A Proposal to Reorient Contact Tracing for COVID

– By Bob England MD

Background on the trends, behavior & transmission of COVID

The events of the past week involving the White House, and the news coverage of them, have demonstrated all too well the continuing ignorance of basic truths regarding the COVID pandemic.  More than 7 months into this, concepts as simple as incubation time, infectiousness, quarantine, and isolation remain misunderstood by public officials and newscasters alike.  Conflicting perceptions are not only the result of inattention to information, but are fueled by deliberate distortion of that information to suit individual interests. 

All of this, combined with a natural exhaustion from social mitigation measures, have driven many to disregard simple protective actions and to engage in truly reckless behavior.  Young people, with their natural sense of immortality, have engaged in unprotected partying.  Both young and old have flocked to high-risk settings such as indoor bars and restaurants, theaters, and gyms.

Others, taking their cue from leaders who proclaim the pandemic to be a hoax or at least a minimal threat, have participated in church services, political rallies, or other mass gatherings with little or no protection.  Still others, responding to social and political pressures, have taken to the streets in mass protests that involve, at best, sporadic social distancing and masking.

All of these, as exemplified by the now infamous White House gathering held to formally announce the latest Supreme Court nomination, have created ample opportunities for so-called “super spreader” events.  What is not generally appreciated by the public, and not adequately addressed by public health agencies, is that the entire pandemic may be primarily driven by such events.

Much has been made of the overall transmission rate, the R0, of the COVID pandemic. But the R0 is merely the mean transmission rate — how many new infections, on average, result from a single existing infection.  That’s an important statistic, and it is true that if the R0 is greater than 1, the epidemic curve is increasing, while if it is less than 1, the number of new infections is decreasing.  But like all averages, there can be a lot of meaningful differences between the mean, median, and mode.

In the case of this pandemic, it appears that many people infected with the virus do not transmit the infection on to anyone else.  In other words, the median and mode of the transmission rate are smaller than the mean.  Perhaps that is due to all the attention to this disease and the desperate desire of most infected persons to not infect others in their family or immediate social sphere.

Rather, most new infections may be occurring in clusters, either in these so called “super-spreader” events or smaller clusters of several infections among participants in group gatherings, wherein a small number of infected persons manage to infect several others at a single opportunity.

This has huge implications for the control of the epidemic, including the appropriate use of contact tracing.

Basics of Contact Tracing

Contact tracing was developed as a public health tool primarily for diseases in which there was a useful intervention for persons known to be exposed.  If we had a preventive treatment that could be given to exposed persons while they were still incubating the infection but not yet ill, we could use contact tracing to find them and treat them, preventing cases of disease.  Examples include tuberculosis, certain sexually transmitted infections, bacterial meningitis, and so forth. 

Even if we didn’t have such a preventive treatment, we sometimes use contract tracing to inform and educate exposed persons if either:

• they could watch for symptoms and avail themselves of early treatment;  or

• modify their behavior or even quarantine themselves so that they didn’t further transmit the disease to others until enough time had passed that we were sure that they could no longer do so. It is with the latter objective that we might try contact tracing with COVID to lessen spread of the disease. That’s the basis for the 14-day quarantine period for those known to have significant exposure to a person with COVID.

However, contact tracing has always been more effective with diseases that are relatively less common, well diagnosed, and thus easier to track.  We’ve never tried contact tracing with influenza, for example, even though it kills an average of 20,000 Americans each year, and sometimes 60-80,000 or more.  It’s simply too widespread, and too often unrecognized, for us to attempt to identify contacts. 

Instead, every year, we declare that there is “community-wide spread” of flu, and we tell everyone to behave as if they may be exposed every single day.  Now we are in a situation where at times, there are more active COVID infections and more community-wide spread than in any typical flu season.  

So how do we use contact tracing to address COVID?

The Reality of COVID Numbers & Contact Tracing

During the peak of COVID transmission in Arizona, we were reporting more than 1,000 cases per day.  All indications are that we should expect new infections in that range during the winter months, soon to come.  Early on, the unavailability of testing, combined with the predominance of asymptomatic or mildly symptomatic infection, led to estimates of perhaps 50-times the number of true infections to reported cases.  This was borne out by early seroprevalence studies in California. 

As testing became more available, estimates from CDC indicated perhaps a ten-fold incidence of true infections to those reported.  For the sake of argument, the rest of this paper will assume that this has been further cut in half, so that the number of true infections to those reported is only 5-fold. This is an optimistic assumption, given wide variability in the sensitivity of tests, the rate of truly asymptomatic infection (varying from 20% to 50%, depending on the study), the continued barriers which exist to testing, and the reporting issues that persist from the increasingly common point-of-service tests.

