Checking back to see who’s been in contact with newly identified patients with an infectious disease is a standard technique for containing an epidemic, but experts argue that it’s particularly important for the coronavirus pandemic.

Trevor Bedford, an epidemiologist at Seattle’s Fred Hutchinson Cancer Research Center, explains why — and lays out a relatively straightforward system for doing contact tracing in combination with testing — in a Twitter thread spun out today.

“We need a huge push to increase the speed and scale of contact tracing, but this doesn’t necessarily require ‘digital’ solutions,” Bedford writes.

The solution suggested by Bedford and his colleagues in the NextTrace effort makes use of mobile device data, but as a supplement to the traditional phone-based and in-phone interviews used in contact tracing.

Other projects, including the ones being discussed by Apple and Google, aim to use proximity data and Bluetooth links, but more care has to be taken to preserve personal privacy. In a white paper, NextTrace says its system would be complementary to those other projects.

One big challenge is that the system relies on early and frequent testing for coronavirus. Studies suggest that nearly half of the virus transmission events occur during the first few days after infection, before the onset of symptoms. That means contacts with infected patients should be identified and tested as soon as possible.

“It is not ethical (or expeditious) to tell an individual ‘they may have been exposed to COVID’ without the ability to get them tested,” the NextTrace team says.

Unfortunately, despite assurances from the White House, easy access to testing is still a problem.

“Given the critical role that testing will play in this venture, we will launch this system only in areas with sufficient testing infrastructure to support the testing load,” the NextTrace team says. The plan is to roll out a pilot project in partnership with public health officials in one region, but NextTrace says “we do not yet have a partnership in place that can be discussed publicly.”

Another challenge has to do with how many personnel would be required to do traditional contact tracing.

Jeffrey Duchin, health officer for Public Health – Seattle & King County, told journalists this week that experts have worked out a formula for determining how many contact tracers would be required to track down COVID-19 cases.

Previously: Coronavirus sleuth outlines his ‘Apollo program’ for bringing down the pandemic

“Using those rough estimates, we’ll need several hundred contact tracers in Washington state, probably about a third of those or so in King County,” Duchin said. “In addition to that large number of contact tracers, we’ll need people to work on isolation and quarantine, social support and all the other wrap-around services that people may need.”

That workforce is far beyond what currently exists. For example, Chris Spitters, health officer for the Snohomish Health District, said that he had the equivalent of four to five full-time workers available to do contact tracing, and that it can take an entire day for just one case to be completed.

“I’ll tell you right now, with 30 cases a day being reported, we could not do that intense a level of contact notification and monitoring,” Spitters said. “It’s probably got to be substantially less than that.”

Digital tools could make things easier. “Unfortunately, no one has real experience with these. It’s hard to know what will work. But anything that seems like it may help, and is not going to take away from the tried and true methods that we trust, should be welcomed,” Duchin said.

Anthony L-T Chen, director of health at the Tacoma-Pierce County Health Department, said an all-digital system may not be necessary. “There are some low-tech solutions, like using call centers and lesser-trained people who can do some of the initial contact and tracing before handing off to higher-level skills,” he said. “I think there are lots of solutions. We just need to be innovative.”

Here’s today’s Twitter thread from Fred Hutch’s Trevor Bedford:

I wanted to give a basic picture of why #TestTraceIsolate is so important to the control of the #COVID19 epidemic and why we need to move quickly when testing and contact tracing in order to create an effective intervention. 1/14

— Trevor Bedford (@trvrb) April 16, 2020

The core idea of why this is so important is that we have evidence that a substantial fraction of transmission occurs in the ~2 days before symptom onset. Figure from @lucaferrettievo et al (https://t.co/jWibJ9CPxr). 2/14 pic.twitter.com/lPPwxOyRpN

— Trevor Bedford (@trvrb) April 16, 2020

Similarly, He et al (https://t.co/WQi8tAnAq1) estimate 44% (95% confidence interval from 25% to 69%) of transmission events occur presymptomatically, with these generally occurring between 0 and 2 days before symptom onset. 3/14

— Trevor Bedford (@trvrb) April 16, 2020

If someone is infected, but not yet symptomatic, there’s no way for them to know that they may be contagious and so they won’t know to take appropriate quarantine measures. 4/14

— Trevor Bedford (@trvrb) April 16, 2020

Identifying and isolating presymptomatic individuals is key to limiting transmission in a #TestTraceIsolate intervention. And for this to happen, everything needs to move very quickly from the alerting of a confirmed case to the quarantine their contacts. 5/14

— Trevor Bedford (@trvrb) April 16, 2020

Here is a median scenario in which Alice is infected at day 0, but does not show symptoms until day 5, despite being infectious at day 3. This results in transmission to Bob at day 4 and Bob, in turn, infecting Carol at day 8. 6/14 pic.twitter.com/oPCdT2b2br

— Trevor Bedford (@trvrb) April 16, 2020

In order for #TestTraceIsolate to be truly effective, Bob needs to be alerted by day 7 to quarantine. This is a tight timeline. It requires that Alice get tested on day 5 when she develops symptoms and rapid turnover of lab assay and alerting. 7/14

— Trevor Bedford (@trvrb) April 16, 2020

Thus, we often have a ~48hr window after an index case develops symptoms to test and alert contacts. This turnaround is highly challenging, but if done in time and Bob is alerted to quarantine we can break the transmission cycle. 8/14 pic.twitter.com/f0QnBcuTbc

— Trevor Bedford (@trvrb) April 16, 2020

Challenges here lie in getting someone with symptoms immediately tested as well as getting contacts rapidly alerted. However, the basic strategy of #TestTraceIsolate as routinely conducted by public health has demonstrated efficacy. 9/14

— Trevor Bedford (@trvrb) April 16, 2020

Or Alice and Bob could just wear masks, which dramatically reduces their viral shed, making transmission much less likely.

— Close, But No Cigar (@noCigar) April 16, 2020

Yes. Mask wearing is complementary to #TestTraceIsolate strategy of quarantine after exposure.

— Trevor Bedford (@trvrb) April 16, 2020

We need a huge push to increase the speed and scale of contact tracing, but this doesn’t necessarily require “digital” solutions. https://t.co/Jz7Q3cUf6M 10/14

— Trevor Bedford (@trvrb) April 16, 2020

Going forward, we should be empowering local and state public health to do their job with contact tracing, while simultaneously thinking about how new technologies can be brought to bear. 11/14

— Trevor Bedford (@trvrb) April 16, 2020

With @nexttrace, we’re trying to thread this needle and think about strategies that can empower speed and scale of #TestTraceIsolate, while adhering to the fundamental process that we know works. 12/14

— Trevor Bedford (@trvrb) April 16, 2020

To this end, we’re prototyping a survey-based contact tracing platform that can be deployed by public health officials around the United States. Our current thinking is available in the GDoc here: https://t.co/WQpIL9Dr8d 13/14

— Trevor Bedford (@trvrb) April 16, 2020

Huge thanks to public health colleagues and to the @nexttrace team including @alliblk, @rebeccaegger, @thefreemanlab, @marcprecipice, @colinmegill, @dbsmasher, @miguelp1120, @trs, @dylanbgeorge and Leah Alpert for help thinking through this problem. 14/14

— Trevor Bedford (@trvrb) April 16, 2020

Does this also require testing asymptomatic people and if so how do you determine frequency?

— Ross MacDonald (@RossMacDonaldMD) April 16, 2020

I believe you can restrict to testing asymptomatic contacts of a confirmed case. You don’t need blanket asymptomatic testing.

— Trevor Bedford (@trvrb) April 16, 2020