key: cord-1047647-kvnmrb9r
authors: McAloon, C. G.; Wall, P.; Butler, F.; Codd, M.; Gormley, E.; Walsh, C.; Duggan, J.; Murphy, T. B.; Nolan, P.; Smyth, B.; O'Brien, K.; Teljeur, C.; Green, M. J.; O'Grady, L.; Culhane, K.; Buckley, C.; Martin, J.; Doyle, S.; Carroll, C.; More, S. J.
title: Numbers of close contacts of individuals infected with SARS-CoV-2 and their association with government intervention strategies.
date: 2021-01-25
journal: nan
DOI: 10.1101/2021.01.20.21250109
sha: 17ad735dcd7e69a9c895500d87e4d87c0d3006c2
doc_id: 1047647
cord_uid: kvnmrb9r

Contact tracing is conducted with the primary purpose of interrupting transmission from individuals who are likely to be infectious to others. Secondary analyses of data on the numbers of close contacts of confirmed cases could also provide an early signal of increases in contact patterns that might precede larger than expected case numbers; evaluate the impact of government interventions on the number of contacts of confirmed cases; or provide information on contact rates between age cohorts for the purpose of epidemiological modelling. We analysed data from 140,204 contacts of 39861 cases in Ireland from 1st May to 1st December 2020. Only "close" contacts were included in the analysis. A close contact was defined as any individual who had had > 15 minutes face-to-face (<2 m) contact with a case; any household contact; or any individual sharing a closed space for longer than 2 hours, in any setting. The number of contacts per case was overdispersed, the mean varied considerably over time, and was temporally associated with government interventions. Negative binomial regression models highlighted greater numbers of contacts within specific population demographics, after correcting for temporal associations. Separate segmented regression models of the number of cases over time and the average number of contacts per case indicated that a breakpoint indicating a rapid decrease in the number of contacts per case in October 2020 preceded a breakpoint indicating a reduction in the number of cases by 11 days. These data were collected for a specific purpose and therefore any inferences must be made with caution. The data are representative of contact rates of cases, and not of the overall population. However, the data may be a more accurate indicator of the likely degree of onward transmission than might be the case if a random sample of the population were taken. Furthermore, since we analysed only the number of close contacts, the total number of contacts per case would have been higher. Nevertheless, this analysis provides useful information for monitoring the impact of government interventions on number of contacts; for helping pre-empt increases or decreases in case numbers, and for triangulating assumptions regarding the contact mixing rates between different age cohorts for epidemiological modelling.

Contact tracing is conducted with the primary purpose of interrupting transmission from 43 individuals who are likely to be infectious to others. Secondary analyses of data on the 44 numbers of close contacts of confirmed cases could also: provide an early signal of increases 45 in contact patterns that might precede larger than expected case numbers; evaluate the impact 46 of government interventions on the number of contacts of confirmed cases; or provide data 47 information on contact rates between age cohorts for the purpose of epidemiological 48 modelling.

Methods 50 We analysed data from 140,204 contacts of 39861 cases in Ireland from 1st May to 1st 51 December 2020. Only 'close' contacts were included in the analysis. A close contact was 52 defined as any individual who had had > 15 minutes face-to-face (<2 m) contact with a case; 53 any household contact; or any individual sharing a closed space for longer than 2 hours, in 54 any setting.

The number of contacts per case was overdispersed, the mean varied considerably over time, 57 and was temporally associated with government interventions. Negative binomial regression 58 models highlighted greater numbers of contacts within specific population demographics, 59 after correcting for temporal associations. Separate segmented regression models of the CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint These data were collected for a specific purpose and therefore any inferences must be made 65 with caution. The data are representative of contact rates of cases, and not of the overall 66 population. However, the data may be a more accurate indicator of the likely degree of 67 onward transmission than might be the case if a random sample of the population were taken. 68 Furthermore, since we analysed only the number of close contacts, the total number of 69 contacts per case would have been higher. Nevertheless, this analysis provides useful CC-BY-NC-ND 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.

