Time trends in social contacts before and during the COVID-19 pandemic: the CONNECT study

Abstract

Background: Since the beginning of the COVID-19 pandemic, many countries, including Canada, have adopted unprecedented physical distancing measures such as closure of schools and non-essential businesses, and restrictions on gatherings and household visits. We described time trends in social contacts for the pre-pandemic and pandemic periods in Quebec, Canada.

Methods: CONNECT is a population-based study of social contacts conducted shortly before (2018/2019) and during the COVID-19 pandemic (April 2020 - February 2021), using the same methodology for both periods. We recruited participants by random digit dialing and collected data by self-administered web-based questionnaires. Questionnaires documented socio-demographic characteristics and social contacts for two assigned days. A contact was defined as a two-way conversation at a distance ≤ 2 m or as a physical contact, irrespective of masking. We used weighted generalized linear models with a Poisson distribution and robust variance (taking possible overdispersion into account) to compare the mean number of social contacts over time and by socio-demographic characteristics.

Results: A total of 1291 and 5516 Quebecers completed the study before and during the pandemic, respectively. Contacts significantly decreased from a mean of 8 contacts/day prior to the pandemic to 3 contacts/day during the spring 2020 lockdown. Contacts remained lower than the pre-COVID period thereafter (lowest = 3 contacts/day during the Christmas 2020/2021 holidays, highest = 5 in September 2020). Contacts at work, during leisure activities/in other locations, and at home with visitors showed the greatest decreases since the beginning of the pandemic. All sociodemographic subgroups showed significant decreases of contacts since the beginning of the pandemic. The mixing matrices illustrated the impact of public health measures (e.g. school closure, gathering restrictions) with fewer contacts between children/teenagers and fewer contacts outside of the three main diagonals of contacts between same-age partners/siblings and between children and their parents.

Conclusion: Physical distancing measures in Quebec significantly decreased social contacts, which most likely mitigated the spread of COVID-19.

Keywords: COVID-19; Infectious disease; Mathematical modeling; Public health; Social contacts; Social distancing measures.

Fig. 1

Time trends in the number of social contacts in the province of Quebec. Pre-COVID: February 1^st 2018 to March 17^th 2019; Spring 2020: April 21^st to May 25^th 2020; Summer 2020: July 3^rd to August 31^st 2020; Fall 2020: October 1^st to December 16^th 2020; Holidays 2020–2021: December 17^th 2020 to January 8^th 2021. 1^st wave: Spring 2020; 2^nd wave: September 2020 to February 2021. *Contacts at work were truncated to a maximum of 40 contacts per day. ^†Contacts for workers in schools were included in the school location. Stringency index: Higher values indicate stricter measures; the mean stringency index for each of the 8 periods was obtained by averaging the daily values of the index. Error bars represent the 95% confidence interval of the total number of social contacts

Fig. 2

Time trends in the number of social contacts in the province of Quebec, according to age of participants and location of contacts. A 0–17 years old^*. B 18–65 years old^†. C  > 65 years old. Pre-COVID: February 1^st 2018 to March 17^th 2019; Spring 2020: April 21^st to May 25^th 2020; Summer 2020: July 3^rd to August 31^st 2020; Fall 2020: October 1^st to December 16^th 2020; Holidays 2020–2021: December 17^th 2020 to January 8^th 2021. 1^st wave: Spring 2020; 2^nd wave: September 2020 to February 2021. *Only the main locations of contacts were included in this figure for 0–17-year-olds (contacts at work were excluded). ^†For adults: contacts at work are truncated to a maximum of 40 contacts per day and they include contacts in schools for school workers and students. Contacts in other locations include transports, leisure and other locations. Error bars represent the 95% confidence interval of the total number of social contacts

Fig. 3

Time trends in the total number of social contacts in the province of Quebec, according to key socio-demographic characteristics (adjusted for age). A Household size. B Household composition. C Level of education (among ≥ 25 years old). D Mother tongue. Pre-COVID: February 1^st 2018 to March 17^th 2019; 1^st wave: April 21^st to May 25^th 2020; Summer 2020: July 3^rd to August 31^st 2020; 2^nd wave: Sept 1^st 2020 to February 26^th 2021. Error bars represent the 95% confidence interval of the total number of social contacts

Fig. 4

Time trends in social contacts matrices. Pre-COVID: February 1^st 2018 to March 17^th 2019; Spring 2020: April 21^st to May 25^th 2020; Summer 2020: July 3^rd to August 31^st 2020; Fall 2020: October 1^st to December 16^th 2020; Holidays 2020–2021: December 17^th 2020 to January 8^th 2021. 1^st wave: Spring 2020; 2^nd wave: September 2020 to February 2021. The matrices of contacts at home include contacts with household members and visitors. The matrices of contacts in other locations include contacts at work, school, transport, leisure, and other locations