d); 5 U.S.C. 145 §552a; 44 U.S.C. §3501 et seq.). CDC and WDHS reviewed the investigation; in addition, the university’s ethics review board determined the activities to be nonresearch public health surveillance.

At the start of the fall semester, 2,187 students had on-campus housing contracts. The median age of on-campus students was 19 years; 60.5% were female, 79.4% White, and 79.7% non-Hispanic/Latino (Table 1). Dormitory student populations range was 176–405 students per dormitory, with a mean of 55 students per occupied dormitory floor and a mean floor occupancy of 85% (Table 2) at semester start. Overall, 74.5% of students shared a bedroom and 81.7% of students shared a living space.

During the semester, 528 (24.1%) COVID-19 cases were identified among on-campus students. The percentage of students diagnosed with COVID-19 was 17.0%–31.1% across dormitories for the fall semester; the lowest percent positivity was in dorm E. All dormitories saw a rise in cases in mid- to late September (Figure 1).

Using a univariable regression model, we found that students who shared a bedroom had 1.52 (95% CI 1.19–1.92) times the odds of receiving a COVID-19 diagnosis as students who lived alone (Figure 2). The effect estimate remained unchanged in the adjusted multivariable regression model (Appendix Table 1). However, in the adjusted model, students with a shared living space (e.g., suites and bedrooms) had 1.80 (95% CI 1.28–2.55) times the odds of testing positive for COVID-19 compared with students living alone. After controlling for shared living status, students from 2 dormitories, dorms D and F, had higher odds for COVID-19 than dorm E students in both models.

CDC has provided guidance on prevention measures to reduce the transmission of SARS-CoV-2 at IHEs (6,8). Similar to other IHE outbreak reports from the United States, the university saw a surge in cases during September (1,3–5). After the surge, the university updated their COVID-19 prevention plan such that residential students were tested weekly instead of biweekly for SARS-CoV-2, messaging on COVID-19 prevention measures increased, and on-campus dining was limited to takeout only for 2 weeks.

Despite this university’s updated COVID-19 prevention plan, students sharing a suite or bedroom had higher odds of being diagnosed with COVID-19. SARS-CoV-2 household transmission studies have shown that households are a significant transmission source for both symptomatic and asymptomatic persons (12,13). For example, a meta-analysis found that the household SARS-CoV-2 secondary attack rate was 16.6% (8,12). Reducing the number of students with roommates or those in suite-style units is needed to limit SARS-CoV-2 transmission.

After adjusting for sharing bedrooms or living spaces, students from 2 dormitories still had higher odds of having COVID-19 than students from the dormitory with the lowest percentage of positive students. This finding could be associated with differing student attitudes and social behaviors towards COVID-19 (14).

Racial and ethnic disparities in COVID-19 incidence have been found in persons <25 years of age in the United States (15). However, we found that Native Hawaiian or other Pacific Islander students had lower odds of having COVID-19 compared with White students, and Hispanic students had lower odds than non-Hispanic students: adjusted odds ratio for Native Hawaiian or other Pacific Islander students was 0.13 (95% CI 0.01–0.63) and for Hispanic students it was 0.56 (95% CI 0.31–0.96). We observed no other associations by age, sex, race category, and dormitory floor occupancy. These results should be interpreted with caution; our findings could be the result of low sample sizes in some groups or residual confounding.

The first limitation of our study is that findings from this IHE may not be generalizable for all IHEs. Second, these results characterize an association between sharing a living space and COVID-19 and do not necessarily indicate roommate transmission. Third, students may have moved out of the dormitory during the semester, causing an underestimation of attack rates and misclassification of those students with roommates or suitemates for the term. Last, because this investigation was cross-sectional in design, a causal relationship cannot be determined.

In summary, sharing a living space or bedroom was associated with increased odds of having COVID-19 even with COVID-19 prevention policies at a Wisconsin university. Reducing the number of students sharing living spaces could further prevent the spread of SARS-CoV-2 on-campus as part of COVID-19 prevention practices at IHEs.