How to Talk About Pandemic Response Strategies

… and how to understand what the experts mean when they talk about them.

This story originally appeared as a public post on my Facebook account, where I have written a number of lengthy posts on the topic of the COVID-19 pandemic, particularly as viewed from an Australian perspective. Although it has already been published there, I have chosen to publish it as my first story here, since I intend to refer to it in my next piece of writing on the topic.

It’s time to discuss the latest hot topic in Australia’s COVID-19 saga. No, I’m not talking about the Daily Tele’s “Dan-Made Disaster” headline, or Star City casino being hit with a whopping $5000 fine over COVID-19-related breaches (how will they ever afford to pay?). I’m talking about the word that is starting to make new headlines, whether it’s numerous ABC headlines about epidemiologists coming out in support, or numerous AFR headlines about business lobby groups coming out against it — that’s right, it’s time to talk about elimination.

But I’m not going to talk about it yet — at least, I’m not going to use this story to make a case for it. Instead, I’m going to try to clear up some of the overwhelming noise surrounding the topic that I see in the media (both social and mainstream). This isn’t about trying to reduce complexity — quite the opposite. We need to grapple with the complexity if we are to have any consideration of the options more meaningful than simply picking which team we want to be on. It’s going to take some time to get through — grab a drink (alcoholic content at your discretion) and make yourself comfortable.

So, what the hell does it all mean? Disclaimer: I am not an epidemiologist. I’m one to two degrees of separation from plenty of epidemiologists, which gives me exactly zero epidemiological qualification of my own. What I will try to share are explanations — with what understanding I’ve managed to gather over the course of this year — of some key concepts in epidemiology, and how they relate to the public health policy decisions facing our state and federal governments. Some elements of these descriptions and understandings could be wrong, because I’m not an epidemiologist! I’m always very happy to learn from someone who knows better or has better sources (and may edit this story as necessary).

In order to have any meaningful discussion, we need to understand what we’re talking about, because if I say “elimination” and to you that means “eradication”, or if I say “suppression” and to you that means “flattening the curve”, then we can’t actually have a discussion about the merits of any given strategy because we aren’t talking about the same thing.

And before I talk about any of the terms above, I want to introduce two terms/symbols used in epidemiological analysis of diseases and their spread. R0 (which should, formatted correctly, be an “R” followed by a subscript zero) and Reff (“R” followed by a subscript “eff”, short for “effective” — can also be denoted “R” followed by a subscript “e”, or “R” with no subscript at all, but it is clearer in unformatted writing online to use “Reff”).

R0 is the basic reproduction number of an infection, and represents the average number of subsequent cases that a single case of that infection is expected to generate directly in an entirely susceptible population — i.e. how many people does one person with a disease infect. It depends on many variables, such as how the infection spreads, how much of the pathogen is needed to cause an infection, how long an infected individual is in a state capable of infecting others, etc etc. It is a fairly good shorthand for answering the question “how contagious is this disease?”, although I’m sure that is an oversimplification (but a useful one). For the virus of the hour — SARS-CoV-2 infection in humans — the R0 sits in a fairly wide range of uncertainty because it is so new and we’re still learning so much about it. A current Australian government publication which cites a review of 12 recent studies, gives values of R0 around 3 (“…estimated the basic R0 to be 3.28 and the median R0 to be 2.79” — I’m not sure as to the distinction between those two, but it’s not significant for this story). In other words, one infected person, with no-one modifying their normal behaviour in any way, will result in about three other people becoming infected. And each of those will infect another three people on average. And so forth. However, this is the average of a distribution that may be very broad — it could be that many infect only one other person, or none at all, while a few individuals infect dozens of others each.