Introduction 77 Tracing contacts of infected individuals for the purpose of monitoring, isolation or testing is a 78 fundamental pillar of communicable disease control [1] . In the control of COVID-19, contact 79 tracing has been employed since the earliest stages of the pandemic [2] as an essential public reported by confirmed cases during the contact tracing process could be used as a more 98 timely metric to evaluate the impact of public health interventions. Furthermore, given that 99 the number of contacts is a key driver of onward transmission, changes in the numeric 100 . CC-BY-NC-ND 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)

The copyright holder for this preprint this version posted January 25, 2021. population within the susceptible, exposed, infectious, and recovered (SEIR) compartments 114 to be described at a point in time [10] . Basic versions of this model assume random mixing 115 within the population, such that the contact rates for different cohorts within the population 116 are homogenous. In reality however, the contact rates across different cohorts in the 117 population is not expected to be uniform. If the contact rates between different cohorts in the 118 population is known, this information may be incorporated into SEIR models, and used to 119 estimate age-specific transmission parameters [11, 12] . Earlier work has quantified the contact CC-BY-NC-ND 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)

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint

The aim of this study therefore was to use national contact tracing data from Ireland to 124 describe the rates of contact between individuals within and between different age cohorts at CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint their body fluids or their laboratory specimen, or being present in the same room when an 170 aerosol generating procedure was undertaken on the case); or passengers on an aircraft sitting 171 within two seats (in any direction) of the case, travel companions or persons providing care, 172 and crew members serving in the section of the aircraft where the index case was seated. For 173 those contacts who shared a closed space (including an office or school setting) with a case 174 for >2 hours, a risk assessment was undertaken taking into consideration the size of the room, 175 ventilation and the distance from the case.

A casual contact was defined separately [14] . However, since these contacts were not 177 generally collected from the contact tracing of routine cases, our study only used those 178 contacts that were defined as 'close'. Research Hub is a secure data repository from which personally identifiable data cannot be 186 exported. These data were accessed through the CSO data hub by the first author for the 187 purpose of this analysis. Access was granted under Section 20(b) of the Statistics Act, 1993, 188 for the purpose of using data collected during the pandemic to aid in the national response. Table S1 . Briefly, Case 1 data were initially filtered to 208 include only those records where the current status entry indicated that contact tracing had 209 been completed. In addition, duplicate cases were removed by selecting the most recent data 210 entry where multiple entries existed for the same case. Contact data were initially filtered to 211 include only close contacts. Then, contacts identified by COVID Tracker were excluded as 212 these were not linked to a specific case in the contact tracing database. In addition, contacts 213 with no recorded primary case were also removed from the dataset. Finally, we also restricted 214 the analysis to contacts identified after May 1st, due to concerns over the variability of the 215 quality of data collection processes prior to this point.

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint Case and Contact data were then joined to create an overall dataset at the level of the contact 217 (that is, with each line representing a contact, with a column indicating the reference ID of 218 the primary case). Ages of the cases and contacts were categorised according to age groups 219 corresponding to school-age children, college-age adults, young adults, middle-age adults and 220 retired adults: 0-17; 18-24; 25-39; 40-64 and greater than 65 years of age. Location of the 221 case was recategorised as "Dublin" and "Rest of Country".

Next, data were collapsed to the case level for analysis (that is, with each line representing a 223 primary case), summarising the overall number of contacts reported by the case (Dataset 1). Table S1 .

After joining these data, collapsing to the case level and conducting the data cleaning steps 283 outlined in Supplementary Material Table S1, Dataset 1 consisted of 39861 case records.

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. from confirmed cases could serve as an early warning for increasing number of cases. 366 We found that changes in the number of contacts per case were temporally associated with 367 the introduction of government interventions. Such associations must be interpreted with care 368 since there is no possibility for a national control population and there is therefore potential 369 for spurious correlation. However, the regional introduction of interventions facilitates a CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint however, the mean number of contacts per case during the lowest level of government 377 contact restrictions (July to mid-September) was two to three times more than at the highest However, these authors found that as contacts increased from mid-April, a corresponding 387 increase in case numbers was not observed, which was potentially related to changes in the 388 nature of the contacts between individuals. It is also likely that the nature of contacts change likely that a critical threshold in contacts must be reached before Rt can be forced less than 1.

