Contacts and behaviours of university students during the COVID-19 pandemic at the start of the 2020/21 academic year

CONQUEST (COroNavirus QUESTionnaire) is an online survey of contacts, behaviour, and COVID-19 symptoms for University of Bristol (UoB) staff/students. We analysed survey results from the start of the 2020/2021 academic year, prior to the second national lockdown (14/09/2020-01/11/2020), where COVID-19 outbreaks led to lockdown of some student halls of residence. The aim of these analyses was to enhance knowledge of student contact patterns to inform infection disease mathematical modelling approaches. Responses captured information on demographics, contacts on the previous day, symptoms and self-isolation during the prior week, and COVID-19 status. 740 students provided 1261 unique records. Of 42 (3%) students testing positive in the prior fortnight, 99% had been self-isolating. The median number of contacts on the previous day was 2 (interquartile range: 1-5), mode: 1, mean: 6.1; 8% had [≥]20 contacts. 57% of student contacts were other UoB students/staff. Most students reported few daily contacts but there was heterogeneity, and some reported many. Around 40% of student contacts were with individuals not affiliated with UoB, indicating potential for transmission to non-students/staff.

1 0 1 mathematical models. We aimed to investigate the reported behaviours and contact patterns 1 0 2 among students of the UoB during the start of the 2020/2021 academic term by carrying out  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 Methods 2 6 6 Students that reported isolating within the last week had a lower mean number of contacts 2 6 7 on the previous day (4.5) than those not isolating (6.4) ( Table 5). The number of "individual" 2 6 8 contacts appeared to be similar between those not isolating (2.3) and those isolating (2.1), 2 6 9 however the "group" contacts were higher among those not isolating (2.5) than those 2 7 0 isolating (1.8), as were "other" contacts (1.6 vs 0.6). Staff had lower mean numbers of 2 7 1 overall contacts on the previous day than students (5.2 vs 6.1), which was driven by having 2 7 2 lower numbers of "group" (1.8 vs 2.4) and "other" contacts (0.6 vs 1.5).

7 3
The mean percentage of "individual" contacts on the previous day that involved touch was 2 7 4 39% (SD: 41.0%) overall, 35% (SD: 42%) for males, and 42% (SD: 41%) for females. Overall, the mean percentage of "individual" contacts on the previous day that were with 2 7 6 household members was 64% (Table 5). There was a higher percentage of household 2 7 7 contacts on the previous day for those who had been isolating within the last 7 days, than for 2 7 8 those who had not been isolating within the last 7 days (84% and 61% respectively). Similar results are seen for the percentage of contacts that were frequent (where the person would 2 8 0 usually meet that particular contact ≥ 4 times a week) as for those seen for household 2 8 1 contacts. 62% of "individual and group" contacts on the previous day were made at the 2 8 2 home of the respondent, and this percentage was lower among those not isolating within the 2 8 3 last 7 days (59%) than among those that had been isolating (80%). Whilst the percentage of 2 8 4 contacts on the previous day made at the university were similar between those that had and 2 8 5 had not been isolating within the last 7 days (10% vs 7%), the percentage of contacts at 2 8 6 other locations was higher among those that had not been isolating in the prior week (35%) 2 8 7 than those that had been isolating (18%). 57% of "individual and group" contacts on the 2 8 8 previous day were with other UoB students or staff -this percentage was lower among those 2 8 9 not isolating within the past week (54%) than those isolating (81%). In comparison to 2 9 0 students, staff had a higher number of contacts on the previous day that involved touch 2 9 1 (57% for staff vs 39% for students). Similar numbers of their "individual and group" contacts 2 9 2 on the previous day were made at home for staff (61%) and students (62%), whilst far fewer 2 9 3 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020 of the contacts of staff on the previous day were either UoB staff or students (16% for staff 2 9 4 vs 54% for students). The mean percentage of the student's "individual" non-UoB contacts 2 9 5 that were household members was 50%.

