Efficacy of contact tracing for the containment of the 2019 novel coronavirus (COVID-19)

Abstract

Objective: Contact tracing is a central public health response to infectious disease outbreaks, especially in the early stages of an outbreak when specific treatments are limited. Importation of novel coronavirus (COVID-19) from China and elsewhere into the UK highlights the need to understand the impact of contact tracing as a control measure.

Design: Detailed survey information on social encounters from over 5800 respondents is coupled to predictive models of contact tracing and control. This is used to investigate the likely efficacy of contact tracing and the distribution of secondary cases that may go untraced.

Results: Taking recent estimates for COVID-19 transmission we predict that under effective contact tracing less than 1 in 6 cases will generate any subsequent untraced infections, although this comes at a high logistical burden with an average of 36 individuals traced per case. Changes to the definition of a close contact can reduce this burden, but with increased risk of untraced cases; we find that tracing using a contact definition requiring more than 4 hours of contact is unlikely to control spread.

Conclusions: The current contact tracing strategy within the UK is likely to identify a sufficient proportion of infected individuals such that subsequent spread could be prevented, although the ultimate success will depend on the rapid detection of cases and isolation of contacts. Given the burden of tracing a large number of contacts to find new cases, there is the potential the system could be overwhelmed if imports of infection occur at a rapid rate.

Keywords: Disease modeling; Epidemiology; communicable diseases; public health policy.

Figure 1

(A) Cartoon example of the encounters made during a day by an infectious index case (central figure) with contacts positioned by their total contact duration. Here, the definition of a contact is someone with whom the index case encountered for 15 min or longer. Some contacts will be identifiable (green), while others will be unidentifiable (orange). A definition of contact that is too restrictive and inappropriate for the infection means some encounters may fail to meet the definition yet may be at risk of infection; these excluded contacts could be identifiable (light grey) or unidentifiable (orange). (B) Examples of ego-centric networks collected by the survey. The participant (ego) is the blue central triangle; circles represent individual contacts, squares represent groups of contacts (size of group indicated). Colours represent social settings of encounters (red=home, cyan=work/school, yellow=travel, pink=other). Larger symbol sizes represent longer contact durations, while a closer proximity to the ego indicates the contact is more frequently encountered.

Figure 2

Distributions associated with transmission and contact tracing. (A) Infectivity over time based on an SEIR model with a latent period of 4 days (Erlang distribution with shape 3), infectious period of 2 days, R₀=3. (B) Frequency distribution of the number contacts over a 14-day period using colours from figure 1A: white is all contacts; blue are those matching the >15 min definition of a close contact; green are those matching the definition that are also assumed to be identifiable (met previously or for more than 1 hour), and therefore traceable. (C) Frequency distribution of the number of secondary cases per index case, again using colours from figure 1A: red is all secondary cases; grey and orange are those that are not traced either through failing to meet the definition of a close contact or because they are assumed to unidentifiable; orange are all secondary cases that are shorter than 15 min or unidentifiable.

Figure 3

Impact of different assumptions for the definition of a close contact on: (A) the total number of contacts traced per index case; (B) the number of secondary contacts that are not traced per index case; and (C) the probability that at least one secondary case is not traced per index case. For (A) and (B) the crosses mark the mean value, boxes contain the 50th percentiles while bars contact the 95th percentiles, and colours correspond to those in figure 1A—distributions are across all respondents to the survey and across stochastic realisations. (Based on an SEIR model with latent period 4 days, infectious period 2 days, R₀=3).

Figure 4

Impact of different values for the initial reproduction number of the primary case; on (A) the number of secondary contacts that are not traced, and (B) the probability that at least one secondary case is not traced. changing this reproduction number does not affect the number of contacts traced. For (A) the crosses mark the mean value, boxes contain the 50th percentiles while bars contact the 95th percentiles—distributions are across all respondents to the survey and across stochastic realisations. (Main results are based on an SEIR model with a latent period of 4 days (and three latent classes), and an infectious period of 2 days; other points, in blue, use a latent period chosen from a lognormal distribution and an infection period between 2 and 3 days, and are based on a model with one, two or three latent and infectious classes).