We observed that the number of contacts per case continued to decrease over that 5-week 400 . CC-BY-NC-ND 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. (Table 2 ). However, within each age 407 category, the number of contacts was not constant. Consequently, a much better model fit 408 was found when age was modelled as a spline.

A number of limitations must be considered when considering this study. Firstly, it must be 410 remembered that these contact rates are of those that were infected with SARS-CoV-2 at 411 particular points in time; that is, they are not a random sample from the population.

Therefore, care is needed in interpretation, in particular when examining temporal 413 associations. For example, at particular points in time in Ireland, older individuals were 414 overrepresented in the numbers of cases. The contacts from these individuals might be 415 expected to be less than in other age cohorts, with many of the most elderly in long term care 416 facilities, and this could be associated temporally with a particular government intervention.

As a consequence, changes, for example a reduction, in the average number of contacts could 418 largely be a consequence of changes to case demographics rather than as a result of 419 government restrictions. 420 However, regional comparison of contact number suggests this is not a significant factor. For 421 example, Figure 1 shows that contacts per case fell much faster in Dublin than the rest of the 422 country following the introduction of regional Level 3 restrictions. Following this, Level 3 423 was extended nationally, with a corresponding rapid drop in contacts per case across the rest 424 . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint of the country. Furthermore, whilst the number of contacts per case is not representative of 425 the overall population, they are representative of the individuals who were infected at that 426 point in time. Therefore, these data are likely more informative with respect to predicting 427 onward transmission than a random sample from the Irish population might have been, and 428 therefore more likely to be of use in forecasting changes in the trajectory of the disease. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint comparable between studies. However, the relative reduction in contacts over time may be 449 compared and is useful to monitor changes in contact behaviour over time. 450 Whilst the contact tracing programme successfully traced the overwhelming majority of 451 contacts, due to a rapid increase in cases late October 2020, approximately 2,000 cases over a 452 48-hr period were not contact traced. These data are therefore not present in our dataset. 453 However, unless these data were systematically different to the overall population of cases at . CC-BY-NC-ND 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)

The copyright holder for this preprint this version posted January 25, 2021. CC-BY-NC-ND 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)

The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint

Modeling contact tracing in outbreaks with application 474 to Ebola

Evaluation of the effectiveness of 476 surveillance and containment measures for the first 100 patients with

Effectiveness 480 of isolation, testing, contact tracing, and physical distancing on reducing transmission of 481

SARS-CoV-2 in different settings: a mathematical modelling study

Routine testing of close contacts of confirmed COVID-19 485 cases-National COVID-19 Contact Management Programme

R: A language and environment for statistical computing. R 521 Foundation for Statistical Computing

Elegant Graphics for Data Analysis

Dates and Times Made Easy with lubridate

The mgcv package

zoo: S3 Infrastructure for Regular and Irregular Time Series

Muggeo VMR segmented: an R Package to Fit Regression Models with Broken-Line 530

ggplot2: elegant graphics for data analysis

Changes in contact 534 patterns shape the dynamics of the COVID-19 outbreak in China

International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

on the Mobility and Relation to Infection Patterns

Inferred duration 540 of infectious period of SARS-CoV-2: rapid scoping review and analysis of available evidence 541 for asymptomatic and symptomatic COVID-19 cases

Incubation period 544 of COVID-19: a rapid systematic review and meta-analysis of observational research

COVID-19 Educational Settings cases and close contacts definitions

CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted January 25, 2021. . CC-BY-NC-ND 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)The copyright holder for this preprint this version posted January 25, 2021. ; https://doi.org/10.1101/2021.01.20.21250109 doi: medRxiv preprint