9 6
Participants that had not been isolating in the prior week had shorter mean contact durations 2 9 7 with their contacts at home (3.3 hours) than those that had been isolating (3.9 hours), and 2 9 8 longer durations of their contacts on the previous day in a location other than home or  In unweighted analyses looking at repeat records from participants, there were 37 records 3 0 2 where a participant self-reported not isolating in the 7 days before one survey completion 3 0 3 date but then isolating in the 7 days before their next survey completion. For these records, 3 0 4 the mean number of contacts was 7.1 (SD: 7.1) for the first survey (when not isolating) and went from isolating to not isolating, where the mean number of contacts on the previous day 3 0 7 went from 8.7 (SD: 19.6) at the first survey to 9.2 (SD: 13.3) at the second. test within the last two weeks, having previously said they had no history of suspected or 3 1 0 confirmed infection with COVID-19 (i.e. new cases). For these records, the mean number of 3 1 1 contacts on the previous day was 7.8 (SD: 8.2) at the first survey and 6.2 (SD: 6.1) at the 3 1 2 second. Only 6 individuals reported current infection, and subsequently reported a previous 3 1 3 infection at the next survey. The mean number of contacts reported by these individuals was 3 1 4 3.9 (SD: 4.0) at the first survey and 5.6 (SD: 6.1) at the second. In the multivariable regression analysis of the number of contacts for the previous day (Table   3 1 7 6), older ages were associated with a lower number of contacts when compared with those 3 1 8 aged 17-24 years. Students in their 4th (or higher) year of study reported higher numbers of 3 1 9 contacts for the previous day than students in their 1st year. Reporting the cardinal COVID-3 2 0 19 symptoms within the last week was associated with a higher number of contacts on the 3 2 1 previous day (versus not having the cardinal COVID-19 symptoms), whilst isolating within 3 2 2 the week before the survey was associated with having fewer contacts on the previous day. In the multivariable regression analysis, participants having a household size of 1 was 3 2 4 associated with higher numbers of contacts than participants having a household size of 2-3.

2 5
Similarly, in comparison to having a household size of 2-3, a household size of 4-5 was 3 2 6 associated with more contacts, whilst not reporting household size was associated with 3 2 7 reporting fewer contacts. COVID-19 status was associated with number of contacts. Those 3 2 8 that had not tested positive for or did not suspect themselves to have had COVID-19 had 3 2 9 lower numbers of contacts on the previous day than those that suspected themselves to 3 3 0 have had COVID-19 more than two weeks prior to the survey. Those testing positive within 3 3 1 the last 2 weeks before survey completion had fewer contacts. Students in catered and self-3 3 2 catered halls had fewer contacts on the previous day then those living in a shared house/flat 3 3 3 but students living in a shared house/flat had fewer contacts than those living with their isolating in the previous week having lower numbers of contacts than that had not been  CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint 9 3 4 1 3 4 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint 1 1 with the COVID-19 guidelines. Students living with their family appeared to report the highest 3 9 4 number of contacts, with those living in catered and self-catered halls reporting lower 3 9 5 numbers of contacts.