Reff is the effective reproduction number of an infection. That represents how the infection in question is actually spreading in a given population, under the influence of whatever we are doing to prevent (hopefully) or promote (yes, sometimes we do that too) the spread of infection within the population being measured. So it includes the effect of things we’re doing in response to the current pandemic (stay-at-home restrictions, physical distancing, masks, extra hand washing, cough etiquette, etc), and things that are already part of the environment (movement and interaction and living patterns — housing, transport, etc), and more. Reff gives us the result of all the various factors affecting the spread of a given infection within a particular population in question. Change anything you’re doing, or change anything about what population you are measuring, and you have the potential to change Reff. It is even possible to shift Reff higher than R0 (for example, chicken pox parties) — because R0 represents what happens if you don’t do anything at all to affect the spread. As the extraordinarily low incidence of influenza-like illness this year has shown, it is entirely possible to change Reff for a particular disease as an indirect result of doing something else (in this case, trying to change Reff for COVID-19). Reff is the magic number, but the problem is we can only measure it by collecting data that is not yet available at the time for which we are measuring it. With COVID-19, the Reff today in any given population will only be measurable in a couple of weeks’ time at the very earliest.

If Reff is greater than 1, the incidence of the disease in the population increases at an exponential rate over time, until such time as the disease starts to run out of new people to infect (which then reduces Reff), resulting in an S-shaped “logistic” curve of infections. If Reff is equal to 1, the incidence of the disease in the population remains constant. If Reff is less than 1, the incidence of the disease decreases over time.

Got that? Excellent. Now we’re ready to start to talk about response strategies for dealing with COVID-19 (as well as at least one that is often included due to misunderstanding, but needs to be clarified because it is rightly considered impossible and is sometimes conflated with a potentially valid strategy in order to undermine it). I’m going to list all the strategies, such as I’m aware of them, and try to consider in brief the merits of each, along with what we know of how these appear to have played out so far. These I will list in descending order of Reff.

“Let ‘er rip”: Reff = R0. No attempt to control the spread, everyone go about their life as if there were no pandemic. This results in exponential growth until the number of either currently infected or recovered (possibly immune) individuals in the population starts to limit the opportunity of new cases to come into contact with uninfected individuals. If there is a goal here (other than a lack of guidance, leadership or will), it is to achieve a population among whom enough are recovered (and therefore presumed immune to fresh infection) that the virus will no longer spread significantly either from the remaining carriers or any new carrier introduced to the population. A number of countries followed this course during the early days of the COVID-19 pandemic, although in many cases either because they didn’t take the disease seriously, or because they felt that taking strong action to limit spread of the virus would cause excessive harm to their economies. Italy was one of the first to follow this path, then rapidly turn away from it as their health system collapsed, and they scared a lot of countries into doing better. Sweden was to be the poster child for holding the line on something close to this strategy (as it was guided by a qualified epidemiologist) — they did try to isolate vulnerable groups, so it’s not strictly no control at all, but close to no control in the wider population beyond reminders to wash your hands. Sweden has… not gone well, by any measure.

“Flatten the curve”: Target Reff >=1, reducing to the extent necessary to avoid uncontrolled exponential growth in the number of cases, in order to prevent overload of health services. This was very popular in the early days of the pandemic — there were some lovely animated graphs going around. At that time, I was convinced this was the desirable option and voiced support for that goal (primarily because I didn’t yet understand anything about the options I’m going to describe below: I told you I’m not an epidemiologist). It has been the strategy favoured by many large countries, to varying degrees of success and failure. Many US states talk about their performance as a proportion of hospital and ICU beds still available. Some have managed to keep spread below levels at which hospitals were overwhelmed, some have emphatically not. At the time of writing, the number of confirmed COVID-19 cases in the US represents just over 1% of their total population, and nearly 140,000 of those have already died. But most countries with “flatten the curve” strategies (many European countries, for example) have done much better than the USA. The ongoing challenge for “flatten the curve” is that there continue to be plenty of active cases of disease in the population, so every aspect of life which affects Reff requires ongoing management to prevent an uncontrolled return to exponential case growth, until one of two things happens: either (a) a vaccine is developed, allowing us to bring Reff below 1 without those controls, or (b) enough of the population have been infected, recovered and developed natural immunity as a result, that removing the controls on spread will not result in overwhelmed health services.