So let’s assume that as we again approach 1,000 case reports per day in Arizona, we are actually experiencing 5,000 new infections per day. Let’s say that if we attempt traditional contact tracing, we’ll somehow interview all of those 1,000 reports and identify 5 close contacts per case.  So that’s 5,000 contacts per day, whom we should then contact and tell to stay in quarantine for 14 days from their last exposure to the case.  5,000 contacts x 14 days means that on any given day, we would have approximately 70,000 people, or ~1% of the entire population, in quarantine.  And that’s an ideal number. 

But each of the 4,000 real infections that we don’t know about will also, in reality, have the same 5 contacts each.  So there will be an additional 20,000 contacts, each and every day that we cannot notify, or 280,000 contacts at any one time who should be in quarantine, but that we can’t know about.

Thus, we’ll only ever be able to notify 20% of those we should, and that’s if every single reported case responds to our call, and divulges every single contact.  That’s far from reality.  In reality, under the best of circumstances, with such a widespread infection and typical compliance, we’ll be lucky if we can notify even 10% of those exposed.

In addition, most of those contacts will be well-known to the reported individual.  Often they will be family members, even living in the same household.  They will mostly have already been informed by the reported case.  If requested, the case will usually also be willing to notify most others.  Thus, simply educating the case and asking him/her to notify those who may have been exposed will usually satisfy.

In other words, while it’s not a bad thing to notify exposed persons, it will have little to no effect toward containing the pandemic.  It’s just not possible to use a traditional contact tracing strategy to turn the epidemic curve around.  Infections must be much less frequent, and nearly always identified, before we can use it effectively in that manner.

The proposed new look-back contact tracing method

So instead, let’s put our effort toward making more of a difference.  Given that most (some estimates are up to ~80%) of infections are due to mass events, let’s ask about them, and pursue them.  Doing so might spur testing and identification of many attendees who would otherwise never have been identified, allowing them to self-notify their own contacts.

In addition, it will draw attention to the otherwise unpublicized events causing mass infection, and may therefore cause behavior change to limit their existence in the future. Even if it doesn’t fully impact behavior in the desired manner, it’s certain to have a better chance at doing so than our current strategy of remaining blinded to the consequences of mass gatherings.

So, rather than always looking forward, and asking each case who they may have exposed from two days prior to symptoms or positive test result and onward, let’s simply tell them that they need to notify such persons themselves.  Then ask the following:

During the 2 weeks prior to your onset of symptoms (or if asymptomatic, 2 weeks prior to positive test):

• Did you attend any group gatherings?  (Such as a party, reception, wedding, funeral, church service, political rally, political protest, sporting event, attendance at a bar, or a restaurant, etc.)

• If yes,

• Roughly how many people were present?  (Not just at your own table, for example).

  1. How long did it last?  How long were you there?

  2. Indoors or outdoors?

  3. Was everyone masked?  Or how many?  Were you?

  4. Who organized it?   Or what was the venue?  (e.g., bar or restaurant)

Then it’s up to the jurisdiction how to pursue.  Ideally, we would contact the organizer to get more details, contact info, and follow-up with each attendee.  More likely, depending on the threshold decided upon by the jurisdiction (see options below), make a public announcement.  To wit:

“Anyone who attended X event at X location on X date should be aware that COVID transmission likely occurred there. You should quarantine yourselves for 14 days from the date of the event, seek testing now and at the completion of the 14 days, and if you have any symptoms, contact your healthcare provider for advice.”

The threshold for such an intervention might be a single case identifying a suspect event, or two or more case reports naming the same event. 

This is a real opportunity for education that might strike home.  It’s purely anecdotal, but in my own experience working in STDs, patients who should have well understood their own risk behaviors and risk factors would suddenly sit up and take notice when notified that they were a contact to HIV.  I was surprised that they were surprised, but there is something in human behavior that we take notice once it becomes more personal.   

So tell people.  “If you were at Restaurant/Bar X on Date Y, it appears that some persons there contracted COVID.  You should….” 

Not only will that make those at a particular bar on a particular night take notice, but anyone who has been foolish enough to patronize the same location on other nights, or any other crowded venues, will also take notice, and hopefully not do it again. I think this would be especially helpful for bars, where large numbers of people congregate seemingly without concern.  Publicize several dates at several bars, and people will think twice before patronizing them again. 

I know this is a long shot.  Not particularly politically correct during a time when the emphasis is on re-opening.  But we need to reopen without ridiculously risky venues being open indoors, without social distancing or masks, or the sporadic superspreading events will continue to fuel our infection rates, and may force the elected officials to shut everything down again. 

Backwards contact tracing won’t just identify more missed cases, it may help communicate the true risks going forward.