9 6
Unsurprisingly given the COVID-19 restrictions in place at the time of our survey, we found a 3 9 7 lower mean number of daily contacts among our student population (6.1) than was found in (26.8) or among the students in that sample (29.9). However, that study also found a large 4 0 0 amount of heterogeneity in the number of contacts. In the Warwick social contacts survey 4 0 1 data, students have more home contacts (3.5) than other participants (2.3), whilst they found 4 0 2 that the majority of contacts for students (82%, 95% confidence interval: 79%-86%) were 4 0 3 either at home or associated with the university, and that students reported 20 (95% 4 0 4 confidence interval: 14.1-28.8) university-related contacts. We also found that a high 4 0 5 percentage of student contacts were either at home or university (around 72%). However, 4 0 6 we found that both students and our comparator sample of staff had 1.6 contacts at home 4 0 7 and that there were only 3.1 university contacts for students on average, with these 30), lower than the Warwick social contacts survey, but still much higher than our mean 4 1 2 value (6.1). The POLYMOD survey found that people of the same age tended to mix with 4 1 3 each other and that people in the main student age group (18-24) have more contacts than 4 1 4 older adults, both of which are corroborated by our findings. In the POLYMOD survey, 4 1 5 around 75% of contacts at home and 50% of school and leisure contacts involved touch, 4 1 6 results that seem closer to our findings for staff (57% of contacts involving touch) than 4 1 7 students (39%), which is likely both down to physical distancing measures and due to 4 1 8 students being less likely to live with family members than staff or other adults. The strengths of this survey include the sample size, longitudinal format, and anonymous 4 2 0 nature that enable us to capture self-reported behaviours of a large number of students 4 2 1 during a key period in the UK's COVID-19 pandemic. In addition, it provides a unique data 4 2 2 source on student behaviour during the pandemic, which will be useful in informing public  However, this study has some limitations. Firstly, the number of contacts was asked for the 4 2 8 previous day, whilst the questions on self-isolation and symptoms asked about the previous 4 2 9 week, and a window of 14-days was used to define current COVID-19. This discrepancy in 4 3 0 time-windows used for different questions could lead to difficulties in interpreting results, 4 3 1 particularly regarding contact patterns for those that had previously been isolating during the 4 3 2 prior week but not on the previous day, possibly leading to higher reported contacts for this participants had a negative test for COVID-19, which would have been useful information. Thirdly, in order to capture sufficient detail on contacts, the questionnaire is fairly long (5-10 4 3 7 minutes) and complicated, which may deter those with lots of contacts or with little available filled in their household sizes, which perhaps shows that some people struggled to answer  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint Selection bias for those who particularly engaged in health-seeking behaviours may have 4 4 4 occurred, as those that are less likely to abide by the guidelines may also be less likely to fill 4 4 5 out the survey. However, while we are not able to identify the proportion of the population 4 4 6 that are not complying with COVID-19 restrictions, we did capture individuals who did not 4 4 7 appear to be compliant that were reporting large numbers of contacts and not isolating when is for students who have had COVID-19. Almost one-fifth of our surveyed student population 4 5 0 had tested positive for COVID-19 or suspected that they had had COVID-19, however, only 4 5 1 around 7% students had had a positive test as of the 1st November(10). Nevertheless, the 4 5 2 true prevalence of COVID-19 in the student population may be greater than 7%, since 4 5 3 students with symptoms may not want to present for a test to avoid the potential of obligatory 4 5 4 isolation for them and their household. There will inevitably be issues regarding recall bias, 4 5 5 particularly when we are asking respondents to estimate when they first think they had 4 5 6 COVID-19 (if this hasn't been confirmed by a positive test), and there will also likely be 4 5 7 issues with response bias, leading to inaccurate or false responses. Our study comes at a crucial time in the COVID-19 pandemic, Autumn 2020, when the 4 5 9 disease is resurgent with high numbers of daily cases, including among university 4 6 0 students (7). The second national UK lockdown, commenced on the 5th November 2020, has 4 6 1 not included the closing of universities (22) and therefore, it is vital that we continue to  It is important to understand the epidemiology of COVID-19 among students due to high people moving from all over the country and world to study, forming new social networks in 4 6 8 the process. Although the student population is mostly young and therefore unlikely to see in the COVID-19 pandemic, as well as crucial data for any future infectious disease 4 7 8 outbreaks. We found that the number of daily contacts for students was much lower than in 4 7 9 pre-COVID-19 studies, which is likely to be due to the COVID-19 restrictions in place. We show that whilst the majority of students report low numbers of contacts on the previous day, transmission to groups other than students. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint 1 3 Acknowledgements: We would like to thank the Elizabeth Blackwell Institute for funding 4 8 9 this research, our RedCap data manager Alison Horne and our PPI group for their feedback 4 9 0 during the development of the survey. We would also like to thank all the participants who 4 9 1 have taken part in this study.

9 2
Competing interests: JGW has received research funding from Gilead Sciences unrelated 4 9 3 to this research. All other authors declare no competing interests. Evaluation at the University of Bristol. HC is additionally funded through an NIHR Career    . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020.  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint 1 5 2 4 . C  h  r  i  s  t  e  n  s  e  n  H  ,  T  u  r  n  e  r  K  ,  T  r  i  c  k  e  y  A  ,  B  o  o  t  o  n  R  D  ,  H  e  m  a  n  i  G  ,  N  i  x  o  n  E  ,  e  t  a  l  .  C  O  V  I  D  -1  9   t  r  a  n  s  m  i  s  s  i  o  n  i  n  a  u  n  i  v  e  r  s  i  t  y  s  e  t  t  i  n  g  :  a  r  a  p  i  d  r  e  v  i  e  w  o  f  m  o  d  e  l  l  i  n  g  s  t  u  d  i  e  s  .  m  e  d  R  x  i  v  .  2  0  2  0  . . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint
*"Individual" contacts were the people that the participant spoke to in person one-on-one, including those in the participant's household and support bubble. "Group" contacts were the contacts that the participant had with large groups of individuals in the same setting (for example, sports teams, tutorials, lectures, religious services, large gatherings with friends and family). "Other" contacts were the many people participants spoke to one-onone in the same setting where the contacts did not have the opportunity to speak to each other (for example, as part of a customer service role in a shop). Not all of the contact types were asked for each category of contacts, so are only comparable to the associated categories indicated here.

Mean (95% confidence interval), Median (IQR)
Students ( . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint Figure 1: Unweighted histograms of the number of overall contacts* on the previous day among a) students (including staff/students); b) staff (excluding staff/students) a) b) . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint *81 students had more than 20 contacts on the previous day; 58 staff had more than 20 contacts on the previous day -full histograms are shown in supplementary figure 1.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20246421 doi: medRxiv preprint Figure 2: Mean number of contacts on the previous day and the proportion of people isolating within the prior week by residence type . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted December 11, 2020. ;https://doi.org/10.1101https://doi.org/10. /2020