Successful contact tracing systems for COVID-19 rely on effective quarantine and isolation

Abstract

Models of contact tracing often over-simplify the effects of quarantine and isolation on disease transmission. We develop a model that allows us to investigate the importance of these factors in reducing the effective reproduction number. We show that the reduction in onward transmission during quarantine and isolation has a bigger effect than tracing coverage on the reproduction number. We also show that intuitively reasonable contact tracing performance indicators, such as the proportion of contacts quarantined before symptom onset, are often not well correlated with the reproduction number. We conclude that provision of support systems to enable people to quarantine and isolate effectively is crucial to the success of contact tracing.

Fig 1. Schematic diagram of the contact tracing model.

Infected individuals are initially not symptomatic (yellow). Some infections eventually develop symptoms (red), others remain asymptomatic for the duration of infection. For the index case who was not traced (0), there is a delay between onset of symptoms (red) and getting tested. Isolation occurs at some point between symptom onset and testing. There is a subsequent delay from testing to the test result being returned (+) and tracing of contacts. Traced contacts (1–4) are quarantined when contacted by public health officials (phone icons) and are isolated and tested immediately on symptom onset. Traced contacts (3) who are already symptomatic prior to being traced are isolated immediately when contacted. Traced contacts (4) that have already isolated prior to being traced are not affected. Contacts that cannot be traced (5) may still get tested and isolated after developing symptoms, but this is likely to take longer. Asymptomatic infections (1) do not get tested or isolated, but will be quarantined if they are a traced contact. Some asymptomatic infections may be untraced, in which case they will not be quarantined or isolated (not shown).

Fig 2. The impact of contact tracing on the effective reproduction number R_eff is strongly affected by the proportion of contacts traced, the tracing speed and the effectiveness of quarantine and isolation.

Graphs show the effective reproduction number relative to the no-control scenario (R_eff/R₀) against the mean tracing time, E(Days to trace), for a range of tracing probabilities and quarantine/isolation effectiveness. R_eff/R₀ = 1 when there is no contact tracing or case isolation.

Fig 3

The proportion of cases quarantined or isolated within 4 days of the index case being quarantined or isolated (A) is the most robust indicator of the performance of the contact tracing system, measured by the reduction in effective reproduction number R_eff relative to the no control case. Other indicators (B)-(D) are not reliably correlated with R_eff across all model parameters. Each plotted point corresponds to one combination of model parameters: fast tracing (same time as test result) of 50% of contacts (stars); fast tracing of 100% of contacts (crosses); slow tracing (mean 3 days after test result) of 50% of contacts (circles); slow tracing of 100% of contacts (pluses); varying effectiveness of isolation and quarantine are shown by different colours (see graph legend). Other parameter values shown in Table 1. The horizontal axes show the proportion of cases meeting the specified performance indicator. Vertical dashed lines show the minimum and maximum values of the indicators computed from New Zealand case data. Indicator (D) can be calculated for all N = 93 cases in the dataset. Indicators (A)-(C) require an index case to be identified, which means they can only be calculated for 81 cases. 37 of these 81 cases have multiple potential index cases. The left-hand dashed line in (A)-(C) shows the result of selecting the index case corresponding to the smallest value of the indicator; the right-hand dashed line in (A)-(C) shows the result of selecting the index case corresponding to the largest value of the indicator.

Fig 4

The proportion of cases quarantined or isolated within 4 days of the index case being quarantined or isolated (A) is still the most robust indicator of the performance of the contact tracing system under different levels of presymptomatic transmission (20%, black ‘*’, 35% red ‘x’, 50% blue ‘o’). Other indicators (B)-(D) are not reliably correlated with R_eff across all model parameters. Each plotted point corresponds to a randomly chosen combination of: tracing coverage p_trace, tracing speed ${\overline{T}}_{trace}$, quarantine effectiveness c_quar, isolation effectiveness c_iso, and probability of testing for untraced clinical cases p_test. Other parameter values shown in Table 1. The horizontal axes show the proportion of cases meeting the specified performance indicator.