Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Media update: 11 September 2020

graph showing the probability of transmission as a function of generation time (left) and time from onset of symptoms to transmission (TOST) (right) for a given duration of the incubation period, relative to the peak probability.
For description, see final point below

Preprint paper released this week on MedRxiv by Ferretti et al. at Oxford University’s Nuffield Department of Medicine. These summary points below can be attributed as quotes to the first author Dr Luca Ferretti, senior researcher at Oxford University’s Nuffield Department of Medicine.

For media questions, quotes or interview requests with one of the authors, please contact Andrea Stewart. WhatsApp / call +44 7528 132 489 (ideally) or email andrea.stewart@bdi.ox.ac.uk

What did our COVID-19 transmission research find?

  • We studied transmission timing in individuals who eventually developed symptoms (i.e. excluding asymptomatic individuals, who never show symptoms). We found that transmission timing is mainly linked to the onset of symptoms, rather than time since infection: infected individuals with a long incubation period tend to transmit later than individuals with early onset of symptoms.
  • Strictly pre-symptomatic transmissions account for ~42% of all transmissions; early symptomatic transmissions (occurring on the day of symptom onset and the next day) account for an additional large fraction ~35% of all transmissions. 
  • On average, the time profile of transmissions peaks around onset of symptoms, with a standard deviation of approximately 3 days. Symptomatic individuals are infectious for longer, but transmissions are reduced after 3 days post onset because of isolation of symptomatic cases.

  • The incubation period affects the rise in COVID-19 infectiousness up to its peak at symptom onset: longer incubation periods correspond to longer pre-symptomatic infectious periods.

  • Current CDC and WHO guidance for contact tracing guidelines supports tracing of contacts for up to 2 days before the onset of symptoms for symptomatic people. For asymptomatic asymptomatic individuals, up to 2 days before a positive test. The UK test and trace programme also traces 2 days of contacts before symptoms, or before a positive test for asymptomatic individuals. 

    • For symptomatic individuals, our analysis suggests that 2 days of tracing before symptoms could miss approximately 16% of all onward transmissions from these individuals in the population. Extending this interval would trace more infected contacts, but with a third of the maximum benefit/cost.

    • For symptomatic individuals with a longer incubation period, these individuals might have a higher proportion of traced contacts missed with only 2 days of tracing, for example, one third of an infected person’s transmissions are missed if their incubation period is 10 days (see figure below).

How does this research evidence inform guidelines and interventions designed to stop COVID-19 transmission? 

  • The large fraction of transmissions that occur either before or shortly after onset of symptoms means that isolation of cases starting more than 2 days after onset of symptoms is far from sufficient to control the epidemic. 

  • Physical distancing, mask wearing, rapid mass testing, manual and digital contact tracing are key non-pharmaceutical interventions that are able to stop transmissions before and around the onset of symptoms, and should be part of an effective COVID-19 control strategy.​​

  • Our results on early symptomatic transmission could help to update sector-specific guidelines and public health messaging. For example, individuals who develop early signs of illness should strictly follow advice and take extra precautions if attending settings with many individuals and / or vulnerable individuals, especially during the first two days after symptoms appear. 

    • Guidance could be updated to suggest those who develop mild or non-specific symptoms should adhere strictly to social distancing, handwashing, mask-wearing and the maximum feasible reduction in contacts during the first 2-3 days after mild symptoms appear. 

  • We recommend the date of infection should be investigated and recorded by contact tracing systems, both manual and digital, whenever possible. Public health authorities should consider appropriately longer pre-symptomatic intervals for contact tracing whenever an estimate of the date of infection is available and precedes symptoms by more than 4-5 days. 

  • Though we found little evidence of transmission from 5 days after onset of symptoms, we urge against individuals being considered "safe" prematurely. Transmissions could be rarer after a few days of symptoms due to individuals already isolating from others, as well as due to lower biological infectiousness, and these precautions should not be relaxed.

  • If the duration of the infectious period for fully asymptomatic individuals is similar to the one for symptomatic individuals, tracing of contacts of fully asymptomatic cases for 2 days before a positive test could miss much of their infectious period.

  • When the temporal profile of infectiousness is used for developing risk scoring for digital contact tracing (apps) / exposure notification systems (ENS), both time of onset of symptoms and (when available) time of exposure should be included.

  • With these improved estimates of infectiousness, ENS could be updated to more accurately notify contacts most at risk of infection. Currently the Google-Apple ENS only distinguishes two levels of risk (low and high); a more detailed characterisation of infectiousness could improve accuracy to reduce the number of false positive and false negative notifications, preventing more transmission with fewer individuals quarantined. 

  • The pre-print Ferretti et al. The timing of COVID-19 transmission. Posted on MedRxiv. 

  • Available soon, estimated 11-12 September a scientific blog summarising this latest preprint paper. 

  • These findings have not been formally peer reviewed. A manuscript is in preparation for submission. 

  • Updated today (11/09/20) Sample data from the preprint paper (main image, above). Figure 3: probability of transmission as a function of generation time (left) and time from onset of symptoms to transmission (TOST) (right) for a given duration of the incubation period, relative to the peak probability. The black line represents the average with respect to the incubation period distribution. The envelopes correspond to the pointwise 95% CI.