“Suppression”: Target Reff <1. As described above, where this is successfully achieved, the number of active cases in the population will decrease over time. The further below 1 Reff is, and the more of the time it stays there, the more rapidly cases will decrease. Under a suppression strategy, Reff may from time to time exceed 1 due to local outbreaks, which may occur because a low level of ongoing cases within the population is expected, but the goal is to detect and respond to any such outbreaks in order to bring Reff back below 1. Obviously, the more promptly any outbreak is detected and the more effectively contacts of known cases are traced, the better the response can be and the sooner Reff can be brought back below 1. Under suppression, some subset of control measures — physical distancing, reduced use of public transport, etc — must remain in place to keep the average Reff below 1, and certain especially high-risk activities (e.g. packed clubs and indoor concerts, choir rehearsals, massive indoor parties, etc) need to remain off-limits until a sufficiently effective vaccine is available. A suppression strategy will never result in widespread immunity in the population without a vaccine. Australia, to date, is rightly held as an example of suppression done right. Neither the Melbourne outbreak, nor the clusters being traced in Sydney, are indications that the strategy has failed. Provided the countermeasures deployed in response are successful in bringing Reff back below 1, our suppression strategy is functioning as intended.

“Elimination”: Target Reff=0. By bringing Reff as close to zero as possible, the goal is to get rid of ALL infections within the population, such that members of that population can go about their lives with no consideration of the risk of being infected by another person within that population (or at least, in the same way that people don’t generally consider the risk of being struck by a meteorite, though it is theoretically possible). This does not mean that there cannot be a single case of the virus anywhere within the state or country in question — just that there cannot be a single case of the virus arising outside of managed quarantine or isolation, i.e. there cannot be cases arising within the community who have had the opportunity to spread the infection to others. If elimination is successful, all local controls to reduce Reff can then be removed, provided the population is carefully monitored for any new case (because, with no controls in place, such a case would spread according to Reff=R0). New Zealand is the obvious example of a (so far) successful elimination strategy. While they still demonstrate ongoing caution in the form of such measures as testing anyone with symptoms and isolating those until they are confirmed negative, there are no constraints on day to day life within the borders of the country. People can pack into crowded clubs with inadequate ventilation. People can go to huge concerts or hold choir rehearsals. People can cram into trains and buses. Life within the country has genuinely returned to a strong semblance of pre-pandemic normal, and has done for what is now going on six weeks. That means that every arrival from a location where spread of the virus has not been eliminated — i.e. any other country — must be held in managed quarantine for fourteen days, and tested and confirmed negative before they leave quarantine. International arrivals are now the only plausible source for re-introduction of the virus into the general population… so they’re taking it pretty seriously (not perfectly — there have been three or four cases of people temporarily escaping mandatory quarantine — but they are taking it seriously).

“Eradication”: Reff irrelevant: zero instances of the virus in any person worldwide. This is not going to happen. Not even when we get a vaccine. Even if it’s a really good vaccine. In all of recorded history, there have been exactly two diseases eradicated due to human intervention: smallpox and rinderpest (I had to look up the latter, as it is a disease of livestock not humans). There are two more diseases currently targeted for eradication — polio and guinea worm disease. That’s it. Every other disease we’ve ever discovered still exists in the world somewhere, and has to be managed with things like treatment and/or vaccination, and SARS-CoV-2 will — with near certainty — likewise remain present in the global population for the indefinite future.

I said I wasn’t going to use this story to make a case for elimination — or indeed for any other strategy — and for now, I won’t. I do intend to write more on the topic, but it’s after 1 a.m. now and — since I live in Melbourne — I’ve got at least five more weeks of stage 3 stay-at-home restrictions ahead of me, so there will be plenty more time to write. In the mean time, let’s make sure we understand what it means, when an epidemiologist or public health expert is quoted in the media as supporting or opposing Australia changing to a national strategy of elimination (and what less qualified folks, like politicians and business leaders, are arguing for when they take up positions on the matter). As always, there are plenty of poor quality arguments being bandied about by unqualified voices in the media. As Sweden has proven, even epidemiologists can make massive errors of judgement in the context of this pandemic, but I think it’s important that — all else being equal — we give greater consideration to those with genuine qualifications and expertise than we do to those (including myself) without. So between now and whenever I get to writing about what I personally think would be the most desirable strategy for our various governments to pursue, feel free to take in as many other opinions — expert and otherwise — as you happen to find… but hopefully, with the above as a guide, you’ll feel more comfortable in knowing what they’re actually talking about.